Power grid load statistical method

Abstract

The invention belongs to the technical field of power systems, and discloses a power grid load statistical method, which specifically comprises the following steps of: 1, collecting data; 2, data preprocessing; step 3, load classification; 4, calculating an adjustable load; 5, establishing a load prediction model; step 6, real-time load monitoring and prediction; step 7, error correction and feedback; 8, comparing data; and ninthly, generating a load statistical report and performing early warning. According to the method, the source of the power grid load data in the statistical region is comprehensively collected, so that the load classification can be more accurate, the load prediction model can predict the power grid load by combining the operation characteristics of the power grid in the statistical region and the change rule of the electrical load, and meanwhile, due to the establishment of the power grid load prediction model, the power grid load prediction efficiency is improved. Therefore, a manager can compare the statistical power grid load data with the prediction data of the load prediction model when counting the power grid load data.

Description

Technical Field

[0001] The invention belongs to the technical field of electric power systems, and in particular relates to a power grid load statistics method. Background Art

[0002] The whole composed of substations of various voltages and transmission and distribution lines in the power system is called the power grid, which includes three units: substation, transmission and distribution. The task of the power grid is to transmit and distribute electric energy and change the voltage. The total load of the power system is the sum of the total power consumed by all electrical equipment in the system. The comprehensive power load of the power system is obtained by adding up the power consumed by industry, agriculture, post and telecommunications, transportation, municipal administration, commerce, and urban and rural residents.

[0003] When the regional power grid is transmitting and distributing electricity, managers need to grasp the actual load situation of the power grid in real time. However, the real-time load situation of the power grid is the total load of the power grid obtained by the actual output of the units in the power grid dispatching organization. Under normal circumstances, managers can obtain the actual load situation of the power grid through calculation. Although the existing statistical method can obtain the current load of the power grid, it is unable to predict and count the power grid load, which makes it impossible for managers to respond to abnormal situations in advance. At the same time, the existing power grid load data is not compared with the data at the same time after statistics, which may lead to the abnormality of the power grid load data may not be discovered in time, so it needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to provide a power grid load statistics method to solve the problems raised in the above background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a power grid load statistics method, the specific steps are as follows:

[0006] Step 1: Collect data

[0007] According to the scope of the power grid managed by the current power grid dispatching organization, the power grid area that needs to be counted is clarified, and the source of the total power grid load data that needs to be collected is determined to ensure that the source of the total power grid load data is stable and reliable, and can provide accurate total power grid load data in real time;

[0008] Step 2: Data preprocessing

[0009] Unified format: All collected grid total load data are converted into a processable format. Secondly, the timestamp of the grid total load is standardized to ensure the time consistency and integrity of the grid total load data.

[0010] Data cleaning: Clean the total load data of the power grid, remove duplicate data, and repair erroneous data and missing values;

[0011] Step 3: Load Classification

[0012] According to the load type and characteristics, the total load of the power grid is divided into two categories: fixed load and adjustable load;

[0013] Among them, fixed load refers to the load whose power demand or power consumption is relatively stable within a certain period of time and is not easy to be adjusted significantly through direct control means; adjustable load refers to the load whose power demand or power consumption can be adjusted to a certain extent through human control or technical means;

[0014] Baseline load identification: The baseline load is the natural load level without grid dispatch intervention. The baseline load of fixed load is identified and calculated. The baseline load of fixed load is usually obtained through historical data analysis and statistics.

[0015] Step 4: Adjustable load calculation

[0016] Quantitative calculation: The adjustable load is obtained by subtracting the baseline load of the fixed load from the total load of the power grid;

[0017] Step 5: Establish a load forecasting model

[0018] Based on the historical baseline load data and adjustable load characteristics, combined with the operation characteristics of the power grid in the statistical area and the change law of power load, a load forecasting model suitable for the power grid is established;

[0019] By comparing the prediction results of the load forecasting model with the actual load data of the power grid, the accuracy and stability of the load forecasting model are evaluated. Cross-validation, mean square error (MSE) and root mean square error (RMSE) are used to evaluate the performance of the load forecasting model. According to the evaluation results, the parameters of the load forecasting model are adjusted and optimized to improve the prediction accuracy.

[0020] Step 6: Real-time load monitoring and forecasting

[0021] The DMS system of the power grid is an important part of the power grid management system. It can realize real-time data collection and monitoring of the power grid, connect the data of the power grid DMS system with the load forecasting model, and thus collect real-time load data of the power grid in real time. These data include time, region, and maximum load value, which can directly reflect the real-time load situation of the power grid;

[0022] Online prediction: Use the trained load prediction model to perform load prediction;

[0023] Step 7: Error Correction and Feedback

[0024] Use the real-time load data of the power grid to correct the error of the prediction results, modify the load prediction model, improve the prediction accuracy, calculate the data error between the prediction time and the actual time, and apply the error mean to the future prediction results to gradually reduce the error;

[0025] Establish a feedback mechanism for load statistics and forecasting, and adjust load classification and load forecasting model parameters according to actual conditions;

[0026] Step 8: Data comparison

[0027] The dispatcher compares the real-time load data of the current regional power grid with the predicted result data of the load forecasting model. If the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is less than 3%, it means that there is no abnormality in the real-time load data of the power grid, and the real-time load data of the power grid is taken for statistics; if the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is greater than 3%, the real-time load data of the power grid is reviewed, and the current power grid is checked for faults;

[0028] Step 9: Load statistics report generation and early warning

[0029] The load forecasting model is used to generate a report on the real-time load data of the power grid, the prediction results of the load forecasting model, and the quantification results of adjustable resources. At the same time, a statistical analysis of the power grid load is performed to calculate key indicators such as the total load, the proportion of various loads, the peak-to-valley difference, and the load rate. Based on the statistical analysis results, a power grid load statistical report is generated; the load forecasting model is used to monitor the power grid load in real time, and thresholds or warning levels are set. Once the load forecasting model predicts that the load exceeds the set value or an abnormal fluctuation occurs, an alarm is triggered immediately, so that the abnormality can be handled in advance.

[0030] Preferably, the total grid load data collected in step one includes real-time power consumption, voltage, current, and distributed photovoltaic output and operating status of adjustable loads.

[0031] Preferably, the total grid load data format described in step 2 needs to be uniformly converted into CSV or JSON format, and the grid load data after the format conversion needs to be checked.

[0032] Preferably, the load classification described in step three divides the grid load into multiple more detailed categories according to the type, function and load characteristics of the electrical equipment, such as residential electricity, industrial electricity, commercial electricity, electric vehicle charging load, and distributed photovoltaic power generation, and classifies the load by the meter installation location and user attribute information.

[0033] Preferably, the load forecasting model in step 5 is established by combining time series analysis, and its specific formula is:

[0034] L t =LSTM(L t ,L t-1 ,...,L t-n )+SA(t)

[0035] in,

[0036] L t represents the predicted value of the power grid load at time t;

[0037] LSTM(L t ,L t-1 ,...,L t-n ) represents the load data of the past n time points using the long short-term memory network LSTM;

[0038] SA(t) represents the seasonal adjustment factor, which is a function based on time t and is used to further adjust the forecast value to match the seasonal load variation.

[0039] Preferably, the total grid load data described in step 2 needs to be aggregated every hour. Each time the total grid load data is aggregated, the average total grid load data within the time period needs to be calculated. The specific calculation formula for the average total grid load data within the time period is:

[0040]

[0041] in,

[0042] AverageLoad T represents the average load in time period T;

[0043] Lt represents the load value at time point t;

[0044] T| represents the total number of time points in time period T.

[0045] Preferably, when generating the power grid load statistics report in step nine, it is necessary to use a visualization tool to present the analysis results in the form of a chart, and the report content needs to include the basic load situation, change trend, and existing problems.

[0046] Preferably, the sources of the total grid load data in step one include smart meters, distribution automation systems, electricity market trading platforms, energy storage systems and distributed power management systems.

[0047] Preferably, the error measurement formula used in the comparison between the actual real-time load data of the power grid and the prediction result data of the load prediction model in step eight is used to calculate the absolute average error, and the error measurement formula is:

[0048]

[0049] in,

[0050] MAE stands for mean absolute error;

[0051] N is the total number of data points compared;

[0052] Indicates the sum of all N data points;

[0053] L actual,i represents the actual grid load value of the ith data point;

[0054] L predicted,i represents the load model prediction value of the i-th data point;

[0055] |L actual,i -L predicted,i | represents taking the absolute value, which is used to ensure that the error is always non-negative.

[0056] The beneficial effects of the present invention are as follows:

[0057] By comprehensively collecting the total load data sources of the statistical area power grid, the classification of load can be more accurate, and the load forecasting model can predict the power grid load in combination with the operating characteristics of the power grid in the statistical area and the changing law of power load. At the same time, due to the establishment of the power grid load forecasting model, the management personnel can compare the actual load data of the statistical power grid with the predicted data of the load forecasting model, so as to calibrate the actual load data of the statistical power grid to avoid data errors, making the actual load data of the power grid more accurate. At the same time, the load forecasting model is used to predict the power grid load, so that the management personnel can respond to abnormal situations in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the operation flow of the present invention. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] like Figure 1 As shown, the embodiment of the present invention provides a method for counting power grid loads, and the specific steps are as follows:

[0061] Step 1: Collect data

[0062] According to the scope of the power grid managed by the current power grid dispatching organization, the power grid area that needs to be counted is clarified, and the source of the total power grid load data that needs to be collected is determined to ensure that the source of the total power grid load data is stable and reliable, and can provide accurate total power grid load data in real time;

[0063] Step 2: Data preprocessing

[0064] Unified format: All collected grid total load data are converted into a processable format. Secondly, the timestamp of the grid total load is standardized to ensure the time consistency and integrity of the grid total load data.

[0065] Data cleaning: Clean the total load data of the power grid, remove duplicate data, and repair erroneous data and missing values;

[0066] Step 3: Load Classification

[0067] According to the load type and characteristics, the total load of the power grid is divided into two categories: fixed load and adjustable load;

[0068] Among them, fixed load refers to the load whose power demand or power consumption is relatively stable within a certain period of time and is not easy to be adjusted significantly through direct control means; adjustable load refers to the load whose power demand or power consumption can be adjusted to a certain extent through human control or technical means;

[0069] Baseline load identification: The baseline load is the natural load level without grid dispatch intervention. The baseline load of fixed load is identified and calculated. The baseline load of fixed load is usually obtained through historical data analysis and statistics.

[0070] Step 4: Adjustable load calculation

[0071] Quantitative calculation: The adjustable load is obtained by subtracting the baseline load of the fixed load from the total load of the power grid;

[0072] Step 5: Establish a load forecasting model

[0073] Based on the historical baseline load data and adjustable load characteristics, combined with the operation characteristics of the power grid in the statistical area and the change law of power load, a load forecasting model suitable for the power grid is established;

[0074] By comparing the prediction results of the load forecasting model with the actual load data of the power grid, the accuracy and stability of the load forecasting model are evaluated. Cross-validation, mean square error (MSE) and root mean square error (RMSE) are used to evaluate the performance of the load forecasting model. According to the evaluation results, the parameters of the load forecasting model are adjusted and optimized to improve the prediction accuracy.

[0075] Step 6: Real-time load monitoring and forecasting

[0076] The DMS system of the power grid is an important part of the power grid management system. It can realize real-time data collection and monitoring of the power grid, connect the data of the power grid DMS system with the load forecasting model, and thus collect real-time load data of the power grid in real time. These data include time, region, and maximum load value, which can directly reflect the real-time load situation of the power grid;

[0077] Online prediction: Use the trained load prediction model to perform load prediction;

[0078] Step 7: Error Correction and Feedback

[0079] Use the real-time load data of the power grid to correct the error of the prediction results, modify the load prediction model, improve the prediction accuracy, calculate the data error between the prediction time and the actual time, and apply the error mean to the future prediction results to gradually reduce the error;

[0080] Establish a feedback mechanism for load statistics and forecasting, and adjust load classification and load forecasting model parameters according to actual conditions;

[0081] Step 8: Data comparison

[0082] The dispatcher compares the real-time load data of the current regional power grid with the predicted result data of the load forecasting model. If the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is less than 3%, it means that there is no abnormality in the real-time load data of the power grid, and the real-time load data of the power grid is taken for statistics; if the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is greater than 3%, the real-time load data of the power grid is reviewed, and the current power grid is checked for faults;

[0083] Step 9: Load statistics report generation and early warning

[0084] The load forecasting model is used to generate a report on the real-time load data of the power grid, the prediction results of the load forecasting model, and the quantification results of adjustable resources. At the same time, a statistical analysis of the power grid load is performed to calculate key indicators such as the total load, the proportion of various loads, the peak-to-valley difference, and the load rate. Based on the statistical analysis results, a power grid load statistical report is generated; the load forecasting model is used to monitor the power grid load in real time, and thresholds or warning levels are set. Once the load forecasting model predicts that the load exceeds the set value or an abnormal fluctuation occurs, an alarm is triggered immediately, so that the abnormality can be handled in advance.

[0085] By comprehensively collecting data from the statistical area, the subsequent prediction of power grid load data can be made more accurate. At the same time, by dividing the total power grid load into fixed load and adjustable load, the fluctuation of power grid load can be predicted, making the real-time power grid load data more accurate. Finally, by comparing the real-time power grid load data with the predicted data, the accuracy of the real-time power grid load data can be ensured, and the possible fluctuation of power grid load in the future can be predicted, so as to handle and respond to abnormal situations in advance.

[0086] Among them, the total grid load data collected in step one includes real-time power consumption, voltage, current, and the operating status of distributed photovoltaic output and adjustable load.

[0087] By collecting total load data of various types of power grids, we can conduct statistics and forecasts on power grid loads in a more comprehensive manner, thereby reducing the impact of other factors on the accuracy of total load data of power grids.

[0088] Among them, the total grid load data format in step 2 needs to be uniformly converted into CSV and JSON formats, and the grid load data after format conversion needs to be checked.

[0089] Standardize data by converting data formats to improve data quality and consistency.

[0090] Among them, the load classification in step three divides the grid load into multiple more detailed categories according to the type, function and load characteristics of the electrical equipment, such as residential electricity, industrial electricity, commercial electricity, electric vehicle charging load, and distributed photovoltaic power generation, and classifies the load according to the meter installation location and user attribute information.

[0091] By classifying the power grid load in more detail, we can deeply analyze the characteristics of various types of loads, including the peak and valley characteristics, fluctuation characteristics, and seasonal change characteristics of the load curve. We can also apply data analysis tools and techniques, such as time series analysis, trend forecasting, and association rule mining, to further explore the laws and characteristics of load changes.

[0092] Among them, the load forecasting model in step 5 is established by combining time series analysis, and its specific formula is:

[0093] L t =LSTM(L t ,L t-1 ,...,L t-n )+SA(t)

[0094] in,

[0095] L t represents the predicted value of the power grid load at time t;

[0096] LSTM(L t ,L t-1 ,...,L t-n ) represents the load data of the past n time points using the long short-term memory network LSTM;

[0097] SA(t) represents the seasonal adjustment factor, which is a function based on time t and is used to further adjust the forecast value to match the seasonal load variation.

[0098] By establishing a load forecasting model, the grid load can be predicted, allowing managers to compare real-time data with predicted data, thereby ensuring the accuracy of the real-time grid load data and avoiding managers from failing to promptly discover abnormal real-time grid load data due to grid failures.

[0099] Among them, the total grid load data in step 2 needs to be aggregated every hour. Each time the total grid load data is aggregated, the average total grid load data within the time period needs to be calculated. The specific calculation formula for the average total grid load data within the time period is:

[0100]

[0101] in,

[0102] AverageLoad T represents the average load in time period T;

[0103] Lt represents the load value at time point t;

[0104] |T| represents the total number of time points in time period T.

[0105] By aggregating the collected data every hour, we can avoid data accumulation and make the data too large for subsequent analysis.

[0106] Among them, when generating the power grid load statistics report in step nine, it is necessary to use visualization tools to present the analysis results in the form of charts, and the report content needs to include the basic load situation, change trends, and existing problems.

[0107] By presenting the report in the form of a chart, the load changes can be displayed more intuitively, making it easier for managers to understand the grid load.

[0108] Among them, the sources of total grid load data in step one include smart meters, distribution automation systems, electricity market trading platforms, energy storage systems and distributed power management systems.

[0109] By conducting comprehensive data collection on the sources of total grid load data, the total grid load data can be made more accurate and facilitate the subsequent classification of grid loads.

[0110] Among them, the error measurement formula used in step eight to compare the actual real-time load data of the power grid with the load forecasting model prediction result data is used to calculate the absolute average error. The error measurement formula is:

[0111]

[0112] in,

[0113] MAE stands for mean absolute error;

[0114] N is the total number of data points compared;

[0115] Indicates the sum of all N data points;

[0116] L actual,i represents the actual grid load value of the ith data point;

[0117] L predicted,i represents the load model prediction value of the i-th data point;

[0118] |L actual,i -L predicted,i | represents taking the absolute value, which is used to ensure that the error is always non-negative.

[0119] By cleaning the data, we can ensure the accuracy of the data, improve the data integrity, and increase its usability and value.

[0120] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0121] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A power grid load statistics method, characterized in that: The specific steps are as follows: Step 1: Collect data According to the scope of the power grid managed by the current power grid dispatching organization, the power grid area that needs to be counted is clarified, and the source of the total power grid load data that needs to be collected is determined to ensure that the source of the total power grid load data is stable and reliable, and can provide accurate total power grid load data in real time; Step 2: Data preprocessing Unified format: All collected grid total load data are converted into a processable format. Secondly, the timestamp of the grid total load is standardized to ensure the time consistency and integrity of the grid total load data. Data cleaning: Clean the total load data of the power grid, remove duplicate data, and repair erroneous data and missing values; Step 3: Load Classification According to the load type and characteristics, the total load of the power grid is divided into two categories: fixed load and adjustable load; Among them, fixed load refers to the load whose power demand or power consumption is relatively stable within a certain period of time and is not easy to be adjusted significantly through direct control means; adjustable load refers to the load whose power demand or power consumption can be adjusted to a certain extent through human control or technical means; Baseline load identification: The baseline load is the natural load level without grid dispatch intervention. The baseline load of fixed load is identified and calculated. The baseline load of fixed load is usually obtained through historical data analysis and statistics. Step 4: Adjustable load calculation Quantitative calculation: The adjustable load is obtained by subtracting the baseline load of the fixed load from the total load of the power grid; Step 5: Establish a load forecasting model Based on the historical baseline load data and adjustable load characteristics, combined with the operation characteristics of the power grid in the statistical area and the change law of power load, a load forecasting model suitable for the power grid is established; By comparing the prediction results of the load forecasting model with the actual load data of the power grid, the accuracy and stability of the load forecasting model are evaluated. Cross-validation, mean square error (MSE) and root mean square error (RMSE) are used to evaluate the performance of the load forecasting model. According to the evaluation results, the parameters of the load forecasting model are adjusted and optimized to improve the prediction accuracy. Step 6: Real-time load monitoring and forecasting The DMS system of the power grid is an important part of the power grid management system. It can realize real-time data collection and monitoring of the power grid, connect the data of the power grid DMS system with the load forecasting model, and thus collect real-time load data of the power grid in real time. These data include time, region, and maximum load value, which can directly reflect the real-time load situation of the power grid; Online prediction: Use the trained load prediction model to perform load prediction; Step 7: Error Correction and Feedback Use the real-time load data of the power grid to correct the error of the prediction results, modify the load prediction model, improve the prediction accuracy, calculate the data error between the prediction time and the actual time, and apply the error mean to the future prediction results to gradually reduce the error; Establish a feedback mechanism for load statistics and forecasting, and adjust load classification and load forecasting model parameters according to actual conditions; Step 8: Data comparison The dispatcher compares the real-time load data of the current regional power grid with the predicted result data of the load forecasting model. If the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is less than 3%, it means that there is no abnormality in the real-time load data of the power grid, and the real-time load data of the power grid is taken for statistics; if the error value between the real-time load data of the power grid and the predicted result data of the load forecasting model is greater than 3%, the real-time load data of the power grid is reviewed, and the current power grid is checked for faults; Step 9: Load statistics report generation and early warning The load forecasting model is used to generate a report on the real-time load data of the power grid, the prediction results of the load forecasting model, and the quantification results of adjustable resources. At the same time, a statistical analysis of the power grid load is performed to calculate key indicators such as the total load, the proportion of various loads, the peak-to-valley difference, and the load rate. Based on the statistical analysis results, a power grid load statistical report is generated; the load forecasting model is used to monitor the power grid load in real time, and thresholds or warning levels are set. Once the load forecasting model predicts that the load exceeds the set value or an abnormal fluctuation occurs, an alarm is triggered immediately, so that the abnormality can be handled in advance.

2. A power grid load statistics method according to claim 1, characterized in that: The total grid load data collected in step 1 includes real-time power consumption, voltage, current, and the operating status of distributed photovoltaic output and adjustable load.

3. A power grid load statistics method according to claim 1, characterized in that: The total grid load data format described in step 2 needs to be uniformly converted into CSV and JSON formats, and the grid load data after format conversion needs to be checked.

4. A power grid load statistics method according to claim 1, characterized in that: The load classification described in step three divides the grid load into multiple more detailed categories according to the type, function and load characteristics of the electrical equipment, such as residential electricity, industrial electricity, commercial electricity, electric vehicle charging load, and distributed photovoltaic power generation, and classifies the load by the meter installation location and user attribute information.

5. A power grid load statistics method according to claim 1, characterized in that: The load forecasting model described in step 5 is established by combining time series analysis, and its specific formula is: L t =LSTM(L t ,L t-1 ,...,L t-n )+SA(t) in, L t represents the predicted value of the power grid load at time t; LSTM(L t ,L t-1 ,...,L t-n ) represents the load data of the past n time points using the long short-term memory network LSTM; SA(t) represents the seasonal adjustment factor, which is a function based on time t and is used to further adjust the forecast value to match the seasonal load variation.

6. A power grid load statistics method according to claim 1, characterized in that: The total grid load data described in step 2 needs to be aggregated every hour. Each time the total grid load data is aggregated, the average total grid load data within the time period needs to be calculated. The specific calculation formula for the average total grid load data within the time period is: in, AverageLoad T represents the average load in time period T; Lt represents the load value at time point t; |T| represents the total number of time points in time period T.

7. A power grid load statistics method according to claim 1, characterized in that: When generating the power grid load statistics report described in step nine, it is necessary to use visualization tools to present the analysis results in the form of charts, and the report content needs to include the basic load situation, change trends, and existing problems.

8. A power grid load statistics method according to claim 1, characterized in that: The sources of the total grid load data described in step one include smart meters, distribution automation systems, electricity market trading platforms, energy storage systems and distributed power management systems.

9. A power grid load statistics method according to claim 1, characterized in that: The error measurement formula used in step eight to compare the actual real-time load data of the power grid with the load forecasting model prediction result data is used to calculate the absolute average error. The error measurement formula is: in, MAE stands for mean absolute error; N is the total number of data points compared; Indicates the sum of all N data points; L actual,i represents the actual grid load value of the ith data point; L predicted,i represents the load model prediction value of the i-th data point; |L actual,i -L predicted,i | represents taking the absolute value, which is used to ensure that the error is always non-negative.

Images