Power grid business data intelligent scheduling method based on multi-source data fusion

Abstract

The invention relates to the field of power grid data scheduling, and discloses a power grid business data intelligent scheduling method based on multi-source data fusion, which comprises the following steps: fusing different business data of a power grid for risk assessment and fault probability assessment of the power grid and user loads, and combining a fault probability assessment result to obtain a power grid business data intelligent scheduling result; and carrying out load intelligent scheduling strategy construction and fault correction strategy construction on the user load and the power grid in the high-risk state. The system can accurately and actively identify the operation risk of the power grid, achieves the intelligent load transfer of main and distribution cooperation, can prevent faults in advance, achieves the quick response when the faults occur, effectively reduces the power-off time and range, guarantees the normal power demands of all walks of life and residents, and improves the power utilization efficiency. And meanwhile, unnecessary equipment overload and redundant operation can be avoided, the operation cost of the power grid is reduced, the service life of equipment is prolonged, and the operation economic benefit of the power grid is improved on the whole.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power grid data scheduling, in particular to a power grid business data intelligent scheduling method based on multi-source data fusion. BACKGROUND

[0002] Traditional power grid operation risk identification often relies on limited key indicators and empirical rules, lacking the ability to comprehensively mine and analyze multi-source business data. For some potential risks under complex conditions, such as the volatility impact of distributed power access and the risk generated by the coupling of extreme weather and power grid operation state, it is difficult to achieve accurate and early identification, making the power grid more passive in responding to risks. On the one hand, current power grid business involves multiple data sources, including D5000 system station data, use sampling system distribution area data, and distribution automation system intelligent switch data. The data has significant differences in format, standard, and spatio-temporal characteristics, making it difficult to deeply integrate and form a unified and effective data resource pool for subsequent risk identification and load transfer analysis. On the other hand, the main grid and distribution network are relatively independent in load management, and the information interaction is lagging and incomplete, lacking effective coordination mechanisms and intelligent decision support systems to achieve accurate and intelligent load transfer. In the face of power grid failures or peak load periods, it is difficult to quickly and reasonably allocate loads between the main grid and distribution network, which may lead to local overload or even large-scale power outages. In addition, the power grid operation state is complex and variable, and existing technical solutions are not fast enough in responding to real-time data updates and dynamic changes in power grid structure and load characteristics, making it difficult to adaptively adjust the corresponding risk identification strategies and load transfer schemes, affecting the reliability and stability of power grid operation.

[0003] With the rapid development of social economy, people's demand for the reliability of power supply is increasing. Accurate and proactive identification of power grid operation risks and intelligent load transfer between the main grid and distribution network can prevent faults from occurring in advance, quickly respond when faults occur, effectively reduce power outage time and range, and ensure normal power demand for various industries and residents. On the other hand, large-scale access of new energy to the power grid, such as distributed photovoltaic and wind power, makes the operation characteristics of the power grid more complex. Through deep fusion of multi-source business data and related key technology research, it is helpful to better integrate these new energy resources, improve the power grid's ability to accept different energy forms, and promote the smooth progress of energy transformation. Intelligent load transfer and accurate risk control can also optimize power grid resource allocation, avoid unnecessary equipment overload and redundant operation, reduce power grid operation costs, extend equipment service life, and improve overall power grid operation economic benefits. SUMMARY

[0004] The present application overcomes the shortcomings of the prior art and provides a power grid business data intelligent scheduling method based on multi-source data fusion.

[0005] To achieve the above object, the technical scheme adopted by the present application is:

[0006] The first aspect of the present application provides a power grid business data intelligent scheduling method based on multi-source data fusion, comprising the following steps:

[0007] Real-time collection of different business data in the power grid, and data heterogeneous fusion of the collected power grid multi-source business data, outputting power grid fusion multi-source business data;

[0008] Risk assessment and fault probability assessment of the target power grid and the target user load in combination with the power grid fusion multi-source business data;

[0009] According to the fault probability assessment result, high-risk user load and high-risk power grid are constructed for intelligent load scheduling strategy and fault correction strategy.

[0010] Further, in a preferred embodiment of the present application, the real-time collection of different business data in the power grid and the data heterogeneous fusion of the collected power grid multi-source business data, outputting power grid fusion multi-source business data, are specifically:

[0011] Determine the target power grid and determine the target user load connected to the target power grid in real time, and calibrate it as the target user load;

[0012] During the power supply of the target power grid to the target user load, real-time multi-source business data collection and processing of the target power grid are carried out, wherein the real-time multi-source business data collection and processing includes collecting real-time power grid state data during the power supply of the target power grid, including real-time voltage value, real-time current value and real-time power value, collecting real-time meteorological data around the target user load and real-time load data of the target user load at the same time;

[0013] Wherein, the load data of the target user load is collected by the electric meter connected to the target user load, and the historical load curve of the target user load is generated, and the real-time load data of the target user load is determined based on the historical load curve of the target user load;

[0014] The real-time power grid state data, the real-time meteorological data around the target user load and the real-time load data of the target user load are collectively referred to as power grid multi-source business data;

[0015] Real-time Kalman filtering of the power grid multi-source business data, and data feature extraction of the power grid multi-source business data in real-time Kalman filtering, wherein the time domain feature and the frequency domain feature of the power grid multi-source business data need to be extracted;

[0016] The multi-modal neural network is introduced, time domain features and frequency domain features of power grid multi-source business data are used for feature vector construction, and the constructed feature vectors are fused based on the multi-modal neural network, and power grid fusion multi-source business data is output.

[0017] Further, in a preferred embodiment of the present application, the power grid fusion multi-source business data is combined to perform risk assessment and failure probability assessment on the target power grid and the target user load, specifically:

[0018] The Newton-Raphson method is introduced to perform weighted least squares processing on the power grid fusion multi-source business data, wherein the weighted least squares processing is an objective function based on the Newton-Raphson method, and the power grid fusion multi-source business data is iteratively calculated until the power grid fusion multi-source business data converges to a preset value, and the iterated power grid fusion multi-source business data is output.

[0019] The iterated power grid fusion multi-source business data includes node voltage values, phase angles and line flow values used for power flow calculation of the target power grid.

[0020] The time series convolution network algorithm is introduced to construct a time series convolution network model, and the power grid fusion multi-source business data is imported into the time series convolution network model for convolution processing, wherein the convolution processing includes causal convolution, dilation convolution and residual connection of the power grid fusion multi-source business data.

[0021] After the convolution processing in the time series convolution network model is completed, the convolved power grid fusion multi-source business data is mapped on the fully connected output layer of the time series convolution network model to obtain the load prediction value of the target power grid.

[0022] The maximum load capacity of the target power grid is determined, and it is judged whether the load prediction value of the target power grid is greater than the maximum load capacity, if yes, the target power grid is determined as high risk.

[0023] The load prediction value of the target power grid and the iterated power grid fusion multi-source business data are combined to perform equipment overload risk analysis on the target user load and the target power grid, and failure probability assessment of the target user load and the target power grid is performed.

[0024] Further, in a preferred embodiment of the present application, the load prediction value of the target power grid and the iterated power grid fusion multi-source business data are combined to perform equipment overload risk analysis on the target user load and the target power grid, and failure probability assessment of the target user load and the target power grid is performed, specifically:

[0025] Based on the iterated power grid fusion multi-source business data, power flow calculation is performed on the target power grid to obtain the power flow distribution of the target power grid.

[0026] based on the target power grid flow distribution, in combination with the target power grid load prediction value, the line load rate in the target power grid and the load rate of the target user load are calculated;

[0027] The preset standard load rate, if the line load rate in the target power grid and the load rate of the target user load are not greater than the standard load rate, the target user load and the target power grid are determined as low risk;

[0028] If there is a line load rate in the target power grid or a load rate of the target user load greater than the standard load rate, the corresponding target user load or target power grid is determined as high risk;

[0029] If the target user load and the target power grid are determined as high risk, the high risk user load and the high risk power grid are obtained, the Bayesian network algorithm is introduced, the mutual dependence probability between the power grid fusion multi-source business data is calculated, and based on the mutual dependence probability between the power grid fusion multi-source business data, the fault condition probability table of the high risk user load and the high risk power grid is constructed, and the target fault condition probability table is calibrated;

[0030] Based on the target fault condition probability table, the fault probability of different fault factors of the high risk user load and the high risk power grid is calculated.

[0031] Further, in one preferred embodiment of the present application, according to the fault probability evaluation result, the high risk user load and the high risk power grid are constructed and the fault correction strategy is constructed, specifically:

[0032] The fault probability of different fault factors of the high risk user load and the high risk power grid is analyzed, and the response fault probability is preset;

[0033] If the fault probability of the fault factor of the high risk user load and the high risk power grid is greater than the response fault probability, the corresponding fault factor is calibrated as the target fault factor;

[0034] The high risk user load and the high risk power grid are modeled for load transfer optimization, and a load transfer optimization model is obtained, wherein the load transfer optimization modeling needs to build a target function and build a constraint condition;

[0035] The target function building of the load transfer optimization modeling is to classify the high risk user load in priority first, then allocate weight according to the priority of the high risk user load, and finally balance the load rate of load transfer according to the weight;

[0036] The constraint condition building of the load transfer optimization modeling is to perform power flow balance constraint and voltage safety constraint on the high risk power grid;

[0037] The particle swarm algorithm is introduced, the particle continuous proportion adjustment is carried out on the load transfer optimization model, and the position of the particle is updated in real time through iterative optimization in the particle continuous proportion adjustment process, the maximum iterative optimization number is preset, when the iterative optimization number of the particle continuous proportion adjustment is equal to the maximum iterative optimization number, the particle continuous proportion adjustment is stopped, and the updated load transfer optimization model is output, and is calibrated as a target load transfer optimization model;

[0038] The load transfer optimization model is run, and a load intelligent scheduling strategy is output, wherein the load intelligent scheduling strategy implements dynamic load transfer scheduling between the high-risk user load and the high-risk power grid according to the current high-risk user load and various load rates of the high-risk power grid and the load prediction value, and ensures that the high-risk user load and the high-risk power grid become low-risk.

[0039] In combination with the target fault factor, a fault correction strategy is constructed and output for the high-risk user load and the high-risk power grid.

[0040] Further, in a preferred embodiment of the present application, the combination of the target fault factor is used to construct and output a fault correction strategy for the high-risk user load and the high-risk power grid, specifically:

[0041] A topological connection graph between the high-risk user load and the high-risk power grid is constructed, and is calibrated as a target topological connection graph.

[0042] The position of the load rate anomaly in the high-risk user load is located through the target topological connection graph, and is calibrated as a load fault position, and the position of the load rate anomaly in the high-risk power grid is located through the target topological connection graph, and is calibrated as a power grid fault position.

[0043] During the output of the load intelligent scheduling strategy, the power grid fault position is analyzed, and the influence range of the load fault position is determined based on the target topological connection graph.

[0044] The high-risk position in the high-risk user load and the high-risk power grid is determined, if the influence range of the load fault position contains the high-risk position, the high-risk user load and the high-risk power grid are immediately controlled to trip and power off, and the output of the load intelligent scheduling strategy is stopped.

[0045] During the tripping and power off of the high-risk user load and the high-risk power grid, a historical data network is introduced, and a strategy for correcting the target fault factor is retrieved and output in combination with the target fault factor.

[0046] If the influence range of the load fault position does not contain the high-risk position, a strategy for correcting the target fault factor is output in combination with the historical data network during the output of the load intelligent scheduling strategy.

[0047] The second aspect of the application also provides a power grid business data intelligent scheduling system based on multi-source data fusion, which comprises a memory and a processor, the memory stores a power grid business data intelligent scheduling method, and the power grid business data intelligent scheduling method is executed by the processor to realize the following steps:

[0048] Real-time collection is performed on different business data in the power grid, data heterogeneous fusion is performed on the collected power grid multi-source business data, and power grid fusion multi-source business data is output.

[0049] Risk assessment and fault probability assessment are performed on the target power grid and the target user load in combination with the power grid fusion multi-source business data.

[0050] According to the fault probability assessment result, load intelligent scheduling strategy construction and fault correction strategy construction are performed on the high-risk user load and the high-risk power grid.

[0051] The technical defects in the background art are solved by the application, and the application has the following beneficial effects: different business data of the power grid are fused to perform risk assessment and fault probability assessment on the power grid and the user load, and in combination with the fault probability assessment result, load intelligent scheduling strategy construction and fault correction strategy construction are performed on the user load and the power grid in a high-risk state. The application can accurately and actively identify the power grid operation risk and realize intelligent load transfer of main and distribution coordination, can prevent faults in advance, quickly responds when faults occur, effectively reduces the power-off time and power-off range, ensures the normal power demand of various industries and residents' life, avoids unnecessary equipment overload and redundant operation, reduces the power grid operation cost, prolongs the service life of the equipment, and improves the overall operation economic benefit of the power grid. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only some embodiments of the application, and for those skilled in the art, other drawings of other embodiments can be obtained without creative labor on the basis of these drawings.

[0053] Figure 1 A flowchart of a power grid business data intelligent scheduling method based on multi-source data fusion is shown;

[0054] Figure 2 A method flowchart of load intelligent scheduling strategy construction and fault correction strategy construction on high-risk user load and high-risk power grid is shown;

[0055] Figure 3A program view of a power grid business data intelligent scheduling system based on multi-source data fusion is shown. DETAILED DESCRIPTION

[0056] In order to enable a more clear understanding of the above-mentioned objects, features and advantages of the present application, the present application will be further described below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0057] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, however, the present application can also be implemented in other manners different from those described herein, and therefore, the scope of protection of the present application is not limited by the specific embodiments disclosed below.

[0058] Figure 1 A flowchart of a power grid business data intelligent scheduling method based on multi-source data fusion is shown, including the following steps:

[0059] S102: Real-time collection of different business data in the power grid, and data heterogeneous fusion of the collected power grid multi-source business data, output of power grid fusion multi-source business data;

[0060] S104: Risk assessment and fault probability assessment of the target power grid and the target user load in combination with the power grid fusion multi-source business data;

[0061] S106: Construction of load intelligent scheduling strategy and construction of fault correction strategy for high-risk user load and high-risk power grid according to the fault probability assessment result.

[0062] Further, in a preferred embodiment of the present application, the real-time collection of different business data in the power grid and the data heterogeneous fusion of the collected power grid multi-source business data to output the power grid fusion multi-source business data are specifically:

[0063] Determination of the target power grid and determination of the user load connected to the target power grid in real time, calibration as the target user load;

[0064] Real-time collection and processing of multi-source business data of the target power grid during power supply to the target user load, wherein the real-time collection and processing of multi-source business data includes collection of real-time power grid state data during power supply of the target power grid, including real-time voltage value, real-time current value and real-time power value, and collection of real-time meteorological data around the target user load and real-time load data of the target user load;

[0065] The load data of the target user load is collected by a meter connected with the target user load, a historical load curve of the target user load is generated, and real-time load data of the target user load is determined based on the historical load curve of the target user load.

[0066] The real-time power grid state data, the target user load surrounding real-time meteorological data and the target user load real-time load data are collectively referred to as power grid multi-source business data.

[0067] The power grid multi-source business data is subjected to real-time Kalman filtering, and data feature extraction is performed on the power grid multi-source business data subjected to real-time Kalman filtering, wherein time domain features and frequency domain features of the power grid multi-source business data need to be extracted.

[0068] A multi-modal neural network is introduced to construct feature vectors of the time domain features and the frequency domain features of the power grid multi-source business data, and data heterogeneous fusion is performed on the constructed feature vectors based on the multi-modal neural network, and power grid fused multi-source business data is output.

[0069] It should be noted that in the present application, the target power grid and the target user load are subjected to multi-source business data integration, including developing efficient data cleaning, conversion and alignment algorithms, removing abnormal values in the data, and integrating the data according to the space-time characteristics, to ensure the quality and availability of the fused data, and to realize data uniformity when the power grid business data is intelligently scheduled. Various data in the power grid multi-source business data will affect the strategy when the data is scheduled. The real-time Kalman filtering algorithm is used for data cleaning, conversion and space-time alignment to realize feature extraction of the data. Among them, the extraction of the time domain features and the frequency domain features of the power grid multi-source business data is the premise of heterogeneous fusion of the data.

[0070] Further, in a preferred embodiment of the present application, the power grid fused multi-source business data is combined to perform risk assessment and fault probability assessment on the target power grid and the target user load, specifically:

[0071] The Newton-Raphson method is introduced to perform weighted least squares processing on the power grid fused multi-source business data, wherein the weighted least squares processing is an objective function based on the Newton-Raphson method, and the power grid fused multi-source business data is iteratively calculated until the power grid fused multi-source business data converges to a preset value, and the iterated power grid fused multi-source business data is output.

[0072] The iterated power grid fused multi-source business data includes node voltage values, phase angles and line flow values used for power flow calculation of the target power grid.

[0073] The time sequence convolution network algorithm is introduced, a time sequence convolution network model is constructed, and the power grid fusion multi-source business data is introduced into the time sequence convolution network model for convolution processing, wherein the convolution processing includes causal convolution, dilated convolution and residual connection on the power grid fusion multi-source business data.

[0074] After the convolution processing in the time sequence convolution network model is completed, the convolved power grid fusion multi-source business data is mapped on the fully connected output layer of the time sequence convolution network model to obtain the load prediction value of the target power grid.

[0075] The load capacity maximum value of the target power grid is determined, and it is judged whether the load prediction value of the target power grid is greater than the load capacity maximum value, if yes, the target power grid is determined as high risk.

[0076] The target user load and the target power grid are subjected to device overload risk analysis and fault probability evaluation in combination with the load prediction value of the target power grid and the iterated power grid fusion multi-source business data.

[0077] It should be noted that the Newton-Raphson method is an algorithm for power flow calculation, calculation of different device load rates and judgment of whether overload. The objective function of the Newton-Raphson method is a calculation function in power flow calculation, which is used to solve nonlinear equations to obtain power flow distribution. Before obtaining the power flow distribution, the node voltage value, phase angle and line flow value of the target power grid for power flow calculation need to be determined, which is the condition for power flow calculation. The time sequence convolution network algorithm is a deep learning model suitable for time sequence prediction, which has the characteristics of strong parallel computing ability, better long-range dependence capture and faster training speed compared with LSTM, and is used in the present application to predict the load prediction value of the target power grid. Convolution processing includes causal convolution, dilated convolution and residual connection on the power grid fusion multi-source business data. Causal convolution is to ensure that the output of the model at a certain time only depends on historical data, avoiding future information leakage; dilated convolution is to capture long-term convolution processing through exponential growth of the dilated rate. Residual connection is to connect the convolved data together to prevent overfitting and map in the fully connected output layer. After obtaining the load prediction value of the target power grid, it is judged whether the load prediction value of the target power grid is greater than the load capacity maximum value. If yes, it proves that the load value will increase after a period of time, resulting in an increase in the failure rate of the power grid and the load rate, so it is determined as high risk.

[0078] Further, in a preferred embodiment of the present application, the device overload risk analysis and fault probability evaluation of the target user load and the target power grid in combination with the load prediction value of the target power grid and the iterated power grid fusion multi-source business data are specifically as follows:

[0079] Based on the iterated power grid fusion multi-source business data, power flow calculation is performed on the target power grid to obtain the power flow distribution of the target power grid.

[0080] Based on the power flow distribution of the target power grid, combined with the load prediction value of the target power grid, the line load rate in the target power grid and the load rate of the target user load are calculated.

[0081] A preset standard load rate is set, and if the line load rate in the target power grid and the load rate of the target user load are both not greater than the standard load rate, the target user load and the target power grid are determined as low risk.

[0082] If the line load rate in the target power grid or the load rate of the target user load is greater than the standard load rate, the corresponding target user load or target power grid is determined as high risk.

[0083] If the target user load and the target power grid are determined as high risk, the high-risk user load and the high-risk power grid are obtained, the Bayesian network algorithm is introduced, the mutual dependence probability between the power grid fusion multi-source business data is calculated, and based on the mutual dependence probability between the power grid fusion multi-source business data, the fault condition probability table of the high-risk user load and the high-risk power grid is constructed, and the target fault condition probability table is calibrated.

[0084] Based on the target fault condition probability table, the fault probability of different fault factors of the high-risk user load and the high-risk power grid is calculated.

[0085] It should be noted that after obtaining the power flow distribution of the target power grid, the line load rate in the target power grid and the load rate of the target user load under the current state can be determined, so as to evaluate the fault probability. The Bayesian network can establish the probability dependence relationship of the fault influencing factors table, such as temperature, equipment itself problem, etc., construct the condition probability table, output the probability of different factors leading to the generation of high-risk user load and high-risk power grid, that is, calculate the fault probability of different fault factors of high-risk user load and high-risk power grid. Through the Bayesian network model, potential risks can be found in advance, which provides a scientific basis for subsequent main and distribution collaborative load transfer, and reduces the probability of occurrence of safety accidents such as power failure.

[0086] Figure 2 A method flow chart for constructing a load intelligent scheduling strategy and a fault correction strategy for high-risk user load and high-risk power grid is shown, including the following steps:

[0087] S202: According to the fault probability evaluation result, the load intelligent scheduling strategy and the fault correction strategy for the high-risk user load and the high-risk power grid are constructed;

[0088] S204: Combined with the target fault factor, the fault correction strategy for the high-risk user load and the high-risk power grid is constructed and output.

[0089] Further, in a preferred embodiment of the present application, the load intelligent scheduling strategy construction and the fault correction strategy construction for high-risk user loads and high-risk power grids according to the fault probability evaluation results are as follows:

[0090] The fault probabilities of different fault factors of high-risk user loads and high-risk power grids are analyzed, and a response fault probability is preset;

[0091] If the fault probability of a fault factor of a high-risk user load and a high-risk power grid is greater than the response fault probability, the corresponding fault factor is marked as a target fault factor;

[0092] The load transfer optimization modeling is performed on the high-risk user loads and the high-risk power grids to obtain a load transfer optimization model, wherein the load transfer optimization modeling needs to build a target function and build a constraint condition;

[0093] The target function building of the load transfer optimization modeling is to first classify the high-risk user loads by priority, then assign weights to the high-risk user loads according to the priority, and finally balance the load transfer rate according to the weights;

[0094] The constraint condition building of the load transfer optimization modeling is to perform power flow balance constraint and voltage safety constraint on the high-risk power grids;

[0095] The particle swarm algorithm is introduced to perform particle continuous proportion adjustment on the load transfer optimization model, and the position of the particle is updated in real time through iterative optimization during the particle continuous proportion adjustment. The maximum number of iterative optimizations is preset. When the number of iterative optimizations of the particle continuous proportion adjustment is equal to the maximum number of iterative optimizations, the particle continuous proportion adjustment is stopped, and the updated load transfer optimization model is output as a target load transfer optimization model;

[0096] The load intelligent scheduling strategy is output by running the load transfer optimization model, wherein the load intelligent scheduling strategy performs dynamic load transfer scheduling between high-risk user loads and high-risk power grids according to various load rates and load prediction values of the current high-risk user loads and high-risk power grids, to ensure that the high-risk user loads and the high-risk power grids become low-risk;

[0097] The fault correction strategy construction and output are performed on the high-risk user loads and the high-risk power grids in combination with the target fault factor.

[0098] It needs to be noted that a main coordination load transfer optimization model needs to be constructed, and multiple factors such as power grid safety constraints, device capacity constraints, load characteristics and user power priority are comprehensively considered, and a particle swarm algorithm is used to generate a load intelligent scheduling strategy. Among them, the load transfer optimization model of high-risk user load and high-risk power grid is constructed, and the objective function needs to be determined. The objective function prioritizes the user load of high priority, minimizes the power loss, minimizes the network loss and balances the load rate, and avoids overloading after load transfer. The constraint conditions include power flow balance constraint and voltage safety constraint, and there are two constraint conditions, which can ensure that there is no overloading during the transfer process. The particle swarm algorithm represents the continuous proportional adjustment of the load transfer optimization model, wherein the load is the particle, and the continuous proportional adjustment is the continuous adjustment of the position of different loads, which realizes the purpose of maintaining dynamic balance during load transfer scheduling. When the iteration optimization times of the particle continuous proportional adjustment are equal to the maximum iteration optimization times, the particle continuous proportional adjustment is stopped, and the obtained target load transfer optimization model is used to output the load intelligent scheduling strategy. Since the front of the house needs to adjust the load for dynamic scheduling, it is proved that there is a fault in the target power grid and user load, so the cause of the fault needs to be determined to repair the fault.

[0099] Further, in a preferred embodiment of the present application, the combination of target fault factors is used to construct and output fault correction strategies for high-risk user loads and high-risk power grids, specifically:

[0100] A topological connection graph between high-risk user loads and high-risk power grids is constructed, and is labeled as a target topological connection graph.

[0101] The position of the load rate anomaly in the high-risk user load is located through the target topological connection graph, and is labeled as a load fault position, and the position of the load rate anomaly in the high-risk power grid is located through the target topological connection graph, and is labeled as a power grid fault position.

[0102] During the output of the load intelligent scheduling strategy, the power grid fault position is analyzed, and the influence range of the load fault position is determined based on the target topological connection graph.

[0103] The high-risk position in the high-risk user load and the high-risk power grid is determined, and if the influence range of the load fault position contains the high-risk position, the high-risk user load and the high-risk power grid are immediately controlled to trip and power off, and the output of the load intelligent scheduling strategy is stopped.

[0104] During the tripping and power off of the high-risk user load and the high-risk power grid, historical data network is introduced, and the strategy for correcting the target fault factor is retrieved and output in combination with the target fault factor.

[0105] If the influence range of the load fault position does not contain the high-risk position, during output of the load intelligent scheduling strategy, in combination with the historical data network, a strategy for correcting the target fault factor is output.

[0106] It should be noted that the topological connection diagram between the high-risk user load and the high-risk power grid can know the fault position and the influence range between the current high-risk user load and the high-risk power grid, and the fault position includes but is not limited to mechanical failure, short-circuit fault point, etc. After locating the fault position, it is necessary to analyze whether the influence range of the fault position will affect the operation of the whole system. For example, if the fault position is the main power supply position of the load, the load will appear problems such as fire due to short-circuit fault, etc., then at this time, power supply needs to be cut off immediately and fault maintenance is needed. The purpose of fault maintenance is achieved by searching the fault factor corresponding to the fault point through the historical data network. If the fault position is a position that does not affect the operation of the whole system, the overload in the system can be eliminated by real-time load scheduling transfer, the power demand of high-priority users such as data centers and hospitals is preferentially guaranteed, and during output of the load intelligent scheduling strategy, in combination with the historical data network, a strategy for correcting the target fault factor is output, so as to meet the purpose of improving power utilization efficiency without affecting fault maintenance.

[0107] As shown in Figure 3 The second aspect of the present application also provides a power grid business data intelligent scheduling system based on multi-source data fusion, which comprises a memory 31 and a processor 32, the memory 31 stores a power grid business data intelligent scheduling method, and the power grid business data intelligent scheduling method is executed by the processor 32 to realize the following steps:

[0108] Real-time collection of different business data in the power grid, data heterogeneous fusion of the collected power grid multi-source business data, and output of power grid fusion multi-source business data;

[0109] Risk assessment and fault probability assessment of the target power grid and the target user load in combination with the power grid fusion multi-source business data;

[0110] According to the fault probability assessment result, the high-risk user load and the high-risk power grid are subjected to load intelligent scheduling strategy construction and fault correction strategy construction.

[0111] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for intelligent dispatching of power grid business data based on multi-source data fusion, characterized in that: The following steps are involved: Collect different business data in the power grid in real time, perform heterogeneous data fusion on the collected multi-source power grid business data, and output the integrated multi-source power grid business data; Combined with multi-source business data of the power grid, risk assessment and failure probability assessment of the target power grid and target user load are carried out; Based on the fault probability assessment results, intelligent load dispatching strategies and fault correction strategies are constructed for high-risk user loads and high-risk power grids.

2. The method for intelligent dispatching of power grid service data based on multi-source data fusion according to claim 1 is characterized in that: The real-time collection of different business data in the power grid and the heterogeneous fusion of the collected multi-source power grid business data are performed to output the fused multi-source power grid business data, specifically: Determine the target power grid and the user load connected to the target power grid in real time, and mark it as the target user load; During the period when the target power grid supplies power to the target user load, real-time collection and processing of multi-source business data of the target power grid is performed. The real-time collection and processing of multi-source business data includes collecting real-time grid status data during the period when the target power grid supplies power, including real-time voltage values, real-time current values, and real-time power values, and collecting real-time meteorological data around the target user load and real-time load data of the target user load; The load data of the target user load is collected through an electric meter connected to the target user load to generate a historical load curve of the target user load, and the real-time load data of the target user load is determined based on the historical load curve of the target user load; The real-time power grid status data, the real-time meteorological data around the target user load, and the real-time load data of the target user load are collectively referred to as the power grid multi-source business data; Performing real-time Kalman filtering on the multi-source business data of the power grid, and extracting data features from the multi-source business data of the power grid using the real-time Kalman filtering, wherein time domain features and frequency domain features of the multi-source business data of the power grid need to be extracted; A multimodal neural network is introduced to construct feature vectors of the time domain characteristics and frequency domain characteristics of the multi-source business data of the power grid, and data heterogeneous fusion is performed on the constructed feature vectors based on the multimodal neural network to output the fused multi-source business data of the power grid.

3. The method for intelligent dispatching of power grid service data based on multi-source data fusion according to claim 1 is characterized in that: The risk assessment and failure probability assessment of the target power grid and target user load are performed by combining the multi-source service data of the power grid, specifically: The Newton-Raphson method is introduced to perform weighted least squares processing on the power grid fusion multi-source service data, wherein the weighted least squares processing is based on the objective function of the Newton-Raphson method, and the power grid fusion multi-source service data is iteratively calculated until the power grid fusion multi-source service data converges to a preset value, and the iterated power grid fusion multi-source service data is output; The iterated power grid fusion multi-source service data includes node voltage values, phase angles, and line power flow values ​​used for power flow calculation of the target power grid; A time series convolutional network algorithm is introduced to construct a time series convolutional network model. The multi-source service data of the power grid fusion is imported into the time series convolutional network model for convolution processing. The convolution processing includes causal convolution, dilated convolution and residual connection on the multi-source service data of the power grid fusion. After the convolution processing is completed in the time series convolutional network model, the convolutional power grid fusion multi-source business data is mapped to the fully connected output layer of the time series convolutional network model to obtain the load forecast value of the target power grid; Determine the maximum load capacity of the target power grid, and judge whether the load forecast value of the target power grid is greater than the maximum load capacity. If so, the target power grid is judged to be high risk; Combined with the load forecast value of the target power grid and the iterative power grid fusion multi-source business data, the equipment overload risk analysis of the target user load and the target power grid is carried out, and the failure probability of the target user load and the target power grid is evaluated at the same time.

4. The method for intelligent dispatching of power grid service data based on multi-source data fusion according to claim 3 is characterized in that: The load forecast value of the target power grid and the iterative power grid fusion multi-source business data are combined to perform equipment overload risk analysis on the target user load and the target power grid, and simultaneously perform failure probability assessment on the target user load and the target power grid, specifically: Based on the iterative grid fusion multi-source business data, the target grid is subjected to power flow calculation to obtain the target grid power flow distribution; Based on the power flow distribution of the target power grid and the load forecast value of the target power grid, the line load rate in the target power grid and the load rate of the target user load are calculated; A standard load rate is preset. If the line load rate in the target power grid and the load rate of the target user load are both not greater than the standard load rate, the target user load and the target power grid are judged to be low risk. If the line load rate in the target power grid or the load rate of the target user load is greater than the standard load rate, the target user load or the target power grid will be judged as high risk. If the target user load and target power grid are judged to be high-risk, the high-risk user load and high-risk power grid are obtained, and the Bayesian network algorithm is introduced to calculate the interdependence probability between the multi-source business data of the power grid fusion. Based on the interdependence probability between the multi-source business data of the power grid fusion, a fault condition probability table of the high-risk user load and the high-risk power grid is constructed and calibrated as the target fault condition probability table; Based on the target fault condition probability table, the failure probabilities of different failure factors of high-risk user loads and high-risk power grids are calculated.

5. The method for intelligent dispatching of power grid service data based on multi-source data fusion according to claim 1 is characterized in that: According to the fault probability assessment results, a load intelligent dispatching strategy and a fault correction strategy are constructed for high-risk user loads and high-risk power grids, specifically: Analyze the failure probability of different failure factors of high-risk user loads and high-risk power grids, and preset the response failure probability; If the failure probability of the high-risk user load and high-risk power grid with a fault factor is greater than the response failure probability, the corresponding fault factor is calibrated as the target fault factor; Carry out load transfer optimization modeling for high-risk user loads and high-risk power grids to obtain a load transfer optimization model. The load transfer optimization modeling requires the establishment of objective functions and constraint conditions; The objective function of load transfer optimization modeling is to first prioritize high-risk user loads, then assign weights based on the priorities of high-risk user loads, and finally balance the load transfer rate based on the weights. Among them, the constraints of load transfer optimization modeling are built to constrain power flow balance and voltage safety for high-risk power grids; The particle swarm algorithm is introduced to perform continuous particle ratio adjustment on the load transfer optimization model. During the process of continuous particle ratio adjustment, the particle positions are updated in real time through iterative optimization. The maximum number of iterative optimization times is preset. When the number of iterative optimization times of the continuous particle ratio adjustment equals the maximum number of iterative optimization times, the continuous particle ratio adjustment is stopped and the updated load transfer optimization model is output and calibrated as the target load transfer optimization model. Running the load transfer optimization model and outputting a load intelligent dispatching strategy, wherein the load intelligent dispatching strategy implements dynamic load transfer scheduling between the high-risk user load and the high-risk power grid based on various load rates and load forecast values ​​of the current high-risk user load and the high-risk power grid, to ensure that the high-risk user load and the high-risk power grid become low-risk; Combined with the target fault factor, fault correction strategies are constructed and output for high-risk user loads and high-risk power grids.

6. The method for intelligent dispatching of power grid business data based on multi-source data fusion according to claim 5 is characterized in that: The above-mentioned fault correction strategy is constructed and output for high-risk user loads and high-risk power grids in combination with the target fault factor, specifically: Construct a topological connection diagram between high-risk user loads and high-risk power grids, and mark it as a target topological connection diagram; The target topology connection diagram is used to locate the location of the high-risk user load that causes the abnormal load rate and mark it as the load fault location. At the same time, the target topology connection diagram is used to locate the location of the high-risk power grid that causes the abnormal load rate and mark it as the power grid fault location. During the output of the load intelligent dispatching strategy, the grid fault location is analyzed, and the impact range of the load fault location is determined based on the target topology connection diagram; Identify high-risk user loads and high-risk locations in the high-risk power grid. If the impact range of the load fault location includes the high-risk location, immediately control the high-risk user load and the high-risk power grid to trip and cut off power, and stop the output of the load intelligent scheduling strategy; During periods of high-risk user loads and high-risk grid tripping and outages, a historical data network is introduced, combined with the target fault factor, to retrieve and output a strategy for correcting the target fault factor. If the impact range of the load fault location does not include the high-risk location, during the output of the load intelligent scheduling strategy, the strategy for correcting the target fault factor is output in combination with the historical data network.

7. The intelligent dispatching system for power grid business data based on multi-source data fusion is characterized by: The power grid business data intelligent dispatching system includes a memory and a processor, and the power grid business data intelligent dispatching method program is stored in the memory. When the power grid business data intelligent dispatching method program is executed by the processor, the power grid business data intelligent dispatching method steps as described in any one of claims 1-6 are implemented.

Images