Power information communication method and system based on abnormal state recognition

Abstract

The application provides a power information communication method and system based on abnormal state identification, relates to the field of power data processing, and comprises the following steps: obtaining control parameter state information and environment index state information; performing fitness analysis on the control parameter state information and the environment index state information to obtain a first abnormal coefficient; when the first abnormal coefficient is less than a first abnormal threshold, performing safety evaluation on a first power equipment based on the environment index state information to obtain a second abnormal coefficient; when the second abnormal coefficient is less than a second abnormal threshold, performing long-time fault analysis based on the control parameter state information to obtain a third abnormal coefficient; and when the third abnormal coefficient is less than a third abnormal threshold, performing health identification on state monitoring data at a first time to generate first identification information and sending the first identification information to a remote client for storage. The technical problem that the abnormal identification in the communication process has poor referenceability due to the fact that the types of power field data are redundant in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of power data processing technology, and specifically to a power information communication method and system based on abnormal state identification. Background Technology

[0002] Power communication primarily involves the real-time collection and remote transmission of various status information from power sites using sensing technology, providing data support for power site management. Traditional power communication methods for identifying abnormal data mainly involve setting thresholds for various status data and then judging abnormal patterns based on these thresholds. However, power site monitoring data has numerous attributes and limited visualization, requiring experts to set abnormal thresholds for different attribute status data, which makes it difficult to guarantee accuracy and objectivity. Summary of the Invention

[0003] This application provides a power information communication method and system based on abnormal state identification, which is used to address the technical problem in the prior art that the abnormal identification of the communication process is of poor reference value due to the complexity of power field data.

[0004] In view of the above problems, this application provides a power information communication method and system based on abnormal state identification.

[0005] The first aspect of this application provides a power information communication method based on abnormal state identification, applied to a power information communication system based on abnormal state identification, comprising: acquiring first-moment state monitoring data of a first power device, wherein the first-moment state monitoring data includes control parameter state information and environmental indicator state information; performing fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation degree of the control parameter state information and the environmental indicator state information; when the first anomaly coefficient is less than a first anomaly threshold, performing a safety assessment on the first power device based on the environmental indicator state information to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the danger level of the first power device; when the second anomaly coefficient is less than a second anomaly threshold, performing long-term fault analysis on the control parameter state information to obtain a third anomaly coefficient, wherein the third anomaly coefficient characterizes the failure probability of the first power device during long-term operation of the control parameter state information; when the third anomaly coefficient is less than a third anomaly threshold, performing health identification on the first-moment state monitoring data, generating first identification information, and sending it to a remote client for storage.

[0006] A second aspect of this application provides a power information communication system based on abnormal state identification, comprising: a monitoring state data acquisition module for acquiring first-moment state monitoring data of a first power device, wherein the first-moment state monitoring data includes control parameter state information and environmental indicator state information; a first anomaly identification module for performing fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation between the control parameter state information and the environmental indicator state information; a second anomaly identification module for performing a safety assessment of the first power device based on the environmental indicator state information when the first anomaly coefficient is less than a first anomaly threshold to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the danger level of the first power device; a third anomaly identification module for performing long-term fault analysis based on the control parameter state information when the second anomaly coefficient is less than a second anomaly threshold to obtain a third anomaly coefficient, wherein the third anomaly coefficient characterizes the failure probability of the first power device during long-term operation of the control parameter state information; and an information communication module for performing health identification on the first-moment state monitoring data when the third anomaly coefficient is less than a third anomaly threshold, generating first identification information, and sending it to a remote client for storage.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] This application evaluates the fitness of control parameter status information and environmental indicator status information from the real-time monitoring data at the first moment. After the fitness evaluation is qualified, a safety evaluation is performed on the first power equipment based on the environmental indicator status information. After the safety evaluation is qualified, long-term fault analysis is performed based on the control parameter status information. The fitness evaluation can verify the possibility of deviation of the uploaded monitoring data, the safety evaluation can analyze the impact of the current environmental state on the equipment, and the long-term fault analysis can determine the stability of the equipment during long-term operation. Then, the data is labeled based on the analysis results of the three dimensions before transmission, which improves the referenceability and visualization of the transmitted data. Attached Figure Description

[0009] Figure 1 A schematic diagram of a power information communication method based on abnormal state identification provided in this application;

[0010] Figure 2 A schematic diagram illustrating the process of obtaining second identification information in a power information communication method based on abnormal state identification provided in this application;

[0011] Figure 3A flowchart illustrating the process of obtaining the first abnormality coefficient in a power information communication method based on abnormal state identification provided in this application;

[0012] Figure 4 This application provides a schematic diagram of the structure of a power information communication system based on abnormal state identification.

[0013] Explanation of reference numerals in the attached figures: 11 monitoring status data acquisition module, 12 first anomaly identification module, 13 second anomaly identification module, 14 third anomaly identification module, and 15 information communication module. Detailed Implementation

[0014] This application provides a power information communication method and system based on abnormal state identification, which addresses the technical problem in the prior art where the complexity of power field data leads to poor referenceability in abnormal identification during the communication process.

[0015] Example 1: As Figure 1 As shown, this application provides a power information communication method based on abnormal state identification, applied to a power information communication system based on abnormal state identification, including the following steps:

[0016] S100: Obtain the first moment status monitoring data of the first power equipment, wherein the first moment status monitoring data includes control parameter status information and environmental indicator status information;

[0017] In a preferred embodiment, the power information communication system based on anomaly state identification is an apparatus for implementing any step of a power information communication method based on anomaly state identification. Optionally, it includes a processor and a memory. The memory is used to store intermediate processing data and flow data of any step of the power information communication method based on anomaly state identification, as well as the processing method. The processor is used to schedule the processing method in the memory to process the intermediate processing data and flow data of any step of the power information communication method based on anomaly state identification.

[0018] The first power equipment refers to any equipment that the power system needs to monitor, and the first moment status monitoring data refers to the monitoring data that corresponds one-to-one with the first power equipment.

[0019] This includes real-time control parameters for equipment operation. Real-time operation control parameters that can be directly uploaded by the equipment are uploaded by the equipment itself, such as specific operation control parameters: output voltage, current, and other information. However, parameters that cannot be directly uploaded, such as the display data of various instruments and display devices, can be acquired by image acquisition devices deployed at the power site to collect image information from the display screen. Then, the monitoring data can be extracted through OCR text recognition. Since OCR text recognition is a relatively mature technology, it will not be elaborated on here.

[0020] It also includes internal environmental information during the operation of the first power equipment, such as environmental state quantities such as temperature, vibration amplitude, and humidity. These environmental state quantities are determined in real time by sensors deployed on the first power equipment.

[0021] Specifically, the real-time control parameters at the first moment are stored as control parameter status information, and the internal environment information at the first moment is stored as environmental indicator status information. These are combined and stored as the status monitoring data at the first moment, set to a pending response state, and await subsequent calls.

[0022] S200: Perform fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation between the control parameter state information and the environmental indicator state information;

[0023] In a preferred embodiment, the first anomaly coefficient characterizes the degree of mismatch between the control parameter status information and the environmental indicator status information. Since the internal environmental state of the equipment, such as vibration data, temperature data, and humidity data, is usually highly correlated with the operating control parameters of the equipment, by evaluating the degree of mismatch between the control parameter status information and the environmental indicator status information, it is possible to analyze whether the uploaded control parameter status information is compatible with the environmental indicator status. The lower the degree of compatibility between the control parameter status information and the environmental indicator status information, the larger the first anomaly coefficient. This verifies the data transmission loss between the control parameter status information and the environmental indicator status, ensuring that the monitoring data for subsequent identification of abnormal states of power equipment has a small transmission loss error, thereby improving the accuracy of identifying abnormal states of power equipment.

[0024] S300: When the first anomaly coefficient is less than the first anomaly threshold, a safety assessment is performed on the first power equipment based on the environmental indicator status information to obtain a second anomaly coefficient, wherein the second anomaly coefficient represents the danger level of the first power equipment;

[0025] In a preferred embodiment, the first anomaly threshold represents the minimum allowable deviation between the control parameter status information and the environmental indicator status information, which is predefined by the user. When the first anomaly coefficient is less than the first anomaly threshold, it indicates that the deviation between the control parameter status information and the environmental indicator status information is low, and therefore the reliability is high, allowing for subsequent processes.

[0026] Furthermore, since the operational safety of the equipment is related to the internal environmental conditions of the equipment, such as operating temperature, operating humidity, and vibration status, a second anomaly coefficient representing the danger level of the first power equipment can be determined based on the environmental indicator status information. The larger the second anomaly coefficient, the lower the operational safety of the equipment. The operational safety assessment result of the power equipment is obtained, which has higher interpretability than the underlying monitoring data and has strong visualization characteristics when transmitted to the remote client.

[0027] S400: When the second abnormal coefficient is less than the second abnormal threshold, based on the long-term fault analysis of the control parameter status information, a third abnormal coefficient is obtained, wherein the third abnormal coefficient represents the failure probability of the first power equipment when the control parameter status information is running for a long time.

[0028] In a preferred embodiment, the second anomaly threshold represents the minimum value of the second anomaly coefficient for the safe operation of the equipment. When the second anomaly coefficient is less than the second anomaly threshold, the first power equipment is considered to be operating relatively safely. Furthermore, while the first power equipment is in a safe state under the current control parameter status information, its safety under long-term operation cannot be determined. Therefore, based on long-term fault analysis of the control parameter status information, a third anomaly coefficient representing the probability of failure of the first power equipment under the long-term operation of the control parameter status information is obtained. This allows the determination of the failure probability of the first power equipment under different durations when the control parameter status information remains unchanged. Through long-term fault analysis, short-term monitoring data anomaly identification is transformed into long-term fault identification, improving the richness of the monitoring data and providing a data foundation for the pre-analysis of anomaly identification.

[0029] S500: When the third abnormality coefficient is less than the third abnormality threshold, the status monitoring data at the first moment is marked with a health label, a first label information is generated, and it is sent to the remote client for storage.

[0030] In a preferred embodiment, the third anomaly threshold represents the fault probability threshold of the anomaly identifier. When the third anomaly coefficient is less than the third anomaly threshold, it is considered that the fault probability is low, and the long-term operating state of the equipment is considered safe. This indicates that the power equipment is in a healthy state. Therefore, the status monitoring data at the first moment is marked with a health identifier, a first identifier is generated, and sent to a remote client for storage.

[0031] By verifying the reliability of monitoring data, assessing its safety, and analyzing long-term faults, highly visual identification data was identified, improving the reference value and accuracy of power information communication.

[0032] Furthermore, such as Figure 2 As shown, it also includes step S600, which further includes the following steps:

[0033] S610: When the first abnormal coefficient is greater than or equal to the first abnormal threshold, a first attribute abnormality identifier is generated; or

[0034] S620: When the second abnormal coefficient is greater than or equal to the second abnormal threshold, a second attribute abnormality identifier is generated; or

[0035] S630: When the third abnormality coefficient is greater than or equal to the third abnormality threshold, a third attribute abnormality identifier is generated;

[0036] S640: Based on the first attribute anomaly identifier, the second attribute anomaly identifier, or the third attribute anomaly identifier, anomaly identifier is generated on the status monitoring data at the first moment, and second identifier information is sent to the remote client for management.

[0037] In a preferred embodiment, if the first anomaly coefficient is greater than or equal to the first anomaly threshold, it indicates that the transmission process of control parameter status information or environmental indicator status information with a high probability has suffered significant loss. Therefore, it is not suitable to perform subsequent abnormal identification of the equipment status. Instead, the anomaly of transmission loss is directly identified in the status monitoring data at the first moment, which is recorded as the first attribute anomaly identifier and sent to the remote client for communication, so as to facilitate the remote client's abnormal analysis of transmission loss.

[0038] When the second anomaly coefficient is greater than or equal to the second anomaly threshold, the first power equipment is considered to be operating in a relatively dangerous manner. In this case, the first moment status monitoring data needs to be marked as dangerous operation of the equipment, recorded as the second attribute anomaly mark, and transmitted to the remote user client so that the remote client can promptly identify and check the first power equipment to eliminate dangerous operation.

[0039] When the third anomaly coefficient is greater than or equal to the third anomaly threshold, it is considered that the fault probability is low and the long-term operation state of the equipment is dangerous. Then, based on the control parameter status information, duration and fault probability that the fault probability is higher than the third anomaly threshold, it is used as the long-term fault analysis identification data of the equipment and recorded as the third attribute anomaly identifier. Based on the first attribute anomaly identifier or the second attribute anomaly identifier or the third attribute anomaly identifier, the status monitoring data at the first moment is anomaly identified to obtain the second identifier information representing the first attribute anomaly identifier or the second attribute anomaly identifier or the third attribute anomaly identifier, which is sent to the remote client for management.

[0040] Furthermore, such as Figure 3 As shown, fitness analysis is performed on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation between the control parameter state information and the environmental indicator state information. Step S200 includes the following steps:

[0041] S210: Using the model of the first power equipment as the acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, perform data mining to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records;

[0042] S220: Perform fitness analysis on the state monitoring record dataset to obtain a first fitness, wherein the first fitness represents the number of times the control parameter state information and the environmental indicator state information co-occur in the N state monitoring record datasets;

[0043] S230: Obtain the first calculation result of N minus the first fitness, determine the proportion of the first calculation result in the N state monitoring record datasets, and generate the first anomaly coefficient.

[0044] In a preferred embodiment, the first anomaly coefficient is preferably obtained as follows:

[0045] The model of the first power equipment is used as the object of data collection, and the control parameter status information and environmental indicator status information are used as the target data for data mining to obtain N status monitoring record datasets, where N is greater than or equal to 150. Any status monitoring record dataset includes control parameter status record data and environmental indicator status information at the same time.

[0046] The first fitness represents the number of times the control parameter state information and the environmental indicator state information co-occur in the N state monitoring record datasets. The control parameter state information and the environmental indicator state information appearing in the same record are considered to co-occur. Preferably, the remote client sequentially sets multiple tolerance intervals for the control parameter state information and the environmental indicator state information. Even if the N state monitoring record datasets deviate from the control parameter state information and the environmental indicator state information, but still fall within the multiple tolerance intervals, the control parameter state information and the environmental indicator state information are still considered to co-occur.

[0047] Furthermore, the first fitness is subtracted from N and set as the first calculation result. Then, the proportion of the first calculation result in the N state monitoring record datasets is calculated to generate the first anomaly coefficient. The preferred calculation formula is as follows:

[0048]

[0049] Where p1 represents the first anomaly coefficient, f represents the first fitness, and Nf represents the first calculation result.

[0050] Furthermore, using the model of the first power equipment as the data acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, data mining is performed to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records. Step S210 includes the following steps:

[0051] S211: Obtain the N value assignment result set by the remote client, wherein the N value assignment result is greater than or equal to 150;

[0052] S212: Divide the N value assignment result into two equal parts to obtain the number of equal parts;

[0053] S213: Using the first power equipment model as the acquisition target, the control parameter status information as the retrieval target, and the environmental indicator status information and the equal division quantity as constraints, perform data mining to obtain the first state monitoring record dataset.

[0054] S214: Using the first power equipment model as the data acquisition target, the environmental indicator status information as the retrieval target, and the control parameter status information and the number of equal divisions as constraints, perform data mining to obtain the second state monitoring record dataset.

[0055] S215: Merge the first state monitoring record dataset and the second state monitoring record dataset to obtain the N state monitoring record datasets.

[0056] In a preferred embodiment, the N value assignment result set by the remote client is obtained, wherein the N value assignment result is greater than or equal to 150, that is, the N value is customized by the user client. The larger the N value, the shorter the processing time, but the lower the accuracy; the larger the N value, the longer the processing time, but the higher the accuracy. The N value is divided by 2 to obtain the number of equal divisions.

[0057] Using the first power equipment model as the data acquisition target, the control parameter status information as the retrieval target, and the environmental indicator status information and the equal division quantity as constraints, data mining is performed to obtain the first state monitoring record dataset. That is, data mining is performed with the environmental indicator status information fault tolerance interval as the constraint, and a monitoring record dataset that meets the equal division quantity is collected. In the first state monitoring record dataset, all environmental indicators meet the environmental indicator status information fault tolerance interval.

[0058] Using the first power equipment model as the data collection target, the environmental indicator status information as the retrieval target, and the control parameter status information and the equal division quantity as constraints, data mining is performed to obtain a second status monitoring record dataset. Specifically, data mining is performed using the control parameter status information tolerance range as a constraint to collect a monitoring record dataset that satisfies the equal division quantity. In the second status monitoring record dataset, all control parameters conform to the control parameter status information tolerance range. The first status monitoring record dataset and the second status monitoring record dataset are then merged to obtain the N status monitoring record datasets.

[0059] Furthermore, when the first anomaly coefficient is less than the first anomaly threshold, a safety assessment is performed on the first power equipment based on the environmental indicator status information to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the hazard level of the first power equipment. Step S300 includes the following steps:

[0060] S310: Using the model of the first power equipment and the environmental indicator status information as constraints, and fault record data as the retrieval target, data mining is performed to obtain equipment fault record data;

[0061] S320: Obtain the environmental indicator deviation threshold set by the remote client, perform cluster analysis on the equipment fault record data, and obtain the fault record data clustering result;

[0062] S330: Based on the clustering results of the fault record data, calculate the fault record frequency and environmental indicator status range of any cluster;

[0063] S340: Calculate the proportion of the fault record frequency of any cluster to the total fault record frequency, and set it as the second anomaly coefficient of the environmental indicator state interval.

[0064] In a preferred embodiment, data mining is performed using the model of the first power equipment and environmental indicator status information as constraints, and fault record data as the retrieval target to obtain equipment fault record data, which refers to event records that affect the normal operation of the equipment.

[0065] The environmental indicator deviation threshold set by the remote client is obtained, and cluster analysis is performed on the equipment fault record data to obtain the fault record data clustering results. The environmental indicator deviation threshold refers to the index deviation clustering threshold set by the client. During clustering, if the index deviation is greater than or equal to the environmental indicator deviation threshold, it is considered as two classes; if it is less than the environmental indicator deviation threshold, it is considered as one class. The clustering is repeated. Finally, the deviation of any two environmental indicator feature values ​​in any two groups of the fault record data clustering results is greater than or equal to the environmental indicator deviation threshold.

[0066] Based on the clustering results of the fault record data, the fault record frequency and environmental indicator state interval of any cluster are statistically analyzed. Any environmental indicator state interval refers to the interval constructed from the minimum to the maximum value of the environmental indicator feature value within any cluster. The fault record frequency refers to the number of clusters of equipment fault record data in any cluster. The proportion of the fault record frequency of any cluster to the total fault record frequency is calculated and set as the second anomaly coefficient of the environmental indicator state interval. The total fault record frequency refers to the sum of the fault record frequencies of multiple clusters.

[0067] Furthermore, when the second anomaly coefficient is less than the second anomaly threshold, a third anomaly coefficient is obtained based on long-term fault analysis of the control parameter status information, wherein the third anomaly coefficient characterizes the failure probability of the first power equipment during long-term operation of the control parameter status information. Step S400 includes the following steps:

[0068] S410: Obtain the long-term fault analysis model embedded in the power information communication system based on abnormal state identification;

[0069] S420: Obtain the first set of analysis durations set by the remote client;

[0070] S430: The control parameter status information is traversed through the first analysis duration set and sequentially input into the long-term fault analysis model to obtain the third abnormal coefficient set.

[0071] In a preferred embodiment, the long-term fault analysis model refers to an intelligent model used to analyze the fault probability of control parameter state information under different operating durations, and is preferably constructed in the following manner:

[0072] Furthermore, obtaining the long-term fault analysis model embedded in the power information communication system based on abnormal state identification, step S410 includes the following steps:

[0073] S411: Using the model of the first power equipment as the data collection object, collect data from the long-term fault analysis model to construct a dataset;

[0074] S412: Set the control parameter status record data and running duration of the dataset constructed by the long-term fault analysis model as input data, set the fault probability identification information of the dataset constructed by the long-term fault analysis model as output supervision data, and train the long-term fault analysis model based on the long short-term memory neural network.

[0075] S413: When the mean square error of the long-term fault analysis model is less than or equal to the expected value of the mean square error, the long-term fault analysis model is embedded in the power information communication system based on abnormal state identification.

[0076] Using the first power equipment model as the data acquisition target, a dataset is constructed from long-term fault analysis models. Any dataset constructed from these long-term fault analysis models includes control parameter status records, operating duration, and fault probability identification information. The fault probability identification information is preferably determined in the following manner:

[0077] Using control parameter status recording data and operating duration as constraints, operation monitoring data is collected. Any set of control parameter status recording data and operating duration corresponds to multiple sets of operation monitoring data. The operation monitoring data includes fault operation recording data. The proportion of fault operation recording data in the corresponding operation monitoring data for any set of control parameter status recording data and operating duration is calculated and recorded as the fault probability identification information for any set of control parameter status recording data and operating duration.

[0078] The control parameter status record data and operating duration of the dataset used to construct the long-term fault analysis model are set as input data, and the fault probability identification information of the dataset is set as output supervision data. The long-term fault analysis model is trained based on a long short-term memory neural network. When the mean squared error (MSE) of the long-term fault analysis model is less than or equal to the expected value of the MSE, the long-term fault analysis model is embedded in the power information communication system based on abnormal state identification. Here, the loss value refers to the deviation between the model output and the output supervision data, and the MSE refers to the output loss value from the start of training on a certain set of data to the end of training on another set of data being less than or equal to the expected value of the MSE. The average loss value of the output of all sets in this training is calculated. When the MSE of the long-term fault analysis model is less than or equal to the expected value of the MSE, the long-term fault analysis model is embedded in the power information communication system based on abnormal state identification. The expected value of the MSE refers to a user-defined convergence loss threshold.

[0079] Obtain the first set of analysis durations set by the remote client. The first set of analysis durations is the duration that the expected control parameter status information can last, which is customized by the remote client. Input the control parameter status information and any one of the durations in the first set of analysis durations into the long-term fault analysis model to obtain the third set of abnormal coefficients, which is used to identify the abnormal running duration in the next step.

[0080] In summary, the embodiments of this application have at least the following technical effects:

[0081] This application embodiment performs a fitness assessment on the control parameter status information and environmental indicator status information of the real-time monitoring data at the first moment. After the fitness assessment is qualified, a safety assessment is performed on the first power equipment based on the environmental indicator status information. After the safety assessment is qualified, long-term fault analysis is performed based on the control parameter status information. The fitness assessment can verify the possibility of deviation of the uploaded monitoring data, the safety assessment can analyze the degree of impact of the current environmental state on the equipment, and the long-term fault analysis can determine the stability of the equipment during long-term operation. Then, the data is identified based on the analysis results of the three dimensions before transmission, which improves the referenceability and visualization of the transmitted data.

[0082] Example 2: Based on the same inventive concept as the power information communication method based on abnormal state identification in the foregoing examples, such as... Figure 4 As shown, this application provides a power information communication system based on abnormal state identification, including:

[0083] The monitoring status data acquisition module 11 is used to acquire the status monitoring data of the first power equipment at the first moment, wherein the status monitoring data at the first moment includes control parameter status information and environmental indicator status information.

[0084] The first anomaly identification module 12 is used to perform fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation between the control parameter state information and the environmental indicator state information.

[0085] The second anomaly identification module 13 is used to perform a safety assessment on the first power equipment based on the environmental indicator status information when the first anomaly coefficient is less than the first anomaly threshold, and obtain a second anomaly coefficient, wherein the second anomaly coefficient represents the danger level of the first power equipment.

[0086] The third anomaly identification module 14 is used to obtain a third anomaly coefficient based on long-term fault analysis of the control parameter status information when the second anomaly coefficient is less than the second anomaly threshold, wherein the third anomaly coefficient represents the probability of failure of the first power equipment when the control parameter status information is running for a long time.

[0087] The information communication module 15 is used to perform health identification on the status monitoring data at the first moment when the third abnormality coefficient is less than the third abnormality threshold, generate first identification information, and send it to the remote client for storage.

[0088] Furthermore, the information communication module 15 further includes the following steps:

[0089] When the first anomaly coefficient is greater than or equal to the first anomaly threshold, a first attribute anomaly identifier is generated; or

[0090] When the second anomaly coefficient is greater than or equal to the second anomaly threshold, a second attribute anomaly identifier is generated; or

[0091] When the third anomaly coefficient is greater than or equal to the third anomaly threshold, a third attribute anomaly identifier is generated;

[0092] Based on the first attribute anomaly identifier, the second attribute anomaly identifier, or the third attribute anomaly identifier, the status monitoring data at the first moment is anomaly identified, second identifier information is generated, and sent to the remote client for management.

[0093] Furthermore, the first anomaly identification module 12 performs the following steps:

[0094] Using the model of the first power equipment as the data acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, data mining is performed to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records.

[0095] A fitness analysis is performed on the state monitoring record dataset to obtain a first fitness, wherein the first fitness represents the number of times the control parameter state information and the environmental indicator state information co-occur in the N state monitoring record datasets;

[0096] Obtain the first calculation result of N minus the first fitness, determine the proportion of the first calculation result in the N state monitoring record datasets, and generate the first anomaly coefficient.

[0097] Furthermore, the first anomaly identification module 12 performs the following steps:

[0098] Obtain the N value assignment result set by the remote client, wherein the N value assignment result is greater than or equal to 150;

[0099] The N value assignment result is divided into two equal parts to obtain the number of equal parts;

[0100] Data mining is performed using the first power equipment model as the data collection target, the control parameter status information as the retrieval target, and the environmental indicator status information and the number of equal divisions as constraints, to obtain the first status monitoring record dataset.

[0101] Using the first power equipment model as the data collection target, the environmental indicator status information as the retrieval target, and the control parameter status information and the number of equal divisions as constraints, data mining is performed to obtain the second state monitoring record dataset.

[0102] The first state monitoring record dataset and the second state monitoring record dataset are merged to obtain the N state monitoring record datasets.

[0103] Furthermore, the second anomaly identification module 13 performs the following steps:

[0104] Data mining was performed using the first power equipment model and environmental indicator status information as constraints, and fault record data as the retrieval target, to obtain equipment fault record data.

[0105] Obtain the environmental indicator deviation threshold set by the remote client, perform cluster analysis on the equipment fault record data, and obtain the fault record data clustering results;

[0106] Based on the clustering results of the fault record data, the frequency of fault records and the range of environmental indicator status for any cluster are statistically analyzed.

[0107] Calculate the proportion of the fault record frequency of any cluster to the total fault record frequency, and set it as the second anomaly coefficient of the environmental indicator state interval.

[0108] Furthermore, the third anomaly identification module 14 performs the following steps:

[0109] Obtain a long-term fault analysis model embedded in a power information communication system based on abnormal state identification;

[0110] Obtain the first set of analysis durations set by the remote client;

[0111] The control parameter status information is traversed through the first analysis duration set and sequentially input into the long-term fault analysis model to obtain the third abnormal coefficient set.

[0112] Furthermore, the third anomaly identification module 14 performs the following steps:

[0113] Using the model of the first power equipment as the data collection object, a dataset is constructed by collecting data from a long-term fault analysis model.

[0114] The control parameter status record data and running duration of the dataset used to construct the long-term fault analysis model are set as input data, and the fault probability identification information of the dataset used to construct the long-term fault analysis model is set as output supervision data. The long-term fault analysis model is trained based on a long short-term memory neural network.

[0115] When the mean square error of the long-term fault analysis model is less than or equal to the expected value of the mean square error, the long-term fault analysis model is embedded in the power information communication system based on abnormal state identification.

[0116] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A power information communication method based on abnormal state identification, characterized in that, Applications include power information communication systems based on anomaly state identification, including: Acquire the first moment status monitoring data of the first power equipment, wherein the first moment status monitoring data includes control parameter status information and environmental indicator status information; A fitness analysis is performed on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the degree of matching deviation between the control parameter state information and the environmental indicator state information; When the first anomaly coefficient is less than the first anomaly threshold, a safety assessment is performed on the first power equipment based on the environmental indicator status information to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the danger level of the first power equipment. When the second anomaly coefficient is less than the second anomaly threshold, a third anomaly coefficient is obtained based on the long-term fault analysis of the control parameter status information, wherein the third anomaly coefficient represents the probability of failure of the first power equipment when the control parameter status information is running for a long time. When the third abnormality coefficient is less than the third abnormality threshold, the status monitoring data at the first moment is marked with a health indicator, a first identification information is generated, and it is sent to the remote client for storage. The process includes performing fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient. This first anomaly coefficient characterizes the degree of mismatch between the control parameter state information and the environmental indicator state information, including: Using the model of the first power equipment as the data acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, data mining is performed to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records. A fitness analysis is performed on the state monitoring record dataset to obtain a first fitness, wherein the first fitness represents the number of times the control parameter state information and the environmental indicator state information co-occur in the N state monitoring record datasets; Obtain the first calculation result of N minus the first fitness, determine the proportion of the first calculation result in the N state monitoring record datasets, and generate the first anomaly coefficient; When the first anomaly coefficient is less than the first anomaly threshold, a safety assessment is performed on the first power equipment based on the environmental indicator status information to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the hazard level of the first power equipment, including: Data mining was performed using the first power equipment model and environmental indicator status information as constraints, and fault record data as the retrieval target, to obtain equipment fault record data. Obtain the environmental indicator deviation threshold set by the remote client, perform cluster analysis on the equipment fault record data, and obtain the fault record data clustering results; Based on the clustering results of the fault record data, the frequency of fault records and the range of environmental indicator status for any cluster are statistically analyzed. Calculate the proportion of the fault record frequency of any cluster to the total fault record frequency, and set it as the second anomaly coefficient of the environmental indicator state interval; When the second anomaly coefficient is less than the second anomaly threshold, a third anomaly coefficient is obtained based on long-term fault analysis of the control parameter status information. The third anomaly coefficient characterizes the probability of failure of the first power equipment during long-term operation of the control parameter status information, including: Obtain a long-term fault analysis model embedded in a power information communication system based on abnormal state identification; Obtain the first set of analysis durations set by the remote client; The control parameter status information is traversed through the first analysis duration set and sequentially input into the long-term fault analysis model to obtain the third abnormal coefficient set.

2. The method as described in claim 1, characterized in that, Also includes: When the first abnormal coefficient is greater than or equal to the first abnormal threshold, a first attribute abnormality identifier is generated; or When the second anomaly coefficient is greater than or equal to the second anomaly threshold, a second attribute anomaly identifier is generated; or When the third anomaly coefficient is greater than or equal to the third anomaly threshold, a third attribute anomaly identifier is generated; Based on the first attribute anomaly identifier, the second attribute anomaly identifier, or the third attribute anomaly identifier, the status monitoring data at the first moment is anomaly identified, second identifier information is generated, and sent to the remote client for management.

3. The method as described in claim 1, characterized in that, Using the model of the first power equipment as the data acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, data mining is performed to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records, including: Obtain the N value assignment result set by the remote client, wherein the N value assignment result is greater than or equal to 150; The N value assignment result is divided into two equal parts to obtain the number of equal parts; Data mining is performed using the first power equipment model as the data collection target, the control parameter status information as the retrieval target, and the environmental indicator status information and the number of equal divisions as constraints, to obtain the first status monitoring record dataset. Using the first power equipment model as the data collection target, the environmental indicator status information as the retrieval target, and the control parameter status information and the number of equal divisions as constraints, data mining is performed to obtain the second state monitoring record dataset. The first state monitoring record dataset and the second state monitoring record dataset are merged to obtain the N state monitoring record datasets.

4. The method as described in claim 1, characterized in that, Obtain a long-term fault analysis model embedded in a power information communication system based on abnormal state identification, including: Using the model of the first power equipment as the data collection object, a dataset is constructed by collecting data from a long-term fault analysis model. The control parameter status record data and running duration of the dataset used to construct the long-term fault analysis model are set as input data, and the fault probability identification information of the dataset used to construct the long-term fault analysis model is set as output supervision data. The long-term fault analysis model is trained based on a long short-term memory neural network. When the mean square error of the long-term fault analysis model is less than or equal to the expected value of the mean square error, the long-term fault analysis model is embedded in the power information communication system based on abnormal state identification.

5. A power information communication system based on abnormal state identification, characterized in that, include: The monitoring status data acquisition module is used to acquire the status monitoring data of the first power equipment at the first moment, wherein the status monitoring data at the first moment includes control parameter status information and environmental indicator status information; The first anomaly identification module is used to perform fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient, wherein the first anomaly coefficient characterizes the matching deviation between the control parameter state information and the environmental indicator state information. The second anomaly identification module is used to perform a safety assessment on the first power equipment based on the environmental indicator status information when the first anomaly coefficient is less than the first anomaly threshold, and obtain a second anomaly coefficient, wherein the second anomaly coefficient represents the danger level of the first power equipment. The third anomaly identification module is used to obtain a third anomaly coefficient based on long-term fault analysis of the control parameter status information when the second anomaly coefficient is less than the second anomaly threshold. The third anomaly coefficient represents the probability of failure of the first power equipment when the control parameter status information is running for a long time. The information communication module is used to identify the health status monitoring data at the first moment when the third abnormality coefficient is less than the third abnormality threshold, generate first identification information, and send it to the remote client for storage. The process includes performing fitness analysis on the control parameter state information and the environmental indicator state information to obtain a first anomaly coefficient. This first anomaly coefficient characterizes the degree of mismatch between the control parameter state information and the environmental indicator state information, including: Using the model of the first power equipment as the data acquisition target, and the control parameter status information and the environmental indicator status information as the retrieval targets, data mining is performed to obtain a dataset of N status monitoring records, where N is greater than or equal to 150 records. A fitness analysis is performed on the state monitoring record dataset to obtain a first fitness, wherein the first fitness represents the number of times the control parameter state information and the environmental indicator state information co-occur in the N state monitoring record datasets; Obtain the first calculation result of N minus the first fitness, determine the proportion of the first calculation result in the N state monitoring record datasets, and generate the first anomaly coefficient; When the first anomaly coefficient is less than the first anomaly threshold, a safety assessment is performed on the first power equipment based on the environmental indicator status information to obtain a second anomaly coefficient, wherein the second anomaly coefficient characterizes the hazard level of the first power equipment, including: Data mining was performed using the first power equipment model and environmental indicator status information as constraints, and fault record data as the retrieval target, to obtain equipment fault record data. Obtain the environmental indicator deviation threshold set by the remote client, perform cluster analysis on the equipment fault record data, and obtain the fault record data clustering results; Based on the clustering results of the fault record data, the frequency of fault records and the range of environmental indicator status for any cluster are statistically analyzed. Calculate the proportion of the fault record frequency of any cluster to the total fault record frequency, and set it as the second anomaly coefficient of the environmental indicator state interval; When the second anomaly coefficient is less than the second anomaly threshold, a third anomaly coefficient is obtained based on long-term fault analysis of the control parameter status information. The third anomaly coefficient characterizes the probability of failure of the first power equipment during long-term operation of the control parameter status information, including: Obtain a long-term fault analysis model embedded in a power information communication system based on abnormal state identification; Obtain the first set of analysis durations set by the remote client; The control parameter status information is traversed through the first analysis duration set and sequentially input into the long-term fault analysis model to obtain the third abnormal coefficient set.

Images