Multi-agent electricity price prediction intelligent computing cloud platform system based on autonomous model evaluation

Abstract

This application proposes a multi-agent intelligent computing cloud platform system for electricity price prediction based on autonomous model evaluation. The system includes: a data acquisition agent for collecting feature data based on power production-related characteristic variables; a data preprocessing agent for preprocessing the feature data; a model evaluation agent for training multiple candidate models using training data corresponding to the feature variables; and a model evaluation agent for performing quantitative evaluations of the candidate models based on stability and accuracy to obtain evaluation results, and determining a target model from the candidate models based on the evaluation results; and an electricity price prediction agent for inputting the preprocessed feature data into the target model for prediction, and outputting the predicted electricity price. By evaluating the candidate models through the model evaluation agent, the model used for predicting electricity prices is optimized, improving the accuracy of electricity price prediction.

Description

Technical Field

[0001] This application relates to the field of cloud computing technology, and in particular to a multi-agent intelligent computing cloud platform system for predicting electricity prices based on autonomous model evaluation. Background Technology

[0002] Electricity price forecasting is a core research area in the electricity market, and the choice of forecasting model directly determines the accuracy and reliability of price predictions. Electricity prices are time series data with covariates, dynamically influenced by multiple factors. Their data characteristics continuously change with variations in electricity market supply and demand, policy adjustments, and the natural environment. Therefore, it is necessary to dynamically select an appropriate forecasting model based on changes in data characteristics.

[0003] Current methods for selecting electricity price forecasting models heavily rely on the human experience of AI experts. Experts must manually analyze the characteristics of electricity price time series data (such as stationarity, nonlinearity, and long-term correlation) and then select a suitable forecasting model from numerous options based on their experience. If data characteristics change, experts must re-analyze and reselect the model manually. This approach has significant technical drawbacks: 1) AI experts are expensive, and they need to continuously monitor changes in electricity market data, resulting in significant time and economic costs; 2) Model selection efficiency is low; the process of manually analyzing data characteristics and selecting a model is time-consuming, making real-time, dynamic model selection impossible and difficult to adapt to the rapid fluctuations in intraday electricity prices; 3) Model selection is highly subjective; differences in the experience and analytical perspectives of different experts can easily lead to biased selection results, affecting forecast accuracy. Summary of the Invention

[0004] This application aims to at least partially address one of the technical problems in the related art.

[0005] Therefore, this application proposes a system, apparatus, electronic device, and storage medium.

[0006] One embodiment of this application proposes a multi-agent intelligent computing cloud platform system for electricity price prediction based on autonomous model evaluation, including: Data acquisition agent, data preprocessing agent, electricity price prediction agent, model evaluation agent, feature evaluation agent; The data acquisition agent is used to collect feature data based on power production-related feature variables, and send the feature data to the data preprocessing agent. The data preprocessing agent is used to preprocess the feature data and send the preprocessed feature data to the electricity price prediction agent. The model evaluation agent is used to autonomously discover candidate models for electricity price and electricity consumption prediction scenarios using training data corresponding to the feature variables, construct a candidate model set, and train and test multiple candidate models; it is also used to perform quantitative evaluation of the candidate models based on stability and accuracy to obtain evaluation results, and determine the target model from the candidate models based on the evaluation results. The electricity price prediction agent is used to input the preprocessed feature data into the target model for prediction and output the predicted value of the electricity price. The feature evaluation agent is used to evaluate the prediction accuracy based on the actual value and the predicted value of the electricity price, and to instruct the model evaluation agent to update the target model based on the prediction accuracy.

[0007] Optionally, the model evaluates the agent for: Based on the characteristics of the aforementioned feature data, candidate models suitable for electricity price forecasting are retrieved; wherein, the candidate models include at least one of the following categories: statistical models, machine learning models, deep learning models, and large models; The training data is processed in a targeted manner according to the type of the candidate model, so that the training data is adapted to the corresponding candidate model; The candidate model is trained using the specially processed training data, and the trained candidate model is used to make predictions to obtain prediction results.

[0008] Optionally, the step of retrieving candidate models suitable for electricity price forecasting based on the characteristics of the feature data includes: Determine the characteristics of the feature data in the time series, and determine the first search keyword; wherein, the characteristics include at least one of the following: stationary time series, nonlinear high-dimensional feature time series, and long-term correlated nonlinear time series; Identify secondary search keywords related to electricity prices; The first search keyword and the second search keyword are combined to perform model retrieval and determine the candidate model.

[0009] Optionally, the targeted processing of the training data according to the type of candidate model includes any one of the following: For the training data input to the statistical model, a time series stationarity test is performed, and the non-stationary training data is differentially processed. For the training data input to the machine learning model, feature dimensionality reduction is performed to remove redundant features from the training data; The training data input to the deep learning model and the large model are subjected to time series sliding window processing to generate training data in the input sequence format.

[0010] Optionally, training the candidate model with the specifically processed training data includes: The training data is divided into a training set, a validation set, and a test set; The candidate model is trained using the training set, and the hyperparameters in the candidate model are optimized based on the validation set to determine the hyperparameter combination corresponding to the candidate model, so as to obtain the trained candidate model.

[0011] Optionally, the step of using the trained candidate model to make predictions to obtain prediction results includes: The test set is input into the trained candidate model to obtain the prediction result.

[0012] Optionally, the step of performing a quantitative evaluation of the candidate model based on stability and accuracy to obtain the evaluation result includes: Calculate the accuracy index based on the predicted results and the actual results in the test set; Based on the prediction accuracy of the candidate model on the validation set and the prediction accuracy of the candidate model on the test set, a stability index is calculated. Calculate the inference efficiency index based on the time taken to obtain the prediction result from the candidate model; The evaluation result is obtained by weighting the accuracy index, the stability index, and the inference efficiency index.

[0013] Optionally, the model evaluation agent is also used for: The weights of the accuracy index, the stability index, and the inference efficiency index are determined based on the electricity price forecast scenario, and the weights are used for weighted calculation to obtain the evaluation result.

[0014] Optionally, determining the target model from the candidate models based on the evaluation results includes: The candidate model with the highest evaluation result is selected as the target model.

[0015] Optionally, the feature evaluation agent is used for: If the prediction accuracy is lower than a preset threshold, the model evaluation agent is instructed to reselect the target model. It is also used to determine the feature variable to be replaced and to re-mine candidate feature variables to replace the feature variable to be replaced.

[0016] Another embodiment of this application proposes an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the system as described in the foregoing aspect.

[0017] Another embodiment of this application proposes a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the system as described in the foregoing aspect.

[0018] Another embodiment of this application proposes a chip including processing circuitry configured to execute the system as described in the foregoing aspect.

[0019] Another embodiment of this application proposes a computer program product that, when executed by a processor, implements the system as described in the foregoing aspect.

[0020] The multi-agent intelligent computing cloud platform system, device, electronic equipment, chip, and storage medium for electricity price prediction based on autonomous model evaluation proposed in this application can achieve the following beneficial effects: Completely eliminate reliance on AI experts and significantly reduce costs: The model evaluation agent achieves autonomous model selection throughout the entire process through automatic retrieval, feature adaptation processing, and multi-dimensional evaluation, replacing the manual analysis and model selection work of AI experts in traditional methods. This completely eliminates expert costs, reduces the model selection cycle, and improves model selection efficiency.

[0021] The objectivity and accuracy of model selection have been significantly improved: a multi-dimensional quantitative evaluation index system was designed and combined with the analytic hierarchy process to achieve dynamic adjustment of weights, avoiding subjective biases caused by human experience, ensuring that the target model is highly compatible with the characteristics of electricity price data, and significantly improving prediction accuracy.

[0022] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0023] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 A schematic diagram of the structure of a multi-agent intelligent computing cloud platform system for electricity price prediction based on autonomous model evaluation, provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a chip proposed in an embodiment of this application. Detailed Implementation

[0024] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0025] The following describes, with reference to the accompanying drawings, an embodiment of the intelligent computing cloud platform system for multi-agent electricity price prediction based on autonomous model evaluation, according to an embodiment of this application.

[0026] This application proposes a multi-agent intelligent computing cloud platform device for electricity price prediction based on autonomous model evaluation.

[0027] Figure 1 This is a schematic diagram of the structure of a multi-agent intelligent computing cloud platform system for electricity price prediction based on autonomous model evaluation, provided in an embodiment of this application.

[0028] like Figure 1 As shown, the system includes: a data acquisition agent 10, a data preprocessing agent 20, an electricity price prediction agent 30, a model evaluation agent 40, and a feature evaluation agent 50. The data acquisition agent is used to collect feature data based on power production-related feature variables, and send the feature data to the data preprocessing agent. The data preprocessing agent is used to preprocess the feature data and send the preprocessed feature data to the electricity price prediction agent. The model evaluation agent is used to autonomously discover candidate models for electricity price and electricity consumption prediction scenarios using training data corresponding to the feature variables, construct a candidate model set, and train and test multiple candidate models; it is also used to perform quantitative evaluation of the candidate models based on stability and accuracy to obtain evaluation results, and determine the target model from the candidate models based on the evaluation results. The electricity price prediction agent is used to input the preprocessed feature data into the target model for prediction and output the predicted value of the electricity price. The feature evaluation agent is used to evaluate the prediction accuracy based on the actual value and the predicted value of the electricity price, and to instruct the model evaluation agent to update the target model based on the prediction accuracy.

[0029] In this embodiment, the data acquisition agent, data preprocessing agent, electricity price prediction agent, model evaluation agent, and feature evaluation agent work together to form a closed-loop automated process of "model selection - prediction feedback - feature optimization - model retraining". The core is to achieve autonomous model selection for electricity price prediction through the model evaluation agent.

[0030] Let the time series of electricity price forecasts be... ,in , The current moment; Forecast values ​​for electricity prices, including day-ahead trading prices. and intraday real-time trading prices ; for The set of characteristic variables at any given time; This is a set of candidate models for electricity price forecasting. This represents the number of candidate models. The core objective of the system is to determine the optimal prediction model that best suits the current data characteristics by evaluating the agent's autonomous model selection. (That is, the target model), and through the collaboration of various intelligent agents, to achieve accurate prediction of electricity prices, so that power plants can make reasonable adjustments to their production plans, reduce production costs, improve production efficiency, and reduce waste of resources.

[0031] Optionally, feature data can be crawled using data crawling techniques to output raw time series data. Initially, data from power plants and meteorological data are crawled according to preset rules to achieve timed / real-time updates of feature data, providing a data foundation for subsequent preprocessing and model evaluation. Among these, the feature variables related to power production include: Power plant-side physical characteristics: These characteristics reflect the physical state of the power production process, directly affect power supply capacity, and are a core physical influencing factor on electricity prices.

[0032] Grid-side physical characteristics: These primarily reflect the physical constraints of power transmission / dispatch. Physical bottlenecks in power transmission directly affect regional electricity prices. Specifically, these include: grid operation-related characteristics and grid dispatch-related characteristics. Load-side physical characteristics: These characteristics primarily reflect the physical demand in the electricity consumption process. Changes in the physical load are a direct cause of electricity price fluctuations. Specifically, they include: regional load-related characteristics. Physical characteristics of power fuels: physical indicators of fuels directly related to power production, including: characteristics of thermal power fuels and characteristics of fuel transportation. Physical characteristics of the power system operating environment: natural physical environment data that are strongly related to power production / operation, specifically including: extreme weather-related characteristics: regional extreme temperatures (maximum / minimum), meteorological level quantification values ​​of strong winds / rainstorms / snowstorms (e.g., strong wind level quantified as 0-12), duration of high temperature / cold wave; equipment operating environment-related characteristics: icing thickness of transmission and transformation lines, dust accumulation rate of photovoltaic panels, ambient temperature at wind turbine hubs.

[0033] Optionally, the feature data may be preprocessed, including: The feature data is automatically cleaned, standardized, and formatted to output standardized data that meets the model input requirements. Specifically, it performs outlier identification and repair (3σ criterion + interpolation), missing value imputation (mean / K nearest neighbors), dimensional normalization (min-max to [0,1] interval), and time series data resampling (matching prediction frequency). The processing flow is fully automated, ensuring the data quality of the input model.

[0034] Optionally, the model evaluates the agent for: Based on the characteristics of the aforementioned feature data, candidate models suitable for electricity price forecasting are retrieved; wherein, the candidate models include at least one of the following categories: statistical models, machine learning models, deep learning models, and large models; The training data is processed in a targeted manner according to the type of the candidate model, so that the training data is adapted to the corresponding candidate model; The candidate model is trained using the specially processed training data, and the trained candidate model is used to make predictions to obtain prediction results.

[0035] In this embodiment, based on in-depth analysis and understanding of the feature data, the model evaluation agent can capture the characteristics of the feature data, thereby retrieving candidate models suitable for electricity price prediction. These candidate models come from, but are not limited to, at least one of the following categories: Statistical models, which are based on statistical principles and methods, analyze and model historical data to predict future electricity price trends; Machine learning models, leveraging the powerful data analysis and pattern recognition capabilities of machine learning algorithms, automatically learn the complex relationship between electricity prices and related characteristic variables from large amounts of data; Deep learning models, which utilize deep neural network structures, can process complex nonlinear data and are particularly suitable for mining deep features and patterns in electricity price data. The large model is an AI system adapted for electricity price prediction scenarios, trained with a large amount of data and ultra-large-scale parameters. It has comprehensive capabilities such as understanding, generation, reasoning, and interaction.

[0036] To ensure that each candidate model can be trained on the most suitable data, the model evaluation agent performs targeted processing on the training data according to the type of candidate model, so that the training data is adapted to the corresponding candidate model.

[0037] The specially processed training data is systematically input into the candidate model for training. During training, the model continuously adjusts its parameters to optimize its fit to the training data. After training, the trained candidate model is used to predict relevant data, thereby obtaining prediction results and providing data support for subsequent model evaluation.

[0038] Optionally, the step of retrieving candidate models suitable for electricity price forecasting based on the characteristics of the feature data includes: Determine the characteristics of the feature data in the time series, and determine the first search keyword; wherein, the characteristics include at least one of the following: stationary time series, nonlinear high-dimensional feature time series, and long-term correlated nonlinear time series; Identify secondary search keywords related to electricity prices; The first search keyword and the second search keyword are combined to perform model retrieval and determine the candidate model.

[0039] In this embodiment, the characteristics exhibited by the feature data over time series are analyzed. These characteristics include at least one of the following: A stationary time series is a data set whose statistical properties (such as mean, variance, etc.) do not change over time. Nonlinear time series indicate that the relationship between data is not a simple linear one, and more complex models are needed to capture it; High-dimensional feature time series means that the feature data contains information in many dimensions, which places high demands on the model's ability to process high-dimensional data; Long-term correlation time series reflect the correlation between data over a relatively long time span.

[0040] By identifying the corresponding primary search keywords based on these characteristics, we can accurately locate models that meet the criteria during the model retrieval process.

[0041] Focusing on the core objective of electricity prices, we identified secondary search keywords related to electricity prices. These keywords may involve factors influencing electricity prices, such as: electricity, electricity price, energy, forecasting, etc., ensuring that the retrieved models are closely related to electricity price forecasts.

[0042] The first and second search keywords are combined to form a precise set of search criteria. Using professional model retrieval tools or platforms, models are retrieved based on these criteria, and candidate models that meet the requirements are selected from the model resources, providing a rich selection for subsequent model training and evaluation.

[0043] Optionally, the targeted processing of the training data according to the type of candidate model includes any one of the following: For the training data input to the statistical model, a time series stationarity test is performed, and the non-stationary training data is differentially processed. For the training data input to the machine learning model, feature dimensionality reduction is performed to remove redundant features from the training data; The training data input to the deep learning model and the large model are subjected to time series sliding window processing to generate training data in the input sequence format.

[0044] In this embodiment, statistical models typically have high requirements for the stationarity of the data. Therefore, for the training data input into the statistical model, the model evaluation agent first performs a rigorous time series stationarity test. Common testing methods include unit root tests (such as the ADF test). If the training data is found to not meet the stationarity requirement, i.e., it is a non-stationary time series, the model evaluation agent will perform differencing. Difference processing transforms the non-stationary series into a stationary series by subtracting each period of the time series data, thereby meeting the input requirements of the statistical model and improving the model's prediction accuracy.

[0045] Machine learning models may face the curse of dimensionality when processing high-dimensional data, meaning that excessive feature dimensions can lead to problems such as excessively long training times and overfitting. Therefore, the model evaluation agent performs feature dimensionality reduction on the training data input to the machine learning model. Common feature dimensionality reduction methods include Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and Factor Analysis. These methods remove redundant features from the training data, retaining the key features that have the greatest impact on the model's predictions, thereby reducing data dimensionality and improving model training efficiency and predictive performance.

[0046] Deep learning models and large models require specific input sequence formats when processing time series data. Therefore, the model evaluation agent performs time series sliding window processing on the training data input to the deep learning model. Specifically, a fixed-length window slides across the time series data, extracting a segment of data as an input sample at each slide, thus generating training data that conforms to the input sequence format of the deep learning model. For example, for a time series of length N, a window size of M, and a step size of 1, N-M+1 input samples will be generated, each containing data from M time steps, enabling the deep learning model to effectively learn patterns and features in the time series.

[0047] Optionally, training the candidate model with the specifically processed training data includes: The training data is divided into a training set, a validation set, and a test set; The candidate model is trained using the training set, and the hyperparameters in the candidate model are optimized based on the validation set to determine the hyperparameter combination corresponding to the candidate model, so as to obtain the trained candidate model.

[0048] In this embodiment, in order to comprehensively evaluate the performance of candidate models and make effective parameter adjustments during training, the model evaluation agent scientifically divides the specially processed training data into training sets. Validation set and test set The training set, comprising a relatively large portion (approximately 70% of the data), is used for learning the model's parameters. The validation set, comprising approximately 20%, is used to adjust the model's hyperparameters during training. The test set, comprising approximately 10%, is used to finally evaluate the generalization ability of the trained model. The data partitioning process must follow the principles of randomness and stratification, ensuring that the data distribution of each subset is similar to that of the original data, in order to avoid bias in model evaluation results due to unreasonable data partitioning.

[0049] During training, the candidate model continuously adjusts its parameters based on the training set data to minimize the error between the predicted results and the true values. Simultaneously, the model evaluation agent optimizes the hyperparameters of the candidate model using validation set data (e.g., through grid search / Bayesian optimization). By trying different hyperparameter combinations on the validation set and evaluating the model's performance metrics (such as accuracy and loss function value) on the validation set, the hyperparameter combination that optimizes the model's performance is selected, thus determining the optimal hyperparameter settings for the candidate model and obtaining the trained candidate model.

[0050] Optionally, the step of using the trained candidate model to make predictions to obtain prediction results includes: The test set is input into the trained candidate model to obtain the prediction result.

[0051] In this embodiment, the optimized candidate models are used to... Perform predictions and output the prediction results of each candidate model. ( , corresponding to the (1 candidate model).

[0052] Optionally, the step of performing a quantitative evaluation of the candidate model based on stability and accuracy to obtain the evaluation result includes: Calculate the accuracy index based on the predicted results and the actual results in the test set; Based on the prediction accuracy of the candidate model on the validation set and the prediction accuracy of the candidate model on the test set, a stability index is calculated. Calculate the inference efficiency index based on the time taken to obtain the prediction result from the candidate model; The evaluation result is obtained by weighting the accuracy index, the stability index, and the inference efficiency index.

[0053] In this embodiment, an accuracy index is calculated based on the prediction results output by the candidate model and the pre-known true results in the test set. The accuracy index includes: Mean squared error (MSE) measures the average of the squares of the errors between the predicted and actual values, reflecting the degree of dispersion of the predicted values. Mean Absolute Error (MAE) calculates the average absolute error between the predicted and actual values, and intuitively reflects the average degree of deviation between the predicted and actual values. Mean absolute percentage error (MAPE) expresses prediction error as a percentage, making it easy to compare the prediction accuracy of models across different data volumes. The mean absolute percentage error (sMAPE) measures the average absolute percentage difference between the predicted and the true values. It presents the error as a percentage, making it easier to understand and compare the prediction accuracy of different models or different datasets. The coefficient of determination (R²) measures the proportion of variation in the dependent variable that can be explained by the independent variable. R² ranges from 0 to 1. The closer it is to 1, the better the model fits the data, meaning that the independent variable can explain most of the variation in the dependent variable.

[0054] These accuracy metrics allow us to precisely assess how close the model's predictions are to the actual values, reflecting the model's predictive accuracy.

[0055] Calculating stability metrics: Model stability is one of the important factors in measuring its performance. To evaluate the stability of candidate models, the model evaluation agent calculates stability metrics based on the prediction accuracy of the candidate model on the validation set and the prediction accuracy on the test set.

[0056] For example, the prediction accuracy index deviation rate can be calculated. As a stability indicator,

[0057] in, This is the prediction performance metric of the candidate model on the validation set. This is the prediction performance metric for the candidate model on the test set.

[0058] A lower stability metric indicates that the candidate model performs consistently across different datasets and has good stability; conversely, a higher metric indicates poor model stability. Calculating stability metrics allows us to understand the performance fluctuations of a model across different data subsets, providing an important reference for selecting a stable and reliable model.

[0059] Besides accuracy and stability, inference efficiency is also a crucial factor to consider in practical applications. The model evaluation agent calculates inference efficiency metrics based on the time taken for candidate models to obtain prediction results. For example, the time required for the model to process test set data and output prediction results can be directly used as an inference efficiency metric, or related metrics such as the amount of data the model can process per unit time can be calculated. Shorter inference time or higher data processing efficiency means that the model can provide prediction results more quickly in practical applications, meeting real-time requirements.

[0060] The model evaluation agent performs a comprehensive evaluation based on the accuracy, stability, and inference efficiency metrics calculated above, using a weighted average method. Each metric is assigned a corresponding weight according to the different requirements for model accuracy, stability, and inference efficiency in actual application scenarios. Through weighted calculation, a comprehensive evaluation result reflecting the performance of candidate models is obtained, providing an accurate basis for selecting the most suitable target model from among many candidate models.

[0061] Optionally, the model evaluation agent is also used for: The weights of the accuracy index, the stability index, and the inference efficiency index are determined based on the electricity price forecast scenario, and the weights are used for weighted calculation to obtain the evaluation result.

[0062] In this embodiment, the Analytic Hierarchy Process (AHP) is used to determine the weights of multi-dimensional evaluation indicators. The weights are dynamically adjusted based on power forecasting demand (day-ahead / intraday, accuracy priority / efficiency priority). Finally, a weighted scoring method is used to comprehensively evaluate each candidate model, and the model with the highest score is selected as the target model. : Dynamic setting of indicator weights: For intraday real-time price predictions (accuracy priority + real-time performance), set the prediction accuracy weight. Stability weights Inference efficiency weight If it is a day-ahead price forecast (robustness), set a weight for forecast accuracy. Stability weights Weight of reasoning efficiency index The weights can be manually adjusted or automatically optimized according to user needs. Indicator standardization: The evaluation indicators are normalized and oriented to eliminate differences in the dimensions and directions of the indicators; Overall score calculation: ,in The first The standardized scores of the prediction accuracy, stability, and inference efficiency of each candidate model.

[0063] Optionally, determining the target model from the candidate models based on the evaluation results includes: The candidate model with the highest evaluation result is selected as the target model.

[0064] In this embodiment, a comprehensive score is selected. The highest model as the target Simultaneously output The results of the model feature importance analysis.

[0065] Optionally, the feature evaluation agent is used for: If the prediction accuracy is lower than a preset threshold, the model evaluation agent is instructed to reselect the target model. It is also used to determine the feature variable to be replaced and to re-mine candidate feature variables to replace the feature variable to be replaced.

[0066] In this embodiment, the feature evaluation agent is used to receive feedback from the electricity price prediction agent. Predicted value at time And at the next moment, that is At any given moment, collect the true value of electricity prices. The prediction accuracy is determined based on a second difference between the actual value and the predicted value; the smaller the second difference, the higher the prediction accuracy.

[0067] If the prediction accuracy is lower than the preset threshold, it indicates that the prediction accuracy is low, meaning that the target model currently used for prediction has low prediction accuracy, and the model needs to be re-retrieved and retrained for prediction.

[0068] Furthermore, the feature variables used may have a low correlation with electricity prices, requiring a re-evaluation of feature variables to find those with a higher correlation to electricity prices.

[0069] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the system as described in the foregoing system embodiments.

[0070] To implement the above embodiments, this application also proposes a computer program product having a computer program stored thereon, which, when executed by a processor, implements the system as described in the foregoing system embodiments.

[0071] To implement the above embodiments, this application also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the system as described in the foregoing system embodiments.

[0072] Figure 2 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. For example, the electronic device 800 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0073] Reference Figure 2 The electronic device 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input / output (I / O) interface 812, sensor component 814, and communication component 816.

[0074] Processing component 802 typically controls the overall operation of electronic device 800, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the system described above. Furthermore, processing component 802 may include one or more modules to facilitate interaction between processing component 802 and other components. For example, processing component 802 may include a multimedia module to facilitate interaction between multimedia component 808 and processing component 802.

[0075] Memory 804 is configured to store various types of data to support the operation of electronic device 800. Examples of this data include instructions for any application or system operating on electronic device 800, contact data, phonebook data, messages, pictures, videos, etc. Memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0076] Power component 806 provides power to various components of electronic device 800. Power component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 800.

[0077] Multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 808 includes a front-facing camera and / or a rear-facing camera. When the electronic device 800 is in an operating mode, such as a shooting mode or a video mode, the front-facing camera and / or the rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0078] Audio component 810 is configured to output and / or input audio signals. For example, audio component 810 includes a microphone (MIC) configured to receive external audio signals when electronic device 800 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 804 or transmitted via communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

[0079] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0080] Sensor assembly 814 includes one or more sensors for providing state assessments of various aspects of electronic device 800. For example, sensor assembly 814 can detect the on / off state of electronic device 800, the relative positioning of components such as the display and keypad of electronic device 800, changes in position of electronic device 800 or a component of electronic device 800, the presence or absence of user contact with electronic device 800, orientation or acceleration / deceleration of electronic device 800, and temperature changes of electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 814 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0081] Communication component 816 is configured to facilitate wired or wireless communication between electronic device 800 and other devices. Electronic device 800 can access wireless networks based on communication standards, such as WiFi, 4G, or 5G, or combinations thereof. In one exemplary embodiment, communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 816 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0082] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the system described above.

[0083] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, which can be executed by a processor 820 of an electronic device 800 to complete the system described above. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0084] To implement the above embodiments, this application also proposes a chip, including: the chip includes processing circuitry configured to execute the system provided in the foregoing embodiments.

[0085] Figure 3 This is a schematic diagram of the structure of a chip according to an embodiment of this application. See also... Figure 3 The diagram shown is a schematic representation of the structure of chip 1100, but it is not limited to this.

[0086] Chip 1100 includes processing circuitry 1101, which is configured to execute any of the above systems.

[0087] In some embodiments, chip 1100 further includes one or more interface circuits 1102. Optionally, the interface circuit 1102 is connected to memory 1103, and the interface circuit 1102 can be used to receive signals from memory 1103 or other devices, and the interface circuit 1102 can be used to send signals to memory 1103 or other devices. For example, the interface circuit 1102 can read instructions stored in memory 1103 and send the instructions to processing circuit 1101.

[0088] In some embodiments, the interface circuit 1102 performs at least one of the communication steps such as sending and / or receiving in the above system, while the processing circuit 1101 performs other steps.

[0089] In some embodiments, the terms interface circuit, interface, transceiver pin, transceiver, etc., can be used interchangeably.

[0090] In some embodiments, chip 1100 further includes one or more memories 1103 for storing instructions. Optionally, all or part of the memories 1103 may be located outside of chip 1100.

[0091] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0092] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0093] Any process or system description in the flowchart or otherwise described herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0094] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0095] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or systems can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0096] Those skilled in the art will understand that all or part of the steps of the system implementing the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the system embodiments.

[0097] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0098] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A multi-agent intelligent computing cloud platform system for electricity price prediction based on autonomous model evaluation, characterized in that, include: Data acquisition agent, data preprocessing agent, electricity price prediction agent, model evaluation agent, feature evaluation agent; The data acquisition agent is used to collect feature data based on power production-related feature variables, and send the feature data to the data preprocessing agent. The data preprocessing agent is used to preprocess the feature data and send the preprocessed feature data to the electricity price prediction agent. The model evaluation agent is used to autonomously discover candidate models for electricity price and electricity consumption prediction scenarios using training data corresponding to the feature variables, construct a candidate model set, and train and test multiple candidate models; it is also used to perform quantitative evaluation of the candidate models based on stability and accuracy to obtain evaluation results, and determine the target model from the candidate models based on the evaluation results. The electricity price prediction agent is used to input the preprocessed feature data into the target model for prediction and output the predicted value of the electricity price. The feature evaluation agent is used to evaluate the prediction accuracy based on the actual value and the predicted value of the electricity price, and to instruct the model evaluation agent to update the target model based on the prediction accuracy.

2. The system according to claim 1, characterized in that, The model evaluation agent is used for: Based on the characteristics of the aforementioned feature data, candidate models suitable for electricity price forecasting are retrieved; wherein, the candidate models include at least one of the following categories: statistical models, machine learning models, deep learning models, and large models; The training data is processed in a targeted manner according to the type of the candidate model, so that the training data is adapted to the corresponding candidate model; The candidate model is trained using the specially processed training data, and the trained candidate model is used to make predictions to obtain prediction results.

3. The system according to claim 2, characterized in that, The step of retrieving candidate models suitable for electricity price forecasting based on the characteristics of the feature data includes: Determine the characteristics of the feature data in the time series, and determine the first search keyword; wherein, the characteristics include at least one of the following: stationary time series, nonlinear high-dimensional feature time series, and long-term correlated nonlinear time series; Identify secondary search keywords related to electricity prices; The first search keyword and the second search keyword are combined to perform model retrieval and determine the candidate model.

4. The system according to claim 3, characterized in that, The targeted processing of the training data according to the type of candidate model includes any one of the following: For the training data input to the statistical model, a time series stationarity test is performed, and the non-stationary training data is differentially processed. For the training data input to the machine learning model, feature dimensionality reduction is performed to remove redundant features from the training data; The training data input to the deep learning model and the large model are subjected to time series sliding window processing to generate training data in the input sequence format.

5. The system according to claim 4, characterized in that, The step of training the candidate model with the specifically processed training data includes: The training data is divided into a training set, a validation set, and a test set; The candidate model is trained using the training set, and the hyperparameters in the candidate model are optimized based on the validation set to determine the hyperparameter combination corresponding to the candidate model, so as to obtain the trained candidate model.

6. The system according to claim 5, characterized in that, The step of using the trained candidate model to make predictions to obtain prediction results includes: The test set is input into the trained candidate model to obtain the prediction result.

7. The system according to claim 6, characterized in that, The quantitative evaluation of the candidate models based on stability and accuracy to obtain evaluation results includes: Calculate the accuracy index based on the predicted results and the actual results in the test set; Based on the prediction accuracy of the candidate model on the validation set and the prediction accuracy of the candidate model on the test set, a stability index is calculated. Calculate the inference efficiency index based on the time taken to obtain the prediction result from the candidate model; The evaluation result is obtained by weighting the accuracy index, the stability index, and the inference efficiency index.

8. The system according to claim 7, characterized in that, The model evaluation agent is also used for: The weights of the accuracy index, the stability index, and the inference efficiency index are determined based on the electricity price forecast scenario, and the weights are used for weighted calculation to obtain the evaluation result.

9. The system according to claim 8, characterized in that, The step of determining the target model from the candidate models based on the evaluation results includes: The candidate model with the highest evaluation result is selected as the target model.

10. The system according to claim 1, characterized in that, The feature evaluation agent is used for: If the prediction accuracy is lower than a preset threshold, the model evaluation agent is instructed to reselect the target model. It is also used to determine the feature variable to be replaced and to re-mine candidate feature variables to replace the feature variable to be replaced.

11. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the system as described in any one of claims 1-10.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the system as described in any one of the preceding claims 1-10.

13. A chip, characterized in that, The chip includes processing circuitry configured to execute the system as described in any one of claims 1-10.

14. A computer program product, characterized in that, It includes a computer program, which, when executed by a processor, implements the system as described in any one of claims 1-10.

Images