Safety state monitoring system and method for intelligent box-type substation

By establishing a fault propagation network model and mining weak anomaly features using a sparse autoencoder network, and combining this with a random walk algorithm to simulate abnormal states, the problem of accurately predicting and early identifying the risk of fault propagation in prefabricated substation equipment was solved, thereby improving the sensitivity of equipment condition monitoring and the level of precision in safety management.

CN121073227AActive Publication Date: 2025-12-05湖南科立电气有限公司

Patent Information

Application Number
CN202511619079.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-06
Publication Date
2025-12-05
Estimated Expiration
2045-11-06

AI Technical Summary

Technical Problem

Existing technologies are insufficient to fully reveal the risk of fault propagation between equipment in prefabricated substations, and traditional monitoring methods are not sensitive enough to subtle anomalies, resulting in equipment faults not being identified and spread in a timely manner.

Method used

A fault propagation network model is established, potential propagation paths are identified using dynamic Bayesian probabilistic inference, weak anomaly features are mined using sparse autoencoder networks, and the evolution trend of abnormal states is simulated using a random walk algorithm to generate dynamic distribution curves of equipment risks.

Benefits of technology

It enables accurate prediction of fault propagation risks and early anomaly identification in prefabricated substations, improves the sensitivity and reliability of equipment condition monitoring, and enhances the level of precision in safety management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121073227A_ABST
    Figure CN121073227A_ABST
Patent Text Reader

Abstract

The invention discloses a safety state monitoring system and method for an intelligent box-type substation, and relates to the technical field of power detection, and the method comprises the steps: constructing an equipment fault propagation path set; mining weak abnormal feature information covered by background noise in the monitoring signal; inputting the weak anomaly feature information into a state evolution model of corresponding equipment; simulating a propagation process of the weak anomaly feature information along an equipment fault propagation path set by using a random walk algorithm, and predicting an evolution trend of an equipment abnormal state; generating an equipment risk dynamic distribution curve according to the evolution trend of the abnormal state and the dynamic deviation degree between the preset safety state of the equipment, and determining the risk level of the current safety state of the box-type substation; according to the invention, early abnormal accurate identification and dynamic risk assessment of the safe state of the box-type substation are realized, and the predictability and reliability of safe operation of the substation are effectively improved.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power detection, in particular to a safety state monitoring system and method for an intelligent box-type substation. BACKGROUND

[0002] As a key infrastructure in the power distribution network, the box-type substation is widely used in urban power distribution, industrial parks, and new energy access scenarios. Its safe and stable operation is of great significance to the overall reliability of the power system. In recent years, with the expansion of the power grid, the diversification of loads, and the improvement of equipment automation, the operating environment of the box-type substation has become increasingly complex. Since the equipment is operated in outdoor or semi-outdoor environments for a long time, the equipment operating state is easily affected by external environmental interference and equipment aging, wear and tear, and other factors, thereby increasing the risk of equipment failure.

[0003] Currently, the traditional method for box-type substation equipment state monitoring mainly focuses on single equipment state monitoring, usually only focusing on isolated analysis of equipment state data and basic alarm threshold setting. This single-point monitoring method ignores the interrelation between the state evolution of devices, making it difficult to fully reveal the potential fault propagation or state degradation trend between various devices in the substation, and thus making it difficult to effectively predict and manage potential safety risks. In addition, due to the presence of background noise in the field monitoring data to some extent, the traditional monitoring method lacks sensitivity to weak abnormal features implied in the data, which may lead to early abnormal signals not being identified in time, thereby increasing the possibility of equipment failure and spread. SUMMARY

[0004] The purpose of the present application is to provide a safety state monitoring system and method for an intelligent box-type substation to solve the problems in the background.

[0005] In order to achieve the above-mentioned purpose, the present application provides the following technical solutions: In a first aspect, the present application provides a safety state monitoring method for an intelligent box-type substation, comprising: establishing a fault propagation network model according to the topological relationship of the box-type substation equipment; identifying potential propagation paths in the fault propagation network model using a dynamic Bayesian probability reasoning method, and constructing a device fault propagation path set; reconstructing the device multi-source monitoring signals based on a sparse auto-encoding network to mine weak abnormal feature information in the monitoring signals hidden by background noise; inputting the weak abnormal feature information into the state evolution model of the corresponding device; simulating the propagation process of the weak abnormal feature information along the device fault propagation path set using a random walk algorithm to predict the evolution trend of the device abnormal state; According to the dynamic deviation degree between the evolution trend of the abnormal state and the predetermined safe state of the equipment, a dynamic distribution curve of the risk of the equipment is generated, and a risk level of the current safe state of the box-type substation is determined.

[0006] In a second aspect, the present application provides a safe state monitoring system for an intelligent box-type substation, which is realized based on the safe state monitoring method for the intelligent box-type substation described above, and comprises: A path construction module is configured to establish a fault propagation network model according to the topological relationship of the equipment of the box-type substation, and to identify potential propagation paths in the fault propagation network model by using a dynamic Bayesian probability reasoning method, thereby constructing a set of equipment fault propagation paths. A feature mining module is configured to perform sparse coding reconstruction on the multi-source monitoring signals of the equipment based on a sparse auto-encoding network, so as to mine weak abnormal feature information in the monitoring signals that is masked by background noise. A trend prediction module is configured to input the weak abnormal feature information into a state evolution model of the corresponding equipment, and to simulate the propagation process of the weak abnormal feature information along the set of equipment fault propagation paths by using a random walk algorithm, thereby predicting the evolution trend of the abnormal state of the equipment. A risk analysis module is configured to generate a dynamic distribution curve of the risk of the equipment according to the dynamic deviation degree between the evolution trend of the abnormal state and the predetermined safe state of the equipment, and to determine a risk level of the current safe state of the box-type substation.

[0007] In the above technical solution, the present application has the following technical effects and advantages: The present application effectively identifies potential fault propagation paths between equipment by establishing a fault propagation network model based on dynamic Bayesian probability reasoning, thereby realizing accurate prediction and monitoring of the fault propagation risk of the box-type substation, and overcoming the defect that traditional single-equipment state monitoring cannot reveal the fault cross-equipment propagation risk, and improving the comprehensiveness and forward-looking nature of equipment operation risk prediction.

[0008] The present application successfully mines weak abnormal feature signals that are difficult to be discovered by traditional monitoring methods and are masked by background noise by using a sparse auto-encoding network to perform sparse coding reconstruction on the multi-source monitoring signals of the equipment, thereby realizing accurate identification and extraction of early weak abnormal states of the equipment, avoiding the problem that potential abnormal states of the equipment cannot be discovered in time and develop into serious faults, and improving the sensitivity and reliability of equipment state monitoring.

[0009] This invention introduces the extracted weak anomaly feature information into a random walk algorithm to simulate state propagation, dynamically predicts the evolution trend of abnormal equipment states, and further determines the risk level based on the dynamic deviation between the predicted trend and the predetermined safe state of the equipment. This enables real-time and dynamic assessment and early warning of the overall safety status risk of the prefabricated substation, effectively improving the precision of substation safety management and accident prevention capabilities. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0011] Figure 1 This is a flowchart of a safety status monitoring method for intelligent prefabricated substations according to the present invention; Figure 2 This is a framework diagram of a safety status monitoring system for intelligent prefabricated substations according to the present invention. Detailed Implementation

[0012] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided to make the description of this application more complete and comprehensive, and to fully convey the concept of the exemplary embodiments to those skilled in the art. The drawings are merely illustrative illustrations of this application and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0013] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of the exemplary embodiments disclosed in this application. However, those skilled in the art will recognize that the technical solutions disclosed in this application can be practiced with one or more specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of the disclosure of this application. Example 1

[0014] like Figure 1 As shown in the figure, this embodiment discloses a method for monitoring the safety status of an intelligent prefabricated substation, including: S101: Establish a fault propagation network model according to the topological relationship of the box-type substation equipment; use a dynamic Bayesian probability reasoning method to identify potential propagation paths in the fault propagation network model, and construct a set of equipment fault propagation paths; In an optional implementation, the set of equipment fault propagation paths is constructed, including: Based on the topological connection relationship of the equipment nodes in the fault propagation network model, the initial state transition structure between the equipment nodes is determined, and an initial Bayesian network model is established to obtain an initial state transition probability matrix of the equipment nodes; It should be noted that the fault propagation network model in this embodiment is a directed graph structure established according to the electrical connection and communication connection topological structure of the actual equipment in the box-type substation. Specifically, each node represents a specific equipment in the substation, and typical equipment includes but is not limited to: transformer, high-voltage circuit breaker, low-voltage distribution cabinet, protection relay, and intelligent sensor, etc. The directed edge of the network represents the direction of the fault propagation path, for example: transformer→high-voltage circuit breaker, high-voltage circuit breaker→low-voltage distribution cabinet, protection relay→high-voltage circuit breaker, etc.

[0015] It should be further noted that in this embodiment, the state of each equipment node is divided into three typical states: normal, abnormal, and fault, for example, taking the high-voltage circuit breaker node as an example: Normal: Each monitoring index of the equipment is within the rated range; Abnormal: The equipment has an abnormal index, such as a reduced insulation resistance, but does not reach the fault threshold of equipment shutdown; Fault: The equipment has a significant fault, such as insulation breakdown, and the equipment is shut down.

[0016] Based on the above state definition and network topological structure, an initial Bayesian network model is established, and an initial state transition probability matrix is obtained , the specific form is as follows:

[0017] In the formula, each probability element is preliminarily determined by historical operation experience or expert experience, and the specific value is determined according to actual historical data.

[0018] Based on the historical fault data of the equipment, a probability likelihood function of state transition is constructed, and the conditional expectation of the state transition probability of the equipment is derived to obtain an initial estimate of the state transition probability; In specific implementation, the historical fault data of the equipment in this embodiment refers to the long-term accumulated historical fault records of each equipment in the box-type substation during operation, and typical data includes: specific time of equipment fault occurrence, duration, fault mode, fault propagation equipment range, and actual number of state transitions of each equipment.

[0019] Based on the historical data, a likelihood function of the state transition probability is constructed, and the specific formula is as follows:

[0020] In the formula, L (Θ) represents the likelihood function; Θ represents the state transition probability parameter to be estimated; represents the i th historical sample; represents the state of the device at the t th moment of the i th historical sample; represents the total number of historical samples; represents the duration length of the observation of a single historical sample.

[0021] Then, the conditional expectation function of the state transition probability is obtained through Bayesian probability derivation:

[0022] In the formula, Θ represents the state transition probability parameter to be estimated; represents the probability parameter estimation value obtained in the i th iteration, which is used as the input initial value of the next iteration of the EM algorithm. The initial estimation value is used as the input of the algorithm, and the maximum expectation algorithm is used for multiple iteration calculation of the state transition probability matrix, and the convergence of the likelihood function is determined to obtain the optimized state transition probability matrix;

[0023] Specifically, the multiple iteration calculation of the state transition probability matrix by using the maximum expectation algorithm and the determination of the convergence of the likelihood function include: According to the initial estimation value of the state transition probability, an alternating iteration process of the expectation step and the maximization step is performed to update the numerical value of the device state transition probability; In implementation, the specific calculation formula of the expectation step (E-step) is as follows:

[0024] In the formula, represents the expected transition times of the state to the state in the i th iteration process; represents the actual observed transition times of the state to the state . The specific calculation method of the maximization step (M-step) is as follows:

[0025]

[0026] The above steps are alternately performed to update the numerical value of the state transition probability.

[0027] ​​​​​The likelihood function value is recalculated after each iteration, and the adjacent two calculation results are compared to determine whether the likelihood function meets the convergence condition; In a specific implementation, whether the difference between the likelihood functions obtained by adjacent two iterations is lower than a set convergence threshold is taken as the convergence judgment basis, and the specific formula is represented as: , wherein the threshold is The threshold is determined by actual experimental data.

[0028] When the likelihood function meets the set convergence condition, the iteration process is stopped, and the optimized state transition probability matrix is obtained. In a specific implementation process, the optimized state transition probability matrix is:

[0029] Each matrix element is the state transition probability obtained after the EM algorithm converges.

[0030] A Monte Carlo simulation process is performed according to the optimized state transition probability matrix, a fault propagation path that meets a set probability threshold is screened out, and a device fault propagation path set is obtained. In a specific implementation, a Monte Carlo random simulation method is used to perform a large number of device state transition random simulation analogs by using the optimized state transition probability matrix. For example, after 10,000 random simulations, the occurrence probability of each fault propagation path is counted: , wherein is the probability of the occurrence of the fault propagation path, is the number of occurrences of the path, is the total number of simulations; and A probability threshold is further set according to actual requirements (which can be determined according to experimental data). Paths with a probability higher than the threshold are regarded as effective propagation paths, and a device fault propagation path set is finally formed, which serves as a basis for subsequent safety state analysis.

[0031] S102: Sparse coding reconstruction is performed on the device multi-source monitoring signals based on a sparse auto-encoding network to mine weak abnormal feature information in the monitoring signals that is masked by background noise. In an optional implementation, the sparse coding reconstruction of the device multi-source monitoring signals includes: A multi-dimensional time series data set of a device operating state is constructed from the multi-source monitoring signals, and an initial feature dimension and a feature space are determined. It should be noted that the multi-source monitoring signals in this embodiment include, but are not limited to, various types of equipment operation monitoring data such as transformer winding temperature, high-voltage circuit breaker operation count, partial discharge signals of electrical equipment, bus voltage and current signals, and temperature and humidity sensing signals of the internal environment of the transformer substation; the above signals are collected in real time by various sensors to form a multi-source signal set of equipment operation status.

[0032] Specifically, based on the different sampling periods and data types of the monitored signals, each signal is synchronized to a unified timestamp, and a multi-dimensional time series dataset is constructed in units of fixed duration (e.g., 5 minutes). Each time series data sample is specifically defined as follows: ;in, For the first Each dimension corresponds to a one-dimensional monitoring signal vector, representing the time series data of monitoring signals for a single type of device. This refers to the number of signal dimensions, i.e., the number of categories of multi-source monitored signals, such as temperature, partial discharge, voltage, current, etc.

[0033] The initial feature dimension and feature space are the data dimensions and the value space of each dimension, as mentioned above. For example, the transformer winding temperature takes values ​​in the real number range (e.g., The partial discharge signal is the value of the discharge pulse frequency (e.g., ).

[0034] Based on the distribution characteristics of multidimensional time series data, the sparse regularization parameters of the sparse autoencoder network are determined, and the initial parameter set of the network is established accordingly. In practical implementation, the sparse autoencoder network (SAE) is used to achieve noise reduction, reconstruction, and weak anomaly feature extraction of multi-source monitoring signals. Sparse regularization parameters (e.g., sparsity parameters) Regularization factor The method for determining the value is as follows: based on the actual distribution characteristics of multidimensional time series data, it is determined through the distribution of a large number of samples.

[0035] Specifically, the loss function (Loss) of a sparse autoencoder network is defined as:

[0036] In the formula: Indicates the first The input signal of each sample, For the corresponding reconstructed output; The total number of samples; , The first term represents the network weights and biases; the second term is the weight decay term to prevent overfitting; the third term is the sparsity regularization term, where... divergence Defined as:

[0037] wherein, is the expected sparsity value, represents the average activation value of the th hidden neuron; , are regularization parameters determined by experimental data.

[0038] The network initial parameter set includes layer weight matrices and bias vectors, and the parameter specific initialization method, for example, the Xavier method is used for initialization.

[0039] The method of combining forward propagation and reverse iteration optimization is used to perform sparse compression coding processing on the multi-dimensional time series data by using the network initial parameter set, and a preliminary sparse feature code is obtained. Specifically, the sparse compression coding processing on the multi-dimensional time series data by using the network initial parameter set comprises: The network initial parameter set is input into the sparse auto-encoding network, and the preliminary coding features of the multi-dimensional time series data are calculated by forward propagation operation; In specific implementation, the sparse auto-encoding network of the embodiment includes an input layer, a hidden coding layer and an output decoding layer, wherein the number of neurons in the hidden coding layer is lower than the dimension of the input layer, so as to achieve the purpose of data compression coding. Given the input multi-dimensional time series data sample X, the forward propagation calculation content is: The hidden coding layer activation output is: ; The output decoding layer reconstruction result is: ; wherein, , are the weight matrix and the bias vector from the input layer to the hidden coding layer, , are the weight matrix and the bias vector from the hidden coding layer to the output layer, is a Sigmoid or ReLU activation function.

[0040] The sparse coding feature vector of the hidden coding layer output is calculated by the above process .

[0041] Based on the difference between the preliminary coding features and the original monitoring signal, a sparse regularization loss function is established and the loss gradient value is calculated; In specific implementation, the sparse regularization loss function is defined as:

[0042] Wherein, the first term is the reconstruction error term, the second term is the regularization term to prevent overfitting, and the third term is the sparse constraint term.

[0043] Based on the above loss function, the gradient of the loss function relative to each parameter of the network is calculated, and the gradient calculation method is, for example, back propagation (BP algorithm), and the specific gradient calculation content is as follows: Output layer error term: ; Hidden encoding layer error term: ; Further, the gradient of each parameter is obtained:

[0044] In the formula: is the derivative of the corresponding activation function.

[0045] According to the loss gradient value, the network parameters are updated, and the reverse iteration optimization is performed to obtain the preliminary sparse feature coding of the multi-dimensional time series data; In specific implementation, the gradient descent algorithm or its improved algorithm (such as Adam optimizer) is used to update the network parameters: Taking the Adam optimization algorithm as an example, the parameter update rule is: ; In the formula: is the learning rate, respectively represent the first moment and the second moment estimate value of the gradient, is a smoothing term to prevent the denominator from being zero.

[0046] Through the above reverse iteration optimization process, multiple iterations are performed until convergence, and the optimized sparse auto-encoding network parameters are obtained, and then the preliminary sparse feature coding of the multi-dimensional time series data is obtained.

[0047] According to the residual distribution between the preliminary sparse feature coding and the original monitoring signal, the effective residual interval is extracted and the feature enhancement reconstruction is performed to mine the weak abnormal feature information hidden by the background noise; In specific implementation, first, the residual between the original signal and the preliminary sparse feature reconstruction signal is calculated: ; Wherein, represents the multi-dimensional monitoring signal residual signal.

[0048] Further, according to the statistical distribution characteristics of the residual, the residual signal whose residual exceeds the preset threshold interval (the threshold is determined by actual historical data or experiment) is selected as the suspected weak abnormal feature interval, denoted as: ; The effective residual signal interval is subjected to feature enhancement reconstruction, and the specific enhancement method is, for example, wavelet transform or empirical mode decomposition (EMD) processing on the effective residual interval. Taking wavelet transform as an example: The effective residual signal interval is subjected to feature enhancement reconstruction, and the specific enhancement method is, for example, wavelet transform or empirical mode decomposition (EMD) processing on the effective residual interval. Taking wavelet transform as an example: Discrete wavelet transform (DWT) is performed to extract high-frequency detail coefficients; The high-frequency coefficients are subjected to threshold denoising and enhancement processing (for example, soft threshold denoising method); The enhanced weak abnormal signal features are reconstructed by inverse wavelet transform.

[0049] Through the above processing, the weak abnormal feature signals hidden by background noise can be further highlighted, so as to facilitate subsequent device abnormal state prediction and risk assessment.

[0050] S103: inputting the weak abnormal feature information into a state evolution model of the corresponding device; simulating the propagation process of the weak abnormal feature information along the set of device fault propagation paths by using a random walk algorithm, and predicting the evolution trend of the device abnormal state; In an optional embodiment, the evolution trend of the predicted device abnormal state comprises: Based on the set of device fault propagation paths, the initial propagation weight of the device nodes in each fault path is calculated, and an initial random walk weight matrix is established; It should be noted that the set of fault propagation paths is obtained from step S101, and is specifically represented as a plurality of fault propagation paths, each path being composed of a sequence of device nodes, for example, path examples: Path 1: transformer → high-voltage circuit breaker → low-voltage distribution cabinet; Path 2: busbar → high-voltage circuit breaker → relay protection device.

[0051] The initial propagation weight calculation method is specifically: the initial propagation probability weight between nodes is determined according to the frequency of actual fault propagation of the device nodes in the historical data or expert experience. Specifically, for any adjacent node pair on each path, the initial propagation weight is defined as: wherein, represents the number of fault propagations from node to node in the historical data or expert experience statistics; and the denominator represents the sum of the fault propagation numbers from node to all subsequent nodes.

[0052] The initial propagation weights between the device nodes in all paths are calculated according to the above method, and an initial random walk weight matrix is constructed, and the matrix form is: denotes the total number of device nodes in the network, and the matrix element denotes the initial probability weight of failure propagation between nodes.

[0053] According to the initial state distribution of the weak abnormal feature information, a path sampling process of random walk is performed to generate a state transition sample set of the abnormal feature information; In a specific implementation, the embodiment first determines the initial state distribution of the abnormal information according to the weak abnormal feature information extracted in step S102 . For example, the abnormal feature information can be represented as the abnormal signal strength of a specific device, and the initial state probability distribution is defined as: ; wherein, denotes the device node The probability of abnormality at the initial abnormality time, and the initial probability is determined by the feature enhancement result in step S102.

[0054] Subsequently, taking the initial probability distribution as the starting point, path sampling is performed under the guidance of the random walk weight matrix by the random walk method. For example, each time the walk extracts a starting abnormal node from the initial distribution, determines the next propagation node according to the initial weight matrix, and gradually performs state transition until the set walk length is reached.

[0055] The path sequence generated each time the random walk is performed is a state transition sample of the abnormal feature information, and after a large number (for example, 100 times) of random walks are performed, a state transition sample set of the abnormal feature information is obtained, which is used for subsequent optimization of the random walk weight matrix.

[0056] According to the state transition sample set, the initial random walk weight matrix is optimized and adjusted to obtain an optimized random walk weight matrix; Specifically, the optimization and adjustment of the initial random walk weight matrix includes: performing clustering analysis on the state transition sample set to obtain clustering results of different state transition modes; In a specific implementation, the embodiment first performs clustering analysis on the state transition sample set generated in the above, and the clustering method, for example, adopts a K-means clustering or DBSCAN density clustering algorithm.

[0057] Exemplarily, taking the K-means clustering method as an example, first, feature extraction is performed on each state transition sample sequence, for example, the frequency of occurrence of each node in the sequence or the number of times of occurrence of a specific node sequence is used as a clustering feature: Specifically, the th sample feature vector is defined as: ; in the formula, For nodes In the state transition sequence The frequency or number of times it appears in; Subsequently, K-means clustering was performed using the aforementioned feature vectors to obtain a set of different state transition patterns: ,in, For the first Clustering The number of clusters is determined by experiments or statistical data.

[0058] Based on the clustering results, the dominant path for state transition between device nodes is determined, and the corresponding elements in the random walk weight matrix are adjusted according to the dominant path. In specific implementation, this embodiment defines the state transition path with the highest percentage of samples within each cluster as the dominant path of that cluster. For example, taking the first... Cluster For example, its advantageous path is defined as: Statistical clusters The frequency of occurrence of each state transition path within the system; The path that appears most frequently is selected as the dominant path, denoted as: ; Subsequently, the elements of the random walk weight matrix are adjusted based on the statistical frequency of each dominant path. The specific adjustment method is as follows: If a node pair in a certain advantageous path The frequency of occurrence was significantly higher than that of the initial random weights. If the set value is not met, the weight will be increased accordingly; If the value is lower than the initial setting, the weight should be reduced appropriately.

[0059] The specific weight adjustment formula can be defined as follows: In the formula: For node pairs in the clustering results The frequency of occurrence of the dominant path; Average frequency for all node pairs; It is an adjustment factor greater than zero, determined based on actual data.

[0060] The initial random walk weight matrix is ​​optimized and adjusted using the method described above, resulting in the optimized random walk weight matrix. .

[0061] The adjusted random walk weight matrix is ​​subjected to probability normalization to obtain the optimized random walk weight matrix. In practice, after obtaining the adjusted random walk weight matrix as described above, this step further performs probability normalization on each element of the matrix, specifically defined by the following formula: wherein, is the node to the node the optimized random walk weight value; the denominator is the node the sum of the adjusted weights to all subsequent nodes; represents the total number of nodes.

[0062] After the above normalization processing, the final optimized random walk weight matrix is obtained, which satisfies the normalization condition of the probability matrix and serves as the basis for subsequent random walk simulation of abnormal feature propagation.

[0063] The propagation process of abnormal feature information is simulated using the optimized random walk weight matrix, so as to predict the evolution trend of the device abnormal state. In implementation, based on the optimized random walk weight matrix obtained above , the random walk method is used for random simulation again to simulate the dynamic propagation process of abnormal feature information in the device network, and the specific implementation process is as follows: First, according to the initial distribution state of weak abnormal features , the starting node of abnormal feature propagation is determined; Second, according to the optimized random walk weight matrix , the next propagation node of abnormal features is determined by random sampling according to the following transition probability in each propagation process: wherein, is the transition probability of abnormal feature information from the current node to the next node . After the above method is executed for multiple times (such as 100 times), the frequency of each device node reached by abnormal feature information is counted, and the propagation trend of abnormal feature information in the future period of time is predicted according to the frequency, and then the overall evolution trend of the device abnormal state is predicted.

[0064] In another optional implementation, the prediction of the evolution trend of the device abnormal state further includes: Based on the random walk algorithm, the stability of the abnormal feature information propagation path is analyzed to obtain the path stability coefficient of each device node. In specific implementation, the stability of the abnormal feature propagation path is defined as the frequency of repeated appearance of the same propagation path of abnormal features in multiple random simulation processes, which represents the stability degree of abnormal features propagating along a specific path. The specific implementation method for calculating the stability coefficient is as follows: First, the random walk simulation process is executed to count the repeated number of abnormal feature information on each propagation path. Secondly, according to the frequency of each node appearing on a specific path, the path stability coefficient of the node is defined , and the specific calculation formula is as follows: ; in the formula, represents the number of times of the node appearing in all path simulation results; represents the total number of all random simulation propagation paths. The higher the path stability coefficient is, the stronger the stability of the abnormal feature propagation through the node is.

[0065] According to the path stability coefficient, the propagation probability weight of the abnormal feature information is adjusted to obtain a corrected abnormal feature propagation probability. In a specific implementation, the step further adjusts the aforementioned optimized random walk weight matrix , and corrects the propagation weight corresponding to each node according to the path stability coefficient obtained above. The specific correction formula is as follows: ; in the formula, is the corrected abnormal feature propagation probability weight; is the path stability coefficient of the node ; and is a weight correction factor, which is determined through actual experiments.

[0066] After the above correction processing, probability normalization is performed again to ensure that the propagation probability meets the basic condition of the probability matrix: ; wherein the denominator is the sum of the corrected weights of the node to all subsequent nodes.

[0067] The random walk simulation process is performed again using the corrected abnormal feature propagation probability to predict the evolution trend of the device abnormal state. In a specific implementation, the random walk simulation process is performed again based on the corrected random walk weight matrix , and the specific process is the same as that in the above embodiment: According to the abnormal feature initial state distribution , a starting node is selected; and according to the corrected propagation probability , state transition is performed to perform random walk simulation. After sufficient simulation, the arrival probability of the abnormal feature propagation of each device node is statistically analyzed to obtain the final evolution trend of the device abnormal state.

[0068] Compared with the first mode, the simulation result corrected by the path stability coefficient can more accurately reflect the actual abnormal propagation process, is more consistent with the physical law and historical statistical law of the actual device abnormal propagation, and thus improves the accuracy and stability of the prediction of the evolution trend of the device abnormal state.

[0069] S104: generating a dynamic distribution curve of the risk of the box-type substation according to the dynamic deviation degree between the evolution trend of the abnormal state and the predetermined safe state of the equipment, and determining the risk level of the current safe state of the box-type substation; In an optional implementation, the determining of the risk level of the current safe state of the box-type substation comprises: According to the evolution trend of the abnormal state of the equipment, the initial deviation degree between each equipment node and the predetermined safe state is determined, and a set of initial deviation indexes is obtained. In a specific implementation, the predetermined safe state of the embodiment is defined as a typical state parameter range of the equipment in normal operation, for example, a normal reference value interval of parameters such as the temperature of the transformer winding, the local discharge frequency of the equipment, and the number of switch actions of the circuit breaker.

[0070] First, the evolution trend of the abnormal state of the equipment predicted in step S103 is quantified as a predicted abnormal state parameter value of each equipment node, for example: The predicted value of the temperature of the transformer winding is 85℃; The predicted value of the number of actions of the circuit breaker is 1500 times; The predicted value of the local discharge frequency is 800Hz.

[0071] Secondly, the difference between the predicted abnormal state parameter value of each node and the predetermined safe state parameter value is calculated, which is defined as the initial deviation degree The specific calculation formula is: In the formula: represents the predicted abnormal state parameter value of the node represents the predetermined safe state parameter value (or the upper limit of the normal value) of the node The initial deviation degree set is represented as: wherein, is the total number of equipment nodes in the box-type substation.

[0072] A risk probability distribution model of the equipment node is established by using the set of initial deviation indexes. In a specific implementation, based on the determined set of initial deviation indexes the risk probability of each equipment node is calculated. Typically, the calculation of the risk probability in the embodiment can adopt an exponential mapping mode, and the specific definition formula is: In the formula: represents the risk probability of the equipment node , and the value range is [0, 1]; represents the initial deviation degree of the equipment node ; and is a risk sensitivity coefficient, and the specific value is determined according to historical operation data or experience.

[0073] The above formula means that the greater the degree of deviation of the equipment, the higher the risk probability; when the degree of deviation is 0 (i.e., the equipment state is completely consistent with the predetermined safe state), the risk probability is 0.

[0074] In the above manner, the risk probability distribution model of the equipment node is established, and is specifically expressed as a risk probability set: .

[0075] In combination with the risk probability distribution of each equipment node, the equipment risk dynamic distribution curve of the overall box-type substation is generated. In a specific implementation, the risk probability distribution set of the equipment node obtained is used in combination with the topological structure or equipment logical layout of the box-type substation to generate the overall risk dynamic distribution curve of the box-type substation based on the risk probability of the equipment node.

[0076] Specifically, the horizontal axis of the overall risk dynamic distribution curve of the box-type substation is the equipment node or node position serial number, and the vertical axis is the risk probability value of the corresponding node; further, the risk probability curve can also be drawn after each equipment node is sorted according to the actual layout.

[0077] Exemplarily, the risk dynamic distribution curve is expressed as: Horizontal coordinate: transformer, high-voltage circuit breaker, low-voltage distribution cabinet, relay protection device, etc. Vertical coordinate: risk probability value corresponding to each node ; The risk dynamic distribution curve is obtained, and the risk probability of each key equipment node of the current box-type substation can be clearly displayed.

[0078] According to the risk distribution of the equipment risk dynamic distribution curve, the risk level of the current safety state of the box-type substation is determined. In a specific implementation, the equipment risk dynamic distribution curve obtained is further analyzed, and the safety risk level of the overall box-type substation is determined through the risk probability distribution characteristics of the equipment node. The specific implementation is as follows: First, the equipment risk level division standard is defined, for example, the risk level of the equipment node can be determined according to the numerical interval of the risk probability value For example, the equipment node risk level is as follows: When , the node risk level is "safe"; When , the node risk level is "general risk"; When , the node risk level is "high risk"; When , the node risk level is "serious risk".

[0079] Secondly, according to the risk level statistics of all equipment nodes in the box transformer substation, the comprehensive risk level evaluation rule of the overall box transformer substation is determined, and a typical mode is as follows: If all key equipment nodes are in the "safe" state, the overall risk level of the box transformer substation is determined as "safe"; If a small number (for example, no more than 20%) of nodes are in the "general risk" state, and the other nodes are safe, the overall risk level is determined as "general risk"; If there is a node risk level of "high risk", or more than 20% of the nodes reach the "general risk" level, the overall risk level is determined as "high risk"; If any node risk level reaches "serious risk", the overall risk level of the box transformer is determined as "serious risk".

[0080] In the above manner, the safety risk level of the current box transformer substation can be accurately determined, so that subsequent risk control measures can be taken in time to ensure the safe operation of the box transformer.

[0081] Embodiment 2 As shown in Figure 2 The embodiment does not detail part as shown in embodiment 1, and the embodiment discloses a safety state monitoring system for intelligent box transformer substation, which comprises: The path construction module 201 is used for establishing a fault propagation network model according to the equipment topology relationship of the box transformer substation; and a dynamic Bayesian probability inference method is used to identify the potential propagation path in the fault propagation network model, and a set of equipment fault propagation paths is constructed; The feature mining module 202 is used for sparse coding reconstruction of the equipment multi-source monitoring signal based on a sparse self-encoding network, so as to mine weak abnormal feature information in the monitoring signal which is covered by background noise; The trend prediction module 203 is used for inputting the weak abnormal feature information into a state evolution model of the corresponding equipment; and a random walk algorithm is used to simulate the propagation process of the weak abnormal feature information along the set of equipment fault propagation paths, so as to predict the evolution trend of the equipment abnormal state; The risk analysis module 204 is used for generating a dynamic distribution curve of the equipment risk according to the dynamic deviation degree between the evolution trend of the abnormal state and the predetermined safety state of the equipment, and determining the risk level of the current safety state of the box transformer substation.

[0082] The above formulas are all dimensionless values, and the formulas are obtained by collecting a large amount of data to simulate the latest real situation, and the preset parameters, weights and threshold values in the formula are set by the person skilled in the art according to the actual situation.

[0083] The foregoing merely illustrates some exemplary embodiments of the application, and it will be appreciated that those skilled in the art will be able to devise various modifications without departing from the spirit and scope of the application. The appended drawings and description are illustrative only, and are not intended to be limiting.

Claims

1. A method for security state monitoring of a smart box-type substation, characterized by, The method comprises the following steps: A fault propagation network model is established according to the topological relationship of the box-type substation equipment; potential propagation paths in the fault propagation network model are identified by using a dynamic Bayesian probability inference method, and a set of equipment fault propagation paths is constructed; A sparse auto-encoding network is used to perform sparse coding reconstruction on the multi-source monitoring signals of the equipment, so as to mine weak abnormal feature information in the monitoring signals that is covered by background noise; The weak abnormal feature information is input into a state evolution model of the corresponding equipment; a random walk algorithm is used to simulate the propagation process of the weak abnormal feature information along the set of equipment fault propagation paths, and the evolution trend of the abnormal state of the equipment is predicted; A dynamic distribution curve of the risk of the equipment is generated according to the dynamic deviation between the evolution trend of the abnormal state and the predetermined safe state of the equipment, and the risk level of the current safe state of the box-type substation is determined.

2. The method for safety state monitoring of a smart cubicle substation according to claim 1, characterized in that, The set of equipment fault propagation paths is constructed by: Based on the topological connection relationship of the equipment nodes in the fault propagation network model, the initial state transition structure between the equipment nodes is determined, and an initial Bayesian network model is established to obtain an initial state transition probability matrix of the equipment nodes; A probability likelihood function of state transition is constructed based on historical fault data of the equipment, and a conditional expectation of the state transition probability is derived to obtain an initial estimate of the state transition probability; The initial estimate value is used as the input of the algorithm, and the expectation-maximization algorithm is used to perform multiple iterations of the state transition probability matrix, and the convergence of the likelihood function is determined to obtain an optimized state transition probability matrix; A Monte Carlo simulation process is performed according to the optimized state transition probability matrix, and fault propagation paths that meet a set probability threshold are selected to obtain the set of equipment fault propagation paths.

3. The method for safety state monitoring of a smart cubicle substation according to claim 2, characterized in that, The multiple iterations of the state transition probability matrix by the expectation-maximization algorithm and the determination of the convergence of the likelihood function comprise: Based on the initial estimate of the state transition probability, an alternating iteration process of the expectation step and the maximization step is performed to update the value of the equipment state transition probability; After each iteration, the likelihood function value is recalculated, and the results of adjacent two calculations are compared to determine whether the likelihood function meets the convergence condition; When the likelihood function meets the set convergence condition, the iteration process is stopped, and the optimized state transition probability matrix is obtained.

4. The method for safety state monitoring of a smart cubicle substation according to claim 3, characterized in that, The sparse coding reconstruction of the multi-source monitoring signals of the equipment comprises: A multi-dimensional time series data set of the equipment operating state is constructed from the multi-source monitoring signals, and the initial feature dimension and feature space are determined; Based on the distribution characteristics of the multi-dimensional time series data, the sparse regularization parameters of the sparse auto-encoding network are determined, and the initial network parameter set is established accordingly; A method combining forward propagation and backward iteration optimization is used to perform sparse compression coding processing on the multi-dimensional time series data by using the initial network parameter set, and a preliminary sparse feature code is obtained; According to the residual distribution between the preliminary sparse feature code and the original monitoring signal, the effective residual interval is extracted and feature enhancement reconstruction is performed to mine the weak abnormal feature information covered by the background noise.

5. The method for safety state monitoring of a smart cubicle substation according to claim 4, characterized in that, The sparse compression coding processing of the multi-dimensional time series data by using the initial network parameter set comprises: inputting the network initial parameter set into a sparse self-encoding network, calculating preliminary encoding features of the multi-dimensional time series data through a forward propagation operation; based on the difference between the preliminary encoding features and the original monitoring signals, establishing a sparse regularization loss function and calculating a loss gradient value thereof; updating the network parameters according to the loss gradient value, performing a reverse iterative optimization, and obtaining preliminary sparse feature encoding of the multi-dimensional time series data.

6. The method for safety state monitoring of a smart cubicle substation according to claim 5, characterized in that, The evolution trend of the device abnormal state is predicted, including: based on the set of device fault propagation paths, calculating the initial propagation weight of the device nodes in each fault path, and establishing an initial random walk weight matrix; based on the initial state distribution of the weak abnormal feature information, performing a path sampling process of random walk to generate a set of state transition samples of the abnormal feature information; optimizing and adjusting the initial random walk weight matrix according to the set of state transition samples to obtain an optimized random walk weight matrix; using the optimized random walk weight matrix to simulate the propagation process of the abnormal feature information, thereby predicting the evolution trend of the device abnormal state.

7. The method for safety state monitoring of a smart cubicle substation according to claim 6, characterized in that, The optimization and adjustment of the initial random walk weight matrix includes: performing cluster analysis on the set of state transition samples to obtain clustering results of different state transition modes; determining the dominant path of state transition between device nodes according to the clustering results, and adjusting the corresponding elements in the random walk weight matrix according to the dominant path; performing probability normalization processing on the adjusted random walk weight matrix to obtain the optimized random walk weight matrix.

8. The method for safety state monitoring of a smart cubicle substation according to claim 7, characterized in that, The evolution trend of the device abnormal state is predicted, including: based on the random walk algorithm, analyzing the stability of the abnormal feature information propagation path to obtain a path stability coefficient of each device node; adjusting the propagation probability weight of the abnormal feature information according to the path stability coefficient to obtain a corrected abnormal feature propagation probability; re-executing the random walk simulation process using the corrected abnormal feature propagation probability to predict the evolution trend of the device abnormal state.

9. The method for safety state monitoring of a smart cubicle substation according to claim 8, characterized in that, The risk level of the current safety state of the box-type substation is determined, including: determining the initial deviation degree between each device node and the predetermined safety state according to the evolution trend of the device abnormal state to obtain an initial deviation index set; using the initial deviation index set to establish a risk probability distribution model of the device nodes; combining the risk probability distribution of each device node to generate a device risk dynamic distribution curve of the box-type substation as a whole; determining the risk level of the current safety state of the box-type substation according to the risk distribution of the device risk dynamic distribution curve.

10. A safety state monitoring system for a smart box-type substation, realized based on the safety state monitoring method for a smart box-type substation according to any one of claims 1-9, characterized in that, including: a path construction module, configured to establish a fault propagation network model according to the device topology relationship of the box-type substation, and identify potential propagation paths in the fault propagation network model by using a dynamic Bayesian probability reasoning method to construct a set of device fault propagation paths; a feature mining module, configured to perform sparse encoding reconstruction on device multi-source monitoring signals based on a sparse self-encoding network to mine weak abnormal feature information in the monitoring signals which is covered by background noise; a trend prediction module configured to input the weak abnormal feature information into a state evolution model of the corresponding device, simulate a propagation process of the weak abnormal feature information along a set of device fault propagation paths by using a random walk algorithm, and predict an evolution trend of an abnormal state of the device; a risk analysis module configured to generate a dynamic distribution curve of a risk of the device according to a dynamic deviation degree between the evolution trend of the abnormal state and a predetermined safe state of the device, and determine a risk level of a current safe state of the box-type substation.

Citation Information

Patent Citations

  • Livestock and poultry behavior pattern abnormity identification method based on distribution estimation algorithm and residual network

    CN120472533A

  • Transformer substation monitoring method and system based on Internet of Things

    CN120498112A

  • Intelligent abnormal operation monitoring and positioning method for polypropylene cable

    CN120891322A

  • Quantum, biological, computer vision, and neural network systems for industrial internet of things

    US20230176557A1

Cited By

  • Intelligent box transformer equipment fault prediction method and system based on big data

    CN122366756A