Intelligent fault judgment method for electric power Internet of Things

Abstract

The invention discloses an intelligent fault judgment method of an electric power Internet of Things, and relates to the technical field of the electric power Internet of Things. In the intelligent fault judgment process of the electric power Internet of Things, data processing modes corresponding to various fault data are deeply analyzed, and a data change smooth coefficient and a data discrete mutation coefficient are calculated; according to the method, the type of the fault data can be accurately determined, different types of data always reflect the operation states and fault features of the equipment in different aspects, the accurate classification enables the subsequent fault judgment to be more targeted, and when the residual fault data is processed in a mode of fusing the fault data processing models, the fault data processing efficiency is improved. The data matching degree evaluation coefficient is calculated in combination with the histogram, and then the comprehensive fault evaluation coefficient is obtained, so that whether the power equipment corresponding to the residual fault data has the fault condition and the severity of the fault can be analyzed objectively and comprehensively.

Description

Technical Field

[0001] The present invention relates to the technical field of the Internet of Electric Power, and particularly to an intelligent fault judgment method for the Internet of Electric Power. Background Art

[0002] With the rapid development of the Internet of Electric Power, the number of devices has increased and the operating environment has become complex, resulting in frequent failures and increased diagnostic difficulty. Traditional fault judgment methods are difficult to meet the requirements of accuracy and efficiency, which has promoted the emergence of the above intelligent fault judgment method, aiming to improve the accuracy and timeliness of fault diagnosis by using advanced technologies.

[0003] The prior art, such as the invention patent application with the publication number CN114709923A, discloses an intelligent fault judgment method for the Internet of Electric Power, including the following steps: S1: Divide the intelligent fault diagnosis system architecture of the Internet of Electric Power into three parts: the perception layer, the communication layer, and the application layer; S2: Establishment of the perception layer; S3: Establishment of the communication layer; S4: Establishment of the application layer; S5: Construction of the maintenance scheduling module. The present invention analyzes the device status information in the Internet of Things information platform through the cooperation of each module, and judges the type of fault occurred to the device according to certain criteria. The fault diagnosis architecture will send the diagnosis result and relevant maintenance decisions to the maintenance personnel and relevant responsible persons in the form of short messages in real time, eliminate the potential fault hazards of the device in time, repair the device, reduce the occurrence of some unnecessary accidents, and effectively guarantee the real-time nature of fault diagnosis.

[0004] Regarding the above solution, there are at least the following technical problems: 1. The above solution lacks the accurate analysis and processing process of fault data, which will lead to a decrease in accuracy and reliability in actual intelligent fault judgment. The above solution lacks the data processing methods corresponding to how to analyze various types of fault data. In a complex Internet of Electric Power environment, different types of fault data have different characteristics and variation laws. If accurate classification and targeted processing are not carried out, the fault diagnosis system will misjudge or miss judge the fault situation, and it will not be able to effectively discover the potential problems of the device in time, thus increasing the risk of device failures and reducing the stability and safety of the power system.

[0005] 2. The above solution lacks the relevant content of processing data and evaluating the abnormal fault judgment situations corresponding to various types of fault data through the fusion of each fault data processing model. In actual applications, a single fault data processing model often has limitations and cannot comprehensively and accurately handle various complex fault scenarios. The lack of a model fusion method will cause the system to be unable to give full play to the advantages of different models when facing diverse fault data, reducing the accuracy rate and adaptability of fault judgment.

[0006] 3. The steps of obtaining the characteristic fault location index corresponding to various fault data in the case of faults in the above solution and evaluating the specific fault data in the fault data corresponding to each power device in the case of faults will result in the inability to quickly and accurately locate which specific data reflects the most critical problem after determining that the device has a fault. It is difficult for maintenance personnel to quickly find the root cause of the fault, prolonging the maintenance time of the device, reducing the recovery efficiency of the power system, and further affecting the reliability and stability of power supply. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent fault judgment method for a power Internet of Things, which solves the problems in the background technology.

[0008] To solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides an intelligent fault judgment method for a power Internet of Things, including: S1. When various fault data generated by each power device are transmitted to a designated power Internet of Things terminal, analyze the data processing methods corresponding to various fault data.

[0009] S2. According to the data processing methods corresponding to various fault data, further evaluate the abnormal fault judgment situations corresponding to various fault data.

[0010] S3. When various fault data are processed by each fault data processing model, if there is a fault situation, calculate the characteristic fault location index corresponding to various fault data in the case of faults.

[0011] S4. According to the characteristic fault location index corresponding to various fault data in the case of faults, further evaluate the specific fault data in the fault data corresponding to each power device in the case of faults.

[0012] The beneficial effects of the present invention are as follows: 1. An intelligent fault judgment method for a power Internet of Things provided by an embodiment of the present invention, in the process of intelligent fault judgment of the power Internet of Things, by deeply analyzing the data processing methods corresponding to various fault data and calculating the data change smoothing coefficient and the data discrete mutation coefficient, it is beneficial to accurately determine the type to which the fault data belongs. Different types of data often reflect different aspects of the operating state and fault characteristics of the device, and accurate classification can make the subsequent fault judgment more targeted.

[0013] 2. When evaluating the abnormal fault judgment situations corresponding to various types of fault data in the embodiments of the present invention, the remaining types of fault data that do not conform to common data types are processed by integrating various fault data processing models, which is conducive to giving full play to the advantages of different models, thereby improving the accuracy of fault judgment corresponding to various types of fault detection data. By integrating various fault data processing models, the limitations of a single model are made up for. When facing complex and changeable power equipment fault situations, fault signals can be identified more comprehensively and accurately, the risks of misjudgment and missed judgment are reduced, and the stable operation of the power system is guaranteed.

[0014] 3. In the process of calculating the characteristic fault location index corresponding to various types of fault data with fault situations and evaluating the specific fault data in the fault data corresponding to each power equipment with fault situations in the embodiments of the present invention, it is conducive to accurately locating the root cause of the fault. When a power equipment fails, it is crucial to quickly and accurately find the key data and components that cause the fault. Screening out the factors most likely to cause the fault from numerous fault monitoring data enables maintenance personnel to carry out maintenance work in a targeted manner, greatly shortening the maintenance time, reducing power outage losses, and improving the availability and reliability of power equipment.

[0015] 4. In the embodiments of the present invention, by calculating the model adaptability index corresponding to each fault data processing model and combining the histogram to calculate the data matching degree evaluation coefficient, and then obtaining the comprehensive fault evaluation coefficient, it is conducive to objectively and comprehensively analyzing whether there are fault situations and the severity of the faults in the power equipment corresponding to each remaining type of fault data. The model adaptability index reflects the performance advantages and disadvantages of each fault data processing model in the current data processing task, and the data matching degree evaluation coefficient provides supplementary information from the perspective of data distribution characteristics. The comprehensive fault evaluation coefficient formed by combining the two can more accurately measure the possibility of fault occurrence and the fault level, comprehensively considering multiple factors, improving the scientificity and reliability of fault judgment, and providing a strong guarantee for the safe and stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic diagram of the implementation steps of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Please refer to Figure 1 As shown, the present invention provides an intelligent fault judgment method for the power Internet of Things. The method includes: S1. When various types of fault data generated by each power device are transmitted to a specified power Internet of Things terminal, analyze the data processing methods corresponding to various types of fault data.

[0020] In a specific embodiment, the process of analyzing the data processing methods corresponding to various types of fault data is as follows: According to the set data collection time points, through the sensor devices set in each power device in the specified power Internet of Things, various types of fault data of each power device at each data collection time point are collected. After the sensor devices transmit various types of fault data generated by each power device to the specified power Internet of Things terminal, by calculating the data change smoothing coefficient and data discrete mutation coefficient corresponding to various types of fault data, which are respectively denoted as DVi and DDMi, where i is the number corresponding to the type of fault data, i = 1, 2,..., n, n represents the total number of types of fault monitoring data, and n is a positive integer. Then, analyze the data types to which various types of fault data belong. The data types to which they belong include continuous data, discrete data, and time-series data. According to the data types to which various types of fault data belong.

[0021] Obtain the data types processed by each fault data processing model from the database, thereby determining the corresponding fault data processing models for various types of fault data when performing fault judgment, and then process various types of fault data through each fault data processing model respectively.

[0022] It should be noted that taking a power transformer as an example, various types of fault data include continuous data such as winding temperature, discrete data such as the number of tap changer operations, and time-series data such as load current, etc.

[0023] In a specific embodiment, the process of calculating the data discrete mutation coefficient corresponding to various types of fault data is as follows: Through the calculation formula: Obtain the data change smoothing coefficient DVi corresponding to various types of fault data. t is the number corresponding to the data collection time point, t = 1, 2,..., m, m represents the total number of data collection time points, and m is a positive integer.

[0024] X i (t + 1)-Xit |represents the absolute value of the difference between the data value corresponding to the i-th type of fault data at the (t + 1)-th data acquisition time point and the data value corresponding to the t-th data acquisition time point, denoted as the absolute value of each difference, min(X i (t + 1)-X it ) represents the minimum value among the absolute values of each difference.

[0025] It should be noted that the "data value" refers to the specific value obtained for each type of fault data at each data acquisition time point. For example, for the voltage monitoring data of a certain power equipment, the voltage value collected at the 1st data acquisition time point is 220 volts, and the voltage value collected at the 2nd data acquisition time point is 218 volts. Then these two voltage values are the corresponding data values.

[0026] In a specific embodiment, the process of calculating the data discrete mutation coefficient corresponding to each type of fault data is as follows: Through the calculation formula:

[0027] the data discrete mutation coefficient DDMi corresponding to each type of fault data is obtained, where λi represents the standard deviation corresponding to the i-th type of fault data, represents the number of absolute values of differences greater than the sum of the mean value of the data value corresponding to the i-th type of fault data and 3 times the standard deviation.

[0028] It should be noted that through the calculation formula: the standard deviation λi corresponding to the i-th type of fault data is obtained. The standard deviation is a statistic used to measure the degree of dispersion of a set of data. In the context of calculating the data discrete mutation coefficient, for example, for the temperature data of a certain power equipment, the standard deviation reflects the dispersion of temperature values relative to the average value. When calculating the data discrete mutation coefficient, if the equipment temperature data is concentrated under normal conditions, the standard deviation is small. When the equipment has faults such as local overheating, the degree of dispersion of the temperature data increases, the standard deviation becomes larger, and the number of absolute values of differences greater than the sum of the mean value and 3 times the standard deviation also changes.

[0029] In the embodiment of the present invention, when evaluating the abnormal situation of fault judgment corresponding to each type of fault data, the remaining type of fault data that does not conform to the common data type is processed by the method of fusing each fault data processing model, which is beneficial to giving full play to the advantages of different models, thereby improving the accuracy of fault judgment corresponding to each type of fault detection data. By fusing each fault data processing model, the limitations of a single model are compensated. When facing complex and changeable power equipment fault situations, fault signals can be identified more comprehensively and accurately, the risks of misjudgment and missed judgment are reduced, and the stable operation of the power system is guaranteed.

[0030] S2. Based on the data processing methods corresponding to various types of fault data, further evaluate the abnormal situations of fault judgments corresponding to various types of fault data.

[0031] In a specific embodiment, the process of analyzing and obtaining the data types to which various types of fault data belong is as follows: Obtain the preset standard data change smoothing coefficient threshold, standard data discrete mutation coefficient threshold, and standard discrete proportion threshold from the database, and denote them as DV′, DDM′, and DDM″ respectively. Compare the data change smoothing coefficient and data discrete mutation coefficient corresponding to various types of fault data with the standard data change smoothing coefficient threshold and standard data discrete mutation coefficient threshold respectively.

[0032] If (DV i ≥DV′) ∧ (DDM i ≤DDM′), it indicates that the data type to which the i-th type of fault data belongs is continuous data, and set the discrete proportion Bi corresponding to various types of fault data.

[0034] If (DV i <DV′) ∧ (DDM i >DDM′) and B i ≥DDM″, it indicates that the data type to which the i-th type of fault data belongs is discrete data. If (DV i <DV′) ∧ (DDM i ≤DDM′), it indicates that the data type to which the i-th type of fault data belongs is time-series data.

[0035] It should be noted that the process of setting the discrete proportion Bi is as follows: For example, when monitoring the operation of power equipment in a substation, voltage, current, and temperature are collected. Within the set time, a total of 100 data points are collected. Among them, there are 40 valid data points for voltage data, 30 valid data points for current data, and 30 valid data points for temperature data. Then the discrete proportion of voltage data is the number of valid data points of voltage data, 40, divided by the total number of data points, 100, and the discrete proportions of current and temperature are obtained in the same way.

[0036] In a specific embodiment, the process of evaluating the abnormal situations of fault judgments corresponding to various types of fault data is as follows: According to the data types to which various types of fault data belong, classify the various types of fault data that do not belong to continuous data, discrete data, and time-series data as the remaining types of fault data. Then, through the method of fusing various fault data processing models, perform data processing on the remaining types of fault data, so as to achieve the purpose of improving the accuracy of fault judgment corresponding to the fault detection data of each type.

[0037] It should be noted that the intelligent fault judgment of various types of fault data through each fault data processing model has been involved in the patent document with the publication number CN117269641A, and will not be elaborated here.

[0038] In a specific embodiment, the data processing of this type of fault data is performed by fusing each fault data processing model, and the specific process is as follows: Denote the fault results corresponding to each remaining type of fault data processed by the fusion method of each fault data processing model as Q qj , where q is the number of each remaining type of fault data, q = 1, 2,..., p, p represents the total number of remaining types of fault data, p is a positive integer, j is the number of each fault data processing model, j = 1, 2,..., y, y represents the total number of fault data processing models, and y is a positive integer;

[0039] Through the calculation formula: Obtain the model adaptability index α corresponding to each fault data processing model j , where acc j , rec j , sta j respectively represent accuracy, recall rate, and stability in coping with environmental changes. μ1, μ2, and μ3 are respectively the set weight factors for accuracy, recall rate, and stability in coping with environmental changes, and calculate the data matching degree evaluation coefficient β based on the histogram corresponding to each fault data processing model j , and thus through the calculation formula: Obtain the comprehensive fault evaluation coefficient γq corresponding to each remaining type of fault data, and thus analyze whether there is a fault in the power equipment corresponding to each remaining type of fault data.

[0040] It should be noted that the accuracy is obtained by the ratio of the number of data correctly judged by the fault data processing model to the total number of judged data. The recall rate is the ratio of the number of fault data correctly judged by the fault data processing model to the actual number of fault data. The stability in coping with environmental changes is obtained by measuring the fluctuation of the output results of the model under different working conditions and environmental changes, such as the variance of the output results. The specific analysis process is the prior art and will not be elaborated here.

[0041] It should also be noted that taking a small hydropower station as an example, assuming there are models A, B, and C. For model A, during the stable power generation period, it accounts for 60%. The accuracy rate is calculated by counting the number of correct judgments on faults and normal states and the total number of judgments. Assuming the accuracy rate is 80%, the recall rate is the ratio of the number of correctly judged faults to the actual number of faults, assuming it is 75%. The stability is measured by comparing the output fluctuations in different environments. Assuming the full score of the stability score is 100 points, it is set to 70 points. The analytic hierarchy process is used to determine the accuracy weight of 0.4. During the flood season, it accounts for 25%. The fuzzy comprehensive evaluation method is used to determine the recall weight of 0.35. The grey relational analysis method is used to determine the stability weight of 0.25 according to the changing environment. The analytic hierarchy process, the fuzzy comprehensive evaluation method, and the grey relational analysis method are all analysis processes of existing technologies and will not be elaborated here.

[0042] In a specific embodiment, the process of analyzing whether there is a fault in the power equipment corresponding to each remaining type of fault data is as follows: Obtain the range intervals corresponding to each preset standard comprehensive fault evaluation coefficient threshold of each power equipment corresponding to each remaining type of fault data from the database, and obtain the corresponding fault levels for the range intervals corresponding to each standard comprehensive fault evaluation coefficient threshold. Compare the comprehensive fault evaluation coefficient corresponding to each remaining type of fault data with the preset standard comprehensive fault evaluation coefficient threshold. If the comprehensive fault evaluation coefficient corresponding to a certain remaining type of fault data belongs to the range interval corresponding to a certain preset standard comprehensive fault evaluation coefficient threshold, it indicates that there is a fault in the power equipment corresponding to this remaining type of fault data, and it indicates that the fault level of the power equipment corresponding to this remaining type of fault data is the fault level corresponding to the range interval corresponding to this standard comprehensive fault evaluation coefficient threshold. Otherwise, it indicates that there is no fault in the power equipment corresponding to this remaining type of fault data.

[0043] In the process of calculating the characteristic fault location index corresponding to various types of fault data with fault situations and evaluating the specific fault data in the fault data corresponding to each power equipment with fault situations in the embodiment of the present invention, it is beneficial to accurately locate the root cause of the fault. When a power equipment fails, it is crucial to quickly and accurately find the key data and components that cause the fault. Screen out the factors most likely to cause the fault from numerous fault monitoring data, enabling maintenance personnel to carry out maintenance work targeted, greatly shortening the maintenance time, reducing power outage losses, and improving the availability and reliability of power equipment.

[0044] S3. When various types of fault data are processed by each fault data processing model, if there is a fault situation, calculate the characteristic fault location index corresponding to various types of fault data with fault situations.

[0045] In a specific embodiment, the calculation of the characteristic fault location index corresponding to various types of fault data with a fault situation is as follows: Denote the class fault data of a certain power device corresponding to a certain data acquisition time point with a fault situation as the characteristic vector E, E = [e1, e2,......, e g , where g represents each data in the class fault data of a certain power device corresponding to a certain data acquisition time point, and g is a positive integer. Input E into the fault data processing model to obtain the corresponding fault probability value F. Set the characteristic perturbation value Δe h , then obtain the new characteristic vector E′, E′ = [e1, e2,...,e h + Δe h ,..., e g , input E′ into the fault data processing model to obtain the corresponding fault probability value F′, and then through the calculation formula: Obtain the characteristic fault location index E h corresponding to the h-th data in the fault detection data of a certain type of a power device with a fault situation, and calculate the characteristic fault location indices corresponding to various types of fault data with a fault situation in this way.

[0046] It should be noted that h takes any integer in the interval [1, g], and the setting process of Δe h is as follows: For example, when monitoring the fault of a power device, the input characteristic vector E includes the voltage e1, current e2, and temperature e3 of the power device. For the voltage characteristic e1, if the current voltage value is 220 volts, then set Δe h = 0.01×e1 = 2.2 volts, and thus obtain a small voltage perturbation value Δe h , and Δe h is used to observe the influence on the fault judgment result output by the fault data processing model when this characteristic changes.

[0047] It should also be noted that taking a fault data processing model for power device fault diagnosis as an example, input the characteristic vector E = [220, 10, 50] composed of the voltage of 220 volts, current of 10 amps, and temperature of 50 °C of the power device collected at a certain data acquisition time point into the fault data processing model. The fault data processing model, according to the internal neuron network structure, weight parameters, and trained algorithms, after a complex calculation process, outputs the fault probability value F = 0.2, which means that according to the judgment of the fault data processing model, the probability of the power device having a fault at this time is 20%. The process of the fault data processing model obtaining the fault probability value through a complex calculation is the prior art and will not be elaborated here too much.

[0048] In the embodiments of the present invention, by calculating the model adaptability index corresponding to each fault data processing model and combining with the histogram to calculate the data matching degree evaluation coefficient, and then obtaining the comprehensive fault evaluation coefficient, it is beneficial to objectively and comprehensively analyze whether there is a fault situation in the power equipment corresponding to each remaining type of fault data and the severity of the fault. The model adaptability index reflects the performance of each fault data processing model in the current data processing task, and the data matching degree evaluation coefficient provides supplementary information from the perspective of data distribution characteristics. The comprehensive fault evaluation coefficient formed by combining the two can more accurately measure the probability of fault occurrence and the fault level, comprehensively considering multiple factors, improving the scientificity and reliability of fault judgment, and providing a strong guarantee for the safe and stable operation of the power system.

[0049] S4. According to the characteristic fault location index corresponding to each type of fault data with a fault situation, further evaluate the specific fault data in the fault data corresponding to each power equipment with a fault situation.

[0050] In a specific embodiment, the process of evaluating the specific fault data in the fault data corresponding to each power equipment with a fault situation is as follows: The characteristic fault location index Eh corresponding to the h-th data in the fault detection data of a certain type corresponding to a power equipment with a fault situation h is sorted in descending order. The probability of the data corresponding to the characteristic fault location index ranked first having a fault is the first, and the probability of the data corresponding to the characteristic fault location index ranked second having a fault is the second, so as to obtain the fault probability of the specific fault data existing in the fault data corresponding to each power equipment, and thus obtain the specific fault data in the fault data corresponding to each power equipment with a fault situation.

[0051] It should be noted that, for example, for a certain power transformer, the fault detection data includes winding temperature, load current, insulation resistance, etc. Here, the h-th data refers to a certain one of the numerous fault detection data. When h = 1, it represents the winding temperature.

[0052] A smart fault judgment method for a power Internet of Things provided by the embodiments of the present invention, in the process of smart fault judgment of the power Internet of Things, by deeply analyzing the data processing methods corresponding to various types of fault data and calculating the data change smoothing coefficient and the data discrete mutation coefficient, it is beneficial to accurately determine the type to which the fault data belongs. Different types of data often reflect different aspects of the operating state and fault characteristics of the equipment. Accurate classification can make the subsequent fault judgment more targeted.

[0053] The above content is only an example and illustration of the concept of the present invention. Those skilled in the art of the present technology can make various modifications, supplements, or use similar methods to substitute for the specific embodiments described, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. An intelligent fault judgment method for the power Internet of Things, characterized in that, Including: S1. When various types of fault data generated by each power equipment are transmitted to a specified power Internet of Things terminal, analyze the data processing methods corresponding to the various types of fault data; S2. According to the data processing methods corresponding to the various types of fault data, further evaluate the abnormal fault judgment situations corresponding to the various types of fault data; S3. When the various types of fault data are processed by each fault data processing model, if there is a fault situation, calculate the characteristic fault location indexes corresponding to the various types of fault data with the fault situation; S4. According to the characteristic fault location indexes corresponding to the various types of fault data with the fault situation, further evaluate the specific fault data in the fault data corresponding to each power equipment with the fault situation.

2. The intelligent fault judgment method of an electric power Internet of Things according to claim 1, characterized in that The analysis of the data processing methods corresponding to the various types of fault data is as follows: According to the set data acquisition time points, through the sensor devices set in each power equipment in the specified power Internet of Things, further collect various types of fault data of each power equipment at each data acquisition time point. After the sensor devices transmit the various types of fault data generated by each power equipment to the specified power Internet of Things terminal, by calculating the data change smoothing coefficient and the data discrete mutation coefficient corresponding to the various types of fault data, which are respectively denoted as DVi and DDMi, where i is the number corresponding to the type of fault data, i = 1, 2,..., n, n represents the total number of types of fault monitoring data, and n is a positive integer, and then analyze the data types to which the various types of fault data belong. The data types to which they belong include continuous data, discrete data, and time-series data. According to the data types to which the various types of fault data belong; Obtain the data types processed by each fault data processing model from the database, thereby determine the respective fault data processing models corresponding to the various types of fault data for fault judgment, and then process the various types of fault data through each fault data processing model respectively.

3. The intelligent fault judgment method of an electric power Internet of Things according to claim 2, characterized in that, The specific calculation process of calculating the data discrete mutation coefficient corresponding to the various types of fault data is as follows: Through the calculation formula: Obtain the data change smoothing coefficient DVi corresponding to various types of fault data. t is the number corresponding to the data acquisition time point. t = 1, 2,......, m, where m represents the total number of data acquisition time points and m is a positive integer. |Xi(t + 1)-Xit| represents the absolute value of the difference between the data value corresponding to the i-th type of fault data at the (t + 1)-th data acquisition time point and the data value corresponding to the t-th data acquisition time point, denoted as the absolute value of each difference, min(X i (t + 1)-X it ) represents the minimum value among the absolute values of each difference.

4. The intelligent fault judgment method of an electric power Internet of Things according to claim 3, characterized in that, The specific process of calculating the data discrete mutation coefficient corresponding to the various types of fault data is as follows: Through the calculation formula: Obtain the data discrete mutation coefficient DDMi corresponding to various types of fault data, where λi represents the standard deviation corresponding to the i-th type of fault data, It represents the number of absolute values of the differences that are greater than the sum of the mean value of the data value corresponding to the i-th type of fault data and 3 times the standard deviation.

5. The intelligent fault judgment method of an electric power Internet of Things according to claim 4, characterized in that, The specific process of analyzing the data types to which the various types of fault data belong is as follows: Obtain the preset standard data change smoothing coefficient threshold, standard data discrete mutation coefficient threshold, and standard discrete proportion threshold from the database, and denote them as DV′, DDM′, and DDM″ respectively. Compare the data change smoothing coefficient and the data discrete mutation coefficient corresponding to the various types of fault data with the standard data change smoothing coefficient threshold and the standard data discrete mutation coefficient threshold respectively; If (DV i ≥DV′) ∧ (DDM i ≤DDM′), it indicates that the data type to which the fault data of the i-th category belongs is continuous data, and the discrete ratio Bi corresponding to each category of fault data is set; If (DV i < DV'), and (DDM i > DDM'), and B i ≥ DDM'', it indicates that the data type of the i-th type of fault data is discrete data. If (DV i < DV') and (DDM i ≤ DDM'), it indicates that the data type of the i-th type of fault data is time-series data.

6. The intelligent fault judgment method of an electric power Internet of Things according to claim 5, characterized in that The specific process of evaluating the abnormal fault judgment situations corresponding to the various types of fault data is as follows: According to the data types to which the various types of fault data belong, record the various types of fault data that do not belong to continuous data, discrete data, and time-series data as each remaining type of fault data, and then process the each remaining type of fault data by the method of fusing each fault data processing model, so as to achieve the purpose of improving the accuracy rate of fault judgment corresponding to the fault detection data of each type.

7. The intelligent fault judgment method of an electric power Internet of Things according to claim 6, characterized in that The data processing of this type of fault data is carried out by fusing each fault data processing model, and the specific process is as follows: Denote the fault results corresponding to each remaining type of fault data processed by the fusion method of each fault data processing model as Q qj , where q is the number of each remaining type of fault data, q = 1, 2,......, p, p represents the total number of remaining type of fault data, p is a positive integer, j is the number of each fault data processing model, j = 1, 2,......, y, y represents the total number of fault data processing models, and y is a positive integer; Through the calculation formula: obtain the model adaptability index α corresponding to each fault data processing model j , where acc j , rec j , sta j respectively represent the accuracy rate, recall rate, and stability in coping with environmental changes. μ1, μ2, and μ3 are respectively the set accuracy rate weight factor, recall rate weight factor, and stability weight factor in coping with environmental changes, and calculate the data matching degree evaluation coefficient β based on the histogram corresponding to each fault data processing model j , thus through the calculation formula: obtain the comprehensive fault evaluation coefficient γq corresponding to each remaining type of fault data, and thus analyze whether there is a fault situation in the power equipment corresponding to each remaining type of fault data.

8. The intelligent fault judgment method of an electric power Internet of Things according to claim 7, characterized in that, The process of analyzing whether there are fault conditions in the power equipment corresponding to each remaining type of fault data is as follows: Obtain the range intervals corresponding to each preset standard comprehensive fault evaluation coefficient threshold of each power equipment corresponding to each remaining type of fault data from the database, and obtain the corresponding fault levels for the range intervals corresponding to each standard comprehensive fault evaluation coefficient threshold. Compare the comprehensive fault evaluation coefficient corresponding to each remaining type of fault data with the preset standard comprehensive fault evaluation coefficient threshold. If the comprehensive fault evaluation coefficient corresponding to a certain remaining type of fault data belongs to the range interval corresponding to a preset standard comprehensive fault evaluation coefficient threshold, it indicates that there is a fault condition in the power equipment corresponding to this remaining type of fault data, and it indicates that the fault level of the power equipment corresponding to this remaining type of fault data is the fault level corresponding to the range interval corresponding to this standard comprehensive fault evaluation coefficient threshold. Otherwise, it indicates that there is no fault condition in the power equipment corresponding to this remaining type of fault data.

9. The intelligent fault judgment method of an electric power Internet of Things according to claim 8, characterized in that, The calculation of the characteristic fault location index corresponding to various types of fault data with fault conditions is as follows: Record the fault-like data of a certain power equipment corresponding to a certain data acquisition time point with a fault situation as the feature vector E, E = [e1, e2,......, e g , where g represents each data in the fault-like data of a certain power equipment corresponding to a certain data acquisition time point, g is a positive integer. Input E into the fault data processing model to obtain the corresponding fault probability value F. Set the feature perturbation value Δe h , then a new feature vector E' is obtained, E' = [e1, e2,...,e h + Δe h ,..., e g . Input E' into the fault data processing model to obtain the corresponding fault probability value F'. Furthermore, through the calculation formula: Obtain the characteristic fault location index E h corresponding to the hth data in the fault detection data of a certain type of power equipment with a fault situation. Calculate the characteristic fault location indexes corresponding to various fault data with a fault situation in this way.

10. The intelligent fault judgment method of an electric power Internet of Things according to claim 9, characterized in that, The process of evaluating the specific fault data in the fault data corresponding to each power equipment with fault conditions is as follows: The characteristic fault location index E corresponding to the h-th data in the fault detection data of a certain type of power equipment with a fault situation h Sort them in descending order. The probability of a fault corresponding to the data with the characteristic fault location index ranked first is the first, and the probability of a fault corresponding to the data with the characteristic fault location index ranked second is the second. Thus, the fault probability of the specific fault data in the fault data corresponding to each power equipment is obtained, and the specific fault data in the fault data corresponding to each power equipment with a fault situation is obtained accordingly.

Images