Power distribution network health assessment method and device based on convolutional fuzzy neural network

Abstract

The invention discloses a power distribution network health assessment method and device based on a convolutional fuzzy neural network, and the technical scheme of the invention comprises the steps: extracting implicit features and modes from operation data through introducing the combination of the convolutional neural network and a fuzzy logic system, and employing the feature extraction capability of the convolutional neural network, thereby achieving the health assessment of a power distribution network. Complexity and uncertainty in operation of the power distribution network are better processed by using the fuzzy neural network, training efficiency and training precision are improved, and network performance is enhanced. And in an actual application stage, the trained convolutional fuzzy neural network model and actual power distribution network operation data are utilized to quickly and efficiently evaluate the health of the power distribution network in real time.

Description

Technical Field

[0001] The present application relates to the field of distribution network operation and maintenance technology, and in particular to a distribution network health assessment method and device based on convolutional fuzzy neural network. Background Art

[0002] With the continuous development of the power system, the operating environment of the distribution network has become more dynamic and complex. Real-time and efficient health diagnosis technology can help to promptly discover and deal with potential problems and avoid large-scale power outages caused by faults. It can be seen that the operating status of the distribution network directly affects the user's power experience and the reliability of power supply. In order to ensure the safe, stable and efficient operation of the power system, it is necessary to grasp the health status of the distribution network at any time according to its own characteristics in order to provide stable power services. Therefore, accurately evaluating the operating status of the distribution network is not only a necessary means to ensure the stability and safety of power supply, but also will help identify potential problems and hidden dangers, so as to take preventive measures and reduce the possibility of failures.

[0003] Most of the existing distribution network health diagnosis methods are based on offline data analysis, which requires a lot of manpower and time investment, which not only increases operating costs but also reduces diagnostic efficiency. Summary of the invention

[0004] The present application provides a distribution network health assessment method and device based on a convolutional fuzzy neural network, which are used to solve the technical problem of low diagnostic efficiency of existing distribution network health diagnosis methods.

[0005] In order to solve the above technical problems, the first aspect of the present application provides a distribution network health assessment method based on convolutional fuzzy neural network, comprising:

[0006] According to the preset distribution network evaluation index system information, determine the weight coefficient corresponding to each distribution network evaluation index;

[0007] Obtaining a preset distribution network operation data sample, and calculating distribution network evaluation index data based on the distribution network operation data sample and the distribution network evaluation index system information;

[0008] According to the general convolutional neural network framework and the fuzzy neural network framework, combined with the distribution network evaluation index system information, an initial distribution network health assessment model is constructed, wherein the convolutional neural network module in the distribution network health assessment model is used to extract features from the distribution network operation data, and the fuzzy neural network module is used to perform fuzzy reasoning operations on the extracted features through fuzzy rules;

[0009] The distribution network health assessment model is trained according to the distribution network assessment index data and the corresponding weight coefficients, and when the convergence of the model output or the number of training iterations reaches a preset condition, the current distribution network health assessment model is output;

[0010] The actual operation data of the distribution network is obtained, and the actual operation data is input into the distribution network health assessment model, so as to obtain the distribution network health assessment result through the operation of the distribution network health assessment model.

[0011] Preferably, the determining of the weight coefficient corresponding to each distribution network evaluation index according to the preset distribution network evaluation index system information specifically includes:

[0012] According to the preset distribution network evaluation index system information, the subjective weight coefficient corresponding to each distribution network evaluation index is calculated through the hierarchical analysis method;

[0013] According to the preset distribution network evaluation index system information, the objective weight coefficient corresponding to each distribution network evaluation index is calculated by using the coefficient of variation method;

[0014] According to the subjective weight coefficient and the objective weight coefficient, a comprehensive weight is obtained according to a preset comprehensive weight calculation formula, and the comprehensive weight is used as the weight coefficient corresponding to the distribution network evaluation index.

[0015] Preferably, the comprehensive weight calculation formula is specifically:

[0016]

[0017] In the formula, Indicates The subjective weight of each indicator, Indicates The objective weight of an indicator, Indicates The comprehensive weight of the indicators, n is the number of indicators.

[0018] Preferably, the distribution network evaluation index specifically includes: reliability index, safety index, economic index and environmental protection index.

[0019] Preferably, the reliability index specifically includes: load margin and voltage margin;

[0020] The safety indicators specifically include: load change rate, voltage qualification rate and line load rate;

[0021] The economic indicators specifically include: line loss rate and transformer load rate;

[0022] Environmental protection indicators include: renewable energy penetration rate.

[0023] At the same time, the second aspect of the present application provides a distribution network health assessment device based on a convolutional fuzzy neural network, comprising:

[0024] A weight coefficient calculation unit, used to determine the weight coefficient corresponding to each distribution network evaluation index according to the preset distribution network evaluation index system information;

[0025] An index data calculation unit, used to obtain a preset distribution network operation data sample, and calculate the distribution network evaluation index data based on the distribution network operation data sample and the distribution network evaluation index system information;

[0026] An initial model building unit is used to build an initial distribution network health assessment model based on a general convolutional neural network framework and a fuzzy neural network framework in combination with the distribution network assessment index system information, wherein the convolutional neural network module in the distribution network health assessment model is used to extract features from distribution network operation data, and the fuzzy neural network module is used to perform fuzzy reasoning operations on the extracted features through fuzzy rules;

[0027] A model training unit, used to train the distribution network health assessment model according to the distribution network assessment index data and the corresponding weight coefficients, and output the current distribution network health assessment model when the convergence of the model output or the number of training iterations reaches a preset condition;

[0028] The distribution network health assessment unit is used to obtain actual operation data of the distribution network, and input the actual operation data into the distribution network health assessment model to obtain a distribution network health assessment result through the operation of the distribution network health assessment model.

[0029] Preferably, the weight coefficient calculation unit is specifically used for:

[0030] According to the preset distribution network evaluation index system information, the subjective weight coefficient corresponding to each distribution network evaluation index is calculated through the hierarchical analysis method;

[0031] According to the preset distribution network evaluation index system information, the objective weight coefficient corresponding to each distribution network evaluation index is calculated by using the coefficient of variation method;

[0032] According to the subjective weight coefficient and the objective weight coefficient, a comprehensive weight is obtained according to a preset comprehensive weight calculation formula, and the comprehensive weight is used as the weight coefficient corresponding to the distribution network evaluation index.

[0033] Preferably, the comprehensive weight calculation formula is specifically:

[0034]

[0035] In the formula, Indicates The subjective weight of each indicator, Indicates The objective weight of an indicator, Indicates The comprehensive weight of the indicators, n is the number of indicators.

[0036] Preferably, the distribution network evaluation index specifically includes: reliability index, safety index, economic index and environmental protection index.

[0037] Preferably, the reliability index specifically includes: load margin and voltage margin;

[0038] The safety indicators specifically include: load change rate, voltage qualification rate and line load rate;

[0039] The economic indicators specifically include: line loss rate and transformer load rate;

[0040] Environmental protection indicators include: renewable energy penetration rate.

[0041] It can be seen from the above technical solutions that this application has the following advantages:

[0042] The technical solution of the present application introduces a combination of convolutional neural networks and fuzzy logic systems, utilizes the feature extraction capabilities of convolutional neural networks, extracts implicit features and patterns from operating data, and utilizes fuzzy neural networks to better handle the complexity and uncertainty in distribution network operation, thereby improving training efficiency and training accuracy and enhancing network performance. In the actual application stage, the convolutional fuzzy neural network model obtained through training is used with actual distribution network operation data to achieve rapid real-time and efficient evaluation of the health of the distribution network, thereby solving the technical problem of low diagnostic efficiency of existing distribution network health diagnosis methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0044] Figure 1 A flow chart of an embodiment of a distribution network health assessment method based on a convolutional fuzzy neural network provided in this application.

[0045] Figure 2 A schematic diagram of a flow chart of an embodiment of a distribution network health assessment device based on a convolutional fuzzy neural network provided in the present application. DETAILED DESCRIPTION

[0046] The embodiments of the present application provide a distribution network health assessment method and device based on a convolutional fuzzy neural network, which are used to solve the technical problem of low diagnostic efficiency of existing distribution network health diagnosis methods.

[0047] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] First, a detailed description of an embodiment of a distribution network health assessment method based on a convolutional fuzzy neural network provided by the present application is as follows:

[0049] See also Figure 1 , this embodiment provides a distribution network health assessment method based on convolutional fuzzy neural network, including:

[0050] Step 101: Determine the weight coefficient corresponding to each distribution network evaluation index according to the preset distribution network evaluation index system information;

[0051] It should be noted that the preset distribution network evaluation index system information mentioned in this embodiment is a specific index system for setting the diagnostic index of the health status of the distribution network from multiple dimensions based on the characteristics of the distribution network. It is divided into the following four dimensions: reliability, safety, economy and environmental protection. Through these multi-dimensional index systems, the health status of the distribution network can be comprehensively and accurately diagnosed and evaluated to ensure its efficient operation in all aspects.

[0052] Among them, reliability refers to the ability of the power system to continuously supply electric energy to power users according to the specified quality standards within a specified time. This embodiment selects load margin and voltage margin as evaluation indicators to reflect the reliability of the distribution network. Load margin refers to the ability of the distribution network to maintain stable power supply when facing sudden load increases or equipment failures. Specifically, the load margin is the ratio of the difference between the actual available load capacity and the currently used load capacity in the power system to the available load capacity. A high load margin means that the distribution network has a large margin to deal with emergencies, thereby improving the reliability and stability of power supply; the voltage margin indicates the deviation range of the actual operating voltage relative to the standard voltage. The voltage margin measures the ability of the power system to maintain the voltage within a stable and acceptable range, and maintain the stability of the voltage even when the load changes or faults occur.

[0053] Safety refers to the ability of the distribution network to continue to supply power to the load for a certain period of time even if a fault occurs during operation. This embodiment selects the load change rate, voltage qualification rate and line load rate as key indicators for evaluating the safety of the distribution network. These indicators comprehensively reflect the stability and reliability of the distribution network in the face of various operational challenges. The load change rate refers to the speed at which the power load in the distribution network changes over time. It reflects the rate at which the load of the distribution network increases or decreases within a certain period of time. The load change rate is an important indicator for evaluating the dynamic response capability of the distribution network, because rapidly changing loads will challenge the stability and reliability of the power system; the voltage qualification rate refers to the ratio between the number of nodes with qualified voltage in the distribution network and the total number of nodes in the entire distribution network. It is one of the important parameters for evaluating the health of the distribution network. The higher the voltage qualification rate, the better the health of the distribution network; the line load rate refers to the ratio of the actual load of the power line in the distribution network to its rated capacity. It is an important indicator for evaluating the operating status and load level of the power line. A high load rate indicates that the line is close to its maximum carrying capacity, which may lead to overload and failure risks.

[0054] Economy refers to the ability of the distribution network to maximize cost-effectiveness during operation and maintenance. The purpose of economic evaluation is to reduce the operation and maintenance costs as much as possible, improve resource utilization efficiency, and achieve economical and efficient power supply under the premise of ensuring reliability and safety. This embodiment selects line loss rate and transformer load rate as key indicators for evaluating the economy of the distribution network. The line loss rate refers to the ratio of power lost on the line to the total power supply power. A lower line loss rate means that less electric energy is lost during transmission, which can improve energy utilization efficiency and reduce the operating cost of power transmission; the transformer load rate refers to the ratio of the actual power load carried by the transformer to its rated capacity. It reflects the utilization efficiency and operating status of the transformer. An appropriate load rate helps to extend the service life of the transformer and improve operating efficiency. Too low a load rate means that the capacity of the transformer is not fully utilized, resulting in a waste of resources; too high a load rate may cause overload operation, increase the risk of failure and maintenance costs.

[0055] Environmental protection refers to the degree of environmental friendliness and sustainable development capabilities of the distribution network during operation. This embodiment selects the renewable energy penetration rate as an indicator to measure the environmental protection of the distribution network. The renewable energy penetration rate refers to the proportion of renewable energy (such as wind energy, solar energy, hydropower, etc.) in the total power generation or total power consumption of the distribution network. It is an important indicator to measure the degree of utilization of renewable energy during the operation of the distribution network. Within a certain range, the higher the penetration rate, the better the environmental protection performance of the system, but beyond this range, it may cause instability in the power system.

[0056] Then, based on the above evaluation indicators, the corresponding weight coefficients of these evaluation indicators are calculated respectively through the preset weight calculation method.

[0057] More specifically, the weight coefficient calculation provided in this embodiment includes: subjective weights calculated based on the analytic hierarchy process and objective weights calculated based on the coefficient of variation method. By combining the analytic hierarchy process with the coefficient of variation method, the subjective weights and objective weights of each evaluation indicator are determined respectively, and then the subjective weights and objective weights are weighted to obtain the comprehensive weights of each evaluation indicator.

[0058] The process of determining the subjective weight based on the hierarchical analysis method in this embodiment can refer to the following example:

[0059] The indicator system for distribution network evaluation is determined and constructed as a hierarchical structure model. The model includes three levels: overall level, standard level and indicator level. The overall level represents the overall goal of the invention for health diagnosis of distribution network; the standard level covers the four main dimensions of distribution network evaluation, namely reliability, safety, economy and environmental protection; the indicator level is refined into specific evaluation indicators under each dimension.

[0060] According to the expert opinions, each evaluation index is compared in pairs. According to the Satty scale, experts need to judge the importance of one index relative to another and assign a corresponding scale value. The Satty scale usually uses a scale of 1 to 9, where: 1 means that the two indicators are equally important, 3 means that one indicator is slightly more important than the other, 5 means that one indicator is more important than the other, 7 means that one indicator is very important than the other, 9 means that one indicator is extremely important than the other, and 2, 4, 6, and 8 represent the intermediate values ​​of the above judgments.

[0061] Construct a judgment matrix. First-level judgment matrix: Construct a judgment matrix of the standard layer relative to the overall layer. Assume that the standard layer has criteria, construct a A matrix with the elements Indicates The criterion is relative to The importance of the two criteria. Secondary judgment matrix: Construct the judgment matrix of the indicator layer relative to the standard layer. For each standard, assume that there is indicators, then construct a A matrix with the elements Indicates The indicator is relative to The importance of an indicator.

[0062] Calculate the eigenvectors and eigenvalues. For each judgment matrix, calculate its maximum eigenvalue and its corresponding eigenvector. After normalization, the eigenvector is the relative weight of each matrix. The judgment matrix is ​​checked for consistency. If the judgment matrix has satisfactory consistency, the weight result is confirmed; otherwise, the element value of the judgment matrix needs to be readjusted until the consistency requirement is met.

[0063] Calculate the total weight. Combine the weight of the standard layer and the weight of the indicator layer to calculate the total weight of each evaluation indicator at the indicator layer relative to the overall layer. The specific method is to multiply the weight of a certain evaluation indicator relative to the standard to which it belongs by the weight of the standard relative to the overall layer to obtain the total weight of the evaluation indicator relative to the overall layer. This process ensures the scientificity and accuracy of the evaluation system and provides a basis for subsequent comprehensive evaluation.

[0064] In addition, the process of determining the objective weight based on the coefficient of variation method in this embodiment can refer to the following example:

[0065] By calculating the coefficient of variation of each evaluation index, the objective weight of each index is determined. The coefficient of variation is used to measure the relative dispersion of data, that is, the size of the fluctuation of data relative to its mean. The index with a larger coefficient of variation has a larger difference in value between different evaluation objects and provides more information, so its weight should be larger; conversely, the index with a smaller coefficient of variation has a smaller difference in value between different evaluation objects and provides less information, so its weight should be smaller.

[0066] Positive conversion and dimensionless processing of indicator data. The purpose of positive conversion is to unify the data direction of all indicators so that the value range and meaning of all indicators are consistent, so as to make reasonable comparisons and calculate weights. The purpose of dimensionless processing is to convert indicator data of different dimensions and units to the same scale through standardization and normalization, so as to eliminate the difficulty and irrationality of comparison caused by dimensional differences. In the evaluation and analysis, the above-mentioned apple orchard indicators are usually divided into positive indicators, interval indicators and reverse indicators. These three types of indicators reflect different meanings and evaluation criteria.

[0067] Positive indicators are indicators whose values ​​are larger, the better. Such indicators usually represent positive and expected effects. The larger the positive index value, the better the health of the system. The dimensionless processing formula of the positive index is:

[0068]

[0069] in: is the dimensionless index value, is the original indicator value, Is an indicator The maximum value of Is an indicator The minimum value of .

[0070] An interval index is one whose optimal value is within a certain range. Exceeding or falling below this range will make the index unsatisfactory. The closer the interval index value is to the middle value of the interval, the better the health of the system. The dimensionless processing formula of the interval index is:

[0071]

[0072] in: is the dimensionless index value, is the original indicator value, Is an indicator The midpoint value of It is the maximum value of the difference between the indicator value and the midpoint value in all samples.

[0073] The reverse index is an index whose value is as small as possible. This type of index usually represents a negative effect that needs to be reduced or avoided. The smaller the reverse index value, the better the health of the system. The positive and dimensionless processing formula of the reverse index is:

[0074]

[0075] First, convert the reverse indicator into a positive indicator. is the converted positive indicator value, is the raw inverse indicator value, is the maximum value of the original reverse index value; then, the converted forward index value is dimensionless, where is the dimensionless index value, and They are respectively the minimum and maximum values ​​of the positive indicator values ​​after all indicators are converted.

[0076] Calculate the coefficient of variation. After all indicator data are normalized and dimensionless, the coefficient of variation of each indicator is calculated. The coefficient of variation is the ratio of the standard deviation to the mean value, which is used to measure the degree of dispersion of the data. The larger the coefficient of variation, the higher the discrimination of the indicator, and the greater the weight should be given. The calculation steps are as follows.

[0077] Calculate the average value for each metric:

[0078]

[0079] in, It is The sample in The dimensionless value of the indicator is is the number of samples.

[0080] Calculate the standard deviation for each metric:

[0081]

[0082] Calculate the coefficient of variation for each indicator:

[0083]

[0084] Determine the objective weight of each indicator. According to the coefficient of variation method, the objective weight of each indicator can be determined by the coefficient of variation. The larger the coefficient of variation, the higher the discrimination of the indicator for the comprehensive evaluation, and the greater the weight should be. The objective weight calculation formula of each indicator is as follows:

[0085]

[0086] in, It is The weight of an indicator.

[0087] Next, the comprehensive weighting method combines subjective weights with objective weights to ensure that the weights of the evaluation indicators reflect both the experience of experts and the data characteristics of the indicators themselves. The formula for comprehensive weighting is as follows:

[0088]

[0089] in, Indicates The subjective weight of each indicator, Indicates The objective weight of an indicator, Indicates The comprehensive weight of the indicators, is the number of indicators.

[0090] Step 102: obtaining a preset distribution network operation data sample, and calculating distribution network evaluation index data based on the distribution network operation data sample and the distribution network evaluation index system information;

[0091] It should be noted that the preset distribution network operation data samples obtained may be historical distribution network operation data, and then, the above distribution network operation data samples are used in combination with the distribution network evaluation index system information to calculate various distribution network evaluation index data respectively, which can be used as training samples.

[0092] Step 103: construct an initial distribution network health assessment model based on a general convolutional neural network framework and a fuzzy neural network framework and combined with distribution network assessment index system information;

[0093] It should be noted that this embodiment introduces a combination of convolutional neural networks and fuzzy logic systems to establish a distribution network health assessment model based on a convolutional fuzzy neural network framework. The model automatically extracts feature maps through a convolutional neural network CNN, and then uses a fuzzy neural network FNN that combines a neural network and a fuzzy logic system to perform fuzzy reasoning through fuzzy rules. The model combines the advantages of CNN and FNN, and can not only automatically extract deep-level features, but also improve the robustness and accuracy of classification through fuzzy reasoning.

[0094] More specifically, a convolutional neural network is constructed to extract feature maps of the input data. In the feature extraction stage, multiple convolutional layers and pooling layers are constructed. The convolutional layer uses convolution operations to extract features from the input training data, improving the gradient vanishing and overfitting problems in deep neural networks; the pooling layer is used to reduce the size of the feature map while retaining key information to further extract features.

[0095] Fuzzification processing. The feature graph is fuzzified, and each feature graph is converted into multiple fuzzy sets, which represent the degree of fuzziness of the feature values. The fuzzy sets include three sets of "negative", "zero" and "positive". Each feature graph will generate three fuzzy graphs, which represent the degree of fuzziness of the feature graph on the three fuzzy sets of "negative", "zero" and "positive". Negative represents the range of lower eigenvalues, zero represents the eigenvalue in the medium range, and positive represents the range of higher eigenvalues. These generated fuzzy graphs will be used in the subsequent fuzzy reasoning steps and output fuzzy values. Finally, all fuzzy reasoning outputs will be converted into definite values ​​through the defuzzification process for the final diagnostic decision. In this way, fuzzification processing can not only retain the important information in the feature graph, but also enhance the model's ability to handle uncertainty and ambiguity through the fuzzy logic system, thereby improving the robustness and accuracy of classification.

[0096] Fuzzy reasoning. Fuzzy reasoning is the process of performing calculations based on fuzzy logic in a fuzzy neural network. This process uses fuzzy rules to infer the fuzzy set of the input and finally generates a fuzzy output. Due to the large number of feature maps in the convolutional fuzzy neural network architecture, directly using the fully connected layer will result in too high computational complexity, so semi-connected layers and independent fuzzy reasoning engines are introduced to optimize the calculations. The basic idea of ​​the semi-connected layer is to partially connect the feature map as an input unit instead of connecting the entire feature map. This can reduce the computational complexity while retaining enough feature information. After the semi-connected layer, each feature map has its own independent fuzzy reasoning engine. These engines reason according to the same fuzzy rules. The fuzzy values ​​generated by the fuzzy reasoning engine are aggregated to form the final fuzzy output. These outputs will be used in the subsequent defuzzification step.

[0097] Defuzzification. The defuzzification layer aggregates the results of all fuzzy inference engines. The defuzzification process formula is as follows:

[0098]

[0099] in, Representation from feature map The inference output , Yes Input In the fuzzy set The membership degree in is the size of the feature map. The defuzzification layer aggregates the output of the inference engine to generate the final model output.

[0100] Step 104: training the distribution network health assessment model according to the distribution network assessment index data and the corresponding weight coefficients, and outputting the current distribution network health assessment model when the convergence of the model output or the number of training iterations reaches a preset condition;

[0101] It should be noted that the distribution network health assessment model constructed based on step 103 uses the distribution network assessment index data obtained in steps 101 to 102 and the weight coefficients corresponding to each index to train the distribution network health assessment model. The selected samples cover various index conditions and health assessment conditions and are diverse. The training process includes forward propagation and back propagation. In the forward propagation, the features are transmitted through each layer and finally summarized in the defuzzification layer to generate the model output; in the back propagation, the model parameters are updated by gradient descent by calculating the loss function to minimize the loss function. Through training, the convolutional fuzzy neural network model will be able to effectively learn and predict samples under different index conditions and health assessment conditions, thereby achieving efficient and accurate classification and evaluation. The training process of the model ensures the diversity and comprehensiveness of the samples, and improves the robustness and applicability of the model in practical applications.

[0102] Step 105: Acquire actual operation data of the distribution network, and input the actual operation data into a distribution network health assessment model to obtain a distribution network health assessment result through calculation of the distribution network health assessment model.

[0103] It should be noted that based on the trained distribution network health assessment model, the parameters of the training results are saved and applied to the diagnosis of the actual distribution network health status. After the convolutional fuzzy neural network training is completed, the saved model can be used to comprehensively evaluate the health status of the distribution network, generate the real-time health status of the distribution network, and classify the current health level of the distribution network according to the health assessment data output by the model. According to the characteristics of the distribution network and the existing evaluation criteria, the health status of the distribution network is divided into five levels, and these health levels are quantified with integers from 0 to 4. The larger the index, the healthier the distribution network. According to the health index, the operating status of the distribution network can be divided into five levels: healthy, sub-healthy, attention, abnormal, and severe. When the network is in a healthy state, only its operating conditions need to be monitored regularly and routine diagnostic tests are carried out to ensure that the system maintains the best operating state; when the network is in a sub-healthy state, it needs to be monitored more frequently and some preventive maintenance work is carried out to prevent the state from deteriorating; when the network is in an attention state, we should closely monitor the trend of the system's health index and formulate an emergency plan to take prompt action when necessary; when the network is in an abnormal state, we need to quickly locate the abnormal data source and formulate a maintenance strategy to restore normal operation as soon as possible; when the network is in a severe state, emergency measures must be taken immediately to prevent the system from crashing, and maintenance and repair work should be carried out as soon as possible to ensure the safety and stability of the system. Through this solution, potential problems can be discovered in a timely manner, thereby guiding the inspection and maintenance work of the distribution network and ensuring the stable operation and efficient management of the distribution network.

[0104] The above is a detailed description of an embodiment of a distribution network health assessment method based on a convolutional fuzzy neural network provided by this application. The following is a detailed description of an embodiment of a distribution network health assessment device based on a convolutional fuzzy neural network provided by this application.

[0105] Please refer to Figure 2 , a distribution network health assessment device based on a convolutional fuzzy neural network provided by this embodiment includes:

[0106] A weight coefficient calculation unit 201, configured to determine the weight coefficients corresponding to each distribution network assessment index according to the preset distribution network assessment index system information;

[0107] An index data calculation unit 202, configured to obtain the preset distribution network operation data samples, and calculate the distribution network assessment index data based on the distribution network operation data samples in combination with the distribution network assessment index system information;

[0108] An initial model construction unit 203, configured to construct an initial distribution network health assessment model according to the general convolutional neural network framework and fuzzy neural network framework in combination with the distribution network assessment index system information;

[0109] A model training unit 204, which is used to train a distribution network health assessment model according to the distribution network evaluation index data and the corresponding weight coefficients, and outputs the current distribution network health assessment model when the convergence or the number of training iterations of the model output reaches a preset condition;

[0110] A distribution network health assessment unit 205, which is used to obtain the actual operation data of the distribution network, input the actual operation data into the distribution network health assessment model, and obtain the distribution network health assessment result through the operation of the distribution network health assessment model.

[0111] Furthermore, the weight coefficient calculation unit 201 is specifically used for:

[0112] According to the preset distribution network evaluation index system information, calculate the subjective weight coefficients corresponding to each distribution network evaluation index through the analytic hierarchy process;

[0113] According to the preset distribution network evaluation index system information, calculate the objective weight coefficients corresponding to each distribution network evaluation index through the coefficient of variation method;

[0114] According to the subjective weight coefficients and the objective weight coefficients, obtain the comprehensive weight according to the preset comprehensive weight calculation formula, and use the comprehensive weight as the weight coefficient corresponding to the distribution network evaluation index.

[0115] Furthermore, the specific form of the comprehensive weight calculation formula is:

[0116]

[0117] In the formula, represents the subjective weight of the th index, represents the objective weight of the th index, represents the comprehensive weight of the th index, and n is the number of indexes.

[0118] Furthermore, the distribution network evaluation indexes specifically include: reliability indexes, safety indexes, economic indexes, and environmental protection indexes.

[0119] Furthermore, the reliability indexes specifically include: load margin and voltage margin;

[0120] The safety indexes specifically include: load change rate, voltage qualification rate, and line load rate;

[0121] The economic indexes specifically include: line loss rate and transformer load rate;

[0122] The environmental protection index includes: renewable energy penetration rate.

[0123] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the terminals, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0124] In several embodiments provided in the present application, it should be understood that the disclosed terminals, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0125] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances so that the embodiments of the present application described herein, for example, can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0126] It should be understood that in the present application, "at least one (item)" means one or more, and "multiple" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that there can be three relationships. For example, "A and / or B" can represent: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after. "At least one (one) of the following" or its similar expression means any combination of these items, including any combination of single item (one) or plural items (ones). For example, at least one (one) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0127] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0128] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0130] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distribution network health assessment method based on convolutional fuzzy neural network, characterized in that: include: According to the preset distribution network evaluation index system information, determine the weight coefficient corresponding to each distribution network evaluation index; Obtaining a preset distribution network operation data sample, and calculating distribution network evaluation index data based on the distribution network operation data sample and the distribution network evaluation index system information; According to the general convolutional neural network framework and the fuzzy neural network framework, combined with the distribution network evaluation index system information, an initial distribution network health assessment model is constructed, wherein the convolutional neural network module in the distribution network health assessment model is used to extract features from the distribution network operation data, and the fuzzy neural network module is used to perform fuzzy reasoning operations on the extracted features through fuzzy rules; The distribution network health assessment model is trained according to the distribution network assessment index data and the corresponding weight coefficients, and when the convergence of the model output or the number of training iterations reaches a preset condition, the current distribution network health assessment model is output; The actual operation data of the distribution network is obtained, and the actual operation data is input into the distribution network health assessment model, so as to obtain the distribution network health assessment result through the operation of the distribution network health assessment model.

2. A distribution network health assessment method based on convolutional fuzzy neural network according to claim 1, characterized in that: Determining the weight coefficient corresponding to each distribution network evaluation index according to the preset distribution network evaluation index system information specifically includes: According to the preset distribution network evaluation index system information, the subjective weight coefficient corresponding to each distribution network evaluation index is calculated through the hierarchical analysis method; According to the preset distribution network evaluation index system information, the objective weight coefficient corresponding to each distribution network evaluation index is calculated by using the coefficient of variation method; According to the subjective weight coefficient and the objective weight coefficient, a comprehensive weight is obtained according to a preset comprehensive weight calculation formula, and the comprehensive weight is used as the weight coefficient corresponding to the distribution network evaluation index.

3. A distribution network health assessment method based on convolutional fuzzy neural network according to claim 2, characterized in that: The comprehensive weight calculation formula is specifically: In the formula, Indicates The subjective weight of each indicator, Indicates The objective weight of an indicator, Indicates The comprehensive weight of the indicators, n is the number of indicators.

4. A distribution network health assessment method based on convolutional fuzzy neural network according to claim 1, characterized in that: The distribution network evaluation indicators specifically include: reliability indicators, safety indicators, economic indicators and environmental protection indicators.

5. A distribution network health assessment method based on convolutional fuzzy neural network according to claim 4, characterized in that: Reliability indicators specifically include: load margin and voltage margin; The safety indicators specifically include: load change rate, voltage qualification rate and line load rate; The economic indicators specifically include: line loss rate and transformer load rate; Environmental protection indicators include: renewable energy penetration rate.

6. A distribution network health assessment device based on convolutional fuzzy neural network, characterized in that: include: A weight coefficient calculation unit, used to determine the weight coefficient corresponding to each distribution network evaluation index according to the preset distribution network evaluation index system information; An index data calculation unit, used to obtain a preset distribution network operation data sample, and calculate the distribution network evaluation index data based on the distribution network operation data sample and the distribution network evaluation index system information; An initial model building unit is used to build an initial distribution network health assessment model based on a general convolutional neural network framework and a fuzzy neural network framework in combination with the distribution network assessment index system information, wherein the convolutional neural network module in the distribution network health assessment model is used to extract features from distribution network operation data, and the fuzzy neural network module is used to perform fuzzy reasoning operations on the extracted features through fuzzy rules; A model training unit, used to train the distribution network health assessment model according to the distribution network assessment index data and the corresponding weight coefficients, and output the current distribution network health assessment model when the convergence of the model output or the number of training iterations reaches a preset condition; The distribution network health assessment unit is used to obtain actual operation data of the distribution network, and input the actual operation data into the distribution network health assessment model to obtain a distribution network health assessment result through the operation of the distribution network health assessment model.

7. A distribution network health assessment device based on convolutional fuzzy neural network according to claim 6, characterized in that: The weight coefficient calculation unit is specifically used for: According to the preset distribution network evaluation index system information, the subjective weight coefficient corresponding to each distribution network evaluation index is calculated through the hierarchical analysis method; According to the preset distribution network evaluation index system information, the objective weight coefficient corresponding to each distribution network evaluation index is calculated by using the coefficient of variation method; According to the subjective weight coefficient and the objective weight coefficient, a comprehensive weight is obtained according to a preset comprehensive weight calculation formula, and the comprehensive weight is used as the weight coefficient corresponding to the distribution network evaluation index.

8. A distribution network health assessment device based on convolutional fuzzy neural network according to claim 7, characterized in that: The comprehensive weight calculation formula is specifically: In the formula, Indicates The subjective weight of each indicator, Indicates The objective weight of an indicator, Indicates The comprehensive weight of the indicators, n is the number of indicators.

9. A distribution network health assessment device based on convolutional fuzzy neural network according to claim 6, characterized in that: The distribution network evaluation indicators specifically include: reliability indicators, safety indicators, economic indicators and environmental protection indicators.

10. A distribution network health assessment device based on convolutional fuzzy neural network according to claim 9, characterized in that: Reliability indicators specifically include: load margin and voltage margin; The safety indicators specifically include: load change rate, voltage qualification rate and line load rate; The economic indicators specifically include: line loss rate and transformer load rate; Environmental protection indicators include: renewable energy penetration rate.

Images