技术领域technical field
本发明属于计算机技术，尤其涉及一种电熔镁炉异常工况识别方法及安全控制方法。The invention belongs to computer technology, and in particular relates to a method for identifying abnormal working conditions of an electric fused magnesium furnace and a method for safety control.
背景技术Background technique
电熔镁是一种重要的耐火材料，它具有以下优点：强绝缘性、高熔点、抗氧化能力强及结构紧密，因此在多种工业生产中得到了广泛应用。电熔镁的原材料为菱镁矿，在大多数情况下，原材料品位低，矿物组成复杂多样。Fused magnesium is an important refractory material, which has the following advantages: strong insulation, high melting point, strong oxidation resistance and compact structure, so it has been widely used in various industrial production. The raw material of fused magnesium is magnesite. In most cases, the raw material is of low grade and the mineral composition is complex and diverse.
电熔镁使用的熔炼设备为三相电熔镁炉，电熔镁炉的电流控制系统跟踪不同工况下的电流设定值完成熔炼过程。在实际应用中，当原料的粒度、成分、熔点等发生变化时，电极底部与熔池之间的距离会产生波动。这种未知的波动将导致电流的波动，之前的电流设定值将不再适用于变化的工况，此时会产生更高的能量消耗，系统性能下降甚至具有安全威胁。The smelting equipment used for fused magnesium is a three-phase fused magnesium furnace, and the current control system of the fused magnesium furnace tracks the current setting value under different working conditions to complete the smelting process. In practical applications, when the particle size, composition, melting point, etc. of the raw materials change, the distance between the bottom of the electrode and the molten pool will fluctuate. This unknown fluctuation will lead to the fluctuation of the current, and the previous current setting value will no longer be suitable for the changed operating conditions, which will result in higher energy consumption, system performance degradation and even security threats.
以排气异常工况为例，当原料的粒度发生变化时，原来的电极动作无法完全使熔炼过程中产生的二氧化碳气体排出，即排气工况的电流设定值不再适用，电熔镁炉内的压强将升高，当压强升高到一定程度时，高温熔浆将随气体一起排出炉外，飞溅的高温熔浆将对设备造成严重损坏，对操作人员的人身安全造成极大威胁。同时，在飞溅产生的过程中，大量的能量损失提高了电熔镁炉的单吨能耗。由于外界的强烈干扰及变量之间的强耦合、非线性的存在，理解变量间相关关系，以及获取异常工况下的模型是十分困难的。Taking the abnormal exhaust condition as an example, when the particle size of the raw material changes, the original electrode action cannot completely discharge the carbon dioxide gas generated during the smelting process, that is, the current setting value of the exhaust condition is no longer applicable, and the fused magnesium The pressure in the furnace will rise. When the pressure rises to a certain level, the high-temperature molten slurry will be discharged out of the furnace together with the gas. The splashing high-temperature molten slurry will cause serious damage to the equipment and pose a great threat to the personal safety of operators. . At the same time, in the process of splash generation, a large amount of energy loss increases the energy consumption per ton of the fused magnesia furnace. Due to the strong external interference, strong coupling between variables, and the existence of nonlinearity, it is very difficult to understand the correlation between variables and obtain models under abnormal conditions.
目前，现场操作人员通过观察到的电流信息，图像信息和声音信息及自己的经验来辨识异常工况，根据识别的异常工况，给出相应的调整方案，大部分通过手动的方式实施，自动化水平较低。人工识别和操作对人为误差敏感，而且操作人员在处理多源信息时能力是有限的，往往会忽略一些变量的影响。在有限的时间内，在巨大的精神压力下，操作人员往往很难制定出有效的决策。操作人员的人为调整方法完全依赖于操作者各自的经验，很难保证决策的及时性及准确性。如有操作不当、疏于检测或因不可抗拒的自然因素引起设备故障而导致生产中断等事故，将会给生产带来巨大的浪费和损失。At present, field operators identify abnormal working conditions through the observed current information, image information and sound information and their own experience, and give corresponding adjustment plans according to the identified abnormal working conditions, most of which are implemented manually and automatically. low level. Manual identification and operation are sensitive to human error, and operators have limited ability to process multi-source information, and often ignore the influence of some variables. In a limited time and under great mental pressure, it is often difficult for operators to make effective decisions. The operator's manual adjustment method is completely dependent on the operator's own experience, and it is difficult to ensure the timeliness and accuracy of decision-making. If there are any accidents such as improper operation, neglect of detection or equipment failure caused by irresistible natural factors, such as production interruption, it will bring huge waste and loss to production.
发明内容SUMMARY OF THE INVENTION
针对现有存在的技术问题，本发明提供一种电熔镁炉异常工况识别及控制方法，上述方法应用于电熔镁炉工况中提高了矿产资源的综合利用率、降低了能耗。In view of the existing technical problems, the present invention provides a method for identifying and controlling abnormal working conditions of an electric fused magnesia furnace.
本发明提供一种电熔镁炉异常工况识别及控制方法，包括：The invention provides a method for identifying and controlling abnormal working conditions of an fused magnesia furnace, comprising:
S1、获取电熔镁炉工况中的预设周期内的在线数据；S1. Obtain online data in a preset period in the working condition of the fused magnesia furnace;
S2、采用相似度匹配策略查看案例库中是否存在与在线数据匹配的案例信息；S2. Use the similarity matching strategy to check whether there is case information matching the online data in the case database;
S3、若存在，依据匹配的案例信息给出当前在线数据的辨识结果，将辨识结果作为当前电熔镁炉工况的异常识别结果；S3. If there is, the identification result of the current online data is given according to the matched case information, and the identification result is used as the abnormal identification result of the current working condition of the fused magnesia furnace;
其中，案例库为预先根据电熔镁炉工况的历史数据建立的各种异常工况的案例信息。Among them, the case library is the case information of various abnormal working conditions established in advance according to the historical data of the working conditions of the fused magnesia furnace.
可选地，所述方法包括：S3a、若案例库中不存在匹配的案例信息，则采用贝叶斯网络推理模型对所述在线数据进行分析，获得辨识结果，将辨识结果作为当前电熔镁炉工况的异常识别结果。Optionally, the method includes: S3a, if there is no matching case information in the case library, use a Bayesian network inference model to analyze the online data, obtain an identification result, and use the identification result as the current fused magnesium Abnormal identification results of furnace operating conditions.
可选地，步骤S2之前，所述方法还包括：Optionally, before step S2, the method further includes:
S2a、根据预设时间段内电熔镁炉工况的历史数据，建立案例库；S2a, establishing a case library according to the historical data of the working conditions of the fused magnesia furnace within a preset time period;
以及根据预设时间段内电熔镁炉工况的历史数据和先验知识，建立贝叶斯网络推理模型。And according to the historical data and prior knowledge of the working conditions of the fused magnesia furnace in a preset time period, a Bayesian network inference model is established.
可选地，步骤S2a包括：Optionally, step S2a includes:
离线收集预设时间段内电熔镁炉工况发生异常的信息；Collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
根据收集的信息，确定异常工况的特征与异常工况中相关变量之间的关系，获得历史数据，根据历史数据建立案例库和贝叶斯网络推理模型；According to the collected information, determine the relationship between the characteristics of abnormal conditions and relevant variables in abnormal conditions, obtain historical data, and build a case library and a Bayesian network inference model based on the historical data;
或者，or,
离线收集预设时间段内电熔镁炉工况发生异常的信息；Collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
根据收集的信息，确定异常工况的特征与异常工况中相关变量之间的关系，获得历史数据，对历史数据进行过滤处理，根据过滤处理后的历史数据建立案例库和贝叶斯网络推理模型。According to the collected information, determine the relationship between the characteristics of abnormal conditions and relevant variables in abnormal conditions, obtain historical data, filter the historical data, and build a case library and Bayesian network inference based on the filtered historical data Model.
可选地，所述S2a包括：Optionally, the S2a includes:
离线收集预设时间段内电熔镁炉工况发生异常的信息；Collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
根据先验知识以及异常工况中相关变量之间的关系，确定贝叶斯网络的节点和结构；Determine the nodes and structure of the Bayesian network according to the prior knowledge and the relationship between relevant variables in abnormal conditions;
根据收集的历史数据及确定的结构，获得贝叶斯网络的参数，从而建立起贝叶斯网络推理模型。According to the collected historical data and the determined structure, the parameters of the Bayesian network are obtained, thereby establishing the Bayesian network inference model.
可选地，所述步骤S1还包括：Optionally, the step S1 further includes:
对获取的在线数据进行过滤处理，以获得过滤处理后的在线数据；Filter the acquired online data to obtain filtered online data;
相应地，步骤S2中采用过滤处理后的在线数据与案例库中的案例信息进行匹配。Correspondingly, in step S2, the online data after filtering is used to match the case information in the case database.
可选地，所述方法还包括：Optionally, the method further includes:
S4、根据异常识别结果，确定设备的剩余生命时间；S4. Determine the remaining life time of the device according to the abnormal identification result;
依据预先建立的剩余生命时间和调整量之间的关系，计算调整量，获得安全控制决策信息，以使控制系统根据安全控制策略信息适应调整；According to the relationship between the pre-established remaining life time and the adjustment amount, the adjustment amount is calculated, and the safety control decision information is obtained, so that the control system can be adjusted according to the safety control strategy information;
其中，剩余生命时间为电熔镁炉工况设备从获取在线数据时的状态到完全无法正常工作的时间；Among them, the remaining life time is the time from when the equipment of the fused magnesia furnace working condition is in the state when the online data is obtained to when it cannot work normally;
所述调整量为电流设定值；剩余生命时间和调整量之间的关系为离线方式确定的关系。The adjustment amount is the current set value; the relationship between the remaining life time and the adjustment amount is determined in an offline manner.
可选地，所述步骤S2包括：Optionally, the step S2 includes:
获取在线数据与案例库中每一案例的变量之间的相似度，判断该相似度是否大于预设阈值，若是，则确定在线数据与大于阈值的最大相似度对应的案例匹配。Obtain the similarity between the online data and the variables of each case in the case database, and determine whether the similarity is greater than a preset threshold, and if so, determine that the online data matches the case corresponding to the maximum similarity greater than the threshold.
可选地，所述方法还包括：Optionally, the method further includes:
S5、若通过贝叶斯网络推理模型获得的异常识别结果确定为当前电熔镁炉工况实际发生的结果，则将该异常识别结果作为案例信息，存储在所述案例库中。S5. If the abnormal identification result obtained by the Bayesian network inference model is determined to be the actual occurrence result of the current working condition of the fused magnesia furnace, the abnormal identification result is used as case information and stored in the case database.
可选地，所述方法还包括：Optionally, the method further includes:
在控制系统根据安全控制策略信息适应调整之后，重复获取电熔镁炉工况中下一预设周期内的在线数据，根据该在线数据的判断过程，以确定步骤S1中的在线数据的异常工况是否排除，若是，则结束，否则，重复获取下一预设周期内的在线数据的安全控制决策信息的过程。After the control system adapts and adjusts according to the safety control strategy information, repeatedly obtains the online data in the next preset period in the working condition of the fused magnesia furnace, and determines the abnormal operation of the online data in step S1 according to the judgment process of the online data. Whether the situation is eliminated, if yes, end, otherwise, repeat the process of acquiring the security control decision information of the online data in the next preset period.
本发明的电熔镁炉异常工况识别及控制方法，可以有效根据电熔镁炉生产过程出现的异常工况，通过在案例库中查找是否存在匹配案例，在不存在匹配的案例信息时，采用贝叶斯网络模型进行异常工况识别，进而根据异常识别结果，制定安全控制决策信息并在控制系统中应用该决策信息，提高了电熔镁炉工况中矿产资源的综合利用率、降低能耗。The method for identifying and controlling the abnormal working conditions of the fused magnesia furnace of the present invention can effectively find out whether there is a matching case in the case database according to the abnormal working conditions occurring in the production process of the fused magnesia furnace, and when there is no matching case information, The Bayesian network model is used to identify abnormal working conditions, and then according to the abnormal identification results, the safety control decision information is formulated and applied in the control system, which improves the comprehensive utilization rate of mineral resources in the working condition of the fused magnesia furnace, reduces the energy consumption.
附图说明Description of drawings
图1A为本发明一实施例提供的电熔镁炉异常工况识别及控制方法的流程示意图；1A is a schematic flowchart of a method for identifying and controlling abnormal working conditions of an fused magnesia furnace according to an embodiment of the present invention;
图1B为本发明一实施例提供的异常工况识别及控制方法的过程示意图；1B is a schematic process diagram of a method for identifying and controlling abnormal operating conditions provided by an embodiment of the present invention;
图2为本发明实施例中控制系统异常工况从轻微程度变为严重程度的生命周期示意图；2 is a schematic diagram of a life cycle in which an abnormal operating condition of the control system changes from a slight degree to a serious degree in an embodiment of the present invention;
图3为当前的电熔镁炉工艺的过程示意图；Fig. 3 is the process schematic diagram of current fused magnesium furnace technology;
图4为本发明实施例中提出的案例推理过程示意图；4 is a schematic diagram of a case reasoning process proposed in an embodiment of the present invention;
图5为本发明实施例中针对电熔镁炉排气异常工况建立的贝叶斯网络推理的示意图；5 is a schematic diagram of a Bayesian network inference established for abnormal exhaust gas conditions of an fused magnesium furnace in an embodiment of the present invention;
图6为排气异常工况辨识结果的对比示意图；Fig. 6 is the comparative schematic diagram of the identification results of abnormal exhaust conditions;
图7为电熔镁炉正常排气工况的电流示意图；Fig. 7 is the electric current schematic diagram of the normal exhaust working condition of fused magnesia furnace;
图8为电熔镁炉异常排气工况的电流示意图；Fig. 8 is the electric current schematic diagram of abnormal exhaust working condition of fused magnesia furnace;
图9为本发明提出的方法的排气异常工况控制效果示意图；FIG. 9 is a schematic diagram of the control effect of the method proposed by the present invention for abnormal working conditions of exhaust gas;
图10为传统方法的排气异常工况控制效果示意图。FIG. 10 is a schematic diagram of the control effect of the conventional method for abnormal working conditions of exhaust gas.
具体实施方式Detailed ways
为了更好的解释本发明，以便于理解，下面结合附图，通过具体实施方式，对本发明作详细描述。In order to better explain the present invention and facilitate understanding, the present invention will be described in detail below with reference to the accompanying drawings and through specific embodiments.
如图3所示，电熔镁炉熔炼过程主要包括加料，加热熔化，排气等工况。通过电弧产生的热量熔化原材料，得到最终的产品。通过设定不同的电流设定值，电流控制系统通过调整电极与熔池的距离来跟踪变化的设定值，调整电弧的大小，以满足不同工况所需的能量。为了获得更高的经济效益，有必要在电熔镁的生产中保持良好的操作性能及生产安全。As shown in Figure 3, the smelting process of the fused magnesium furnace mainly includes working conditions such as feeding, heating and melting, and exhausting. The heat generated by the arc melts the raw materials to obtain the final product. By setting different current setting values, the current control system tracks the changing setting value by adjusting the distance between the electrode and the molten pool, and adjusts the size of the arc to meet the energy required for different working conditions. In order to obtain higher economic benefits, it is necessary to maintain good operating performance and production safety in the production of fused magnesium.
以下实施例中的方法所涉及的装置包括：电熔镁炉异常工况识别及安全控制系统(下述简称控制系统)、上位机、PLC、现场传感变送结构。其中现场传感变送结构包括电流检测仪表、图像检测仪表、声音检测仪表等。在电熔镁炉过程现场安装各种检测仪表，检测仪表将采集的信号通过Profibus-DP总线送到PLC，PLC通过以太网定时将采集信号传送到上位机，上位机把接受的数据传给电熔镁炉异常工况识别及安全控制系统，该系统识别异常工况，并制定安全控制决策用于排除异常工况。The devices involved in the methods in the following embodiments include: an abnormal working condition identification and safety control system of the fused magnesia furnace (hereinafter referred to as a control system), a host computer, a PLC, and an on-site sensor transmission structure. The on-site sensor transmission structure includes current detection instruments, image detection instruments, and sound detection instruments. Various detection instruments are installed on the site of the fused magnesia furnace process. The detection instruments send the collected signals to the PLC through the Profibus-DP bus. The PLC periodically transmits the collected signals to the upper computer through the Ethernet, and the upper computer transmits the received data to the electricity. Magnesium melting furnace abnormal working condition identification and safety control system, the system identifies abnormal working conditions, and makes safety control decisions to eliminate abnormal working conditions.
上述装置的各部分功能简要说明如下：The functions of each part of the above device are briefly described as follows:
①现场传感变送部分：包括电流检测、图像检测、声音检测等检测仪表，由传感器组成，负责过程数据的采集与传送；①On-site sensor transmission part: including current detection, image detection, sound detection and other detection instruments, composed of sensors, responsible for the collection and transmission of process data;
②PLC：负责把采集的信号A/D转换，并通过以太网把信号传送给上位机。②PLC: Responsible for A/D conversion of the collected signal, and transmit the signal to the host computer through Ethernet.
本实施例中，PLC的控制器可采用Simens 400系列的CPU 414-2，该控制器具有Profibus DP口连接分布式IO；为PLC配备以太网通讯模块，用于上位机访问PLC数据；PLC的控制器和以太网通讯模块放置在中央控制室中的PLC柜中。In this embodiment, the controller of the PLC can use the CPU 414-2 of the Simens 400 series, the controller has a Profibus DP port to connect distributed IO; the PLC is equipped with an Ethernet communication module for the host computer to access the PLC data; The controller and Ethernet communication module are placed in the PLC cabinet in the central control room.
③上位机：收集本地PLC数据，传送给电熔镁炉异常工况识别及安全控制系统。③Host computer: collect local PLC data and transmit it to the abnormal working condition identification and safety control system of the fused magnesia furnace.
上位机本实施例中可选用i7联想计算机，采用WINDOW XP操作系统。The host computer in this embodiment can be selected as an i7 Lenovo computer, which adopts the WINDOW XP operating system.
④电熔镁炉异常工况识别及安全控制系统对异常工况进行识别，并制定安全控制决策用于排除异常工况。④ The abnormal working condition identification and safety control system of the fused magnesia furnace shall identify the abnormal working conditions, and make safety control decisions to eliminate the abnormal working conditions.
电熔镁炉异常工况识别及安全控制系统在i7联想计算机上，采用C#2008编程软件，电熔镁炉异常工况识别及安全控制算法采用Matlab2010a编程软件；The abnormal working condition identification and safety control system of the fused magnesia furnace is based on the i7 Lenovo computer, using C#2008 programming software, and the abnormal working condition identification and safety control algorithm of the fused magnesia furnace adopts the Matlab2010a programming software;
PLC与异常控制系统的信号传送软件是采用C#2008编程软件。The signal transmission software of PLC and abnormal control system adopts C#2008 programming software.
在电熔镁炉过程现场安装检测仪表，检测仪表将采集的信号通过Profibus-DP传送到PLC中，PLC定时将采集信号通过以太网传送给上位机，上位机把接收的数据传给电熔镁炉异常工况识别及安全控制系统，该系统识别异常工况，并制定安全控制决策用于排除异常。The detection instrument is installed on the site of the fused magnesia furnace process. The detection instrument transmits the collected signals to the PLC through Profibus-DP. The PLC periodically transmits the collected signals to the upper computer through the Ethernet, and the upper computer transmits the received data to the fused magnesium. A furnace abnormal working condition identification and safety control system, which identifies abnormal working conditions and formulates safety control decisions to eliminate abnormalities.
结合图1A和图1B所示，本实施例的电熔镁炉异常工况识别及控制方法包括：1A and 1B, the method for identifying and controlling abnormal working conditions of an fused magnesia furnace in this embodiment includes:
S1、获取电熔镁炉工况中的预设周期内的在线数据。S1. Obtain online data within a preset period in the working condition of the fused magnesia furnace.
在实际应用中，可对获取的在线数据进行过滤处理，以获得过滤处理后的在线数据。In practical applications, the acquired online data may be filtered to obtain filtered online data.
例如，使用滤波技术去除在线数据的噪声，获取去除噪声的在线数据。For example, use filtering techniques to remove noise from online data to obtain denoised online data.
S2、采用相似度匹配策略查看案例库中是否存在与在线数据匹配的案例信息。S2. Use a similarity matching strategy to check whether there is case information matching the online data in the case database.
举例来说，若步骤S1中的在线数据为过滤处理后的在线数据，则可采用过滤处理后的在线数据与案例库中的案例信息进行匹配。For example, if the online data in step S1 is the filtered online data, the filtered online data can be used to match the case information in the case database.
另外，步骤S2可具体包括：In addition, step S2 may specifically include:
获取在线数据与案例库中每一案例的变量之间的相似度(例如，通过相似度准则，确定相似度)，判断该相似度是否大于预设阈值，若是，则确定在线数据与大于阈值的最大相似度对应的案例匹配；若相似度均不大于预设阈值，可认为案例库中案例与在线数据均不匹配。Obtain the similarity between the online data and the variables of each case in the case database (for example, determine the similarity through the similarity criterion), and judge whether the similarity is greater than the preset threshold, and if so, determine whether the online data is greater than the threshold. The case corresponding to the maximum similarity is matched; if the similarity is not greater than the preset threshold, it can be considered that the case in the case database does not match the online data.
当前，欧式距离是计算相似度的一种方式，本实施例中计算在线数据变量与离线数据变量间的欧式距离即相似度。获取在线数据与案例库中每一案例的相似度。Currently, the Euclidean distance is a way to calculate the similarity. In this embodiment, the Euclidean distance between the online data variable and the offline data variable is calculated, that is, the similarity. Obtain the similarity between online data and each case in the case library.
本实施例中的预设阈值可为预先通过专家经验或业内技术人员知识确定的阈值。The preset threshold in this embodiment may be a threshold determined in advance through expert experience or knowledge of persons skilled in the art.
特别地，该步骤中的案例库为预先根据电熔镁炉工况的历史数据建立的各种异常工况的案例信息。In particular, the case library in this step is case information of various abnormal working conditions established in advance according to the historical data of the working conditions of the fused magnesia furnace.
S3、若存在，依据匹配的案例信息给出当前在线数据的辨识结果，将辨识结果作为当前电熔镁炉工况的异常识别结果。S3. If it exists, the identification result of the current online data is given according to the matched case information, and the identification result is used as the abnormal identification result of the current working condition of the fused magnesia furnace.
可选地，在具体实现过程中，该方法还可包括下述的步骤S4：Optionally, in a specific implementation process, the method may further include the following step S4:
S4、根据异常识别结果，确定设备的剩余生命时间；依据预先建立的剩余生命时间和调整量之间的关系，计算调整量，获得安全控制决策信息，以使控制系统根据安全控制策略信息适应调整；即，将安全控制决策信息应用于控制系统，若异常工况被排除，则结束。否则，重复上述步骤，获取新的异常工况调节措施。S4. Determine the remaining life time of the equipment according to the abnormal identification result; calculate the adjustment amount according to the relationship between the pre-established remaining life time and the adjustment amount, and obtain the safety control decision information, so that the control system can adapt and adjust according to the safety control strategy information ; That is, apply the safety control decision information to the control system, and end if the abnormal condition is eliminated. Otherwise, repeat the above steps to obtain new adjustment measures for abnormal working conditions.
其中，上述的剩余生命时间为电熔镁炉工况设备从获取在线数据时的状态到完全无法正常工作的时间；Among them, the remaining life time mentioned above is the time from the state of the fused magnesia furnace working condition equipment from the state when the online data is obtained to the complete failure to work normally;
所述调整量为电流设定值；剩余生命时间和调整量之间的关系为离线方式确定的关系。The adjustment amount is the current set value; the relationship between the remaining life time and the adjustment amount is determined in an offline manner.
其中，剩余生命时间为电熔镁炉工况设备能够持续正常运转的时间，即设备的剩余生命时间是指从现在的状态开始到设备完全无法正常工作的时间。Among them, the remaining life time is the time that the equipment can continue to operate normally under the working condition of the fused magnesia furnace, that is, the remaining life time of the equipment refers to the time from the current state to the time when the equipment is completely unable to work normally.
本实施例中，在线数据是表征量，是随时间改变的；针对电熔镁炉，调整量是电流设定值，它是一个离线量，定下来就是固定不变了。本实施例中，异常的发生是因为设定值不合适，所以要改设定值。In this embodiment, the online data is the characteristic quantity, which changes with time; for the fused magnesia furnace, the adjustment quantity is the current setting value, which is an offline quantity, and it is fixed when it is set. In this embodiment, the abnormality occurs because the set value is inappropriate, so the set value needs to be changed.
上述的剩余生命时间和调整量之间的关系是离线就建立好的，异常程度和剩余生命时间之间的计算规则也是离线建立好的；根据‘异常识别结果’及‘异常程度和剩余时间的关系’，计算剩余生命时间；根据‘剩余生命时间’及‘剩余生命时间和调整量之间的关系’，计算调整量。The above relationship between remaining life time and adjustment amount is established offline, and the calculation rule between abnormality degree and remaining life time is also established offline; relationship', calculate the remaining life time; calculate the adjustment amount according to the "remaining life time" and the "relationship between the remaining life time and the adjustment amount".
在实际应用中，在控制系统根据安全控制策略信息适应调整之后，重复获取电熔镁炉工况中下一预设周期内的在线数据，根据该在线数据的判断过程，以确定步骤S1中的在线数据的异常工况是否排除，若是，则结束，否则，重复获取安全控制决策信息的过程。In practical application, after the control system adapts and adjusts according to the safety control strategy information, it repeatedly acquires the online data in the next preset period in the working condition of the fused magnesia furnace, and determines the online data in step S1 according to the judgment process of the online data. Whether the abnormal working condition of the online data is excluded, if so, end, otherwise, repeat the process of obtaining the safety control decision information.
在本实施例中方法通过采集当前工况的在线数据，进而在案例库中查看是否有相似案例，若存在相似案例，则将该相似案例作为辨识结果，进而根据辨识结果，获取安全控制决策信息，以便控制系统根据所述安全控制决策信息适应调整，即将获得的决策信息应用于出现异常工况的控制系统(即电熔镁炉异常工况识别及安全控制系统)，若异常工况排除，则结束，否则，重新识别异常工况，获取决策方案。In this embodiment, the method collects the online data of the current working condition, and then checks whether there are similar cases in the case database. If there is a similar case, the similar case is used as the identification result, and then the safety control decision information is obtained according to the identification result. , so that the control system can adapt and adjust according to the safety control decision-making information, and the obtained decision-making information will be applied to the control system with abnormal working conditions (ie, the abnormal working condition identification and safety control system of the fused magnesia furnace). If the abnormal working condition is eliminated, Then end, otherwise, re-identify abnormal working conditions and obtain a decision-making plan.
由此可对电熔镁炉的生产工况进行现场指导，提高能效，降低能耗，避免经济损失。In this way, it is possible to provide on-site guidance for the production conditions of the fused magnesia furnace, improve energy efficiency, reduce energy consumption, and avoid economic losses.
在一种可选的实现方式中，如图1A所示，若上述步骤S2中例库中不存在匹配的案例信息，此时，可执行下述的步骤S3a：In an optional implementation manner, as shown in FIG. 1A , if there is no matching case information in the example library in the above step S2, at this time, the following step S3a can be performed:
S3a、若案例库中不存在匹配的案例信息，则采用贝叶斯网络推理模型对所述在线数据进行分析，获得辨识结果，将辨识结果作为当前电熔镁炉工况的异常识别结果；S3a. If there is no matching case information in the case database, use a Bayesian network inference model to analyze the online data, obtain an identification result, and use the identification result as the abnormal identification result of the current working condition of the fused magnesia furnace;
进而可执行上述步骤S4等步骤。Further steps such as the above-mentioned step S4 can be performed.
也就是说，在步骤S2中，如果匹配的案例的相似度小于给定阈值，则将在线数据划分为不同的程度等级，并将划分程度等级后的在线数据作为证据输入到贝叶斯网络单元，进行贝叶斯网络推理，推理得到的结果中，拥有最大后验概率的异常程度为异常工况的识别结果。That is to say, in step S2, if the similarity of the matched cases is less than a given threshold, the online data is divided into different degree levels, and the online data after dividing the degree levels is input into the Bayesian network unit as evidence , and conduct Bayesian network inference. Among the results obtained by inference, the abnormality degree with the largest posterior probability is the recognition result of abnormal working conditions.
进一步地，贝叶斯网络给出的辨识结果经过证实后作为一个新的案例存入案例库中，实现案例库补充。Further, the identification result given by the Bayesian network is verified as a new case and stored in the case database to realize the supplement of the case database.
也就是说，在具体实现过程中，上述方法还可包括下述的图1A中未示出的步骤S5：That is to say, in a specific implementation process, the above method may further include the following step S5 not shown in FIG. 1A :
S5、若通过贝叶斯网络推理模型获得的异常识别结果确定为当前电熔镁炉工况实际发生的结果，则将该异常识别结果作为案例信息，存储在所述案例库中。S5. If the abnormal identification result obtained by the Bayesian network inference model is determined to be the actual occurrence result of the current working condition of the fused magnesia furnace, the abnormal identification result is used as case information and stored in the case database.
本实施例能有效根据电熔镁炉生产过程出现的异常工况，通过案例库中案例推理及贝叶斯网络推理模型进行异常工况识别，通过剩余生命预测的概念，根据异常识别结果，制定安全控制决策。This embodiment can effectively identify the abnormal working conditions according to the abnormal working conditions in the production process of the fused magnesia furnace, through the case reasoning in the case database and the Bayesian network reasoning model, and formulate the formula according to the abnormal identification results through the concept of remaining life prediction. Security control decisions.
在具体应用中，在前述的步骤S2之前，上述方法还可包括下述的步骤S2a，如图1B所示。In a specific application, before the aforementioned step S2, the above method may further include the following step S2a, as shown in FIG. 1B .
S2a、根据预设时间段内电熔镁炉工况的历史数据，建立案例库；S2a, establishing a case library according to the historical data of the working conditions of the fused magnesia furnace within a preset time period;
以及根据预设时间段内电熔镁炉工况的历史数据和先验知识，建立贝叶斯网络推理模型。And according to the historical data and prior knowledge of the working conditions of the fused magnesia furnace in a preset time period, a Bayesian network inference model is established.
举例来说建立案例库的步骤可包括：For example, the steps of building a case library may include:
S2a1、离线收集预设时间段内电熔镁炉工况发生异常的信息；S2a1. Collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
S2a2、根据收集的信息，确定异常工况的特征与异常工况中相关变量之间的关系，获得历史数据，根据历史数据建立案例库和贝叶斯网络推理模型。S2a2. According to the collected information, determine the relationship between the characteristics of the abnormal working condition and the relevant variables in the abnormal working condition, obtain historical data, and establish a case database and a Bayesian network inference model according to the historical data.
或者，在另一可选的实现方式中，建立案例库的步骤可包括：Or, in another optional implementation manner, the step of establishing the case library may include:
S2a1’、离线收集预设时间段内电熔镁炉工况发生异常的信息；S2a1', collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
S2a2’、根据收集的信息，确定异常工况的特征与异常工况中相关变量之间的关系，获得历史数据，对历史数据进行过滤处理，根据过滤处理后的历史数据建立案例库和贝叶斯网络推理模型。S2a2', according to the collected information, determine the relationship between the characteristics of the abnormal working condition and the relevant variables in the abnormal working condition, obtain historical data, filter the historical data, and establish a case library and Bayeux based on the filtered historical data s network inference model.
在该子步骤中，为避免噪声对历史数据的影响，使用滤波技术去除历史数据中的噪声。In this sub-step, in order to avoid the influence of noise on the historical data, a filtering technique is used to remove the noise in the historical data.
通过历史中电熔镁炉异常工况的分析，确定异常工况的特征及相关变量间的关系。基于找到的关系，表达异常识别规则，例如；“如果_前提，那么_结论”。相关变量作为规则的前提，异常工况的程度作为案例的结论，最终，案例库被建立起来。Through the analysis of abnormal working conditions of fused magnesia furnace in history, the characteristics of abnormal working conditions and the relationship between related variables are determined. Based on the relationships found, anomaly recognition rules are expressed, eg; "if _ premise, then _ conclusion". The relevant variables are used as the premise of the rules, the degree of abnormal conditions is used as the conclusion of the case, and finally, the case base is established.
特别说明的是，本发明实施例中采用案例库中案例推理技术进行在线数据分析。其中，案例推理是一种人工智能方法，主要是利用历史上出现的相似案例解决新的问题。目前，案例推理过程可包括四个基本过程：检索、重用、修正和保存。在检索相似案例时需要先确定相似度计算方法，最常用的相似衡量方式为计算欧式距离。It is particularly noted that, in the embodiment of the present invention, the case reasoning technology in the case database is used for online data analysis. Among them, case reasoning is an artificial intelligence method that mainly uses similar cases in history to solve new problems. Currently, the case reasoning process can include four basic processes: retrieval, reuse, revision, and preservation. When retrieving similar cases, the similarity calculation method needs to be determined first. The most commonly used similarity measurement method is to calculate the Euclidean distance.
另外，前述的建立贝叶斯网络推理模型的步骤可包括：In addition, the aforementioned steps of establishing a Bayesian network inference model may include:
第一步、离线收集预设时间段内电熔镁炉工况发生异常的信息；The first step is to collect offline information about abnormal working conditions of the fused magnesia furnace within a preset time period;
第二步、根据先验知识以及异常工况中相关变量之间的关系，确定贝叶斯网络的节点和结构；The second step is to determine the nodes and structure of the Bayesian network according to the prior knowledge and the relationship between relevant variables in abnormal conditions;
第三步、根据收集的历史数据及确定的结构，获得贝叶斯网络的参数，从而建立起贝叶斯网络推理模型。The third step is to obtain the parameters of the Bayesian network according to the collected historical data and the determined structure, thereby establishing a Bayesian network inference model.
特别地，在构建贝叶斯网络参数时，需要对历史数据进行处理，除了滤波以外，还要将所有异常信息中相关变量划分等级，划分等级的数量和阈值通过专家知识或试错法决定。根据划分了等级的数据计算参数。In particular, when constructing Bayesian network parameters, historical data needs to be processed. In addition to filtering, all relevant variables in abnormal information must be classified into grades. The number of grades and thresholds are determined by expert knowledge or trial and error. The parameters are calculated from the graded data.
本实施例中的贝叶斯网络是一种不确定知识的表达方法。贝叶斯网络的结构可以表达变量间的相关关系，参数表示节点变量间的依赖程度。当收集到相关变量作为证据输入给建立好的贝叶斯网络推理模型时，可通过推理机制得到关注变量的后验概率。The Bayesian network in this embodiment is an expression method of uncertain knowledge. The structure of Bayesian network can express the correlation between variables, and the parameters represent the degree of dependence between node variables. When the relevant variables are collected as evidence and input to the established Bayesian network inference model, the posterior probability of the variable of interest can be obtained through the inference mechanism.
上述利用历史数据建立案例库，当新问题出现时，利用相似度匹配，在案例库中寻找相似案例，当匹配到的案例的相似程度大于给定阈值时，利用检索到的案例给出异常识别结果；当匹配到的案例的相似程度小于给定阈值时，利用贝叶斯网络进行异常工况识别，给出辨识结果。The above uses historical data to build a case database. When a new problem occurs, similarity matching is used to find similar cases in the case database. When the similarity of the matched cases is greater than a given threshold, the retrieved cases are used to identify abnormality. Results: When the similarity of the matched cases is less than the given threshold, the Bayesian network is used to identify abnormal conditions, and the identification results are given.
对于安全控制决策部分，引入剩余生命时间的概念，建立系统剩余生命时间与调整量间的关系。在获得辨识结果之后，首先计算电熔镁炉异常工况的剩余生命时间，进而利用剩余生命时间与调整量间的关系，获得调整量，以实现电熔镁炉的安全控制。For the safety control decision-making part, the concept of remaining life time is introduced, and the relationship between the remaining life time of the system and the adjustment amount is established. After the identification results are obtained, the remaining life time of the fused magnesia furnace under abnormal conditions is calculated first, and then the relationship between the remaining life time and the adjustment amount is used to obtain the adjustment amount to realize the safety control of the fused magnesia furnace.
本发明实施例利用案例推理和贝叶斯网络进行异常工况识别，利用设备的剩余生命时间的概念，提供安全控制方案。The embodiment of the present invention uses case reasoning and Bayesian network to identify abnormal working conditions, and uses the concept of remaining life time of equipment to provide a security control scheme.
为较好的理解上述方法的内容，以下结合实验结果及工艺/工况过程进行具体说明。In order to better understand the content of the above method, a specific description is given below in combination with the experimental results and the process/working condition process.
电熔镁炉的工艺如图3所示，主要设备包括变压器、电流回路、电极升降装置、三相电极及熔炉。电熔镁炉熔炼过程主要包括加料，加热熔化，排气等工况。通过电弧产生的热量熔化原材料，得到最终的产品。通过设定不同的电流设定值，电流控制系统调整电极与熔池的距离来跟踪变化的设定值，调整电弧的大小，以满足不同工况所需的能量。每隔10-15分钟，向炉内填料一次，上面的工况循环进行，直到炉子被填满为止。The process of the fused magnesia furnace is shown in Figure 3. The main equipment includes a transformer, a current loop, an electrode lifting device, a three-phase electrode and a furnace. The smelting process of fused magnesia furnace mainly includes feeding, heating and melting, exhaust and other working conditions. The heat generated by the arc melts the raw materials to obtain the final product. By setting different current setting values, the current control system adjusts the distance between the electrode and the molten pool to track the changing setting value and adjust the size of the arc to meet the energy required for different working conditions. Fill the furnace every 10-15 minutes, and cycle through the above conditions until the furnace is filled.
步骤01：建立案例库及贝叶斯网络推理模型Step 01: Build a case base and a Bayesian network inference model
在电熔镁熔炼的过程中，会产生大量的二氧化碳气体，为避免炉内压强过大造成熔浆飞溅，需要调整电流的设定值，电流控制系统跟踪电流设定值，调整电极的位置，使电极上下移动，使电极和原料之间产生缝隙，二氧化碳气体顺利排出。但是当原材料的粒度变化时，电流控制系统若跟踪原有的设定值，则电极和原料之间的缝隙无法使二氧化碳气体顺利排出，炉内压强过大，熔浆将随气体一起飞溅，产生排气异常工况。In the process of fused magnesium smelting, a large amount of carbon dioxide gas will be produced. In order to avoid the splash of molten slurry caused by excessive pressure in the furnace, the set value of the current needs to be adjusted. The current control system tracks the set value of the current and adjusts the position of the electrode. The electrode is moved up and down to create a gap between the electrode and the raw material, and the carbon dioxide gas is discharged smoothly. However, when the particle size of the raw material changes, if the current control system tracks the original set value, the gap between the electrode and the raw material cannot allow the carbon dioxide gas to be discharged smoothly, and the pressure in the furnace is too large, and the molten slurry will splash together with the gas, resulting in Exhaust abnormal conditions.
在进行排气异常工况识别时，操作人员会关注三个方面的信息：电流信息、图像信息和声音信息。在排气异常工况发生时，电流变化率和电流跟踪误差会发生变化；电极和熔池之间的电弧是主要的声源信号，当炉内的状态发生变化时，电弧声音的幅值和频率会发生变化；当发生飞溅时，操作人员将观察到高温熔浆喷出炉外，因此图像信息可以作为辅助变量来辨识异常工况。When identifying abnormal exhaust conditions, the operator will pay attention to three aspects of information: current information, image information and sound information. When abnormal exhaust conditions occur, the current change rate and current tracking error will change; the arc between the electrode and the molten pool is the main sound source signal. When the state in the furnace changes, the amplitude and frequency of the arc sound changes; when splashing occurs, the operator will observe the high-temperature molten slurry spraying out of the furnace, so the image information can be used as an auxiliary variable to identify abnormal conditions.
将排气异常工况划分为三种程度：轻度、中度和重度。基于专家经验，在异常进化的不同阶段会有不同的信息起主要作用。在轻度排气异常时，声音信息起主要作用；在中度异常时，声音信息和电流信息起主要作用；在重度异常时，电流信息和图像信息起主要作用。The abnormal exhaust conditions are divided into three degrees: mild, moderate and severe. Based on expert experience, different information plays a major role at different stages of anomalous evolution. In the case of mild exhaust abnormality, the sound information plays the main role; in the moderate abnormality, the sound information and the current information play the main role; in the severe abnormality, the current information and the image information play the main role.
将异常的声音信号划分为两种程度：轻度和重度。通过魏格纳维尔分布分析，飞溅的特征频率能够被提取，为200Hz。当排气异常工况为轻度和中度时，声音信号的幅值会提高；当排气异常工况为重度时，由于能量的释放，声音信号的幅值会降低。因此针对声音信号，选择下面的特征变量：飞溅特征频率的短时能量和飞溅特征频率的幅值。Divide abnormal sound signals into two degrees: mild and severe. Through the analysis of the Wegnerville distribution, the characteristic frequency of the splash can be extracted, which is 200 Hz. When the abnormal exhaust conditions are mild and moderate, the amplitude of the sound signal will increase; when the abnormal exhaust conditions are severe, the amplitude of the sound signal will decrease due to the release of energy. Therefore, for the sound signal, the following characteristic variables are selected: the short-term energy of the splash characteristic frequency and the amplitude of the splash characteristic frequency.
将异常的电流信号划分为三种程度：轻度、中度和重度。选择电流跟踪误差和电流变化率为其主要特征变量。The abnormal current signal is divided into three degrees: mild, moderate and severe. The current tracking error and current rate of change are chosen as their main characteristic variables.
将异常的图像信号划分为两种程度：轻度和重度。当排气异常工况严重时，炉口范围内的图像亮度会提高，在图像处理中，亮度的变化用灰度来体现。在图像的三个主要颜色分量中，红色分量占主要作用。因此针对图像信号，选择下面的特征变量：平均灰度的变化、平均灰度短时能量的变化、灰度方差的变化、灰度的丰度和红色分量的变化，其中灰度的丰度是指超过正常图像平均灰度的比例。排气异常工况中所有的相关特征被总结在表1中。The abnormal image signal is divided into two degrees: mild and severe. When the abnormal working condition of exhaust gas is serious, the brightness of the image in the furnace mouth will increase. In the image processing, the change of brightness is reflected by grayscale. Of the three main color components of an image, the red component dominates. Therefore, for the image signal, the following characteristic variables are selected: the change of the average gray level, the change of the short-term energy of the average gray level, the change of the gray level variance, the change of the gray level abundance and the change of the red component, where the gray level abundance is Refers to the ratio that exceeds the average gray level of a normal image. All relevant characteristics of the exhaust abnormal conditions are summarized in Table 1.
表1.排气异常特征及相关关系Table 1. Exhaust anomaly characteristics and correlation
为避免噪声对历史数据的影响，使用滤波技术去除历史数据中的噪声。通过排气异常工况的分析及确定的异常工况的特征，建立案例库，形式见表2。特征A-I作为规则的前提，排气异常工况的程度作为案例的结果。In order to avoid the influence of noise on historical data, filtering techniques are used to remove noise in historical data. Through the analysis of the abnormal working conditions of exhaust gas and the characteristics of the determined abnormal working conditions, a case library is established, and the form is shown in Table 2. The characteristics A-I are used as the premise of the rule, and the degree of abnormal exhaust conditions is used as the result of the case.
表2.案例库的结构Table 2. Structure of the case library
为建立贝叶斯网络推理模型，需要将所有相关变量划分等级，划分等级的数量和阈值通过专家知识或试错法决定。通过专家知识和排气异常工况的分析，能够确定排气异常的贝叶斯网络的节点和结构。利用数据获取与预处理单元，可以获得贝叶斯网络的条件概率表，即贝叶斯网络的参数，最终，贝叶斯网络被建立起来，见图5。In order to build a Bayesian network inference model, all relevant variables need to be graded, and the number of grades and thresholds are determined by expert knowledge or trial and error. Through expert knowledge and analysis of abnormal exhaust conditions, the nodes and structures of the Bayesian network of exhaust anomalies can be determined. Using the data acquisition and preprocessing unit, the conditional probability table of the Bayesian network can be obtained, that is, the parameters of the Bayesian network. Finally, the Bayesian network is established, as shown in Figure 5.
步骤2：在线异常工况辨识Step 2: Online abnormal condition identification
具体过程如下：The specific process is as follows:
1)在线数据被预处理后，在案例库中进行相似度匹配。通过相似度准则，计算相似度，找到最相似的案例。如果匹配的案例的相似度大于给定阈值，则使用该案例进行异常工况识别。通过专家经验确定阈值的大小。1) After the online data is preprocessed, similarity matching is performed in the case base. Through the similarity criterion, calculate the similarity and find the most similar case. If the similarity of a matched case is greater than a given threshold, the case is used for abnormal condition identification. The size of the threshold is determined by expert experience.
2)如果匹配的案例的相似度均小于给定阈值，则将在线数据划分为不同的程度等级，划分程度等级的在线数据作为证据输入到贝叶斯网络推理模型，进行贝叶斯网络推理，推理得到的结果中，拥有最大后验概率的异常程度为异常工况的识别结果。2) If the similarity of the matched cases is less than the given threshold, the online data will be divided into different degree levels, and the online data of the divided degree levels will be input into the Bayesian network inference model as evidence, and the Bayesian network inference will be performed. Among the results obtained by inference, the abnormal degree with the largest posterior probability is the recognition result of abnormal conditions.
3)在获取了新问题的异常识别结果之后，新的问题的有用性需要被检测来判断它是否能够作为新的案例存储在案例库中。类似地，由贝叶斯网络推理模型获得的辨识结果也需要通过验证才能存储在案例库中作为新的案例。案例推理的具体过程见图4。3) After obtaining the anomaly identification results of the new problem, the usefulness of the new problem needs to be checked to determine whether it can be stored in the case database as a new case. Similarly, the identification results obtained by the Bayesian network inference model also need to be verified before being stored in the case database as a new case. The specific process of case reasoning is shown in Figure 4.
下面主要说明匹配的案例的相似度小于给定阈值的情况。通过分析实际的情况，排气工况可能发生的事件被总结在表3中。The following mainly describes the case where the similarity of matched cases is less than a given threshold. By analyzing the actual situation, the events that may occur in the exhaust operating conditions are summarized in Table 3.
表3.排气工况可能发生的事件Table 3. Possible Events for Exhaust Conditions
表3中的每个事件均包含9个变量，变量A-H被划分为三个程度，分别用数字1-3表示，含义分别为正常、轻微异常和严重异常。变量I被划分为四种程度，分别用数字1-4表示，含义分别为正常、轻微异常、中度异常和严重异常。以事件10为例，特征A-B的状态是正常，特征C-I 的状态是严重异常。其它事件的含义可以用类似的方式获得。Each event in Table 3 contains 9 variables. Variables A-H are divided into three degrees, which are represented by numbers 1-3, meaning normal, slightly abnormal, and severely abnormal. Variable I is divided into four degrees, which are represented by numbers 1-4, meaning normal, slightly abnormal, moderately abnormal, and severely abnormal. Taking event 10 as an example, the status of features A-B is normal, and the status of features C-I is severely abnormal. The meaning of other events can be obtained in a similar manner.
表3中的事件将作为证据，通过贝叶斯网络的推理得到异常识别结果，如表4所示。The events in Table 3 will be used as evidence to obtain anomaly identification results through the inference of Bayesian network, as shown in Table 4.
表4.针对表3中的事件的识别结果Table 4. Identification results for the events in Table 3
表4中，辨识结果1-4代表排气工况的4个状态：正常、轻度异常、中度异常和重度异常，以事件10为例，其辨识结果为严重异常。其它事件的辨识结果的含义可以用类似的方式获得。表3中的事件是按异常的程度排序的，针对表4的辨识结果可知，其辨识结果是符合实际异常工况辨识经验的。In Table 4, the identification results 1-4 represent four states of the exhaust working condition: normal, mildly abnormal, moderately abnormal, and severely abnormal. Taking event 10 as an example, the identification result is a serious abnormality. The meaning of the recognition results of other events can be obtained in a similar manner. The events in Table 3 are sorted according to the degree of abnormality. According to the identification results in Table 4, the identification results are in line with the actual abnormal working condition identification experience.
为更好的说明本发明实施例的方法，将提出的方案与传统的仅使用电流信息进行异常工况识别的方法进行对比，结果如图6所示。在图6中可以看出，当排气异常工况的程度为轻度和中度时，提出的方案能够获得辨识结果，而传统的方法仅在异常程度为严重时才能获得辨识结果。In order to better illustrate the method of the embodiment of the present invention, the proposed scheme is compared with the traditional method of identifying abnormal working conditions using only current information, and the result is shown in FIG. 6 . It can be seen in Figure 6 that the proposed scheme can obtain identification results when the degree of abnormal exhaust conditions is mild and moderate, while the traditional method can obtain identification results only when the degree of abnormality is severe.
步骤3：制定安全控制决策信息。Step 3: Develop security control decision information.
针对电熔镁炉排气异常，其异常控制方案是在原来排气工况电流设定值yj(t)(j＝1,2,3)的基础上，给出补偿值Δyj(t)，让电流控制系统跟踪新的电流设定值yj′(t)＝yj(t)+Δyj(t)(j＝1,2,3)，其中j代表三相电极，使异常工况逐渐恢复。电流补偿值的设计与异常的程度和系统的剩余生命时间有关，当系统的剩余生命时间越短时，异常的严重程度越高，电流的补偿值越大。所以，通过异常工况的识别结果，计算系统的剩余生命时间。通过预先建立的剩余生命时间和调整量的关系，计算调整量的大小。当系统的异常辨识结果来源于案例推理时，使用下面的公式计算剩余生命时间，异常程度的进化过程见图2。For the abnormal exhaust of the fused magnesia furnace, the abnormal control scheme is to give the compensation value Δyj (t ), let the current control system track the new current setting value yj ′(t)=yj (t)+Δyj (t) (j=1,2,3), where j represents the three-phase electrode, making the abnormal Conditions gradually recovered. The design of the current compensation value is related to the degree of abnormality and the remaining life time of the system. When the remaining life time of the system is shorter, the severity of the abnormality is higher, and the compensation value of the current is larger. Therefore, through the identification results of abnormal working conditions, the remaining life time of the system is calculated. According to the relationship between the remaining life time and the adjustment amount established in advance, the size of the adjustment amount is calculated. When the abnormal identification result of the system comes from case reasoning, the following formula is used to calculate the remaining life time, and the evolution process of the abnormal degree is shown in Figure 2.
式中，R(t)代表剩余生命时间。τi代表第i个异常状态的持续时间。ki(t)是状态持续时间的系数，用以表达时刻t在第i个异常状态上的持续时间。I(t)代表在t时刻的异常状态，Ii代表状态i的下限。In the formula, R(t) represents the remaining life time. τi represents the duration of the i-th abnormal state. ki (t) is the coefficient of state duration, which is used to express the duration of time t on the i-th abnormal state. I(t) represents the abnormal state at time t, and Ii represents the lower limit of state i.
当系统的异常辨识结果来源于贝叶斯网络推理模型时，使用下面的公式(3)-(5)计算剩余生命时间When the abnormal identification result of the system comes from the Bayesian network inference model, the following formulas (3)-(5) are used to calculate the remaining life time
(3)式中，R(t)表示所有状态剩余生命时间的权重和。P(X＝i)代表第i个状态的后验概率。每个状态的剩余时间ri(t)用式(4)计算。对于不同状态的系数ki(t)用式(5)计算。In formula (3), R(t) represents the weight sum of the remaining life time of all states. P(X=i) represents the posterior probability of the ith state. The remaining timeri (t) for each state is calculated by equation (4). The coefficients ki (t) for different states are calculated by equation (5).
在获得了剩余生命时间之后，需要建立剩余生命时间与调整量之间的关系。将排气异常程度为轻度、中度和重度时的调整量粗略的设为和其中j代表三相电极。选择下面的点来拟合关系Δy＝f(R(t))，(τ1+τ2+τ3,0)和其中为工艺允许的最大调整量。通过直线连接这些点，构成分段函数的形式。After the remaining life time is obtained, the relationship between the remaining life time and the adjustment amount needs to be established. Roughly set the adjustment amount when the degree of exhaust abnormality is mild, moderate, and severe to and where j represents the three-phase electrode. Choose the points below to fit the relationship Δy=f(R(t)), (τ1 +τ2 +τ3 ,0) and in It is the maximum adjustment allowed by the process. Connect these points by straight lines, forming the form of a piecewise function.
这样，通过异常识别结果，计算剩余生命时间。通过剩余生命时间和调整量之间的关系，计算出调整量的大小，即给出安全控制决策。In this way, the remaining life time is calculated from the abnormal identification result. Through the relationship between the remaining life time and the adjustment amount, the size of the adjustment amount is calculated, that is, the safety control decision is given.
图7和图8分别给出了正常的排气工况和异常的排气工况。由图8 可以看出，由于没有调整好电流的设定值，二氧化碳气体没有顺利从炉中排出，造成电流的波动越来越大。安全控制方案是给出电流设定值的补偿值，使电流控制系统跟踪新的电流设定值以排除异常工况。Figures 7 and 8 show normal exhaust conditions and abnormal exhaust conditions, respectively. It can be seen from Fig. 8 that the carbon dioxide gas was not discharged from the furnace smoothly due to the failure to adjust the set value of the current, causing the current to fluctuate more and more. The safety control scheme is to give the compensation value of the current set value, so that the current control system can track the new current set value to eliminate abnormal conditions.
步骤4：实施决策排除异常工况Step 4: Implement decisions to eliminate abnormal conditions
针对图8产生的异常工况，将本发明制定的安全控制决策应用于控制系统，效果如图9所示。从图9中可以看出，在第7个采样点，电流波动超出正常范围，并且这种现象继续持续。安全控制决策实施之后，在第13个采样点，电流波动下降，大约在第19个采样点，异常工况被排除。为了体现本发明的优越性，本发明与传统的仅使用电流信息进行安全决策的方法进行比较，传统方法的控制效果如图10所示。从图10可以看出，在第7个采样点电流波动超出正常范围，并且这种现象继续持续，直到第20个采样点，异常程度变为严重时，传统方法才给出安全决策，之后系统逐渐恢复正常。In view of the abnormal working condition generated in FIG. 8 , the safety control decision made by the present invention is applied to the control system, and the effect is shown in FIG. 9 . As can be seen from Figure 9, at the 7th sampling point, the current fluctuation is outside the normal range, and this phenomenon continues. After the safety control decision was implemented, at the 13th sampling point, the current fluctuation decreased, and around the 19th sampling point, abnormal conditions were excluded. In order to demonstrate the superiority of the present invention, the present invention is compared with the traditional method that only uses current information to make safety decisions. The control effect of the traditional method is shown in FIG. 10 . It can be seen from Figure 10 that the current fluctuation at the 7th sampling point exceeds the normal range, and this phenomenon continues until the 20th sampling point, when the degree of abnormality becomes serious, the traditional method does not give a safety decision, and then the system gradually returned to normal.
通过上面的实例，表明了本发明实施例的方法能够识别异常工况，并且根据识别的异常工况制定有效的安全措施，措施实施后能够有效的使异常工况恢复为正常，并且和传统仅使用电流信息进行异常工况识别和安全控制的方法相比，本发明提出的方法更加有效且具有更好的性能。进一步地，上述方法对于提高矿产资源的综合利用率，降低能耗，减少环境污染，促进安全生产，都有重大的意义。Through the above examples, it is shown that the method of the embodiment of the present invention can identify abnormal working conditions, and formulate effective safety measures according to the identified abnormal working conditions. After the measures are implemented, the abnormal working conditions can be effectively restored to normal. Compared with the method of using current information for abnormal working condition identification and safety control, the method proposed by the present invention is more effective and has better performance. Further, the above method is of great significance for improving the comprehensive utilization rate of mineral resources, reducing energy consumption, reducing environmental pollution, and promoting safe production.
以上结合具体实施例描述了本发明的技术原理，这些描述只是为了解释本发明的原理，不能以任何方式解释为对本发明保护范围的限制。基于此处解释，本领域的技术人员不需要付出创造性的劳动即可联想到本发明的其它具体实施方式，这些方式都将落入本发明的保护范围之内。The technical principles of the present invention have been described above with reference to specific embodiments. These descriptions are only for explaining the principles of the present invention, and cannot be interpreted as limiting the protection scope of the present invention in any way. Based on the explanations herein, those skilled in the art can think of other specific embodiments of the present invention without creative efforts, and these methods will all fall within the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710324622.3ACN106959662B (en) | 2017-05-10 | 2017-05-10 | A kind of identification of electric melting magnesium furnace unusual service condition and control method |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710324622.3ACN106959662B (en) | 2017-05-10 | 2017-05-10 | A kind of identification of electric melting magnesium furnace unusual service condition and control method |
| Publication Number | Publication Date |
|---|---|
| CN106959662A CN106959662A (en) | 2017-07-18 |
| CN106959662Btrue CN106959662B (en) | 2019-03-08 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710324622.3AExpired - Fee RelatedCN106959662B (en) | 2017-05-10 | 2017-05-10 | A kind of identification of electric melting magnesium furnace unusual service condition and control method |
| Country | Link |
|---|---|
| CN (1) | CN106959662B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108960256A (en)* | 2018-06-28 | 2018-12-07 | 东软集团股份有限公司 | A kind of determination method, device and equipment of components damage degree |
| CN111385106B (en)* | 2018-12-11 | 2022-03-01 | 华为技术有限公司 | A method, device and device for identifying the root cause of failure |
| CN110059359A (en)* | 2019-03-21 | 2019-07-26 | 江苏东方国信工业互联网有限公司 | System and method for controlling furnace body process based on big data analysis |
| CN111598137A (en)* | 2020-04-24 | 2020-08-28 | 北京金山云网络技术有限公司 | Method, apparatus and electronic device for providing reasoning service |
| CN113064390B (en)* | 2021-03-17 | 2022-03-01 | 国网辽宁省电力有限公司辽阳供电公司 | Case reasoning-based active warning method for pollutant emission of cement production enterprise |
| CN113237332B (en)* | 2021-04-14 | 2022-02-25 | 东北大学 | A method for identifying working conditions of fused magnesia furnace |
| CN113705601A (en)* | 2021-05-21 | 2021-11-26 | 国网辽宁省电力有限公司鞍山供电公司 | Method for identifying running state of electric smelting magnesium furnace based on random forest |
| CN115354144B (en)* | 2022-08-29 | 2024-01-30 | 湖南华菱涟源钢铁有限公司 | Roller quenching process operation parameter setting method based on case matching |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102637019B (en)* | 2011-02-10 | 2014-04-02 | 武汉科技大学 | Intelligent integrated fault diagnosis method and device in industrial production process |
| CN102765643B (en)* | 2012-05-31 | 2015-06-17 | 天津大学 | Elevator fault diagnosis and early-warning method based on data drive |
| CN104298225B (en)* | 2014-09-25 | 2017-07-04 | 中国石油化工股份有限公司 | Chemical process unusual service condition causality inference pattern is modeled and graphical representation method |
| Publication number | Publication date |
|---|---|
| CN106959662A (en) | 2017-07-18 |
| Publication | Publication Date | Title |
|---|---|---|
| CN106959662B (en) | A kind of identification of electric melting magnesium furnace unusual service condition and control method | |
| WO2024108871A1 (en) | Shield construction early warning system and early warning method based on edge computing architecture | |
| CN111949700A (en) | Intelligent safety guarantee real-time optimization method and system for petrochemical device | |
| CN106372799B (en) | A Grid Security Risk Prediction Method | |
| Li et al. | Abnormal condition identification and safe control scheme for the electro-fused magnesia smelting process | |
| CN105159235A (en) | Comprehensive coordination control method and system for rotary kiln during calcining process | |
| CN114632624B (en) | Electric dust removal operation optimization system and optimization method | |
| CN108571320B (en) | A multi-stage control method of shearer for intelligent fully mechanized mining face | |
| CN118278620A (en) | Infrastructure power consumption monitoring and management system | |
| CN115358626A (en) | Safety inspection early warning system for mineral processing production equipment | |
| CN116823175B (en) | Intelligent operation and maintenance method and system for petrochemical production informatization system | |
| CN116991130A (en) | An intelligent automated control system and method for petrochemical production | |
| CN119916766B (en) | PCL-based automatic control method for carbon black master batch production equipment | |
| CN119416935A (en) | Sewage water quality prediction method and system based on deep learning | |
| CN115078892B (en) | State remote monitoring system for single-machine large-transmission frequency converter | |
| CN113608560B (en) | A control system for the preparation process of alumina lye | |
| CN118152970B (en) | A device status trend perception method based on edge computing algorithm | |
| CN106950946B (en) | An abnormal control method for hydrometallurgy based on optimization principle | |
| CN102660656B (en) | Speed control system and speed control method for granulating and dewatering rotary drum of blast furnace | |
| Li et al. | Abnormal condition identification for the electro-fused magnesia smelting process | |
| CN117720157A (en) | Monitoring method and system for industrial brine degassing process | |
| CN116700197B (en) | Industrial control monitoring, analyzing and early warning system and analyzing and early warning processing method | |
| US20240369983A1 (en) | Alarm flood management system and method | |
| CN119225256B (en) | An intelligent safety control system for one-button pulverizing of coal for ironmaking | |
| CN104503243A (en) | Hammerstein-Wiener model based predictive control method of ladle refining furnace electrode regulation system |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | Granted publication date:20190308 | |
| CF01 | Termination of patent right due to non-payment of annual fee |