Fault identification model construction method and device, computer device and storage medium

Abstract

The application relates to a fault identification model construction method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring historical event information, each piece of historical event information comprising at least two information subsegments; comparing each information subsegment of each piece of historical event information with an information category set to obtain a corresponding target information subsegment and a category identifier of the corresponding target information subsegment; arranging each target information subsegment in a preset order according to the category identifiers of the target information subsegments corresponding to each piece of historical event information to generate corresponding aggregated event information, wherein the preset order is determined according to the time characteristics, equipment characteristics and power grid fault types of historical events; generating sample data according to each piece of aggregated event information and inputting the sample data into an original fault identification model for training to obtain a target fault identification model. The method can effectively improve the accuracy of the fault identification model.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, computer device, and storage medium for constructing a fault identification model. Background Technology

[0002] With the development of computer technology, the use of neural network models to achieve autonomous identification of power grid alarm events and improve the intelligent level of power grid equipment operation monitoring is of great significance for improving the efficiency of daily monitoring and handling of accidents and anomalies.

[0003] In existing technologies, fault identification models are obtained by inputting historical alarm data into a deep learning model for training, but the accuracy of the generated fault identification model in identifying faults is low. Summary of the Invention

[0004] Therefore, it is necessary to provide a fault identification model method, apparatus, computer equipment, and storage medium to address the aforementioned technical problems, which can effectively improve the accuracy of the fault identification model.

[0005] Firstly, this application provides a method for constructing a fault identification model, including:

[0006] Retrieve historical event information, each of which includes at least two information sub-fields;

[0007] Each information segment of each historical event is compared with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers.

[0008] Based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate the corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics and power grid fault type of the historical event.

[0009] Sample data is generated based on the information from each aggregated event, and the sample data is input into the original fault identification model for training to obtain the target fault identification model, which is used to identify faults in power grid equipment.

[0010] In one embodiment, each information segment of each historical event information is compared with a set of information categories to obtain the corresponding target information segment and the category identifier of the corresponding target information segment, including:

[0011] Each information segment of each historical event is matched with each reference information segment in the information category set to obtain the matching result;

[0012] Based on the matching results, the target sub-segment information corresponding to each historical event is determined, and the corresponding category identifier is determined in the information category set based on the corresponding target sub-segment information.

[0013] In one embodiment, based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate corresponding aggregated event information, including:

[0014] Based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate a target information sequence corresponding to each historical event information.

[0015] The target information sequences corresponding to historical event information are classified and aggregated to obtain aggregated event information.

[0016] In one embodiment, before arranging the target information segments in a preset order according to the category identifier of each target information segment corresponding to each historical event information to generate the corresponding aggregated event information, the method further includes:

[0017] Semantic analysis is performed on historical event information to extract time feature information, equipment feature information, and power grid fault type information corresponding to the historical event information;

[0018] The preset sequence is generated sequentially according to the time characteristic information, equipment characteristic information, and power grid fault type information.

[0019] In one embodiment, sample data is generated based on information from each aggregated event, and the sample data is input into the original fault identification model for training to obtain the target fault identification model, including:

[0020] The information of each aggregated event is input into the original encoding model for training to obtain the target encoding model. The target encoding model is used to generate the corresponding target encoding sequence based on the target event information.

[0021] Input the information of each aggregated event into the target encoding model to generate target sample data;

[0022] The target sample data is input into the original fault identification model for training to obtain the target fault identification model.

[0023] Secondly, this application provides a fault identification method, including:

[0024] Get current event information;

[0025] The current event information is input into the target fault identification model to determine the fault type corresponding to the current event information. The target fault identification model is based on historical event information, each of which includes at least two information segments. Each information segment of each historical event information is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and input into the original fault identification model for training.

[0026] Thirdly, this application provides a fault identification model construction apparatus, comprising:

[0027] The acquisition module is used to acquire historical event information, and each historical event information includes at least two information sub-fields;

[0028] The matching module is used to compare each information segment of each historical event information with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers.

[0029] The aggregation module is used to arrange the target information segments in a preset order according to the category identifier of each target information segment corresponding to each historical event information, and generate the corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics and power grid fault type of the historical event.

[0030] The training module is used to generate sample data based on the information of each aggregated event, and input the sample data into the original fault identification model for training to obtain the target fault identification model. The target fault identification model is used to identify faults in power grid equipment.

[0031] Fourthly, this application provides a fault identification device, comprising:

[0032] The acquisition module is used to acquire current event information;

[0033] The identification module is used to input current event information into the target fault identification model to determine the fault type corresponding to the current event information. The target fault identification model is based on acquired historical event information, each of which includes at least two information segments. Each information segment of each historical event information is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and input into the original fault identification model for training.

[0034] Fifthly, this application also provides a computer device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0035] Retrieve historical event information, each of which includes at least two information sub-fields;

[0036] Each information segment of each historical event is compared with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers.

[0037] Based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate the corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics and power grid fault type of the historical event.

[0038] Sample data is generated based on the information from each aggregated event, and the sample data is input into the original fault identification model for training to obtain the target fault identification model, which is used to identify faults in power grid equipment.

[0039] Sixthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0040] Retrieve historical event information, each of which includes at least two information sub-fields;

[0041] Each information segment of each historical event is compared with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers.

[0042] Based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate the corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics and power grid fault type of the historical event.

[0043] Sample data is generated based on the information from each aggregated event, and the sample data is input into the original fault identification model for training to obtain the target fault identification model, which is used to identify faults in power grid equipment.

[0044] The aforementioned fault identification model construction method, apparatus, computer equipment, and storage medium acquire historical event information, each historical event information including at least two information segments; compare each information segment of each historical event information with an information category set to obtain the corresponding target information segment and its category identifier, the information category set containing a preset number of correspondences between reference information segments and reference category identifiers; based on the category identifiers of each target information segment corresponding to each historical event information, arrange the target information segments in a preset order to generate corresponding aggregated event information, the preset order being determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events; generate sample data based on each aggregated event information, and input the sample data into the original fault identification model for training to obtain the target fault identification model, which is used to identify faults in power grid equipment. In this way, by classifying and aggregating the features of the information segments of each historical event information, the constructed sample data can effectively mine the effective data features of each historical event, and finally, the fault identification model trained based on the aggregated event information effectively improves the model's accuracy. Attached Figure Description

[0045] Figure 1 This is an application environment diagram of the fault identification model construction method in one embodiment;

[0046] Figure 2 This is a flowchart illustrating a fault identification model construction method in one embodiment;

[0047] Figure 3 This is a flowchart illustrating a method for determining a category identifier in one embodiment;

[0048] Figure 4 This is a schematic diagram of the process for generating aggregated event information in one embodiment;

[0049] Figure 5 This is a flowchart illustrating the process of generating a preset order in one embodiment;

[0050] Figure 6 This is a flowchart illustrating the process of obtaining the target fault identification model in one embodiment;

[0051] Figure 7 This is a flowchart illustrating a fault identification method in one embodiment;

[0052] Figure 8 This is a structural block diagram of a fault identification model construction device in one embodiment;

[0053] Figure 9 This is a structural block diagram of a fault identification device in one embodiment;

[0054] Figure 10 This is a schematic diagram of a pre-defined standardized model in one embodiment;

[0055] Figure 11 This is a schematic diagram of the structure of the Skip-gram model neural network in one embodiment;

[0056] Figure 12 This is a schematic diagram of the structure of an LSTM Cell in one embodiment;

[0057] Figure 13 This is a schematic diagram of a feature sequence generation method in one embodiment;

[0058] Figure 14 This is a schematic diagram of the fault sequence identification process in one embodiment;

[0059] Figure 15 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0061] The fault identification model construction method provided in this application can be applied to, for example... Figure 1 The application environment shown. For example... Figure 1As shown, the application environment includes terminal 102. Computer device 102 acquires historical event information; compares each information segment of each historical event information with a set of information categories to obtain the corresponding target information segment and its category identifier; based on the category identifiers of each target information segment corresponding to each historical event information, arranges the target information segments in a preset order to generate corresponding aggregated event information; generates sample data based on each aggregated event information, and inputs the sample data into the original fault identification model for training to obtain the target fault identification model. The computer device can be, but is not limited to, various personal computers, servers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc.

[0062] In one embodiment, such as Figure 2 As shown, a method for constructing a fault identification model is provided, which can be applied to... Figure 1 Taking computer device 102 as an example, the following steps are included:

[0063] Step S202: Obtain historical event information, where each historical event information includes at least two information sub-fields.

[0064] Among them, the information sub-field is a fragment of historical event information that carries specific information, such as a time adverbial fragment used to identify time attributes, a fragment used to identify the specific equipment model that malfunctioned, or a fragment used to identify the type of malfunction or the location of the malfunction.

[0065] Specifically, the computer equipment obtains full or partial historical event information of the target power grid from the database of the target server.

[0066] Step S204: Compare each information segment of each historical event information with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment.

[0067] The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. The reference category identifier is used to characterize the segment category of the corresponding reference information segment. For example, when the reference category identifier is a time category identifier, it indicates that all reference information segments corresponding to the time category identifier are time information segments, used to characterize the time information of the corresponding event.

[0068] Specifically, the computer device compares each information segment of each historical event information with each reference information segment in the information category set in turn. When a match is successful, the reference information segment that the current information segment matches successfully is taken as the target information segment corresponding to the current information segment, and the category identifier corresponding to the target information segment is taken as the target category identifier.

[0069] Step S206: Based on the category identifier of each target information segment corresponding to each historical event information, arrange each target information segment in a preset order to generate the corresponding aggregated event information.

[0070] The preset sequence is determined based on the time characteristics of historical events, equipment characteristics, and power grid fault types.

[0071] Specifically, the computer equipment sorts each target information segment according to the target information segments and their category identifiers determined in the aforementioned steps. The sorting order is determined based on the time characteristics, equipment characteristics, and power grid fault types of historical events. Technicians can flexibly sort the time characteristics, equipment characteristics, and power grid fault types as needed. For example, the sorting order can be time-equipment-fault type or equipment-fault type-time. Then, based on the new arrangement of each historical event information, the historical event information is aggregated and analyzed. Historical event information with the same semantic arrangement is aggregated into the same category of events, thereby completing the classification and aggregation of all historical event information.

[0072] Step S208: Generate sample data based on the information of each aggregated event, and input the sample data into the original fault identification model for training to obtain the target fault identification model.

[0073] Among them, the target fault identification model is used to identify faults in power grid equipment.

[0074] Specifically, the computer device obtains information on each aggregated event according to the aforementioned steps, generates corresponding event feature vectors based on the aggregated event information, uses the event feature vectors as training sample data, inputs them into the original fault identification model, trains the original fault identification model until the training stopping condition is met, and obtains the target fault identification model.

[0075] The aforementioned fault identification model construction method acquires historical event information, each of which includes at least two information segments. Each information segment of each historical event is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event, the target information segments are arranged in a preset order to generate corresponding aggregated event information. This preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and input into the original fault identification model for training, resulting in a target fault identification model used to identify faults in power grid equipment. In this way, by classifying and aggregating the features of the information segments of each historical event, the constructed sample data can effectively mine the effective data features of each historical event. Finally, the fault identification model trained based on the aggregated event information effectively improves the model's accuracy.

[0076] In one embodiment, such as Figure 3 As shown, each information segment of each historical event is compared with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment, including:

[0077] Step S302: Match each information segment of each historical event information with each reference information segment in the information category set to obtain the matching result.

[0078] Specifically, the computer device performs correlation calculations on the feature vector corresponding to each information segment of each historical event information and the feature vector corresponding to each reference information segment to obtain the correlation degree. Then, by judging whether the correlation degree is within the preset correlation range, the matching result is obtained.

[0079] Step S304: Determine the target sub-segment information corresponding to each historical event information based on the matching results, and determine the corresponding category identifier in the information category set based on the corresponding target sub-segment information.

[0080] Specifically, the computer device determines the target sub-segment information corresponding to each information sub-segment of each historical event information according to the aforementioned steps, and then uses the reference category identifier corresponding to the target sub-segment information in the information category set as the category identifier of the corresponding sub-segment information of the corresponding event information.

[0081] In this embodiment, each information segment of each historical event information is matched with each reference information segment in the information category set to obtain the matching result. Based on the matching result, the target segment information corresponding to each historical event information is determined, and the corresponding category identifier is determined in the information category set based on the corresponding target segment information. This enables the target segment information and the corresponding category identifier of each historical event information to be determined quickly and effectively, thereby improving the accuracy of determining the attributes of historical event information.

[0082] In one embodiment, such as Figure 4 As shown, based on the category identifiers of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate corresponding aggregated event information, including:

[0083] Step S402: Based on the category identifier of each target information segment corresponding to each historical event information, arrange each target information segment in a preset order to generate a target information sequence corresponding to each historical event information.

[0084] Specifically, the computer device rearranges the target segments corresponding to each historical event information determined in the aforementioned steps according to the category identifier and preset order of each target information segment, generates a new semantic information expression, and then generates a target information sequence corresponding to each historical event information based on the semantic information expression.

[0085] Step S402: Classify and aggregate the target information sequences corresponding to the historical event information to obtain aggregated event information.

[0086] Specifically, after the computer device obtains the target information sequence corresponding to each historical event information in the aforementioned steps, it classifies the target information sequence corresponding to each historical event information into a category, divides the target information sequences of the same category into aggregated event information of the same category, and thus realizes the classification and aggregation of each historical event information.

[0087] In this embodiment, based on the category identifier of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate a target information sequence corresponding to each historical event information. The target information sequences corresponding to the historical event information are then classified and aggregated to obtain aggregated event information. This achieves classification and aggregation based on the effective key information segments in each historical event information, which can effectively improve the accuracy of classification and aggregation.

[0088] In one embodiment, such as Figure 5As shown, before generating the corresponding aggregated event information by arranging the target information segments in a preset order according to the category identifiers of each target information segment corresponding to each historical event information, the process also includes:

[0089] Step S502: Perform semantic analysis on the historical event information to extract the time feature information, equipment feature information, and power grid fault type information corresponding to the historical event information.

[0090] Specifically, the computer device acquires historical expression statements, which contain multiple types of time feature segments, equipment feature segments, and power grid fault type segments. These historical expression statements are then used as training sample data and input into the original semantic recognition model for training until the training stop condition is met. Training is then stopped, and the target semantic recognition model is obtained. The initial semantic recognition model can be a neural network model, a deep learning model, etc. Historical event information is then input into the target semantic recognition model, which outputs the time feature segments, equipment feature segments, and power grid fault type information corresponding to each historical event.

[0091] Step S504: Generate a preset sequence in the order of time feature information, equipment feature information and power grid fault type information.

[0092] Specifically, the computer equipment determines the time characteristics, equipment characteristics, and power grid fault types of each historical event information based on the aforementioned steps. Technicians can then sort the time characteristics, equipment characteristics, and power grid fault type information according to the specific application scenario to obtain the expression sentences corresponding to the preset order.

[0093] In this embodiment, semantic analysis is performed on historical event information to extract time feature information, equipment feature information, and power grid fault type information corresponding to the historical event information. Then, a preset order is generated according to the order of time feature information, equipment feature information, and power grid fault type information to improve the reliability of the expression sentences corresponding to the generated preset order.

[0094] In one embodiment, such as Figure 6 As shown, sample data is generated based on the information from each aggregated event, and the sample data is input into the original fault identification model for training to obtain the target fault identification model, including:

[0095] Step 602: Input the information of each aggregated event into the original encoding model for training to obtain the target encoding model.

[0096] The target encoding model is used to generate the corresponding target encoding sequence based on the target event information. The original encoding model can be a Word2vec model, an LSTM model (such as...). Figure 12As shown), the Word2vec model includes the CBOW model (continuous word band model) and the Skip-gram model (skip-word model, such as...). Figure 11 As shown, the CBOW model is used to predict the generation probability of the currently studied word based on the context of the target text, while the Skip-gram model is used to predict the generation probability of words in the context of the target text based on the current word. Both the CBOW model and the Skip-gram model can be trained using the softmax activation function. In addition, the target encoding sequence can be one-hot encoded.

[0097] Specifically, the computer device uses the aggregated event information determined in the aforementioned steps as training sample data and inputs it into the original encoding model for training until the training stopping condition is met, thereby obtaining the target encoding model.

[0098] Step 604: Input the information of each aggregated event into the target encoding model to generate target sample data.

[0099] Step 606: Input the target sample data into the original fault identification model for training to obtain the target fault identification model.

[0100] In this embodiment, target sample data is generated by inputting the information of each aggregated event into the target encoding model. The target sample data is then input into the original fault identification model for training to obtain the target fault identification model. The embedded encoding vector generated by the target encoding model fully preserves the relative feature relationship between the information of each aggregated event, which facilitates the subsequent classification by the classifier and risk assessment calculation, and effectively improves the accuracy of the sample data.

[0101] In one embodiment, such as Figure 7 As shown, a fault identification method is provided, which can be applied to... Figure 1 Taking computer device 102 as an example, the following steps are included:

[0102] Step S702: Obtain current event information.

[0103] Among them, the current event information refers to alarm events received from the power grid at the current moment. The alarm information includes information such as the substation address, voltage, bay equipment, and signal behavior.

[0104] Step S704: Input the current event information into the target fault identification model to determine the fault type corresponding to the current event information.

[0105] The target fault identification model is developed by acquiring historical event information, each of which includes at least two information segments. Each information segment of each historical event is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event, the target information segments are arranged in a preset order to generate corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and then input into the original fault identification model for training.

[0106] In this embodiment, current event information from the power grid is input into the target fault identification model to identify the fault type. The target fault identification model is constructed by classifying and aggregating the features of information segments of various historical event information. The constructed sample data can effectively mine the effective data features of each historical event. Finally, the fault identification model trained based on the aggregated event information effectively improves the accuracy of fault identification.

[0107] This application also provides an application scenario in which the above-mentioned fault identification model construction method is applied to a scenario of fault identification in the power grid. Specifically, the application of the fault identification model construction method in this application scenario is as follows:

[0108] Power grid alarm information event sequences reflect changes in the operating status of power grid equipment. Besides normal equipment status changes, they also include information on the state evolution when equipment malfunctions. With the expansion of power grid equipment scale and the improvement of intelligent monitoring levels, the amount of alarm information logs accumulated in the power grid is enormous, and the types of power grid alarm information are diverse. Moreover, due to differences in collection standards and other issues, the actual received alarm information often exhibits multiple synonymous terms, which brings difficulties to the processing and analysis of power grid alarm information event sequences. This is especially evident during sudden faults, when a large number of alarm messages emerge simultaneously, and signals unrelated to the fault quickly overwhelm valuable signals. If relying solely on dispatch personnel to distinguish and identify signals, event analysis becomes extremely difficult and prone to missing important alarm information, posing a significant risk to the safe operation of the power grid.

[0109] To address the problems of existing technologies, the power grid fault alarm information event sequence identification method provided by this invention first aggregates collected historical alarm information events to obtain aggregated power grid alarm information events, thereby reducing the number of event types that need to be analyzed. To facilitate feature analysis of the event sequences using machine learning or deep learning methods, it is first necessary to obtain numerical representations of the events. Based on an embedded representation method model in natural language processing, the aggregated power grid alarm information events are subjected to unsupervised training to generate a high-dimensional word vector sequence containing information features. The high-dimensional word vector sequence is then used to extract alarm information event sample sequences according to a sliding time window, achieving fault event sequence classification.

[0110] Considering the diverse types of power grid alarm information, and the fact that actual received alarm information often contains multiple synonyms due to differences in collection standards, this presents difficulties for processing and analyzing power grid alarm information event sequences. In this embodiment, power grid alarm information events can be parsed according to a preset standardized model to obtain a structured event representation, such as... Figure 10 As shown in Table 1, after processing all event information in the substation's historical logs, a structured representation of all types of event information is obtained. For a 110kV substation, approximately 130,000 types of event information were obtained from five years of alarm information log statistics. Therefore, it is necessary to classify and aggregate the historical event information, as shown in Table 1 below:

[0111] Table 1. Examples of partial alarm information event aggregation effects

[0112]

[0113] Subsequently, a deep learning model was built. The sequence of alarm information event samples within the sliding time window was used to train the deep learning model, such as... Figure 13 , Figure 14 As shown in the figure. Finally, the trained deep learning model is used to identify the event sequence of power grid fault alarm information, which can effectively solve the problems of difficulty in analyzing events, high monitoring pressure, and easy omission of monitoring, and improve monitoring efficiency, the ability to identify important events in alarm information, and the ability to control the power grid status in real time.

[0114] In this embodiment, the accurate identification of power grid fault alarm information event sequences is based on the extraction of semantic features from historical power grid alarm information events. Then, the extracted semantic features are used to classify and aggregate each historical power grid alarm information event. Finally, a deep learning model is trained based on the aggregated sample data. The resulting fault identification model can identify power grid faults in a timely, accurate, and automatic manner from massive alarm information, which helps to realize intelligent monitoring of power grid status, assists dispatchers in handling faults, and improves the accuracy of fault identification.

[0115] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0116] In one embodiment, such as Figure 8 As shown, a fault identification model construction device is provided. This device can be a software module, a hardware module, or a combination of both integrated into a computer device. Specifically, the device includes: an acquisition module 802, a matching module 804, an aggregation module 806, and a training module 808, wherein:

[0117] The acquisition module 802 is used to acquire historical event information, and each historical event information includes at least two information sub-fields;

[0118] The matching module 804 is used to compare each information segment of each historical event information with the information category set to obtain the corresponding target information segment and the category identifier of the corresponding target information segment. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers.

[0119] The aggregation module 806 is used to arrange the target information segments in a preset order according to the category identifier of each target information segment corresponding to each historical event information to generate the corresponding aggregated event information. The preset order is determined according to the time characteristics, equipment characteristics and power grid fault type of the historical event.

[0120] The training module 808 is used to generate sample data based on the information of each aggregated event, and input the sample data into the original fault identification model for training to obtain the target fault identification model. The target fault identification model is used to identify faults in power grid equipment.

[0121] In one embodiment, the matching module 804 is further configured to match each information segment of each historical event information with each reference information segment in the information category set to obtain a matching result; determine the target segment information corresponding to each historical event information based on the matching result; and determine the corresponding category identifier in the information category set based on the corresponding target segment information.

[0122] In one embodiment, the aggregation module 806 is further configured to arrange each target information segment in a preset order according to the category identifier of each target information segment corresponding to each historical event information, thereby generating a target information sequence corresponding to each historical event information; and to classify and aggregate each target information sequence corresponding to the historical event information to obtain each aggregated event information.

[0123] In one embodiment, the aggregation module 806 is further configured to perform semantic analysis on historical event information, extract time feature information, equipment feature information, and power grid fault type information corresponding to the historical event information, and generate a preset order in sequence according to the order of time feature information, equipment feature information, and power grid fault type information.

[0124] In one embodiment, the training module 808 is further configured to input the information of each aggregated event into the original encoding model for training to obtain the target encoding model, the target encoding model being used to generate the corresponding target encoding sequence based on the target event information; input the information of each aggregated event into the target encoding model to generate target sample data; and input the target sample data into the original fault identification model for training to obtain the target fault identification model.

[0125] The aforementioned fault identification model construction device acquires historical event information, each of which includes at least two information segments. Each information segment of each historical event is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event, the target information segments are arranged in a preset order to generate corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and input into the original fault identification model for training to obtain a target fault identification model. This target fault identification model is used to identify faults in power grid equipment. In this way, by classifying and aggregating the features of the information segments of each historical event, the constructed sample data can effectively mine the effective data features of each historical event. Finally, the fault identification model trained based on the aggregated event information effectively improves the model's accuracy.

[0126] Specific limitations regarding the fault identification model construction device can be found in the limitations of the fault identification model construction method described above, and will not be repeated here. Each module in the aforementioned fault identification model construction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0127] In one embodiment, such as Figure 9 As shown, a fault identification device is provided. This device can be a software module, a hardware module, or a combination of both as part of a computer device. Specifically, the device includes: an acquisition module 902 and an identification module 904, wherein:

[0128] Module 902 is used to obtain current event information;

[0129] The identification module 904 is used to input current event information into the target fault identification model to determine the fault type corresponding to the current event information. The target fault identification model is based on acquired historical event information, each of which includes at least two information segments. Each information segment of each historical event information is compared with an information category set to obtain the corresponding target information segment and its category identifier. The information category set contains a preset number of correspondences between reference information segments and reference category identifiers. Based on the category identifiers of each target information segment corresponding to each historical event information, the target information segments are arranged in a preset order to generate corresponding aggregated event information. The preset order is determined based on the time characteristics, equipment characteristics, and power grid fault type of the historical events. Sample data is generated based on each aggregated event information and input into the original fault identification model for training.

[0130] The aforementioned fault identification device identifies fault types by inputting current event information from the power grid into a target fault identification model. The target fault identification model is constructed by classifying and aggregating the features of information segments of various historical event information. The constructed sample data can effectively mine the effective data features of each historical event. Finally, the fault identification model trained based on the aggregated event information effectively improves the accuracy of fault identification.

[0131] For specific limitations regarding the fault identification device, please refer to the limitations of the fault identification method above, which will not be repeated here. Each module in the aforementioned fault identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0132] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 15 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a fault identification model construction method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0133] Those skilled in the art will understand that Figure 15 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0134] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0135] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0136] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0138] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for constructing a failure identification model, characterized in that, The method comprises: acquiring historical event information, each historical event information comprising at least two information subsegments; comparing each information subsegment of each historical event information with an information category set to obtain a corresponding target information subsegment and a category identifier of the corresponding target information subsegment, the information category set comprising a corresponding relationship between a preset number of reference information subsegments and reference category identifiers; arranging the target information subsegments in a preset order according to the category identifiers of the target information subsegments corresponding to each historical event information to generate a target information sequence corresponding to each historical event information, the preset order being determined according to time characteristics, device characteristics and power grid fault types of historical events; classifying and aggregating the target information sequences based on the device characteristics and the power grid fault types to obtain aggregated event information; generating sample data according to the aggregated event information and inputting the sample data into an original fault recognition model for training to obtain a target fault recognition model, the target fault recognition model being used for identifying faults of power grid devices.

2. The method of claim 1, wherein, The comparison of each information subsegment of each historical event information with an information category set to obtain a corresponding target information subsegment and a category identifier of the corresponding target information subsegment comprises: matching each information subsegment of each historical event information with each reference information subsegment in the information category set to obtain a matching result; determining target subsegment information corresponding to each historical event information according to the matching result and determining a corresponding category identifier in the information category set according to the corresponding target subsegment information.

3. The method of claim 1, wherein, Before the arrangement of the target information subsegments in a preset order according to the category identifiers of the target information subsegments corresponding to each historical event information to generate corresponding aggregated event information, the method further comprises: performing semantic analysis on the historical event information to extract time characteristic information, device characteristic information and power grid fault type information corresponding to the historical event information; generating a preset order in the order corresponding to the time characteristic information, the device characteristic information and the power grid fault type information.

4. The method of claim 1, wherein, The generation of sample data according to the aggregated event information and the input of the sample data into an original fault recognition model for training to obtain a target fault recognition model comprises: inputting the aggregated event information into an original encoding model for training to obtain a target encoding model, the target encoding model being used for generating a corresponding target encoding sequence according to target event information; inputting the aggregated event information into the target encoding model to generate target sample data; inputting the target sample data into an original fault recognition model for training to obtain a target fault recognition model.

5. A failure identification method characterized by, The method comprises: acquiring current event information; The current event information is input into a target fault identification model to determine a fault type corresponding to the current event information; the target fault identification model is obtained according to historical event information, each piece of historical event information including at least two information segments; each information segment of each piece of historical event information is compared with an information category set to obtain a corresponding target information segment and a category identifier of the corresponding target information segment, the information category set including a corresponding relationship between a preset number of reference information segments and reference category identifiers; each target information segment is arranged in a preset order according to the category identifiers of the target information segments corresponding to each piece of historical event information to generate a target information sequence corresponding to each piece of historical event information, the preset order being determined according to time characteristics of historical events, device characteristics and power grid fault types; the target information sequences are classified and aggregated based on the device characteristics and the power grid fault types to obtain each aggregated event information; sample data is generated according to each aggregated event information, and the sample data is input into an original fault identification model for training to obtain the target fault identification model.

6. A failure identification model construction apparatus characterized by comprising: The device comprises: an acquisition module configured to acquire historical event information, each piece of historical event information including at least two information segments; a matching module configured to compare each information segment of each piece of historical event information with an information category set to obtain a corresponding target information segment and a category identifier of the corresponding target information segment, the information category set including a corresponding relationship between a preset number of reference information segments and reference category identifiers; an aggregation module configured to arrange each target information segment in a preset order according to the category identifiers of the target information segments corresponding to each piece of historical event information to generate a target information sequence corresponding to each piece of historical event information, the preset order being determined according to time characteristics of historical events, device characteristics and power grid fault types, and the target information sequences being classified and aggregated based on the device characteristics and the power grid fault types to obtain each aggregated event information; a training module configured to generate sample data according to each aggregated event information, and input the sample data into an original fault identification model for training to obtain a target fault identification model, the target fault identification model being used to identify faults of power grid devices.

7. A failure identification device characterized by comprising: The device comprises: an acquisition module configured to acquire current event information; The identification module is configured to input current event information into a target fault identification model to determine a fault type corresponding to the current event information; the target fault identification model is generated according to historical event information, each piece of historical event information including at least two information segments; each information segment of each piece of historical event information is compared with an information category set to obtain a corresponding target information segment and a category identifier of the corresponding target information segment, the information category set including a corresponding relationship between a preset number of reference information segments and reference category identifiers; each target information segment is arranged in a preset order according to the category identifiers of the target information segments corresponding to each piece of historical event information to generate a target information sequence corresponding to each piece of historical event information, the preset order being determined according to time characteristics, device characteristics and power grid fault types of historical events; the target information sequences are classified and aggregated based on the device characteristics and the power grid fault types to obtain each aggregated event information; sample data is generated according to each aggregated event information, and the sample data is input into an original fault identification model for training.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 5.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 5.

10. A computer program product comprising a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 5. The computer program, when executed by the processor, implements the steps of the method of any one of claims 1 to 5.

Images