Power system fault protection system and method based on multi-modal data fusion

Abstract

The invention discloses an electric power system fault protection system and method based on multi-modal data fusion, and relates to the technical field of electric power systems.The method comprises the steps that an electric power equipment fault analysis model is constructed, and historical electric power fault data is collected to train the electric power equipment fault analysis model; inputting the electrical measurement abnormal condition, the equipment state abnormal condition and the environment meteorological abnormal condition into the power equipment fault analysis model to judge abnormal equipment, positioning a fault point of the abnormal equipment, analyzing the abnormal condition of a module corresponding to the fault point based on the electrical measurement data of the fault point of the abnormal equipment, and determining the abnormal condition of the module. Meanwhile, the influence degree of equipment outage caused by the fault of the fault point is evaluated, the equipment fault severity degree is analyzed based on the abnormal condition of the module corresponding to the fault point and the influence degree of equipment outage caused by the fault of the fault point, and an equipment first-aid repair sequence is generated based on the equipment fault severity degree; the accuracy of power failure degree judgment and the emergency response capability of a power system are improved.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a power system fault protection system and method based on multimodal data fusion. Background Technology

[0002] As the scale of power systems expands and their structures become increasingly complex, the reliability and safety of power equipment, as a core component of the power grid, are directly related to the stable development of the entire society. Therefore, efficient and accurate fault diagnosis and maintenance of power equipment is a key focus of the power industry.

[0003] Currently, power equipment fault management and maintenance technologies still have many shortcomings. For example, formulating fixed periodic maintenance plans based on equipment manufacturers' recommendations or historical operating experience, while reducing the failure rate to some extent, ignores the actual health status and individual differences of the equipment, easily leading to over-maintenance or under-maintenance, resulting in resource waste and potential risks. On the other hand, the maintenance mode of installing sensors on the equipment to monitor its key status parameters in real time and issuing alarms and scheduling maintenance when the parameters exceed preset thresholds usually involves different systems collecting and analyzing data independently. This makes it difficult to integrate multi-dimensional information for accurate diagnosis, especially unable to identify complex faults caused by the coupling of multiple factors, and difficult to accurately locate specific faulty components or fault points. When multiple devices alarm simultaneously or have potential risks, existing technologies usually rely on human experience to determine the priority of emergency repairs, failing to quantify and assess the potential impact of equipment failures. This may result in critical equipment not being handled in a timely manner, causing suboptimal resource allocation and greater losses.

[0004] To address the aforementioned problems, this invention provides a power system fault protection system and method based on multimodal data fusion. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a power system fault protection system and method based on multimodal data fusion. This invention ensures that limited emergency repair resources can be prioritized for handling the most urgent and impactful faults, thereby maximizing the efficiency of resource allocation and improving the accuracy of power fault severity assessment and the emergency response capability of the power system.

[0006] To achieve the above objectives, this invention provides a power system fault protection method based on multimodal data fusion, comprising the following specific steps: S1. Collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and analyze abnormal electrical measurement conditions, abnormal equipment status conditions and abnormal environmental meteorological conditions of power equipment; S2. Construct a power equipment fault analysis model and collect historical power fault data to train the power equipment fault analysis model; S3. Input electrical measurement anomalies, equipment status anomalies, and environmental meteorological anomalies into the power equipment fault analysis model to identify the equipment that is malfunctioning and locate the fault point of the malfunctioning equipment. S4. Based on the electrical measurement data of the fault point of the abnormal equipment, analyze the abnormal status of the module corresponding to the fault point, and at the same time assess the impact of the equipment shutdown caused by the fault point. S5. Analyze the severity of equipment failure based on the abnormal status of the module corresponding to the fault point and the impact of equipment downtime caused by the fault point, and generate an equipment repair sequence based on the severity of the equipment failure.

[0007] Preferably, step S1 includes the following specific steps: S11. Install high-precision current transformers and voltage transformers to collect electrical measurement data during the operation of power equipment. The electrical measurement data includes current and voltage. Obtain abnormal current values ​​based on the ratio of current imbalance rate to safe current imbalance rate. The current imbalance rate is obtained by dividing the difference between the maximum phase current and the minimum phase current by the average phase current. Obtain abnormal voltage values ​​based on the ratio of voltage deviation to safe voltage deviation. The voltage deviation is obtained by dividing the absolute value of the difference between the actual voltage and the nominal voltage by the nominal voltage. Obtain abnormal electrical measurement values ​​by weighted summation of current abnormal values ​​and voltage abnormal values. S12. Install temperature sensors and vibration sensors to collect equipment status monitoring data during the operation of power equipment. The equipment status monitoring data includes equipment temperature and vibration frequency. Obtain abnormal equipment temperature values ​​based on the ratio of equipment temperature to safe equipment temperature, obtain abnormal vibration values ​​based on the ratio of vibration frequency to safe vibration frequency, and obtain abnormal equipment status values ​​by weighted summation of abnormal equipment temperature values ​​and abnormal vibration values. S13. Install a micro-weather station to collect environmental meteorological data during the operation of power equipment. The environmental meteorological data includes environmental temperature and environmental humidity. Obtain abnormal values ​​of environmental temperature based on the ratio of environmental temperature to safe environmental temperature, obtain abnormal values ​​of environmental humidity based on the ratio of environmental humidity to safe environmental humidity, and obtain abnormal values ​​of environmental meteorology based on the weighted sum of abnormal values ​​of environmental temperature and abnormal values ​​of environmental humidity.

[0008] Preferably, step S2 includes the following specific steps: S21. Collect historical power failure data, which includes historical electrical measurement data, historical equipment status monitoring data, historical environmental meteorological data, and historical failure records. The historical failure records include the historical failure time point, historical failure equipment ID, and historical failure type. S22. Based on historical power fault data, obtain historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies. Create a time-indexed label sequence with an initial value of 0, representing normal equipment operation. Obtain the faulty equipment ID and fault time point corresponding to the historical electrical measurement anomaly, historical equipment status anomaly, and historical environmental meteorological anomaly values ​​at each time point. In the label sequence, with the fault occurrence time point as the center, obtain the corresponding preset time window and mark the corresponding fault type. The preset time window can be set to 10 minutes before the fault time to 5 minutes after the fault time. Set the label value within this time window from 0... Modify the code to correspond to the fault type. Mark the remaining time as normal for the equipment. Define a review window and sliding step size. Starting from the beginning of the label sequence, slide forward one time point at a time. Extract historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies that include the review window duration as input features. Use the label corresponding to the end time of the review window as the output target, i.e., use the fault type corresponding to the end time of the window as the output target. Repeat this process until all time points have been traversed. For example, if the review window is set to 30 minutes and the sliding step size is set to 1 minute, starting from the beginning of the timeline, the first sample obtains all historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies at time points [t0, t1, ..., t29] as input features. Take the label at time point t29 as the output target. Slide the window for 1 minute. The next sample obtains the data for [t1, t2, ..., t30] and the label for t30. Repeat this process until the end of the timeline. S23. Obtain the feature dataset and label dataset and divide them into a 70% training set, a 15% validation set, and a 15% test set. Define an input layer that matches the shape of the training dataset. Add two LSTM layers, specifying 128 units per layer. Set `return_sequences=False` so that it only returns the output of the last time step. Add one or more Dense layers after the LSTM layers, using the ReLU activation function to learn feature combinations. Add a final Dense layer with the number of units equal to the total number of fault types + 1. The fully connected layer converts the feature vector output by the LSTM into dimensions corresponding to the number of categories. The Softmax activation function converts these dimensions into a probability distribution, with the sum of the probabilities of all categories being 1. Use the Softmax activation function to output the number of nodes and the probability of each fault category. The number of nodes equals the total number of fault types + 1, where 1 represents the device. Normally, for example, if the device has overload, short circuit, or insulation aging, the number of output nodes is 4, corresponding to the probabilities of normal, overload, short circuit, and insulation aging, respectively. EarlyStopping is set to monitor the loss on the validation set. If the loss does not decrease for 5 consecutive epochs, training is stopped, and the optimal weights are restored. ModelCheckpoint is set to save the best-performing model on the validation set after each epoch. The model.fit() function is called, passing in the training set data, validation set data, setting the number of training epochs to 100, the batch size to 64, and a callback function. Training is started, and the changes in loss and accuracy during training are observed. After training is complete, the best-performing model file saved by ModelCheckpoint is loaded, model.evaluate() is called, inputting the test set data, and the final loss and accuracy are recorded.

[0009] Preferably, step S3 includes the following specific steps: S31. Input the abnormal values ​​of electrical measurements, abnormal values ​​of equipment status, and abnormal values ​​of environmental meteorology into the power equipment fault analysis model, output the number of nodes and the probability of each fault category, obtain the ID of abnormal equipment with a number of nodes greater than 1, and set the fault category corresponding to the maximum probability of the fault category as the fault category of the abnormal equipment. S32. Obtain the set of modules contained in the abnormal device, and obtain the failure probability of each module based on the failure probability calculation formula, wherein the failure probability calculation formula is: ,in, To assign an influence weight to module mᵢ, iterate through the faulty devices, obtain the number of other modules directly connected to module mᵢ and the maximum number of connections among all modules, and then obtain the influence weight of module mᵢ by dividing the number of other modules directly connected to module mᵢ by the maximum number of connections among all modules. Let mᵢ be the prior fault probability of module mᵢ. The normalization constant represents the total probability of equipment failure. It is calculated by traversing all possible combinations of module states and summing the weighted probabilities of each combination leading to system failure. The module with the highest failure probability is set as the failure point of the abnormal equipment.

[0010] Preferably, step S4 includes the following specific steps: S41. Obtain electrical measurement data of the fault point of the abnormal equipment, including the current and voltage at the fault point; S42. Obtain the abnormal value of the fault point current based on the ratio of the fault point current to the module rated current, obtain the abnormal value of the fault point voltage based on the ratio of the fault point voltage to the module rated voltage, and obtain the abnormal evaluation value of the module corresponding to the fault point by weighted summation of the abnormal value of the fault point current and the abnormal value of the fault point voltage. S43. Obtain the duration of continuous loss caused by equipment downtime. The duration of continuous loss is obtained by the historical average power restoration time of the equipment. Abnormal values ​​of the duration of continuous loss are obtained based on the ratio of the duration of continuous loss to the maximum tolerable interruption duration. The maximum tolerable interruption duration is obtained through a short-term interruption tolerance test. S44. Obtain data on equipment affected by the power outage. This data includes the total number of affected equipment, the data generation rate of each affected equipment, and the distance between each affected equipment and the out-of-operation equipment. Obtain anomalies in the scale of the power outage based on the formula for calculating anomalies in the scale of the power outage. The formula for calculating anomalies in the scale of the power outage is as follows: ,in, The total number of devices affected by the power outage. The basic data impact of the nth power outage-affected device is obtained by multiplying the data generation rate of the affected device by the duration of the power outage. The distance influence weight between the nth power outage-affected device and the out-of-operation device is calculated using the following formula: Obtain The distance attenuation coefficient is obtained from historical power outage data, and the calculation formula is as follows: Obtain For a distance of The intensity of the impact was obtained by calculating the percentage decrease in the equipment's processing rate based on the actual measured values. The initial impact intensity at the fault location is theoretically set to a baseline value of 1. This represents the actual distance between the fault point and the observation point. The distance between the equipment affected by the nth power outage and the equipment that is out of service; S45. Obtain the assessment value of the impact of equipment shutdown caused by the failure point by weighted summation of outliers in the scale of power outage impact and outliers in the duration of loss.

[0011] Preferably, step S5 includes the following specific steps: S51. Obtain the equipment failure severity assessment value by weighted summation of the abnormal assessment value of the module corresponding to the fault point and the impact assessment value of the equipment downtime caused by the fault point. S52. Sort the equipment failure severity assessment values ​​in descending order to generate an equipment repair sequence.

[0012] This invention also provides a power system fault protection system based on multimodal data fusion, comprising: The equipment operation anomaly analysis module is used to collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and to analyze electrical measurement anomalies, equipment status anomalies and environmental meteorological anomalies of power equipment; The fault analysis model building module is used to build power equipment fault analysis models and collect historical power fault data to train the power equipment fault analysis models. The fault location module is used to input abnormal electrical measurement conditions, abnormal equipment status conditions, and abnormal environmental weather conditions into the power equipment fault analysis model to identify the abnormal equipment and locate the fault point of the abnormal equipment. The Fault Point Impact Analysis Module is used to analyze the abnormal status of the corresponding module based on the electrical measurement data of the fault point of the abnormal equipment, and at the same time assess the degree of impact of the equipment downtime caused by the fault point. The emergency repair sequence analysis module is used to analyze the severity of equipment failure based on the abnormal status of the module corresponding to the fault point and the impact of equipment downtime caused by the fault point, and to generate an emergency repair sequence based on the severity of the equipment failure.

[0013] The present invention also proposes an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the power system fault protection method based on multimodal data fusion as described above.

[0014] The present invention also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the power system fault protection method based on multimodal data fusion as described above.

[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention collects electrical measurement data, equipment status monitoring data, and environmental meteorological data during the operation of power equipment; analyzes abnormal electrical measurements, abnormal equipment status, and abnormal environmental meteorological conditions of the power equipment; constructs a power equipment fault analysis model; collects historical power fault data to train the power equipment fault analysis model; inputs abnormal electrical measurements, abnormal equipment status, and abnormal environmental meteorological conditions into the power equipment fault analysis model to identify abnormal equipment and locate the fault point of the abnormal equipment; analyzes the abnormal status of the corresponding module based on the electrical measurement data of the fault point; assesses the impact of the equipment shutdown caused by the fault point; analyzes the severity of the equipment fault based on the abnormal status of the corresponding module and the impact of the equipment shutdown caused by the fault point; and generates an equipment repair sequence based on the severity of the equipment fault, ensuring that limited repair resources can be prioritized for the most urgent and impactful fault handling, maximizing the efficiency of resource allocation, improving the accuracy of power fault severity judgment, and enhancing the emergency response capability of the power system. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the 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.

[0017] Figure 1 This is a schematic diagram of the power system fault protection method based on multimodal data fusion according to the present invention; Figure 2 This is a schematic diagram of the S1 flow of the power system fault protection method based on multimodal data fusion according to the present invention; Figure 3 This is a schematic diagram of the S4 process of the power system fault protection method based on multimodal data fusion according to the present invention; Figure 4 This is a schematic diagram of the overall framework of the power system fault protection system based on multimodal data fusion according to the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0019] Please see Figure 1This invention provides a power system fault protection method based on multimodal data fusion, comprising the following specific steps: S1. Collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and analyze abnormal electrical measurement conditions, abnormal equipment status conditions and abnormal environmental meteorological conditions of power equipment; Please see Figure 2 In this embodiment, S1 includes the following specific steps: S11. Install high-precision current transformers and voltage transformers to collect electrical measurement data during the operation of power equipment. The electrical measurement data includes current and voltage. Obtain the current anomaly value based on the ratio of the current imbalance rate to the safe current imbalance rate. The current imbalance rate is obtained by dividing the difference between the maximum phase current and the minimum phase current by the average phase current. Obtain the voltage anomaly value based on the ratio of the voltage deviation to the safe voltage deviation. The voltage deviation is obtained by dividing the absolute value of the difference between the actual voltage and the nominal voltage by the nominal voltage. Obtain the electrical measurement anomaly value by weighted summation of the current anomaly value and the voltage anomaly value. S12. Install temperature sensors and vibration sensors to collect equipment status monitoring data during the operation of power equipment. The equipment status monitoring data includes equipment temperature and vibration frequency. Obtain abnormal equipment temperature values ​​based on the ratio of equipment temperature to safe equipment temperature, obtain abnormal vibration values ​​based on the ratio of vibration frequency to safe vibration frequency, and obtain abnormal equipment status values ​​by weighted summation of abnormal equipment temperature values ​​and abnormal vibration values. S13. Install a micro-weather station to collect environmental meteorological data during the operation of power equipment. The environmental meteorological data includes environmental temperature and environmental humidity. Obtain abnormal values ​​of environmental temperature based on the ratio of environmental temperature to safe environmental temperature, obtain abnormal values ​​of environmental humidity based on the ratio of environmental humidity to safe environmental humidity, and obtain abnormal values ​​of environmental meteorology based on the weighted sum of abnormal values ​​of environmental temperature and abnormal values ​​of environmental humidity.

[0020] Based on industry standards, the safe current imbalance rate can be set to 10%, the safe voltage deviation can be set to ±5%, the safe ambient temperature can be set to 40℃, and the safe ambient humidity can be set to 80%. The safe equipment temperature and safe vibration frequency are obtained by collecting new equipment operation data or according to equipment design specifications. Changes in current and voltage reflect initial faults such as insulation aging and poor contact. High temperature and high humidity affect the natural heat dissipation and forced cooling effect of the equipment, accelerate insulation aging and metal corrosion, and easily cause short circuits. Multi-source anomaly fusion can identify the coupled effects of subtle changes in power equipment and environmental deterioration, and can improve the accuracy of power equipment anomaly detection.

[0021] S2. Construct a power equipment fault analysis model and collect historical power fault data to train the power equipment fault analysis model; In this embodiment, S2 includes the following specific steps: S21. Collect historical power failure data, which includes historical electrical measurement data, historical equipment condition monitoring data, historical environmental meteorological data, and historical failure records. The historical failure records include the historical failure time point, historical failure equipment ID, and historical failure type. S22. Based on historical power fault data, obtain historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies. Create a time-indexed label sequence with an initial value of 0, representing normal equipment operation. Obtain the faulty equipment ID and fault time point corresponding to the historical electrical measurement anomaly, historical equipment status anomaly, and historical environmental meteorological anomaly values ​​at each time point. In the label sequence, with the fault occurrence time point as the center, obtain the corresponding preset time window and mark the corresponding fault type. The preset time window can be set to [10 minutes before the fault time -> 5 minutes after the fault time]. Record the label values ​​within this time window. The code is changed from 0 to the corresponding fault type. The remaining time marks the equipment as normal. A review window and sliding step size are defined. Starting from the beginning of the label sequence, each time point is slid forward by one time point. Historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies containing the review window duration are extracted as input features. The label corresponding to the end time of the review window is used as the output target, that is, the fault type corresponding to the end time of the window is used as the output target. This process is repeated until all time points are traversed. For example, the review window is set to 30 minutes and the sliding step size is set to 1 minute. Starting from the beginning of the timeline, the first sample obtains all historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies at time points [t0, t1, ..., t29] as input features. The label at time point t29 is taken as the output target. The window slides for 1 minute. The next sample obtains the data of [t1, t2, ..., t30] and the label of t30. This process is repeated until the end of the timeline. S23. Obtain the feature dataset and label dataset and divide them into a 70% training set, a 15% validation set, and a 15% test set. The training set is used for model learning, the validation set is used to adjust hyperparameters and prevent overfitting during training, and the test set is used to finally evaluate model performance. Define an input layer that matches the shape of the training dataset. Add two LSTM layers, specifying the number of units in each layer as 128. Set `return_sequences=False` so that it only returns the output of the last time step. Add one or more Dense layers after the LSTM layers, using the ReLU activation function to learn feature combinations. Add a final Dense layer with the number of units equal to the total number of fault types + 1. The fully connected layer converts the feature vector output by the LSTM into dimensions corresponding to the number of categories. The Softmax activation function converts these dimensions into a probability distribution, with the sum of the probabilities of all categories being 1. Use the Softmax activation function to output the number of nodes and the probability of each fault category. The number of nodes equals the total number of fault types plus 1, where 1 represents normal equipment operation. For example, if the equipment has overload, short circuit, and insulation aging, the output node count is 4, corresponding to the probabilities of normal operation, overload, short circuit, and insulation aging, respectively. EarlyStopping is set to monitor the loss on the validation set. If the loss does not decrease for 5 consecutive epochs, training is stopped and the optimal weights are restored. ModelCheckpoint is set to save the best-performing model on the validation set after each epoch. The model.fit() function is called, passing in the training set data, validation set data, setting the number of training epochs to 100, the batch size to 64, and a callback function to start training. The changes in loss and accuracy during training are observed. After training, the best-performing model file saved by ModelCheckpoint is loaded, and model.evaluate() is called, inputting the test set data, and the final loss and accuracy are recorded.

[0022] S3. Input electrical measurement anomalies, equipment status anomalies, and environmental meteorological anomalies into the power equipment fault analysis model to identify the equipment that is malfunctioning and locate the fault point of the malfunctioning equipment. In this embodiment, S3 includes the following specific steps: S31. Input the abnormal values ​​of electrical measurements, abnormal values ​​of equipment status, and abnormal values ​​of environmental meteorology into the power equipment fault analysis model, output the number of nodes and the probability of each fault category, obtain the ID of abnormal equipment with a number of nodes greater than 1, and set the fault category corresponding to the maximum probability of the fault category as the fault category of the abnormal equipment. S32. Obtain the set of modules contained in the abnormal device, and obtain the failure probability of each module based on the failure probability calculation formula, wherein the failure probability calculation formula is: ,in, To determine the influence weight of module mᵢ, we can iterate through the faulty devices, obtain the number of other modules directly connected to module mᵢ and the maximum number of connections among all modules, and then obtain the influence weight of module mᵢ by dividing the number of other modules directly connected to module mᵢ by the maximum number of connections among all modules. The prior failure probability of module mᵢ can be obtained through historical failure probability statistics or MTBF (Mean Time Between Failures). The normalization constant represents the total probability of equipment failure, ensuring that the sum of the posterior probabilities of all modules is 1. The calculation method is to traverse all possible combinations of module states and sum the weighted probabilities of each combination leading to system failure. The module corresponding to the maximum failure probability is set as the failure point of the abnormal equipment.

[0023] S4. Based on the electrical measurement data of the fault point of the abnormal equipment, analyze the abnormal status of the module corresponding to the fault point, and at the same time assess the impact of the equipment shutdown caused by the fault point. Please see Figure 3 In this embodiment, S4 includes the following specific steps: S41. Obtain electrical measurement data of the fault point of the abnormal equipment, including the current and voltage at the fault point; S42. Obtain the abnormal value of the fault point current based on the ratio of the fault point current to the module rated current, obtain the abnormal value of the fault point voltage based on the ratio of the fault point voltage to the module rated voltage, and obtain the abnormal evaluation value of the module corresponding to the fault point by weighted summation of the abnormal value of the fault point current and the abnormal value of the fault point voltage. S43. Obtain the continuous loss time caused by equipment downtime. The continuous loss time is obtained through the historical average power restoration time of the equipment. Abnormal values ​​of the continuous loss time are obtained based on the ratio of the continuous loss time to the maximum tolerable interruption time. The maximum tolerable interruption time is obtained through short-term interruption tolerance test. By simulating short-term power outage, the equipment response time, function retention capability and recovery time are recorded. For example, the maximum duration for which the equipment can maintain core functions after power outage is tested. S44. Obtain data on equipment affected by the power outage. This data includes the total number of affected equipment, the data generation rate of each affected equipment, and the distance between each affected equipment and the out-of-operation equipment. Obtain anomalies in the scale of the power outage based on the formula for calculating anomalies in the scale of the power outage. The formula for calculating anomalies in the scale of the power outage is as follows: ,in, The total number of devices affected by the power outage. The basic data impact of the nth power outage-affected device is obtained by multiplying the data generation rate of the affected device by the duration of the power outage. The distance influence weight between the nth power outage-affected device and the out-of-operation device is calculated using the following formula: Obtain The distance attenuation coefficient is obtained from historical power outage data, and the calculation formula is as follows: Obtain For a distance of The intensity of the impact was obtained by calculating the percentage decrease in the equipment's processing rate based on the actual measured values. The initial impact intensity at the fault location is theoretically set to a baseline value of 1. This represents the actual distance between the fault point and the observation point. The distance between the nth power outage-affected device and the out-of-operation device is given. The number of devices affected by the power outage is obtained through an electrical single-line diagram. The electrical single-line diagram clearly shows the topology of the power system through a simplified single-line representation. The faulty device is located in the single-line diagram, and its electrical connection point is identified. Through the topology of the single-line diagram, the upstream power supply and downstream load of the faulty device are traced. According to the power supply path, the directly affected device is the device directly connected to the fault point, and the indirectly affected device is the device powered through the directly affected device. The number of directly affected and indirectly affected devices is counted. S45. Obtain the assessment value of the impact of equipment shutdown caused by the failure point by weighted summation of outliers in the scale of power outage impact and outliers in the duration of loss.

[0024] S5. Analyze the severity of equipment failure based on the abnormal status of the module corresponding to the fault point and the impact of equipment downtime caused by the fault point, and generate an equipment repair sequence based on the severity of the equipment failure.

[0025] In this embodiment, S5 includes the following specific steps: S51. Obtain the equipment failure severity assessment value by weighted summation of the abnormal assessment value of the module corresponding to the fault point and the impact assessment value of the equipment downtime caused by the fault point. S52. Sort the equipment failure severity assessment values ​​in descending order to generate an equipment repair sequence.

[0026] In this embodiment, the weights are obtained using the coefficient of variation method. The steps are as follows: calculate the mean and standard deviation of each indicator, then divide the standard deviation of each indicator by its mean to obtain the coefficient of variation of that indicator, which represents the degree of dispersion of the indicator data. Then, add the coefficients of variation of each indicator together, and finally calculate the proportion of each coefficient of variation to the total to obtain the weight coefficient of each indicator.

[0027] Please see Figure 4The present invention also provides a power system fault protection system based on multimodal data fusion, including: an equipment operation anomaly analysis module, used to collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and analyze electrical measurement anomalies, equipment status anomalies and environmental meteorological anomalies of power equipment; The fault analysis model building module is used to build power equipment fault analysis models and collect historical power fault data to train the power equipment fault analysis models. The fault location module is used to input abnormal electrical measurement conditions, abnormal equipment status conditions, and abnormal environmental weather conditions into the power equipment fault analysis model to identify the abnormal equipment and locate the fault point of the abnormal equipment. The Fault Point Impact Analysis Module is used to analyze the abnormal status of the corresponding module based on the electrical measurement data of the fault point of the abnormal equipment, and at the same time assess the degree of impact of the equipment downtime caused by the fault point. The emergency repair sequence analysis module is used to analyze the severity of equipment failure based on the abnormal status of the module corresponding to the failure point and the impact of equipment downtime caused by the failure point, and to generate an emergency repair sequence based on the severity of the equipment failure.

[0028] The power system fault protection system based on multimodal data fusion provided by this invention adopts the power system fault protection method based on multimodal data fusion in the above embodiments, which can solve the technical problem. The beneficial effects of the power system fault protection system based on multimodal data fusion provided by this invention are the same as those of the power system fault protection method based on multimodal data fusion provided in the above embodiments. Moreover, other technical features of the power system fault protection system based on multimodal data fusion are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0029] The present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the power system fault protection method based on multimodal data fusion in the above embodiments.

[0030] The electronic device provided by the present invention employs the power system fault protection method based on multimodal data fusion in the above embodiments, which can solve the technical problem. The beneficial effects of the electronic device provided by the present invention are the same as those of the power system fault protection method based on multimodal data fusion provided in the above embodiments, and other technical features of the electronic device are the same as those disclosed in the method of the previous embodiment, which will not be repeated here.

[0031] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof, and specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples in the description of the above embodiments.

[0032] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

[0033] The present invention provides a computer-readable storage medium having computer-readable program instructions stored thereon, the computer-readable program instructions being used to execute the power system fault protection method based on multimodal data fusion in the above embodiments.

[0034] The computer-readable storage medium provided by the present invention stores computer-readable program instructions for executing the above-described power system fault protection method based on multimodal data fusion, which can solve the technical problem. The beneficial effects of the computer-readable storage medium provided by the present invention are the same as those of the power system fault protection method based on multimodal data fusion provided in the above embodiments, and will not be repeated here.

[0035] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, as well as combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0036] The above are only some embodiments of the present invention and do not limit the patent scope of the present invention. All equivalent structural transformations made under the technical concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A power system fault protection method based on multimodal data fusion, characterized in that, The specific steps include the following: S1. Collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and analyze abnormal electrical measurement conditions, abnormal equipment status conditions and abnormal environmental meteorological conditions of power equipment; S2. Construct a power equipment fault analysis model and collect historical power fault data to train the power equipment fault analysis model; S3. Input electrical measurement anomalies, equipment status anomalies, and environmental meteorological anomalies into the power equipment fault analysis model to identify the equipment that is malfunctioning and locate the fault point of the malfunctioning equipment. S4. Based on the electrical measurement data of the fault point of the abnormal equipment, analyze the abnormal status of the module corresponding to the fault point, and at the same time assess the impact of the equipment shutdown caused by the fault point. S5. Analyze the severity of equipment failure based on the abnormal status of the module corresponding to the fault point and the impact of equipment downtime caused by the fault point, and generate an equipment repair sequence based on the severity of the equipment failure.

2. The power system fault protection method based on multimodal data fusion according to claim 1, characterized in that, S1 includes the following specific steps: S11. Collect electrical measurement data during the operation of power equipment. The electrical measurement data includes current and voltage. Obtain abnormal current values ​​based on the ratio of current imbalance rate to safe current imbalance rate. The current imbalance rate is obtained by dividing the difference between the maximum phase current and the minimum phase current by the average phase current. Obtain abnormal voltage values ​​based on the ratio of voltage deviation to safe voltage deviation. The voltage deviation is obtained by dividing the absolute value of the difference between the actual voltage and the nominal voltage by the nominal voltage. Obtain abnormal electrical measurement values ​​by weighted summation of current abnormal values ​​and voltage abnormal values. S12. Collect equipment status monitoring data during the operation of power equipment. The equipment status monitoring data includes equipment temperature and vibration frequency. Obtain abnormal equipment temperature values ​​based on the ratio of equipment temperature to safe equipment temperature, obtain abnormal vibration values ​​based on the ratio of vibration frequency to safe vibration frequency, and obtain abnormal equipment status values ​​by weighted summation of abnormal equipment temperature values ​​and abnormal vibration values. S13. Collect environmental meteorological data during the operation of power equipment. The environmental meteorological data includes environmental temperature and environmental humidity. Obtain abnormal values ​​of environmental temperature based on the ratio of environmental temperature to safe environmental temperature, obtain abnormal values ​​of environmental humidity based on the ratio of environmental humidity to safe environmental humidity, and obtain abnormal values ​​of environmental meteorology based on the weighted sum of abnormal values ​​of environmental temperature and abnormal values ​​of environmental humidity.

3. The power system fault protection method based on multimodal data fusion according to claim 2, characterized in that, S2 includes the following specific steps: S21. Collect historical power failure data, which includes historical electrical measurement data, historical equipment status monitoring data, historical environmental meteorological data, and historical failure records. The historical failure records include the historical failure time point, historical failure equipment ID, and historical failure type. S22. Based on historical power fault data, obtain historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies respectively. Create a time-indexed label sequence. Obtain the fault equipment ID and fault time point corresponding to the historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies at each time point. In the label sequence, with the fault occurrence time point as the center, obtain the corresponding preset time window to mark the corresponding fault type. Mark the equipment as normal for the remaining time. Define the review window and sliding step size. Starting from the beginning of the label sequence, slide one time point backward each time. Extract the historical electrical measurement anomalies, historical equipment status anomalies, and historical environmental meteorological anomalies containing the review window duration as input features. Use the label corresponding to the end time of the review window as the output target. Repeat this process until all time points have been traversed. S23. Obtain the feature dataset and label dataset and divide them into 70% training set, 15% validation set and 15% test set. Define an input layer that matches the shape of the training dataset, use the Softmax activation function, and output the number of nodes and the probability of each fault category.

4. The power system fault protection method based on multimodal data fusion according to claim 3, characterized in that, S3 includes the following specific steps: S31. Input the abnormal values ​​of electrical measurements, abnormal values ​​of equipment status, and abnormal values ​​of environmental meteorology into the power equipment fault analysis model, output the number of nodes and the probability of each fault category, obtain the ID of abnormal equipment with a number of nodes greater than 1, and set the fault category corresponding to the maximum probability of the fault category as the fault category of the abnormal equipment. S32. Obtain the set of modules contained in the abnormal device, and obtain the failure probability of each module based on the failure probability calculation formula, wherein the failure probability calculation formula is: ,in, The influence weight of module mᵢ Let mᵢ be the prior fault probability of module mᵢ. The normalization constant represents the total probability of equipment failure. The module corresponding to the maximum failure probability is set as the failure point of the abnormal equipment.

5. The power system fault protection method based on multimodal data fusion according to claim 4, characterized in that, S4 includes the following specific steps: S41. Obtain electrical measurement data of the fault point of the abnormal equipment, including the current and voltage at the fault point; S42. Obtain the abnormal value of the fault point current based on the ratio of the fault point current to the module rated current, obtain the abnormal value of the fault point voltage based on the ratio of the fault point voltage to the module rated voltage, and obtain the abnormal evaluation value of the module corresponding to the fault point by weighted summation of the abnormal value of the fault point current and the abnormal value of the fault point voltage. S43. Obtain the continuous loss time caused by equipment downtime, and obtain the abnormal value of continuous loss time based on the ratio of continuous loss time to the maximum tolerable interruption time. S44. Obtain data on equipment affected by the power outage. This data includes the total number of affected equipment, the data generation rate of each affected equipment, and the distance between each affected equipment and the out-of-operation equipment. Obtain anomalies in the scale of the power outage based on the formula for calculating anomalies in the scale of the power outage. The formula for calculating anomalies in the scale of the power outage is as follows: ,in, The total number of devices affected by the power outage. The basic data impact of the nth power outage-affected device is obtained by multiplying the data generation rate of the affected device by the duration of the power outage. The distance influence weight between the nth power outage-affected device and the out-of-operation device is calculated using the following formula: Obtain This is the distance attenuation coefficient. The distance between the equipment affected by the nth power outage and the equipment that is out of service; S45. Obtain the assessment value of the impact of equipment shutdown caused by the failure point by weighted summation of outliers in the scale of power outage impact and outliers in the duration of loss.

6. The power system fault protection method based on multimodal data fusion according to claim 5, characterized in that, S5 includes the following specific steps: S51. Obtain the equipment failure severity assessment value by weighted summation of the abnormal assessment value of the module corresponding to the fault point and the impact assessment value of the equipment downtime caused by the fault point. S52. Sort the equipment failure severity assessment values ​​in descending order to generate an equipment repair sequence.

7. A power system fault protection system based on multimodal data fusion, used to implement the power system fault protection method based on multimodal data fusion as described in any one of claims 1-6, characterized in that, include: The equipment operation anomaly analysis module is used to collect electrical measurement data, equipment status monitoring data and environmental meteorological data during the operation of power equipment, and to analyze electrical measurement anomalies, equipment status anomalies and environmental meteorological anomalies of power equipment; The fault analysis model building module is used to build power equipment fault analysis models and collect historical power fault data to train the power equipment fault analysis models. The fault location module is used to input abnormal electrical measurement conditions, abnormal equipment status conditions, and abnormal environmental weather conditions into the power equipment fault analysis model to identify the abnormal equipment and locate the fault point of the abnormal equipment. The Fault Point Impact Analysis Module is used to analyze the abnormal status of the corresponding module based on the electrical measurement data of the fault point of the abnormal equipment, and at the same time assess the degree of impact of the equipment downtime caused by the fault point. The emergency repair sequence analysis module is used to analyze the severity of equipment failure based on the abnormal status of the module corresponding to the failure point and the impact of equipment downtime caused by the failure point, and to generate an emergency repair sequence based on the severity of the equipment failure.

8. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, the computer program being configured to implement the power system fault protection method based on multimodal data fusion as described in any one of claims 1-6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the power system fault protection method based on multimodal data fusion as described in any one of claims 1-6.

Images