A Fault Early Warning Method for Distribution Networks under Comprehensive Multiple Factors

Abstract

The present invention relates to a method for early warning of distribution network faults under comprehensive multiple factors, including: obtaining multi-source fault information of the preprocessed distribution network; the multi-source fault information includes: human factor information, meteorological factor information, and power equipment factor information; using a time series prediction algorithm to predict the power equipment factor characteristics within the time period [t0, t0 + t']; extracting the power equipment factor characteristics when the power equipment is operating normally and when a fault occurs according to historical power equipment factor information, and using the FCM clustering algorithm to calculate the membership degree of the power equipment factor characteristics within the time period [t0, t0 + t'] belonging to the fault state; calculating the accidental fault trigger factor at the current moment t0 according to the human factor information and the meteorological factor information; calculating the fault trigger value at the moment t0 based on the membership degree of the power equipment factor characteristics within the time period [t0, t0 + t'] belonging to the fault state and the accidental fault trigger factor at the moment t0, and performing fault early warning.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network fault warning, and particularly relates to a distribution network fault warning method under comprehensive multiple factors. Background Technique

[0002] The power system is a leading industry in the national economic system. As an important link in the power system, the distribution network is used to connect the main network and users, and undertakes the important task of delivering electric energy to users. The quality of its operation directly affects the user's power consumption experience; in addition, power outages not only cause huge economic losses to society, but also affect people's lives, disrupt social order, and even endanger national defense security. According to statistics, more than 80% of power outage accidents are caused by distribution network faults. Therefore, it is of great significance to carry out fault warning for the distribution network.

[0003] Most of the existing distribution network fault warning methods carry out fault warning based on the operation information of the distribution network (for example, load, voltage, current, etc.), and insufficiently consider the real-time state of the distribution network, ignoring the influence of external factors such as weather; at the same time, it is found in the later fault tracing that a large number of distribution network faults are caused by human factors, such as the personnel in the distribution network area ignoring the safety distance and illegal theft, etc.; while the existing fault warning methods do not consider the influence of human factors, ignore the real-time state information of the distribution network, and cannot evaluate the possibility of inducing faults during the development of real-time state information, resulting in low accuracy of the existing fault warning models for fault warning. Summary of the Invention

[0004] In order to solve the problems in the background technique, one aspect of the present invention provides a distribution network fault warning method under comprehensive multiple factors, including:

[0005] S1: Obtain the multi-source fault information of the preprocessed distribution network; wherein, the multi-source fault information includes: human factor information, meteorological factor information, and power equipment factor information;

[0006] S2: Extract features from the power equipment factor information to obtain power equipment factor features;

[0007] S3: Use the time series prediction algorithm to predict the power equipment factor features from the current moment t 0 starting, to the power equipment factor features within the time period of t 0 +t' after;

[0008] S4: Extract the power equipment factor features when the power equipment is operating normally and when it fails from the historical power equipment factor information, and use the FCM clustering algorithm to calculate the membership degree of the power equipment factor features within the time period of [t 0 , t 0 +t'] belonging to the failure time;

[0009] S5: Calculate the current time t based on the human factor information and meteorological factor information 0 The accidental failure triggering factors;

[0010] S6: According to [t 0 , t 0 +t′] time period, the power equipment factor characteristics belong to the membership degree at the time of failure and the time t 0 The accidental fault trigger factor is calculated to obtain the time t 0 Fault trigger value; judgment time t 0 Whether the fault trigger value is greater than the set threshold, if so, a fault warning is issued.

[0011] Preferably, the power equipment factor information includes but is not limited to: voltage, current, frequency, and temperature of the power equipment during operation;

[0012] The feature extraction of the power equipment factor information includes:

[0013] S21: Obtain the power equipment factor information at time t, and input the power equipment factor information at time t into the encoder of the deep autoencoder network for encoding to obtain the encoded intermediate features;

[0014] S22: The intermediate features from time tt″ to time t are combined into a feature sequence, and the context information of the feature sequence is integrated using the LSTM network to obtain the power equipment factor features at time t; repeat steps S21-S22 to obtain the current time t 0 , and the current time t 0 Characteristics of power equipment factors at all previous moments.

[0015] Preferably, the prediction is from the current time t 0 Start, then t 0 The characteristics of power equipment factors in the +t′ period include:

[0016] S31: Extraction 0 -T to t 0 Characteristics of power equipment factors within the time period;

[0017] S32: According to t 0 -T to t 0 The characteristics of power equipment factors within a time period are predicted using time series prediction algorithms from the current time t 0 Start, then t 0 Characteristics of power equipment factors during the +t′ time period.

[0018] Preferably, the time series prediction algorithm includes but is not limited to: LSTM algorithm and TCN algorithm.

[0019] Preferably, calculating the membership degree of the power equipment factor characteristics during the time period of [t 0 , t 0 +t′] belonging to the fault by using the FCM clustering algorithm includes:

[0020] S41: Using the FCM clustering algorithm to divide the power equipment factor characteristics during normal operation and fault of the power equipment into two groups of normal operation data and fault operation data, and calculating the clustering center of each group;

[0021] S42: Calculating the membership degree of the power equipment factor characteristics during the time period of [t 0 , t 0 +t′] belonging to the fault according to the clustering centers of the two groups of normal operation data and fault operation data.

[0022] Preferably, the membership degree of the power equipment factor characteristics during the time period of [t 0 , t 0 +t′] belonging to the fault includes:

[0023]

[0024] where u 2 represents the membership degree of the power equipment factor characteristics during the time period of [t 0 , t 0 +t′] belonging to the fault; x represents the power equipment factor characteristics during the time period of [t 0 , t 0 +t′]; c 2 represents the clustering center of the fault operation data group; m represents the fuzzy exponent; c = 2 represents the number of groups; c 1 represents the clustering center of the normal operation data group; ||x - c k || represents the Euclidean distance between x and c k .

[0025] Preferably, the step S5 includes:

[0026] S51: Extracting the personnel information of the distribution network area at time t 0 in the human factor information and calculating the human factor correction coefficient, where the personnel information includes the category of personnel, the body temperature of personnel, and the shortest distance between personnel and distribution network equipment;

[0027] The human factor correction coefficient P includes:

[0028]

[0029] where N represents the total number of personnel in the distribution network area; P iDenote the fault induction factor of the \(i\)-th person within the distribution network area; \(C\) i Denote the category of person \(i\); \(C\) i \( = 0\) indicates that person \(i\) belongs to internal staff; \(C\) i \( = 1\) indicates that person \(i\) belongs to other persons; \(\varepsilon\) denotes the set temperature threshold; \(\varphi\) denotes the set safety distance threshold; Denote the body temperature of person \(i\); \(a\) denotes the weight parameter; Denote the shortest distance of person \(i\) from the distribution network equipment;

[0030] S52: Obtain the weather conditions during the time period \([t\) 0 , \(t\) 0 +\(t'\)] according to the weather forecast, and judge whether extreme weather will occur during the time period \([t\) 0 , \(t\) 0 +\(t'\)]; if so, obtain the number of times of historical extreme weather occurrences based on historical meteorological factor information and the number of times of faults occurring under extreme weather conditions Calculate the weather factor correction coefficient

[0031]

[0032] S53: Calculate the accidental fault trigger factor \(\alpha = P\times W\) according to the human factor correction coefficient and the weather factor correction coefficient.

[0033] Preferably, the fault trigger value at the moment \(t\) 0 includes:

[0034] \(\beta = u\) 2 \(\times\alpha\)

[0035] where \(\beta\) represents the fault trigger value, \(\alpha\) represents the accidental fault trigger factor, and \(u\) 2 represents the membership degree when the power equipment factor characteristics during the time period \([t\) 0 , \(t\) 0 +\(t'\)] belong to the fault.

[0036] Another aspect of the present invention provides a distribution network fault warning device under multiple factors, which is characterized by including a processor and a memory; the memory is used to store a computer program; the processor is connected to the memory and is used to execute the computer program stored in the memory, so that the distribution network fault warning device under multiple factors executes the distribution network fault warning method under multiple factors described above.

[0037] Another aspect of the present invention provides a computer-readable storage medium, storing a program, which is characterized in that when the program is executed by a processor, the distribution network fault warning method under multiple factors described above is implemented.

[0038] The present invention has at least the following beneficial effects

[0039] By acquiring information including human factors, meteorological factors, and factors of the power equipment itself, and calculating the fault trigger values for each time period through an early warning method to conduct early warning for each time period, the present invention comprehensively considers the equipment factor information, extracts the power equipment factor characteristics during normal operation and fault of the power equipment based on historical power equipment factor information, calculates the fault trigger values considering human factor information and meteorological factor information. The early warning method of the present invention integrates multi-source fault information, considers the accidental fault trigger situation under human factors and weather factors, has universality for the fault early warning of the distribution network, and is of great significance for ensuring the safe and stable operation of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a schematic flow chart of the method of the present invention;

[0041] Figure 2 is a training schematic diagram of the time series prediction algorithm of this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following specific examples are used to illustrate the embodiments of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the drawings provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0043] Among them, the drawings are only for illustrative purposes, showing only schematic diagrams, not physical diagrams, and should not be construed as a limitation to the present invention; in order to better illustrate the embodiments of the present invention, some components in the drawings will be omitted, enlarged or reduced, and do not represent the dimensions of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0044] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components; in the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc. indicating the orientation or positional relationship, they are based on the orientation or positional relationship shown in the accompanying drawings. This is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the accompanying drawings are only for illustrative purposes and cannot be construed as a limitation of the present invention. For those of ordinary skill in the art, the specific meanings of the above terms can be understood according to specific circumstances.

[0045] Please refer to Figure 1 , one aspect of the present invention provides a method for warning of distribution network faults under multiple factors, including:

[0046] S1: Obtain multi-source fault information of the preprocessed distribution network; wherein, the multi-source fault information includes: human factor information, meteorological factor information, and power equipment factor information;

[0047] In this embodiment, the multi-source fault information is obtained from the information collection device installed at the distribution network terminal, equipment management information, and other relevant information channels. The multi-source fault information includes human factor information, meteorological factor information, and power equipment factor information. The human factor information includes: the type of personnel in the distribution network area, the number of personnel, the body temperature of personnel, and the distance between personnel and distribution network equipment; wherein, the body temperature of personnel can be monitored by an infrared thermal imaging perception camera for the body temperature of all personnel in the area, and at the same time, the distance between personnel and power grid equipment in the power grid area can be monitored by devices such as cameras; the meteorological factors include: temperature, humidity, wind force, and weather category, and the weather category is, for example, thunderstorm, haze, and tornado, etc.; and the equipment factor information includes: voltage, current, and temperature during equipment operation; the power equipment in the distribution network mainly includes: transformers and overhead lines.

[0048] In this embodiment, it is necessary to preprocess the multi-source fault information. The preprocessing includes: standardizing the multi-source fault information using zero-mean normalization; and normalizing different types of data using linear function normalization.

[0049] S2: Extract features from the power equipment factor information to obtain power equipment factor features;

[0050] The extraction of features from the power equipment factor information includes:

[0051] S21: Obtain the power equipment factor information at time t, and input the power equipment factor information at time t into the encoder of the deep autoencoder network for encoding to obtain the encoded intermediate features;

[0052] S22: Compose the intermediate features from the moment t - t″ to the moment t into a feature sequence, and use the LSTM network to fuse the context information of the feature sequence to obtain the power equipment factor feature at the moment t; repeat steps S21 - S22 to obtain the current moment t 0 and the current moment t 0 and the power equipment factor features at all previous moments

[0053] Preferably, the power equipment factor information includes, but is not limited to: voltage, current, frequency, and temperature during the operation of the power equipment;

[0054] In this embodiment, the deep auto - encoding network consists of an encoder part and a decoder part. The encoding part is used to encode the input features into a low - dimensional space representation, and the decoder is used to reconstruct the low - dimensional space representation into the input features; by using the encoder of the deep auto - encoding network to encode the power equipment factor information into a low - dimensional space to obtain intermediate features, the computational complexity of subsequent calculations can be reduced, and the timeliness of early warning can be improved.

[0055] In this embodiment, the context information of the feature sequence is fused through the LSTM network to obtain the power equipment factor feature at the moment t, that is, the intermediate features from the moment t - t″ to the moment t are composed into a feature sequence, and the context information of the feature sequence is fused through the LSTM network to obtain the power equipment factor feature at the moment t. The power equipment factor feature at the moment t integrates the information in the time period [t - t″, t]. The information such as the voltage and current of the power equipment has a certain correlation in time series. Through the LSTM network, this correlation can be learned to obtain the power equipment factor feature at the moment t, which can improve the accuracy of subsequent early warning; the LSTM network consists of multiple unit modules, and each unit module consists of an input gate, a forget gate, and an output gate; the input of each unit module is the hidden state h output by the previous unit module t-1 and the feature c t-1 , and the output is h t and c t , then the feature c last output by the last unit module is the power equipment factor feature.

[0056] S3: Use the time - series prediction algorithm to predict the power equipment factor features from the current moment t 0 starting from, to the power equipment factor features within the time period of t 0 + t′ after the current moment;

[0057] Preferably, the prediction of the power equipment factor features from the current moment t 0 starting from, to the power equipment factor features within the time period of t 0 + t′ after the current moment includes:

[0058] S31: Extract t0 -T to t 0 Power equipment factor characteristics within the time period;

[0059] S32: According to t 0 -T to t 0 The power equipment factor characteristics within the time period use the time series prediction algorithm to predict from the current moment t 0 starting, to the subsequent t 0 +t' time period of power equipment factor characteristics.

[0060] Preferably, the time series prediction algorithm prediction includes but is not limited to: LSTM algorithm and TCN algorithm. In this embodiment, the time series prediction algorithm can adopt relatively traditional time series modeling methods, such as naive prediction method, simple average method, moving average method, AR, MA, ARMA, etc.; it can also be based on machine learning methods, such as random forest, Xgboost, LightGBM, RNN, LSTM algorithm, and TCN algorithm, etc. In this embodiment, the LSTM algorithm and TCN algorithm are preferably used. As an improved recurrent neural network structure, LSTM overcomes the defect of only being able to remember short-term information, selectively forgets or retains historical information, avoids gradient disappearance, and is beneficial to the modeling of long time series; the TCN algorithm uses a temporal convolutional network to predict, which balances short-term burst memory and long-term memory, and the prediction result is relatively accurate.

[0061] Please refer to Figure 2 , taking the time series prediction algorithm as the LSTM algorithm as an example, as Figure 2 shown. Data_model_i (i = 1, 2, 3, 4) is the historical data collected within Δt 1 as the training data of LSTM, Data_test_i (i = 1, 2, 3, 4) represents the data to be predicted within Δt 2 time. After each model training is completed, both slide backward by Δt 2 to continue the next round of training and prediction. In order to ensure the prediction accuracy, the step size of Δt 2 should be much smaller than Δt 1 .

[0062] S4: Extract the power equipment factor characteristics when the power equipment is operating normally and when it fails according to the historical power equipment factor information, and use the FCM clustering algorithm to calculate the membership degree that the power equipment factor characteristics within [t 0 , t 0 +t'] time period belong to the failure state;

[0063] Preferably, the use of the FCM clustering algorithm to calculate [t 0 , t 0The membership degrees of the power equipment factor characteristics in the time period of [t, t + t′] belonging to the faulty state include:

[0064] S41: Use the FCM clustering algorithm to divide the power equipment factor characteristics during normal operation and faulty operation of the power equipment into two groups, namely normal operation data and faulty operation data, and calculate the clustering center of each group;

[0065] S42: Calculate the membership degrees of the power equipment factor characteristics in the time period of [t, t + t′] belonging to the faulty state according to the clustering centers of the two groups of normal operation data and faulty operation data. 0 t 0 + t′]

[0066] Preferably, the membership degrees of the power equipment factor characteristics in the time period of [t, t + t′] belonging to the faulty state include: 0 t 0 + t′]

[0067]

[0068] where u 2 represents the membership degree of the power equipment factor characteristics in the time period of [t, t + t′] belonging to the faulty state; x represents the power equipment factor characteristics in the time period of [t, t + t′]; c 0 t 0 + t′] 0 t 0 + t′] 2 represents the clustering center of the faulty operation data group; m represents the fuzzy exponent; c = 2 represents the number of groups; c 1 represents the clustering center of the normal operation data group; ||x - c k || represents the Euclidean distance between x and c k .

[0069] When a fault occurs in the distribution network, the fault data at this moment will mostly change compared with the normal operation state. In the time series, the characteristic information at the moment of fault occurrence is different from that during the normal operation of the system, which is called an outlier and can be used as the main basis for judging whether a fault has occurred. The FCM clustering algorithm can use the membership degree to determine the membership degree of each feature quantity belonging to the normal operation state or the faulty state, and construct an objective function based on distance and membership degree to find its optimal value. As the sampled data changes, the membership degree also changes, which can depict the change trend of the distribution network operation state and then determine the fault boundary conditions of the distribution network.

[0070] S5: Calculate the accidental fault trigger factor at the current moment t 0 according to the human factor information and meteorological factor information;

[0071] Preferably, the step S5 includes:

[0072] S51: Extract the information of the moment t in the human factor information 0 of the personnel in the distribution network area and calculate the human factor correction coefficient. The personnel information includes the category of the personnel, the body temperature of the personnel, and the shortest distance between the personnel and the distribution network equipment;

[0073] The human factor correction coefficient P includes:

[0074]

[0075] where N represents the total number of personnel in the distribution network area; P i represents the fault induction factor of the i-th person in the distribution network area; C i represents the category of the person i; C i = 0 indicates that the person i belongs to the internal staff; C i = 1 indicates that the person i belongs to the other personnel; ε represents the set temperature threshold; φ represents the set safety distance threshold; represents the body temperature of the person i; a represents the weight parameter; represents the shortest distance between the person i and the distribution network equipment;

[0076] S52: Obtain the weather conditions during the time period [t 0 , t 0 + t'] according to the weather forecast, and judge whether extreme weather will occur during the time period [t 0 , t 0 + t']; if so, obtain the number of times of historical extreme weather occurrence and the number of times of faults occurring under extreme weather conditions according to the historical meteorological factor information and calculate the weather factor correction coefficient

[0077] S53: Calculate the accidental fault trigger factor α = P × W according to the human factor correction coefficient and the weather factor correction coefficient.

[0078] In this embodiment, by considering human factors to calculate the accidental fault trigger factor to correct the fault trigger value, the universality and accuracy of the distribution network fault warning can be improved. For the staff of the distribution network, by monitoring the body temperature of the distribution network staff and the distance from the power equipment, it is possible to prevent faults caused by the operation of the distribution network by the staff when their bodies are abnormal. For non-internal staff, by monitoring the distance between non-internal personnel and grid equipment, it is possible to prevent faults caused by non-staff ignoring the safety distance. The weather factors mainly consider the impact of extreme weather conditions on the distribution network. Extreme weather includes: temperature greater than 40 degrees Celsius, strong wind (wind force greater than level 5), extreme heavy rainfall (rainfall reaching 50 millimeters), humidity greater than (80% RH), lightning strikes, frost and snowstorms, etc. The above situations are all extreme weather. The specific situation can be set by relevant experts in this field according to the impact degree of different weather on the distribution network through research.

[0079] S6: According to the membership degree of the power equipment factor characteristics in the time period of [t 0 , t 0 + t′] belonging to the fault and the accidental fault trigger factor at time t 0 , calculate the fault trigger value at time t 0 ; Determine whether the fault trigger value at time t 0 is greater than the set threshold. If so, perform a fault warning.

[0080] Preferably, the fault trigger value at time t 0 includes:

[0081] β = u 2 × α

[0082] where β represents the fault trigger value, α represents the accidental fault trigger factor, and u 2 represents the membership degree of the power equipment factor characteristics in the time period of [t 0 , t 0 + t′] belonging to the fault.

[0083] On the other hand, the present invention provides a distribution network fault warning device under comprehensive multiple factors, which is characterized by including a processor and a memory; the memory is used to store a computer program; the processor is connected to the memory and is used to execute the computer program stored in the memory, so that the distribution network fault warning device under comprehensive multiple factors executes the distribution network fault warning method under comprehensive multiple factors described above.

[0084] On yet another aspect, the present invention provides a computer-readable storage medium storing a program, which is characterized in that when the program is executed by a processor, it implements the distribution network fault warning method under comprehensive multiple factors described above.

[0085] In summary, the present invention obtains information including human factors, meteorological factors, and factors of the power equipment itself, calculates the fault trigger values for each time period through the early warning method, and conducts early warning for each time period. The present invention comprehensively considers the equipment factor information, extracts the characteristics of power equipment factors during normal operation and faults of the power equipment according to historical power equipment factor information, calculates the fault trigger values considering human factor information and meteorological factor information. The early warning method of the present invention integrates multi-source fault information, considers the accidental fault trigger situation under human factors and weather factors, has universality for the fault early warning of the distribution network, and is of great significance for ensuring the safe and stable operation of the distribution network.

[0086] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0087] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the present technical solution, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A distribution network fault early warning method based on multiple factors, characterized in that: include: S1: Acquire multi-source fault information of the distribution network after preprocessing; wherein the multi-source fault information includes: human factor information, meteorological factor information and power equipment factor information; S2: extracting the power equipment factor information to obtain the power equipment factor features; The power equipment factor information includes but is not limited to: voltage, current, frequency, and temperature of the power equipment during operation; the power equipment in the distribution network mainly includes: transformers and overhead lines; The feature extraction of the power equipment factor information includes: S21: Obtain the power equipment factor information at time t, and input the power equipment factor information at time t into the encoder of the deep autoencoder network for encoding to obtain the encoded intermediate features; S22: The intermediate features from time tt″ to time t are combined into a feature sequence, and the context information of the feature sequence is fused using the LSTM network to obtain the power equipment factor features at time t; steps S21-S22 are repeated to obtain the power equipment factor features at the current time t0 and all times before the current time t0; S3: predict the power equipment factor characteristics from the current time t0 to the next time period t0+t′ using the time series prediction algorithm based on the historical power equipment factor characteristics; S4: extract the power equipment factor characteristics when the power equipment is operating normally and when it fails based on the historical power equipment factor information, and use the FCM clustering algorithm to calculate the membership of the power equipment factor characteristics in the time period [t0, t0+t′] when it fails; The method of using the FCM clustering algorithm to calculate the degree of membership of the power equipment factor characteristics in the time period [t0, t0+t′] when the power equipment factor characteristics belong to a fault includes: S41: using the FCM clustering algorithm to divide the power equipment factor characteristics when the power equipment is in normal operation and when it is faulty into two groups: normal operation data and faulty operation data, and finding the cluster center of each group; S42: Calculate the degree of membership of the power equipment factor feature in the time period [t0, t0+t′] belonging to the fault according to the cluster centers of the two groups of normal operation data and fault operation data; S5: Calculate the accidental fault triggering factor at the current time t0 according to the human factor information and the meteorological factor information; The step S5 comprises: S51: extracting personnel information of the distribution network area at time t0 from the human factor information and calculating the human factor correction coefficient, wherein the personnel information includes the type of personnel, the body temperature of the personnel, and the shortest distance between the personnel and the distribution network equipment; The human factor correction coefficient P includes: Where N represents the total number of people in the distribution network area; P i represents the fault inducing factor of the ith person in the distribution network area; C i Indicates the category of person i; C i =0 means that person i is an internal staff member; C i =1 means that person i belongs to the rest of the people; ε represents the set temperature threshold; φ represents the set safety distance threshold; represents the body temperature of person i; a represents the weight parameter; It indicates the shortest distance between person i and the distribution network equipment; S52: Obtain the weather conditions in the time period [t0, t0+t′] according to the weather forecast, and determine whether extreme weather will occur in the time period [t0, t0+t′]; if so, obtain the number of extreme weather events in history based on historical meteorological factor information and the number of failures that occurred during extreme weather conditions Calculate the weather correction factor S53: Calculate the accidental fault trigger factor α=P×W according to the human factor correction coefficient and the weather factor correction coefficient; S6: Calculate the fault trigger value at time t0 based on the degree of membership of the power equipment factor characteristics in the time period [t0, t0+t′] when the fault occurs and the accidental fault trigger factor at time t0; determine whether the fault trigger value at time t0 is greater than the set threshold, and if so, issue a fault warning.

2. The distribution network fault early warning method based on comprehensive multi-factors according to claim 1 is characterized in that: The prediction starts from the current time t0 and ends in the time period t0+t′, and the power equipment factor characteristics include: S31: extracting the power equipment factor characteristics in the time period from t0-T to t0; S32: Based on the power equipment factor characteristics in the time period from t0-T to t0, the power equipment factor characteristics in the time period from the current time t0 to the next time period t0+t′ are predicted using a time series prediction algorithm.

3. The distribution network fault early warning method under comprehensive multi-factors according to claim 1 is characterized in that: The membership degree of the power equipment factor characteristics in the time period [t0, t0+t′] when it belongs to a fault includes: Among them, u2 represents the degree of membership of the power equipment factor characteristics in the time period [t0, t0+t′] belonging to the fault; x represents the power equipment factor characteristics in the time period [t0, t0+t′]; c2 represents the cluster center of the fault operation data group; m represents the fuzziness index; c=2 represents the number of groups; c1 represents the cluster center of the normal operation data group; ||xc k || represents x and c k The Euclidean distance of .

4. The distribution network fault early warning method based on comprehensive multi-factors according to claim 1 is characterized in that: The fault trigger value at time t0 includes: β＝u2×α Among them, β represents the fault trigger value, α represents the accidental fault trigger factor, and u2 represents the membership degree of the power equipment factor characteristics in the time period [t0, t0+t′] belonging to the fault.

5. A distribution network fault early warning device based on comprehensive multi-factors, characterized in that: It comprises a processor and a memory; the memory is used to store a computer program; the processor is connected to the memory and is used to execute the computer program stored in the memory, so that the distribution network fault early warning device under comprehensive multiple factors executes the distribution network fault early warning method under comprehensive multiple factors described in any one of claims 1 to 4.

6. A computer-readable storage medium storing a program, characterized in that: When the program is executed by the processor, a distribution network fault early warning method based on comprehensive multiple factors as described in any one of claims 1 to 4 is implemented.

Images