Power market power parameter auxiliary decision-making method and device based on multi-model fusion

Abstract

This application relates to a method and apparatus for auxiliary decision-making of power parameters in the power market based on multi-model fusion. The main steps include: training a neural network classification model using a feature dataset to predict feature data for a target day; establishing multiple machine learning algorithm models to obtain the optimal adjustment ratio using the feature data and adjustment ratio as loss functions, and selecting the optimal model for each; obtaining multiple adjustment ratios for the target day based on the optimal model; and performing a co-directional judgment between the multiple adjustment ratios for the target day and the feature data to select the final adjustment ratio. This method integrates multiple machine learning models using artificial intelligence technology to seek the intrinsic relationship between different reporting strategies and power generation benefits, optimize power trading decisions, flexibly respond to complex and volatile market environments, determine the best operating strategy, and improve the reporting efficiency and accuracy of the power market.

Description

Technical Field

[0001] This application relates to the field of reinforcement learning model application technology in electricity market transactions, specifically to a method and device for auxiliary decision-making of electricity market parameters based on multi-model fusion. Background Technology

[0002] Currently, according to the rules of the electricity spot market, new energy power generation companies participating in electricity spot trading are generally required to submit day-ahead declarations as required by market rules. This means that market participants submit their power generation and consumption plans, bids, and other information for the operating day to the electricity trading institution or dispatching agency one day in advance (generally the day before the operating day). This provides market participants with an opportunity to plan ahead and participate in market competition, helping to improve the efficiency of electricity market declarations and the rationality of resource allocation.

[0003] Currently, most new energy power generation companies use three main methods for their day-ahead reporting: the first is to directly use the raw power forecast data from the wind power forecasting system as the market reporting plan; the second is to manually adjust the raw power forecast data to form the market reporting plan; and the third is to use software systems to generate the market reporting plan. However, the first method relies entirely on the raw power forecast data, failing to comprehensively consider various factors such as medium- and long-term contract conditions, market supply and demand, and market assessment rules. Furthermore, due to significant deviations in the power forecast results, the reporting plan lacks comprehensive consideration of factors and in-depth data analysis and mining, resulting in an insufficiently comprehensive market reporting plan. The second method, although it makes some adjustments to the raw power forecast data, relies excessively on individual abilities and experience, without a clear quantitative processing of the intermediate steps. In the actual market reporting plan, some factors are subject to human randomness, and the process is inefficient. The third approach, while utilizing information technology, is based on historical transaction data. It employs statistical analysis to analyze price trends and performs error analysis between short-term power forecasts and actual power generation. The error results are used as the basis for short-term power adjustments to generate market declaration plans. This approach only uses conventional mathematical statistics methods, has simple logic, does not fully consider data correlation, and lacks in-depth analysis and mining of the full dataset. This results in low accuracy of the generated market declaration plans, making it difficult to maximize the reduction of power generation costs and improve power generation efficiency. It is also detrimental to improving the efficiency and accuracy of power market declarations. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, this application provides a method and apparatus for auxiliary decision-making based on power market parameters using multi-model fusion, specifically adopting the following technical solution:

[0005] A power market parameter-assisted decision-making method based on multi-model fusion includes the following steps:

[0006] Step 1: Obtain sample data of the electricity market corresponding to the collection time t on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least the day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, clearing power data, short-term power data, adjustment ratio data, and environmental parameters; where d = 1, 2, ..., D, and D is the upper limit of the number of days;

[0007] Step 2: Analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range;

[0008] Step 3: Calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and process the difference between the day-ahead price and the real-time price into the original feature dataset through hot coding. Then, expand the original feature dataset by window sliding processing to obtain the expanded feature dataset.

[0009] Step 4: Train the neural network classification model using the expanded feature dataset, and predict the feature data for the target day based on the trained neural network classification model;

[0010] Step 5: Establish multiple machine learning algorithm models to obtain the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and train and validate different machine learning algorithm models respectively. Select the model with the most data collection times when the adjustment ratio is consistent with the direction of feature data as the optimal model of the corresponding machine learning algorithm model; the machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm.

[0011] Step Six: Input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day;

[0012] Step 7: Compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day to determine the final adjustment ratio.

[0013] Optionally: When obtaining sample data of the electricity market in step one, the data is collected in days and each day is divided into several equally spaced collection times, and sample data is obtained for each collection time.

[0014] Optional: In step one, a data normalization method is used to preprocess the sample dataset.

[0015] First, obtain the maximum and minimum values ​​of different sample data in the sample dataset, and then classify the sample data according to different times of the same day to obtain the sample units for day d.

[0016] The difference between the sample data at time t is obtained based on the maximum and minimum values ​​of the corresponding sample data.

[0017] Based on the sample unit of day d, obtain the sample data at time t in day d, and obtain the normalized sample data corresponding to time t in day d based on the difference between the sample data at time t in day d and the sample data corresponding to time t in day d.

[0018] Optionally: Step two, which involves analyzing existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range, includes:

[0019] Convert the cleared power data corresponding to the collection time t on day d in the sample dataset into power system output data;

[0020] Calculate the adjustment ratio between the power system output data and the short-term power data at the time of data collection on day d at time t.

[0021] The maximum and minimum values ​​of the adjustment ratio are obtained by statistical analysis of the sample dataset using statistical methods, and the maximum and minimum values ​​of the adjustment ratio are used as the range of the adjustment ratio.

[0022] The probability distribution of the adjustment ratio for each collection time in the historical days is calculated using a probability density curve function.

[0023] Optionally: Step three, which involves processing the difference between the day-to-day price and the real-time price into feature data using a hot coding method, includes:

[0024] Obtain the day-ahead price and real-time price corresponding to the data collection time t on day d;

[0025] Calculate the difference between the day-ahead price and the real-time price at time t on day d;

[0026] The difference between the previous day's price and the real-time price at time t on day d is compared with the threshold 0 to determine the direction of the price difference: when the difference between the previous day's price and the real-time price is less than 0, the direction of the price difference is -1; when the difference between the previous day's price and the real-time price is equal to 0, the direction of the price difference is 0; when the difference between the previous day's price and the real-time price is greater than 0, the direction of the price difference is 1.

[0027] The obtained price difference directions are summarized to obtain characteristic data.

[0028] Optionally, step three, which involves expanding the original feature dataset using a window sliding process to obtain an expanded feature dataset, includes:

[0029] Set the window width W and sliding step size ΔW for the time window, and arrange the feature data of the original feature dataset in chronological order to obtain the sorted original feature dataset;

[0030] The starting position of the sorted original feature dataset is shifted one sliding step ΔW to the right according to the window width W of the time window, and the feature data at the time window position is set to 0 to obtain new feature data.

[0031] Optionally: Step four, which involves training the neural network classification model using the augmented feature dataset, includes:

[0032] First, the expanded feature dataset is classified by day to obtain different feature units;

[0033] Secondly, a density-based clustering algorithm is used to cluster different feature units of the expanded feature dataset, and the feature units containing cluster centers similar to the target date are selected; the following method is used to determine similarity to the target date:

[0034]

[0035] where X di The feature dataset for the target day; X d This is the feature dataset for day d; dist(X) di ,X d ) is the Euclidean distance calculation function; X' di,t X' represents the normalized feature data collected at time t on the target day; d,t The normalized feature data is the data collected at time t on day d; T is the total number of collection times.

[0036] The feature data of the target day is sorted by Euclidean distance with different feature units, and the feature data corresponding to the feature units that meet the preset distance are selected as training data.

[0037] Finally, the obtained training data is used to train the preset neural network classification model to obtain the trained neural network classification model.

[0038] Optionally: Step five, which selects the model with the highest number of data acquisition moments when the adjustment ratio is consistent with the feature data direction, as the optimal model for the corresponding machine learning algorithm, includes:

[0039] First, a training set is selected based on the expanded feature dataset, and different machine learning algorithm models are trained on the training set for a preset number of times. After each training is completed, the corresponding machine learning algorithm model is saved.

[0040] For the same type of machine learning algorithm model, a pre-set validation set is input into the machine learning algorithm model saved after different training iterations, and the adjustment ratio and adjustment direction of the output are obtained respectively; the relationship between the adjustment ratio and the adjustment direction is as follows:

[0041]

[0042] Then, it is determined whether the adjustment direction output by the machine learning algorithm model in each training session is consistent with the direction of the feature data in the validation set:

[0043]

[0044] where R d,t R_diff represents the adjustment ratio for the data collection time t on day d; d,t The adjustment direction for the data collection time t on day d; RP_diff d,t This indicates whether the adjustment direction at time t on day d is consistent with the direction of the feature data; Price_diff d,t This indicates the direction of the feature data collected at time t on day d;

[0045] Finally, the number of data collection moments in each training session where the adjustment direction of the machine learning algorithm model output is consistent with the direction of the feature data in the validation set is counted, and the model with the most corresponding data collection moments is selected as the optimal model for the corresponding machine learning algorithm model.

[0046] Optionally: Step seven, which involves making a judgment in the same direction between the obtained adjustment ratios for the target day and the characteristic data of the target day, and selecting the final adjustment ratio, includes:

[0047] The optimal models based on different machine learning algorithms respectively obtain the adjustment ratio and adjustment direction for the target day, where the relationship between the adjustment ratio and adjustment direction is as follows:

[0048]

[0049] Count the number of different machine learning algorithm models in different adjustment directions at the same data acquisition time:

[0050]

[0051] where R i,t R_diff is the adjustment ratio of the output of the i-th machine learning algorithm model at the t-th acquisition time; i,t Output the adjustment direction at time t for the i-th machine learning algorithm model; I is the number of different machine learning algorithm models; RNpos t RNNeg represents the number of samples collected at time t that are in the positive direction. tRNorig represents the number of samples collected at time t in the negative direction. t The quantity that is not adjusted at time t;

[0052] Select the adjustment direction with the most occurrences at the current acquisition time and compare it with the characteristic data direction at the same acquisition time on the target day: if the adjustment direction selected at the current acquisition time is consistent with the characteristic data direction on the target day, calculate the average adjustment ratio corresponding to the adjustment direction selected at the current acquisition time as the final adjustment ratio; if the adjustment direction selected at the current acquisition time is inconsistent with the characteristic data direction on the target day, no adjustment is made at the current acquisition time.

[0053] Furthermore, this application also discloses a power market power parameter auxiliary decision-making device based on multi-model fusion, the device comprising:

[0054] The sample data acquisition module is used to acquire sample data of the electricity market corresponding to the collection time t on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, clearing power data, short-term power data, adjustment ratio data, and environmental parameters; where d = 1, 2, ..., D, and D is the upper limit of the number of days;

[0055] The adjustment range determination module is used to analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range;

[0056] The feature data augmentation module is used to calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and to process the difference between the day-ahead price and the real-time price into the original feature dataset through hot encoding. Then, the original feature dataset is augmented by window sliding processing to obtain the augmented feature dataset.

[0057] The target feature prediction module is used to train the neural network classification model using an expanded feature dataset, and to predict the feature data of the target day based on the trained neural network classification model.

[0058] The optimal model selection module is used to establish multiple machine learning algorithm models for obtaining the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and to train and validate different machine learning algorithm models respectively. The model with the most data collection times when the adjustment ratio is consistent with the direction of feature data is selected as the optimal model of the corresponding machine learning algorithm model. The machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm.

[0059] The adjustment ratio output module is used to input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day;

[0060] The adjustment ratio determination module is used to compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day and select the final adjustment ratio.

[0061] Beneficial effects

[0062] The technical solution of this application achieves the following beneficial effects:

[0063] The electricity market parameter auxiliary decision-making method of this application integrates multiple machine learning models according to certain rules using artificial intelligence technology, thereby seeking the intrinsic relationship between different reporting strategies and power generation benefits. It can optimize multi-dimensional and multi-period electricity trading decisions, flexibly respond to complex and ever-changing market environments, quantify day-ahead reporting issues, eliminate the influence of human factors, and further explore the correlation of data. The constructed day-ahead reporting auxiliary decision-making model can determine the optimal operating strategy to improve the reporting efficiency and accuracy of the electricity market. Attached Figure Description

[0064] Figure 1 This is a flowchart of the power market power parameter auxiliary decision-making method in the embodiments of this application.

[0065] Figure 2 This is a schematic diagram of the direction of feature data of the target day predicted by the neural network classification model in the embodiments of this application.

[0066] Figure 3 This is a schematic diagram showing the target day adjustment ratio obtained by different machine learning algorithm models in the embodiments of this application.

[0067] Figure 4 This is a schematic diagram of the final adjustment ratio of the target date obtained in the embodiments of this application.

[0068] Figure 5 This is a structural diagram of the power market power parameter auxiliary decision-making device in the embodiments of this application.

[0069] Figure 6 This is a structural diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0070] The present application will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present application and should not be construed as limiting the scope of protection of the present application. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present application.

[0071] Combination Figure 1 As shown in the illustration, this application specifically discloses a power market power parameter auxiliary decision-making method based on multi-model fusion, which includes the following steps:

[0072] Step 1: Obtain sample data of the electricity market corresponding to the collection time t on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least the day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, clearing power data, short-term power data, adjustment ratio data, and environmental parameters; where d = 1, 2, ..., D, and D is the upper limit of the number of days;

[0073] Step 2: Analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range;

[0074] Step 3: Calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and process the difference between the day-ahead price and the real-time price into the original feature dataset through hot coding. Then, expand the original feature dataset by window sliding processing to obtain the expanded feature dataset.

[0075] Step 4: Train the neural network classification model using the expanded feature dataset, and predict the feature data for the target day based on the trained neural network classification model;

[0076] Step 5: Establish multiple machine learning algorithm models to obtain the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and train and validate different machine learning algorithm models respectively. Select the model with the most data collection times when the adjustment ratio is consistent with the direction of feature data as the optimal model of the corresponding machine learning algorithm model; the machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm.

[0077] Step Six: Input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day;

[0078] Step 7: Compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day to determine the final adjustment ratio.

[0079] Specifically, in step one of this application embodiment, when obtaining sample data of the electricity market, the data is collected on a daily basis and each day is divided into several equally spaced collection times. Sample data for each collection time is obtained separately. Preferably, this embodiment uses a daily basis with collection times divided at 15-minute intervals to collect and organize the data, thereby obtaining data from 96 collection times per day.

[0080] Furthermore, since the collected raw sample data are based on different data units and contain some dirty data (such as inaccurate, incomplete, inconsistent, or non-standard data), data cleaning is necessary. In step one of this embodiment, a data normalization method is used to preprocess the sample dataset to avoid bias caused by differences in data levels affecting the model's generalization ability. By standardizing the data to a fixed range, such as [0,1] or [-1,1], the model training speed is accelerated and the prediction accuracy is improved. The specific preprocessing steps are as follows:

[0081] First, obtain the maximum and minimum values ​​of different sample data in the sample dataset, and then classify the sample data according to different times of the same day to obtain the sample unit for day d. It should be noted that the maximum and minimum values ​​obtained are selected within the same sample data range in the sample dataset. For example, when obtaining the maximum and minimum values ​​of a certain sample data, you can obtain the same type of sample data in the historical days, and then filter out the maximum and minimum values ​​based on the corresponding sample data in the historical days.

[0082] The difference between the sample data at time t is obtained based on the maximum and minimum values ​​of the corresponding sample data.

[0083] Based on the sample units of day d, obtain the sample data at time t in day d, and obtain the normalized sample data corresponding to time t in day d based on the difference between the sample data at time t in day d and the sample data corresponding to time t in day d:

[0084]

[0085] Where X' d,t X represents the normalized sample data collected at time t on day d; d,t This refers to the sample data collected at time t on day d. The minimum value of the sample data at time t in the sample dataset; It represents the maximum value of the sample data at time t in the sample dataset.

[0086] More specifically, in step two of this application embodiment, the adjustment ratio range is obtained by analyzing the existing adjustment ratio data in the sample dataset. The adjustment ratio is a crucial decision-making factor in the electricity market, requiring market participants to comprehensively consider factors such as market price fluctuations, changes in supply and demand, their own costs and benefits, and policy and regulatory changes, in order to maximize economic benefits and ensure stable operation of the power system. In this embodiment, the allowable range of the adjustment ratio is determined by analyzing the collected historical sample data, and adjustments are then made within this range. The specific steps include:

[0087] First, the cleared power data corresponding to the collection time t on day d in the sample dataset is converted into power system output data;

[0088] Subsequently, the adjustment ratios for the power system output data and short-term power data corresponding to the data collection time t on day d were calculated respectively:

[0089]

[0090] where R d,t The adjustment ratio for the data collection time t on day d; Power_clean d,t This refers to the power system output data collected at time t on day d; Power_short d,t This refers to the short-term power data collected at time t on day d.

[0091] Then, statistical methods were used to analyze the sample dataset to obtain the maximum value of the adjustment ratio, R. max and minimum value R min The maximum and minimum values ​​of the adjustment ratio are used as the adjustment ratio range R. range =[R min ,R max ];

[0092] Finally, the probability distribution of the adjustment ratio for each collection time in the historical days is calculated using the probability density curve function to obtain the optimal adjustment ratio for each collection time.

[0093] In detail, in step three of this application's embodiments, the difference between the day-to-day price and the real-time price is processed into feature data using a hot coding method. Hot coding is an encoding method that converts categorical variables into numerical variables. It represents each category as a binary vector, avoiding the problem of comparing numerical values ​​between different categories and enabling the model to better handle categorical variables. Its specific steps include:

[0094] First, obtain the day-ahead price and real-time price corresponding to the data collection time t on day d;

[0095] Then, the difference between the day-ahead price and the real-time price at time t on day d is calculated, i.e.:

[0096] Price_diff d,t =Price_riqian d,t -Price_shishi d,t ;

[0097] Then, the difference between the previous day's price and the real-time price at time t on day d is compared with a threshold of 0 to determine the direction of the price difference: when the difference between the previous day's price and the real-time price is less than 0, the direction of the price difference is -1; when the difference between the previous day's price and the real-time price is equal to 0, the direction of the price difference is 0; when the difference between the previous day's price and the real-time price is greater than 0, the direction of the price difference is 1. The process of determining the direction of the price difference is shown in the following formula:

[0098]

[0099] Among them, Price_diff d,t Price_riqian represents the price difference at time t on day d. d,t Price_shishi is the day-ahead price at time t on day d. d,t Price_hot represents the real-time price at time t on day d. d,t The direction of the price difference at time t on day d;

[0100] Finally, by summing up the obtained price difference directions, the corresponding feature data can be obtained.

[0101] Furthermore, in step three of this application embodiment, the original feature dataset is subjected to window sliding processing to expand it into an expanded feature dataset. The specific steps include:

[0102] First, set the window width W and sliding length ΔW of the time window, and then arrange the feature data of the original feature dataset in chronological order to obtain the sorted original feature dataset.

[0103] The starting position of the sorted original feature dataset is shifted one sliding step ΔW to the right according to the window width W of the time window, and the feature data at the time window position is set to 0 to obtain new feature data.

[0104] In this embodiment, a time window of 96 points is used to perform lag processing on the existing feature dataset. Each feature data point is shifted backward by one time window length to become the new feature data. For example, if feature1 = [1,2,3,4,5,6,...], and the time window length W is 3, then feature1 is shifted backward by W points sequentially. That is, data point 1 is moved to the position of data point 4, data point 2 is moved to the position of data point 5, data point 3 is moved to the position of data point 6, and so on, resulting in feature2 = [0,0,0,1,2,3,...]. The first T data points of feature2 are filled with 0.

[0105] In detail, step four of this application embodiment requires predicting the direction of the feature data for the target day. Specifically, firstly, an expanded feature dataset is used to train a preset neural network classification model, and then the trained neural network classification model is used to predict the direction of the feature data for the target day. The specific steps include:

[0106] First, the expanded feature dataset is classified by day to obtain different feature units;

[0107] Secondly, a density-based clustering algorithm (such as the DBSCAN time series similarity clustering algorithm) is used to cluster different feature units of the expanded feature dataset, and the feature units containing the cluster centers similar to the target day are selected; the following method is used to determine similarity to the target day:

[0108]

[0109] where X di The feature dataset for the target day; X d This is the feature dataset for day d; dist(X) di ,X d ) is the Euclidean distance calculation function; X' di,t X' represents the normalized feature data collected at time t on the target day; d,t The normalized feature data is the data collected at time t on day d; T is the total number of collection times.

[0110] The feature data of the target day is sorted by Euclidean distance with different feature units, and the feature data corresponding to the feature units that meet the preset distance are selected as training data.

[0111] Finally, the obtained training data is used to train a pre-defined neural network classification model to obtain the trained neural network classification model. For example... Figure 2 As shown, it is a schematic diagram of the direction of feature data for the target day predicted by a neural network classification model that has been trained.

[0112] In addition, such as Figure 3 As shown, in step five of this application embodiment, the data obtained in the first four steps are used as input data, market and weather data are used as environmental factors, the action range is the adjustment ratio range obtained in step two, the product of feature data and adjustment ratio is used as the loss function, and multiple machine learning algorithm models for obtaining the optimal adjustment ratio are established, such as long short-term neural network LSTM, time series learning transformer, reinforcement learning PPO, etc.

[0113] Specifically, the steps in step five of selecting the model with the highest number of data collection times when the adjustment ratio is consistent with the direction of the feature data as the optimal model for the corresponding machine learning algorithm include:

[0114] First, a training set is selected based on the expanded feature dataset, and different machine learning algorithm models are trained on the training set for a preset number of times. After each training, the corresponding machine learning algorithm model is saved. For example, in this embodiment, for different models, a single training session uses six months of data as training data, the most recent 10 days of data for the target date as validation data, and the data at point T on a single day as the sequence length for training. The number of training steps is set to M*T steps, and a total of N training sessions are conducted. The model is saved for each training session.

[0115] For the same type of machine learning algorithm model, a pre-set validation set is input into the machine learning algorithm model saved after different training iterations, and the adjustment ratio and adjustment direction of the output are obtained respectively; the relationship between the adjustment ratio and the adjustment direction is as follows:

[0116]

[0117] Then, it is determined whether the adjustment direction output by the machine learning algorithm model in each training session is consistent with the direction of the feature data in the validation set:

[0118]

[0119] where R d,t R_diff represents the adjustment ratio for the data collection time t on day d; d,t The adjustment direction for the data collection time t on day d; RP_diff d,t This indicates whether the adjustment direction at time t on day d is consistent with the direction of the feature data; Price_diff d,t This indicates the direction of the feature data collected at time t on day d;

[0120] Finally, the number of data collection moments in each training session where the adjustment direction of the machine learning algorithm model output was consistent with the direction of the feature data in the validation set was counted:

[0121]

[0122] Where RP_num is the number of data collection moments when the adjustment direction of the model output is consistent with the feature data direction of the validation set; M is the number of days in the validation set; and T is the number of data collection moments per day.

[0123] Ultimately, the model with the highest number of corresponding data collection times (RP_num) was selected as the optimal model for the corresponding machine learning algorithm.

[0124] Based on the above process, this embodiment can select the best model for different machine learning algorithm models and obtain multiple adjustment ratios for different types of best models.

[0125] Furthermore, in step seven of this embodiment, the multiple adjustment ratios for the target day are judged in the same direction as the characteristic data of the target day to obtain the final adjustment ratio. The specific steps include:

[0126] First, the optimal model based on different machine learning algorithms is used to obtain the adjustment ratio and adjustment direction for the target day. The relationship between the adjustment ratio and adjustment direction is as follows:

[0127]

[0128] Then, the number of different machine learning algorithm models in different adjustment directions at the same data acquisition time was counted:

[0129]

[0130] where R i,t R_diff is the adjustment ratio of the output of the i-th machine learning algorithm model at the t-th acquisition time; i,t Output the adjustment direction at time t for the i-th machine learning algorithm model; I is the number of different machine learning algorithm models; RNpos t RNNeg represents the number of samples collected at time t that are in the positive direction. t RNorig represents the number of samples collected at time t in the negative direction. t The quantity that is not adjusted at time t;

[0131] Then, the number of times the direction was adjusted at the t-th acquisition time is denoted as RN. t The corresponding direction is denoted as the adjustment direction R_frist_diff at that acquisition moment. t The process is as follows:

[0132] RN t =max(RNpos t ,RNneg t ,RNorig t );

[0133]

[0134] Based on the above process, the adjustment direction with the most data at the current acquisition time can be selected and compared with the characteristic data direction at the same acquisition time in the target day: such as Figure 4 As shown, if the adjustment direction selected at the current acquisition time is consistent with the feature data direction of the target day, then the average adjustment ratio corresponding to the selected adjustment direction at the current acquisition time is calculated as the final adjustment ratio; if the adjustment direction selected at the current acquisition time is inconsistent with the feature data direction of the target day, then no adjustment is made at the current acquisition time. The above process can be expressed as:

[0135]

[0136] Where R_finally t Price_object_diff represents the final adjustment ratio for the data collected at time t on the target day. t R_first_diff represents the direction of the feature data collected at time t on the target day. t The adjustment direction is defined by the data collection time t on the target day; I represents the number of different machine learning algorithm models; R i,t This is the adjustment ratio of the output of the i-th machine learning algorithm model at the t-th acquisition time.

[0137] Furthermore, in combination Figure 5 As shown, this application also discloses a power market power parameter auxiliary decision-making device based on multi-model fusion, the device comprising:

[0138] The sample data acquisition module is used to acquire sample data of the electricity market corresponding to the collection time t on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, clearing power data, short-term power data, adjustment ratio data, and environmental parameters; where d = 1, 2, ..., D, and D is the upper limit of the number of days;

[0139] The adjustment range determination module is used to analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range;

[0140] The feature data augmentation module is used to calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and to process the difference between the day-ahead price and the real-time price into the original feature dataset through hot encoding. Then, the original feature dataset is augmented by window sliding processing to obtain the augmented feature dataset.

[0141] The target feature prediction module is used to train the neural network classification model using an expanded feature dataset, and to predict the feature data of the target day based on the trained neural network classification model.

[0142] The optimal model selection module is used to establish multiple machine learning algorithm models for obtaining the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and to train and validate different machine learning algorithm models respectively. The model with the most data collection times when the adjustment ratio is consistent with the direction of feature data is selected as the optimal model of the corresponding machine learning algorithm model. The machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm.

[0143] The adjustment ratio output module is used to input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day;

[0144] The adjustment ratio determination module is used to compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day and select the final adjustment ratio.

[0145] The apparatus provided in this application embodiment can achieve... Figure 1 To avoid repetition, the various processes implemented in the method embodiments will not be described again here.

[0146] like Figure 6 As shown in the illustration, this application also provides an electronic device, including a processor and a memory, and a program or instructions stored in the memory and executable on the processor, which, when executed by the processor, implement as follows: Figure 1 The various processes of the method embodiments shown are all capable of achieving the same technical effect, and will not be described again here to avoid repetition.

[0147] This application embodiment also provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the above-described functionality. Figure 1 The various processes described in the embodiments of the method described herein can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0148] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the above-described... Figure 1 The various processes described in the embodiments of the method described herein can achieve the same technical effect, and will not be repeated here to avoid repetition.

[0149] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0150] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components may be combined, or integrated into another device, or some features may be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0152] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0153] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0154] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0155] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a device (which may be a terminal or platform, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0156] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A power market parameter-assisted decision-making method based on multi-model fusion, characterized in that, Includes the following steps: Step 1: Obtain sample data of the electricity market corresponding to the t-th collection time on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least the day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, cleared power data, short-term power data, adjustment ratio data, and environmental parameters; wherein D represents the maximum number of days; Step 2: Analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range: Convert the cleared power data corresponding to the collection time t on day d in the sample dataset into power system output data; Calculate the adjustment ratio between the power system output data and the short-term power data at the time of data collection on day d at time t. The maximum and minimum values ​​of the adjustment ratio are obtained by statistical analysis of the sample dataset using statistical methods, and the maximum and minimum values ​​of the adjustment ratio are used as the range of the adjustment ratio. The probability distribution of the adjustment ratio at each collection time in the historical days is calculated using a probability density curve function. Step 3: Calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and process the difference between the day-ahead price and the real-time price into the original feature dataset through hot coding. Then, expand the original feature dataset by window sliding processing to obtain the expanded feature dataset. Step 4: Train the neural network classification model using the expanded feature dataset, and predict the feature data for the target day based on the trained neural network classification model; Step 5: Establish multiple machine learning algorithm models to obtain the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and train and validate different machine learning algorithm models respectively. Select the model with the most data collection times when the adjustment ratio is consistent with the direction of feature data as the optimal model of the corresponding machine learning algorithm model; the machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm. Step Six: Input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day; Step 7: Compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day to determine the final adjustment ratio.

2. The power market power parameter auxiliary decision-making method according to claim 1, characterized in that, In step one, when obtaining sample data of the electricity market, the data is collected in days and each day is divided into several equally spaced collection times. Sample data is collected for each collection time.

3. The power market power parameter auxiliary decision-making method according to claim 2, characterized in that, In step one, a data normalization method is used to preprocess the sample dataset: First, obtain the maximum and minimum values ​​of different sample data in the sample dataset, and then classify the sample data according to different times of the same day to obtain the sample units for day d. The difference between the sample data at time t is obtained based on the maximum and minimum values ​​of the corresponding sample data. Based on the sample unit of day d, obtain the sample data at time t in day d, and obtain the normalized sample data corresponding to time t in day d based on the difference between the sample data at time t in day d and the sample data corresponding to time t in day d.

4. The power market power parameter auxiliary decision-making method according to claim 1, characterized in that, The step three, which processes the difference between the day-to-day price and the real-time price into feature data using a hot coding method, includes: Obtain the day-ahead price and real-time price corresponding to the data collection time t on day d; Calculate the difference between the day-ahead price and the real-time price at time t on day d; The difference between the previous day's price and the real-time price at time t on day d is compared with the threshold 0 to determine the direction of the price difference: when the difference between the previous day's price and the real-time price is less than 0, the direction of the price difference is -1; when the difference between the previous day's price and the real-time price is equal to 0, the direction of the price difference is 0; when the difference between the previous day's price and the real-time price is greater than 0, the direction of the price difference is 1. The obtained price difference directions are summarized to obtain characteristic data.

5. The power market power parameter auxiliary decision-making method according to claim 4, characterized in that, The step of expanding the original feature dataset to obtain the expanded feature dataset in step three includes: Set the window width of the time window W and sliding step size The feature data of the original feature dataset are arranged in chronological order to obtain the sorted original feature dataset. The starting positions of the sorted original feature dataset are arranged according to the window width of the time window. W Move forward one sliding step at a time Then, the feature data at the time window position is set to 0 to obtain new feature data.

6. The power market power parameter auxiliary decision-making method according to claim 1, characterized in that, The step four, which involves training the neural network classification model using the expanded feature dataset, includes: First, the expanded feature dataset is classified by day to obtain different feature units; Secondly, a density-based clustering algorithm is used to cluster different feature units of the expanded feature dataset, and the feature units containing cluster centers similar to the target date are selected; the following method is used to determine similarity to the target date: ； in The feature dataset for the target day; For the first d The feature dataset of the day; This is the function for calculating Euclidean distance; For the target day t Normalized feature data at the time of acquisition; For the first d Heavenly t Normalized feature data after acquisition time; T is the total number of acquisition times; The feature data of the target day is sorted by Euclidean distance with different feature units, and the feature data corresponding to the feature units that meet the preset distance are selected as training data. Finally, the obtained training data is used to train the preset neural network classification model to obtain the trained neural network classification model.

7. The power market power parameter auxiliary decision-making method according to claim 1, characterized in that, The step in step five, which selects the model with the highest number of data acquisition moments when the adjustment ratio is consistent with the direction of the feature data, as the optimal model for the corresponding machine learning algorithm, includes: First, a training set is selected based on the expanded feature dataset, and different machine learning algorithm models are trained on the training set for a preset number of times. After each training is completed, the corresponding machine learning algorithm model is saved. For the same type of machine learning algorithm model, a pre-set validation set is input into the machine learning algorithm model saved after different training iterations, and the adjustment ratio and adjustment direction of the output are obtained respectively; the relationship between the adjustment ratio and the adjustment direction is as follows: ； Then, it is determined whether the adjustment direction output by the machine learning algorithm model in each training session is consistent with the direction of the feature data in the validation set: ； in R d,t For the first d Heavenly t Adjustment ratio for data acquisition time; R_diff d,t For the first d Heavenly t Adjusting the direction of data acquisition; RP_diff d,t Indicates the first d Heavenly t Is the adjustment direction at the time of data acquisition consistent with the direction of the feature data? Price_diff d,t Indicates the first d Heavenly t The direction of characteristic data at the time of acquisition; Finally, the number of data collection moments in each training session where the adjustment direction of the machine learning algorithm model output is consistent with the direction of the feature data in the validation set is counted, and the model with the most corresponding data collection moments is selected as the optimal model for the corresponding machine learning algorithm model.

8. The power market power parameter auxiliary decision-making method according to claim 1, characterized in that, Step seven, which involves comparing the obtained adjustment ratios for the target day with the characteristic data of the target day and selecting the final adjustment ratio, includes: The optimal models based on different machine learning algorithms respectively obtain the adjustment ratio and adjustment direction for the target day, where the relationship between the adjustment ratio and adjustment direction is as follows: ； Count the number of different machine learning algorithm models in different adjustment directions at the same data acquisition time: ； in R i,t For the first i The output of the machine learning algorithm model is the first t Adjustment ratio for data acquisition time; R_diff i,t For the first i The output of the machine learning algorithm model is the first t Adjusting the direction of data acquisition; I The number of different machine learning algorithm models; RNpos t For the first t The number of samples collected at the moment they are in the positive direction; RNneg t For the first t The number of items collected at the time of the negative direction; RNorig t For the first t The number of samples collected at any given time is not adjusted. Select the adjustment direction with the most occurrences at the current acquisition time and compare it with the characteristic data direction at the same acquisition time on the target day: if the adjustment direction selected at the current acquisition time is consistent with the characteristic data direction on the target day, calculate the average adjustment ratio corresponding to the adjustment direction selected at the current acquisition time as the final adjustment ratio; if the adjustment direction selected at the current acquisition time is inconsistent with the characteristic data direction on the target day, no adjustment is made at the current acquisition time.

9. A power market power parameter auxiliary decision-making device based on multi-model fusion, characterized in that, The device includes: The sample data acquisition module is used to acquire sample data of the electricity market corresponding to the t-th collection time on day d, summarize them into a sample dataset, and preprocess the sample dataset; the sample dataset includes at least day-ahead price, real-time price, market day-ahead clearing price, market real-time clearing price, cleared electricity volume data, short-term power data, adjustment ratio data, and environmental parameters; wherein D represents the maximum number of days; The adjustment range determination module is used to analyze the existing adjustment ratio data in the sample dataset to obtain the adjustment ratio range. Convert the cleared power data corresponding to the collection time t on day d in the sample dataset into power system output data; Calculate the adjustment ratio between the power system output data and the short-term power data at the time of data collection on day d at time t. The maximum and minimum values ​​of the adjustment ratio are obtained by statistical analysis of the sample dataset using statistical methods, and the maximum and minimum values ​​of the adjustment ratio are used as the range of the adjustment ratio. The probability distribution of the adjustment ratio at each collection time in the historical days is calculated using a probability density curve function. The feature data augmentation module is used to calculate the difference between the day-ahead price and the real-time price in the electricity market based on the sample dataset, and to process the difference between the day-ahead price and the real-time price into the original feature dataset through hot encoding. Then, the original feature dataset is augmented by window sliding processing to obtain the augmented feature dataset. The target feature prediction module is used to train the neural network classification model using an expanded feature dataset, and to predict the feature data of the target day based on the trained neural network classification model. The optimal model selection module is used to establish multiple machine learning algorithm models for obtaining the optimal adjustment ratio using feature data and adjustment ratio as loss functions, and to train and validate different machine learning algorithm models respectively. The model with the most data collection times when the adjustment ratio is consistent with the direction of feature data is selected as the optimal model of the corresponding machine learning algorithm model. The machine learning algorithm model includes at least long short-term neural network algorithm, time series learning algorithm, and reinforcement learning algorithm. The adjustment ratio output module is used to input the data collected on the target day into the optimal model of different machine learning algorithm models to obtain multiple adjustment ratios for the target day; The adjustment ratio determination module is used to compare the multiple adjustment ratios obtained for the target day with the characteristic data of the target day and select the final adjustment ratio.

Images