CN111060815A - GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method - Google Patents
GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method Download PDFInfo
- Publication number
- CN111060815A CN111060815A CN201911303417.4A CN201911303417A CN111060815A CN 111060815 A CN111060815 A CN 111060815A CN 201911303417 A CN201911303417 A CN 201911303417A CN 111060815 A CN111060815 A CN 111060815A
- Authority
- CN
- China
- Prior art keywords
- layer
- input
- hidden layer
- circuit breaker
- output
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000003745 diagnosis Methods 0.000 title claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 25
- 238000012544 monitoring process Methods 0.000 claims abstract description 16
- 238000005457 optimization Methods 0.000 claims abstract description 9
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000013528 artificial neural network Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 21
- 230000004913 activation Effects 0.000 claims description 16
- 230000002068 genetic effect Effects 0.000 claims description 13
- 230000002457 bidirectional effect Effects 0.000 claims description 12
- 230000000306 recurrent effect Effects 0.000 claims description 11
- 230000035772 mutation Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000013473 artificial intelligence Methods 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 3
- 238000010353 genetic engineering Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 22
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical group [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 11
- 230000007246 mechanism Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 125000004122 cyclic group Chemical group 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000004907 flux Effects 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- 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
- G01R31/327—Testing of circuit interrupters, switches or circuit-breakers
- G01R31/3271—Testing of circuit interrupters, switches or circuit-breakers of high voltage or medium voltage devices
- G01R31/3275—Fault detection or status indication
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
本发明公开了基于GA‑Bi‑RNN的高压断路器故障诊断方法，具体为：S1、利用分合闸线圈电流在线监测系统实时监测得到分合闸线圈电流数据，并将该数据分为训练集和测试集一同作为输入变量；S2、初始化权值，将训练集样本数据输入至Bi‑RNN中，采用GA作为误差反传优化更新每一代的特征信息参量，并将其作为输入，以均方误差作为适应度，以一定迭代次数为模型终止条件，选择预测特征量最优组合，完成模型训练；S3、将得到的测试集样本数据输入至训练好的故障诊断模型中，由故障诊断模型对输入的分合闸线圈电流数据进行处理，完成高压断路器故障诊断分类。本方法能更加准确有效地判断断路器的故障类型，进而有效率的完成检修。
The invention discloses a fault diagnosis method for a high-voltage circuit breaker based on GA-Bi-RNN, which comprises the following steps: S1. Real-time monitoring is used to obtain the current data of the opening and closing coil by using an online monitoring system of the opening and closing coil current, and the data is divided into a training set and the test set as input variables; S2, initialize the weights, input the training set sample data into Bi-RNN, use GA as the error back-propagation optimization to update the feature information parameters of each generation, and use it as the input, with the mean square The error is used as the fitness, and a certain number of iterations is used as the termination condition of the model, and the optimal combination of predicted feature quantities is selected to complete the model training; S3. Input the obtained test set sample data into the trained fault diagnosis model, and the fault diagnosis model will analyze the data. The input current data of the opening and closing coil is processed to complete the fault diagnosis and classification of the high-voltage circuit breaker. The method can more accurately and effectively judge the fault type of the circuit breaker, and then complete the maintenance efficiently.
Description
技术领域technical field
本发明属于高压断路器故障在线监测技术领域，具体涉及一种基于GA-Bi-RNN的高压断路器故障诊断方法。The invention belongs to the technical field of high-voltage circuit breaker fault online monitoring, and in particular relates to a high-voltage circuit breaker fault diagnosis method based on GA-Bi-RNN.
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。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。分合闸线圈电流可作为高压断路器机械故障诊断所用的丰富信息。The high-voltage circuit breaker is the most important control and protection device in the power system, which is related to the reliability and safety of power transmission, distribution and power consumption. High-voltage circuit breakers can perform a variety of operations under system fault and non-fault conditions. The circuit breaker can also close, carry and break the normal current of the running circuit, and can also close, carry and break the specified overload current within the specified time. High-voltage circuit breakers generally use electromagnets as the first control element for operation, and most of the operating mechanisms are DC electromagnets. When a current passes through the coil, a magnetic flux is generated in the magnet, and the moving iron core is affected by the magnetic force, which makes the circuit breaker open or close. The opening and closing coil currents can be used as rich information for the mechanical fault diagnosis of high-voltage circuit breakers.
现有的高压断路器故障诊断的方法有很多，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 fault diagnosis of 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 its membership functions and fuzzy rules are determined. There are certain human factors in the process; the radial basis neural network provides a better structural system for the fault diagnosis of circuit breakers, but there are reasons that cannot explain its own reasoning process and reasoning basis, and the neural network cannot work normally when the data is insufficient. Disadvantages of work.
双向循环神经网络(Bi-RNN)是一种对数据序列建模的神经网络，其处理方式与前馈神经网络有着本质上的不同，双向循环神经网络只处理一个单一的输入单元和上一个时间点的隐藏层信息。这使得双向循环神经网络能够更加自由和动态的获取输入的信息，而不受到定长输入空间的限制，具有良好的容错能力、并行处理能力和自学习能力。但是由于其学习过程稍过单一，训练过程中可能会存在训练不完整的缺陷；因此，利用遗传算法(GA)来优化双向循环神经网络可以解决这一问题，并且使其权重更新直到设定误差范围之内，能在有效解决上述问题的同时，更加准确快速的对故障进行分类。Bi-directional recurrent neural network (Bi-RNN) is a neural network that models data sequences. Its processing method is fundamentally different from feed-forward neural networks. Bi-directional recurrent neural networks only deal with a single input unit and the previous time. The hidden layer information of the point. This enables the bidirectional recurrent neural network to obtain the input information more freely and dynamically without being restricted by the fixed-length input space, and has good fault tolerance, parallel processing and self-learning capabilities. However, due to its single learning process, there may be defects of incomplete training during the training process; therefore, the use of genetic algorithm (GA) to optimize the bidirectional recurrent neural network can solve this problem, and its weights are updated until the setting error Within the scope, it can effectively solve the above problems and classify the faults more accurately and quickly.
发明内容SUMMARY OF THE INVENTION
本发明的目的是提供一种基于GA-Bi-RNN的高压断路器故障诊断方法，采用双向循环神经网络分析故障特征信号，结合遗传算法进行参数优化，在弥补人工神经网络诊断不足的同时，能更加准确有效地判断断路器的故障类型，进而有效率的完成检修。The purpose of the present invention is to provide a fault diagnosis method for high-voltage circuit breakers based on GA-Bi-RNN, which adopts bidirectional cyclic neural network to analyze fault characteristic signals, and combines genetic algorithm for parameter optimization. It can more accurately and effectively judge the fault type of the circuit breaker, and then complete the maintenance efficiently.
本发明所采用的技术方案是，基于GA-Bi-RNN的高压断路器故障诊断方法，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 diagnosis method based on GA-Bi-RNN, which is specifically implemented according to the following steps:
步骤1、利用分合闸线圈电流在线监测系统实时监测得到分合闸线圈电流数据，并将该数据分为训练集和测试集一同作为输入变量；Step 1. Real-time monitoring of the opening and closing coil current online monitoring system is used to obtain the opening and closing coil current data, and the data is divided into a training set and a test set together as input variables;
步骤2、初始化权值，将训练集样本数据输入至Bi-RNN中，采用GA作为误差反传优化更新每一代的特征信息参量，并将其作为输入，以均方误差作为适应度，以一定迭代次数为模型终止条件，选择预测特征量最优组合，完成模型训练；Step 2. Initialize the weights, input the training set sample data into Bi-RNN, use GA as the error back-propagation to optimize and update the feature information parameters of each generation, and use it as the input, take the mean square error as the fitness, and use the mean square error as the fitness. The number of iterations is the termination condition of the model, and the optimal combination of predicted feature quantities is selected to complete the model training;
步骤3、将经步骤1得到的测试集样本数据输入至经步骤2训练好的故障诊断模型中，由故障诊断模型对输入的分合闸线圈电流数据进行处理，完成高压断路器故障诊断分类。Step 3: Input the test set sample data obtained in step 1 into the fault diagnosis model trained in step 2, and the fault diagnosis model processes the input current data of the opening and closing coils to complete the fault diagnosis and classification of the high-voltage circuit breaker.
本发明的特点还在于，The present invention is also characterized in that,
步骤1中，所述分合闸线圈电流在线监测系统，包括过程层、间隔层和站控层；In step 1, the on-line monitoring system for the opening and closing coil current includes a process layer, an interval layer and a station control layer;
过程层采集提取高压断路器分合闸线圈电流的特征信息参数，利用在线监测系统对所采集的数据进行预处理并完成特征信息参量的分析计算；所述间隔层由变电站断路器IED和以太网组成，该间隔层是将经过过程层监测并处理的特征信息参量利用CAN总线和变电站断路器IED之间的通信传送至变电站断路器IED，再通过以太网采用IEC 61850系列标准协议将数据上传至站控层监控中心；所述站控层是对站内的设备进行远程监控，并且接收由间隔层传输的特征信息参量，结合人工智能神经网络对该断路器进行实时故障诊断。The process layer collects and extracts the characteristic information parameters of the opening and closing coil current of the high-voltage circuit breaker, uses the online monitoring system to preprocess the collected data and completes the analysis and calculation of the characteristic information parameters; the interval layer consists of the substation circuit breaker IED and Ethernet The partition layer is to transmit the characteristic information parameters monitored and processed by the process layer to the substation circuit breaker IED using the communication between the CAN bus and the substation circuit breaker IED, and then upload the data to the substation circuit breaker IED through the Ethernet using the IEC 61850 series standard protocol. The monitoring center of the station control layer; the station control layer is to remotely monitor the equipment in the station, and to receive the characteristic information parameters transmitted by the bay layer, and to carry out real-time fault diagnosis of the circuit breaker in combination with the artificial intelligence neural network.
步骤2具体按照以下方法实施：Step 2 is implemented as follows:
步骤2.1、初始化权值，将所有权值初始化为一个随机数[0,1]；Step 2.1, initialize the weight value, and initialize the ownership value to a random number [0,1];
步骤2.2、经步骤2.1后，从训练集中提取一个样例X，并将该样例X输入到双向循环神经网络中，并给出它的目标输出向量，并将其记作O；Step 2.2. After step 2.1, extract an example X from the training set, and input the example X into the bidirectional recurrent neural network, and give its target output vector, and denote it as O;
输入层的输入与隐藏层的输出之间存在下列函数关系：The following functional relationship exists between the input of the input layer and the output of the hidden layer:
其中，
其中，f()为sigmoid函数：Among them, f() is the sigmoid function:
隐含层的输出S(t)与输出层的输出O(t)之间存在下列函数关系：There is the following functional relationship between the output S(t) of the hidden layer and the output O(t) of the output layer:
O(t)＝g(YS(t)) (7)O(t)=g(YS(t)) (7)
其中，Y是隐含层连接到输出层的权重矩阵，g()为softmax函数:where Y is the weight matrix connecting the hidden layer to the output layer, and g() is the softmax function:
其中，x为隐含层输入值，i为隐含层节点个数，随机生成权重矩阵
步骤2.3、经步骤2.2后，从前层向后层依次计算，得到双向循环神经网络的输出值O(t)，其中对于隐含层某一时刻某一节点的激活函数netj(t)用公式表示：Step 2.3. After step 2.2, calculate sequentially from the front layer to the back layer to obtain the output value O(t) of the bidirectional recurrent neural network, where the activation function netj(t) of a node at a certain moment in the hidden layer is expressed by the formula :
其中，n表示输入层节点的个数，i(t)表示t时刻隐含层节点个数，Vji表示该时刻该节点所在层连接后层的权重矩阵，θj表示一个偏置参数，更新隐含层节点激活函数的计算方式：Among them, n represents the number of nodes in the input layer, i(t) represents the number of nodes in the hidden layer at time t, V ji represents the weight matrix of the layer after the connection of the layer where the node is located, θ j represents a bias parameter, update The calculation method of the activation function of the hidden layer node:
hj(t)＝f(netj(t)) (11)hj(t)=f(netj(t)) (11)
其中，m表示的是隐含层节点的总个数，l(t-1)表示t-1时刻隐含层节点，Vjl表示该时刻该节点所在层连接后层的权重矩阵；hj(t)表示上一次更新隐含层节点的激活函数；Among them, m represents the total number of hidden layer nodes, l (t-1) represents the hidden layer node at time t-1, V jl represents the weight matrix of the layer after the connection of the layer where the node is located at this moment; hj(t ) represents the activation function of the last updated hidden layer node;
输出层的激活函数netk(t)：The activation function netk(t) of the output layer:
yk(t)＝g(netk(t)) (13)yk(t)=g(netk(t)) (13)
其中，j(t)表示t时刻隐含层节点，θk表示一个偏置参数，Wkj表示该时刻该节点所在层连接后层的权重矩阵，yk(t)表示的是输出层节点的激活函数；Among them, j(t) represents the hidden layer node at time t, θ k represents a bias parameter, W kj represents the weight matrix of the layer after the node is connected at this moment, and yk(t) represents the activation of the output layer node function;
步骤2.4、经步骤2.3后，采用遗传算法作为误差反传优化，将优化后每一代更新的特征信息参量作为输入，以均方误差作为适应度，以一定的迭代次数为模型终止条件，选择预测特征量最优组合。Step 2.4. After step 2.3, the genetic algorithm is used as the error back propagation optimization, the characteristic information parameters updated in each generation after optimization are used as input, the mean square error is used as the fitness, and a certain number of iterations is used as the model termination condition, and the prediction is selected. The optimal combination of feature quantities.
步骤2.4中，遗传误差反传的具体过程如下：In step 2.4, the specific process of genetic error back propagation is as follows:
一个标准的遗传算法为SCA＝(C,E,P0,M,Φ,δ,ψ,T)，其中，C是GA编码方法，E是GA的适应度函数，P0是初始种群，M是种群大小，Φ是选择操作，δ是GA的交叉操作，ψ是GA的变异操作，T是GA的终止操作条件；以防进入局部最优；A standard genetic algorithm is SCA=(C, E, P 0 , M, Φ, δ, ψ, T), where C is the GA coding method, E is the fitness function of GA, P 0 is the initial population, M is the population size, Φ is the selection operation, δ is the crossover operation of GA, ψ is the mutation operation of GA, T is the termination operation condition of GA; to prevent entering the local optimum;
(a)、编码：(a), coding:
根据所求精度将采用11位二进制数对连接权和阈值进行编码，其中第1位为符号位，其余10位编码的对应关系为：According to the required precision, 11-bit binary numbers will be used to encode the connection weight and threshold, where the first bit is the sign bit, and the corresponding relationship of the remaining 10-bit codes is:
其中δ＝(1.0-0.0)/(210-1)＝0.00098；where δ=(1.0-0.0)/(210-1)=0.00098;
(b)、遗传操作：(b), genetic manipulation:
为了提高模型的运行速度和收敛能力，计算交叉率Pc和变异率Pm，具体如下：In order to improve the running speed and convergence ability of the model, the crossover rate P c and the mutation rate P m are calculated as follows:
式中，fmax为最大个体适应度，favg为平均个体适应度，f′为执行交叉操作个体中的最大适应度，f是执行变异操作个体中的最大适应度；In the formula, f max is the maximum individual fitness, f avg is the average individual fitness, f' is the maximum fitness among individuals performing crossover operations, and f is the maximum fitness among individuals performing mutation operations;
(c)、目标函数(c), the objective function
利用模型的输出量与训练样本的期望输出的差之和的最小值作为目标函数，即Use the minimum value of the sum of the difference between the output of the model and the expected output of the training sample as the objective function, that is,
式中，YBi-RNN-GA为Bi-RNN模型的输出值，Ydata为训练样本的期望输出，N为样本个数；In the formula, Y Bi-RNN-GA is the output value of the Bi-RNN model, Y data is the expected output of the training sample, and N is the number of samples;
(d)、个体适应度(d), individual fitness
式中，Cmax选为种群最大的个体适应度。In the formula, Cmax is selected as the maximum individual fitness of the population.
本发明的有益效果是：The beneficial effects of the present invention are:
(1)本发明方法从优化神经网络模型和优化特征参数选取两个角度对模型故障诊断能力进行提升，双向循环神经网络作为深度神经网络模型，可以从原始数据提取到更加抽象和更具有代表性的特征，能够更加自由和动态的获取输入的信息，而不受到定长输入空间的限制，具有良好的容错能力、并行处理能力和自学习能力；(1) The method of the present invention improves the fault diagnosis capability of the model from the perspective of optimizing the neural network model and optimizing the selection of characteristic parameters. As a deep neural network model, the bidirectional recurrent neural network can be extracted from the original data to be more abstract and more representative. It can obtain the input information more freely and dynamically without being limited by the fixed-length input space, and has good fault tolerance, parallel processing and self-learning capabilities;
(2)本发明方法利用遗传算法优化连接权值，使其权重更新直到设定误差范围内，具有较好的局部与全局搜索能力，可以有效提高高压断路器故障诊断的速度与准确率；(2) The method of the present invention utilizes the genetic algorithm to optimize the connection weights, so that the weights are updated until the set error range, has better local and global search capabilities, and can effectively improve the speed and accuracy of fault diagnosis of high-voltage circuit breakers;
(3)本发明方法在双向循环网络训练过程中引入Dropout技术来防止过拟合，增强了模型的泛化能力。(3) The method of the present invention introduces Dropout technology in the training process of the bidirectional cyclic network to prevent overfitting and enhance the generalization ability of the model.
附图说明Description of drawings
图1是本发明基于GA-Bi-RNN的高压断路器故障诊断方法中采用的分合闸线圈电流在线监测系统的架构图；Fig. 1 is the framework diagram of the on-line monitoring system of opening and closing coil current adopted in the fault diagnosis method of high-voltage circuit breaker based on GA-Bi-RNN of the present invention;
图2是本发明基于GA-Bi-RNN的高压断路器故障诊断方法的流程图；Fig. 2 is the flow chart of the high-voltage circuit breaker fault diagnosis method based on GA-Bi-RNN of the present invention;
图3是本发明基于GA-Bi-RNN的高压断路器故障诊断方法的故障诊断模型；Fig. 3 is the fault diagnosis model of the high-voltage circuit breaker fault diagnosis method based on GA-Bi-RNN of the present invention;
图4是本发明方法中实施例1中涉及的分合闸线圈电流特性曲线。FIG. 4 is the current characteristic curve of the opening and closing coil involved in Embodiment 1 of the method of the present invention.
具体实施方式Detailed ways
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 with reference to the accompanying drawings and specific embodiments.
本发明提供一种基于GA-Bi-RNN的高压断路器故障诊断方法，如图1-3所示，具体按照以下步骤实施：The present invention provides a high-voltage circuit breaker fault diagnosis method based on GA-Bi-RNN, as shown in Figures 1-3, which is specifically implemented according to the following steps:
步骤1、利用分合闸线圈电流在线监测系统实时监测得到分合闸线圈电流数据，并将该数据分为训练集和测试集一同作为输入变量；Step 1. Real-time monitoring of the opening and closing coil current online monitoring system is used to obtain the opening and closing coil current data, and the data is divided into a training set and a test set together as input variables;
步骤1中，分合闸线圈电流在线监测系统，如图1所示，包括过程层、间隔层和站控层；过程层采集提取高压断路器分合闸线圈电流的特征信息参数，利用在线监测系统对所采集的数据进行预处理并完成特征信息参量的分析计算；间隔层由变电站断路器IED和以太网组成，该间隔层是将经过过程层监测并处理的特征信息参量利用CAN总线和变电站断路器IED之间的通信传送至变电站断路器IED，再通过以太网采用IEC 61850系列标准协议将数据上传至站控层监控中心；站控层是对站内的设备进行远程监控，并且接收由间隔层传输的特征信息参量，结合人工智能神经网络对该断路器进行实时故障诊断。In step 1, the opening and closing coil current online monitoring system, as shown in Figure 1, includes a process layer, an interval layer and a station control layer; the process layer collects and extracts the characteristic information parameters of the opening and closing coil current of the high-voltage circuit breaker, and uses the online monitoring The system preprocesses the collected data and completes the analysis and calculation of characteristic information parameters; the bay layer is composed of substation circuit breaker IED and Ethernet. The communication between the circuit breaker IEDs is transmitted to the substation circuit breaker IED, and then the data is uploaded to the monitoring center of the station control layer through the Ethernet using the IEC 61850 series standard protocol; The characteristic information parameters of layer transmission are combined with artificial intelligence neural network to carry out real-time fault diagnosis of the circuit breaker.
步骤2、初始化权值，将训练集样本数据输入至Bi-RNN中，采用GA作为误差反传优化更新每一代的特征信息参量，并将其作为输入，以均方误差作为适应度，以一定迭代次数为模型终止条件，选择预测特征量最优组合，完成模型训练；Step 2. Initialize the weights, input the training set sample data into Bi-RNN, use GA as the error back-propagation to optimize and update the feature information parameters of each generation, and use it as the input, take the mean square error as the fitness, and use the mean square error as the fitness. The number of iterations is the termination condition of the model, and the optimal combination of predicted feature quantities is selected to complete the model training;
步骤2具体按照以下方法实施：Step 2 is implemented as follows:
步骤2.1、初始化权值，将所有权值初始化为一个随机数[0,1]；Step 2.1, initialize the weight value, and initialize the ownership value to a random number [0,1];
步骤2.2、经步骤2.1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输入到双向循环神经网络中，并给出它的目标输出向量，并将其记作O；Step 2.2. After step 2.1, extract a sample X from the training set (taking the five sets of data in Example 1 as the training set), and input the sample X into the bidirectional recurrent neural network, and give its target output vector, and denote it as O;
输入层的输入与隐藏层的输出之间存在下列函数关系：The following functional relationship exists between the input of the input layer and the output of the hidden layer:
其中，
其中，f()为sigmoid函数：Among them, f() is the sigmoid function:
隐含层的输出S(t)与输出层的输出O(t)之间存在下列函数关系：There is the following functional relationship between the output S(t) of the hidden layer and the output O(t) of the output layer:
O(t)＝g(YS(t)) (7)O(t)=g(YS(t)) (7)
其中，Y是隐含层连接到输出层的权重矩阵，g()为softmax函数:where Y is the weight matrix connecting the hidden layer to the output layer, and g() is the softmax function:
其中，x为隐含层输入值，i为隐含层节点个数，随机生成权重矩阵
步骤2.3、经步骤2.2后，从前层向后层依次计算，得到双向循环神经网络的输出值O(t)，其中对于隐含层某一时刻某一节点的激活函数netj(t)用公式表示：Step 2.3. After step 2.2, calculate sequentially from the front layer to the back layer to obtain the output value O(t) of the bidirectional recurrent neural network, where the activation function netj(t) of a node at a certain moment in the hidden layer is expressed by the formula :
其中，n表示输入层节点的个数，i(t)表示t时刻隐含层节点个数，Vji表示该时刻该节点所在层连接后层的权重矩阵，θj表示一个偏置参数，更新隐含层节点激活函数的计算方式：Among them, n represents the number of nodes in the input layer, i(t) represents the number of nodes in the hidden layer at time t, V ji represents the weight matrix of the layer after the connection of the layer where the node is located, θ j represents a bias parameter, update The calculation method of the activation function of the hidden layer node:
hj(t)＝f(netj(t)) (11)hj(t)=f(netj(t)) (11)
其中，m表示的是隐含层节点的总个数，l(t-1)表示t-1时刻隐含层节点，Vjl表示该时刻该节点所在层连接后层的权重矩阵；hj(t)表示上一次更新隐含层节点的激活函数。Among them, m represents the total number of hidden layer nodes, l (t-1) represents the hidden layer node at time t-1, V jl represents the weight matrix of the layer after the connection of the layer where the node is located at this moment; hj(t ) represents the activation function of the last updated hidden layer node.
输出层的激活函数netk(t)：The activation function netk(t) of the output layer:
yk(t)＝g(netk(t)) (13)yk(t)=g(netk(t)) (13)
其中，j(t)表示t时刻隐含层节点，θk表示一个偏置参数，Wkj表示该时刻该节点所在层连接后层的权重矩阵，yk(t)表示的是输出层节点的激活函数(可以与隐含层节点的h是同一个激活函数)。Among them, j(t) represents the hidden layer node at time t, θ k represents a bias parameter, W kj represents the weight matrix of the layer after the node is connected at this moment, and yk(t) represents the activation of the output layer node function (can be the same activation function as h of the hidden layer node).
步骤2.4、经步骤2.3后，采用遗传算法作为误差反传优化，将优化后每一代更新的特征信息参量作为输入，以均方误差作为适应度，以一定的迭代次数为模型终止条件，选择预测特征量最优组合。Step 2.4. After step 2.3, the genetic algorithm is used as the error back propagation optimization, the characteristic information parameters updated in each generation after optimization are used as input, the mean square error is used as the fitness, and a certain number of iterations is used as the model termination condition, and the prediction is selected. The optimal combination of feature quantities.
步骤2.4中，遗传误差反传的具体过程如下：In step 2.4, the specific process of genetic error back propagation is as follows:
一个标准的遗传算法为SCA＝(C,E,P0,M,Φ,δ,ψ,T)，其中，C是GA编码方法，E是GA的适应度函数，P0是初始种群，M是种群大小，Φ是选择操作，δ是GA的交叉操作，ψ是GA的变异操作，T是GA的终止操作条件；以防进入局部最优；A standard genetic algorithm is SCA=(C, E, P 0 , M, Φ, δ, ψ, T), where C is the GA coding method, E is the fitness function of GA, P 0 is the initial population, M is the population size, Φ is the selection operation, δ is the crossover operation of GA, ψ is the mutation operation of GA, T is the termination operation condition of GA; to prevent entering the local optimum;
(a)、编码：(a), coding:
根据所求精度将采用11位二进制数对连接权和阈值进行编码，其中第1位为符号位，其余10位编码的对应关系为：According to the required precision, 11-bit binary numbers will be used to encode the connection weight and threshold, where the first bit is the sign bit, and the corresponding relationship of the remaining 10-bit codes is:
其中δ＝(1.0-0.0)/(210-1)＝0.00098；where δ=(1.0-0.0)/(210-1)=0.00098;
(b)、遗传操作：(b), genetic manipulation:
为了提高模型的运行速度和收敛能力，计算交叉率Pc和变异率Pm，具体如下：In order to improve the running speed and convergence ability of the model, the crossover rate P c and the mutation rate P m are calculated as follows:
式中，fmax为最大个体适应度，favg为平均个体适应度，f′为执行交叉操作个体中的最大适应度，f是执行变异操作个体中的最大适应度；In the formula, f max is the maximum individual fitness, f avg is the average individual fitness, f' is the maximum fitness among individuals performing crossover operations, and f is the maximum fitness among individuals performing mutation operations;
(c)、目标函数(c), the objective function
利用模型的输出量与训练样本的期望输出的差之和的最小值作为目标函数，即Use the minimum value of the sum of the difference between the output of the model and the expected output of the training sample as the objective function, that is,
式中，YBi-RNN-GA为Bi-RNN模型的输出值，Ydata为训练样本的期望输出，N为样本个数；In the formula, Y Bi-RNN-GA is the output value of the Bi-RNN model, Y data is the expected output of the training sample, and N is the number of samples;
(d)、个体适应度(d), individual fitness
式中，Cmax选为种群最大的个体适应度。In the formula, Cmax is selected as the maximum individual fitness of the population.
步骤3、将经步骤1得到的测试集样本数据输入至经步骤2训练好的故障诊断模型中，由故障诊断模型对输入的分合闸线圈电流数据进行处理，完成高压断路器故障诊断分类。Step 3: Input the test set sample data obtained in step 1 into the fault diagnosis model trained in step 2, and the fault diagnosis model processes the input current data of the opening and closing coils to complete the fault diagnosis and classification of the high-voltage circuit breaker.
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 t 0 as the zero point of the command time to extract the fault characteristic parameters I 1 , I 2 , I 3 , t 1 , t 2 , t 3 , t 4 , t 5 to monitor the state of the circuit breaker, and obtain ten groups of fault sample data. Ten groups of fault sample data include normal mechanism (A), operating voltage is too low (B), the closing iron core is jammed at the beginning stage (C), the operating mechanism is jammed (D), and the closing iron core has a large empty stroke (E). ), the data collection status is shown in Table 1;
表1故障样本数据Table 1 Fault sample data
分合闸线圈电流的特性曲线如图4所示，可知：The characteristic curve of the opening and closing 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) Stage I, t=t 0 ~ t 1 ; the coil starts to be energized at time t 0 , and the iron core starts to move at time t 1 ; t 0 is the moment when the circuit breaker opening and closing commands are issued, which is the opening and closing action of the circuit breaker Timing starting point; t 1 is the time when 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; this stage is characterized by an exponential rise in the current and a static state of the iron core; the time at this stage is related to the control power supply voltage and 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=t 1 ~ t 2 ; in this stage, the iron core starts to move, and the current drops; t 2 is the valley point of the control current, which means that the iron core has touched the load of the operating machine and significantly decelerated 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=t 2 ~ t 3 ; in this stage, the iron core stops moving, and the current increases exponentially.
(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=t 3 ~ t 4 ; 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=t 4 ~ t 5 ; in the current breaking stage, the auxiliary switch is broken at this stage, an arc is generated between the auxiliary switch contacts and is elongated, and the arc voltage rises rapidly, forcing the current to rapidly decrease, until it goes out.
分析图4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)。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 in Fig. 4 shows that the current from t 0 to t 1 can reflect the state of the coil (for example, whether the resistance is normal). The change of current from t 1 to t 2 indicates whether the iron core motion structure is jammed or not, and the mechanical load of tripping and energy releasing changes; t 2 is generally the moment when the moving contact starts to move, and after t 2 , the mechanism is driven by the transmission system. The process of opening and closing the moving contact, that is, the moving process of the moving contact; t 4 is the moment when the auxiliary contact of the circuit breaker is cut off; the change of current from t 0 to t 4 can reflect the work of the transmission system of the mechanical operating mechanism Happening.
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故障类型输出设定Table 2 Fault type output settings
本发明提供的一种基于GA-Bi-RNN的高压断路器故障诊断方法，采用双向循环神经网络分析故障特征信号，结合遗传算法进行参数优化，在弥补人工神经网络诊断不足的同时，能更加准确迅速的判断断路器故障类型。The present invention provides a fault diagnosis method for high-voltage circuit breakers based on GA-Bi-RNN, which adopts bidirectional cyclic neural network to analyze fault characteristic signals, and combines genetic algorithm for parameter optimization, which not only makes up for the shortage of artificial neural network in diagnosis, but also can be more accurate. Quickly determine the type of circuit breaker failure.
Claims (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911303417.4A CN111060815B (en) | 2019-12-17 | 2019-12-17 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911303417.4A CN111060815B (en) | 2019-12-17 | 2019-12-17 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111060815A true CN111060815A (en) | 2020-04-24 |
| CN111060815B CN111060815B (en) | 2021-09-14 |
Family
ID=70302107
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911303417.4A Active CN111060815B (en) | 2019-12-17 | 2019-12-17 | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111060815B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN112729411A (en) * | 2021-01-14 | 2021-04-30 | 金陵科技学院 | Distributed drug warehouse environment monitoring method based on GA-RNN |
| CN113358157A (en) * | 2021-06-10 | 2021-09-07 | 国网甘肃省电力公司兰州供电公司 | RST-PNN-GA-based power equipment temperature rise detection and early warning method |
| CN113469222A (en) * | 2021-06-10 | 2021-10-01 | 国网甘肃省电力公司兰州供电公司 | RST-PNN-GA-based high-voltage circuit breaker fault detection method |
| CN115389812A (en) * | 2022-10-28 | 2022-11-25 | 国网信息通信产业集团有限公司 | Artificial neural network short-circuit current zero prediction method and prediction terminal |
| CN116087692A (en) * | 2023-04-12 | 2023-05-09 | 国网四川省电力公司电力科学研究院 | Distribution network tree line discharge fault identification method, system, terminal and medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1153488A1 (en) * | 1999-02-12 | 2001-11-14 | Deutsche Telekom AG | Method for monitoring the transmission quality of an optical transmission system, notably an optical wavelength-division multiplex network |
| CN106291351A (en) * | 2016-09-20 | 2017-01-04 | 西安工程大学 | Primary cut-out fault detection method based on convolutional neural networks algorithm |
| 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 |
| CN108569607A (en) * | 2018-06-22 | 2018-09-25 | 西安理工大学 | Elevator faults method for early warning based on bidirectional valve controlled Recognition with Recurrent Neural Network |
| 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 |
| CN109726200A (en) * | 2018-12-06 | 2019-05-07 | 国网甘肃省电力公司信息通信公司 | Fault location system and method for power grid information system based on bidirectional deep neural network |
| US20190146474A1 (en) * | 2016-05-09 | 2019-05-16 | Strong Force Iot Portfolio 2016, Llc | Methods and systems of industrial production line with self organizing data collectors and neural networks |
| CN110118928A (en) * | 2018-02-05 | 2019-08-13 | 西安交通大学 | A kind of circuit breaker failure diagnostic method based on Back Propagation Algorithm |
| CN110261109A (en) * | 2019-04-28 | 2019-09-20 | 洛阳中科晶上智能装备科技有限公司 | A kind of Fault Diagnosis of Roller Bearings based on bidirectional memory Recognition with Recurrent Neural Network |
-
2019
- 2019-12-17 CN CN201911303417.4A patent/CN111060815B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1153488A1 (en) * | 1999-02-12 | 2001-11-14 | Deutsche Telekom AG | Method for monitoring the transmission quality of an optical transmission system, notably an optical wavelength-division multiplex network |
| US20190146474A1 (en) * | 2016-05-09 | 2019-05-16 | Strong Force Iot Portfolio 2016, Llc | Methods and systems of industrial production line with self organizing data collectors and neural networks |
| CN106291351A (en) * | 2016-09-20 | 2017-01-04 | 西安工程大学 | Primary cut-out fault detection method based on convolutional neural networks algorithm |
| 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 |
| CN110118928A (en) * | 2018-02-05 | 2019-08-13 | 西安交通大学 | A kind of circuit breaker failure diagnostic method based on Back Propagation Algorithm |
| CN108569607A (en) * | 2018-06-22 | 2018-09-25 | 西安理工大学 | Elevator faults method for early warning based on bidirectional valve controlled Recognition with Recurrent Neural Network |
| 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 |
| CN109726200A (en) * | 2018-12-06 | 2019-05-07 | 国网甘肃省电力公司信息通信公司 | Fault location system and method for power grid information system based on bidirectional deep neural network |
| CN110261109A (en) * | 2019-04-28 | 2019-09-20 | 洛阳中科晶上智能装备科技有限公司 | A kind of Fault Diagnosis of Roller Bearings based on bidirectional memory Recognition with Recurrent Neural Network |
Non-Patent Citations (1)
| Title |
|---|
| 黄新波 等: "智能断路器机械特性在线监测技术和状态评估", 《高压电器》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN112729411A (en) * | 2021-01-14 | 2021-04-30 | 金陵科技学院 | Distributed drug warehouse environment monitoring method based on GA-RNN |
| CN112729411B (en) * | 2021-01-14 | 2022-09-13 | 金陵科技学院 | A distributed drug warehouse environment monitoring method based on GA-RNN |
| CN113358157A (en) * | 2021-06-10 | 2021-09-07 | 国网甘肃省电力公司兰州供电公司 | RST-PNN-GA-based power equipment temperature rise detection and early warning method |
| CN113469222A (en) * | 2021-06-10 | 2021-10-01 | 国网甘肃省电力公司兰州供电公司 | RST-PNN-GA-based high-voltage circuit breaker fault detection method |
| CN115389812A (en) * | 2022-10-28 | 2022-11-25 | 国网信息通信产业集团有限公司 | Artificial neural network short-circuit current zero prediction method and prediction terminal |
| CN116087692A (en) * | 2023-04-12 | 2023-05-09 | 国网四川省电力公司电力科学研究院 | Distribution network tree line discharge fault identification method, system, terminal and medium |
| CN116087692B (en) * | 2023-04-12 | 2023-06-23 | 国网四川省电力公司电力科学研究院 | Distribution network tree line discharge fault identification method, system, terminal and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111060815B (en) | 2021-09-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111060815B (en) | GA-Bi-RNN-based high-voltage circuit breaker fault diagnosis method | |
| CN109270442B (en) | DBN-GA neural network-based high-voltage circuit breaker fault detection method | |
| CN106291351B (en) | High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm | |
| CN106856322B (en) | A kind of flexible direct current power distribution network intelligent protection system based on neural network | |
| CN115081316A (en) | DC/DC converter fault diagnosis method and system based on improved sparrow search algorithm | |
| CN113486078A (en) | Distributed power distribution network operation monitoring method and system | |
| CN112147494A (en) | Mechanical fault detection method for high-voltage vacuum circuit breaker | |
| CN110118928B (en) | Breaker fault diagnosis method based on error inverse propagation algorithm | |
| CN107450016A (en) | Fault Diagnosis for HV Circuit Breakers method based on RST CNN | |
| CN112464556A (en) | AC contactor electric service life prediction method based on long-short term memory neural network | |
| CN114266301B (en) | Intelligent power equipment fault prediction method based on graph convolution neural network | |
| CN101739025A (en) | Immunity genetic algorithm and DSP failure diagnostic system based thereon | |
| CN109031114A (en) | A kind of modeling of spring actuator mechanism circuit-breaker and method for diagnosing faults | |
| CN112327098A (en) | Power distribution network fault section positioning method based on low-voltage distribution network comprehensive monitoring unit | |
| CN114785573A (en) | Intelligent substation process layer network abnormal flow detection method based on deep learning | |
| Chen et al. | Remaining useful life prediction of turbofan engine based on temporal convolutional networks optimized by genetic algorithm | |
| CN116629831A (en) | Switch machine health management method and system based on hidden semi-Markov model | |
| CN113536509A (en) | A microgrid topology identification method based on graph convolutional network | |
| CN112924813A (en) | Power distribution network short-circuit fault monitoring method and device based on electrical data | |
| CN109901064A (en) | Fault Diagnosis Method of High Voltage Circuit Breaker Based on ICA-LVQ | |
| CN116401572A (en) | A method and system for fault diagnosis of transmission lines based on CNN-LSTM | |
| CN118568653B (en) | State perception and fault diagnosis method of switchgear for combined electrical appliances based on multiple characteristic parameters | |
| CN113326663A (en) | Railway relay running state evaluation method based on extreme learning machine | |
| CN112085043B (en) | Intelligent monitoring method and system for network security of transformer substation | |
| CN113762591A (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy |
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 |