A method and system for intelligent fault diagnosis and recovery of coal mine power supply network

Abstract

The present invention relates to the field of power supply network fault recovery, and specifically, to an intelligent fault diagnosis and recovery method and system for coal mine power supply networks. The method obtains the fault nodes of the power supply network system; based on the characteristics of the above-mentioned fault nodes, determines the priority order between the above-mentioned fault nodes; the characteristics of the above-mentioned fault nodes include the equipment load of the fault nodes, the contribution of the fault nodes to the stability of the power supply network, and the load recovery adjustment amount of the fault nodes; and adjusts the voltage of the above-mentioned fault nodes according to the above-mentioned priority order until the voltage of the power supply network system is restored to a normal voltage. The present invention drives the voltage in the surrounding area to stabilize and reduces the overall voltage deviation by restoring the fault nodes with high priority, so as to improve the voltage recovery efficiency of the power supply network system.

Description

Technical Field

[0001] The present invention relates to the field of power supply network fault recovery, and in particular to an intelligent fault diagnosis and recovery method and system for a coal mine power supply network. Background Art

[0002] As one of the most important infrastructures in the industrial production process, the stability of the power supply network is directly related to the safe production and economic benefits of industrial production. Traditional power supply networks are usually composed of substations, transmission lines, distribution equipment, etc., which are used to transmit electricity to various production links such as mining, transportation, and processing. With the continuous expansion of industrial production, the power supply network is facing an increasingly complex operating environment and increasing load pressure. Therefore, the timely diagnosis and effective recovery of power supply network faults have become the key to ensuring the safety of industrial production. Intelligent fault diagnosis and recovery technology relies on advanced artificial intelligence, automatic control and data analysis technology. By monitoring and analyzing various parameters in the power system, faults can be detected and responded to in real time, thereby reducing the impact of faults on production.

[0003] In the process of fault diagnosis and recovery of traditional power supply networks, manual inspections, equipment monitoring and simplified fault location methods are usually relied on. Traditional fault diagnosis methods are often based on empirical rules and preset thresholds, and use the monitoring signals of the equipment (such as current, voltage, etc.) to determine whether the system has a fault. Although these methods can provide preliminary warnings in some simple fault situations, due to the diversity of fault types and the complexity of the power network, traditional methods have problems such as low diagnostic accuracy and long response time. In addition, traditional fault recovery also mainly relies on manual operations and preset recovery processes, with slow recovery speed and susceptible to human misjudgment. Therefore, the limitations of traditional technologies often lead to untimely system recovery when facing complex power grid faults, affecting the continuity and safety of industrial production. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent fault diagnosis and recovery method and system for a coal mine power supply network, so as to solve the problem in the prior art that power supply network faults cannot be promptly and efficiently recovered.

[0005] An embodiment of the present invention provides an intelligent fault diagnosis and recovery method for a coal mine power supply network, which is applied to a power supply network system, and the method includes the following steps: obtaining a faulty node of the power supply network system; determining a priority order between the faulty nodes based on characteristics of the faulty nodes; the characteristics of the faulty nodes include the equipment load of the faulty nodes, the contribution of the faulty nodes to the power supply network stability, and the load recovery adjustment amount of the faulty nodes; and adjusting the voltage of the faulty nodes according to the priority order until the voltage of the power supply network system returns to a normal voltage.

[0006] Optionally, determining the priority order between the faulty nodes based on the characteristics of the faulty nodes includes: obtaining a first weight coefficient corresponding to the device load and a second weight coefficient corresponding to the power supply network stability contribution; calculating a first product of the first weight coefficient and the device load, calculating a second product of the second weight coefficient and the power supply network stability contribution, and calculating the sum of the first product and the second product to obtain a priority evaluation value; determining the priority order between the faulty nodes according to the order of size between the priority evaluation values.

[0007] Optionally, determining the priority order between the faulty nodes based on the characteristics of the faulty nodes further includes: if there are some faulty nodes having the same priority evaluation values, obtaining the load recovery adjustment amounts corresponding to the faulty nodes; and determining the priority order between the faulty nodes according to the size order of the load recovery adjustment amounts.

[0008] Optionally, the acquiring the load recovery adjustment amount corresponding to the fault node includes: acquiring the fault voltage and the normal voltage of the fault node and the maximum load recovery adjustment amount of the fault node; calculating the difference between the fault voltage and the normal voltage according to the fault voltage and the normal voltage of the fault node; and calculating the load recovery adjustment amount of the fault node according to the difference, the maximum load recovery adjustment amount of the fault node and the normal voltage of the fault node;

[0009] The calculation formula of the load recovery adjustment amount of the fault node is as follows:

[0010] ,

[0011] in, Indicates the serial number is The load recovery adjustment amount of the fault node, Indicates the serial number is The maximum load recovery adjustment amount of the fault node, Indicates the serial number is The difference between the fault voltage and the normal voltage of the fault node, Indicates the serial number is The normal voltage of the fault node.

[0012] Optionally, adjusting the voltage of the faulty node according to the priority order until the voltage of the power supply network system recovers to a normal voltage includes: obtaining a load recovery adjustment amount of the faulty node; adjusting the voltage of the faulty node according to the priority order and the load recovery adjustment amount of the faulty node; obtaining the load recovery adjustment amount of each node of the power supply network system after the voltage of the faulty node is adjusted; and adjusting the voltage of each node of the power supply network system according to the load recovery adjustment amount of each node of the power supply network system, so that the voltage of the power supply network system recovers to a normal voltage.

[0013] Optionally, the method further includes: limiting the total amount of load shedding during the process of the voltage of the power supply network system recovering to a normal voltage to be less than a preset total amount of load shedding threshold.

[0014] Optionally, the method also includes: limiting the load shedding speed of each node of the power supply network system during the process of the voltage of the power supply network system recovering to a normal voltage to be greater than a preset load shedding speed lower limit threshold, and the load shedding speed of each node of the power supply network system is less than a preset load shedding speed upper limit threshold.

[0015] Optionally, the method further includes: limiting the voltage fluctuation of each node of the power supply network system to be less than a preset fluctuation voltage threshold during the process of the voltage of the power supply network system recovering to a normal voltage.

[0016] Optionally, obtaining the faulty node of the power supply network system includes: establishing a fault monitoring model for each node of the power supply network system based on a convolutional neural network architecture; and using the fault detection model to perform real-time monitoring on each node of the power supply network system to obtain the faulty node of the power supply network system.

[0017] Compared with the prior art, the intelligent fault diagnosis and recovery method for coal mine power supply network provided by the present invention has the following beneficial effects:

[0018] An intelligent fault diagnosis and recovery method for a coal mine power supply network provided by an embodiment of the present invention obtains the fault nodes of the power supply network system; based on the characteristics of the fault nodes, the priority order between the fault nodes is determined; the characteristics of the fault nodes include the equipment load of the fault nodes, the contribution of the fault nodes to the stability of the power supply network, and the load recovery adjustment of the fault nodes; the voltage of the fault nodes is adjusted according to the priority order until the voltage of the power supply network system is restored to normal voltage. By restoring the fault nodes with high priority first, the voltage in the surrounding area is driven to stabilize, and the overall voltage deviation is reduced, so as to improve the voltage recovery efficiency of the power supply network system.

[0019] An embodiment of the present invention provides an intelligent fault diagnosis and recovery system for a coal mine power supply network, the system comprising: an acquisition module, used to acquire the faulty nodes of the power supply network system; a determination module, used to determine the priority order between the faulty nodes based on the characteristics of the faulty nodes; the characteristics of the faulty nodes include the equipment load of the faulty nodes, the contribution of the faulty nodes to the power supply network stability and the load recovery adjustment amount of the faulty nodes; a recovery module, used to adjust the voltage of the faulty nodes according to the priority order until the voltage of the power supply network system returns to normal voltage.

[0020] The beneficial effect of the intelligent fault diagnosis and recovery system for a coal mine power supply network provided by the present invention is that it can achieve the same technical effect as the intelligent fault diagnosis and recovery method for a coal mine power supply network described above, and will not be described here to avoid repetition. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0022] Figure 1 A schematic flow chart of an intelligent fault diagnosis and recovery method for a coal mine power supply network provided by an embodiment of the present invention;

[0023] Figure 2 It is a schematic flow chart of a specific intelligent fault diagnosis and recovery method for a coal mine power supply network in an embodiment of the present invention;

[0024] Figure 3 A schematic flow chart of a method for intelligent diagnosis of power supply network system faults in an embodiment of the present invention;

[0025] Figure 4 A schematic flow chart of data collection and preprocessing in an embodiment of the present invention;

[0026] Figure 5 is a schematic flow chart of a method for building a solution platform in an embodiment of the present invention;

[0027] Figure 6 The present invention is a schematic diagram of the structure of an intelligent fault diagnosis and recovery system for a coal mine power supply network in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0029] The embodiment of the present invention provides an intelligent fault diagnosis and recovery method for a coal mine power supply network, which is applied to a power supply network system. Figure 1 A schematic flow chart of an intelligent fault diagnosis and recovery method for a coal mine power supply network is shown; the method comprises the following steps:

[0030] S110, obtaining a faulty node of the power supply network system.

[0031] Optionally, the above step S110 includes: establishing a fault monitoring model for each node of the power supply network system based on a convolutional neural network architecture; using the above fault detection model to perform real-time monitoring on each node of the power supply network system to obtain the faulty nodes of the power supply network system.

[0032] S120: Determine the priority order between the faulty nodes based on the characteristics of the faulty nodes.

[0033] The characteristics of the above-mentioned fault node include the equipment load of the fault node, the contribution of the fault node to the stability of the power supply network and the load recovery adjustment amount of the fault node.

[0034] Optionally, the above-mentioned step S120 includes: obtaining a first weight coefficient corresponding to the above-mentioned equipment load and a second weight coefficient corresponding to the above-mentioned power supply network stability contribution; calculating a first product of the first weight coefficient and the above-mentioned equipment load, calculating a second product of the second weight coefficient and the above-mentioned power supply network stability contribution, calculating the sum of the first product and the second product, and obtaining a priority evaluation value; determining the priority order between the above-mentioned fault nodes according to the order of size between the priority evaluation values.

[0035] Optionally, the above step S120 also includes: if there are some faulty nodes with the same priority evaluation values, then obtaining the load recovery adjustment values ​​of the corresponding faulty nodes; and determining the priority order between the corresponding faulty nodes according to the order of magnitude of the load recovery adjustment values.

[0036] It should be noted that the power supply network includes more than one path, and the path is formed by connecting nodes. A path contains multiple nodes, and a node may be connected to more than one device, which together constitute the basic elements of the power supply network, are interrelated and affect each other. The embodiment of the present invention is explained by taking the determination of the priority between nodes as an example, and can also be used to determine the priority between each path / device to further improve the efficiency of power supply network recovery.

[0037] Optionally, the step of obtaining the load recovery adjustment amount of the corresponding fault node includes: obtaining the fault voltage and normal voltage of the fault node and the maximum load recovery adjustment amount of the fault node; calculating the difference between the fault voltage and the normal voltage based on the fault voltage and the normal voltage of the fault node; calculating the load recovery adjustment amount of the fault node based on the difference, the maximum load recovery adjustment amount of the fault node and the normal voltage of the fault node.

[0038] The calculation formula for the load recovery adjustment amount of the above fault node is as follows:

[0039] ,

[0040] in, Indicates the serial number is The load recovery adjustment amount of the faulty node, Indicates the serial number is The maximum load recovery adjustment of the failed node, Indicates the serial number is The difference between the fault voltage and the normal voltage of the fault node, Indicates the serial number is The normal voltage of the fault node.

[0041] It should be noted that the maximum load recovery adjustment amount of each fault node is affected by the load amount of each fault node and can be obtained in advance. The function of the above-mentioned load recovery adjustment amount calculation formula of the fault node is to calculate the load amount that needs to be removed during the recovery process through the voltage change, so as to ensure that the restored voltage level can be restored to a safe range.

[0042] S130, adjusting the voltage of the above-mentioned faulty node according to the above-mentioned priority order until the voltage of the power supply network system returns to a normal voltage.

[0043] It should be noted that the faulty node with a higher priority has the priority for voltage adjustment.

[0044] Optionally, the above step S130 includes: obtaining the load recovery adjustment amount of the faulty node; adjusting the voltage of the above faulty node according to the above priority order and the load recovery adjustment amount of the above faulty node; obtaining the load recovery adjustment amount of each node of the power supply network system after the voltage of the above faulty node is adjusted; adjusting the voltage of each node of the power supply network system according to the load recovery adjustment amount of each node of the power supply network system, so that the voltage of the power supply network system is restored to a normal voltage. In this way, the faulty nodes with a high recovery priority are restored first, driving the voltage in the surrounding area to stabilize, so that when other nodes in the surrounding area perform voltage adjustment, they will not experience serious voltage fluctuations due to the faulty nodes with a high recovery priority, avoiding the problem of repeated adjustment of other nodes in the surrounding area; through the above reasonable load shedding operation, the supply and demand relationship of the power grid can be balanced faster, the voltage can be stabilized, and the voltage difference can be reduced, so that the voltage of the power supply network system can be restored to a normal voltage as soon as possible.

[0045] Optionally, the method further includes: limiting the total amount of load shedding during the process of the voltage of the power supply network system recovering to a normal voltage to be less than a preset total amount of load shedding threshold.

[0046] It should be noted that the total amount of load shedding is the amount of load actually shedded during the process of voltage restoration in the power supply network system; the load recovery adjustment amount of the fault node is the preset load shedding amount calculated by the above-mentioned load recovery adjustment amount calculation formula for the fault node; the load recovery adjustment amount of each fault node can be used as a reference for the actual load shedding amount of each fault node.

[0047] Optionally, the above method also includes: limiting the load shedding speed of each node of the power supply network system during the process of the power supply network system voltage recovering to normal voltage to be greater than a preset load shedding speed lower limit threshold, and the load shedding speed of each node of the power supply network system is less than a preset load shedding speed upper limit threshold.

[0048] Optionally, the method further includes: limiting the voltage fluctuation of each node of the power supply network system to be less than a preset fluctuation voltage threshold during the process of the voltage of the power supply network system being restored to a normal voltage.

[0049] The intelligent fault diagnosis and recovery method for the coal mine power supply network provided by the embodiment of the present invention obtains the fault nodes of the power supply network system; determines the priority order between the fault nodes based on the characteristics of the fault nodes; and adjusts the voltage of the fault nodes according to the priority order until the voltage of the power supply network system is restored to the normal voltage. By restoring the fault nodes with high restoration priority first, the voltage in the surrounding area is driven to stabilize and the overall voltage deviation is reduced; because the restoration order of the fault nodes is determined, the power supply network system is prompted to stabilize faster, so as to improve the voltage recovery efficiency of the power supply network system.

[0050] The embodiment of the present invention also provides a specific intelligent fault diagnosis and recovery method for a coal mine power supply network. Figure 2 A schematic flow chart of a specific intelligent fault diagnosis and recovery method for a coal mine power supply network is shown, the method comprising the following steps:

[0051] S202, collecting system voltage and current data and performing preprocessing.

[0052] Among them, preprocessing includes data acquisition and A / D conversion, filtering, and normalization.

[0053] S204, constructing input data features of the convolutional neural network model.

[0054] S206, construct a convolutional neural network model.

[0055] S208, model training stage.

[0056] According to steps S202-S206, the network model based on the convolutional neural network is trained so that it has the ability to recognize input features.

[0057] S210, model testing phase.

[0058] The convolutional neural network model constructed in steps S202-S208 is tested, with the test set as input, and the output of the network model is compared with the faults marked by the experts.

[0059] S212, apply the accurately tested model to actual engineering.

[0060] The accurate test model constructed in step S210 is applied to actual projects.

[0061] S214, construct a fault recovery solution based on voltage reconstruction.

[0062] S216, building a model to solve the optimization objective function of the fault recovery solution solution platform.

[0063] S218, calculating the solution platform.

[0064] Calculate the solution platform built in step S216;

[0065] S210, screening and comparing the calculation results, selecting the optimal recovery path to implement the fault recovery plan.

[0066] The above-mentioned specific intelligent fault diagnosis and recovery method of the coal mine power supply network realizes the intelligent diagnosis and efficient recovery of the coal mine power supply network fault by introducing advanced convolutional neural networks (CNN) and voltage reconstruction technology. Compared with the traditional fault diagnosis method that relies on manual and empirical rules, the scheme can quickly and accurately identify and locate various types of faults, especially complex multiple fault conditions, and effectively improve the accuracy and response speed of fault detection. At the same time, the fault recovery model based on voltage reconstruction not only optimizes the selection of recovery paths, but also dynamically adjusts the recovery strategy through intelligent algorithms to ensure the safety and efficiency of the fault recovery process. The technical solution provided by the embodiment of the present invention greatly shortens the fault handling time, avoids delays or improper recovery caused by human operation errors, ensures the continuity and stability of the coal mine power supply system, and provides a strong guarantee for the safe operation of coal mine production. At the same time, intelligent technical means enable the system fault recovery to be completed automatically, greatly reducing the dependence on manual intervention, and improving the overall automation level and intelligent management capabilities of the coal mine power supply network.

[0067] The embodiment of the present invention also provides a method for intelligent diagnosis of power supply network system faults. Figure 3 A schematic flow chart of a method for intelligent diagnosis of power supply network system faults is shown. The method for intelligent diagnosis of power supply network system faults comprises the following steps:

[0068] Step S302: Collecting voltage and current data of the power supply network system and performing preprocessing.

[0069] It should be noted that the above preprocessing includes data acquisition and A / D conversion, filtering, and normalization.

[0070] For example, in the data collection stage, a smart energy meter is used for real-time data collection, and the sampling frequency is set to 500 Hz, which can ensure that the voltage and current data with high frequency changes are accurately captured. Figure 4 A schematic flow chart of data acquisition and preprocessing is shown in FIG. The collected data is first converted from analog signals to digital signals through A / D conversion. Then, the data is analyzed in the frequency domain using Fourier transform. Fourier transform can convert time domain signals into frequency domain signals, so that the frequency components are clearly expressed, which is convenient for further filtering and feature extraction. The Fourier transform formula is as follows:

[0071] ,

[0072] in, represents the frequency domain signal after Fourier transform, is the time domain signal, is the frequency. The result of Fourier transform can be used to analyze the frequency components of the signal, filter out high-frequency noise or irrelevant signals, and thus enhance the reliability of the data. After filtering, normalization is performed to scale all data to a uniform range (usually 0 to 1). The normalized data generates data samples and provides standardized input for subsequent model training.

[0073] Step S304: Based on the above voltage and current data, construct the input data features of the convolutional neural network model.

[0074] For example, when constructing the input data features of the convolutional neural network, firstly, the normalized data is subjected to expert judgment. The expert judges the characteristics of faults and non-faults by analyzing the changing trend and frequency distribution of the data. Then, the data is segmented, and the time series data is divided into multiple small segments to ensure that each segment contains enough information for subsequent learning. For each segment of data, the data is labeled to indicate whether a fault has occurred or the type of fault. Then, a database is established to save the labeled data by category.

[0075] After the data is labeled and segmented, the sample division phase begins. The data samples are divided into training data sets, test data sets, and validation data sets according to a certain ratio (such as 6:2:2). The training data set is used for model training, the test data set is used for model evaluation, and the validation data set is used for tuning and preventing overfitting during the training process. In addition, in order to ensure that the convolutional neural network can better learn the features in the data, it is necessary to formulate training, testing, and validation rules, including the selection of loss functions, the setting of optimization algorithms (such as Adam or SGD), and evaluation indicators (such as accuracy, F1 value, etc.).

[0076] Step S306: Construct a convolutional neural network model.

[0077] For example, when building a convolutional neural network model, it is necessary to first establish a convolutional layer structure. In the setting of the number of nodes in the input layer and output layer of the convolutional neural network, the number of nodes in the input layer and output layer needs to be reasonably set according to the dimension and number of categories of the actual data. The number of nodes in the input layer is determined by the characteristic dimension of the data. If each data sample contains multiple voltage and current data channels (such as current and voltage sampling points), the number of nodes in the input layer is the total number of dimensions of these data. The number of nodes in the output layer is determined by the number of fault classifications. If the fault types are divided into fault and non-fault categories, the number of nodes in the output layer is 2.

[0078] The setting of the convolution layer needs to calculate the number and size of the convolution kernels according to the empirical formula. Assume that the number of convolution kernels is determined by the number of input layer nodes. , the number of nodes in the output layer , the input range of the activation function With output range The formula is as follows:

[0079] , (1)

[0080] in, is the number of convolution kernels, is a constant between 0 and 9, usually adjusted experimentally. is the number of nodes in the input layer, is the number of nodes in the output layer, is the input range of the activation function, is the output range of the activation function. This formula can reasonably estimate the number and size of convolution kernels according to the actual data scale, thereby optimizing the design of the convolution layer.

[0081] The convolution layer extracts local features from the data through convolution operations. The convolution operation can be expressed by the following formula:

[0082] ,

[0083] in, Represents input data, is the convolution kernel, is the output after convolution. The function of the convolution layer is to extract local features in the input data by sliding the convolution kernel and output the feature map.

[0084] Next, a pooling layer structure is established. The pooling layer usually uses maximum pooling or average pooling, which aims to reduce dimensionality and computational complexity while retaining important features in the data. The pooling layer simplifies the data by selecting the maximum value or average value in the region.

[0085] It should be noted that the above pooling layer structure is explained as follows: the pooling layer is used to reduce the dimension of the feature map output by the convolution layer while retaining important features. The calculation of the pooling step size and the feature map size is the key to designing the pooling layer. The pooling step size is generally determined by the number of samples in the training set and the validation set. A larger step size can speed up the training process, but may lose some detailed information. The size of each dimension feature map after pooling can be calculated using the following formula:

[0086] ,

[0087] in, is the size of the feature map after pooling, is the feature map size before pooling, is the size of the pooling window, The pooling operation reduces the computational complexity by reducing the dimension of each feature map, while making the convolutional neural network more fault-tolerant and more robust to small changes in the input data.

[0088] When designing the pooling layer, it is also necessary to calculate the reduction in the number of feature maps after pooling. The number of reduced features is usually determined by calculating the pooling results of each dimension to ensure that too many valuable features are not lost during the dimensionality reduction process.

[0089] In this way, the convolutional neural network can not only effectively diagnose faults in the coal mine power supply network, but also provide accurate data support for subsequent fault recovery.

[0090] Then, a Dropout layer structure is established to prevent overfitting of the neural network. During the training process, the Dropout layer randomly discards neurons, making the network structure different each time, thereby enhancing the generalization ability of the model.

[0091] Then, a fully connected layer structure is established to synthesize the extracted features, and finally the prediction layer generates the final output, which is the fault category or the predicted probability of the fault. Each layer is activated by a nonlinear activation function (such as ReLU or Sigmoid), so that the network has sufficient expressive power.

[0092] In the final construction of the model, these layers are connected sequentially to form a complete convolutional neural network structure, which can effectively learn fault characteristics from input data and make accurate fault diagnosis.

[0093] For example, assume that in a coal mine power supply network, the collected current data fluctuates significantly when a fault occurs. In the data collection stage, a sampling frequency of 500Hz is used, and the captured data will be converted into a frequency domain signal through Fourier transform to filter out the higher frequency noise. After normalization, the generated data samples will be used to construct the input features of the convolutional neural network model. Assume that through expert judgment, it is found that certain features of this segment of data (such as sudden current values, specific frequency components in the frequency domain) are highly correlated with the fault. In the model training stage, by learning a large number of data samples, the convolutional neural network can automatically extract these features, and train according to the labeled results, and finally generate a network model that can accurately diagnose whether the current fluctuation is caused by the fault. After testing and verification, the model can accurately identify and classify faults, thereby providing decision support for subsequent fault recovery plans.

[0094] Step S308: training a network model based on a convolutional neural network.

[0095] Exemplarily, a network model based on a convolutional neural network is trained to enable it to have the ability to recognize input features. The convolutional neural network training phase is the core of the convolutional neural network model optimization, and setting a suitable training cycle and maximum number of iterations is crucial to the training effect. In an embodiment of the present invention, the training cycle is set to 4000 times, the maximum number of iterations of the training set is 2000 times, the maximum number of iterations of the validation set is set to 4000 times, and the network learning rate is set to 0.001. The learning rate controls the step size of each parameter update. A smaller learning rate helps to fine-tune the model parameters, but will increase the training time, and a larger learning rate will accelerate convergence, but may cause the optimal solution to be missed. Therefore, choosing a suitable learning rate has an important influence on the convergence speed and final effect of the model.

[0096] During the training process, the model parameters are randomly initialized to avoid falling into the local optimal solution. The samples in the training set will be repeatedly used to train the convolutional neural network until the set maximum number of iterations is reached. At the same time, at the end of each iteration cycle, the performance indicators of the model (such as accuracy, loss value, etc.) are calculated through the validation set, and the model parameters are adjusted according to the results of the validation set to prevent overfitting and ensure the generalization ability of the model on new data.

[0097] Step S310: Test and optimize the constructed convolutional neural network model.

[0098] Exemplarily, the test set is used as input, the network model output result is compared with the faults marked by the expert, and the network model is optimized until the network model output result is consistent with the faults marked by the expert.

[0099] Step S312: Apply the constructed and tested accurate model to the actual project.

[0100] For example, in the fault diagnosis of coal mine power supply network, the input data contains multiple voltage and current channels, each channel collects 1000 sampling points, and the data dimension is 1000. Therefore, the number of nodes in the input layer is 1000. In the case of two types of faults, the number of nodes in the output layer is 2. By applying the above empirical formula, the number of convolution kernels of the convolution layer can be calculated. For example, assuming that the constant in the above formula (1) is , the number of nodes in the input layer is 1000, the number of nodes in the output layer is 2, the input range of the activation function is -1 to 1, and the output range is 0 to 1. Substitute it into the formula to calculate the number of convolution kernels.

[0101] At the same time, when the feature map is reduced in dimension through the pooling layer, the pooling window size is set to 2 and the step size is set to 2. Assuming that the size of the feature map before pooling is 28×28, the size of the feature map of each dimension after pooling is 14×14, and the number of feature maps will be reduced accordingly, thereby speeding up the calculation and reducing memory usage.

[0102] In addition, there are other ways to apply the constructed accurate test model to actual projects: according to the actual power supply network system status, the data collection and preprocessing steps generate a sample set related to the fault, providing basic data for subsequent fault diagnosis and recovery. In this process, the generation of the fault sample set is based on the dynamic changes of voltage and current data, and the power grid status is trained through the convolutional neural network to obtain key features that can reflect the system status. Next, the trained convolutional neural network model is used to input the network node set S1, and the abnormal probability and fault category of the node are output to obtain the node set S2. The calculation of the network node set S2 lays the foundation for the subsequent node classification. Specifically, the convolutional neural network model can output the abnormal probability of each node by processing the input features, and usually selects nodes with a probability of more than 70% as abnormal nodes. For nodes predicted to be abnormal, the depth first search (DFS) algorithm is used to traverse the nodes to obtain the node set adjacent to the abnormal node, and these adjacent nodes are further classified according to the categories predicted by the convolutional neural network to obtain the normal node set S3 and the abnormal node set S4. The key to this step is to effectively expand the impact range of abnormal nodes and comprehensively evaluate the areas affected by the fault through network topology relationships combined with the adjacent relationships between nodes.

[0103] Optionally, assume that the convolutional neural network predicts two categories: current overload fault and voltage drop fault. After performing DFS traversal on the node A predicted to be abnormal to obtain the adjacent nodes B, C, and D, if the voltage and current data of node B are highly similar to the characteristic data of the current overload fault (such as the current value is continuously higher than the normal threshold by a certain proportion and fluctuates violently) after analysis, and match the sample data pattern of the current overload fault during the convolutional neural network training, then node B is classified into the abnormal node set S4; if the voltage and current data of node C are relatively stable and within the normal range, and meet the characteristics of normal operation, then node C is classified into the normal node set S3; for node D, if its voltage data has a short-term small drop but recovers quickly and the overall trend is close to normal, which does not meet the severity standard of the voltage drop fault, then it is also classified into the normal node set S3.

[0104] The present invention also provides a method for constructing a solution platform. Figure 5 A schematic flow chart of a method for constructing a solution platform is shown, and the steps of the method for constructing a solution platform include:

[0105] Step S502: establishing a voltage reconstruction objective function and constructing a voltage reconstruction recovery model.

[0106] The goal of voltage reconstruction is to restore the power supply network to a voltage state close to that before the fault as soon as possible after the fault occurs. Assume that the voltage of the power supply network after the fault occurs is , while the voltage when no fault occurs is , the goal of voltage reconstruction is to minimize the voltage deviation, that is, to restore each device to its normal voltage state. The objective function can be expressed as:

[0107] ,

[0108] in, is the voltage after the fault occurs, is the voltage before the fault occurs, is the total number of devices. The purpose of this objective function is to reconstruct the voltage by minimizing the voltage deviation to ensure that the voltage level of the power supply network returns to normal.

[0109] Step S504, setting constraints corresponding to the voltage reconstruction and recovery model, and completing the construction of the solution platform.

[0110] Equipment recovery priority, safe operation constraints, total load shedding constraints and load shedding speed constraints are all closely related to the voltage difference.

[0111] For example, the device recovery priority is: ,in is the load of the device, is the contribution of the device to the system stability, and In practical applications, by monitoring and evaluating the load and system stability contribution of different devices, the weight factor is reasonably set according to the device type, importance and impact of failure. and , get the priority evaluation value , thereby determining the recovery priority of each device. During the fault recovery process, priority is given to Devices with high values ​​are restored, which will affect the selection of the restoration path and the restoration order, and further affect the efficiency of the entire restoration process and the stability of the system, and indirectly optimize the parameters in relationship functions such as voltage reconstruction. Because the change in the device recovery order will cause changes in the grid voltage and power flow distribution, the system will tend to a stable state faster, and parameters such as voltage will return to the normal range faster.

[0112] The determination of device recovery priority will affect the order of device recovery, and thus affect the speed and path of voltage recovery. Devices with high recovery priority will be recovered first, and their voltage will approach the normal voltage more quickly, thereby reducing the overall voltage deviation. For example, key devices will be restored first due to their high priority, which can drive the voltage in the surrounding areas to stabilize, making the voltage difference of the entire network develop in the direction of reduction.

[0113] For example, safe operation constraints: Taking voltage as an example, the safe operation constraints are ,in is the voltage during the recovery process, is the normal operating voltage of the equipment, is the preset voltage fluctuation threshold. During the fault recovery process, the voltage is monitored to ensure that it always meets the constraint. If the voltage approaches or exceeds the constraint range, the system will adjust the recovery strategy, such as adjusting the load shedding amount or the recovery path, to ensure that the voltage is stable within a safe range. This directly limits the voltage variation range, making the voltage deviation calculation in the voltage reconstruction objective function more reasonable, avoiding system instability caused by excessive voltage fluctuations, thereby optimizing the parameter selection of the entire recovery process, ensuring that the recovery process is carried out in a stable and safe direction, avoiding equipment damage or new faults caused by voltage fluctuations, ensuring the safe and stable operation of the power grid, and indirectly affecting other voltage-related parameters to change within a reasonable range, promoting the overall recovery of the system.

[0114] During the fault recovery process, the voltage is monitored to ensure that it always meets the constraint. If the voltage approaches or exceeds the constraint range, the system will adjust the recovery strategy; for example, adjust the load shedding amount or recovery path to ensure that the voltage is stable within a safe range, avoid new faults or equipment damage caused by excessive voltage fluctuations, and thus ensure the safety of the voltage reconstruction process.

[0115] For example, the total load shedding constraint is expressed as: ,in Yes Equipment The amount of load that needs to be reduced, is the total load shedding amount, It is the preset load shedding threshold. During the fault recovery process, the load shedding amount is calculated based on the real-time status of the power grid and voltage reconstruction requirements. , and ensure that their sum does not exceed . When load shedding is required, the system will comprehensively consider the conditions of each device and the overall load reduction requirements, and reasonably allocate the load shedding amount. This constraint limits the total amount of load shedding, avoiding excessive system instability due to excessive load shedding, and also affects the distribution of voltage and power flow. By controlling the total amount of load shedding, the power grid can maintain a certain load balance during the recovery process, which helps to stabilize the voltage, and then optimize the parameters in the voltage reconstruction objective function and other related relationship functions, ensuring the effectiveness and stability of the recovery process, so that the system can gradually resume normal operation under stable load conditions.

[0116] Load shedding speed constraint: The load shedding speed constraint is set to ,in It is Load shedding speed of busbars, and are the minimum and maximum allowable load shedding speeds, respectively. During the fault recovery process, the load shedding speed is monitored and controlled in real time to ensure that it meets the constraint condition. Too fast a load shedding speed may cause drastic fluctuations in the system voltage and frequency, causing system instability; too slow a load shedding speed may not be able to relieve the grid pressure in time, affecting the recovery effect. By reasonably controlling the load shedding speed, the system can make a smooth transition and avoid adverse effects on grid stability caused by too fast or too slow load adjustment, thereby optimizing the parameters in the recovery process and ensuring that parameters such as voltage and power flow are within a reasonable dynamic change range, which helps to achieve efficient and stable fault recovery, ensure the safe operation of the grid, and provide a stable operating environment for processes such as voltage reconstruction, which is conducive to the optimization of the parameters of the correlation function.

[0117] When the fault is restored, the system will reasonably allocate the load shedding amount according to the real-time status of the power grid to avoid excessive instability of the system due to excessive load shedding. Reasonable load shedding operation can balance the supply and demand relationship of the power grid, stabilize the voltage, and reduce the voltage difference. Controlling the load shedding speed can make the system transition smoothly and avoid the adverse effects on the stability of the power grid caused by too fast or too slow load adjustment, which will help stabilize the voltage, reduce the voltage difference, and optimize the voltage reconstruction objective function, so that the power grid can be restored to a voltage state close to that before the fault as soon as possible, ensuring the stable operation of the coal mine power supply network.

[0118] For example, power flow constraints: power flow constraints are crucial for controlling the voltage and load flow in the power grid. Power flow constraints specifically include voltage constraints, line power flow constraints, generator power flow constraints, and load power flow constraints.

[0119] Here, we take the line flow constraint as an example. The generator flow constraint and load flow constraint are similar. The line flow constraint reflects the change of the line admittance matrix after the fault occurs. Assume that the line admittance matrix before the fault is , and the line admittance matrix after the fault is , then the difference in line power flow can be expressed as:

[0120] ,

[0121] here, represents the line admittance matrix before the fault, represents the line admittance matrix after the fault, The difference between the two indicates the change in line admittance caused by the fault. By adjusting the line power flow constraints, line overload or instability can be effectively avoided.

[0122] After the above solution platform is solved, it can be determined which nodes need to be prioritized during the recovery process to reduce the impact of the fault on the power grid. The recovery amount of each node is based on the voltage constraint, and the maximum value of the recovery amount can be calculated by the following formula:

[0123] ,

[0124] in, Indicates the total load recovery adjustment of the power supply network system, is the total number of nodes that recover from load shedding, Represents the load recovery adjustment amount of each node. The focus of the screening process is to select the node with the strongest recovery capability and maximize the efficiency of grid restoration on this basis.

[0125] The voltage reconstruction objective function middle, After a failure, the device The voltage, Is the device before the failure The goal is to minimize this voltage deviation. Equipment recovery priority, safe operation constraint, total load shedding constraint and load shedding speed constraint are all closely related to the voltage difference. The determination of equipment recovery priority will affect the order of equipment recovery, and then affect the voltage recovery speed and path. Equipment with high recovery priority will be restored first, and its voltage will approach the normal voltage faster, thereby reducing the overall voltage deviation. The safe operation constraint directly limits the allowable range of the voltage difference, ensuring that the voltage will not deviate excessively from the normal range during the recovery process, avoiding new faults or equipment damage caused by excessive voltage fluctuations, and ensuring the safety of the voltage reconstruction process. The total load shedding constraint and the load shedding speed constraint indirectly affect the voltage difference by adjusting the load. Reasonable load shedding operation can balance the supply and demand relationship of the power grid, stabilize the voltage, and make the voltage difference develop in the direction of reduction, which is helpful to achieve the optimization of the voltage reconstruction objective function, so that the power grid can be restored to a voltage state close to the pre-fault state as soon as possible, and ensure the stable operation of the coal mine power supply network.

[0126] The embodiment of the present invention provides a specific coal mine power supply network intelligent fault diagnosis and recovery method, which relies on advanced convolutional neural network technology and voltage reconstruction fault recovery scheme to achieve efficient diagnosis, accurate recovery and safety assurance of the coal mine power supply network when a fault occurs. The method provided by the embodiment of the present invention not only improves the accuracy of fault identification and recovery through multiple innovative steps and optimization algorithms, but also significantly improves the stability and reliability of the power grid. Its beneficial effects are mainly reflected in the following aspects:

[0127] Efficient fault diagnosis and classification

[0128] The method provided by the embodiment of the present invention adopts convolutional neural network for fault diagnosis, and through the precise processing of voltage and current data, it can quickly and accurately identify the faulty nodes in the coal mine power supply network. Compared with the traditional fault detection method, the convolutional neural network can automatically learn the key features in the network, thereby reducing the interference of human factors and improving the accuracy and real-time performance of fault diagnosis. Especially when facing complex and dynamically changing power grid conditions, the convolutional neural network model can flexibly adapt to different types of faults and provide high-quality data support for subsequent recovery decisions.

[0129] Innovative application of voltage reconstruction fault recovery scheme

[0130] The method provided by the embodiment of the present invention constructs a fault recovery model based on voltage reconstruction, so that the embodiment of the present invention can dynamically adjust the recovery strategy according to the current state of the power grid and the specific situation of the fault node. The voltage reconstruction objective function not only considers the accuracy of the restored voltage, but also introduces multiple factors such as equipment recovery priority and safe operation constraints to ensure the safety and stability during the recovery process. In particular, by setting the total load shedding and load shedding speed constraints, secondary faults caused by overload or too fast recovery are avoided during the recovery process, effectively protecting the normal operation of the power grid equipment and the power supply system.

[0131] Adaptive recovery path optimization

[0132] The method provided in the embodiment of the present invention introduces a solution platform based on an optimization objective function in the fault recovery process, and uses multiple constraints to optimize the recovery path. By comprehensively considering factors such as voltage, load, flow constraints, and load shedding speed, the embodiment of the present invention can select the optimal recovery path in a complex power grid. This optimization strategy not only reduces the recovery time, but also avoids system instability and load fluctuations to the greatest extent, and improves the efficiency and accuracy of power grid recovery.

[0133] Intelligent decision making during failure recovery

[0134] The method provided by the embodiment of the present invention introduces a depth-first search algorithm in the recovery process, so that when a fault occurs, the embodiment of the present invention can automatically determine the relative position of abnormal nodes and normal nodes and their voltage mean values ​​according to the topological relationship between nodes, and give priority to restoring nodes whose voltage is lower than the safe voltage. Combined with the voltage mean value calculation and the optimization of the load shedding speed, the scientificity and efficiency of the system recovery are ensured, thereby achieving precise control and intelligent decision-making in the system recovery process.

[0135] Enhanced system stability and reliability

[0136] Since the embodiment of the present invention adopts a comprehensive fault diagnosis and recovery method, and quantifies and optimizes each link in the recovery process, the embodiment of the present invention can ensure that the coal mine power supply network can quickly and accurately restore normal operation after a fault occurs. This not only reduces the impact of the fault on the coal mine power supply system and reduces the power outage time, but also avoids the chain reaction caused by the spread of the fault, and enhances the overall stability and reliability of the power grid.

[0137] Save manual intervention and improve automation level

[0138] The method provided by the embodiment of the present invention realizes a highly intelligent fault diagnosis and recovery process through the automated integration of data acquisition, convolutional neural network model, optimization algorithm and fault recovery model. Compared with the traditional manual intervention method, the embodiment of the present invention reduces the intervention of manual operation, improves work efficiency, and enables the coal mine power supply network to self-recover without manual intervention, greatly improving the level of automation and reducing the risk of human error.

[0139] Strong adaptability, able to cope with complex power grid environments

[0140] The method provided by the embodiment of the present invention has strong adaptability and can cope with coal mine power supply networks of different scales and complexities. Through the deep learning ability of convolutional neural networks, the embodiment of the present invention can automatically adjust the fault diagnosis model and recovery strategy according to the actual situation of the power grid, thereby adapting to the challenges of different equipment and different fault types in the coal mine power supply network. Whether it is voltage fluctuations, equipment failures or line short circuits, various fault conditions can be quickly and accurately identified and restored.

[0141] Improving coal mine production safety assurance capabilities

[0142] The method provided by the embodiment of the present invention can significantly reduce the safety risks caused by coal mine power supply network failures and improve the safety production guarantee capability of coal mines through rapid and accurate fault diagnosis and recovery. It ensures that when a fault occurs, the power system can be restored to normal operation as soon as possible, providing continuous and stable power support for the production process of the coal mine, which helps to improve the production efficiency of the coal mine and ensure the safety of the working environment of the miners.

[0143] In summary, the method provided by the embodiment of the present invention significantly improves the fault response speed and recovery efficiency of the coal mine power supply network through an innovative intelligent fault diagnosis and recovery method, effectively ensures the stable operation of the power grid, and provides strong technical support for the safe production of coal mines.

[0144] The embodiment of the present invention provides an intelligent fault diagnosis and recovery system for a coal mine power supply network. Figure 6 The structure diagram of an intelligent fault diagnosis and recovery system for a coal mine power supply network is shown, and the system includes:

[0145] The acquisition module 602 is used to acquire the faulty node of the power supply network system.

[0146] The determination module 604 is used to determine the priority order between the above-mentioned faulty nodes based on the characteristics of the above-mentioned faulty nodes.

[0147] The characteristics of the above-mentioned fault node include the equipment load of the fault node, the contribution of the fault node to the stability of the power supply network and the load recovery adjustment amount of the fault node.

[0148] The recovery module 606 is used to adjust the voltage of the above-mentioned faulty node according to the above-mentioned priority order until the voltage of the power supply network system is restored to a normal voltage.

[0149] The beneficial effect of the intelligent fault diagnosis and recovery system for a coal mine power supply network provided by the present invention is that it can achieve the same technical effect as the intelligent fault diagnosis and recovery method for a coal mine power supply network described above, and will not be described here to avoid repetition.

[0150] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the control device through a computer, and the program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the above-mentioned method embodiments, wherein the storage medium may be a memory, a disk, an optical disk, etc.

[0151] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0152] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0153] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. An intelligent fault diagnosis and recovery method for a coal mine power supply network, characterized in that: Applied to a power supply network system, the method comprises the following steps: Obtaining a fault node of the power supply network system; Based on the characteristics of the faulty nodes, determining the priority order between the faulty nodes; the characteristics of the faulty nodes include the equipment load of the faulty nodes, the contribution of the faulty nodes to the power supply network stability, and the load recovery adjustment amount of the faulty nodes; adjusting the voltage of the faulty node according to the priority order until the voltage of the power supply network system returns to a normal voltage; The determining the priority order between the faulty nodes based on the characteristics of the faulty nodes includes: Obtaining a first weight coefficient corresponding to the device load and a second weight coefficient corresponding to the power supply network stability contribution; Calculate a first product of the first weight coefficient and the device load, calculate a second product of the second weight coefficient and the power supply network stability contribution, calculate the sum of the first product and the second product, and obtain a priority evaluation value; Determining the priority order between the faulty nodes according to the order of magnitude between the priority evaluation values; The determining the priority order between the faulty nodes based on the characteristics of the faulty nodes further includes: If the priority evaluation values ​​of some of the faulty nodes are the same, then obtaining the load recovery adjustment amount corresponding to the faulty nodes; The priority order between the corresponding fault nodes is determined according to the order of magnitudes between the load recovery adjustment amounts.

2. The method according to claim 1, characterized in that The obtaining of the load recovery adjustment amount corresponding to the faulty node includes: Acquire a fault voltage and a normal voltage of the fault node and a maximum load recovery adjustment amount of the fault node; Calculating a difference between the fault voltage and the normal voltage according to the fault voltage and the normal voltage of the fault node; Calculating a load recovery adjustment amount of the fault node according to the difference, a maximum load recovery adjustment amount of the fault node and a normal voltage of the fault node; The calculation formula of the load recovery adjustment amount of the fault node is as follows: ， in, Indicates the serial number is The load recovery adjustment amount of the fault node, Indicates the serial number is The maximum load recovery adjustment amount of the fault node, Indicates the serial number is The difference between the fault voltage and the normal voltage of the fault node, Indicates the serial number is The normal voltage of the fault node.

3. The method according to claim 1, characterized in that The step of adjusting the voltage of the faulty node according to the priority order until the voltage of the power supply network system returns to a normal voltage includes: Obtaining a load recovery adjustment amount of the faulty node; adjusting the voltage of the fault node according to the priority order and the load recovery adjustment amount of the fault node; Obtaining a load recovery adjustment amount of each node of the power supply network system after the voltage of the fault node is adjusted; The voltage of each node of the power supply network system is adjusted according to the load recovery adjustment amount of each node of the power supply network system, so that the voltage of the power supply network system is restored to a normal voltage.

4. The method according to claim 3, characterized in that The method further comprises: The total amount of load shedding during the process of limiting the voltage of the power supply network system to recover to a normal voltage is less than a preset total amount of load shedding threshold.

5. The method according to claim 3, characterized in that: The method further comprises: Limiting the load shedding speed of each node of the power supply network system to be greater than a preset load shedding speed lower limit threshold during the process of restoring the voltage of the power supply network system to a normal voltage, and the load shedding speed of each node of the power supply network system to be less than a preset load shedding speed upper limit threshold.

6. The method according to claim 3, characterized in that The method further comprises: In the process of limiting the voltage of the power supply network system to recover to a normal voltage, the voltage fluctuation of each node of the power supply network system is less than a preset fluctuation voltage threshold.

7. The method according to claim 1, characterized in that The obtaining of the fault node of the power supply network system comprises: Based on the convolutional neural network architecture, a fault monitoring model for each node of the power supply network system is established; The fault detection model is used to perform real-time monitoring on each node of the power supply network system to obtain the faulty nodes of the power supply network system.

8. An intelligent fault diagnosis and recovery system for coal mine power supply network, characterized in that: The system comprises: An acquisition module, used for acquiring a fault node of a power supply network system; A determination module, configured to determine a priority order between the faulty nodes based on the characteristics of the faulty nodes; the characteristics of the faulty nodes include the equipment load of the faulty nodes, the contribution of the faulty nodes to the stability of the power supply network, and the load recovery adjustment amount of the faulty nodes; A recovery module, used for adjusting the voltage of the faulty node according to the priority order until the voltage of the power supply network system is restored to a normal voltage; The determining the priority order between the faulty nodes based on the characteristics of the faulty nodes includes: Obtaining a first weight coefficient corresponding to the device load and a second weight coefficient corresponding to the power supply network stability contribution; Calculate a first product of the first weight coefficient and the device load, calculate a second product of the second weight coefficient and the power supply network stability contribution, calculate the sum of the first product and the second product, and obtain a priority evaluation value; Determining the priority order between the faulty nodes according to the order of magnitude between the priority evaluation values; The determining the priority order between the faulty nodes based on the characteristics of the faulty nodes further includes: If the priority evaluation values ​​of some of the faulty nodes are the same, then obtaining the load recovery adjustment amount corresponding to the faulty nodes; The priority order between the corresponding fault nodes is determined according to the order of magnitudes between the load recovery adjustment amounts.

Images