A power grid-based distributed load early warning management method and system

Abstract

The application discloses a power grid-based distributed load early warning management method and system, relates to the field of power system automation, and solves the technical problem that the prior art ignores the priority state of each region during early warning, so that when a plurality of regions simultaneously perform data analysis, the region in urgent need of early warning cannot be timely and effectively analyzed, and the early warning efficiency is low; the region analysis priority is generated according to regional power grid data; a regional power grid topology graph is constructed; the regional predicted load is generated according to the region analysis priority and the regional power grid topology graph, and the equipment adjustment parameter is generated according to the regional predicted load; the alarm signal is generated according to the regional predicted load, and the power grid data of each region is analyzed from multiple angles, so that the important region can be preferentially paid attention to during load early warning, and the corresponding equipment parameter is dynamically adjusted according to the load prediction of each region, so that the accuracy of data is improved during high-load early warning. The efficiency and accuracy of early warning are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power system automation, and particularly relates to a distributed load early warning management method and system based on a power grid. BACKGROUND

[0002] With the rapid development of social economy and the continuous improvement of people's living standards, the demand for electricity continues to grow, and the load of the power grid presents new characteristics such as an increasing peak-valley difference, enhanced volatility, and rising risk of local overload. Especially under the background of frequent extreme weather, large-scale access of new energy, and rapid development of new types of loads such as electric vehicles, the traditional power grid operation and management mode is facing severe challenges. How to accurately predict the load trend and timely identify potential operation risks has become a key problem to ensure the safe and stable operation of the power grid.

[0003] The prior art ignores the priority state of each region during early warning when the load of the power grid is early warned, so that when data analysis is simultaneously performed in several regions, the region in urgent need of early warning cannot be effectively analyzed in time, resulting in low early warning efficiency. Therefore, the load early warning of the power grid still needs to be further improved. SUMMARY

[0004] The application aims to at least solve one of the technical problems existing in the prior art. To this end, the application provides a distributed load early warning management method and system based on a power grid, which is used to solve the technical problem that the prior art ignores the priority state of each region during early warning, so that when data analysis is simultaneously performed in several regions, the region in urgent need of early warning cannot be effectively analyzed in time, resulting in low early warning efficiency.

[0005] To achieve the above-mentioned purpose, a first aspect of the application provides a distributed load early warning management method based on a power grid, comprising:

[0006] obtaining region data; the region data comprises a region ID and corresponding region grid data and region environment data of the region ID;

[0007] generating a region analysis priority according to the region grid data; the region analysis priority refers to the priority of analyzing the region data;

[0008] constructing a region grid topology graph;

[0009] generating a region predicted load according to the region analysis priority and the region grid topology graph; the region predicted load refers to a load prediction value of the region in a future time period;

[0010] generating a device adjustment parameter according to the region predicted load; and assigning the device adjustment parameter to a corresponding device parameter; the device adjustment parameter refers to a parameter value of a related device adjusted according to the load prediction value;

[0011] generating an alarm signal according to the regional predicted load.

[0012] The application can focus on the key region in the load warning by comprehensively analyzing the power grid data of each region from multiple dimensions through the above steps, and can dynamically adjust the related equipment parameters according to the load prediction results of each region, thereby further improving the data accuracy of the warning under the high load state and effectively enhancing the efficiency and accuracy of the warning method.

[0013] Further, the generating of the regional analysis priority according to the regional power grid data comprises:

[0014] obtaining regional power grid data; the regional power grid data comprises a plurality of power grid parameters;

[0015] extracting data integrity, data accuracy and data timeliness corresponding to the regional power grid data;

[0016] calculating the regional data quality QSZ by weighted fusion of the data integrity, data accuracy and data timeliness;

[0017] extracting a load rate, a voltage deviation and a fault frequency corresponding to the regional power grid data;

[0018] calculating the regional abnormal score QYP by weighted fusion of the load rate, the voltage deviation and the fault frequency;

[0019] calculating the regional analysis priority by a formula, and the formula satisfies:

[0020] QFY=QSZ α1 ×QYP α2 ; wherein, α1 and α2 represent adjustment index coefficients, and α1 and α2 are both greater than 0.

[0021] Further, the adjustment index coefficients are generated by a coefficient generation model, comprising:

[0022] obtaining the average regional data quality, the regional data quality standard deviation, the regional abnormal score and the regional false alarm rate corresponding to the coefficient generation model and the regional ID;

[0023] integrating the average regional data quality, the regional data quality standard deviation, the regional abnormal score and the regional false alarm rate into regional analysis data;

[0024] inputting the regional analysis data corresponding to the regional ID into the coefficient generation model to obtain a plurality of adjustment index coefficients corresponding to the regional ID;

[0025] The coefficient generation model is constructed by an artificial intelligence model, comprising:

[0026] Obtain a plurality of historical regional analysis data and historical adjustment index coefficients corresponding to the region ID;

[0027] Divide the plurality of historical regional analysis data and historical adjustment index coefficients corresponding to the region ID into training data, validation data and test data; and perform data preprocessing on the training data, validation data and test data to obtain a training set, a validation set and a test set;

[0028] Select an artificial intelligence model as a base model;

[0029] Train the base model through the training set, and adjust the learning rate and other hyperparameters on the validation set to obtain a pre-trained model;

[0030] Finally, obtain a coefficient generation model with the region ID corresponding to the region analysis data as input and the region ID corresponding to a plurality of adjustment index coefficients as output by verifying the pre-trained model on the test set.

[0031] Further, the regional power grid topology is constructed by the following method, comprising:

[0032] Obtain regional power grid data and regional environment data;

[0033] Define a power grid node based on the power grid equipment in the regional power grid data; the power grid node includes a plurality of node attributes;

[0034] Define an edge connection based on the device connection relationship in the regional power grid data; the edge connection includes a plurality of connection attributes;

[0035] Map the regional environment data to the power grid node as a node attribute;

[0036] Substitute the regional power grid data into a plurality of power grid nodes and a plurality of edge connections to obtain a regional power grid topology.

[0037] The present application constructs the regional power grid topology corresponding to each region ID, and maps the regional environment data to the regional power grid topology, so that the information expression ability of the regional power grid topology is stronger, and the subsequent estimation of the regional load prediction corresponding to each region ID with time series data is more accurate, thereby improving the accuracy and efficiency of the load warning.

[0038] Further, the regional predicted load is generated according to the regional analysis priority and the regional power grid topology, comprising:

[0039] Obtain the regional analysis priority, the regional power grid topology and the regional power grid data;

[0040] Sort the region ID in descending order according to the regional analysis priority to obtain an analysis sequence;

[0041] sequentially selecting a region ID in the analysis sequence;

[0042] extracting a plurality of influence parameters of the region ID in the region grid data in the sliding time window; the influence parameters refer to the grid data capable of affecting the future power load prediction;

[0043] integrating the plurality of influence parameters in the sliding time window and the region environmental data corresponding to the sliding time window into a prediction sequence;

[0044] inputting the prediction sequence corresponding to the region ID and the region grid topology map into a load prediction model to obtain the region predicted load corresponding to the region ID; the load prediction model is constructed by an artificial intelligence model.

[0045] The application sorts the region analysis priorities corresponding to each region ID, sequentially selects the region with high region analysis priority to perform region load prediction, so that the load warning can be performed in a timely and effective manner according to the actual situation of each region, and the efficiency of the grid load warning is improved.

[0046] Further, the load prediction model is constructed by an artificial intelligence model, including:

[0047] obtaining a plurality of historical prediction sequences, historical region grid topology maps and historical region predicted loads corresponding to the region ID;

[0048] dividing the plurality of historical prediction sequences, historical region grid topology maps and historical region predicted loads corresponding to the region ID into training data, verification data and test data; and performing data preprocessing on the training data, verification data and test data to obtain a training set, a verification set and a test set;

[0049] selecting an artificial intelligence model as a base model;

[0050] training the base model by the training set, and adjusting the learning rate and other hyperparameters on the verification set to obtain a pre-trained model;

[0051] verifying the pre-trained model on the test set, and finally obtaining a load prediction model with the input of the prediction sequence and the region grid topology map corresponding to the region ID, and the output of the region predicted load corresponding to the region ID.

[0052] Further, the generation of the device adjustment parameter according to the region predicted load includes:

[0053] obtaining the region predicted load;

[0054] when the region predicted load is greater than or equal to the corresponding region load threshold, setting a plurality of device adjustment parameters in the next time period to the maximum value of the device parameters;

[0055] Otherwise, the several device adjustment parameters STC in the next time period are calculated by a calculation formula t+1,i The calculation formula satisfies the following formula:

[0056] Wherein, t represents the number of time period, k represents the sensitivity index, k>0; STC t+1,i represents the several device adjustment parameters in the next time period, i represents the number of adjustment device; STC i,max and STC i,min respectively represent the maximum value and the minimum value of the device adjustment parameters of the i th adjustment device; QJF th and QF min respectively represent the emergency load threshold and the minimum load benchmark; QYF t+1 represents the regional predicted load in the next time period.

[0057] The application analyzes the relationship between the regional predicted load and the device adjustment parameters of each region, dynamically adjusts the device parameter value corresponding to the region in the case of large change of regional predicted load change, can improve the accuracy of power grid data corresponding to the region with high regional predicted load, and improves the accuracy of load warning.

[0058] Further, the alarm signal generated according to the regional predicted load comprises:

[0059] Obtain the regional predicted load corresponding to the several region IDs;

[0060] Determine whether the regional predicted load of the region ID is greater than the regional load threshold corresponding to the region ID;

[0061] Yes, generate an overload alarm signal of the region ID;

[0062] No, determine whether the regional predicted load of the region ID is greater than the D times regional load threshold corresponding to the region ID; wherein, D is a proportional coefficient, D∈(0, 1);

[0063] Yes, generate an overload risk warning signal of the region ID; No, do nothing.

[0064] Further, the regional load threshold is generated by the following method, comprising:

[0065] Obtain the historical same period regional load basic value QFJ corresponding to the region ID in the next time period;

[0066] Obtain the regional predicted load trend QYFQ corresponding to the region ID in the next time period;

[0067] Obtain the regional environment data corresponding to the region ID in the next time period; the regional environment data comprises several environment parameters;

[0068] The regional environmental trend QHQ corresponding to the region ID in the next time period is calculated using a formula; the formula satisfies the following equation:

[0069] Where t represents the time period number, j represents the environmental parameter number, and the total number of environmental parameters is J, HC t+1,j This represents the specific value of the j-th environmental parameter in the next time period; HC base,j HCQ represents the historical average value corresponding to the j-th environmental parameter; j This represents the weight of the j-th environmental parameter;

[0070] Based on the nonlinear relationships between the historical regional load baseline value QFJ, the regional predicted load trend QYFQ, the regional environmental trend QHQ and the regional load threshold, a regional load threshold calculation function QFYJF(QFJ, QYFQ, QHQ) is constructed.

[0071] The regional load threshold calculation function satisfies the following calculation formula:

[0072] QFYJF t+1,n (QFJ t+1,n QYFQ t+1,n QHQ t+1,n )=QFJ t+1,n ×(1+β1×QYFQ t+1，n +β2×QHQ t+1，n ); where n represents the number corresponding to the region ID, β1 and β2 represent the load trend influence coefficient and the environmental trend influence coefficient, respectively, and β1 and β2 ∈ (0, 1);

[0073] Substituting the historical regional load baseline value, regional predicted load trend, and regional environmental trend corresponding to the regional ID into the regional load calculation function yields the regional load threshold corresponding to the regional ID.

[0074] Another aspect of the present invention provides a distributed load early warning management system based on a power grid, comprising: a data acquisition module, a data analysis module, and an early warning module; the data acquisition module and the data analysis module are connected together; the data analysis module and the early warning module are connected together.

[0075] The data acquisition module acquires regional data through a data acquisition device; the regional data includes a regional ID and its corresponding regional power grid data and regional environmental data.

[0076] The data analysis module generates regional analysis priorities according to regional power grid data, constructs a regional power grid topology map, generates regional predicted loads according to the regional analysis priorities and the regional power grid topology map, generates device adjustment parameters according to the regional predicted loads, and assigns the device adjustment parameters to corresponding device parameters, and generates an alarm signal according to the regional predicted loads.

[0077] The early warning module prompts according to the alarm signal.

[0078] Compared with the prior art, the application has the following beneficial effects:

[0079] 1. The application generates regional analysis priorities according to regional power grid data, constructs a regional power grid topology map, generates regional predicted loads according to the regional analysis priorities and the regional power grid topology map, generates device adjustment parameters according to the regional predicted loads, and assigns the device adjustment parameters to corresponding device parameters, and generates an alarm signal according to the regional predicted loads, which analyzes the power grid data of each region from multiple angles, so as to pay attention to important regions during load warning, and dynamically adjusts the corresponding device parameters according to the load prediction of each region, so as to improve the accuracy of data during high load warning. The efficiency and accuracy of the warning are improved.

[0080] 2. The application analyzes the power grid data collected by each region, analyzes the power grid data from multiple angles and quantifies the power grid data, then calculates the regional analysis priorities of each region by calculating the adjustment index coefficient generated by the pre-trained coefficient generation model combined with other related data, which ensures that the actual power grid data in each region can be used to calculate the regional priorities, improves the calculation accuracy and efficiency of the regional analysis priorities, and provides strong data support for subsequent load warning.

[0081] 3. The application dynamically adjusts the regional load threshold corresponding to each region ID by analyzing the historical power grid data and the current power grid data of each region in different time periods, so that the regional load threshold corresponding to each region in different time periods is not consistent for a long time, which reduces the false alarm of load warning and improves the accuracy of power grid load warning. BRIEF DESCRIPTION OF DRAWINGS

[0082] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0083] Figure 1A flow chart of a power grid-based distributed load early warning management method of the present application;

[0084] Figure 2 A principle schematic diagram of a power grid-based distributed load early warning management system of the present application. DETAILED DESCRIPTION

[0085] The technical solutions of the present application will be described clearly and completely below in conjunction with embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0086] Please refer to Figure 1 The first aspect embodiment of the present application provides a power grid-based distributed load early warning management method, comprising:

[0087] Obtaining regional data; the regional data comprises a regional ID and corresponding regional power grid data and regional environment data;

[0088] Generating a regional analysis priority according to the regional power grid data; the regional analysis priority refers to the priority of analyzing the regional data;

[0089] Constructing a regional power grid topology map;

[0090] Generating a regional predicted load according to the regional analysis priority and the regional power grid topology map; the regional predicted load refers to a predicted load value of the region in a future time period;

[0091] Generating a device adjustment parameter according to the regional predicted load; and assigning the device adjustment parameter to a corresponding device parameter; the device adjustment parameter refers to adjusting the parameter value of the related device according to the predicted load value;

[0092] Generating an alarm signal according to the regional predicted load.

[0093] In the embodiment, the regional analysis priority is generated according to the regional power grid data, comprising:

[0094] Obtaining regional power grid data; the regional power grid data comprises a plurality of power grid parameters;

[0095] Extracting data integrity, data accuracy and data timeliness corresponding to the regional power grid data; the data integrity is to measure the missing proportion of the data field, the data accuracy is to measure the noise level of the data, and the data timeliness is to measure the delay time of the data; in the embodiment, the data integrity, the data accuracy and the data timeliness are all controlled within the range of 0-1;

[0096] The regional data quality QSZ is calculated by weighting and integrating data integrity, data accuracy, and data timeliness.

[0097] The formula for calculating regional data quality satisfies:

[0098] QSZ = γ1×SW + γ2×SZ + γ3×SS; where γ1, γ2, and γ3 represent the weight coefficients corresponding to data integrity SW, data accuracy SZ, and data timeliness SS, respectively, γ1, γ2, and γ3 ∈ (0, 1), and the specific values ​​are set according to experience. In this embodiment, γ1, γ2, and γ3 are set to 0.3, 0.4, and 0.3, respectively; the larger the weight coefficient, the greater the influence of the corresponding data on the calculation of regional data quality;

[0099] Extract the load factor, voltage deviation, and fault frequency corresponding to the regional power grid data; the load factor is expressed as the ratio of the current load to the rated capacity of the region; the voltage deviation is a quantification of the voltage stability in the region, and the current voltage deviation is mapped to the range of 0-1 through historical voltage deviation; the fault frequency refers to the frequency of faults occurring in the past period, and the currently obtained fault frequency is mapped to the range of 0-1 through historical fault frequency; for example, if the fault frequency in a certain region is 3 times in the past 24 hours and the historical maximum fault frequency is 5 times, then the corresponding fault frequency is mapped to 0.6 in the range of 0-1.

[0100] The regional anomaly score QYP is calculated by weighting and fusing load factor, voltage deviation and fault frequency.

[0101] The formula for calculating the regional anomaly score satisfies:

[0102] QSZ = γ4 × FL + γ5 × DP + γ6 × GP; where γ4, γ5, and γ6 represent the weighting coefficients corresponding to load factor FL, voltage deviation DP, and fault frequency GP, respectively. γ4, γ5, and γ6 ∈ (0, 1), and the specific values ​​are set according to experience. In this embodiment, γ4, γ5, and γ6 are set to 0.6, 0.3, and 0.1, respectively. The larger the weighting coefficient, the greater the influence of the corresponding data when calculating the regional anomaly score.

[0103] The priority of regional analysis is calculated using a formula that satisfies the following:

[0104] QFY = QSZ α1 ×QYP α2 ; where α1 and α2 represent the adjustment index coefficients, and both α1 and α2 are greater than 0; the priority of regional analysis increases with the increase of regional data quality and regional anomaly score.

[0105] The embodiment performs multi-angle analysis and quantitative processing on the power grid data collected in each region, combines other related data, inputs into the pre-trained coefficient generation model, calculates the adjustment index coefficient reflecting the characteristics of each region, and then determines the analysis priority of each region, ensures that the priority calculation is fully based on the actual power grid operation of the region, effectively improves the accuracy and efficiency of regional analysis, and provides strong data support for subsequent load early warning.

[0106] The adjustment index coefficient in the embodiment is generated by a coefficient generation model, which includes:

[0107] The average regional data quality, regional data quality standard deviation, regional anomaly score, and regional false alarm rate corresponding to the coefficient generation model and the region ID are obtained; the average regional data quality refers to the mean of the historical regional data quality corresponding to the region ID in the past period of time, the regional data quality standard deviation measures the data quality fluctuation in the past period of time; the regional false alarm rate refers to the proportion of the number of false alarms in the past period of time; the specific value of the past period of time is set according to experience, and in the embodiment, the past period of time is set to the past 24 hours;

[0108] The average regional data quality, regional data quality standard deviation, regional anomaly score, and regional false alarm rate are integrated into regional analysis data;

[0109] The regional analysis data corresponding to the region ID is input into the coefficient generation model to obtain a plurality of adjustment index coefficients corresponding to the region ID;

[0110] The coefficient generation model is constructed by an artificial intelligence model, which includes:

[0111] A plurality of historical regional analysis data and historical adjustment index coefficients corresponding to the region ID are obtained;

[0112] The plurality of historical regional analysis data and historical adjustment index coefficients corresponding to the region ID are divided into training data, validation data, and test data; and the training data, validation data, and test data are preprocessed to obtain a training set, a validation set, and a test set; the ratio between the training set, the test set, and the validation set is 7:2:1;

[0113] An artificial intelligence model is selected as a base model; the artificial intelligence model includes a neural network model and the like;

[0114] The base model is trained by the training set, and the learning rate and other hyperparameters are adjusted on the validation set to obtain a pre-trained model;

[0115] The pre-trained model is verified on the test set, and finally a coefficient generation model with the input of regional analysis data corresponding to the region ID and the output of a plurality of adjustment index coefficients corresponding to the region ID is obtained.

[0116] The regional power grid topology map in this embodiment is constructed by the following methods, including:

[0117] Obtain regional power grid data and regional environment data;

[0118] Define power grid nodes based on power grid equipment in the regional power grid data; the power grid nodes include a plurality of node attributes; the power grid nodes include transformer substations, transformers, and user nodes, etc.; the node attributes include node ID, node type, and node coordinates, etc.;

[0119] Define edge connections based on the connection relationship of the equipment in the regional power grid data; the edge connections include a plurality of connection attributes; the edge connections include connection relationships; the connection attributes include connection type and connection length, etc.;

[0120] Map the regional environment data to the power grid nodes as node attributes; that is, the regional environment data corresponding to each power grid node is one of the node attributes of the power grid node;

[0121] Substitute the regional power grid data into a plurality of power grid nodes and a plurality of edge connections to obtain the regional power grid topology map.

[0122] This embodiment constructs a corresponding power grid topology map for each regional ID, and maps the environment data into these topology maps, thereby enhancing the information expression capability of the power grid topology map. This method enables more accurate prediction of the load situation of each regional ID when combined with time series data for analysis, thereby improving the accuracy and efficiency of load early warning, not only improving the understanding and analysis capability of the power grid operation state, but also providing a solid foundation for realizing more scientific and reasonable load management.

[0123] The regional predicted load in this embodiment is generated according to the regional analysis priority and the regional power grid topology map, including:

[0124] Obtain the regional analysis priority, the regional power grid topology map, and the regional power grid data;

[0125] Sort the regional IDs in descending order according to the regional analysis priority to obtain an analysis sequence;

[0126] Select the regional IDs in the analysis sequence in turn;

[0127] Extract a plurality of influence parameters of the regional ID in the regional power grid data within a sliding time window; the influence parameters refer to the power grid data that can affect the future power load prediction; the sliding time window is a time window size for obtaining historical data, which is set to 72 hours in this embodiment; the plurality of influence parameters include voltage, current, equipment state, and feeder load, etc.;

[0128] The several influence parameters in the sliding time window and the regional environment data corresponding to the sliding time window are integrated into a prediction sequence; that is, the several influence parameters corresponding to a time point and the regional environment data corresponding to the time point are integrated into analysis data corresponding to the time point, and then the several time points in the sliding time window are integrated into a prediction sequence in the order of the time points; the regional environment data includes temperature, humidity, wind speed, and the like;

[0129] The prediction sequence corresponding to the regional ID and the regional power grid topology graph are input into a load prediction model to obtain a regional predicted load corresponding to the regional ID; the load prediction model is constructed by an artificial intelligence model.

[0130] The load prediction model in the embodiment is constructed by an artificial intelligence model, including:

[0131] A plurality of historical prediction sequences corresponding to the regional ID, historical regional power grid topology graphs, and historical regional predicted loads are obtained;

[0132] The plurality of historical prediction sequences corresponding to the regional ID, the historical regional power grid topology graphs, and the historical regional predicted loads are divided into training data, verification data, and test data; and the training data, the verification data, and the test data are preprocessed to obtain a training set, a verification set, and a test set; the ratio between the training set, the test set, and the verification set is 7:2:1;

[0133] An artificial intelligence model is selected as a basic model; the artificial intelligence model used in the embodiment includes a convolutional neural network model and an LSTM model; the convolutional neural network model is used to process the regional power grid topology graph, and the LSTM model is used to process the prediction sequence;

[0134] The basic model is trained by the training set, and the learning rate and other hyperparameters are adjusted on the verification set to obtain a pre-trained model;

[0135] The pre-trained model is verified on the test set, and finally a load prediction model with the input of the prediction sequence corresponding to the regional ID and the regional power grid topology graph and the output of the regional predicted load corresponding to the regional ID is obtained.

[0136] The device adjustment parameter generated according to the regional predicted load in the embodiment includes:

[0137] The regional predicted load is obtained;

[0138] When the regional predicted load is greater than or equal to a regional load threshold corresponding thereto, a plurality of device adjustment parameters in a next time period are set to maximum values corresponding to device parameters; the specific value of the time period is set according to experience, and the regional predicted load in the future 1 day is predicted in the embodiment, so the size of the time period is 1 day;

[0139] Otherwise, the several device adjustment parameters STC in the next time period are calculated by a calculation formula t+1,i The calculation formula satisfies the following formula:

[0140] Wherein, t represents the number of time period, k represents the sensitivity index, k>0, the specific value is set according to experience, and k is set to 2 in the embodiment; STC t+1,i represents the several device adjustment parameters in the next time period, and i represents the number of the adjustment device; STC i,max and STC i,min respectively represent the maximum value and the minimum value of the device adjustment parameter of the i th adjustment device; QJF th and QF min respectively represent the emergency load threshold and the minimum load benchmark, the specific value is set according to experience, QF min is set to 100 kW in the embodiment, and QJF th is set to 90% of the maximum load benchmark, and the maximum load benchmark is set to 1000 kW in the embodiment, that is, QJF th is set to 900 kW; QYF t+1 represents the regional predicted load in the next time period; the adjustment device in the embodiment can be divided into three categories, including SCADA device, smart meter device and weather API device, the device adjustment parameter corresponding to the adjustment device is the collection frequency, the SCADA device is used to collect bus voltage, current and power factor, etc., the smart meter device is used to collect the user load curve and harmonic content in the region, and the weather API device is used to obtain environmental data; the higher the regional predicted load is, the higher the collection frequency in the next time period in the region will be, so as to ensure the data accuracy in the next time period.

[0141] The embodiment can improve the accuracy of the power grid data in the region with high predicted load by deeply analyzing the relationship between the regional predicted load and the device adjustment parameter, dynamically adjusting the device parameter corresponding to the region when the predicted load of the region is detected to change significantly, thereby enhancing the accuracy of the load warning; not only the power grid operation efficiency is optimized, but also the response can be more timely and accurate when the load changes significantly.

[0142] The alarm signal generated according to the regional predicted load in the embodiment includes:

[0143] Obtaining the regional predicted load corresponding to the several region IDs;

[0144] Judging whether the regional predicted load of the region ID is greater than the regional load threshold corresponding to the region ID;

[0145] Yes, generating the overload alarm signal of the region ID;

[0146] No, judge whether the regional prediction load of the region ID is greater than the D times regional load threshold corresponding to the region ID; wherein, D is a proportion coefficient, D ∈ (0, 1), the specific value is set according to experience, and D is set to 0.7 in the embodiment;

[0147] Yes, generate a region ID overload risk warning signal; No, do nothing.

[0148] The regional load threshold in the embodiment is generated by the following method, comprising:

[0149] Obtain the historical same period regional load basic value QFJ corresponding to the region ID in the next time period; the historical same period regional load basic value represents the baseline value of the load threshold corresponding to the region ID in the historical same period, and the historical same period regional load basic value corresponding to a region in the region ID in the embodiment is 650 kW;

[0150] Obtain the regional prediction load trend QYFQ corresponding to the region ID in the next time period; the regional prediction load trend represents the trend change of the prediction load in the past M hours before the prediction time point, QYFQ ∈ [-1, 1], when QYFQ > 0, it means that the prediction load shows an upward trend, otherwise a downward trend; M is a positive integer, and M is set to 24 hours in the embodiment;

[0151] Obtain the regional environment data corresponding to the region ID in the next time period; the regional environment data includes a plurality of environment parameters;

[0152] Calculate the regional environment trend QHQ corresponding to the region ID in the next time period by the calculation formula; the calculation formula satisfies the following formula:

[0153] Wherein, t represents the number of time period, j represents the number of environment parameter, the total number of environment parameters is J, HC t+1,j represents the specific value of the jth environment parameter in the next time period, which can be obtained by weather forecast; HC base,j represents the historical same period average value corresponding to the jth environment parameter; HCQ j represents the weight of the jth environment parameter, and the specific value is set according to experience; in the embodiment, the environment parameters include temperature, humidity and wind speed, and the weights corresponding to temperature, humidity and wind speed are set to 0.7, 0.2 and 0.1 respectively; the regional environment trend increases with the increase of the deviation degree of the environment parameter from the historical same period average value;

[0154] Based on the nonlinear relationships between the historical regional load baseline value QFJ, the regional predicted load trend QYFQ, the regional environmental trend QHQ and the regional load threshold, a regional load threshold calculation function QFYJF(QFJ, QYFQ, QHQ) is constructed.

[0155] The regional load threshold calculation function satisfies the following calculation formula:

[0156] QFYJF t+1,n (QFJ t+1,n QYFQ t+1,n QHQ t+1,n )=QFJ t+1,n ×(1+β1×QYFQ t+1，n +β2×QHQ t+1，n ); where n represents the number corresponding to the region ID, β1 and β2 represent the load trend influence coefficient and the environmental trend influence coefficient, respectively, β1 and β2∈(0,1); the specific values ​​are set according to experience, and in this embodiment, β1 and β2 are set to 0.5 and 0.3, respectively;

[0157] Substituting the historical regional load baseline value, regional predicted load trend, and regional environmental trend corresponding to the regional ID into the regional load calculation function yields the regional load threshold corresponding to the regional ID.

[0158] This embodiment analyzes historical and current power grid data for each region at different times and dynamically adjusts the load threshold corresponding to each region ID. This allows the load threshold for each region to adaptively adjust over time according to actual operating conditions, avoiding the limitations of fixed thresholds, effectively reducing false alarms in load warnings, and significantly improving the accuracy and reliability of power grid load warnings.

[0159] Please see Figure 2 Another embodiment of this application provides a distributed load early warning management system based on a power grid, including: a data acquisition module, a data analysis module, and an early warning module; the data acquisition module and the data analysis module are electrically and / or communicatively connected; the data analysis module and the early warning module are electrically and / or communicatively connected.

[0160] Data acquisition module: Acquires regional data through data acquisition equipment; regional data includes the regional ID and its corresponding regional power grid data and regional environmental data; the data acquisition equipment includes several sensors, etc.

[0161] The data analysis module generates regional analysis priorities according to regional power grid data, constructs a regional power grid topology map, generates regional predicted loads according to the regional analysis priorities and the regional power grid topology map, generates device adjustment parameters according to the regional predicted loads, and assigns the device adjustment parameters to corresponding device parameters, and generates alarm signals according to the regional predicted loads.

[0162] The early warning module prompts according to the alarm signals, and the alarm signals include a regional ID overload alarm signal and a regional ID overload risk early warning signal.

[0163] Some data in the above formula is calculated by removing the dimension, and the formula is obtained by software simulation of a large amount of collected data to obtain a formula closest to the actual situation; the preset parameters and the preset threshold in the formula are set by a person skilled in the art according to the actual situation or obtained by a large amount of data simulation.

[0164] The working principle of the present application is as follows: regional data is obtained, regional analysis priorities are generated according to regional power grid data, a regional power grid topology map is constructed, regional predicted loads are generated according to the regional analysis priorities and the regional power grid topology map, device adjustment parameters are generated according to the regional predicted loads, and the device adjustment parameters are assigned to corresponding device parameters, and alarm signals are generated according to the regional predicted loads. The power grid data of each region is analyzed from multiple angles, so that important regions can be given priority attention during load early warning, and the corresponding device parameters are dynamically adjusted according to the load prediction of each region, so that the accuracy of data is improved during high load early warning. The efficiency and accuracy of early warning are improved, and the prior state of each region during early warning is ignored in the prior art, so that the region in urgent need of early warning cannot be analyzed in time and effectively when a number of regions are simultaneously analyzed, resulting in low early warning efficiency.

[0165] The above embodiments are only used to illustrate the technical method of the present application and not to limit it. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A power grid-based distributed load early warning management method, characterized in that, The method comprises the following steps: obtaining area data; the area data comprises an area ID and corresponding area power grid data and area environment data; generating an area analysis priority according to the area power grid data; the area analysis priority refers to the priority of analyzing the area data; constructing an area power grid topology map; generating a predicted load of the area according to the area analysis priority and the area power grid topology map; the predicted load of the area refers to a predicted load value of the area in a future time period; generating a device adjustment parameter according to the predicted load of the area; and assigning the device adjustment parameter to a corresponding device parameter; the device adjustment parameter refers to adjusting the parameter value of the related device according to the predicted load value; generating an alarm signal according to the predicted load of the area; the method of generating an area analysis priority according to area power grid data comprises the following steps: obtaining area power grid data; extracting data integrity, data accuracy and data timeliness corresponding to the area power grid data; calculating the quality of area data QSZ by weighted fusion of data integrity, data accuracy and data timeliness; extracting the load rate, voltage deviation and fault frequency corresponding to the area power grid data; calculating the area abnormal score QYP by weighted fusion of the load rate, voltage deviation and fault frequency; calculating the area analysis priority by formula, and the calculation formula satisfies: ;in, and These are respectively represented as adjustment index coefficients. and All are greater than 0.

2. The power grid-based distributed load early warning management method according to claim 1, characterized in that, the adjustment index coefficient is generated by a coefficient generation model, comprising: obtaining the average area data quality, area data quality standard deviation, area abnormal score and area false alarm rate corresponding to the coefficient generation model and the area ID; integrating the average area data quality, area data quality standard deviation, area abnormal score and area false alarm rate into area analysis data; inputting the area analysis data corresponding to the area ID into the coefficient generation model to obtain a plurality of adjustment index coefficients corresponding to the area ID; the coefficient generation model is constructed by an artificial intelligence model, comprising: obtaining a plurality of historical area analysis data and historical adjustment index coefficients corresponding to the area ID; dividing the plurality of historical area analysis data and historical adjustment index coefficients corresponding to the area ID into training data, validation data and test data; and performing data preprocessing on the training data, validation data and test data to obtain a training set, a validation set and a test set; selecting an artificial intelligence model as a base model; training the base model through the training set, and adjusting the learning rate and other hyperparameters on the validation set to obtain a pre-trained model; verifying the pre-trained model on the test set to finally obtain a coefficient generation model with the area ID corresponding to the area analysis data as input and a plurality of adjustment index coefficients corresponding to the area ID as output. 3.The grid-based distributed load early warning management method according to claim 1, characterized in that, the area power grid topology map is constructed by the following method, comprising: obtaining area power grid data and area environment data; defining a power grid node based on the power grid equipment in the area power grid data; the power grid node comprises a plurality of node attributes; defining an edge connection based on the device connection relationship in the area power grid data; the edge connection comprises a plurality of connection attributes; mapping the area environment data to the power grid node as a node attribute; substituting the area power grid data into a plurality of power grid nodes and a plurality of edge connections to obtain the area power grid topology map.

4. The power grid-based distributed load early warning management method according to claim 1, characterized in that, The region prediction load is generated according to the region analysis priority and the region power grid topology, and the region prediction load comprises the following steps: obtaining the region analysis priority, the region power grid topology and the region power grid data; sequentially selecting the region ID in the analysis sequence; extracting a plurality of influence parameters of the region ID in the region power grid data in the sliding time window; the influence parameters refer to the power grid data which can affect the future power load prediction; integrating the plurality of influence parameters in the sliding time window and the region environmental data corresponding to the sliding time window into a prediction sequence; inputting the prediction sequence corresponding to the region ID and the region power grid topology into a load prediction model to obtain the region prediction load corresponding to the region ID; the load prediction model is constructed by an artificial intelligence model. The load prediction model is constructed by an artificial intelligence model, and the load prediction model comprises the following steps:

5. The grid-based distributed load early warning management method according to claim 4, characterized in that, obtaining a plurality of historical prediction sequences corresponding to the region ID, historical region power grid topologies and historical region prediction loads; dividing the plurality of historical prediction sequences corresponding to the region ID, the historical region power grid topologies and the historical region prediction loads into training data, verification data and test data; and performing data preprocessing on the training data, the verification data and the test data to obtain a training set, a verification set and a test set; selecting an artificial intelligence model as a basic model; training the basic model by using the training set, and adjusting the learning rate and other hyperparameters on the verification set to obtain a pre-trained model; verifying the pre-trained model on the test set, and finally obtaining a load prediction model with the prediction sequence corresponding to the region ID and the region power grid topology as input and the region prediction load corresponding to the region ID as output. The device adjustment parameter is generated according to the region prediction load, and the device adjustment parameter comprises the following steps:

6. The grid-based distributed load early warning management method according to claim 1, characterized in that, obtaining the region prediction load; when the region prediction load is greater than or equal to the corresponding region load threshold, setting a plurality of device adjustment parameters in the next time period to the maximum value of the device parameter; The alarm signal is generated according to the region prediction load, and the alarm signal comprises the following steps: Otherwise, a number of device adjustment parameters for the next time period are calculated by a calculation formula The calculation formula satisfies the following equation: ; where t denotes a number of a time period, k denotes a sensitivity index, k > 0; denotes a number of devices adjustment parameters of the next time period, i denotes a number of adjustment devices; and denote a maximum value and a minimum value of the device adjustment parameters of the i-th adjustment device, respectively; and denote an emergency load threshold and a minimum load reference, respectively; denotes a zone forecast load of the next time period.

7. The grid-based distributed load early warning management method according to claim 1, characterized in that, obtaining the region prediction load corresponding to a plurality of region IDs; determining whether the region prediction load of the region ID is greater than the region load threshold corresponding to the region ID; yes, generating an overload alarm signal of the region ID; no, determining whether the region prediction load of the region ID is greater than the D times of the region load threshold corresponding to the region ID; wherein D is a proportion coefficient, and D [0, 1]; yes, generating an overload risk early warning signal of the region ID; no, performing no operation. The region load threshold is generated by the following method, and the region load threshold comprises the following steps:

8. A grid-based distributed load warning management method according to claim 6 or claim 7, characterized in that, obtaining the historical same period region load basic value QFJ, the region prediction load trend QYFQ and the region environmental data corresponding to the region ID in the next time period; the region environmental data comprises a plurality of environmental parameters; calculating the region environmental trend QHQ corresponding to the region ID in the next time period by using a calculation formula; the calculation formula satisfies the following formula: constructing a region load threshold calculation function QFYJF(QFJ, QYFQ, QHQ) according to the nonlinear relationship between the historical same period region load basic value QFJ, the region prediction load trend QYFQ, the region environmental trend QHQ and the region load threshold; ; wherein t denotes a number of a time period, j denotes a number of an environmental parameter, a total number of the environmental parameters being J, denotes a specific value of the jth environmental parameter at the next time period; denotes a historical same-period average corresponding to the jth environmental parameter; denotes a weight of the jth environmental parameter; ​ The historical and periodic regional load basic value corresponding to the region ID, the regional predicted load trend and the regional environment trend are substituted into a regional load calculation function to obtain a regional load threshold corresponding to the region ID.

9. A power grid-based distributed load early warning management system applied to the power grid-based distributed load early warning management method of any one of claims 1-8, characterized in that, Comprise: The data acquisition module and the data analysis module are connected with each other; The data acquisition module: acquires regional data through a data acquisition device;The regional data comprises a region ID and corresponding regional power grid data and regional environment data; The data analysis module: generates a regional analysis priority according to the regional power grid data;Constructs a regional power grid topology map;Generate a regional predicted load according to the regional analysis priority and the regional power grid topology map;Generate a device adjustment parameter according to the regional predicted load, and assign the device adjustment parameter to the corresponding device parameter; Generate an alarm signal according to the regional predicted load; The regional analysis priority is generated according to the regional power grid data, comprising: Obtain regional power grid data; Extract the data integrity, data accuracy and data timeliness corresponding to the regional power grid data; The regional data quality QSZ is calculated by weighted fusion method according to the data integrity, data accuracy and data timeliness; Extract the load rate, voltage deviation and fault frequency corresponding to the regional power grid data; The regional abnormal score QYP is calculated by weighted fusion method according to the load rate, voltage deviation and fault frequency; The regional analysis priority is calculated by formula, and the calculation formula satisfies: ;in, and These are respectively represented as adjustment index coefficients. and All are greater than 0.

Images