技术领域technical field
本发明属于高压断路器检测方法技术领域，具体涉及一种基于卷积神经网络算法的高压断路器故障检测方法。The invention belongs to the technical field of high-voltage circuit breaker detection methods, and in particular relates to a high-voltage circuit breaker fault detection method based on a convolutional neural network algorithm.
背景技术Background technique
高压断路器是电力系统最主要的控制与保护装置，关系到输电、配电及用电的可靠性、安全性。高压断路器能在系统故障与非故障情况下实现多种操作。断路器也是能关合、承载、开断运行回路正常电流，也能在规定时间内关合、承载及开断规定的过载电流。High-voltage circuit breakers are the most important control and protection devices in power systems, which are related to the reliability and safety of power transmission, power distribution and power consumption. High voltage circuit breakers are capable of multiple operations under system fault and non-fault conditions. The circuit breaker can also close, carry and break the normal current of the operating circuit, and can also close, carry and break the specified overload current within the specified time.
高压断路器一般都以电磁铁为操作的第一控制元件，在操动机构中大部分是直流电磁铁。当线圈中通过电流时，在磁铁内产生磁通，动铁芯受磁力影响，使断路器分闸或合闸。合分闸线圈中的电流可作为高压断路器机械故障诊断所用的丰富信息。High-voltage circuit breakers generally use electromagnets as the first control element to operate, and most of the operating mechanisms are DC electromagnets. When the current passes through the coil, magnetic flux is generated in the magnet, and the moving iron core is affected by the magnetic force to make the circuit breaker open or close. The current in the closing and opening coils can be used as a wealth of information for the diagnosis of mechanical faults in high voltage circuit breakers.
现有的高压断路器故障检修的方法有很多，其中涉及各种人工智能算法，如：模糊控制能用精确的数学工具将模糊的概念或自然语言清晰化，但其隶属函数和模糊规则的确定过程存在一定的人为因素；径向基神经网络为断路器的故障诊断问题提供了一种比较好的结构体系，但存在着无法解释自己的推理过程和推理依据以及数据不充分时神经网络无法正常工作的缺点。There are many existing methods for troubleshooting high-voltage circuit breakers, which involve various artificial intelligence algorithms, such as: fuzzy control can use precise mathematical tools to clarify fuzzy concepts or natural language, but the determination of its membership functions and fuzzy rules There are certain human factors in the process; the radial basis neural network provides a better structural system for the fault diagnosis problem of circuit breakers, but there are problems that cannot explain the reasoning process and reasoning basis of itself and the neural network cannot work normally when the data is insufficient. The downside of the job.
神经网络算法具有良好的容错能力、并行处理能力和自学习能力，可处理环境信息复杂，背景知识不清楚，推理规则不明确情况下的问题，而且运行速度快、自适应性能好、具有较高的分辨率，尤其是权值共享网络使之更类似于生物神经网络，降低了网络模型的复杂度，减少了权值的数量。采用神经网络算法，避免了传统识别算法中复杂的特征提取和数据重建过程。The neural network algorithm has good fault tolerance, parallel processing ability and self-learning ability, and can deal with complex environmental information, unclear background knowledge, and unclear inference rules. It also has fast running speed, good adaptive performance, and high The resolution, especially the weight sharing network makes it more similar to the biological neural network, which reduces the complexity of the network model and reduces the number of weights. The neural network algorithm is used to avoid the complex feature extraction and data reconstruction process in the traditional recognition algorithm.
发明内容Contents of the invention
本发明的目的在于提供一种基于卷积神经网络算法的高压断路器故障检测方法，采用卷积神经网络分析故障特征信号，在弥补人工神经网络检测的不足的同时，能更加准确有效地判断断路器的故障类型，进而能够有效率的检修。The purpose of the present invention is to provide a high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm. The convolutional neural network is used to analyze the fault characteristic signal. While making up for the shortcomings of the artificial neural network detection, it can judge the circuit breaker more accurately and effectively. The fault type of the device can be checked and repaired efficiently.
本发明所采用的技术方案是，基于卷积神经网络算法的高压断路器故障检测方法，具体按照以下步骤实施：The technical solution adopted in the present invention is a high-voltage circuit breaker fault detection method based on a convolutional neural network algorithm, which is specifically implemented according to the following steps:
步骤1、先将磁平衡式霍尔电流传感器分别与断路器分合闸线圈、数据处理系统连接，构建出分合闸线圈电流在线监测系统；然后利用分合闸线圈电流在线监测系统实时监测得到的分合闸线圈电流数据，并将实时监测得到的分合闸线圈电流数据作为输入变量；Step 1. First connect the magnetic balance Hall current sensor with the opening and closing coil of the circuit breaker and the data processing system to build an online monitoring system for the opening and closing coil current; then use the real-time monitoring of the opening and closing coil current online monitoring system to obtain The current data of the opening and closing coils, and the current data of the opening and closing coils obtained by real-time monitoring are used as input variables;
步骤2、利用基于卷积神经网络的深度学习算法构建故障类型预测模型，将经步骤1得到的一部分分合闸线圈电流数据输入到构建故障类型预测模型中进行训练；Step 2. Construct a fault type prediction model using a convolutional neural network-based deep learning algorithm, and input a part of the opening and closing coil current data obtained in step 1 into the fault type prediction model for training;
步骤3、将经步骤1得到的一部分分合闸线圈电流数据输入到经步骤2训练好的故障类型预测模型中，由故障类型预测模型对输入的分合闸线圈电流数据进行处理，完成对高压断路器故障检测。Step 3. Input part of the opening and closing coil current data obtained in step 1 into the fault type prediction model trained in step 2, and the fault type prediction model processes the input opening and closing coil current data to complete the high-voltage Circuit breaker fault detection.
本发明的特点还在于：The present invention is also characterized in that:
步骤1中构建出的分合闸线圈电流在线监测系统，其结构为，包括有单片机，单片机分别与电源模块、信息处理单元、4G通信模块、Zigbee通信模块、数据存储单元连接；电源模块分别与太阳能发电模块、蓄电池连接；信息处理单元的输入端与磁平衡式霍尔电流传感器连接。The on-line monitoring system for opening and closing coil current constructed in step 1 has a structure including a single-chip microcomputer, and the single-chip microcomputer is respectively connected with a power module, an information processing unit, a 4G communication module, a Zigbee communication module, and a data storage unit; the power module is respectively connected with a The solar power generation module and the storage battery are connected; the input end of the information processing unit is connected with the magnetic balance Hall current sensor.
单片机的型号为STM32F407。The model of the microcontroller is STM32F407.
在步骤2中，建立故障类型预测模型具体按照以下方法实施：In step 2, the establishment of a fault type prediction model is implemented in the following ways:
建立卷积神经网络，首先要确定卷积过程，具体按照以下方法实施：To establish a convolutional neural network, the convolution process must first be determined, and the specific implementation is as follows:
首先确定滤波器的大小，采用卷积核Kernelij来对上一层滤波器的一个特征进行加权得到xi*Kernelij；之后对xi*Kernelij进行求和后加偏移，具体按照以下算法实施：First determine the size of the filter, and use the convolution kernel Kernelij to weight a feature of the upper filter to obtain xi *Kernelij ; then add the offset to xi *Kernelij after summing, specifically as follows Algorithm implementation:
在式(1)中：xi为上一层的滤波器中的一个特征，Mj为神经元j对应的滤波器，为第l层的神经元i的第j个对应的权值，Bl为第1层的唯一偏移；In formula (1): xi is a feature in the filter of the previous layer, Mj is the filter corresponding to neuron j, is the jth corresponding weight of neuron i in layer l, and Bl is the unique offset of layer 1;
抽样层采用下采样的方法确定，具体方法如下：The sampling layer is determined by the method of downsampling, the specific method is as follows:
采用mean-pooling，首先将滤波器中的所有值求均值；然后将采样出的信息乘以可训练参数，再加上可训练偏置，将得到的结果通过激活函数计算，即能得到神经元的输出；其中，激活函数采用sigmoid函数；Using mean-pooling, first average all the values in the filter; then multiply the sampled information by the trainable parameters, plus the trainable bias, and calculate the result through the activation function to get the neuron The output; where the activation function uses the sigmoid function;
第一层的输出算法具体如下：The output algorithm of the first layer is as follows:
在式(2)中：β为第1层的可训练参数，Bl为可训练偏置，Mj为神经元j对应的滤波器。In formula (2): β is the trainable parameter of the first layer, Bl is the trainable bias, and Mj is the filter corresponding to neuron j.
在步骤2中，对故障类型预测模型进行训练，具体按照以下步骤实施：In step 2, the fault type prediction model is trained, specifically implemented according to the following steps:
步骤A、初始化权值，即将所有权值初始化为一个较小的随机数，具体是将所有权值初始化为一个随机数[0，1]；Step A, initialize the weight value, that is, initialize the ownership value to a small random number, specifically, initialize the ownership value to a random number [0, 1];
步骤B、经步骤A后，从训练集(以实施例1为例将其中的五组数据作为训练集)中提取一个样例X，并将该样例X输入到卷积神经网络中，并给出它的目标输出向量，并将其记作D；Step B, after step A, extract a sample X from the training set (taking embodiment 1 as an example, five groups of data wherein are used as the training set), and input the sample X into the convolutional neural network, and Give its target output vector, and denote it as D;
步骤C、经步骤B后，从前层向后层依次计算，得到卷积神经网络的输出值Y，对于各个层的计算方法具体如下：Step C, after step B, calculate sequentially from the front layer to the back layer to obtain the output value Y of the convolutional neural network. The calculation method for each layer is as follows:
对于卷积层，采用如下算法进行计算：For the convolutional layer, the following algorithm is used for calculation:
在式(1)中：xi为上一层的滤波器中的一个特征，Mj为神经元j对应的滤波器，为第1层的神经元i的第j个对应的权值，Bl为第1层的唯一偏移；In formula (1): xi is a feature in the filter of the previous layer, Mj is the filter corresponding to neuron j, is the jth corresponding weight of neuron i in the first layer, and Bl is the unique offset of the first layer;
对于卷积层的输出值，最终要添加sigmoid函数进行非线性变换；For the output value of the convolutional layer, the sigmoid function is finally added for nonlinear transformation;
对于抽样层，具体采用如下算法进行计算：For the sampling layer, the following algorithm is used for calculation:
在式(2)中：β为第1层的可训练参数，Bl为可训练偏置，Mj为神经元j对应的滤波器，xi为上一层的滤波器中的一个特征；In formula (2): β is the trainable parameter of the first layer, Bl is the trainable bias, Mj is the filter corresponding to neuron j, and xi is a feature in the filter of the previous layer;
对于全链接层，直接采用多层人工神经网络的方法进行计算，具体算法如下：For the full connection layer, the method of multi-layer artificial neural network is directly used for calculation, and the specific algorithm is as follows:
在式(3)中，Bl为可训练偏置，xi为上一层的滤波器中的一个特征,wji为第l层的结点j到第l+1层的节点i的权值，f(x)为sigmoid函数；In formula (3), Bl is the trainable bias, xi is a feature in the filter of the previous layer, and wji is the weight from node j of layer l to node i of layer l+1 value, f(x) is the sigmoid function;
步骤D、待步骤C完成后，反向(即从后层向前层)依次计算各层的误差项，具体按照以下步骤实施：Step D, after step C is completed, calculate the error terms of each layer in reverse order (that is, from the back layer to the front layer), and specifically implement according to the following steps:
步骤a、计算输出层的误差，具体按照以下方法实施：Step a, calculating the error of the output layer, specifically implemented according to the following method:
设定输出层共有M个结点，则对输出层的结点k的误差项为：Assuming that there are M nodes in the output layer, the error term for node k in the output layer is:
δk＝(dk-yk)yk(1-yk) (4)；δk = (dk -yk )yk (1-yk ) (4);
在式(4)中：dk为结点k的目标输出，yk为结点k的预测输出；In formula (4): dk is the target output of node k, and yk is the predicted output of node k;
步骤b、经步骤a后，计算中间全链接层的误差，具体方法如下：Step b, after step a, calculate the error of the middle fully connected layer, the specific method is as follows:
设定当前层为第1层，共有L个结点，第1+1层共M个节点；Set the current layer as layer 1, with a total of L nodes, and a total of M nodes in layer 1+1;
则对于第1层的节点j的误差项具体如下：Then the error term for node j in the first layer is as follows:
在式(5)中：hj为结点j的输出，wjk为第l层的结点j到第l+1层的节点k的权值，M为滤波器大小，δk为节点j的误差项；In formula (5): hj is the output of node j, wjk is the weight of node j in layer l to node k in layer l+1, M is the filter size, and δk is node j the error term;
步骤c、经步骤b后，对卷积层的误差项进行计算，具体算法与步骤b中涉及的算法相同，具体如下：Step c, after step b, calculate the error term of the convolutional layer, the specific algorithm is the same as the algorithm involved in step b, specifically as follows:
在式(5)中：hj为结点j的输出，wjk为第l层的结点j到第l+1层的节点k的权值，M为滤波器大小，δk为节点j的误差项；In formula (5): hj is the output of node j, wjk is the weight of node j in layer l to node k in layer l+1, M is the filter size, and δk is node j the error term;
步骤d、待步骤a～步骤c后，再从后层向前层逐层依次计算出各权值的调整量，即第n轮迭代的节点j的第k个所输入的权向量的改变量，涉及的具体算法如下：Step d, after step a to step c, calculate the adjustment amount of each weight value layer by layer from the back layer to the front layer, that is, the change amount of the kth input weight vector of node j in the nth round of iteration , the specific algorithm involved is as follows:
在式(6)中，N为输入变量个数，n为迭代层数，δk为节点j的误差项，hj为结点j的输出；In formula (6), N is the number of input variables, n is the number of iteration layers, δk is the error term of node j, and hj is the output of node j;
阀值改变量ΔBk(n)具体按照以下算法经计算获得：The threshold value change amount ΔBk (n) is specifically calculated according to the following algorithm:
在式(7)中：a为迭代层数，N为输入变量个数，n为迭代层数，δk为节点j的误差项；In formula (7): a is the number of iteration layers, N is the number of input variables, n is the number of iteration layers, and δk is the error term of node j;
步骤e、经步骤d后，调整各权值，以获得更新后的权值wjk(n+1)，具体按照以下算法实施：Step e, after step d, adjust each weight value to obtain the updated weight value wjk (n+1), specifically implement according to the following algorithm:
wjk(n+1)＝wjk(n)+Δwjk(n) (8)；wjk (n+1)=wjk (n)+Δwjk (n) (8);
在式(8)中：wjk(n)为第n层的结点j到第l+1层的节点k的权值，Δwjk(n)为第n轮迭代的节点所输入的权向量的改变量；In formula (8): wjk (n) is the weight value from node j of the nth layer to node k of the l+1 layer, Δwjk (n) is the weight vector input by the node of the nth iteration the amount of change;
更新后的阀值Bk(n+1)具体按照以下算法经计算获得：The updated threshold Bk (n+1) is specifically calculated according to the following algorithm:
Bk(n+1)＝Bk(n)+ΔBk(n) (9)；Bk (n+1) = Bk (n) + ΔBk (n) (9);
在式(9)中：Bk(n+1)为更新后的阀值，ΔBk(n)为阀值的改变量；In formula (9): Bk (n+1) is the updated threshold value, and ΔBk (n) is the change amount of the threshold value;
步骤f，重复步骤b～步骤e，直到误差函数小于设定的阀值；Step f, repeating steps b to e until the error function is smaller than the set threshold;
其中，误差函数E具体表示为如下形式：Among them, the error function E is specifically expressed as the following form:
在式(10)中，dk为dk为结点k的目标输出，yk为yk为结点k的预测输出，M为滤波器大小，k为节点数。In formula (10), dk is the target output of nodek , yk is the predicted output of nodek , M is the filter size, and k is the number of nodes.
本发明的有益效果在于：The beneficial effects of the present invention are:
(1)本发明基于卷积神经网络算法的高压断路器故障检测方法，通过磁平衡式霍尔电流传感器准确感知合分闸线圈中的电流波形，并通过STM32F407、电源模块、信息处理单元、4G通信模块、Zigbee通信模块、数据存储单元等实现电流的A/D转化、信号处理和数据通信等功能。(1) The high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention accurately senses the current waveform in the closing and opening coils through the magnetic balance Hall current sensor, and through STM32F407, power module, information processing unit, 4G The communication module, Zigbee communication module, data storage unit, etc. realize the functions of current A/D conversion, signal processing and data communication.
(2)本发明基于卷积神经网络算法的高压断路器故障检测方法，首先确定输入/输出设计，通过对10组数据作为卷积神经网络的输入向量，进行归一化处理将故障类型进行量化编码；其次构造卷积神经网络故障类型预测模型。(2) The high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention first determines the input/output design, and quantifies the fault type by performing normalization processing on 10 sets of data as the input vector of the convolutional neural network Encoding; secondly, construct a convolutional neural network fault type prediction model.
(3)本发明基于卷积神经网络算法的高压断路器故障检测方法，用故障预测模型分析故障类型，将实时监测到的分合闸线圈电流数据输入到模型中即可得到故障类型。(3) The high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention uses a fault prediction model to analyze the fault type, and inputs the real-time monitored opening and closing coil current data into the model to obtain the fault type.
综上所述，与现有的方法相比：本发明基于卷积神经网络算法的高压断路器故障检测方法，在数据表示方式上能够消除输入数据中与学习任务无关因素的改变对学习性能的影响，同时保留学习任务中有用的信息；将其应用在高压断路器故障诊断上，能更加准确的判断故障类型和进行状态维修。In summary, compared with the existing methods: the present invention based on the convolutional neural network algorithm high-voltage circuit breaker fault detection method can eliminate the impact on the learning performance of changes in the input data that have nothing to do with the learning task in terms of data representation. influence while retaining the useful information in the learning task; applying it to the fault diagnosis of high-voltage circuit breakers can more accurately judge the fault type and perform condition-based maintenance.
附图说明Description of drawings
图1是本发明基于卷积神经网络算法的高压断路器故障检测方法中采用的分合闸线圈电流在线监测系统的结构示意图；Fig. 1 is the structure schematic diagram of the opening and closing coil current online monitoring system adopted in the high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention;
图2是本发明基于卷积神经网络算法的高压断路器故障检测方法中涉及的卷积神经网络的结构图；Fig. 2 is the structural diagram of the convolutional neural network involved in the high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention;
图3是本发明基于卷积神经网络算法的高压断路器故障检测方法的流程图；Fig. 3 is the flow chart of the high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention;
图4是实施例1中涉及的合/分闸线圈电流的特性曲线。FIG. 4 is a characteristic curve of closing/opening coil current involved in Embodiment 1. FIG.
图中，1.单片机，2.电源模块，3.信息处理单元，4.磁平衡式霍尔电流传感器，5.4G通信模块，6.Zibbee通信模块，7.太阳能发电模块，8.蓄电池，9.数据存储单元，10.断路器分合闸线圈。In the figure, 1. SCM, 2. Power module, 3. Information processing unit, 4. Magnetic balance Hall current sensor, 5.4G communication module, 6. Zibbee communication module, 7. Solar power generation module, 8. Battery, 9 . Data storage unit, 10. Circuit breaker opening and closing coil.
具体实施方式detailed description
下面结合附图和具体实施方式对本发明进行详细说明。The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.
本发明基于卷积神经网络算法的高压断路器故障检测方法，如图3所示，具体按照以下步骤实施：The high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention, as shown in Figure 3, is specifically implemented according to the following steps:
步骤1、先将磁平衡式霍尔电流传感器4分别与断路器分合闸线圈10、数据处理系统连接，构建出分合闸线圈电流在线监测系统；然后利用分合闸线圈电流在线监测系统实时监测得到的分合闸线圈电流数据，并将实时监测得到的分合闸线圈电流数据作为输入变量；Step 1. First connect the magnetic balance Hall current sensor 4 with the circuit breaker opening and closing coil 10 and the data processing system to construct an online monitoring system for the opening and closing coil current; then use the online monitoring system for the opening and closing coil current in real time The current data of the opening and closing coil obtained by monitoring, and the current data of the opening and closing coil obtained by real-time monitoring are used as input variables;
其中，构建出的分合闸线圈电流在线监测系统，其结构如图1所示，包括有单片机1，且单片机1的型号为STM32F407，单片机1分别与电源模块2、信息处理单元3、4G通信模块5、Zigbee通信模块6、数据存储单元9连接，电源模块2分别与太阳能发电模块7、蓄电池8连接，信息处理单元3的输入端与磁平衡式霍尔电流传感器4连接，磁平衡式霍尔电流传感器4与断路器分合闸线圈10连接；Among them, the constructed on-line monitoring system for opening and closing coil current has a structure as shown in Figure 1, including a single-chip microcomputer 1, and the model of the single-chip microcomputer 1 is STM32F407, and the single-chip microcomputer 1 communicates with the power module 2, information processing unit 3, and 4G respectively Module 5, Zigbee communication module 6, data storage unit 9 are connected, power supply module 2 is connected with solar power generation module 7, storage battery 8 respectively, the input terminal of information processing unit 3 is connected with magnetic balance type Hall current sensor 4, magnetic balance type Hall current sensor The current sensor 4 is connected with the circuit breaker opening and closing coil 10;
通过电源模块2和太阳能发电模块7为整个分合闸线圈电流在线监测系统提供电能，蓄电池8用来存储多余的电量，以备不时之需，单片机1通过4G通信模块5、Zigbee通信模块6对外进行通信，单片机1与信息处理单元3连接，信息处理单元3与磁平衡式霍尔电流传感器4连接，磁平衡式霍尔电流传感器4与断路器分合闸线圈10连接，相互配合能够对获取的电流数据，并对获取的数据进行处理，并将数据信息保存于数据存储单元9内。The power supply module 2 and the solar power generation module 7 provide electric energy for the entire opening and closing coil current on-line monitoring system, and the storage battery 8 is used to store excess power for emergency use. For communication, the single-chip microcomputer 1 is connected with the information processing unit 3, the information processing unit 3 is connected with the magnetic balance type Hall current sensor 4, and the magnetic balance type Hall current sensor 4 is connected with the circuit breaker opening and closing coil 10, and mutual cooperation can obtain current data, and process the acquired data, and store the data information in the data storage unit 9.
步骤2、利用基于卷积神经网络的深度学习算法构建故障类型预测模型，将经步骤1得到的一部分分合闸线圈电流数据输入到构建故障类型预测模型中进行训练；Step 2. Construct a fault type prediction model using a convolutional neural network-based deep learning algorithm, and input a part of the opening and closing coil current data obtained in step 1 into the fault type prediction model for training;
建立故障类型预测模型，具体按照以下方法实施：Establish a fault type prediction model, and implement it according to the following methods:
建立卷积神经网络，首先要确定卷积过程，如图2所示，具体按照以下方法实施：To establish a convolutional neural network, the convolution process must first be determined, as shown in Figure 2, and the specific implementation is as follows:
卷积层(即C层)和抽样层(即S层)；Convolutional layer (ie C layer) and sampling layer (ie S layer);
首先确定滤波器的大小，采用卷积核Kernelij来对上一层滤波器的一个特征进行加权得到xi*Kernelij；之后对xi*Kernelij进行求和后加偏移，具体按照以下算法实施：First determine the size of the filter, and use the convolution kernel Kernelij to weight a feature of the upper filter to obtain xi *Kernelij ; then add the offset to xi *Kernelij after summing, specifically as follows Algorithm implementation:
在式(1)中：xi为上一层的滤波器中的一个特征，Mj为神经元j对应的滤波器，为第l层的神经元i的第j个对应的权值，Bl为第1层的唯一偏移；In formula (1): xi is a feature in the filter of the previous layer, Mj is the filter corresponding to neuron j, is the jth corresponding weight of neuron i in layer l, and Bl is the unique offset of layer 1;
抽样层采用下采样的方法确定，具体方法如下：The sampling layer is determined by the method of downsampling, the specific method is as follows:
采用mean-pooling(均值池化)，首先将滤波器中的所有值求均值；然后将采样出的信息乘以可训练参数，再加上可训练偏置，将得到的结果通过激活函数计算，即能得到神经元的输出；其中，激活函数采用sigmoid函数；Using mean-pooling (mean pooling), first average all the values in the filter; then multiply the sampled information by the trainable parameters, plus the trainable bias, and calculate the result through the activation function, That is, the output of the neuron can be obtained; where the activation function adopts the sigmoid function;
第一层的输出算法具体如下：The output algorithm of the first layer is as follows:
在式(2)中：β为第1层的可训练参数，Bl为可训练偏置，Mj为神经元j对应的滤波器；In formula (2): β is the trainable parameter of the first layer, Bl is the trainable bias, and Mj is the filter corresponding to neuron j;
对故障类型预测模型进行训练，具体按照以下步骤实施：To train the fault type prediction model, follow the steps below:
步骤A、初始化权值，即将所有权值初始化为一个较小的随机数，具体是将所有权值初始化为一个随机数[0，1]；Step A, initialize the weight value, that is, initialize the ownership value to a small random number, specifically, initialize the ownership value to a random number [0, 1];
步骤B、经步骤A后，从训练集(以实施例1为例将其中的五组数据作为训练集)中提取一个样例X，并将该样例X输入到卷积神经网络中，并给出它的目标输出向量，并将其记作D；Step B, after step A, extract a sample X from the training set (taking embodiment 1 as an example, five groups of data wherein are used as the training set), and input the sample X into the convolutional neural network, and Give its target output vector, and denote it as D;
步骤C、经步骤B后，从前层向后层依次计算，得到卷积神经网络的输出值Y，对于各个层的计算方法具体如下：Step C, after step B, calculate sequentially from the front layer to the back layer to obtain the output value Y of the convolutional neural network. The calculation method for each layer is as follows:
对于卷积层，采用如下算法进行计算：For the convolutional layer, the following algorithm is used for calculation:
在式(1)中：xi为上一层的滤波器中的一个特征，Mj为神经元j对应的滤波器，为第1层的神经元i的第j个对应的权值，Bl为第1层的唯一偏移；In formula (1): xi is a feature in the filter of the previous layer, Mj is the filter corresponding to neuron j, is the jth corresponding weight of neuron i in the first layer, and Bl is the unique offset of the first layer;
对于卷积层的输出值，最终要添加sigmoid函数进行非线性变换；For the output value of the convolutional layer, the sigmoid function is finally added for nonlinear transformation;
对于抽样层，具体采用如下算法进行计算：For the sampling layer, the following algorithm is used for calculation:
在式(2)中：β为第1层的可训练参数，Bl为可训练偏置，Mj为神经元j对应的滤波器，xi为上一层的滤波器中的一个特征；In formula (2): β is the trainable parameter of the first layer, Bl is the trainable bias, Mj is the filter corresponding to neuron j, and xi is a feature in the filter of the previous layer;
对于全链接层，直接采用多层人工神经网络的方法进行计算，具体算法如下：For the full connection layer, the method of multi-layer artificial neural network is directly used for calculation, and the specific algorithm is as follows:
在式(3)中，Bl为可训练偏置，xi为上一层的滤波器中的一个特征,wji为第l层的结点j到第l+1层的节点i的权值，f(x)为sigmoid函数；In formula (3), Bl is the trainable bias, xi is a feature in the filter of the previous layer, and wji is the weight from node j of layer l to node i of layer l+1 value, f(x) is the sigmoid function;
步骤D、待步骤C完成后，反向(即从后层向前层)依次计算各层的误差项，具体按照以下步骤实施：Step D, after step C is completed, calculate the error terms of each layer in reverse order (that is, from the back layer to the front layer), and specifically implement according to the following steps:
步骤a、计算输出层的误差，具体按照以下方法实施：Step a, calculating the error of the output layer, specifically implemented according to the following method:
设定输出层共有M个结点，则对输出层的结点k的误差项为：Assuming that there are M nodes in the output layer, the error term for node k in the output layer is:
δk＝(dk-yk)yk(1-yk) (4)；δk = (dk -yk )yk (1-yk ) (4);
在式(4)中：dk为结点k的目标输出，yk为结点k的预测输出；In formula (4): dk is the target output of node k, and yk is the predicted output of node k;
步骤b、经步骤a后，计算中间全链接层的误差，具体方法如下：Step b, after step a, calculate the error of the middle fully connected layer, the specific method is as follows:
设定当前层为第1层，共有L个结点，第1+1层共M个节点；Set the current layer as layer 1, with a total of L nodes, and a total of M nodes in layer 1+1;
则对于第1层的节点j的误差项具体如下：Then the error term for node j in the first layer is as follows:
在式(5)中：hj为结点j的输出，wjk为第l层的结点j到第l+1层的节点k的权值，M为滤波器大小，δk为节点j的误差项；In formula (5): hj is the output of node j, wjk is the weight of node j in layer l to node k in layer l+1, M is the filter size, and δk is node j the error term;
步骤c、经步骤b后，对卷积层的误差项进行计算，具体算法与步骤b中涉及的算法相同，具体如下：Step c, after step b, calculate the error term of the convolutional layer, the specific algorithm is the same as the algorithm involved in step b, specifically as follows:
在式(5)中：hj为结点j的输出，wjk为第l层的结点j到第l+1层的节点k的权值，M为滤波器大小，δk为节点j的误差项；In formula (5): hj is the output of node j, wjk is the weight of node j in layer l to node k in layer l+1, M is the filter size, and δk is node j the error term;
步骤d、待步骤a～步骤c后，再从后层向前层逐层依次计算出各权值的调整量，即第n轮迭代的节点j的第k个所输入的权向量的改变量，涉及的具体算法如下：Step d, after step a to step c, calculate the adjustment amount of each weight value layer by layer from the back layer to the front layer, that is, the change amount of the kth input weight vector of node j in the nth round of iteration , the specific algorithm involved is as follows:
在式(6)中，N为输入变量个数，n为迭代层数，δk为节点j的误差项，hj为结点j的输出；In formula (6), N is the number of input variables, n is the number of iteration layers, δk is the error term of node j, and hj is the output of node j;
阀值改变量ΔBk(n)具体按照以下算法经计算获得：Threshold value change ΔBk (n) is specifically calculated according to the following algorithm:
在式(7)中：a为迭代层数，N为输入变量个数，n为迭代层数，δk为节点j的误差项；In formula (7): a is the number of iteration layers, N is the number of input variables, n is the number of iteration layers, and δk is the error term of node j;
步骤e、经步骤d后，调整各权值，以获得更新后的权值wjk(n+1)，具体按照以下算法实施：Step e, after step d, adjust each weight value to obtain the updated weight value wjk (n+1), specifically implement according to the following algorithm:
wjk(n+1)＝wjk(n)+Δwjk(n) (8)；wjk (n+1)=wjk (n)+Δwjk (n) (8);
在式(8)中：wjk(n)为第n层的结点j到第l+1层的节点k的权值，Δwjk(n)为第n轮迭代的节点所输入的权向量的改变量；In formula (8): wjk (n) is the weight value from node j of the nth layer to node k of the l+1 layer, Δwjk (n) is the weight vector input by the node of the nth iteration the amount of change;
更新后的阀值Bk(n+1)具体按照以下算法经计算获得：The updated threshold Bk (n+1) is specifically calculated according to the following algorithm:
Bk(n+1)＝Bk(n)+ΔBk(n) (9)；Bk (n+1) = Bk (n) + ΔBk (n) (9);
在式(9)中：Bk(n+1)为更新后的阀值，ΔBk(n)为阀值的改变量；In formula (9): Bk (n+1) is the updated threshold value, and ΔBk (n) is the change amount of the threshold value;
步骤f，重复步骤b～步骤e，直到误差函数小于设定的阀值；Step f, repeating steps b to e until the error function is smaller than the set threshold;
其中，误差函数E具体表示为如下形式：Among them, the error function E is specifically expressed as the following form:
在式(10)中，dk为dk为结点k的目标输出，yk为yk为结点k的预测输出，M为滤波器大小，k为节点数。In formula (10), dk is the target output of nodek , yk is the predicted output of nodek , M is the filter size, and k is the number of nodes.
步骤3、将经步骤1得到的一部分分合闸线圈电流数据输入到经步骤2训练好的故障类型预测模型中，由故障类型预测模型对输入的分合闸线圈电流数据进行处理，完成对高压断路器故障检测。Step 3. Input part of the opening and closing coil current data obtained in step 1 into the fault type prediction model trained in step 2, and the fault type prediction model processes the input opening and closing coil current data to complete the high-voltage Circuit breaker fault detection.
实施例Example
以t0为命令时间的零点提取故障特征参数I1，I2，I3，t1，t2，t3，t4，t5对断路器进行状态监测，获取十组故障样本数据，这十组故障样本数据包括机构正常(A)、操作电压过低(B)、合闸铁心开始阶段由卡涩(C)、操作机构有卡涩(D)及合闸铁心空行程太大(E)，数据采集状况具体如表1所示；Take t0 as the zero point of the command time to extract the fault characteristic parameters I1, I2, I3, t1, t2, t3, t4, t5 to monitor the state of the circuit breaker, and obtain ten sets of fault sample data. These ten sets of fault sample data include normal mechanism ( A), the operating voltage is too low (B), the closing iron core is jammed at the beginning (C), the operating mechanism is jammed (D), and the empty travel of the closing iron core is too large (E). The data collection conditions are shown in Table 1 shown;
表1故障样本数据Table 1 Fault sample data
合/分闸线圈电流的特性曲线如图4所示，可知：The characteristic curve of closing/opening coil current is shown in Figure 4, it can be seen that:
(1)阶段Ⅰ，t＝t0～t1；线圈在t0时刻开始通电，到t1时刻铁心开始运动；t0为断路器分、合闸命令下达时刻，是断路器分、合动作计时起点；T1为线圈中电流、磁通上升到足以驱动铁心运动，即铁心开始运动的时刻；这一阶段的特点是电流呈指数上升，铁心静止；这一阶段的时间与控制电源电压及线圈电阻有关。(1) Phase I, t=t0～t1; the coil starts to be energized at t0, and the iron core starts to move at t1; t0 is the moment when the circuit breaker opens and closes the command, and it is the starting point of the circuit breaker’s opening and closing action timing; T1 is The current and magnetic flux in the coil rise enough to drive the iron core to move, that is, the moment when the iron core starts to move; the characteristic of this stage is that the current rises exponentially, and the iron core is stationary; the time of this stage is related to the control power supply voltage and the coil resistance.
(2)阶段Ⅱ，t＝t1～t2；在这一阶段，铁心开始运动，电流下降；t2为控制电流的谷点，代表铁心已经触动操作机械的负载而显著减速或停止运动。(2) Stage II, t=t1~t2; in this stage, the iron core starts to move, and the current drops; t2 is the valley point of the control current, which means that the iron core has touched the load of the operating machine and significantly slowed down or stopped moving.
(3)阶段Ⅲ，t＝t2～t3；这一阶段铁心停止运动，电流又呈指数上升。(3) Stage III, t=t2～t3; at this stage, the iron core stops moving, and the current rises exponentially again.
(4)阶段Ⅳ，t＝t3～t4；这一阶段是阶段Ⅲ的延续，电流达到近似的稳态。(4) Stage IV, t=t3~t4; this stage is the continuation of stage III, and the current reaches an approximate steady state.
(5)阶段Ⅴ，t＝t4～t5；电流开断阶段，此阶段辅助开关分断，在辅助开关触头间产生电弧并被拉长，电弧电压快速升高，迫使电流迅速减小，直到熄灭。(5) Stage Ⅴ, t=t4～t5; current breaking stage, the auxiliary switch is broken at this stage, an arc is generated between the contacts of the auxiliary switch and is elongated, and the arc voltage rises rapidly, forcing the current to decrease rapidly until it goes out .
分析电流波形可知，t0～t1时间电流可以反映线圈的状态(如：电阻是否正常)。t＝t1～t2时间电流的变化表征铁心运动结构有无卡涩，脱扣、释能机械负载变动的情况；t2一般是动触头开始运动时刻，从t2以后是机构通过传动系统带动动触头分、合闸的过程，即动触头运动的过程；t4为断路器的辅助触点切断的时刻；t0～t4时间电流的变化可以反映机械操动机构传动系统的工作情况。Analysis of the current waveform shows that the current from t0 to t1 can reflect the state of the coil (such as whether the resistance is normal). The change of the current at t=t1～t2 indicates whether the core movement structure is jammed, tripped, and the mechanical load of the energy release changes; t2 is generally the moment when the moving contact starts to move. After t2, the mechanism drives the moving contact through the transmission system. The process of opening and closing the switch is the process of moving the moving contact; t4 is the moment when the auxiliary contact of the circuit breaker is cut off; the change of current from t0 to t4 can reflect the working condition of the mechanical operating mechanism transmission system.
故障类型的输出采用进制数来表示，具体如表2所示：The output of the fault type is represented by a decimal number, as shown in Table 2:
表2故障类型输出表示Table 2 Fault type output representation
本发明基于卷积神经网络算法的高压断路器故障检测方法正确率为96.6％。The correct rate of the high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention is 96.6%.
本发明基于卷积神经网络算法的高压断路器故障检测方法，采用卷积神经网络分析故障特征信号，在弥补人工神经网络检测的不足的同时，能更加准确有效地判断断路器的故障类型，进而能够有效率的检修。The high-voltage circuit breaker fault detection method based on the convolutional neural network algorithm of the present invention adopts the convolutional neural network to analyze the fault characteristic signal, and can more accurately and effectively judge the fault type of the circuit breaker while making up for the deficiency of the artificial neural network detection, and then Can be repaired efficiently.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610835299.1ACN106291351B (en) | 2016-09-20 | 2016-09-20 | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610835299.1ACN106291351B (en) | 2016-09-20 | 2016-09-20 | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm |
| Publication Number | Publication Date |
|---|---|
| CN106291351Atrue CN106291351A (en) | 2017-01-04 |
| CN106291351B CN106291351B (en) | 2019-01-18 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610835299.1AExpired - Fee RelatedCN106291351B (en) | 2016-09-20 | 2016-09-20 | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm |
| Country | Link |
|---|---|
| CN (1) | CN106291351B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107238507A (en)* | 2017-06-20 | 2017-10-10 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of industrial equipment failure prediction method based on deep learning |
| CN107449994A (en)* | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN107450016A (en)* | 2017-07-24 | 2017-12-08 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on RST CNN |
| CN107490760A (en)* | 2017-08-22 | 2017-12-19 | 西安工程大学 | The circuit breaker failure diagnostic method of fuzzy neural network is improved based on genetic algorithm |
| CN108520080A (en)* | 2018-05-11 | 2018-09-11 | 武汉理工大学 | System and method for marine diesel generator failure prediction and health status online assessment |
| CN108734202A (en)* | 2018-04-27 | 2018-11-02 | 西安工程大学 | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP |
| CN109270442A (en)* | 2018-08-21 | 2019-01-25 | 西安工程大学 | High-voltage circuitbreaker fault detection method based on DBN-GA neural network |
| CN110174610A (en)* | 2019-05-17 | 2019-08-27 | 温州大学 | A method of obtaining A.C. contactor electric life based on convolutional neural networks |
| CN110297178A (en)* | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel generating set fault diagnosis and detection device and method based on deep learning |
| CN110296802A (en)* | 2019-06-26 | 2019-10-01 | 国网辽宁省电力有限公司葫芦岛供电公司 | Shaft tower bolt looseness method of discrimination based on vibrating speed sensors waveform acquisition |
| CN110333440A (en)* | 2019-07-04 | 2019-10-15 | 深圳供电局有限公司 | Detection method and detection device for circuit breaker, computer equipment and storage medium |
| CN110673024A (en)* | 2019-09-10 | 2020-01-10 | 国网浙江省电力有限公司嘉兴供电公司 | A high-voltage circuit breaker fault early warning method based on the power Internet of things |
| CN110705038A (en)* | 2019-09-10 | 2020-01-17 | 国网浙江省电力有限公司嘉兴供电公司 | A kind of high-voltage circuit breaker life cycle assessment and fault early warning method |
| CN110703075A (en)* | 2019-09-10 | 2020-01-17 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker quick overhaul method based on big data technology |
| CN111044890A (en)* | 2019-11-22 | 2020-04-21 | 国网浙江省电力有限公司嘉兴供电公司 | Comprehensive test device and method for high-voltage circuit breaker based on ubiquitous power Internet of things |
| CN111060304A (en)* | 2020-02-03 | 2020-04-24 | 国网江苏省电力有限公司宿迁供电分公司 | Circuit breaker fault diagnosis method based on neural network |
| CN111060815A (en)* | 2019-12-17 | 2020-04-24 | 西安工程大学 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
| CN111272405A (en)* | 2020-03-26 | 2020-06-12 | 广西电网有限责任公司电力科学研究院 | A method and system for diagnosing mechanical faults of high-voltage circuit breakers |
| CN111638649A (en)* | 2020-05-13 | 2020-09-08 | 江苏泛锐思智能技术有限公司 | Method and device for monitoring operation safety of electric appliance |
| CN111913103A (en)* | 2020-08-06 | 2020-11-10 | 国网福建省电力有限公司 | Fault detection method for spring energy storage operating structure circuit breaker |
| CN112348159A (en)* | 2019-08-07 | 2021-02-09 | 青岛鼎信通讯股份有限公司 | Fault arc neural network optimization training method |
| CN112485655A (en)* | 2020-10-29 | 2021-03-12 | 国网湖北省电力有限公司电力科学研究院 | Accurate monitoring device and method for tripping and closing loop of substation circuit breaker |
| CN113030717A (en)* | 2021-03-12 | 2021-06-25 | 广东电网有限责任公司 | Circuit breaker fault diagnosis method and circuit based on electric signal deep learning |
| CN113191192A (en)* | 2021-03-25 | 2021-07-30 | 云南电网有限责任公司玉溪供电局 | Breaker fault detection method based on wavelet analysis and fuzzy neural network algorithm |
| CN113671361A (en)* | 2021-06-25 | 2021-11-19 | 海南电网有限责任公司海南输变电检修分公司 | High-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion |
| CN114509664A (en)* | 2020-11-17 | 2022-05-17 | Abb瑞士股份有限公司 | System for monitoring circuit breaker |
| CN116662750A (en)* | 2023-05-12 | 2023-08-29 | 中国空分工程有限公司 | Air compressor fault detection method based on deep learning |
| CN117590223A (en)* | 2024-01-18 | 2024-02-23 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1022494A2 (en)* | 1998-12-23 | 2000-07-26 | Eaton Corporation | Fault detection system and method for solenoid controlled actuators of a transmission system |
| CN101261192A (en)* | 2008-04-24 | 2008-09-10 | 南京因泰莱配电自动化设备有限公司 | 24KV high-voltage vacuum breaker mechanical features on-line monitoring method and monitoring device |
| CN202562733U (en)* | 2012-03-30 | 2012-11-28 | 山西省电力公司太原供电分公司 | High voltage breaker mechanical property on-line live-line detection system |
| CN103078404A (en)* | 2012-12-31 | 2013-05-01 | 广州供电局有限公司 | Intelligentized switch cabinet monitoring system and method |
| CN203054194U (en)* | 2012-12-31 | 2013-07-10 | 广州供电局有限公司 | Monitoring system for breaking coil and closing coil of breaker |
| CN104538222A (en)* | 2014-12-27 | 2015-04-22 | 中国西电电气股份有限公司 | High-voltage switch phase selection controller based on artificial neural network and method |
| CN104657908A (en)* | 2014-11-12 | 2015-05-27 | 国家电网公司 | New breaker heating amount calculating method |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1022494A2 (en)* | 1998-12-23 | 2000-07-26 | Eaton Corporation | Fault detection system and method for solenoid controlled actuators of a transmission system |
| CN101261192A (en)* | 2008-04-24 | 2008-09-10 | 南京因泰莱配电自动化设备有限公司 | 24KV high-voltage vacuum breaker mechanical features on-line monitoring method and monitoring device |
| CN202562733U (en)* | 2012-03-30 | 2012-11-28 | 山西省电力公司太原供电分公司 | High voltage breaker mechanical property on-line live-line detection system |
| CN103078404A (en)* | 2012-12-31 | 2013-05-01 | 广州供电局有限公司 | Intelligentized switch cabinet monitoring system and method |
| CN203054194U (en)* | 2012-12-31 | 2013-07-10 | 广州供电局有限公司 | Monitoring system for breaking coil and closing coil of breaker |
| CN104657908A (en)* | 2014-11-12 | 2015-05-27 | 国家电网公司 | New breaker heating amount calculating method |
| CN104538222A (en)* | 2014-12-27 | 2015-04-22 | 中国西电电气股份有限公司 | High-voltage switch phase selection controller based on artificial neural network and method |
| Title |
|---|
| 胡正平等: "卷积神经网络分类模型在模式识别中的新进展", 《燕山大学学报》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107238507A (en)* | 2017-06-20 | 2017-10-10 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of industrial equipment failure prediction method based on deep learning |
| CN107449994A (en)* | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN107450016A (en)* | 2017-07-24 | 2017-12-08 | 西安工程大学 | Fault Diagnosis for HV Circuit Breakers method based on RST CNN |
| CN107490760A (en)* | 2017-08-22 | 2017-12-19 | 西安工程大学 | The circuit breaker failure diagnostic method of fuzzy neural network is improved based on genetic algorithm |
| CN108734202A (en)* | 2018-04-27 | 2018-11-02 | 西安工程大学 | A kind of Fault Diagnosis for HV Circuit Breakers method based on improved BP |
| CN110297178A (en)* | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel generating set fault diagnosis and detection device and method based on deep learning |
| CN108520080A (en)* | 2018-05-11 | 2018-09-11 | 武汉理工大学 | System and method for marine diesel generator failure prediction and health status online assessment |
| CN109270442A (en)* | 2018-08-21 | 2019-01-25 | 西安工程大学 | High-voltage circuitbreaker fault detection method based on DBN-GA neural network |
| CN110174610A (en)* | 2019-05-17 | 2019-08-27 | 温州大学 | A method of obtaining A.C. contactor electric life based on convolutional neural networks |
| CN110296802A (en)* | 2019-06-26 | 2019-10-01 | 国网辽宁省电力有限公司葫芦岛供电公司 | Shaft tower bolt looseness method of discrimination based on vibrating speed sensors waveform acquisition |
| CN110333440A (en)* | 2019-07-04 | 2019-10-15 | 深圳供电局有限公司 | Detection method and detection device for circuit breaker, computer equipment and storage medium |
| CN112348159A (en)* | 2019-08-07 | 2021-02-09 | 青岛鼎信通讯股份有限公司 | Fault arc neural network optimization training method |
| CN110673024A (en)* | 2019-09-10 | 2020-01-10 | 国网浙江省电力有限公司嘉兴供电公司 | A high-voltage circuit breaker fault early warning method based on the power Internet of things |
| CN110705038A (en)* | 2019-09-10 | 2020-01-17 | 国网浙江省电力有限公司嘉兴供电公司 | A kind of high-voltage circuit breaker life cycle assessment and fault early warning method |
| CN110703075A (en)* | 2019-09-10 | 2020-01-17 | 国网浙江省电力有限公司嘉兴供电公司 | High-voltage circuit breaker quick overhaul method based on big data technology |
| CN110705038B (en)* | 2019-09-10 | 2023-09-26 | 国网浙江省电力有限公司嘉兴供电公司 | A high-voltage circuit breaker life cycle assessment and fault early warning method |
| CN110673024B (en)* | 2019-09-10 | 2022-05-13 | 国网浙江省电力有限公司嘉兴供电公司 | A high-voltage circuit breaker fault early warning method based on the power Internet of things |
| CN110703075B (en)* | 2019-09-10 | 2022-03-15 | 国网浙江省电力有限公司嘉兴供电公司 | A rapid maintenance method for high-voltage circuit breakers based on big data technology |
| CN111044890B (en)* | 2019-11-22 | 2022-03-15 | 国网浙江省电力有限公司嘉兴供电公司 | Comprehensive test device and method for high-voltage circuit breaker based on ubiquitous power Internet of things |
| CN111044890A (en)* | 2019-11-22 | 2020-04-21 | 国网浙江省电力有限公司嘉兴供电公司 | Comprehensive test device and method for high-voltage circuit breaker based on ubiquitous power Internet of things |
| CN111060815B (en)* | 2019-12-17 | 2021-09-14 | 西安工程大学 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
| CN111060815A (en)* | 2019-12-17 | 2020-04-24 | 西安工程大学 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
| CN111060304A (en)* | 2020-02-03 | 2020-04-24 | 国网江苏省电力有限公司宿迁供电分公司 | Circuit breaker fault diagnosis method based on neural network |
| CN111272405B (en)* | 2020-03-26 | 2022-08-16 | 广西电网有限责任公司电力科学研究院 | High-voltage circuit breaker mechanical fault diagnosis method and system |
| CN111272405A (en)* | 2020-03-26 | 2020-06-12 | 广西电网有限责任公司电力科学研究院 | A method and system for diagnosing mechanical faults of high-voltage circuit breakers |
| CN111638649A (en)* | 2020-05-13 | 2020-09-08 | 江苏泛锐思智能技术有限公司 | Method and device for monitoring operation safety of electric appliance |
| CN111913103A (en)* | 2020-08-06 | 2020-11-10 | 国网福建省电力有限公司 | Fault detection method for spring energy storage operating structure circuit breaker |
| CN111913103B (en)* | 2020-08-06 | 2022-11-08 | 国网福建省电力有限公司 | A kind of fault detection method of spring energy storage operation structure circuit breaker |
| CN112485655A (en)* | 2020-10-29 | 2021-03-12 | 国网湖北省电力有限公司电力科学研究院 | Accurate monitoring device and method for tripping and closing loop of substation circuit breaker |
| CN112485655B (en)* | 2020-10-29 | 2022-05-31 | 国网湖北省电力有限公司电力科学研究院 | Accurate monitoring device and method for tripping and closing loop of substation breaker |
| CN114509664A (en)* | 2020-11-17 | 2022-05-17 | Abb瑞士股份有限公司 | System for monitoring circuit breaker |
| CN113030717A (en)* | 2021-03-12 | 2021-06-25 | 广东电网有限责任公司 | Circuit breaker fault diagnosis method and circuit based on electric signal deep learning |
| CN113191192A (en)* | 2021-03-25 | 2021-07-30 | 云南电网有限责任公司玉溪供电局 | Breaker fault detection method based on wavelet analysis and fuzzy neural network algorithm |
| CN113671361A (en)* | 2021-06-25 | 2021-11-19 | 海南电网有限责任公司海南输变电检修分公司 | High-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion |
| CN116662750A (en)* | 2023-05-12 | 2023-08-29 | 中国空分工程有限公司 | Air compressor fault detection method based on deep learning |
| CN117590223A (en)* | 2024-01-18 | 2024-02-23 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| CN117590223B (en)* | 2024-01-18 | 2024-04-30 | 南京飞腾电子科技有限公司 | Online monitoring system and method for circuit breaker |
| Publication number | Publication date |
|---|---|
| CN106291351B (en) | 2019-01-18 |
| Publication | Publication Date | Title |
|---|---|---|
| CN106291351B (en) | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm | |
| CN109270442B (en) | DBN-GA neural network-based high-voltage circuit breaker fault detection method | |
| CN111060815B (en) | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method | |
| CN109917205B (en) | A device and method for fault diagnosis of solenoid valve based on feature extraction and multi-layer perceptron | |
| CN110221200A (en) | A kind of omnipotent breaker attachment method for diagnosing faults based on deep learning | |
| CN106856322B (en) | A kind of flexible direct current power distribution network intelligent protection system based on neural network | |
| CN107450016A (en) | Fault Diagnosis for HV Circuit Breakers method based on RST CNN | |
| CN107423839A (en) | A kind of method of the intelligent building microgrid load prediction based on deep learning | |
| CN106600050A (en) | BP neural network-based ultra-short load prediction method | |
| CN113434493A (en) | Non-invasive load decomposition method based on Transformer | |
| CN105891708A (en) | ANFIS-based condition assessment for high-voltage circuit breaker operating mechanism | |
| CN114839523A (en) | Circuit breaker fault detection system and method based on CNN combined LSTM algorithm | |
| CN117131407A (en) | Relay protection strategy generation method, device, electronic equipment and storage medium | |
| CN101702537A (en) | Method of Self-Adaptive Handling of Faults in Distribution Network Terminals | |
| Chen et al. | Remaining useful life prediction of turbofan engine based on temporal convolutional networks optimized by genetic algorithm | |
| CN113536509A (en) | A microgrid topology identification method based on graph convolutional network | |
| CN117031291A (en) | Method and terminal for detecting available battery charging capacity | |
| CN114113896A (en) | Modular multi-level converter type high-voltage direct-current transmission fault diagnosis method | |
| CN112085043B (en) | Intelligent monitoring method and system for network security of transformer substation | |
| Fu et al. | Fault diagnosis of power electronic device based on wavelet and neural network | |
| Pei et al. | Mechanical remaining Life Prediction of Circuit Breaker based on long-term and short-term memory Neural Network | |
| CN112100767A (en) | Aero-engine service life prediction method based on singular value decomposition and GRU | |
| CN113469222A (en) | RST-PNN-GA-based high-voltage circuit breaker fault detection method | |
| Chen et al. | A transformer vibration amplitude prediction method via fusion of multi-signals | |
| Liwen et al. | Fault diagnosis of distribution network based on artificial intelligence |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right | Effective date of registration:20210812 Address after:710065 No. 11301, block B, Huajing Business Plaza, No. 20, Fenghui South Road, Zhangba street, high tech Zone, Xi'an, Shaanxi Province Patentee after:XI'AN JIN POWER ELECTRICAL Co.,Ltd. Address before:710048 No. 19 Jinhua South Road, Shaanxi, Xi'an Patentee before:XI'AN POLYTECHNIC University | |
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee | Granted publication date:20190118 |