Load control simulation method, system and equipment for simulating extreme weather power load scene and medium
By decomposing historical power load data and constructing a load time series prediction model, combined with an extreme weather feature library and regulation commands, the problem that load control simulation devices cannot simulate extreme weather load changes was solved. This enabled accurate training scenarios and verification of regulation effects, and improved the ability of operation and maintenance personnel to cope with extreme weather.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-03-10
AI Technical Summary
Existing load control simulation training devices cannot accurately reproduce the load change patterns under extreme weather conditions, and cannot provide accurate training scenarios for the formulation and regulation of load control strategies under extreme weather conditions, resulting in insufficient response capabilities of operation and maintenance personnel when real extreme weather occurs.
By acquiring historical power load and weather data for the target area, decomposing them into non-extreme and extreme weather data, a power load time series prediction model is constructed. Weighted superposition calculations are then performed using an extreme weather power load feature library. Combined with rigid and flexible adjustment commands, a load control simulation process is realized.
It achieves accurate reproduction of power load fluctuation characteristics under extreme weather conditions, provides accurate training scenarios, provides verification means for the formulation and regulation of load control strategies under extreme weather conditions, and improves the practical skills of operation and maintenance personnel.
Smart Images
Figure CN121642918A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of power load management simulation training, and in particular to a load control simulation method, system, device and medium for simulating an extreme weather power load scene. BACKGROUND
[0002] In the process of new power system construction, high-proportion intermittent and volatile uncontrollable units such as wind power and photovoltaic are connected to the power grid, and new loads such as electric vehicle charging piles and distributed energy storage emerge in large numbers, which significantly increases the uncertainty of the power supply side and the load side. The traditional power balance mode of "source following load" has been difficult to adapt to the operation requirements of the new power system. More seriously, under the background of global climate warming, extreme weather events occur frequently, such as continuous high temperature in summer, extreme cold wave in winter, and strong typhoon in coastal areas, which further aggravates the pressure of power supply. Due to the high risk and difficulty in replicating critical scenarios of power grid operation, there are great limitations in actual operation training, so the industry generally uses load control simulation training devices to carry out power worker post training to improve the employees' cognitive level and practical skills of load management process.
[0003] However, the mainstream load control simulation training device can simulate the strategy formulation, instruction issuance and rigid / soft adjustment functions when the load gap occurs under normal working conditions, but lacks the ability to accurately reproduce the load fluctuation characteristics under extreme weather conditions, and cannot simulate the load mutation rules caused by extreme weather, such as the surge of air conditioning load under high temperature weather and the sudden drop of industrial load under typhoon weather. Furthermore, it cannot provide accurate training scenarios for load control strategy formulation, regulation process deduction and regulation effect verification under extreme weather scenarios, which leads to insufficient response capability of operation and maintenance personnel when real extreme weather occurs, and poses a potential risk to the safe operation of the power grid. In the existing related technology, some schemes only focus on load prediction methods under abnormal weather conditions, only solve the problem of prediction accuracy, and do not involve simulation training scenario construction and regulation operation training. Traditional simulation devices focus on process simulation in normal scenarios and have not broken through the technical bottleneck of extreme weather scenario replication. SUMMARY
[0004] In view of the above existing problems, the present application is proposed.
[0005] Therefore, the present application provides a load control simulation method, system, device and medium for simulating an extreme weather power load scene, which solves the problems that the existing load control simulation technology cannot reproduce extreme weather load scenarios, cannot simulate the load mutation rules caused by extreme weather, and cannot provide accurate training scenarios for load control strategy formulation and regulation effect verification under extreme weather scenarios.
[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a load control simulation method for simulating an extreme weather power load scenario, comprising the following steps: obtaining historical power load data and historical weather data of a target area, and preprocessing the historical power load data; decomposing the preprocessed historical power load data into non-extreme weather power load data and extreme weather power load data according to a preset extreme weather determination condition; constructing a power load time series prediction model, using the power load time series prediction model to predict a basic power load value at a future time, and reading an extreme weather power load value of a corresponding type from an extreme weather power load feature library; superimposing and calculating the basic power load value and the extreme weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather; comparing the comprehensive predicted power load value with a dispatchable power consumption to determine whether a power gap is generated, and issuing an adjustment instruction according to the determination result.
[0007] As a preferred scheme of the load control simulation method for simulating an extreme weather power load scenario, the extreme weather determination condition comprises a temperature greater than a preset high temperature threshold, a temperature less than a preset low temperature threshold, or reaching a preset typhoon grade.
[0008] As a preferred scheme of the load control simulation method for simulating an extreme weather power load scenario, the step of decomposing the preprocessed historical power load data into non-extreme weather power load data and extreme weather power load data comprises: identifying an extreme weather occurrence period according to the historical weather data, extracting historical power load data in the period as extreme weather power load data, and storing the extreme weather power load data to an extreme weather power load feature library; and taking the historical power load data after the extreme weather period as non-extreme weather power load data, wherein data gaps generated by the extreme weather period are filled with power load data at the same time of adjacent dates.
[0009] The beneficial effects of the preferred technical scheme are: by identifying the extreme weather period to extract corresponding load data and store the load data to the feature library, and filling the missing values with load data at the same time of adjacent dates, the time sequence continuity of the non-extreme weather load data can be effectively maintained, and the prediction model training deviation caused by data missing can be avoided.
[0010] As a preferred scheme of the load control simulation method for simulating an extreme weather power load scenario, in the step of constructing a power load time sequence prediction model, the normalized power load value at time t is represented as Lt, the step length is set as N, Lt, Lt+1, Lt+2 to Lt+N-1 are taken as inputs, and Lt+N is taken as output to organize a training data set; the constructed training data set is randomly divided into a training set, a validation set and a test set, wherein the sample quantity of the training set accounts for more than 50% of the total sample quantity; the power load time sequence prediction model is trained by using the training set, and the performance of the power load time sequence prediction model is evaluated by using the validation set and the test set.
[0011] As a preferred scheme of the load control simulation method for simulating an extreme weather power load scenario, in the step of constructing a power load time sequence prediction model, the normalized power load value at time t is represented as Lt, the step length is set as N, Lt, Lt+1, Lt+2 to Lt+N-1 are taken as inputs, and Lt+N is taken as output to organize a training data set; the constructed training data set is randomly divided into a training set, a validation set and a test set, wherein the sample quantity of the training set accounts for more than 50% of the total sample quantity; the power load time sequence prediction model is trained by using the training set, and the performance of the power load time sequence prediction model is evaluated by using the validation set and the test set.
[0012] The preferred technical scheme has the beneficial effects that: the multiple sub-models of different time granularities are constructed to respectively predict the load values after 15 minutes, 30 minutes, 1 hour and 2 hours, which can meet the prediction requirements in different regulation and control scenarios and improve the time resolution and practicability of load prediction.
[0013] As a preferred scheme of the load control simulation method for simulating an extreme weather power load scenario, in the step of obtaining a comprehensive predicted power load value under future extreme weather, the steps include: According to the set extreme weather type and quantified value, the extreme weather power load value at the same time is read from the extreme weather power load characteristic library; the basic power load value is L0, and the extreme weather power load value is L1; the calculation formula of the comprehensive predicted power load value L is represented as: ; Wherein, a and b are greater than or equal to 0, the sum of a and b is equal to 1, and b is greater than a.
[0014] The beneficial effects of the preferred technical scheme are that the weighted superposition method is used to calculate the comprehensive predicted load value, and the weight of the extreme weather load is greater than that of the basic load, so that the actual influence degree of the extreme weather on the power load can be more accurately reflected, and the accuracy of the load prediction under the extreme weather scenario is improved.
[0015] As a preferred scheme of the load control simulation method for simulating the extreme weather power load scenario, the step of issuing the adjustment instruction according to the judgment result comprises: when the power gap is generated, the power gap is decomposed to the contracted users participating in orderly power utilization and demand response according to the load management scheme, and rigid adjustment instructions and flexible adjustment instructions are issued to the load control simulation device; the load control simulation device performs tripping power-off operation on the rigid adjustment load according to the rigid adjustment instruction, and performs reducing operation load operation on the flexible adjustment load according to the flexible adjustment instruction; the load control simulation device returns the rigid adjustment result and the flexible adjustment result to the master station, the rigid adjustment result comprises tripping execution state and execution time, and the flexible adjustment result comprises actual operation load and execution time; the master station calculates the load response rate according to the returned adjustment result and visually displays it.
[0016] The beneficial effects of the preferred technical scheme are that the rigid adjustment instruction and the flexible adjustment instruction are cooperatively issued, the adjustment result is returned, and the load response rate is calculated, so that a complete load regulation closed loop process is realized, and quantifiable training effect evaluation basis is provided for the operation and maintenance personnel.
[0017] In the second aspect, the application provides a load control simulation system for simulating an extreme weather power load scenario, comprising: a data processing module, which acquires historical power load data and historical weather data of a target region, pre-processes the historical power load data, and decomposes the pre-processed historical power load data into non-extreme weather power load data and extreme weather power load data according to a preset extreme weather judgment condition; a load prediction module, which predicts a basic power load value at a future time by using a power load time series prediction model, and reads an extreme weather power load value of a corresponding type from the extreme weather power load feature library; a load management module, which superimposes and calculates the basic power load value and the extreme weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather; a load control simulation device, which compares the comprehensive predicted power load value with a dispatchable power consumption, judges whether a power gap is generated, and issues an adjustment instruction according to a judgment result.
[0018] In the third aspect, the application provides an electronic device, comprising: a memory and a processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions, which realize the steps of the load control simulation method for simulating an extreme weather power load scenario.
[0019] In a fourth aspect, the present application provides a computer readable storage medium storing computer executable instructions, which realize the steps of the load control simulation method for simulating an extreme weather power load scenario when executed by a processor.
[0020] Compared with the prior art, the present application has the following beneficial effects: by decomposing historical power load data into non-extreme weather power load data and extreme weather power load data, respectively constructing a time series prediction model and an extreme weather power load feature library, and calculating a comprehensive predicted power load value by using a weighted superposition method, the present application realizes accurate reproduction of power load fluctuation characteristics under extreme weather, solves the technical defect that the existing load control simulation technology cannot simulate the load mutation law caused by extreme weather, and provides an accurate training scene for load control strategy formulation, regulation process deduction, and regulation effect verification under an extreme weather scenario.
[0021] In addition, by constructing load prediction sub-models of multiple time granularities, the present application can output load prediction results of different time scales to meet diversified regulation and control decision requirements; at the same time, by issuing rigid regulation and flexible regulation instructions in coordination and returning the regulation results in real time and evaluating the load response rate, a complete load regulation simulation closed loop is constructed, which provides a scene-realistic, process-complete, and effect-quantifiable training support for new power system load management post training, and effectively improves the practical operation ability of operation and maintenance personnel to cope with extreme weather. BRIEF DESCRIPTION OF DRAWINGS
[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.
[0023] Figure 1 The overall flowchart of the load control simulation method for simulating an extreme weather power load scenario according to an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION
[0024] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.
[0025] Embodiment 1, reference Figure 1 For an embodiment of the present application, a load control simulation method for simulating extreme weather power load scenarios is provided, comprising the following steps: S100, obtaining historical power load data and historical weather data of a target area, and preprocessing the historical power load data.
[0026] S200, according to a preset extreme weather judgment condition, decomposing the preprocessed historical power load data into non-extreme weather power load data and extreme weather power load data.
[0027] S300, constructing a power load time series prediction model, predicting a basic power load value at a future time by using the power load time series prediction model, and reading an extreme weather power load value of a corresponding type from the extreme weather power load feature library.
[0028] S400, superimposing and calculating the basic power load value and the extreme weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather.
[0029] S500, comparing the comprehensive predicted power load value with a dispatchable power consumption to determine whether a power gap is generated, and issuing an adjustment instruction according to the determination result.
[0030] It should be noted that in the process of new power system construction, intermittent and volatile uncontrollable units such as wind power and photovoltaic are highly connected to the power grid, and new loads such as electric vehicle charging piles and distributed energy storage emerge in large numbers, which significantly increases the uncertainty of the power supply side and the load side. Under the background of global warming, extreme weather events occur frequently, such as continuous high temperature in summer, extreme cold wave in winter, and strong typhoon in coastal areas, which further aggravates the pressure of power supply. Although the existing load control simulation training device can simulate the load adjustment function under normal working conditions, it lacks the ability to accurately reproduce the load fluctuation characteristics under extreme weather, and cannot simulate the load mutation rules caused by extreme weather, such as the surge of air conditioning load under high temperature weather and the sudden drop of industrial load under typhoon weather. At the same time, due to the difficulty of reproducing extreme weather scenarios in real power grids, operation and maintenance personnel lack targeted practical training opportunities, which leads to insufficient response capability when real extreme weather occurs, and potential risks to the safe operation of power grids.
[0031] Therefore, in order to solve the problem of extreme weather scenario reproduction and load control simulation training, through the steps of S100-S500, first, the historical power load data and historical weather data of the target area are acquired and preprocessed, and the load data is divided into two parts of non-extreme weather load and extreme weather load; then a power load time series prediction model is constructed to predict the basic load value, and the extreme weather load value in the extreme weather power load feature library is combined for weighted superposition to realize accurate reproduction of power load fluctuation characteristics under extreme weather; finally, the integrated predicted load value is compared with the dispatchable electricity consumption to judge the power gap, and adjustment instructions are issued according to the judgment result to realize a complete load control simulation process, and to provide accurate practical training support for load management personnel under extreme weather scenarios.
[0032] Embodiment 2, refer to Figure 1 For an embodiment of the present application, based on the above embodiment, a load control simulation method for simulating an extreme weather power load scenario is provided.
[0033] In the present application, S100, the historical power load data and historical weather data of the target area are acquired, and the historical power load data is preprocessed.
[0034] The preprocessing step of the historical power load data includes S1.1-S1.3: S1.1, the historical power load data of the target area is acquired from the power dispatching system, and the historical weather data of the corresponding period is acquired from the meteorological department.
[0035] Specifically, the power load refers to the active power of the user, and the power load referred to in the present application is the total load of all power users in the target area. The power load data is collected every 15 minutes, and the collection time is 0:00, 0:15, 0:30 every day, and so on, until 23:45, a total of 96 data points per day. The historical weather data includes meteorological parameters such as air temperature, wind speed, wind direction, and typhoon grade, which maintains a corresponding relationship with the power load data in the time dimension.
[0036] S1.2, the acquired historical power load data is processed for missing value filling and outlier removal.
[0037] Specifically, for the missing values in the historical power load data, linear interpolation or adjacent date and time load data is used for filling; for the outliers, 3σ criterion or box plot method based on statistics is used for identification, and the abnormal data points exceeding the normal range are removed or corrected to ensure that the data quality meets the subsequent model training requirements.
[0038] S1.3, standardize the historical power load data after the filling and elimination processing.
[0039] Specifically, the power load value at t time is represented as Lt, the load data is standardized by using a Z-score standardization method or a Min-Max normalization method, the influence of different dimensions and orders of magnitude is eliminated, the data distribution is more uniform, and the training and convergence of the subsequent time sequence prediction model are facilitated.
[0040] It should be noted that the extreme weather determination condition includes that the air temperature is greater than a preset high temperature threshold, the air temperature is less than a preset low temperature threshold, or a preset typhoon grade is reached.
[0041] Specifically, the extreme weather referred to in the application includes: extremely hot weather, i.e., the air temperature is greater than 35℃; extremely cold weather, i.e., the air temperature is less than -10℃; typhoon weather, according to the typhoon warning grade issued by the Central Meteorological Observatory. The above threshold values can be adjusted and set according to the climate characteristics and power consumption rules of the target region.
[0042] In an optional implementation, the preprocessing in step S100 can also include data smoothing processing, the load data is smoothed by using a moving average method or an exponential smoothing method, the influence of short-term random fluctuations is eliminated, the main trend characteristics of load variation are extracted, and a more stable data basis is provided for subsequent load decomposition and prediction model construction.
[0043] In another optional implementation, the historical weather data in step S100 can also be acquired in real time by calling a meteorological service API interface, and a local meteorological database is established to store and manage, automatic association and matching of meteorological data and power load data are realized, and the automation degree and timeliness of data acquisition are improved.
[0044] In the embodiment of the application, S200, according to a preset extreme weather determination condition, the preprocessed historical power load data is decomposed into non-extreme weather power load data and extreme weather power load data.
[0045] It should be noted that the extreme weather determination condition includes that the air temperature is greater than a preset high temperature threshold, the air temperature is less than a preset low temperature threshold, or a preset typhoon grade is reached.
[0046] Specific steps include A1-A2: A1, according to the historical weather data, identify the time period when the extreme weather occurs, extract the historical power load data in the time period as the extreme weather power load data, and store it in an extreme weather power load feature library.
[0047] Specifically, the historical weather data is compared with the preset extreme weather judgment condition to identify time periods that meet the extreme weather condition, including extremely hot time periods, extremely cold time periods, and typhoon time periods; the corresponding historical power load data in these time periods is extracted, classified and labeled according to the extreme weather type and quantitative value, and an extreme weather power load feature library is established. The feature library records the load fluctuation characteristics under different types of extreme weather, such as the electricity peak caused by the surge of air conditioning load under high temperature weather, and the electricity valley caused by the sudden drop of industrial load under typhoon weather, etc., providing data support for subsequent load prediction of extreme weather scenarios.
[0048] A2, the historical power load data after the extreme weather period is removed is used as non-extreme weather power load data, and the data gap caused by the extreme weather period is filled with the power load data at the same time of the adjacent date.
[0049] Specifically, the load data corresponding to the extreme weather period is removed from the preprocessed historical power load data, and the remaining data is used as non-extreme weather power load data for training the power load time series prediction model under normal working conditions. For the data gap caused by the removal of the extreme weather period, the non-extreme weather load data at the same time of the adjacent date (such as the previous day or the next day) is used to fill the data gap, so as to maintain the time sequence continuity and integrity of the load data and avoid the model training deviation caused by data loss.
[0050] In an optional embodiment, the extreme weather period identification in step S200 can also use a sliding window method, which sets a window of a certain length to scan the weather data. When a plurality of time points in the window continuously meet the extreme weather judgment condition, the time period is determined as an extreme weather period, avoiding misjudgment caused by short-term meteorological fluctuations and improving the accuracy and stability of the extreme weather period identification.
[0051] In another optional embodiment, the data gap filling in step S200 can also use a similar day filling method, that is, according to the date type (weekday / weekend / holiday), seasonal characteristics and other factors, similar non-extreme weather dates are selected from the historical data according to the characteristics of the missing date, and the load data at the same time of the similar day is used to fill the data gap, so that the filled data is more consistent with the actual electricity consumption rule.
[0052] In the embodiments of the present application, S300, a power load time series prediction model is constructed, the power load time series prediction model is used to predict the basic power load value at the future time, and the corresponding type of extreme weather power load value is read from the extreme weather power load feature library.
[0053] The step of constructing the power load time series prediction model includes B1-B3: B1, the normalized power load value at time t is represented by Lt, the step size is set to N, and the training data set is organized with Lt, Lt+1, Lt+2 to Lt+N-1 as input and Lt+N as output.
[0054] Specifically, the non-extreme weather power load data obtained in step S200 is organized into a training data set required for supervised learning in a sliding window manner. The step size N is set to the length of the historical load sequence, for example, N=96 indicates that the load data of the past 24 hours (one data point every 15 minutes) is used as input features.
[0055] Table 1: Organization form of training data set:
[0056] By traversing the entire non-extreme weather load data sequence through the sliding window, a large number of input-output sample pairs are generated to constitute a complete training data set.
[0057] B2, the constructed training data set is randomly divided into a training set, a validation set and a test set, wherein the number of samples in the training set accounts for more than 50% of the total number of samples.
[0058] Specifically, the training data set is divided into three non-overlapping subsets by random sampling, and each sample can only belong to one of the subsets. The training set is used for learning and optimization of model parameters, the validation set is used for optimization of model hyperparameters and implementation of early stopping strategy, and the test set is used for evaluation of final performance of the model. The sample proportion of the three subsets can be set to 6:2:2, i.e., the training set accounts for 60%, the validation set accounts for 20%, and the test set accounts for 20%, ensuring that the number of samples in the training set accounts for more than 50% of the total number of samples, and providing sufficient learning samples for the model.
[0059] B3, the training set is used to train the power load time series prediction model, and the validation set and the test set are used to evaluate the performance of the power load time series prediction model.
[0060] Specifically, a plurality of time series prediction models are trained based on the training set data, including autoregressive integrated moving average model (ARIMA, Autoregressive Integrated Moving Average Model), deep neural network long short-term memory network (LSTM, Long Short-Term Memory), Transformer, etc. The prediction performance of each model on the validation set and the test set is evaluated by using mean square error (MSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and other indicators, the fitting ability and generalization ability of each model are compared and analyzed, and the model with the best generalization performance is selected as the final power load time series prediction model.
[0061] It also needs to be known that the power load time series prediction model includes a plurality of sub-models of different time granularities, and the steps include B4-B6: B4, respectively, using data sequences with time intervals of 15 minutes, 30 minutes, 1 hour and 2 hours to construct training data sets of different time granularities.
[0062] Specifically, to meet the prediction needs under different regulation scenarios, the non-extreme weather power load data is resampled according to different time intervals. For data sequences with a time interval of 15 minutes, the original collected data is directly used; for data sequences with a time interval of 30 minutes, the adjacent two 15-minute data points are averaged or sampled; for data sequences with a time interval of 1 hour and 2 hours, resampling processing is performed accordingly. Based on data sequences of different time intervals, training data sets of corresponding time granularities are respectively organized and constructed according to the manner of step B1.
[0063] B5, for each time granularity training data set, respectively train each basic model, compare and analyze the performance of each basic model, and select the basic model with the best generalization performance as the sub-model corresponding to the time granularity.
[0064] Specifically, for the training data sets of 15 minutes, 30 minutes, 1 hour and 2 hours, a plurality of basic models such as ARIMA, LSTM and Transformer are trained. Because data of different time granularities have different time series characteristics and fluctuation rules, the performance of each basic model under different granularities may differ. Through performance comparison and analysis on the validation set and the test set, the basic model with the best generalization performance is selected for each time granularity as the sub-model corresponding to the granularity, and a power load time series prediction model system of multiple time granularities is constructed.
[0065] B6, in prediction, input the sequence of N power load values before the current time after standardization, use the sub-models of each time granularity to predict the basic power load values after 15 minutes, 30 minutes, 1 hour and 2 hours, and perform reverse standardization on the prediction results.
[0066] Specifically, in the actual prediction phase, first, the real-time load data of the N time points before the current time is obtained and processed according to the same standardization method in the training phase; then the standardized load sequence is input into the sub-models corresponding to the time granularities of 15 minutes, 30 minutes, 1 hour and 2 hours, respectively, to obtain the standardized prediction results of each time scale; finally, the prediction results are reverse standardized to convert them into actual load values, and the basic power load prediction values after different times in the future are obtained.
[0067] Further, according to the set extreme weather type and quantification value, the extreme weather power load value corresponding to the same time is read from the extreme weather power load characteristic library.
[0068] Specifically, according to the extreme weather information at the future time obtained from the weather forecast, or according to the extreme weather type and quantification value set artificially (such as extremely hot weather with a temperature of 38 degrees Celsius, extremely cold weather with a temperature of-15 degrees Celsius, and a 10-level typhoon), the historical extreme weather period matching the same or similar conditions is queried in the extreme weather power load characteristic library, and the extreme weather power load value at the same time of the period is read as the input of the subsequent load superposition calculation.
[0069] In an optional embodiment, the model training in step B3 can also use the method of ensemble learning to weight and integrate or vote the prediction results of multiple basic models to construct an ensemble prediction model. By complementing the advantages of different models, the stability and accuracy of the prediction results are improved, and the risk of overfitting or underfitting of a single model is reduced.
[0070] In another optional embodiment, the model selection in step B5 can also use the automatic machine learning (AutoML) technology to automatically search for the optimal model structure and hyperparameter combination. Through methods such as Bayesian optimization, grid search, or random search, the optimal model configuration under each time granularity is automatically found in the preset search space, reducing the workload of manual parameter tuning and improving the efficiency and scientificity of model selection.
[0071] In the embodiments of the present application, S400, the basic power load value and the extreme weather power load value are superposed and calculated according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather.
[0072] The step of obtaining a comprehensive predicted power load value under future extreme weather includes C1-C2: C1, according to the set extreme weather type and quantification value, the extreme weather power load value corresponding to the same time is read from the extreme weather power load characteristic library.
[0073] Specifically, the extreme weather condition to be simulated is determined according to the extreme weather warning information obtained according to the weather forecast or the extreme weather type and quantitative value manually set according to the training scene. The extreme weather type includes extreme heat, extreme cold, typhoon, etc., and the quantitative value includes a specific air temperature value or a typhoon grade. In the extreme weather power load feature library, the load data under the same or similar extreme weather condition in history is matched and read according to the extreme weather type and quantitative value to obtain the extreme weather power load value corresponding to the prediction time. For example, if the extreme heat weather with an air temperature of 38 degrees Celsius at a future time is set, the extreme heat period with an air temperature in the range of 37-39 degrees Celsius in history is queried from the feature library, and the load data at the same time of the period is read as the extreme weather power load value.
[0074] C2, the base power load value is L0, and the extreme weather power load value is L1; The calculation formula of the comprehensive predicted power load value L is represented as: L = aL0 + bL1; wherein a and b are greater than or equal to 0, the sum of a and b is equal to 1, and b is greater than a.
[0075] Specifically, the base power load value predicted by the power load time sequence prediction model in step S300 is denoted as L0, and the extreme weather power load value read from the extreme weather power load feature library in step C1 is denoted as L1. The comprehensive predicted power load value L is calculated by using a weighted superposition method, and the calculation formula is L = aL0 + bL1. Wherein, the weight coefficients a and b are non-negative numbers, and satisfy the constraint condition a + b = 1 to ensure the rationality of the comprehensive predicted value. Since the load fluctuation is mainly affected by the extreme weather in the extreme weather scene, the weight b of the extreme weather power load value is set to be greater than the weight a of the base power load value, so as to more accurately reflect the actual influence degree of the extreme weather on the power load. For example, a = 0.3 and b = 0.7 can be set, which means that the extreme weather load contributes 70% and the base load contributes 30% in the comprehensive predicted load. The specific value of the weight coefficient can be adjusted and optimized according to the power consumption characteristics of the target area and the influence degree of the extreme weather.
[0076] In an optional embodiment, the weight coefficient in step C2 can also be dynamically adjusted, and the weight distribution is adaptively adjusted according to the severity of the extreme weather. When the extreme weather quantitative value is more extreme (such as higher air temperature or higher typhoon grade), the extreme weather load weight b is correspondingly increased; when the extreme weather quantitative value approaches the threshold boundary, the weight b is correspondingly reduced, so as to dynamically match the weight coefficient with the severity of the extreme weather and improve the accuracy of the comprehensive prediction.
[0077] In the embodiment of the present application, S500, the comprehensive predicted power load value is compared with the dispatchable power consumption to determine whether a power gap is generated, and a regulation instruction is issued according to the determination result.
[0078] Specifically, the comprehensive predicted power load value under future extreme weather obtained in step S400 is compared with the dispatchable power consumption of the current power grid. The dispatchable power consumption refers to the maximum power supply capability that the power grid can provide under the current operating state. If the comprehensive predicted power load value is greater than the dispatchable power consumption, it is determined that a power gap is generated, and the power gap value is equal to the difference between the comprehensive predicted power load value and the dispatchable power consumption. If the comprehensive predicted power load value is less than or equal to the dispatchable power consumption, it is determined that no power gap is generated, and the power supply capability of the power grid can meet the load demand.
[0079] Among them, the step of issuing a regulation instruction according to the determination result includes D1~D4: D1, when a power gap is generated, the power gap is decomposed to the contracted users participating in orderly power utilization and demand response according to the load management scheme, and rigid regulation instructions and flexible regulation instructions are issued to the load control simulation device.
[0080] Specifically, when it is determined that a power gap is generated, the load management process is started by the negative control master station, and the power gap is decomposed and distributed to the users who have signed up to participate in load management according to the preset orderly power utilization scheme and demand response scheme. The orderly power utilization scheme is for rigid load users such as industrial production, and realizes load reduction through round robin power limiting; the demand response scheme is for flexible load users such as air conditioners and lighting, and realizes load regulation through incentive guidance. The negative control master station distributes the regulation tasks of a number of orderly power utilization users (for example, 2 households) and a number of demand response users (for example, 4 households) to the load control simulation device, and issues rigid regulation instructions and flexible regulation instructions to the load control terminal according to the preset process. The rigid regulation instruction contains parameters such as target user identification, tripping time, and power-off duration; the flexible regulation instruction contains parameters such as target user identification, load reduction amplitude, and regulation period.
[0081] D2, the load control simulation device performs tripping and power-off operation on the rigid regulation load according to the rigid regulation instruction, and performs reducing operation on the flexible regulation load according to the flexible regulation instruction.
[0082] Specifically, after receiving adjustment commands from the master station, the load control terminal in the load control simulation device executes corresponding load adjustment operations. For rigid adjustment commands, the load control terminal controls the simulated high-voltage circuit breaker or molded case circuit breaker to trip and disconnect the power supply to the rigidly regulated load (such as industrial production equipment), forcibly cutting off the load power and achieving immediate load reduction. For flexible adjustment commands, the load control terminal sends power adjustment signals to the flexiblely regulated load (such as air conditioning and lighting equipment) through the intelligent control interface, reducing the equipment's operating power to the level required by the command, achieving flexible load reduction while ensuring basic usage needs are met.
[0083] D3. The load control simulation device transmits the rigid adjustment results and flexible adjustment results back to the main station. The rigid adjustment results include the trip execution status and execution time, and the flexible adjustment results include the actual operating load and execution time.
[0084] After performing adjustment operations, the load control terminal transmits the adjustment results back to the load control master station via communication equipment. Rigid adjustment results include: trip execution status (success / failure), trip execution time, current circuit breaker status, and power outage duration. Flexible adjustment results include: actual operating load value, load adjustment range, adjustment execution time, and equipment operating status. The master station receives and stores the adjustment result data transmitted from each load control terminal for subsequent adjustment effect evaluation and training assessment.
[0085] D4. The main station calculates the load response rate based on the feedback adjustment results and displays it visually.
[0086] Specifically, the load control master station calculates the load response rate for each user and the overall load based on the returned adjustment results data, and evaluates the load adjustment effect. The formula for calculating the load response rate is: Load response rate = (baseline load - actual load) / baseline load × 100%; The formula for calculating the baseline load is as follows: Baseline load = (1 / 5) × Σ (load at the same time for the 5 working days before demand response when there is no demand response). The baseline load is the average load at the same time during the five working days prior to the demand response period. The load response rate reflects the ratio of the actual load reduction by users to the baseline load; a higher response rate indicates better load regulation. The main station displays data such as load response rate, regulation execution status, and load curve changes through a visualization interface, including data reports, trend curves, bar charts, and other formats, providing intuitive data support for training assessments and effectiveness evaluations.
[0087] In an optional embodiment, the power gap decomposition in step D1 can also be intelligently allocated by using an optimization algorithm to minimize the impact on user power consumption or maximize the adjustment economy as the objective function, taking into account factors such as user load capacity, response capability, compensation price, etc., and solving the optimal gap allocation scheme through linear programming or genetic algorithm to achieve optimal allocation of load adjustment resources.
[0088] In another optional embodiment, the visualization in step D4 can also integrate a geographic information system (GIS) to display the geographic location, adjustment state and response effect of each adjustment user on an electronic map to achieve spatial visualization management of load adjustment. At the same time, it supports historical data playback function, which can analyze the training process to help operation and maintenance personnel summarize experience and improve skills.
[0089] In summary, by decomposing historical power load data into non-extreme weather power load data and extreme weather power load data, respectively constructing time series prediction model and extreme weather power load feature library, and calculating the comprehensive predicted power load value by using weighted superposition method, the precise reproduction of power load fluctuation characteristics under extreme weather is realized, and the technical defect that the existing load control simulation technology cannot simulate the load mutation law caused by extreme weather is solved, which provides precise training scene for load control strategy making, regulation process deduction and regulation effect verification under extreme weather scene.
[0090] In addition, by constructing load prediction sub-models with multiple time granularities, the application can output load prediction results of different time scales to meet diversified regulation and control decision-making needs; at the same time, through the collaborative issuance of rigid regulation and flexible regulation instructions and the real-time feedback of regulation results and load response rate evaluation, a complete load regulation simulation closed loop is constructed, which provides scene real, process complete and effect quantifiable training support for new power system load management post training, and effectively improves the practical operation ability of operation and maintenance personnel to cope with extreme weather.
[0091] Embodiment 3, the above is a schematic scheme of a load control simulation method for simulating an extreme weather power load scene. It should be noted that the technical scheme of the load control simulation system for simulating the extreme weather power load scene belongs to the same concept as the technical scheme of the load control simulation method for simulating the extreme weather power load scene described above. The technical scheme of the load control simulation system for simulating the extreme weather power load scene in this embodiment is not described in detail, and the details can be referred to the description of the technical scheme of the load control simulation method for simulating the extreme weather power load scene.
[0092] The embodiment also provides a load control simulation system for simulating an extreme weather power load scene, comprising: The data processing module acquires historical power load data and historical weather data of a target area, pre-processes the historical power load data, and decomposes the pre-processed historical power load data into non-extreme-weather power load data and extreme-weather power load data according to a preset extreme-weather judgment condition. The load prediction module predicts a basic power load value at a future time by using a power load time series prediction model, and reads an extreme-weather power load value of a corresponding type from the extreme-weather power load feature library. The load management module superimposes the basic power load value and the extreme-weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather. The load control simulation device compares the comprehensive predicted power load value with a dispatchable power consumption, judges whether a power gap is generated, and issues an adjustment instruction according to a judgment result.
[0093] The embodiment also provides an electronic device suitable for load control simulation of a simulated extreme-weather power load scene, which comprises a memory and a processor.
[0094] The embodiment also provides a storage medium having a computer program stored thereon, and the program is executed by a processor to implement the load control simulation method of the simulated extreme-weather power load scene.
[0095] The storage medium provided by the embodiment belongs to the same inventive concept as the load control simulation method of the simulated extreme-weather power load scene provided by the above embodiment, and the technical details not described in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.
[0096] Through the above description of the embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a FLASH, a hard disk, or an optical disc, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present application.
[0097] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A load control simulation method for simulating an extreme weather power load scenario, characterized in that, The method comprises the following steps: obtaining historical power load data and historical weather data of a target area, and preprocessing the historical power load data; decomposing the preprocessed historical power load data into non-extreme weather power load data and extreme weather power load data according to preset extreme weather judgment conditions; constructing a power load time series prediction model, predicting a basic power load value at a future time by using the power load time series prediction model, and reading an extreme weather power load value of a corresponding type from the extreme weather power load feature library; superimposing and calculating the basic power load value and the extreme weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather; comparing the comprehensive predicted power load value with a dispatchable power consumption to determine whether a power gap is generated, and issuing an adjustment instruction according to a determination result.
2. The load control simulation method of simulating an extreme weather power load scenario as claimed in claim 1, characterized in that, The extreme weather judgment conditions include that the air temperature is greater than a preset high temperature threshold, the air temperature is less than a preset low temperature threshold, or a preset typhoon grade is reached.
3. The load control simulation method of simulating extreme weather power load scenarios as claimed in claim 2, wherein, The step of decomposing the preprocessed historical power load data into non-extreme weather power load data and extreme weather power load data comprises: identifying an extreme weather occurrence period according to the historical weather data, extracting historical power load data in the period as extreme weather power load data, and storing the extreme weather power load data to an extreme weather power load feature library; regarding the historical power load data after the extreme weather period is removed as non-extreme weather power load data, and filling data gaps generated in the extreme weather period with power load data at the same time on adjacent dates.
4. The load control simulation method of simulating an extreme weather power load scenario as claimed in claim 3, characterized in that, The step of constructing the power load time series prediction model comprises: denoting a standardized power load value at t time as Lt, setting a step length as N, taking Lt, Lt+1, Lt+2 to Lt+N-1 as inputs and taking Lt+N as an output to organize a training data set; randomly splitting the constructed training data set into a training set, a validation set and a test set, wherein the sample quantity of the training set accounts for more than 50% of the total sample quantity; training the power load time series prediction model by using the training set, and performing performance evaluation on the power load time series prediction model by using the validation set and the test set.
5. The load control simulation method of simulating an extreme weather power load scenario as claimed in claim 4, characterized in that, The power load time series prediction model comprises a plurality of sub-models of different time granularities, and the step comprises: constructing training data sets of different time granularities by using data sequences with time intervals of 15 minutes, 30 minutes, 1 hour and 2 hours respectively; training each basic model for each time granularity training data set, comparing and analyzing the performance of each basic model, and selecting a basic model with the best generalization performance as a sub-model corresponding to the time granularity; in prediction, inputting a sequence of N standardized power load values before a current time, predicting basic power load values after 15 minutes, 30 minutes, 1 hour and 2 hours by using sub-models of each time granularity respectively, and performing reverse standardization processing on the prediction results.
6. The load control simulation method of simulating an extreme weather power load scenario as claimed in claim 5, characterized in that, The step of obtaining a comprehensive predicted power load value under future extreme weather comprises: According to the set extreme weather type and quantitative value, an extreme weather power load value corresponding to the same time is read from the extreme weather power load characteristic library; Supposing the basic power load value is L0 and the extreme weather power load value is L1; The calculation formula of the comprehensive predicted power load value L is represented as: ； Wherein, a and b are greater than or equal to 0, the sum of a and b is equal to 1, and b is greater than a.
7. The load control simulation method of simulating extreme weather power load scenarios as claimed in claim 6, wherein, The step of issuing the adjustment instruction according to the judgment result comprises: When the power gap is generated, the power gap is decomposed to the contracted users participating in the orderly power utilization and demand response according to the load management scheme, and rigid adjustment instructions and flexible adjustment instructions are issued to the load control simulation device; The load control simulation device performs tripping power-off operation on the rigid adjustment load according to the rigid adjustment instruction and performs reducing operation load operation on the flexible adjustment load according to the flexible adjustment instruction; The load control simulation device returns the rigid adjustment result and the flexible adjustment result to the master station, the rigid adjustment result comprises tripping execution state and execution time, and the flexible adjustment result comprises actual operation load and execution time; The master station calculates the load response rate according to the returned adjustment result and performs visual display.
8. A load control simulation system for simulating extreme weather power load scenarios, applying the method according to any one of claims 1 to 7, characterized in that Comprise: The data processing module acquires historical power load data and historical weather data of a target area, pre-processes the historical power load data, and decomposes the pre-processed historical power load data into non-extreme weather power load data and extreme weather power load data according to a preset extreme weather judgment condition; The load prediction module predicts a basic power load value at a future time by using a power load time series prediction model, and reads an extreme weather power load value of a corresponding type from the extreme weather power load characteristic library; The load management module superimposes and calculates the basic power load value and the extreme weather power load value according to a preset proportion to obtain a comprehensive predicted power load value under future extreme weather; The load control simulation device compares the comprehensive predicted power load value with a dispatchable power utilization amount, judges whether a power gap is generated, and issues an adjustment instruction according to the judgment result. 9.An electronic device comprising: a memory and a processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions, which realize the steps of the load control simulation method for simulating an extreme weather power load scene according to any one of claims 1 to 7. 10.A computer readable storage medium storing computer executable instructions, which realize the steps of the load control simulation method for simulating an extreme weather power load scene according to any one of claims 1 to 7 when executed by a processor.