Abnormal node electricity price intelligent identification method and system based on multi-dimensional backtracking analysis

Abstract

The invention discloses an intelligent identification method and system for analyzing abnormal node electricity prices based on multi-dimensional backtracking. The method comprises the steps of collecting node real-time electricity prices, node synchronous real-time historical electricity prices and multi-dimensional data indexes influencing the node electricity prices in real time; the node electricity price and the multi-dimensional data are preprocessed, and a multi-dimensional real-time electricity price feature set is constructed. According to historical electricity price data, an electricity price prediction model of a BP nervous system introducing an adaptive fuzzy neural network is established, a multi-dimensional reference electricity price feature set is constructed, and an electricity price normal range is delimited. And comparing the multi-dimensional real-time electricity price feature set with the multi-dimensional reference electricity price feature set in real time, performing multi-dimensional backtracking analysis on the abnormal node electricity price, and analyzing the reason of the electricity price abnormality. And according to the backtracking analysis result, classification is carried out according to abnormal reasons, and an abnormal alarm is sent to technicians. According to the method, the abnormal electricity price and the reason thereof are accurately positioned, the problem response speed and the processing efficiency are improved, and powerful technical support is provided for stable operation of the electricity market.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis. Background Art

[0002] The accuracy and rationality of electricity price execution are crucial to the stable operation of the power market. However, due to the complexity of electricity price execution rules, the large number of customers and the diverse electricity consumption, the traditional electricity price execution monitoring method mainly relies on manual experience and policy analysis, and lacks effective technical means to accurately identify and locate abnormal situations in electricity price execution. This not only leads to a large workload and low efficiency in abnormal management, but also high costs, and it is difficult to meet the requirements of lean management of current power companies. In this context, this patented technology provides a more comprehensive analysis of electricity price fluctuations by real-time collection and analysis of electricity prices and related multidimensional data, such as supply and demand relationships, weather, policies and other factors. Advanced data preprocessing techniques, including time series association analysis and density-based clustering algorithms, are used to identify and clean abnormal data and improve data quality. Combined with adaptive fuzzy neural networks and BP neural networks, an electricity price prediction model is constructed to improve the accuracy and adaptability of predictions. Real-time monitoring of electricity prices is achieved, and an alarm is automatically issued when an abnormality is detected to help market participants respond in a timely manner. Through multidimensional backtracking analysis technology, the causes of abnormal electricity prices are deeply explored to provide decision support for market supervision and risk management. Summary of the invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by the present invention is that due to the complexity of the electricity price execution rules, the large number of customers and the diverse electricity usage, the traditional electricity price execution monitoring method mainly relies on manual experience and policy analysis, and lacks effective technical means to accurately identify and locate abnormal situations in electricity price execution. This not only leads to a large workload and low efficiency in abnormal management, but also high costs, and it is difficult to meet the current requirements of lean management of power companies.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis, comprising: real-time collection of node real-time electricity prices, node real-time historical electricity prices during the same period, and multi-dimensional data indicators that affect node electricity prices.

[0006] The node electricity prices and multi-dimensional data are preprocessed to construct a multi-dimensional real-time electricity price feature set.

[0007] According to the historical electricity price data, an electricity price prediction model which introduces the BP neural system with adaptive fuzzy neural network is established, a multi-dimensional benchmark electricity price feature set is constructed, and the normal range of electricity price is defined.

[0008] Compare the multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set in real time, conduct multi-dimensional back-tracing analysis on abnormal node electricity prices, and analyze the causes of abnormal electricity prices.

[0009] Based on the results of the retrospective analysis, the abnormalities are classified according to their causes and abnormal alarms are issued to technical personnel.

[0010] As a preferred solution of the method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis described in the present invention, the multi-dimensional data indicators affecting node electricity prices include six indicators classified based on influencing factors, including supply and demand relationship, power generation cost, transmission loss, policy factors, market mechanism and weather conditions.

[0011] As a preferred solution of the method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis described in the present invention, wherein: the node electricity prices and multi-dimensional data are pre-processed to construct a multi-dimensional real-time electricity price feature set, including calculating the time series correlation of data features, identifying and cleaning abnormal data based on a density clustering algorithm, and then constructing a multi-dimensional real-time electricity price feature set.

[0012] The node electricity prices and multi-dimensional data collected by the acquisition module are integrated into a data set, and a data space is established based on the data set, and the data are arranged in the space.

[0013] Calculate the density of data feature points, set the threshold according to the actual denoising requirements, and divide the data in the data space into two groups, expressed as:

[0014]

[0015] Among them, θ represents the density of data feature points, i represents data, k represents the data dimension in space, and a represents the distribution size of data in space.

[0016] According to the two divided groups, the data division boundary is adjusted to ensure that the data within one boundary fully meets the noise requirements. The denoised data is normalized and expressed as:

[0017]

[0018] Among them, z j represents the normalized data in standard format, j represents the denoised data, and n j Represents the characteristic quantity of the denoised data, represents the mean of the data feature quantity, s represents the standard deviation of the data feature points, and η represents the density coefficient.

[0019] Calculate the temporal correlation of data features, map the data to complex space, extract data features, build a network module for data feature extraction, and establish a scattered feature matrix.

[0020] The time series module is used to perform global time series correlation on the original data features and calculate the ordered power function, which is expressed as:

[0021]

[0022] Where f represents the ordered power function, G j Indicates the order of data j, G max and G min They represent the maximum and minimum order in the data space respectively.

[0023] Based on the time series of the data, the temporal correlation of the data features is calculated, which is expressed as:

[0024]

[0025] Among them, δ represents the temporal correlation of data features, L represents the length of data features, Represents the time series mean.

[0026] As a preferred solution of the method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis described in the present invention, the method for identifying and cleaning abnormal data based on the density clustering algorithm includes feeding back the data in the original data space according to the calculated characteristic time series correlation, setting cluster points of the data and its adjacent data, and determining the clustering of the data. The Euclidean distance between the data and its cluster points is calculated, expressed as:

[0027]

[0028] Among them, d represents the Euclidean distance between the cluster point and the data in space, o represents the cluster point of data clustering, and l j Indicates the location of the data in space, l o Represents the location of the cluster points in space.

[0029] Density calculation is performed based on the clustered data, expressed as:

[0030]

[0031] Among them, θ' represents the spatial density of the clustered data, x represents the cluster connection point, and m represents the length of the data sequence.

[0032] According to the density clustering calculation, the density clustering algorithm is iterated until the calculated value of the data density reaches the optimal value and no longer changes. According to the data density calculation result, the boundary value of anomaly identification is set, which is expressed as:

[0033] θ′=μ±3σ

[0034] Among them, μ represents the density mean, and σ represents the density standard deviation. Filter the data that does not belong to the boundary values.

[0035] The identified abnormal data is cleaned and divided into two types of abnormalities: local missing and long-term missing according to the different time series correlation of abnormal data. The time series length of missing data is solved through the matrix, which is expressed as:

[0036]

[0037] Among them, h represents the length of the time series where the data is missing, and c represents the current dimension of the data.

[0038] Establish a minimization missing function, expressed as:

[0039] J=||hY|| 2

[0040] Among them, J represents the missing data filled by the abnormal data, h represents the time series length of the missing data, and Y represents the time series length of the normal data with the same time series correlation as the abnormal data. The missing part of the abnormal data is filled.

[0041] Perform outlier detection on the filled data to determine whether its missingness meets the requirements. The abnormal data that failed to be filled are returned for matrix solution and filling again until all the data are filled. Feedback mapping of the filled data in the original data space is performed and it is imported back into the database.

[0042] As a preferred solution of the method for intelligent identification of abnormal node electricity prices based on multi-dimensional backtracking analysis described in the present invention, the electricity price prediction model of the BP neural system introducing an adaptive fuzzy neural network includes that the BP neural network consists of an input layer, a hidden layer and an output layer, each layer contains multiple nodes, and the nodes are connected by connection weight coefficients.

[0043] The input layer is responsible for receiving nodes from external signals, the hidden layer is responsible for processing the intermediate layer of information, and the output layer is responsible for generating nodes for the final output of the network.

[0044] The signal of the input layer is transmitted to the output layer through the hidden layer. If the output of the output layer is consistent with the expected output, the learning process ends. If not, it enters the back propagation stage.

[0045] Based on the error between the expected output and the actual output, the weights are adjusted through the back-propagation process to reduce the error. The weights are updated using the gradient descent method.

[0046] The connection weights between the input layer and the hidden layer, as well as the connection weights between the hidden layer and the output layer are updated through the process of forward propagation and back propagation.

[0047] The output of all neurons in the hidden layer is expressed as:

[0048]

[0049] y i =f H (net i ),i=1,2,...,m

[0050] Among them, net i represents the activation value of the i-th node. m represents the number of nodes. w ji Represents the connection weight between the input layer and the hidden layer. i represents the output of the hidden layer. H The activation function of the node is the Sigmoid function, which is expressed as:

[0051]

[0052] The output formula of all neurons in the output layer is expressed as:

[0053]

[0054] Among them, f l (l=1, 2, ..., t) is an activation function set as a linear function.

[0055] Adaptive fuzzy neural network is introduced, which consists of five main layers, including input layer, rule layer, normalization layer, output calculation layer and summary layer.

[0056] The input layer is responsible for the fuzzification of the input signal. The output of each node is the membership function value of the input variable, expressed as:

[0057]

[0058] Among them, x represents the input of node i, and y represents the input of node i. i represents a fuzzy set, B i Represents a fuzzy set. Indicates A i , B i The membership function value of .

[0059] The rule layer is responsible for calculating the applicability of each fuzzy rule, which is expressed as:

[0060]

[0061] in, Indicates A i , B i The membership function value of . represents the membership function, Represents the membership function. The output of each node represents the credibility of the rule.

[0062] The normalization layer normalizes the applicability of the rules, expressed as:

[0063]

[0064] The output calculation layer calculates the output of the fuzzy rule for each node, which is expressed as:

[0065]

[0066] The aggregation layer aggregates the outputs of all fuzzy rules to produce the final output, which is expressed as:

[0067]

[0068] in, Represents the output of the third layer. i ,q i ,r i} represents the parameter set of this node.

[0069] As a preferred solution of the method for intelligently identifying abnormal node electricity prices based on multidimensional backtracking analysis described in the present invention, wherein: the demarcation of the normal range of electricity prices includes combining the method based on the mean and standard deviation and the method based on the percentile to evaluate the normal range of electricity prices, and calculating the arithmetic mean and standard deviation according to the multidimensional benchmark electricity price feature set to determine the central trend and fluctuation degree of electricity price data, which is expressed as:

[0070]

[0071] Where, μ represents the arithmetic mean of the ith electricity price data point. σ represents the standard deviation of the ith electricity price data point. i represents the i-th electricity price data point, and n represents the number of data points.

[0072] Based on the assumption that electricity price data follows a normal distribution, the normal range is defined as [μ-kσ,μ+kσ], where k is an empirical value.

[0073] Use the 5th percentile P low and the 95th percentile P high To define the abnormal values ​​outside the normal range. Considering the central trend of the data and the extreme values ​​of the data, the normal range of the final electricity price is defined.

[0074] As a preferred solution of the method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis described in the present invention, wherein: the multi-dimensional backtracking analysis of abnormal node electricity prices and the analysis of the abnormal electricity price causes include comparing and identifying the real-time electricity prices with the normal range of the benchmark electricity prices in real time to determine the abnormal node electricity prices. For the detected anomalies, determine which characteristic values ​​deviate from the normal range.

[0075] The multi-dimensional real-time electricity price feature set of abnormal node electricity prices and the multi-dimensional benchmark electricity price feature set are compared and analyzed using causal reasoning technology. If the node electricity price is abnormal, but other features remain stable, it is considered that the electricity price itself is abnormal. If the node electricity price is abnormal, and the feature follows the abnormality, it is considered that the feature causes the node electricity price abnormality.

[0076] Based on the results of the retrospective analysis, a detailed report is prepared, classified according to the cause of the abnormality, and an abnormality alarm is issued to the technical staff.

[0077] An intelligent identification system for abnormal node electricity prices based on multi-dimensional backtracking analysis, characterized in that: it includes:

[0078] The collection module collects the real-time electricity price of the node, the real-time historical electricity price of the node during the same period, and the multi-dimensional data indicators that affect the node electricity price.

[0079] The preprocessing module preprocesses the node electricity price and multi-dimensional data to construct a multi-dimensional real-time electricity price feature set.

[0080] The calculation module establishes an electricity price prediction model based on historical electricity price data by introducing a BP neural system with an adaptive fuzzy neural network, constructs a multi-dimensional benchmark electricity price feature set, and defines a normal range for electricity prices.

[0081] The analysis module compares the multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set in real time, performs multi-dimensional retrospective analysis on the abnormal node electricity prices, and analyzes the causes of abnormal electricity prices. Based on the retrospective analysis results, the abnormal causes are classified and abnormal alarms are issued to technical personnel.

[0082] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0083] A computer-readable storage medium stores a computer program, which implements the steps of the method described above when executed by a processor.

[0084] Beneficial effects of the present invention: The present invention constructs a multidimensional real-time electricity price feature set by collecting node electricity prices and related multidimensional data indicators in real time and using advanced preprocessing technology. At the same time, an electricity price prediction model is established with the help of a BP neural network improved by an adaptive fuzzy neural network to determine the normal range of electricity prices. By comparing the real-time and benchmark electricity price feature sets in real time, combined with multidimensional backtracking analysis, abnormal electricity prices and their causes can be accurately located. In addition, the system also has the functions of automatic report generation and abnormal alarm, which improves the problem response speed and processing efficiency, and provides strong technical support for the stable operation of the power market. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0086] Figure 1 An overall flow chart of a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis provided in the first embodiment of the present invention.

[0087] Figure 2 A BP neural network structure diagram of a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis provided in the first embodiment of the present invention.

[0088] Figure 3 A flow chart of a BP neural network and adaptive fuzzy neural network combined prediction method based on a multi-dimensional backtracking analysis abnormal node electricity price intelligent identification method and system provided in the first embodiment of the present invention.

[0089] Figure 4 A table of node electricity price influencing factors of a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis is provided in the first embodiment of the present invention.

[0090] Figure 5 A multi-dimensional backtracking analysis result recording table diagram of a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis is provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0091] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0092] Example 1, reference Figure 1 to Figure 5 , is an embodiment of the present invention, and provides a method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis, comprising:

[0093] S1: Real-time collection of node real-time electricity prices, node real-time historical electricity prices during the same period, and multi-dimensional data indicators that affect node electricity prices.

[0094] Six indicators are classified based on influencing factors, including supply and demand, power generation costs, transmission losses, policy factors, market mechanisms and weather conditions.

[0095] The supply and demand relationship includes industrial electricity consumption, residential electricity consumption, new energy power generation, and traditional energy power generation.

[0096] The cost of power generation includes fuel costs (including but not limited to coal prices and natural gas prices) and equipment maintenance costs.

[0097] Transmission losses include transmission loss rates, the state of the grid infrastructure, and especially information related to grid upgrades.

[0098] Policy factors include carbon tax policies and subsidy policies, including but not limited to various fiscal incentives provided by the government.

[0099] The market mechanism includes the bidding results, that is, the quotation results of power generation enterprises in the wholesale market, and the retail electricity price, that is, the price paid by the final consumers.

[0100] Weather conditions include extreme weather events (such as storms, floods, etc.) and temperature fluctuations, including changes in air temperature and their impact on electricity demand.

[0101] S2: Preprocess node electricity prices and multidimensional data to construct a multidimensional real-time electricity price feature set.

[0102] The time series correlation of data features is calculated, and abnormal data is identified and cleaned based on the density clustering algorithm, thereby constructing a multi-dimensional real-time electricity price feature set.

[0103] The node electricity prices and multi-dimensional data collected by the acquisition module are integrated into a data set, and a data space is established based on the data set, and the data are arranged in the space.

[0104] Calculate the density of data feature points, set the threshold according to the actual denoising requirements, and divide the data in the data space into two groups, expressed as:

[0105]

[0106] Among them, θ represents the density of data feature points, i represents data, k represents the data dimension in space, and a represents the distribution size of data in space.

[0107] According to the two divided groups, the data division boundary is adjusted to ensure that the data within one boundary fully meets the noise requirements. The denoised data is normalized and expressed as:

[0108]

[0109] Among them, z j represents the normalized data in standard format, j represents the denoised data, and n j Represents the characteristic quantity of the denoised data, represents the mean of the data feature quantity, s represents the standard deviation of the data feature points, and η represents the density coefficient.

[0110] Calculate the temporal correlation of data features, map the data to complex space, extract data features, build a network module for data feature extraction, and establish a scattered feature matrix.

[0111] The time series module is used to perform global time series correlation on the original data features and calculate the ordered power function, which is expressed as:

[0112]

[0113] Where f represents the ordered power function, G j Indicates the order of data j, G max and G min They represent the maximum and minimum order in the data space respectively.

[0114] Based on the time series of the data, the temporal correlation of the data features is calculated, which is expressed as:

[0115]

[0116] Among them, δ represents the temporal correlation of data features, L represents the length of data features, Represents the time series mean.

[0117] Feedback the data in the original data space according to the calculated characteristic time series correlation, set the cluster points of the data and its adjacent data, and determine the clustering of the data. Calculate the Euclidean distance between the data and its cluster points, expressed as:

[0118]

[0119] Among them, d represents the Euclidean distance between the cluster point and the data in space, o represents the cluster point of data clustering, and l j Indicates the location of the data in space, l o Represents the location of the cluster points in space.

[0120] Density calculation is performed based on the clustered data, expressed as:

[0121]

[0122] Among them, θ' represents the spatial density of the clustered data, x represents the cluster connection point, and m represents the length of the data sequence.

[0123] According to the density clustering calculation, the density clustering algorithm is iterated until the calculated value of the data density reaches the optimal value and no longer changes. According to the data density calculation result, the boundary value of anomaly identification is set, which is expressed as:

[0124] θ′=μ±3σ

[0125] Among them, μ represents the density mean, and σ represents the density standard deviation. Filter the data that does not belong to the boundary values.

[0126] The identified abnormal data is cleaned and divided into two types of abnormalities: local missing and long-term missing according to the different time series correlation of abnormal data. The time series length of missing data is solved through the matrix, which is expressed as:

[0127]

[0128] Among them, h represents the length of the time series where the data is missing, and c represents the current dimension of the data.

[0129] Establish a minimization missing function, expressed as:

[0130] J=||hY|| 2

[0131] Among them, J represents the missing data filled by the abnormal data, h represents the time series length of the missing data, and Y represents the time series length of the normal data with the same time series correlation as the abnormal data. The missing part of the abnormal data is filled.

[0132] Perform outlier detection on the filled data to determine whether its missingness meets the requirements. The abnormal data that failed to be filled are returned for matrix solution and filling again until all the data are filled. Feedback mapping of the filled data in the original data space is performed and it is imported back into the database.

[0133] S3: Based on the historical electricity price data, an electricity price prediction model is established by introducing the BP neural system with an adaptive fuzzy neural network, a multi-dimensional benchmark electricity price feature set is constructed, and the normal range of electricity prices is defined.

[0134] The BP neural network consists of an input layer, a hidden layer, and an output layer. Each layer contains multiple nodes, and the nodes are connected by connection weight coefficients. The network structure is as follows Figure 2 shown.

[0135] The input layer is responsible for receiving nodes from external signals, the hidden layer is responsible for processing the intermediate layer of information, and the output layer is responsible for generating nodes for the final output of the network.

[0136] The signal of the input layer is transmitted to the output layer through the hidden layer. If the output of the output layer is consistent with the expected output, the learning process ends. If not, it enters the back propagation stage.

[0137] Based on the error between the expected output and the actual output, the weights are adjusted through the back-propagation process to reduce the error. The weights are updated using the gradient descent method.

[0138] The connection weights between the input layer and the hidden layer, as well as the connection weights between the hidden layer and the output layer are updated through the process of forward propagation and back propagation.

[0139] The output of all neurons in the hidden layer is expressed as:

[0140]

[0141] y i =f H (net i ),i=1,2,...,m

[0142] Among them, net i represents the activation value of the i-th node. m represents the number of nodes. w ji Represents the connection weight between the input layer and the hidden layer. i represents the output of the hidden layer. H The activation function of the node is the Sigmoid function, which is expressed as:

[0143]

[0144] The output formula of all neurons in the output layer is expressed as:

[0145]

[0146] Among them, f l (l=1, 2, ..., t) is an activation function set as a linear function.

[0147] Adaptive fuzzy neural network is introduced. Adaptive fuzzy neural network consists of five main layers, including input layer, rule layer, normalization layer, output calculation layer and summary layer. The process of BP neural network and adaptive fuzzy neural network combined prediction method is as follows Figure 3 shown.

[0148] The input layer is responsible for the fuzzification of the input signal. The output of each node is the membership function value of the input variable, expressed as:

[0149]

[0150] Among them, x represents the input of node i, and y represents the input of node i. i represents a fuzzy set, B i Represents a fuzzy set. Indicates A i , B i The membership function value of .

[0151] The rule layer is responsible for calculating the applicability of each fuzzy rule, which is expressed as:

[0152]

[0153] in, Indicates A i , B i The membership function value of . represents the membership function, Represents the membership function. The output of each node represents the credibility of the rule.

[0154] The normalization layer normalizes the applicability of the rules, expressed as:

[0155]

[0156] The output calculation layer calculates the output of the fuzzy rule for each node, which is expressed as:

[0157]

[0158] The aggregation layer aggregates the outputs of all fuzzy rules to produce the final output, which is expressed as:

[0159]

[0160] in, Represents the output of the third layer. i ,q i ,r i} represents the parameter set of this node.

[0161] The normal range of electricity prices is evaluated by combining the mean and standard deviation-based method with the percentile-based method. According to the multi-dimensional benchmark electricity price feature set, the arithmetic mean and standard deviation are calculated to determine the central trend and volatility of the electricity price data, which is expressed as:

[0162]

[0163] Where, μ represents the arithmetic mean of the ith electricity price data point. σ represents the standard deviation of the ith electricity price data point. i represents the i-th electricity price data point, and n represents the number of data points.

[0164] Based on the assumption that electricity price data follows a normal distribution, the normal range is defined as [μ-kσ,μ+kσ], where k is an empirical value.

[0165] It should be noted that the value of k is usually 2 or 3.

[0166] Using the 5th percentile (P low ) and the 95th percentile (P high ) to define abnormal values ​​outside the normal range. Considering the central trend of the data and the extreme values ​​of the data, the normal range of the final electricity price is defined.

[0167] S4: Real-time comparison of multi-dimensional real-time electricity price feature set and multi-dimensional benchmark electricity price feature set, multi-dimensional back-analysis of abnormal node electricity prices, and analysis of abnormal electricity price causes. Based on the back-analysis results, the abnormal causes are classified and abnormal alarms are issued to technical personnel.

[0168] The real-time electricity price is compared with the normal range of the benchmark electricity price in real time to identify abnormal node electricity prices. For detected anomalies, determine which characteristic values ​​deviate from the normal range.

[0169] The multi-dimensional real-time electricity price feature set of abnormal node electricity prices and the multi-dimensional benchmark electricity price feature set are compared and analyzed using causal reasoning technology. If the node electricity price is abnormal, but other features remain stable, it is considered that the electricity price itself is abnormal. If the node electricity price is abnormal, and the feature follows the abnormality, it is considered that the feature causes the node electricity price abnormality.

[0170] It should be noted that the abnormal nodes are intelligently analyzed to analyze the possible causes of the abnormality. Based on the collected multi-dimensional data indicators that affect the node electricity price, the factors that may be involved in the indicators are intelligently analyzed, such as Figure 4 .

[0171] Furthermore, based on the results of multidimensional retrospective analysis, language editing is introduced to automatically generate a detailed report, such as Figure 5 The report should include the following:

[0172] Abnormal node records include the name of the abnormal node and the specific value of the abnormal node.

[0173] Abnormal node address record.

[0174] Abnormal time records include recording the specific time point when the abnormality occurs and describing the time period for which the abnormality lasts, so as to assess the impact scope of the abnormality.

[0175] The feature category to which the abnormal node belongs.

[0176] It should be noted that the first-level classification includes: abnormal supply and demand relationship, abnormal power generation cost, abnormal transmission loss, abnormal policy factors, abnormal market mechanism, and abnormal weather conditions.

[0177] The secondary classifications include: electricity demand, electricity supply, fuel cost, operation and maintenance cost, line loss, energy policy, bidding mechanism, extreme weather, and temperature change.

[0178] The three-level classification includes: industrial electricity consumption, residential electricity consumption, new energy power generation, traditional energy power generation, coal prices, natural gas prices, equipment maintenance costs, transmission loss rate, grid upgrade status, carbon tax policy, subsidy policy, bidding results, retail electricity prices, extreme weather events, and temperature fluctuations.

[0179] Cause analysis: Conduct detailed cause analysis on abnormal nodes, including direct causes, indirect causes, and possible triggering events.

[0180] Other anomalies, including detailed descriptions of other anomalies.

[0181] Send the detailed report of the back-analysis to the technicians to issue an abnormality alarm. Update the content of the abnormality report according to the feedback of the technicians and send the updated report again.

[0182] The above embodiments also include an intelligent identification system for abnormal node electricity prices based on multi-dimensional backtracking analysis, specifically:

[0183] The collection module collects the real-time electricity price of the node, the real-time historical electricity price of the node during the same period, and the multi-dimensional data indicators that affect the node electricity price.

[0184] The preprocessing module preprocesses the node electricity price and multi-dimensional data to construct a multi-dimensional real-time electricity price feature set.

[0185] The calculation module establishes an electricity price prediction model based on historical electricity price data by introducing a BP neural system with an adaptive fuzzy neural network, constructs a multi-dimensional benchmark electricity price feature set, and defines a normal range for electricity prices.

[0186] The analysis module compares the multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set in real time, performs multi-dimensional retrospective analysis on the abnormal node electricity prices, and analyzes the causes of abnormal electricity prices. Based on the retrospective analysis results, the abnormal causes are classified and abnormal alarms are issued to technical personnel.

[0187] The computer device may be a server. The computer device includes a processor, a memory, an input / output interface (I / O for short) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data cluster data of a power monitoring system. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method is implemented.

[0188] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0189] Example 2 is an embodiment of the present invention, which provides a method and system for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0190] The experiment first selected several typical power nodes, including industrial nodes, commercial nodes and residential nodes, and collected real-time electricity price data of these nodes, historical electricity price data of the same period, and multi-dimensional data indicators that affect node electricity prices (supply and demand relationship, power generation cost, transmission loss, policy factors, market mechanism, weather conditions). After preprocessing, these data formed a multi-dimensional real-time electricity price feature set.

[0191] In the experiment, we invented a BP neural network electricity price prediction model based on an improved adaptive fuzzy neural network, and combined with a density clustering algorithm to pre-process the node electricity price data, thereby forming a multi-dimensional benchmark electricity price feature set. Real-time node real-time electricity prices and multi-dimensional influencing indicator data of industrial areas, commercial areas and residential areas are collected in real time to ensure the accuracy and timeliness of the data. The density clustering algorithm is used to identify and clean abnormal data and construct a multi-dimensional real-time electricity price feature set. In this step, the temporal correlation of data features is calculated, and the data is denoised by combining multi-dimensional feature extraction technology. The adaptive fuzzy neural network is combined with the BP neural network to establish an electricity price prediction model. By training historical electricity price data, a multi-dimensional benchmark electricity price feature set is generated, and the normal fluctuation range of electricity prices is delineated. The multi-dimensional real-time electricity price feature set is compared with the benchmark electricity price feature set in real time, and the cause of abnormal electricity prices is identified and located through retrospective analysis. The experiment compares and analyzes this result with the existing technology. The existing technology mainly relies on traditional linear regression models and manual experience judgments, lacks the flexibility of multi-dimensional data processing and adaptive models, and is prone to misjudgment and missed judgments.

[0192]

[0193] Through the analysis of the above table data, it is obvious that our invention has the advantage in identifying abnormal node electricity prices. First, from the comparison between the actual electricity price and the historical electricity price, it can be found that our invention can still maintain efficient identification when there are slight fluctuations in electricity prices. For example, at the industrial zone node, our invention accurately identified the electricity price deviation of 0.01 yuan / kWh through the application of adaptive fuzzy neural network, and determined that the node was abnormal, while the existing technology misjudged it as normal due to insufficient sensitivity to fluctuations.

[0194] Secondly, based on multi-dimensional retrospective analysis, our invention can effectively capture the linkage effects of multiple dimensions, such as the impact of weather conditions and sudden changes in supply and demand on electricity prices, and locate the root cause of abnormal electricity prices through precise time series correlation analysis. This multi-dimensional analysis capability is not available in existing linear regression models. Traditional methods can usually only consider electricity price fluctuations in a single dimension and cannot simultaneously integrate multiple influencing factors for judgment, resulting in low recognition accuracy.

[0195] In practical applications, the prediction model invented by us has significantly improved the accuracy and flexibility of electricity price anomaly detection, and effectively reduced the false alarm rate and missed alarm rate. In addition, because our invention introduces advanced data preprocessing and abnormal data cleaning technology, the identification results are more accurate, which can provide reliable technical support for the stable operation of the power market.

[0196] In summary, combined with experimental data, our invention has significant innovation and technical advantages in identifying abnormal node electricity prices, and is significantly superior to traditional methods of existing technologies.

[0197] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis, characterized in that: include: Real-time collection of node real-time electricity prices, node real-time historical electricity prices over the same period, and multi-dimensional data indicators that affect node electricity prices; Preprocess node electricity prices and multi-dimensional data to construct a multi-dimensional real-time electricity price feature set; According to the historical electricity price data, an electricity price prediction model is established by introducing the BP neural system with adaptive fuzzy neural network, a multi-dimensional benchmark electricity price feature set is constructed, and the normal range of electricity price is defined; Compare the multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set in real time, conduct multi-dimensional retrospective analysis on abnormal node electricity prices, and analyze the causes of abnormal electricity prices; Based on the results of the retrospective analysis, the abnormalities are classified according to their causes and abnormal alarms are issued to technical personnel.

2. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 1, characterized in that: The multi-dimensional data indicators that affect node electricity prices include six indicators classified based on influencing factors, including supply and demand relationship, power generation cost, transmission loss, policy factors, market mechanism and weather conditions.

3. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 2 is characterized in that: The preprocessing of the node electricity price and the multidimensional data to construct the multidimensional real-time electricity price feature set includes calculating the time series correlation of the data features, identifying and cleaning the abnormal data based on the density clustering algorithm, and then constructing the multidimensional real-time electricity price feature set; Integrate the node electricity prices and multi-dimensional data collected by the acquisition module into a data set, establish a data space based on the data set, and arrange the data in the space; Calculate the density of data feature points, set the threshold according to the actual denoising requirements, and divide the data in the data space into two groups, expressed as: Among them, θ represents the density of data feature points, i represents data, k represents the data dimension in space, and a represents the distribution size of data in space; According to the two divided groups, the data division boundary is adjusted to ensure that the data within a boundary fully meets the noise requirements; the denoised data is normalized and expressed as: Among them, z j represents the normalized data in standard format, j represents the denoised data, and n j Represents the characteristic quantity of the denoised data, represents the mean of the data feature quantity, s represents the standard deviation of the data feature points, and η represents the density coefficient; Calculate the temporal correlation of data features, map the data to complex space, extract data features, build a network module for data feature extraction, and establish a scattered feature matrix; The time series module is used to perform global time series correlation on the original data features and calculate the ordered power function, which is expressed as: Where f represents the ordered power function, G j Indicates the order of data j, G max and G min Respectively represent the maximum and minimum order in the data space; Based on the time series of the data, the temporal correlation of the data features is calculated, which is expressed as: Among them, δ represents the temporal correlation of data features, L represents the length of data features, Represents the time series mean.

4. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 3 is characterized in that: The method of identifying and cleaning abnormal data based on the density clustering algorithm includes feeding back the data in the original data space according to the calculated characteristic time series correlation degree, setting the cluster points of the data and its adjacent data, and determining the clustering affiliation of the data; calculating the Euclidean distance between the data and its cluster points, expressed as: Among them, d represents the Euclidean distance between the cluster point and the data in space, o represents the cluster point of data clustering, and l j Indicates the location of the data in space, l o Represents the location of cluster points in space; Density calculation is performed based on the clustered data, expressed as: Among them, θ' represents the spatial density of the clustered data, x represents the cluster connection point, and m represents the length of the data sequence; According to the density clustering calculation, the density clustering algorithm is iterated until the calculated value of the data density reaches the optimal value and no longer changes; according to the data density calculation result, the boundary value of anomaly identification is set, which is expressed as: θ′=μ±3σ Among them, μ represents the density mean, σ represents the density standard deviation; filter out the data that does not belong to the boundary values; The identified abnormal data is cleaned and divided into two types of abnormalities: local missing and long-term missing according to the different time series correlation of abnormal data. The time series length of missing data is solved through the matrix, which is expressed as: Among them, h represents the length of the time series where the data is missing, and c represents the current dimension of the data; Establish a minimization missing function, expressed as: J=||h-Y|| 2 Among them, J represents the missing data filled by abnormal data, h represents the time series length of missing data, and Y represents the time series length of normal data with the same time series correlation as abnormal data; the missing part of abnormal data is filled; Perform outlier detection on the filled data to determine whether its missingness meets the requirements; return the abnormal data that failed to be filled to re-solve the matrix and fill it until all the data are filled; feedback map the filled data in the original data space and import it back to the database.

5. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 4 is characterized in that: The electricity price prediction model of the BP neural system introducing the adaptive fuzzy neural network includes: the BP neural network is composed of an input layer, a hidden layer and an output layer, each layer contains a plurality of nodes, and the nodes are connected by connection weight coefficients; The input layer is responsible for receiving nodes from external signals, the hidden layer is responsible for the intermediate layer that processes information, and the output layer is responsible for generating nodes for the final output of the network; The signal of the input layer is transmitted to the output layer through the hidden layer. If the output of the output layer is consistent with the expected output, the learning process ends; if not, it enters the back propagation stage; Based on the error between the expected output and the actual output, the weights are adjusted through the back-propagation process to reduce the error; the weights are updated using the gradient descent method; Update the connection weights between the input layer and the hidden layer, and between the hidden layer and the output layer through the process of forward propagation and back propagation; The output of all neurons in the hidden layer is expressed as: y i =f H (net i ),i=1,2,...,m Among them, net i represents the activation value of the i-th node; m represents the number of nodes; w ji Represents the connection weight between the input layer and the hidden layer; y i represents the output of the hidden layer; f H The activation function of the node is the Sigmoid function, which is expressed as: The output formula of all neurons in the output layer is expressed as: Among them, f l (l=1,2,...,t) is the activation function set as a linear function; Adaptive fuzzy neural network is introduced. Adaptive fuzzy neural network consists of five main layers, including input layer, rule layer, normalization layer, output calculation layer and summary layer; The input layer is responsible for the fuzzification of the input signal. The output of each node is the membership function value of the input variable, expressed as: Among them, x represents the input of node i, y represents the input of node i; A i represents a fuzzy set, B i represents a fuzzy set; Indicates A i , B i The membership function value of ; The rule layer is responsible for calculating the applicability of each fuzzy rule, which is expressed as: in, Indicates A i , B i The membership function value of ; represents the membership function, Represents the membership function; the output of each node represents the credibility of the rule; The normalization layer normalizes the applicability of the rules, expressed as: The output calculation layer calculates the output of the fuzzy rule for each node, which is expressed as: The aggregation layer aggregates the outputs of all fuzzy rules to produce the final output, which is expressed as: in, represents the output of the third layer; {p i ,q i ,r i } represents the parameter set of this node.

6. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 5, characterized in that: Defining the normal range of electricity prices includes evaluating the normal range of electricity prices by combining the method based on the mean and standard deviation and the method based on the percentile, and calculating the arithmetic mean and standard deviation based on the multi-dimensional benchmark electricity price feature set to determine the central trend and fluctuation degree of the electricity price data, which is expressed as: Where, μ represents the arithmetic mean of the ith electricity price data point; σ represents the standard deviation of the ith electricity price data point; x i represents the i-th electricity price data point, and n represents the number of data points; Based on the assumption that electricity price data follows a normal distribution, the normal range is defined as [μ-kσ,μ+kσ], where k is an empirical value; Use the 5th percentile P low and the 95th percentile P high To define outliers outside the normal range; considering the central trend of the data and the extreme values ​​of the data, the normal range of the final electricity price is defined.

7. The method for intelligently identifying abnormal node electricity prices based on multi-dimensional backtracking analysis according to claim 6, characterized in that: The multi-dimensional retrospective analysis of the abnormal node electricity price includes comparing the real-time electricity price with the normal range of the benchmark electricity price to determine the abnormal node electricity price; For detected anomalies, determine which feature values ​​deviate from the normal range; The multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set of abnormal node electricity prices are compared and analyzed using causal reasoning technology. If the node electricity price is abnormal while other features remain stable, it is considered that the electricity price itself is abnormal; if the node electricity price is abnormal and the features follow the abnormality, it is considered that the features cause the node electricity price to be abnormal. Based on the results of the retrospective analysis, a detailed report is prepared, classified according to the cause of the abnormality, and an abnormality alarm is issued to the technical staff.

8. An intelligent identification system for abnormal node electricity prices based on multi-dimensional backtracking analysis using the method according to any one of claims 1 to 7, characterized in that: The collection module collects the real-time electricity price of the node, the real-time historical electricity price of the node during the same period, and the multi-dimensional data indicators that affect the node electricity price; The preprocessing module preprocesses the node electricity price and multi-dimensional data to construct a multi-dimensional real-time electricity price feature set; The calculation module establishes an electricity price prediction model based on historical electricity price data by introducing the BP neural system with adaptive fuzzy neural network, constructs a multi-dimensional benchmark electricity price feature set, and defines the normal range of electricity prices; The analysis module compares the multi-dimensional real-time electricity price feature set and the multi-dimensional benchmark electricity price feature set in real time, performs multi-dimensional back-tracing analysis on the electricity prices of abnormal nodes, and analyzes the causes of abnormal electricity prices. Based on the back-tracing analysis results, the abnormal causes are classified and abnormal alarms are issued to technical personnel.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images