CN111585671B - Electromagnetic interference monitoring and identification method of power LTE wireless private network - Google Patents
Electromagnetic interference monitoring and identification method of power LTE wireless private networkDownload PDFInfo
- Publication number
- CN111585671B CN111585671BCN202010297105.3ACN202010297105ACN111585671BCN 111585671 BCN111585671 BCN 111585671BCN 202010297105 ACN202010297105 ACN 202010297105ACN 111585671 BCN111585671 BCN 111585671B
- Authority
- CN
- China
- Prior art keywords
- signal
- features
- spectrum
- interference
- identification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Classifications
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/001—Measuring interference from external sources to, or emission from, the device under test, e.g. EMC, EMI, EMP or ESD testing
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/382—Monitoring; Testing of propagation channels for resource allocation, admission control or handover
Landscapes
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- General Physics & Mathematics (AREA)
- Monitoring And Testing Of Transmission In General (AREA)
Abstract
Translated fromChinese
Description
Translated fromChinese技术领域technical field
本发明属于无线电频谱检测和干扰分析领域，特别涉及一种电力LTE无线专网电磁干扰监测及识别方法。The invention belongs to the field of radio spectrum detection and interference analysis, and in particular relates to a method for monitoring and identifying electromagnetic interference of a power LTE wireless private network.
背景技术Background technique
目前，电力系统已经在多个地市开展了基于LTE技术的电力无线专网建设工作，有效支撑了电力业务的通信承载需求，实现终端侧业务的灵活泛在接入，促进了业务的智能化发展，无线通信在电力应用取得一定成果的同时，存在的一些问题也逐步暴露，其中干扰问题对网络的性能影响尤为突出，甚至直接导致电力业务中断，因此，迫切需要对电力系统中无线电频谱进行实时监测、对电力无线电专网网络干扰分析、判断和查处。At present, the power system has carried out the construction of power wireless private network based on LTE technology in many cities, which effectively supports the communication bearer requirements of power services, realizes flexible and ubiquitous access to terminal-side services, and promotes the intelligentization of services. While the development of wireless communication has achieved certain results in power applications, some problems have gradually been exposed. Among them, the interference problem has a particularly prominent impact on the performance of the network, and even directly leads to the interruption of power services. Real-time monitoring, analysis, judgment and investigation of the interference of the power radio private network network.
频率资源是影响电力无线专网的网络性能、建设成本等的重要因素，现在主要采用230MHz离散频段和1.8GHz的连续频段，在基于不同频段的专网系统中存在多业务频率资源分配、同频异系统共享资源等情况，如何高效使用频率、快速定位干扰是提高电力无线专网服务质量的下一步重点工作，频谱监测是采用技术手段和一定的设备对无线电发射的基本参数和频谱特性参数进行测量，可以实现对数字信号的频谱特性分析；对频段利用率和频带占有度统计测试分析；测试统计指配频率使用情况，以便进行合理、有效地频率指配；并对非法电台和干扰源测向定位进行查处。Frequency resources are an important factor affecting the network performance and construction cost of the power wireless private network. Currently, the 230MHz discrete frequency band and the 1.8GHz continuous frequency band are mainly used. In the private network system based on different frequency bands, there are multi-service frequency resource allocation, co-frequency In the case of different systems sharing resources, how to efficiently use the frequency and quickly locate the interference is the next key work to improve the service quality of the power wireless private network. The measurement can realize the analysis of the spectrum characteristics of the digital signal; the statistical test and analysis of the frequency band utilization rate and the frequency band occupancy; Check the location.
电力无线应用场景多、覆盖范围广，缺少智能化的电磁环境综合评估机制，无法为频率资源管理提供决策依据，当前频率分配手段粗放，主要依靠人工经验和业务需求粗略估计进行分配，科学性、自动化、精细化有待提高，且电磁环境评估只能通过手持式仪表进行小范围的扫频，灵活性不足，且实时性不强。There are many application scenarios and wide coverage of power wireless, and there is a lack of an intelligent comprehensive evaluation mechanism for electromagnetic environment, which cannot provide decision-making basis for frequency resource management. The current frequency allocation methods are extensive, mainly relying on manual experience and rough estimation of business needs. Automation and refinement need to be improved, and the electromagnetic environment assessment can only be performed by a small-scale frequency sweep through a hand-held instrument, which lacks flexibility and is not very real-time.
目前专网的频段都是多行业共享使用的频段，相邻频段的信号泄露、重叠等原因会对无线专网的稳定造成干扰，影响业务接入质量，同时，各类非法电台的出现，给电力无线专网的可靠运行带来了更多的隐患，必须寻找可行方法准确判断干扰类型和干扰溯源，进而采取相关措施，以免造成业务大面积掉线，影响电力业务承载质量。At present, the frequency bands of the private network are shared by multiple industries. The signal leakage and overlapping of adjacent frequency bands will interfere with the stability of the wireless private network and affect the quality of service access. The reliable operation of the power wireless private network brings more hidden dangers. It is necessary to find a feasible method to accurately determine the type of interference and trace the source of the interference, and then take relevant measures to avoid large-scale service disconnection and affect the power service bearing quality.
发明内容SUMMARY OF THE INVENTION
本发明所要解决的技术问题是：克服现有技术不足，提供一种能够显著提高识别精度的面向电力LTE无线专网电磁干扰监测及识别方法。The technical problem to be solved by the present invention is: to overcome the deficiencies of the prior art, and to provide a method for monitoring and identifying electromagnetic interference in a power-oriented LTE wireless private network that can significantly improve the identification accuracy.
本发明为解决技术问题所采用的技术方案是：The technical scheme adopted by the present invention for solving the technical problem is:
一种电力LTE无线专网电磁干扰监测及识别方法，包括两个过程：频谱监测和干扰识别，具体步骤如下：A method for monitoring and identifying electromagnetic interference in an electric power LTE wireless private network includes two processes: spectrum monitoring and interference identification, and the specific steps are as follows:
步骤1，利用传感器将外界输入的非电压信号转换成电压信号，送给信号调理设备处理；Step 1, use the sensor to convert the non-voltage signal input from the outside into a voltage signal, and send it to the signal conditioning equipment for processing;
步骤2，信号调理设备将输入的电压信号进行放大、衰减和滤波，最后将处理过的信号送入频谱分析仪；Step 2, the signal conditioning device amplifies, attenuates and filters the input voltage signal, and finally sends the processed signal to the spectrum analyzer;
步骤3，频谱分析仪对输入的经步骤2处理的信号进行A/D转换，完成测试数据的采集；Step 3, the spectrum analyzer performs A/D conversion on the input signal processed in step 2 to complete the collection of test data;
步骤4，对步骤3采集的数据进行分析计算、处理，实现干扰源识别，所述干扰源识别过程如下：Step 4, analyze, calculate, and process the data collected in step 3 to realize interference source identification. The interference source identification process is as follows:
步骤4-1，数据预处理：对原始测试数据利用小波变换进行消噪，对原始测试数据进行数据归一化操作，对发射特性曲线进行包络和延拓处理，通过降维操作将高维向量映射到低维空间；Step 4-1, data preprocessing: use wavelet transform to de-noise the original test data, perform data normalization operation on the original test data, perform envelope and continuation processing on the emission characteristic curve, and reduce the high-dimensional data through dimension reduction operations. The vector is mapped to a low-dimensional space;
步骤4-2，特征提取：提取采集信号的特征，包括峰值特征、包络特征、谐波特征，采用傅里叶变换对周期性信号和振铃信号进行频域特征提取；采用快速傅里叶变换对离散信号进行频域特征提取；采用短时傅里叶变换对脉冲信号进行频谱特征提取；采用小波变换对非平稳信号进行频域特征提取；Step 4-2, feature extraction: extract the features of the collected signal, including peak features, envelope features, and harmonic features, and use Fourier transform to extract frequency domain features for periodic signals and ringing signals; use fast Fourier transform Transform to extract frequency domain features for discrete signals; use short-time Fourier transform to extract spectral features for pulse signals; use wavelet transform to extract frequency domain features for non-stationary signals;
步骤4-3，特征选择；特征选择包括两个内容，第一是选择训练样本的统计特征评估算法对特征进行评估，计算每一个特征对分类效果的贡献值；第二是利用主成分分析算法挑选出部分属性，针对不同的电磁兼容测试对象，选择不同的特征作为辨识对象，从当前的向量属性集合中选择区分能力强的子集，剔除了干扰分类效果的冗余特征或无效特征，形成一个紧凑的特征集；Step 4-3, feature selection; feature selection includes two contents, the first is to select the statistical feature evaluation algorithm of the training samples to evaluate the features, and calculate the contribution value of each feature to the classification effect; the second is to use the principal component analysis algorithm Select some attributes, select different features as identification objects for different electromagnetic compatibility test objects, select a subset with strong distinguishing ability from the current vector attribute set, eliminate redundant features or invalid features that interfere with the classification effect, and form a compact feature set;
步骤4-4，学习和训练；利用基于卷积神经网络的深度学习算法，学习采集信号特征与干扰源之间的关系，得到基于干扰源识别的分类模型；Step 4-4, learning and training; using a deep learning algorithm based on a convolutional neural network to learn the relationship between the characteristics of the collected signal and the interference source, and obtain a classification model based on the identification of the interference source;
卷积的数学形式如下：The mathematical form of convolution is as follows:
y＝f(∑x*wij+b)y=f(∑x*wij +b)
式中，x是卷积层输入特征图；y是卷积层输出特征图；wij是二维卷积核；b是偏置项；f(·)是激活函数，如sigmoid函数或tanh函数或ReLU函数；where x is the input feature map of the convolution layer; y is the output feature map of the convolution layer; wij is the two-dimensional convolution kernel; b is the bias term; f( ) is the activation function, such as the sigmoid function or the tanh function or ReLU function;
卷积神经网络使用多个过滤器、多个卷积层、多个池化层，最后连接一个全连接层和一个softmax层，最终输出对应输入在每一类输出上的概率分布；The convolutional neural network uses multiple filters, multiple convolution layers, multiple pooling layers, and finally connects a fully connected layer and a softmax layer, and finally outputs the probability distribution corresponding to the input on each type of output;
整个网络的输出为The output of the entire network is
o＝f(n-1)(f(n-2)(...f(1)(x)))o=f(n-1) (f(n-2) (...f(1) (x)))
式中，x是卷积层输入特征图；f(·)是激活函数，如sigmoid函数或tanh函数或ReLU函数；In the formula, x is the input feature map of the convolution layer; f( ) is the activation function, such as the sigmoid function or the tanh function or the ReLU function;
根据类别属性已知的样本集与其向量取值的对应关系，形成对应模式分类方法的一系列参数；According to the corresponding relationship between the known sample set of category attribute and its vector value, a series of parameters corresponding to the pattern classification method are formed;
步骤4-5，分类识别；对测试信号提取特征，采用步骤4-4得到的分类模型对类别属性未知的测试数据集进行干扰源分类，如果测试数据集中的类别属性已知，这个过程就可以用来验证分类器的分类效果，如果分类效果理想，该分类模型就可以运用到实际流量分类中。Step 4-5, classification and identification; extract features from the test signal, and use the classification model obtained in step 4-4 to classify the interference source for the test data set with unknown category attributes. If the category attributes in the test data set are known, this process can be done. It is used to verify the classification effect of the classifier. If the classification effect is satisfactory, the classification model can be applied to the actual traffic classification.
进一步的，在步骤1中，所述传感器包括电流探头、电压探头、近场探头以及接收天线；进行传导发射测试时，频谱仪与电源阻抗稳定网络或电流探头相连；当进行辐射发射测试时，频谱仪与接收天线相连；在天线间的耦合度测试中，将车载天线连接到接收机上，并由此来分析天线的耦合度。Further, in step 1, the sensor includes a current probe, a voltage probe, a near-field probe and a receiving antenna; when conducting a conducted emission test, the spectrum analyzer is connected to a power supply impedance stabilization network or a current probe; when conducting a radiated emission test, The spectrum analyzer is connected with the receiving antenna; in the coupling degree test between the antennas, the vehicle antenna is connected to the receiver, and the coupling degree of the antenna is analyzed.
进一步的，在步骤2中，信号调理设备包括信号放大器、线性阻抗衰减器、带阻滤波器和高通滤波器。Further, in step 2, the signal conditioning equipment includes a signal amplifier, a linear impedance attenuator, a band-stop filter and a high-pass filter.
进一步的，步骤3中，所述频谱分析仪对步骤2中的数据频谱监测过程如下：在高频段采用扫描外差式频谱监测设计，通过多级变频处理，将输入信号变至较低的中频上；在中频段采用傅里叶频谱监测设计，对变频到中频的信号进行A/D采样量化，变为数字信号，再利用数字中频技术、傅里叶变换完成频谱分析。Further, in step 3, the process of monitoring the data spectrum in step 2 by the spectrum analyzer is as follows: a sweeping heterodyne spectrum monitoring design is adopted in the high frequency band, and the input signal is changed to a lower intermediate frequency through multi-stage frequency conversion processing. In the middle frequency band, the Fourier spectrum monitoring design is adopted, and the A/D sampling and quantization are performed on the signal converted to the intermediate frequency, which is converted into a digital signal, and then the spectrum analysis is completed by using the digital intermediate frequency technology and Fourier transform.
本发明的积极有益效果是：The positive beneficial effects of the present invention are:
1、本发明的频谱监测在高频段和中频分别采用扫描外差式频谱监测设计和傅里叶频谱监测设计，能够发挥了扫描外差式频谱设计测量频率范围宽，傅里叶频谱监测设计频率分辨力强的优势，大大提高了频谱监测的性能。1. The spectrum monitoring of the present invention adopts the scanning heterodyne spectrum monitoring design and the Fourier spectrum monitoring design respectively in the high frequency band and the intermediate frequency, which can bring into play that the scanning heterodyne spectrum design has a wide measurement frequency range, and the Fourier spectrum monitoring design frequency The advantage of strong resolution greatly improves the performance of spectrum monitoring.
2、本发明中干扰识别问题转化为机器学习的分类问题，在对采集数据进行数据预处理，特征提取和特征选择后，利用基于卷积神经网络的深度学习算法，学习采集信号特征与干扰源之间的关系，得到基于干扰源识别的分类模型，能够显著提高识别精度。2. The problem of interference identification in the present invention is transformed into a classification problem of machine learning. After data preprocessing, feature extraction and feature selection are performed on the collected data, a deep learning algorithm based on a convolutional neural network is used to learn the characteristics of the collected signal and the source of interference. A classification model based on interference source identification is obtained, which can significantly improve the identification accuracy.
3、本发明能够针对电力系统需要，将电磁频谱将频谱监测和干扰识别功能紧密结合，适用1.8GLTE无线专网。3. The present invention can closely combine the electromagnetic spectrum with the functions of spectrum monitoring and interference identification according to the needs of the power system, and is suitable for the 1.8GLTE wireless private network.
附图说明Description of drawings
图1是本发明中频谱监测和干扰识别的总体框图；Fig. 1 is the overall block diagram of spectrum monitoring and interference identification in the present invention;
图2是本发明中扫频外差式频谱监测原理图；Fig. 2 is the schematic diagram of frequency sweep heterodyne spectrum monitoring in the present invention;
图3是本发明中傅里叶式频谱监测原理图；3 is a schematic diagram of Fourier spectrum monitoring in the present invention;
图4是本发明中干扰源识别流程图；Fig. 4 is the interference source identification flow chart in the present invention;
图5是本发明实际应用下基于多用户、多任务模式的数据流程图。FIG. 5 is a data flow diagram based on a multi-user and multi-task mode under the practical application of the present invention.
具体实施方式Detailed ways
下面结合附图1、图2、图3、图4、图5和具体实施例对本发明的技术方案做进一步的详细说明：Below in conjunction with accompanying drawing 1, Fig. 2, Fig. 3, Fig. 4, Fig. 5 and specific embodiment, the technical scheme of the present invention is described in further detail:
一种电力LTE无线专网电磁干扰监测及识别方法，包括两个过程：频谱监测和干扰识别，具体步骤如下：A method for monitoring and identifying electromagnetic interference in an electric power LTE wireless private network includes two processes: spectrum monitoring and interference identification, and the specific steps are as follows:
步骤1，利用传感器将外界输入的非电压信号转换成电压信号，送给信号调理设备处理；Step 1, use the sensor to convert the non-voltage signal input from the outside into a voltage signal, and send it to the signal conditioning equipment for processing;
步骤2，信号调理设备将输入的电压信号进行放大、衰减和滤波，最后将处理过的信号送入频谱分析仪；Step 2, the signal conditioning device amplifies, attenuates and filters the input voltage signal, and finally sends the processed signal to the spectrum analyzer;
步骤3，频谱分析仪对输入的经步骤2处理的信号进行A/D转换，完成测试数据的采集；Step 3, the spectrum analyzer performs A/D conversion on the input signal processed in step 2 to complete the collection of test data;
步骤4，对步骤3采集的数据进行分析计算、处理，实现干扰源识别，所述干扰源识别过程包括数据预处理、特征提取、特征选择、学习和训练、分类识别。Step 4, analyze, calculate, and process the data collected in step 3 to realize interference source identification. The interference source identification process includes data preprocessing, feature extraction, feature selection, learning and training, and classification and identification.
进一步的，在步骤1中，所述传感器包括电流探头、电压探头、近场探头以及接收天线；进行传导发射测试时，频谱仪与电源阻抗稳定网络或电流探头相连；当进行辐射发射测试时，频谱仪与接收天线相连；在天线间的耦合度测试中，将车载天线连接到接收机上，并由此来分析天线的耦合度。Further, in step 1, the sensor includes a current probe, a voltage probe, a near-field probe and a receiving antenna; when conducting a conducted emission test, the spectrum analyzer is connected to a power supply impedance stabilization network or a current probe; when conducting a radiated emission test, The spectrum analyzer is connected with the receiving antenna; in the coupling degree test between the antennas, the vehicle antenna is connected to the receiver, and the coupling degree of the antenna is analyzed.
进一步的，在步骤2中，信号调理设备包括信号放大器、线性阻抗衰减器、带阻滤波器和高通滤波器。Further, in step 2, the signal conditioning equipment includes a signal amplifier, a linear impedance attenuator, a band-stop filter and a high-pass filter.
如图1所示，本发明中包括两个基本功能：频谱监测和干扰源识别，频谱分析仪用于频谱监测，机器学习算法用于干扰识别，As shown in Figure 1, the present invention includes two basic functions: spectrum monitoring and interference source identification, spectrum analyzer is used for spectrum monitoring, machine learning algorithm is used for interference identification,
本发明中，电磁干扰监测在高频段采用了扫描外差式的设计，通过多级变频处理，将输入信号变至较低的中频上，而在中频段则采用了傅里叶频谱仪设计，对变频到中频的信号进行A/D采样量化，变为数字信号，再利用数字中频技术、FFT变换完成频谱分析。In the present invention, the electromagnetic interference monitoring adopts the scanning heterodyne design in the high frequency band, and through the multi-stage frequency conversion processing, the input signal is changed to the lower intermediate frequency, and the Fourier spectrum analyzer design is adopted in the intermediate frequency band. A/D sampling and quantization is performed on the signal converted to the intermediate frequency, and then becomes a digital signal, and then the digital intermediate frequency technology and FFT transformation are used to complete the spectrum analysis.
如图2所示，扫描外差式频谱监测利用自动调谐的方式，通过改变本地振荡器的频率来不断进行混频，得到一个固定的中频，其频率变换的原理与超外差式收音机的变频原理相同，只不过把扫频振荡器用作本振而已。As shown in Figure 2, the scanning heterodyne spectrum monitoring uses the method of automatic tuning to continuously mix the frequency by changing the frequency of the local oscillator to obtain a fixed intermediate frequency. The principle is the same, except that the swept frequency oscillator is used as the local oscillator.
扫频外差式频谱仪主要有输入通道、混频电路、中频处理电路、检波和视频滤波等组成，本地振荡器信号由扫描发生器控制，当输入一个信号后，就与与本振信号在混频器中进行差频，只有当差频信号的频率在中频滤波器的带宽内，中频放大器才输出正比于输入信号幅度的一个信号，然后经检波、放大后输出在显示器上，这样就可以通过扫描发生器连续调节本振信号的频率来不断选频，从而达到了频谱测量的目的。The swept frequency heterodyne spectrum analyzer is mainly composed of input channel, frequency mixing circuit, intermediate frequency processing circuit, detection and video filtering, etc. The local oscillator signal is controlled by the scanning generator. The frequency difference is performed in the mixer. Only when the frequency of the difference frequency signal is within the bandwidth of the IF filter, the IF amplifier outputs a signal proportional to the amplitude of the input signal, and then it is detected and amplified and then output on the display. The sweep generator continuously adjusts the frequency of the local oscillator signal to continuously select the frequency, thus achieving the purpose of spectrum measurement.
傅里叶式频谱监测是基于快速傅里叶变换实现频谱分析，傅里叶式频谱监测的原理框图如图3所示，输入信号经一个低通滤波器滤除测量频带之外的高频分量，然后经过一个ADC对信号进行采样量化，从而将输入信号转变为了数字信号，然后对得到的数字信号进行FFT变换，即可得到输入信号的频谱信息，包括频率、相位、调制等信息。Fourier spectrum monitoring is based on fast Fourier transform to realize spectrum analysis. The principle block diagram of Fourier spectrum monitoring is shown in Figure 3. The input signal is filtered by a low-pass filter to remove high-frequency components outside the measurement band. , and then sample and quantize the signal through an ADC to convert the input signal into a digital signal, and then perform FFT transformation on the obtained digital signal to obtain the spectral information of the input signal, including frequency, phase, modulation and other information.
如图4所示，本发明干扰识别方案是先建立关键设备的模板数据库，然后将受扰设备端的测试结果作为待辨识数据，将其通过干扰源辨识算法和模板库中的数据进行比较，最后辨识出干扰源。As shown in Figure 4, the interference identification scheme of the present invention is to first establish a template database of key equipment, then use the test result of the disturbed equipment as the data to be identified, compare it with the data in the template library through the interference source identification algorithm, and finally Identify the source of interference.
干扰识别过程包括：数据预处理、特征提取、特征选择、学习和训练、分类识别。The interference identification process includes: data preprocessing, feature extraction, feature selection, learning and training, and classification and identification.
数据预处理：本发明的测试结果包含有多种噪声信号，需要进行消噪处理，对原始测试数据利用小波变换进行消噪，对原始测试数据进行数据归一化操作，对发射特性曲线进行包络和延拓处理，通过降维操作将高维向量映射到低维空间；Data preprocessing: The test results of the present invention contain a variety of noise signals, which need to be de-noised. The original test data is de-noised by using wavelet transform, the data normalization operation is performed on the original test data, and the emission characteristic curve is packaged. Network and continuation processing, which maps high-dimensional vectors to low-dimensional spaces through dimensionality reduction operations;
特征提取：现场电磁兼容测试结果具有数据记录，它已经是数值化的对象，本身就是一种特征；提取采集信号的特征，包括峰值特征、包络特征、谐波特征，采用傅里叶变换对周期性信号和振铃信号进行频域特征提取；采用快速傅里叶变换对离散信号进行频域特征提取；采用短时傅里叶变换对脉冲信号进行频谱特征提取；采用小波变换对非平稳信号进行频域特征提取；Feature extraction: The on-site electromagnetic compatibility test results have data records, which are already numerical objects, and are a feature in themselves; extract the features of the collected signal, including peak features, envelope features, and harmonic features, and use Fourier transform to analyze them. Frequency domain feature extraction for periodic signals and ringing signals; fast Fourier transform for discrete signals for frequency domain feature extraction; short-time Fourier transform for pulse signal spectral feature extraction; wavelet transform for non-stationary signals Perform frequency domain feature extraction;
特征选择：特征选择包括两个内容，第一是选择训练样本的统计特征评估算法对特征进行评估，计算每一个特征对分类效果的贡献值；第二是利用主成分分析算法挑选出部分属性，针对不同的电磁兼容测试对象，可以选择不同的特征作为辨识对象，如电台类的对象特征选择峰值特征和谐波特征较为合适，电源类的对象特征更合适选择包络特征；从当前的向量属性集合中选择区分能力强的子集，剔除了干扰分类效果的冗余特征或无效特征，形成一个紧凑的特征集；Feature selection: Feature selection includes two contents. The first is to select the statistical feature evaluation algorithm of the training samples to evaluate the features, and to calculate the contribution of each feature to the classification effect; the second is to use the principal component analysis algorithm to select some attributes. For different electromagnetic compatibility test objects, different features can be selected as identification objects. For example, it is more suitable to select the peak feature and harmonic feature for the object feature of the radio class, and the envelope feature is more suitable for the object feature of the power supply class; from the current vector attribute Select a subset with strong discriminating ability in the set, eliminate redundant features or invalid features that interfere with the classification effect, and form a compact feature set;
学习和训练：利用基于卷积神经网络的深度学习算法，学习采集信号特征与干扰源之间的关系，得到基于干扰源识别的分类模型；Learning and training: Use the deep learning algorithm based on convolutional neural network to learn the relationship between the characteristics of the collected signal and the interference source, and obtain a classification model based on the identification of the interference source;
卷积的数学形式如下：The mathematical form of convolution is as follows:
y＝f(∑x*wij+b)y=f(∑x*wij +b)
式中，x是卷积层输入特征图；y是卷积层输出特征图；wij是二维卷积核；b是偏置项；f(·)是激活函数，如sigmoid函数或tanh函数或ReLU函数；where x is the input feature map of the convolution layer; y is the output feature map of the convolution layer; wij is the two-dimensional convolution kernel; b is the bias term; f( ) is the activation function, such as the sigmoid function or the tanh function or ReLU function;
卷积神经网络使用多个过滤器、多个卷积层、多个池化层，最后连接一个全连接层和一个softmax层，最终输出对应输入在每一类输出上的概率分布；The convolutional neural network uses multiple filters, multiple convolution layers, multiple pooling layers, and finally connects a fully connected layer and a softmax layer, and finally outputs the probability distribution corresponding to the input on each type of output;
整个网络的输出为The output of the entire network is
o＝f(n-1)(f(n-2)(...f(1)(x)))o=f(n-1) (f(n-2) (...f(1) (x)))
式中，x是卷积层输入特征图；f(·)是激活函数，如sigmoid函数或tanh函数或ReLU函数；In the formula, x is the input feature map of the convolution layer; f( ) is the activation function, such as the sigmoid function or the tanh function or the ReLU function;
根据类别属性已知的样本集与其向量取值的对应关系，形成对应模式分类方法的一系列参数；According to the corresponding relationship between the known sample set of category attribute and its vector value, a series of parameters corresponding to the pattern classification method are formed;
分类识别：对测试信号提取特征，采用步骤4-4得到的分类模型对类别属性未知的测试数据集进行干扰源分类，如果测试数据集中的类别属性已知，这个过程就可以用来验证分类器的分类效果，如果分类效果理想，该分类模型就可以运用到实际流量分类中。Classification and identification: Extract features from the test signal, and use the classification model obtained in steps 4-4 to classify the sources of interference in the test data set with unknown class attributes. If the class attributes in the test data set are known, this process can be used to verify the classifier. If the classification effect is ideal, the classification model can be applied to the actual traffic classification.
本发明在实际应用场景中，频谱监测和干扰源定位系统采可用B/S架构，每个小组可分为多个小型监测站、移动监测车通过网络和服务器通信，多个小组组成多任务模式，每个小组可以执行不同的任务，监测控制系统软件平台支持多个用户操作，在后期扩展用户可以通过手持终端连接到监测控制系统，进行测试项的指令下发，可以指定某个小组进行干扰定位测试或者监测频谱等测试，用户也可以直接在内部网络通过浏览器登陆账号进行指令下发，图5描述了本发明基于多用户、多任务模式实际应用下的数据流图。In the actual application scenario of the present invention, the spectrum monitoring and interference source positioning system adopts the available B/S architecture, each group can be divided into multiple small monitoring stations, the mobile monitoring vehicle communicates with the server through the network, and multiple groups form a multi-task mode , each group can perform different tasks, and the monitoring and control system software platform supports multiple user operations. In the later expansion, users can connect to the monitoring and control system through a handheld terminal to issue instructions for test items, and can designate a group to interfere For testing such as positioning test or monitoring spectrum, the user can also directly log in to the account through the browser on the internal network to issue instructions.
以上所述仅为本发明的较佳实施方式，本发明的保护范围并不以上述实施方式为限，但凡本领域普通技术人员根据本发明所揭示内容所作的等效修饰或变化，皆应纳入权利要求书中记载的保护范围内。The above descriptions are only the preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, but any equivalent modifications or changes made by those of ordinary skill in the art based on the contents disclosed in the present invention should be included in the within the scope of protection described in the claims.
Claims (3)
Translated fromChinesePriority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010297105.3ACN111585671B (en) | 2020-04-15 | 2020-04-15 | Electromagnetic interference monitoring and identification method of power LTE wireless private network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010297105.3ACN111585671B (en) | 2020-04-15 | 2020-04-15 | Electromagnetic interference monitoring and identification method of power LTE wireless private network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111585671A CN111585671A (en) | 2020-08-25 |
| CN111585671Btrue CN111585671B (en) | 2022-06-10 |
Family
ID=72111536
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010297105.3AExpired - Fee RelatedCN111585671B (en) | 2020-04-15 | 2020-04-15 | Electromagnetic interference monitoring and identification method of power LTE wireless private network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111585671B (en) |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112165365B (en)* | 2020-09-29 | 2022-05-10 | 封开低频时码授时台 | A wireless spectrum interference monitoring system and method for a low frequency time code monitoring system |
| CN112365186A (en)* | 2020-11-27 | 2021-02-12 | 中国电建集团海外投资有限公司 | Health degree evaluation method and system for electric power information system |
| CN113114398B (en)* | 2021-04-14 | 2022-01-28 | 中南大学 | Communication interference existence detection method based on residual connection and dilation convolution |
| CN115314134A (en)* | 2021-05-08 | 2022-11-08 | 深圳市万普拉斯科技有限公司 | Wi-Fi signal self-adaptive control method, system and mobile terminal |
| CN113569742B (en)* | 2021-07-29 | 2023-04-07 | 西南交通大学 | Broadband electromagnetic interference source identification method based on convolutional neural network |
| CN114070679B (en)* | 2021-10-25 | 2023-05-23 | 中国电子科技集团公司第二十九研究所 | Pulse intelligent classification-oriented frequency-phase characteristic analysis method |
| CN113868483B (en)* | 2021-12-02 | 2022-03-25 | 国网江苏省电力有限公司苏州供电分公司 | A Forensic Analysis Method of Wireless Spectrum Interference Based on Alliance Chain |
| CN114325258B (en)* | 2021-12-03 | 2022-07-19 | 西南交通大学 | A Transformer Bushing Insulation Evaluation Method Considering Multiple Resonant Frequency |
| CN114839466B (en)* | 2022-05-24 | 2023-02-28 | 温岭市天泰电气有限公司 | EMC electromagnetic compatibility test system for water pump and test method thereof |
| CN115758106A (en)* | 2022-10-08 | 2023-03-07 | 中国科学院国家天文台 | A Method of Extraction and Statistical Analysis of Radio Astronomy Interference Signals |
| CN116125172B (en)* | 2022-12-30 | 2023-08-11 | 北京长鹰恒容电磁科技有限公司 | Electromagnetic environment data acquisition and feature analysis method |
| CN115842597B (en)* | 2023-02-09 | 2023-05-02 | 成都华乾科技有限公司 | Black broadcast detection method and system based on multi-source data fusion |
| CN116056244B (en)* | 2023-03-07 | 2023-08-25 | 浙江万胜智能科技股份有限公司 | Public network wireless communication resource scheduling method and system based on remote module |
| CN117559673B (en)* | 2024-01-11 | 2024-03-26 | 希格玛电气(珠海)有限公司 | Wireless sensing system based on magnetic resonance energy supply and ring main unit |
| CN118337308B (en)* | 2024-06-11 | 2024-08-09 | 武汉海华信通科技有限公司 | Positioning method and system based on power communication radio interference source |
| CN118869122B (en)* | 2024-09-24 | 2024-12-03 | 国能榆林能源有限责任公司 | A communication performance testing method and system for underground charging and discharging facilities |
| CN119154972A (en)* | 2024-11-11 | 2024-12-17 | 深圳市瑞迅通信息技术有限公司 | Spectrum analysis device, method and storage medium for radio monitoring system |
| CN119583003A (en)* | 2024-12-13 | 2025-03-07 | 中国人民解放军国防科技大学 | Anti-electromagnetic interference device and method for short-range unmanned aerial vehicles |
| CN119519663B (en)* | 2025-01-22 | 2025-06-27 | 珠海新立电子科技有限公司 | Anti-interference signal processing method of FPC (Flexible printed Circuit) sensing module and related equipment |
| CN120320871B (en)* | 2025-06-16 | 2025-08-19 | 南京博网软件科技有限公司 | Spectrum monitoring and interference analysis method for power grid wireless communication |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en)* | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN106911603A (en)* | 2017-03-07 | 2017-06-30 | 北京工业大学 | A kind of broadband monitoring pattern Imitating signal modulation style real-time identification method |
| CN108135003A (en)* | 2017-12-25 | 2018-06-08 | 广东海格怡创科技有限公司 | The construction method and system of interference type identification model |
| CN108509911A (en)* | 2018-04-03 | 2018-09-07 | 电子科技大学 | Interference signal recognition methods based on convolutional neural networks |
| CN110197127A (en)* | 2019-05-06 | 2019-09-03 | 安徽继远软件有限公司 | Wireless signal detection and electromagnetic interference categorizing system and method based on deep learning |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103812577B (en)* | 2012-11-06 | 2017-10-13 | 西南交通大学 | The automatic recognition system and its method of improper radio signal |
| WO2017127421A1 (en)* | 2016-01-18 | 2017-07-27 | Qoscience, Inc. | Method and apparatus for the detection of distortion or corruption of cellular communication signals |
- 2020
- 2020-04-15CNCN202010297105.3Apatent/CN111585671B/ennot_activeExpired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722759A (en)* | 2012-05-17 | 2012-10-10 | 河海大学 | Method for predicting power supply reliability of power grid based on BP neural network |
| CN106911603A (en)* | 2017-03-07 | 2017-06-30 | 北京工业大学 | A kind of broadband monitoring pattern Imitating signal modulation style real-time identification method |
| CN108135003A (en)* | 2017-12-25 | 2018-06-08 | 广东海格怡创科技有限公司 | The construction method and system of interference type identification model |
| CN108509911A (en)* | 2018-04-03 | 2018-09-07 | 电子科技大学 | Interference signal recognition methods based on convolutional neural networks |
| CN110197127A (en)* | 2019-05-06 | 2019-09-03 | 安徽继远软件有限公司 | Wireless signal detection and electromagnetic interference categorizing system and method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111585671A (en) | 2020-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111585671B (en) | Electromagnetic interference monitoring and identification method of power LTE wireless private network | |
| Weidling et al. | A framework for RF spectrum measurements and analysis | |
| CN101334434B (en) | Electromagnetic environment test system for extracting electromagnetic leakage signal by utilizing wavelet transformation | |
| CN111428561B (en) | A broadband electromagnetic interference source identification method and system, storage medium and equipment | |
| CN112183339A (en) | Electromagnetic environment assessment method, device and server | |
| CN114584227B (en) | Automatic burst signal detection method | |
| Li | A New Robust Signal Recognition Approach Based on Holder Cloud Features under Varying SNR Environment. | |
| CN105842561A (en) | Background signal eliminating method suitable for on-site electromagnetic interference detection | |
| Huang et al. | FM broadcast monitoring using artificial intelligence | |
| CN117768278A (en) | Modulation identification method based on hybrid complex neural network | |
| CN118611784B (en) | A radio frequency link calibration method for a mobile sensing terminal | |
| CN106100762A (en) | A kind of weak signal of communication detection method of cyclo-stationary analysis of spectrum | |
| CN120336703A (en) | A broadband signal parameter identification method based on wavelet denoising and improved VMD-Prony algorithm | |
| Xu et al. | Individual radio transmitter identification based on spurious modulation characteristics of signal envelop | |
| CN106556760A (en) | A kind of automated test control system of the vehicle mounted multimedia based on cloud computing | |
| Pan et al. | Classification of Power Quality Disturbances Using Resnet with Channel Attention Mechanism | |
| CN101806834A (en) | Kalman filter-based signal real-time time-frequency spectrometer | |
| CN113300986A (en) | Unmanned aerial vehicle image transmission signal and hotspot signal identification method, medium and computer equipment | |
| CN116125178A (en) | Comprehensive determination method and system for electromagnetic radiation blocking effect of frequency-using equipment | |
| CN116707675A (en) | Radio signal detection, radio signal abnormality detection method and device | |
| CN113518049B (en) | Modulation identification method based on fractional low-order polar coordinates and deep learning | |
| Liu et al. | A novel signal separation algorithm for wideband spectrum sensing in cognitive networks | |
| CN114401055A (en) | Intelligent frequency spectrum detection system | |
| Le Truong et al. | An Application of Deep Learning YOLOv5 Framework to Intelligent Radio Spectrum Monitoring | |
| Charalambous et al. | Modeling wireless fading channels via stochastic differential equations: Identification and estimation based on noisy measurements |
Legal Events
| 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | Granted publication date:20220610 |