A method, system, equipment, and storage medium for online monitoring of the operating status of secondary equipment in a substation.

Abstract

This invention discloses an online monitoring method, system, equipment, and storage medium for the operating status of secondary equipment in substations, belonging to the field of power system monitoring technology. The method involves acquiring the status categories of each secondary device and constructing a secondary circuit model. The theoretical values ​​for each monitoring point are calculated using the model, and the residual vector is obtained by comparing the theoretical and measured values. The operating status is then determined based on the residual vector. A target fault set is selected by analyzing the similarity between the fault feature vector corresponding to a virtual fault and the residual vector. Within the target fault set, a correlation index for each virtual fault is determined based on the correlation between the disturbance vector and the residual vector, and a suspected fault set is selected based on the correlation index. Within the suspected fault set, the fault intensity of each virtual fault is calculated by constructing a fault location model. Virtual faults with a fault intensity greater than a set intensity value are used as detection results to provide early warnings for the operating status of the secondary equipment.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, specifically to a method, system, equipment, and storage medium for online monitoring of the operating status of secondary equipment in substations. Background Technology

[0002] Substations are key nodes in the operation of power systems. The higher the voltage level of a substation, the greater its role and the higher the requirements for its safety and reliability. The secondary equipment of intelligent substations is the foundation for the safe and stable operation of the power grid and the primary equipment of the substation.

[0003] Currently, patent application CN119557740A discloses a method for fault location of secondary equipment in substations based on a spatiotemporal graph convolutional network model. The method includes: acquiring historical data of the secondary equipment and constructing a fault feature set including operating status information, SV / GOOSE reception status information, and sampled values; sorting the fault feature data according to time series and corresponding them one-to-one with the corresponding secondary equipment faults; extracting and filtering highly correlated fault features to construct a filtered fault feature set; performing multimodal feature fusion processing on the fault feature set data based on time series to obtain fault feature fusion data; and constructing a three-layer 1D-CNN network. An STGCN model with a single-layer GCN network is used to simultaneously learn the spatial and temporal features of the feature map data. The fault feature fusion data is divided into a training set and a test set. The training set is used for model learning, and the test set is used to evaluate the model's performance and accuracy. The trained STGCN model is obtained. A dynamic threshold for the number of fault feature information under adaptive weather conditions is set to detect secondary equipment faults. When a secondary equipment fault is detected, the fault location is diagnosed and located according to the fault inference rules. If the inference rules cannot accurately locate the fault location, the fault feature fusion data is input into the trained STGCN model, and the fault location of the secondary equipment is identified and located through the model.

[0004] The above method filters fault features and constructs fused data based on the filtered fault feature set. It then uses the fused data and a complex STGCN model to detect faults in secondary equipment. However, there are a large number of secondary equipment in the secondary circuit, and when a fault occurs in the secondary circuit, only a small number of secondary equipment are faulty. Collecting fused data from all secondary equipment to detect faults in secondary equipment would undoubtedly greatly increase the computational load of fault detection, resulting in low efficiency in monitoring the operating status of secondary equipment. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention provides a method, system, equipment and storage medium for online monitoring of the operating status of secondary equipment in substations.

[0006] Therefore, the technical problem solved by this invention is: how to solve the problem that when there are a large number of secondary devices in the secondary circuit, but only a small number of secondary devices are faulty when a fault occurs in the secondary circuit, collecting fused data from all secondary devices to realize fault detection would undoubtedly greatly increase the computational load of fault detection, resulting in low monitoring efficiency of the operating status of secondary devices.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an online monitoring method for the operating status of secondary equipment in a substation, comprising,

[0008] The status categories of each secondary device are obtained and a secondary loop model is constructed. The theoretical values ​​of each monitoring point are calculated through the secondary loop model. The residual vector is obtained by comparing the theoretical values ​​with the measured values, and the operating status is determined based on the residual vector.

[0009] The target fault set is selected by similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault;

[0010] In the target fault set, the correlation index of each virtual fault is determined based on the correlation between the disturbance vector and the residual vector, and a set of suspected faults is selected based on the correlation index.

[0011] In the set of suspected faults, the fault intensity of each virtual fault is calculated by constructing a fault location model. Virtual faults with a fault intensity greater than the set intensity value are used as detection results to provide early warning of the operating status of secondary equipment.

[0012] As a preferred embodiment of the online monitoring method for the operating status of substation secondary equipment according to the present invention, the method for obtaining the status category of each secondary equipment and constructing a secondary circuit model includes: obtaining the status category of each secondary equipment through image input; and constructing a numerical model reflecting the actual operating status of the secondary circuit based on the connection relationship and electrical characteristics of the secondary equipment.

[0013] As a preferred embodiment of the online monitoring method for the operating status of substation secondary equipment according to the present invention, the status categories include at least one of the following: pressure plate status, changeover handle status, air switch status, current terminal connection status, voltage terminal connection status, internal cabinet cable connection status, inter-cabinet cable connection status, and optical fiber connection status.

[0014] In a preferred embodiment of the online monitoring method for the operating status of secondary equipment in a substation as described in this invention, the target fault set is filtered by similarity analysis between the fault feature vector and the residual vector corresponding to a virtual fault, including:

[0015] A virtual fault database is pre-built, which includes various fault types of each secondary device;

[0016] By applying perturbations to the virtual faults, the fault feature vectors corresponding to each virtual fault are obtained;

[0017] Based on the similarity calculation between the fault feature vector and the residual vector, the similarity index of each virtual fault is determined.

[0018] The target fault set is selected based on the similarity index.

[0019] As a preferred embodiment of the online monitoring method for the operating status of secondary equipment in a substation as described in this invention, the suspected fault set is selected based on the correlation index, and the correlation index of each virtual fault is normalized and arranged in descending order.

[0020] Calculate the sum of the first K relevance indicators. If the sum of the relevance indicators is less than the indicator set value, increase K by 1 until the sum of the relevance indicators is not less than the indicator set value. Then, take the virtual faults corresponding to the first K relevance indicators as the suspected fault set.

[0021] As a preferred embodiment of the online monitoring method for the operating status of secondary equipment in a substation according to the present invention, wherein: the virtual fault with a fault intensity greater than the intensity set value is used as the detection result, including comparing the fault intensity with the intensity set value;

[0022] The system acquires virtual faults whose fault intensity exceeds the set intensity value and outputs the corresponding secondary equipment and fault type as the detection result.

[0023] This invention provides an online monitoring system for the operating status of secondary equipment in substations.

[0024] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an online monitoring system for the operating status of secondary equipment in a substation, comprising: an acquisition module, used to acquire the status category of each secondary equipment and construct a secondary circuit model;

[0025] The calculation module is used to calculate the theoretical value of each monitoring point through the secondary loop model, compare the theoretical value with the measured value to obtain the residual vector, and determine the operating status based on the residual vector.

[0026] The filtering module is used to filter the target fault set by similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault; it performs the determination of the correlation index of each virtual fault based on the correlation between the perturbation vector and the residual vector, and selects the suspected fault set based on the correlation index.

[0027] The output module is used to obtain fault detection results based on fault intensity and to provide early warnings on the operating status of secondary equipment.

[0028] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the online monitoring method for the operating status of secondary equipment in a substation.

[0029] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the online monitoring method for the operating status of secondary equipment in a substation.

[0030] The beneficial effects of this invention are as follows: This invention automatically obtains the status categories of secondary equipment through image recognition and constructs an accurate secondary loop model. It calculates the residual vector between the theoretical and measured values ​​of each monitoring point, thereby achieving real-time quantitative evaluation of the operating status of the secondary loop. It quickly filters the target fault set through similarity analysis and further filters the suspected fault set based on the correlation index. The hierarchical screening reduces the fault detection range from all virtual faults to at least a few suspected faults, avoiding comprehensive analysis of all secondary equipment and improving fault detection efficiency. It constructs a fault location model in the suspected fault set, accurately calculates the fault intensity of each virtual fault through sparsity constraints and penalty weight mechanisms, and accurately identifies the real faults and their types by combining the intensity setting value, thereby achieving rapid and accurate fault location. Attached Figure Description

[0031] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 The above is a flowchart of an online monitoring method for the operating status of secondary equipment in a substation, provided as an embodiment of the present invention.

[0033] Figure 2 This is a computer equipment diagram of an online monitoring method for the operating status of secondary equipment in a substation, provided as an embodiment of the present invention.

[0034] Figure 3 This is a schematic diagram of the fault intensity of each virtual fault in a candidate fault set of an online monitoring method for the operating status of secondary equipment in a substation, provided in an embodiment of the present invention. Detailed Implementation

[0035] To make the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0036] Example 1, referring to Figures 1-3 This is one embodiment of the present invention, which provides an online monitoring method for the operating status of secondary equipment in a substation, comprising:

[0037] S100: Obtain the status category of each secondary device and construct a secondary loop model. Calculate the theoretical value of each monitoring point through the secondary loop model, compare the theoretical value with the measured value to obtain the residual vector, and determine the operating status based on the residual vector.

[0038] S200: Target fault set is selected by similarity analysis between the fault feature vector and residual vector corresponding to the virtual fault; in the target fault set, the correlation index of each virtual fault is determined based on the correlation between the disturbance vector and the residual vector, and the suspected fault set is selected based on the correlation index.

[0039] S300: In the suspected fault set, the fault intensity of each virtual fault is calculated by constructing a fault location model. Virtual faults with fault intensity greater than the set intensity value are used as detection results to provide early warning of the operating status of secondary equipment.

[0040] It should be noted that substation secondary circuits contain numerous secondary devices, which are interconnected through complex electrical connections to form a complete control, protection, and measurement system. During operation, the status of these secondary devices directly determines the topology and electrical characteristics of the secondary circuits. When one or more secondary devices malfunction, the electrical parameters at each monitoring point will deviate from normal values. Traditional methods require comprehensive testing of all secondary devices, resulting in enormous computational burden and low efficiency.

[0041] Therefore, to address the aforementioned issues in operation monitoring and fault location, steps S100 to S300 are implemented as follows: First, a numerical model that accurately reflects the true operating state of the secondary circuit is constructed. The theoretical values ​​for each monitoring point are calculated using the model and compared with the measured values ​​to generate a residual vector, enabling rapid judgment of the operating state. When an anomaly is detected, a hierarchical screening strategy is adopted. First, similarity analysis is used to quickly filter out the target fault set from all virtual faults. Then, correlation calculation is used to further refine the selection of the suspected fault set, significantly narrowing the fault detection range. Finally, a fault location model is constructed from the suspected fault set to calculate the fault intensity of each virtual fault and compare it with the set value, accurately identifying the real faults and achieving precise online monitoring of the substation's secondary equipment.

[0042] Example 2 is an embodiment of the present invention.

[0043] In this embodiment, step S100 involves obtaining the state category of each secondary device and constructing a secondary loop model, including the following steps A1-A2:

[0044] A1: Obtain the status category of each secondary device through image input.

[0045] Specifically, the status categories include at least one of the following: pressure plate status, changeover handle status, air switch status, current terminal connection status, voltage terminal connection status, internal cabinet cable connection status, inter-cabinet cable connection status, and optical fiber connection status.

[0046] Secondary equipment refers to various devices in a substation used to monitor and control primary equipment, such as relay protection devices, automatic devices, control switches, and measuring instruments. The status category of secondary equipment refers to the real-time status of these devices at a certain moment, and is a key input affecting the operating status of the secondary circuit.

[0047] Suppose that the on-site camera captures a CT circuit engagement plate of a certain protection device in a vertical position, the target detection model will identify it as "closed" and input this status information (e.g., encoded as "1") into the secondary circuit model; the secondary circuit model will then calculate the theoretical current value flowing through the circuit based on this status.

[0048] In this way, by automatically acquiring and inputting the precise status category of the secondary equipment through image recognition, the secondary circuit model can calculate the theoretical values ​​of the electrical parameters of each monitoring point in real time and accurately under the current secondary circuit status.

[0049] The status category of each secondary device is injected into the secondary loop model to calculate the theoretical value of each monitoring point; the theoretical value and the measured value of each monitoring point are compared to obtain the residual vector; if the magnitude of the residual vector is less than the preset value, the operating status is normal; otherwise, fault detection is performed.

[0050] Fault detection includes: calculating the disturbance vector of each virtual fault, which includes various fault types of secondary equipment; defining the inner product between the disturbance vector and the residual vector as the priority of each virtual fault, and selecting a candidate fault set based on the priority; in the candidate fault set, calculating the fault intensity of each virtual fault in the candidate fault set with the goal of minimizing the difference between the virtual residual and the residual vector caused by the virtual fault, and taking the virtual fault with the fault intensity greater than the intensity threshold as the detection result.

[0051] By comparing the theoretical and measured values ​​at each monitoring point, a residual vector is obtained. The monitoring point is the location in the secondary circuit where electrical parameters are monitored in real time, such as the voltage or current input terminals of a specific relay, or the auxiliary contact status of a switch. The location of the monitoring point is manually selected. The measured values ​​are the values ​​of the electrical parameters acquired in real time by sensors deployed at the monitoring points.

[0052] After calculating the theoretical values ​​of the electrical parameters at each monitoring point, they are compared with the measured value vectors collected at the same time. The difference between the two is calculated to obtain the residual vector. The measured value vector includes the measured values ​​of the electrical parameters at each monitoring point, and the residual vector includes the residual values ​​of the electrical parameters at each monitoring point.

[0053] Current moment residual vector Electrical parameters of the monitoring point residual value for:

[0054]

[0055] in, For the current moment residual vector Electrical parameters of the monitoring point The residual value; For the current moment Electrical parameters of monitoring points The measured value; For the current moment Electrical parameters of monitoring points The theoretical value.

[0056] Each element of the residual vector represents the degree of deviation between the theoretical and actual values ​​of an electrical parameter at a monitoring point. Ideally, when the secondary equipment is functioning normally, the residual vector should be a zero vector or a small value within the allowable error range for each residual. When the secondary equipment is faulty, the theoretical and measured values ​​will differ significantly, leading to an increase in the magnitude of the residual vector. Understandably, if there are 10 monitoring points, with monitoring points 4 and 7 having two electrical parameters each, and the remaining monitoring points each having one electrical parameter, then the residual vector will be a 12-row, 1-column vector.

[0057] Assume that monitoring point 5 is the voltage value of a certain device, with a theoretical value of 220V; however, due to a fault in the air switch of the device, the measured value is 0V. Therefore, the value at the corresponding position of this voltage value in the residual vector is -220V.

[0058] In this way, by calculating the degree of deviation between the theoretical value and the actual value, the complex system state is transformed into a specific residual vector that can be accurately quantified.

[0059] Calculating the fault strength of each virtual fault in the candidate fault set includes:

[0060] Construct the objective function, objective function for:

[0061]

[0062] in, The objective function is... The fault intensity vector includes the candidate fault set. The fault intensity of each virtual fault in the system; For the current moment The residual vector; For candidate fault set The matrix formed by the disturbance vectors of each virtual fault; The regularization coefficient is . and Candidate fault sets Virtual fault The penalty weights and fault strengths; the fault strength vector corresponding to the minimum value of the objective function includes the fault strengths of each virtual fault in the candidate fault set; This is a virtual fault.

[0063] A2: Based on the connection relationships and electrical characteristics of secondary equipment, construct a numerical model that reflects the actual operating state of the secondary circuit. It can calculate the theoretical quantities of electrical parameters at each monitoring point through circuit simulation based on the input state category of each secondary equipment. These electrical parameters include at least one of current, voltage, power, and inductance.

[0064] Specifically, if the magnitude of the residual vector is less than a preset value, the operation is normal; otherwise, fault detection is performed.

[0065] The magnitude of the residual vector is calculated. The magnitude reflects the deviation between the real-time state and the normal state of the secondary circuit under the state category of each secondary device. The larger the magnitude, the more the real-time state of the secondary circuit deviates from the normal state.

[0066] The magnitude of the residual vector is compared with a preset value. If the magnitude is less than the preset value, the secondary circuit is considered to be operating normally. Otherwise, if the magnitude exceeds the preset value, it indicates that there is one or more faults in the secondary circuit, and fault detection is performed.

[0067] In this embodiment, step S200 involves filtering the target fault set through similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault, including the following steps B1-B4:

[0068] B1: Pre-build a virtual fault database, which includes all fault types of each secondary device.

[0069] Specifically, a virtual fault refers to any specific type of fault that may occur in any secondary equipment, such as "false tripping of pressure plate No. 1" or "faulty tripping of circuit breaker No. 3". The disturbance vector can be regarded as the fault fingerprint of each virtual fault, which can characterize the changes in electrical parameters at each monitoring point when a virtual fault occurs.

[0070] The preset disturbance is to apply the fault type in the virtual fault to the secondary equipment in the virtual fault. For example, for the virtual fault "the No. 2 changeover switch is erroneously moved from position A to position B", the disturbance operation is to force the state of the switch from A to B in the model.

[0071] The inner product between the disturbance vector and the residual vector is defined as the priority of each virtual fault, and a set of candidate faults is selected based on the priority.

[0072] Since the number of virtual faults is large, in order to improve the efficiency of fault detection, we first quickly screen the virtual faults, calculate the priority of each virtual fault, and select a small set of candidate faults based on the priority, without having to analyze all virtual faults.

[0073] Specifically, the inner product between the disturbance vector and the residual vector is defined as the priority of each virtual fault; virtual fault priority for:

[0074]

[0075] in, For virtual faults Priority; Let be the perturbation vector of the virtual fault. for transpose, For the current moment The residual vector.

[0076] Virtual Fault Priority The physical meaning is the residual vector. The projection of the residual vector onto the direction of the virtual fault disturbance vector. If the residual vector is very similar to the disturbance vector of a virtual fault, it indicates that the residual vector is likely caused by the virtual fault. The inner product of the residual vector and the disturbance vector of the virtual fault will be large, and the corresponding priority will be higher, indicating that the virtual fault is more likely to cause the current system anomaly.

[0077] After calculating the priorities of all virtual faults, a candidate fault set is selected based on these priorities. Specifically, selecting the candidate fault set based on priorities includes: normalizing the priorities of each virtual fault and arranging them in descending order; before calculation... The sum of priorities, in response to a priority sum being less than a priority threshold, will... Increase by 1 until the sum of priorities is not less than the priority threshold, then move the previous priority level to the next level. Each priority level corresponds to a virtual fault, which is used as a candidate fault set.

[0078] in, The initial value is 1, and The value increases by 1 sequentially, with a priority threshold of 0.9.

[0079] In this way, the search scope for virtual faults is greatly narrowed through filtering, reducing the analysis object from all virtual faults to... A candidate fault set consisting of virtual faults.

[0080] B2: By applying perturbations to the virtual faults, the fault feature vectors corresponding to each virtual fault are obtained.

[0081] Specifically, the secondary loop model is run under the baseline condition to obtain the baseline values ​​for each monitoring point;

[0082] Apply a preset disturbance to the secondary equipment corresponding to the virtual fault, keep the state category of other secondary equipment unchanged, run the secondary loop model again, and obtain the disturbance value of each monitoring point;

[0083] Based on the difference between the disturbance value and the baseline value, a fault feature vector of the virtual fault is constructed.

[0084] Virtual Fault priority for:

[0085]

[0086] in, For virtual faults Priority; Let be the perturbation vector of the virtual fault. for transpose, For the current moment The residual vector.

[0087] Penalty weight for virtual faults It is negatively correlated with the priority of the virtual fault, and positively correlated with the correlation between the perturbation vector of the virtual fault and the perturbation vector of other virtual faults in the candidate fault set.

[0088] The lower the priority of a virtual fault, the lower the probability that the residual vector is caused by the virtual fault, and the less suspicious the virtual fault is. Therefore, the virtual fault should be given a greater penalty weight. At the same time, if the perturbation vector of a virtual fault is very similar to the perturbation vectors of other virtual faults, it means that the virtual fault is likely to cause confusion and should also be given a greater penalty to avoid misjudgment.

[0089] Virtual Fault Penalty weight Satisfying the relation:

[0090]

[0091] in, For virtual faults The penalty weight; For virtual faults The variance inflation factor is used to characterize virtual faults. The correlation between the perturbation vector and the perturbation vectors of other virtual faults in the candidate fault set. For virtual faults Priority.

[0092] B3: Based on the similarity calculation between the fault feature vector and the residual vector, determine the similarity index of each virtual fault.

[0093] Specifically, a similarity index for each virtual fault is determined based on the similarity calculation between the fault feature vector and the residual vector. The similarity index is a metric used to quantify the degree of matching between the residual vector and the fault feature vector of each virtual fault. The higher the similarity index, the greater the likelihood that the virtual fault caused the current residual vector.

[0094] The principle behind the similarity index calculation: The fault feature vector of a virtual fault is essentially the change pattern of electrical parameters at each monitoring point when the fault occurs. When a fault actually occurs in the secondary circuit, specific electrical parameter deviations will be generated at each monitoring point, and these deviations constitute the residual vector. If the pattern of the residual vector is highly similar to the fault feature vector pattern of a virtual fault, it indicates that the currently observed residual is likely caused by the virtual fault.

[0095] The similarity index is calculated using the normalized vector inner product. The similarity index for virtual fault i is:

[0096]

[0097] in, Let be the similarity index of virtual fault i, with a value range of [-1, 1]; Let be the perturbation vector of the virtual fault. for transpose, For the current moment The residual vector; For vectors The 2-norm; For vectors The 2-norm.

[0098] This similarity metric is actually the cosine of the angle between two vectors. When the value is close to 1, it indicates that the directions of the fault feature vector and the residual vector are basically consistent, and the virtual fault highly matches the actual fault. When the value is close to 0, it indicates that the two vectors are orthogonal, and the virtual fault is unrelated to the actual fault. When the value is negative, it indicates that the two vectors are in opposite directions, and the virtual fault cannot be an actual fault.

[0099] From a physical perspective, the similarity index reflects the degree to which the fault feature vector of a virtual fault can explain the residual vector. Consider a simple case where the residual vector is entirely caused by a single virtual fault; in this case, the residual vector should be proportional to the fault feature vector of that virtual fault, and the similarity index would be 1. In reality, the residual vector may be generated by the combined effect of multiple virtual faults or be affected by measurement noise; therefore, the similarity index will be less than 1.

[0100] By using similarity metrics, virtual faults similar to the residual vector pattern can be quickly identified. This step is equivalent to a coarse screening of all possible virtual faults, excluding those that are obviously irrelevant, thus significantly narrowing down the scope of subsequent precise analysis. This hierarchical screening strategy ensures both the accuracy of fault detection and improves computational efficiency.

[0101] B4: Filter the target fault set based on similarity index.

[0102] Specifically, the correlation indices of each virtual fault are normalized and then arranged in descending order;

[0103] Let K be the screening quantity variable (K is a positive integer, with an initial value of 1). Calculate the sum of the first K relevance indicators. In response to the sum of the relevance indicators being less than the indicator set value, increase K by 1 until the sum of the relevance indicators is not less than the indicator set value. Then, take the virtual faults corresponding to the first K relevance indicators as the suspected fault set.

[0104] Specifically, the target fault set is selected based on a similarity index. The target fault set is a subset of virtual faults initially selected from all virtual faults that are similar to the residual vector pattern. This subset contains the virtual faults most likely to cause the current fault phenomenon. By selecting the target fault set, the objects of subsequent precise analysis can be reduced from all M virtual faults to dozens or even a dozen virtual faults, significantly reducing computational complexity.

[0105] The selection strategy for the target fault set is as follows: After normalizing the similarity index of each virtual fault, they are arranged in descending order. The purpose of normalization is to eliminate the influence of differences in the amplitude of feature vectors of different faults, so that the similarity indexes can be compared on a uniform scale. The normalized similarity index is as follows:

[0106]

[0107] in, The similarity index is the normalized version of virtual fault i; The similarity index for virtual fault i; This represents the maximum similarity metric among all virtual faults. This represents the minimum similarity metric among all virtual faults. After normalization, The value range is [0, 1]. The most similar virtual fault normalized similarity index is 1, and the least similar is 0.

[0108] The relationship between the target fault set and the suspected fault set: It should be noted that the target fault set is the initial screening result based on similarity indicators, while the suspected fault set is the result of further precise screening based on the target fault set. The screening process from the target fault set to the suspected fault set needs to consider the correlation between virtual faults and the interpretability of the residual vector.

[0109] In the target fault set, although each virtual fault and residual vector has high similarity, there may be a high degree of correlation between different virtual faults. For example, a fault in the misalignment of a pressure plate and another switch fault connected in series with it may have very similar fault feature vectors because both will cause the same branch to break. In this case, although both virtual faults are in the target fault set, only one is actually a real fault.

[0110] Therefore, when screening the set of suspected faults from the set of target faults, it is necessary to further analyze the correlation between the disturbance vector (i.e., the fault feature vector) and the residual vector of each virtual fault. The correlation here considers not only the similarity of the vector directions (which has already been considered by the similarity index), but also the degree of matching of the vector magnitudes and the independence between the virtual faults.

[0111] In this embodiment, the correlation degree is calculated using the method mentioned in claim 1, namely, based on the inner product between the perturbation vector and the residual vector, taking into account the magnitude of the perturbation vector. The correlation index for virtual fault i... for:

[0112]

[0113] in, For virtual faults Priority; Let be the perturbation vector of the virtual fault. for transpose, For the current moment The residual vector.

[0114] Compared to the similarity index, the relevance index places more emphasis on the magnitude of the projection of the residual vector onto the direction of the perturbation vector, rather than just the consistency of the direction.

[0115] For each virtual fault in the target fault set, its correlation index is calculated, and a suspected fault set is selected by accumulating the correlation index according to the method described in claims 5 and 6. Specifically, after normalizing the correlation index of each virtual fault in the target fault set, they are arranged in descending order, and the sum of the first K correlation indices is calculated. When the sum of the correlation indices reaches the set value (e.g., 0.9), the first K virtual faults are taken as the suspected fault set.

[0116] For example, suppose the target fault set obtained by filtering through similarity metrics contains 15 virtual faults, and their normalized similarity metrics are as follows:

[0117] [0.95, 0.88, 0.82, 0.75, 0.68, 0.61, 0.55, 0.48, 0.42, 0.35, 0.28, 0.22, 0.15, 0.09, 0.03]

[0118] The cumulative similarity threshold is set to 0.9.

[0119] Calculate the cumulative similarity: the cumulative normalized similarity of the first virtual fault is 0.95 / total; the first two are (0.95 + 0.88) / total; and so on. Assuming the total is 7.26, the cumulative similarity of the first five virtual faults is:

[0120] (0.95+0.88+0.82+0.75+0.68) / 7.26≈0.56;

[0121] The first 6 are:

[0122] (0.95+0.88+0.82+0.75+0.68+0.61) / 7.26≈0.64;

[0123] The first 7 are:

[0124] (0.95+0.88+0.82+0.75+0.68+0.61+0.55) / 7.26≈0.72;

[0125] The first 8 are approximately 0.78; the first 9 are approximately 0.84; the first 10 are approximately 0.89; and the first 11 are approximately 0.93.

[0126] Therefore, when K=11, the cumulative similarity first exceeds 0.9, and the first 11 virtual faults are output as the target fault set. Then, the correlation index is calculated for these 11 virtual faults to further filter out the suspected fault set for subsequent accurate fault intensity calculation.

[0127] In this embodiment, the virtual fault with a fault intensity greater than the set intensity value in step S300 is used as the detection result, including the following steps C1-C2:

[0128] C1: Compare the fault strength with the strength setting value.

[0129] Specifically, the fault intensity is compared with the intensity setpoint. The fault intensity reflects the confidence level of the actual existence of each virtual fault in the suspected fault set. By comparing it with the intensity setpoint, the actual faults can be identified from the suspected fault set, and those virtual faults that, although they have some correlation with the residual vector, have not actually occurred can be filtered out.

[0130] The fault intensity vector is calculated using the described fault location model. Each of its elements This represents the fault intensity of the virtual fault. The fault intensity value ranges from 0 to 1.

[0131] when A value close to 1 indicates that the virtual fault exists at near full strength, meaning the fault has completely occurred. For example, if a pressure plate should be in the closed position but is actually in the open position, the fault strength of this "pressure plate misalignment" fault is close to 1.

[0132] when A value around 0.5 indicates that the virtual fault exists with partial intensity, possibly corresponding to an intermediate state in the fault's development process. For example, a contact point may have poor contact but not yet completely disconnected; in this case, the intensity of the "poor contact" fault might be 0.6.

[0133] when A value close to 0 indicates that the virtual fault is essentially nonexistent or its contribution to the current residual vector is negligible. Although the virtual fault is in the suspected fault set, after accurate fault intensity calculation, its actual impact is found to be very small.

[0134] The intensity setting is the threshold for determining whether a virtual fault actually exists. The selection of the intensity setting requires a trade-off between fault detection sensitivity and false alarm rate. Setting the intensity setting too low will increase the false alarm rate, causing some faults that do not actually exist to be identified as existing; setting the intensity setting too high will reduce detection sensitivity, potentially missing some real but minor faults.

[0135] In a preferred embodiment, the intensity setpoint is determined using a statistical analysis method. Historical operating data of the substation is used to statistically analyze the fault intensity distribution of each virtual fault under normal operating conditions (i.e., no real faults occurring). Due to measurement noise and model errors, even in a fault-free state, the elements of the fault intensity vector will not be strictly zero, but will exhibit small fluctuations. By statistically analyzing the amplitude of these fluctuations, a reasonable intensity setpoint can be determined.

[0136] Specifically, monitoring data from the substation under normal operating conditions is collected for N consecutive days. The fault intensity vector is calculated M times per day (e.g., once per hour), resulting in N×M sets of fault intensity data. For each virtual fault, the mean and standard deviation of its fault intensity under normal conditions are calculated. Intensity setpoint. It can be set as follows:

[0137]

[0138] in, This represents the average of the mean fault intensity of all virtual faults under normal conditions. is the average of the standard deviations of the fault intensity of all virtual faults under normal conditions; k is the confidence coefficient, usually between 2 and 3. When k=2, it corresponds to a 95% confidence level; when k=3, it corresponds to a 99.7% confidence level.

[0139] Statistical analysis of 30 days of normal operation data of a 220kV substation yielded the following results: , If k=3, then the strength set value is... =0.05 + 3 × 0.08 = 0.29. Considering the need for a certain safety margin in practical applications, the final strength setting value was set to 0.3.

[0140] Layered judgment based on multi-level intensity settings: A multi-level intensity setting is used for stratified judgment, dividing the severity of the fault into multiple levels. For example, three intensity thresholds can be set: For minor faults, Moderate fault threshold, This is the threshold for severe faults.

[0141] When 0.3 < If the value is less than 0.5, it is considered a minor fault, and a low-level warning is issued, suggesting that it be addressed during planned maintenance.

[0142] When 0.5≤ If the value is less than 0.7, it is considered a moderate fault, and a medium-level warning is issued. It is recommended to arrange a special inspection and handling in the near future.

[0143] C2: Acquire virtual faults whose fault intensity exceeds the set intensity value, and output the corresponding secondary equipment and fault type as the detection result.

[0144] Specifically, in the candidate fault set, with the goal of minimizing the difference between the virtual residual and the residual vector caused by the virtual fault, the penalty weight of each virtual fault in the candidate fault set is calculated as the fault intensity of the virtual fault, and the virtual fault with the fault intensity greater than the intensity threshold is taken as the detection result.

[0145] Within the candidate fault set, multiple virtual faults may occur concurrently, or when there is similarity between the perturbation vectors of different virtual faults, it is difficult to accurately distinguish which virtual fault(s) caused the residual vector, which can complicate fault detection. Therefore, the goal is to minimize the difference between the virtual residuals and residual vectors caused by virtual faults in order to accurately calculate the fault intensity of each virtual fault in the candidate fault set.

[0146] Calculating the fault intensity of each virtual fault in the candidate fault set includes: constructing an objective function, the objective function... for:

[0147]

[0148] in, The objective function is... The fault intensity vector includes the candidate fault set. The fault intensity of each virtual fault in the system; For the current moment The residual vector; For candidate fault set The matrix formed by the disturbance vectors of each virtual fault; The regularization coefficient is . and Candidate fault sets Virtual fault The penalty weights and fault strengths; the fault strength vector corresponding to the minimum value of the objective function includes the fault strengths of each virtual fault in the candidate fault set; This is a virtual fault.

[0149] Among these methods, optimization algorithms such as particle swarm optimization and hill climbing can be used to obtain the fault intensity vector corresponding to the minimum value of the objective function. Please refer to [link to relevant documentation]. Figure 3 This is a schematic diagram of the fault intensity of each virtual fault in the candidate fault set according to this embodiment. Figure 3 Taking a real-world example from a 220kV substation, this paper presents the fault intensity calculation results for typical virtual faults in a candidate fault set. Here, VT stands for Voltage Transformer, CT for Current Transformer, pressure plate for relay protection devices used to control the activation and deactivation of protection functions, and circuit breaker for air circuit breakers. Figure 3 "4 VT switch malfunction" indicates that the switching switch of the 4 voltage transformer circuit malfunctioned to the open position; "1 circuit breaker malfunction" indicates that the 1 air switch malfunctioned and disconnected; "3 CT switch malfunction" indicates that the switching switch of the 3 current transformer circuit malfunctioned to the open position.

[0150] Among them, the fault intensity of virtual faults The value is between 0 and 1. A non-zero value indicates that the virtual fault actually exists, and its magnitude represents the confidence level of the virtual fault. For candidate fault set The perturbation vector of each virtual fault is a matrix of size A rows and K columns, where A is the number of elements in the perturbation vector and K is the set of candidate faults. The number of virtual faults in the middle; A K-row, 1-column vector containing the candidate fault set. The fault strength of K virtual faults; Let A be a vector with 1 column and 1 row, used to represent the virtual residuals caused by virtual faults. The goal is to find a fault intensity vector such that the virtual residuals caused by these virtual faults are... with measured value vector The error between them is minimal; Its key role is to make the fault intensity vector To become sparse, that is, to make As many elements as possible should be zero; because at any given time, only a few secondary devices in the secondary circuit actually experience a fault. It is a regularization coefficient greater than zero, used to balance fitting accuracy with the fault strength vector. Sparsity, regularization coefficient The value is 0.5.

[0151] In one embodiment, the penalty weight for virtual faults This indicates the level of attention given to each virtual fault. The penalty weight for a virtual fault is negatively correlated with its priority and, consequently, with the virtual fault itself. The perturbation vector of the candidate fault is positively correlated with the perturbation vectors of other virtual faults in the candidate fault set.

[0152] It is understandable that the lower the priority of a virtual fault, the lower the probability that the residual vector is caused by that virtual fault, and the less suspicious the virtual fault is. Therefore, a greater penalty weight should be given to the virtual fault to avoid classifying a less suspicious virtual fault as a real fault. At the same time, if the perturbation vector of a virtual fault is very similar to the perturbation vectors of other virtual faults, it means that the virtual fault is likely to cause confusion, and a greater penalty should also be given so that the optimization algorithm will prioritize virtual faults that are not easily confused as real faults and avoid misjudgment.

[0153] Specifically, virtual faults Penalty weight Satisfying the relation:

[0154]

[0155] in, For virtual faults The penalty weight; For virtual faults The variance inflation factor is used to characterize virtual faults. The correlation between the perturbation vector and the perturbation vectors of other virtual faults in the candidate fault set. For virtual faults Priority.

[0156] Among them, virtual faults variance inflation factor The method for obtaining it is: constructing a virtual fault. The weighted model is as follows:

[0157]

[0158] in, A weighted model for virtual faults. Virtual faults for candidate fault sets Besides The perturbation vector of another virtual fault, This represents the number of other virtual faults in the candidate fault set. For the first There are several undetermined coefficients; after determining the values ​​of all undetermined coefficients in the weighted model using the least squares method, the determination coefficients of the weighted model are calculated. Virtual fault variance inflation factor for:

[0159]

[0160] It should be noted that, among them, To calculate the coefficient of determination for the weighted model; The value of is in the range [0, 1], indicating that the disturbance vectors of K−1 other virtual faults can explain the disturbance vector. The percentage; if A value close to 1 indicates that the perturbation vector It can be perfectly represented by a linear combination of other perturbation vectors, indicating a significant correlation; if A value close to 0 indicates that the perturbation vector If it cannot be represented by other vectors, it indicates that there is no correlation.

[0161] Finally, the fault intensity of each virtual fault in the candidate fault set is compared with a preset intensity threshold. All virtual faults with a fault intensity greater than the intensity threshold are identified, and the secondary equipment and fault type corresponding to each virtual fault are used as the detection result. The intensity threshold is set to 0.6.

[0162] Thus, from the rapid screening of candidate fault sets to the accurate calculation of fault intensity vectors, the root cause of faults can be accurately identified, enabling high-precision monitoring of the operating status of substation secondary equipment.

[0163] In summary, by acquiring the status categories of each secondary device, constructing an accurate secondary circuit model, calculating residual vectors, filtering suspected fault sets, and calculating fault intensity and comparing it with intensity setpoints, this invention achieves efficient and accurate online monitoring of the operating status of substation secondary equipment. This method can quickly locate the fault location and type when a fault occurs in the secondary circuit, providing maintenance personnel with reliable decision-making basis and improving the operational reliability and maintenance efficiency of substation secondary equipment.

[0164] Example 3, referring to Figure 2 and Figure 3 This embodiment provides an online monitoring system for the operating status of secondary equipment in a substation, including an acquisition module for acquiring the status categories of each secondary device and constructing a secondary circuit model;

[0165] The calculation module is used to calculate the theoretical values ​​of each monitoring point through the secondary loop model, compare the theoretical values ​​with the measured values ​​to obtain the residual vector, and determine the operating status based on the residual vector.

[0166] The filtering module is used to filter the target fault set by similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault; it performs the determination of the correlation index of each virtual fault based on the correlation between the perturbation vector and the residual vector, and selects the suspected fault set based on the correlation index.

[0167] The output module is used to obtain fault detection results based on fault intensity and to provide early warnings on the operating status of secondary equipment.

[0168] This embodiment also provides an electronic device applicable to an online monitoring method for the operating status of secondary equipment in a substation, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the online monitoring method for the operating status of secondary equipment in a substation as proposed in the above embodiment.

[0169] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements an online monitoring method for the operating status of secondary equipment in a substation as proposed in the above embodiments.

[0170] The storage medium proposed in this embodiment and the method for online monitoring of the operating status of secondary equipment in a substation proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0171] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0172] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for online monitoring of the operating status of secondary equipment in a substation, characterized in that: include, The status categories of each secondary device are obtained and a secondary loop model is constructed. The theoretical values ​​of each monitoring point are calculated through the secondary loop model. The residual vector is obtained by comparing the theoretical values ​​with the measured values, and the operating status is determined based on the residual vector. By analyzing the similarity between the fault feature vector and the residual vector corresponding to the virtual fault, a set of target faults is selected. In the set of target faults, the correlation index of each virtual fault is determined based on the correlation between the disturbance vector and the residual vector, and a set of suspected faults is selected based on the correlation index. The step of filtering the target fault set by similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault includes pre-constructing a virtual fault library, which includes each fault type of each secondary device. By applying perturbations to the virtual faults, the fault feature vectors corresponding to each virtual fault are obtained. include, Run the secondary loop model under the baseline condition to obtain the baseline values ​​of each monitoring point; Apply a preset disturbance to the secondary equipment corresponding to the virtual fault, keep the state category of other secondary equipment unchanged, run the secondary loop model again, and obtain the disturbance value of each monitoring point; Based on the difference between the disturbance value and the reference value, a fault feature vector of the virtual fault is constructed; Based on the similarity calculation between the fault feature vector and the residual vector, the similarity index of each virtual fault is determined. The target fault set is selected based on the similarity index; The step of selecting a set of suspected faults based on relevance indices includes normalizing the relevance indices of each virtual fault and arranging them in descending order. Calculate the sum of the first K relevance indicators. If the sum of the relevance indicators is less than the indicator set value, increase K by 1 until the sum of the relevance indicators is not less than the indicator set value. Then, take the virtual faults corresponding to the first K relevance indicators as the suspected fault set. In the set of suspected faults, the fault intensity of each virtual fault is calculated by constructing a fault location model. Virtual faults with a fault intensity greater than the set intensity value are used as detection results to provide early warning of the operating status of secondary equipment.

2. The method for online monitoring of the operating status of secondary equipment in a substation as described in claim 1, characterized in that: The process involves obtaining the status categories of each secondary device and constructing a secondary circuit model, including obtaining the status categories of each secondary device through image input; and constructing a numerical model that reflects the actual operating state of the secondary circuit based on the connection relationships and electrical characteristics of the secondary devices.

3. The method for online monitoring of the operating status of secondary equipment in a substation as described in claim 2, characterized in that: The status categories include at least one of the following: pressure plate status, changeover handle status, air switch status, current terminal connection status, voltage terminal connection status, internal cabinet cable connection status, inter-cabinet cable connection status, and optical fiber connection status.

4. The online monitoring method for the operating status of secondary equipment in a substation as described in claim 3, characterized in that: The virtual faults whose fault intensity is greater than the intensity set value are used as detection results, including: Compare the fault intensity with the intensity setting value; The system acquires virtual faults whose fault intensity exceeds the set intensity value and outputs the corresponding secondary equipment and fault type as the detection result.

5. An online monitoring system for the operating status of secondary equipment in a substation, employing the online monitoring method for the operating status of secondary equipment in a substation as described in any one of claims 1 to 4, characterized in that, include: The acquisition module is used to acquire the status categories of each secondary device and construct a secondary loop model; The calculation module is used to calculate the theoretical value of each monitoring point through the secondary loop model, compare the theoretical value with the measured value to obtain the residual vector, and determine the operating status based on the residual vector. The filtering module is used to filter the target fault set by similarity analysis between the fault feature vector and the residual vector corresponding to the virtual fault; it performs the determination of the correlation index of each virtual fault based on the correlation between the perturbation vector and the residual vector, and selects the suspected fault set based on the correlation index. The output module is used to obtain fault detection results based on fault intensity and to provide early warnings on the operating status of secondary equipment.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the online monitoring method for the operating status of substation secondary equipment according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the online monitoring method for the operating status of substation secondary equipment according to any one of claims 1 to 4.

Images