Big data-based electricity selling average price abnormity identification system and method

Abstract

The invention relates to the field of electricity selling average price abnormity identification, and particularly discloses an electricity selling average price abnormity identification system and method based on big data, and the method comprises the steps: firstly obtaining historical electricity selling average price data, market factor data and user real-time electricity consumption behavior data, then carrying out the feature extraction and correlation analysis of the three data through a deep learning technology, and carrying out the recognition of the historical electricity selling average price data, the market factor data and the user real-time electricity consumption behavior data; and finally, through a classifier, whether an electricity selling average price abnormity early warning is generated is judged, so that an electricity selling company is helped to formulate a more effective pricing strategy and risk management measures in a fierce market to adapt to market changes.

Description

Technical Field

[0001] The present application relates to the field of identifying abnormal average electricity selling prices, and more specifically, to a system and method for identifying abnormal average electricity selling prices based on big data. Background Art

[0002] The continuous advancement of the market-oriented reform of the power industry has brought new opportunities and challenges to the power industry chain. The price mechanism in the power spot market has been gradually established, and the power sales market has become more active, but this has also raised the requirements for the professional level of practitioners. In this context, power sales companies are facing more intense market competition and prominent business risks. In order to cope with complex market data and price fluctuations, scientific decision-making has become crucial to ensure the company's profits.

[0003] In existing technologies, the identification of abnormal average electricity sales prices mainly relies on manual analysis and rules. This method is not only inefficient, but may also lead to inaccurate evaluation results due to human factors.

[0004] Therefore, a system and method for identifying abnormal average electricity selling prices based on big data is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a system and method for identifying abnormal average electricity selling price based on big data, which first obtains historical average electricity selling price data, market factor data and user real-time electricity consumption behavior data, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a classifier to determine whether to issue an abnormal average electricity selling price warning, thereby helping power sales companies to formulate more effective pricing strategies and risk management measures in a highly competitive market to adapt to market changes.

[0006] According to one aspect of the present application, a system for identifying abnormal average electricity selling price based on big data is provided, which includes:

[0007] The module for acquiring the average electricity price data is used to acquire the historical average electricity price data, market factor data and real-time electricity consumption behavior data of users;

[0008] The average electricity selling price data extraction module is used to extract the average electricity selling price change multimodal correlation feature vector and the user's real-time electricity consumption behavior semantic feature vector from the historical average electricity selling price data, the market factor data and the user's real-time electricity consumption behavior data;

[0009] The average electricity selling price abnormality warning judgment module is used to determine whether to issue an average electricity selling price abnormality warning based on the multimodal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior.

[0010] According to another aspect of the present application, a method for identifying abnormal average electricity selling price based on big data is provided, which includes:

[0011] Obtain historical average electricity sales price data, market factor data, and real-time electricity consumption behavior data of users;

[0012] Extracting a multimodal correlation feature vector of average electricity selling price change and a semantic feature vector of user real-time electricity consumption behavior from the historical average electricity selling price data, the market factor data and the user real-time electricity consumption behavior data;

[0013] Based on the multimodal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior, it is determined whether to issue an abnormal warning of the average electricity selling price.

[0014] Compared with the prior art, the present application provides a system and method for identifying average electricity selling price anomalies based on big data. The system first obtains historical average electricity selling price data, market factor data, and user real-time electricity consumption behavior data, and then uses deep learning technology to perform feature extraction and correlation analysis on the three. Finally, a classifier is used to determine whether to issue an average electricity selling price anomaly warning, thereby helping power sales companies to formulate more effective pricing strategies and risk management measures in the highly competitive market to adapt to market changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a system for identifying abnormal average electricity selling price based on big data according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of an average electricity selling price data extraction module in a system for identifying average electricity selling price anomalies based on big data according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a historical average electricity selling price feature extraction unit in a system for identifying average electricity selling price anomalies based on big data according to an embodiment of the present application.

[0019] Figure 4 The present invention is a block diagram of an abnormal warning module for the average electricity selling price in the average electricity selling price abnormality identification system based on big data according to an embodiment of the present application.

[0020] Figure 5The present invention is a flowchart of a method for identifying anomalies in average electricity selling prices based on big data according to an embodiment of the present application.

[0021] Figure 6 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0023] Figure 1 : is a block diagram of a system for identifying abnormal average electricity prices based on big data according to an embodiment of the present application. Figure 1 As shown, according to the big data-based average electricity selling price anomaly identification system 100 according to the embodiment of the present application, it includes: an average electricity selling price data acquisition module 110, which is used to obtain historical average electricity selling price data, market factor data and user real-time electricity consumption behavior data; an average electricity selling price data extraction module 120, which is used to extract the average electricity selling price change multimodal correlation feature vector and the user real-time electricity consumption behavior semantic feature vector from the historical average electricity selling price data, the market factor data and the user real-time electricity consumption behavior data; an average electricity selling price anomaly warning judgment module 130, which is used to judge whether to issue an average electricity selling price anomaly warning based on the average electricity selling price change multimodal correlation feature vector and the user real-time electricity consumption behavior semantic feature vector.

[0024] In the above-mentioned average electricity price anomaly identification system 100 based on big data, the average electricity price data acquisition module 110 is used to obtain historical average electricity price data, market factor data and real-time electricity consumption behavior data of users. It should be understood that with the continuous advancement of the market-oriented reform of electricity, the power industry chain has ushered in new opportunities and challenges. The price mechanism of the electricity spot market has been gradually established, making the electricity sales market more active, which has also increased the requirements for the professional level of practitioners. In this context, power sales companies are facing increasingly fierce market competition and significant business risks. In order to effectively respond to complex market data and price fluctuations, scientific decision-making has become extremely important to ensure that the company can make stable profits. However, the current technical means mainly rely on manual analysis and fixed rules when identifying abnormal average electricity prices. This method is not only inefficient, but also may lead to inaccurate judgment results due to human factors. Therefore, in the technical solution of the present application, by obtaining historical average electricity sales price data, market factor data and users' real-time electricity consumption behavior data, and combining deep learning technology, it is determined whether it is necessary to issue an abnormal warning of the average electricity sales price, so as to help power sales companies formulate more accurate pricing strategies and risk management measures in a highly competitive market environment, so as to better adapt to market changes.

[0025] Specifically, obtaining historical average electricity price data, market factor data, and user real-time electricity consumption behavior data is a key preliminary step in establishing an abnormal average electricity price identification system based on big data. First, historical average electricity price data refers to the electricity price information within a certain time range in the past. These data include the average electricity price at each time point, price fluctuations, and cyclical trends in price changes. Secondly, market factor data covers various factors that affect electricity market prices. These factors include but are not limited to changes in electricity supply and demand, policy adjustments, fuel price fluctuations, weather conditions, and economic indicators. These data help the system understand market dynamics and their potential impact on the average electricity price, thereby accurately predicting prices and detecting anomalies in a variety of market environments. Finally, user real-time electricity consumption behavior data involves the user's electricity usage in the current time period. These data include the user's real-time electricity consumption, electricity consumption mode, load changes, and short-term trends in electricity consumption behavior. Through the comprehensive analysis of these diverse data, the system can better understand the changing rules of the average electricity price and issue early warnings when anomalies occur.

[0026] In the above-mentioned average electricity price anomaly identification system 100 based on big data, the average electricity price data extraction module 120 is used to extract the multimodal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior from the historical average electricity price data, the market factor data and the user's real-time electricity consumption behavior data. In this way, the complex influencing factors of electricity price changes can be structured and quantified, so that the system can understand and analyze the laws of electricity price changes from multiple dimensions. This comprehensive feature extraction method can effectively improve the accuracy and comprehensiveness of the anomaly identification system, so that the system can reliably issue abnormal warnings in a changing market environment, thereby optimizing the operating efficiency and stability of the power market.

[0027] Figure 2 FIG. 1 is a block diagram of an average electricity price data extraction module in a system for identifying average electricity price anomalies based on big data according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the average electricity selling price data extraction module 120 includes: a historical average electricity selling price feature extraction unit 121, which is used to perform feature extraction on the historical average electricity selling price data to obtain a historical average electricity selling price semantic association feature vector; a market factor data feature extraction unit 122, which is used to perform feature extraction on the market factor data to obtain an average electricity selling price change market factor semantic feature vector; an average electricity price change multimodal feature association unit 123, which is used to associate the historical average electricity selling price semantic association feature vector with the average electricity price change market factor semantic feature vector to obtain the average electricity price change multimodal association feature vector; a user real-time electricity consumption feature extraction unit 124, which is used to perform feature extraction on the user's real-time electricity consumption behavior data to obtain the user's real-time electricity consumption behavior semantic feature vector.

[0028] It should be understood that the historical average electricity price data records the changes in electricity prices over the past period of time, including the price level, fluctuation range and cyclical trend. By extracting features from these data, the basic patterns and laws of electricity prices can be extracted. These patterns and laws usually include seasonal fluctuations, long-term trends and cyclical changes, which are of great significance for identifying the normal fluctuation range and abnormal fluctuation of electricity prices. Feature extraction converts complex time series data into structured feature vectors, so that the data can be further analyzed and processed. The feature vector of the historical average electricity price contains not only the numerical information of the electricity price, but also the semantic information of the change of the electricity price. This semantic information may involve the upward or downward trend of the electricity price, the peak and valley of the price, the fluctuation range, etc. These features can more accurately reflect the changes in electricity prices and provide more accurate input data for anomaly detection.

[0029] Furthermore, market factor data covers various external influencing factors in the electricity market, such as fuel prices, policy changes, economic indicators, weather conditions, etc., which have a significant impact on electricity price fluctuations. Through feature extraction, these complex market factors can be transformed into structured feature vectors to capture their specific impact on electricity prices. This transformation enables these complex market dynamics to be quantified and analyzed, providing key background information for electricity price forecasting. Feature extraction compresses various information in market factor data into representative semantic feature vectors, which can accurately reflect the impact of market factors on electricity prices. The semantic feature vectors of these market factors can help establish electricity price models and reveal the relationship between different market factors and their combined effect on electricity prices.

[0030] Furthermore, associating the semantic association feature vector of the historical average electricity price and the semantic feature vector of the market factors of the average electricity price change to obtain the multimodal association feature vector of the average electricity price change is a key step to improve the accuracy and comprehensiveness of the electricity price prediction and anomaly detection system. Among them, the semantic association feature vector of the historical average electricity price and the semantic feature vector of market factors provide the internal pattern and external influence of electricity price changes respectively. The historical average electricity price feature vector captures the historical trend, cyclical fluctuations and laws of electricity prices, while the market factor feature vector reflects the impact of external factors on electricity prices. By associating these two feature vectors, the historical pattern of electricity prices can be combined with the current market conditions to form a comprehensive perspective, so as to more accurately understand the reasons for electricity price changes and their complex driving mechanisms.

[0031] In particular, feature extraction of real-time user electricity consumption behavior data is a key step to improve the operational efficiency of the power system and the accuracy of user demand forecasting. Real-time user electricity consumption behavior data provides instant information on power consumption, including power consumption, power consumption time, power consumption pattern, etc. By extracting features from these data, the core features of user electricity consumption behavior can be refined, such as daily peak power consumption, seasonal changes in power consumption, sudden load changes, etc. These feature vectors can accurately reflect the actual situation of user electricity consumption and provide dynamic demand data for the power system.

[0032] Figure 3 FIG. 1 is a block diagram of a historical average electricity price feature extraction unit in a system for identifying abnormal average electricity price based on big data according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the historical average electricity selling price feature extraction unit 121 includes: a historical average electricity selling price data preprocessing subunit 1211, used to perform data preprocessing on the historical average electricity selling price data to obtain a historical average electricity selling price text association feature matrix; a historical average electricity selling price convolutional encoding subunit 1212, used to pass the historical average electricity selling price text association feature matrix through a historical average electricity selling price text association convolutional neural network to obtain the historical average electricity selling price semantic association feature vector.

[0033] It should be understood that historical average electricity sales price data usually includes electricity price records in time series, which may contain noise, missing values ​​or outliers. Through data preprocessing, these data can be cleaned and sorted, and inaccurate information can be removed or corrected, thereby improving the quality and consistency of the data. These steps unify electricity price data from different time periods and different sources into a standard format and range. For example, through standardization, the dimensional effect in the data can be eliminated, so that electricity price data from different time periods can be compared on the same scale.

[0034] Furthermore, processing the historical average electricity price text association feature matrix through the historical average electricity price text association convolutional neural network (CNN) to obtain the historical average electricity price semantic association feature vector is an important step to improve the accuracy of electricity price forecasting and market analysis. The historical average electricity price text association feature matrix integrates the time series information of electricity price data and the text description information related to it. This feature matrix usually contains the time series data of electricity price changes and related market background information. By inputting these data into the convolutional neural network, the powerful feature extraction ability of CNN can be used to automatically learn representative semantic features from the complex matrix. This feature extraction can capture the potential patterns and trends in electricity price data and improve the understanding of electricity price changes. Among them, the convolutional neural network can effectively extract local features when processing text data, and refine information layer by layer through convolutional layers and pooling layers. In this way, CNN can capture the inherent laws of electricity price changes and convert them into semantic association feature vectors, which is crucial for understanding the complex reasons for electricity price changes. Specifically, each layer of the historical average electricity selling price text association convolutional neural network is used to perform convolution processing, mean pooling processing based on the local feature matrix, and nonlinear activation processing on the input data in the forward pass of the layer, so that the last layer of the historical average electricity selling price text association convolutional neural network outputs the historical average electricity selling price semantic association feature vector, wherein the input of the historical average electricity selling price text association convolutional neural network is the historical average electricity selling price text association feature matrix.

[0035] In a specific embodiment of the present application, the historical average electricity selling price data preprocessing subunit 1211 includes: converting the historical average electricity selling price data into multiple historical average electricity selling price text vectors; performing historical average electricity selling price feature encoding on the multiple historical average electricity selling price text vectors through multiple fully connected layers to obtain multiple historical average electricity selling price text feature vectors; and arranging the multiple historical average electricity selling price text feature vectors in two dimensions to obtain the historical average electricity selling price text association feature matrix.

[0036] It should be understood that historical average electricity price data usually contains rich time series information, which involves electricity price records and fluctuations at different time points. However, directly processing these raw data may ignore the semantic level and contextual information. By converting electricity price data into multiple historical average electricity price text vectors, the changing trends and patterns of electricity prices can be expressed in the form of text, which helps to convert time series data into a more structured and semantic form.

[0037] Furthermore, multiple historical average electricity price text vectors are respectively encoded with historical average electricity price features through multiple fully connected layers in order to deeply extract the features of electricity price data and effectively convert them into a format suitable for analysis and modeling. The fully connected layer plays an important role in the deep learning model, and is mainly responsible for further nonlinear transformation and integration of the input features. The historical average electricity price text vector is encoded with a fully connected layer, which can map the original text vector into a high-dimensional feature space. This mapping not only retains the key features of electricity price data, but also extracts more complex and abstract feature representations through multi-level nonlinear transformations. Among them, the use of multiple fully connected layers allows the model to learn and refine features at different levels. Each fully connected layer can abstract the data at different levels, thereby gradually extracting more representative features. Through this hierarchical processing, the model can identify more complex relationships from simple electricity price data, such as trends, cyclical changes in electricity prices, and their association with external factors. This hierarchical feature extraction helps to fully understand the information in electricity price data and form high-quality feature vectors. Specifically, each fully-connected layer is used to perform fully-connected encoding on the input vector in the forward pass of the layer to extract high-dimensional implicit features of the eigenvalues ​​at each position in the input vector, wherein the input of the first layer of the multiple fully-connected layers is the multiple historical average electricity selling price text vectors, the input of the second to last layers of the multiple fully-connected layers is the output vector of the previous layer, and the output of the last layer of the multiple fully-connected layers is multiple historical average electricity selling price text feature vectors.

[0038] Furthermore, the purpose of arranging multiple historical average electricity price text feature vectors in two dimensions to obtain the historical average electricity price text associated feature matrix is ​​to systematically organize and process the multidimensional features in the electricity price data, thereby improving the effect of data analysis and model training. The two-dimensional arrangement integrates multiple feature vectors into a matrix, which helps to display the multidimensional features of the electricity price data in a structured form. Each historical average electricity price text feature vector represents the electricity price information over a period of time. By arranging these vectors into a two-dimensional matrix, the relationship between the time series information and features of the electricity price data can be intuitively presented.

[0039] In a specific embodiment of the present application, the market factor data feature extraction unit 122 includes: passing the market factor data through a converter-based market factor context encoder to obtain a market factor semantic embedding feature vector; passing the market factor semantic embedding feature vector through a long short-term memory neural network model-based market factor semantic encoder to obtain the average electricity selling price change market factor semantic feature vector.

[0040] It should be understood that market factor data usually includes multiple types of information, such as economic indicators, policy changes, market sentiment, etc., which often have complex contextual relationships. Traditional feature extraction methods may not be able to effectively capture these contextual relationships, while the transformer-based context encoder can capture long-range dependencies in the data through the self-attention mechanism. The self-attention mechanism allows the model to pay attention to the impact of other related market factors when processing each market factor, thereby more comprehensively understanding the complex relationship between market factors. The transformer-based encoder can convert market factor data into semantic embedding feature vectors, which can effectively compress high-dimensional and complex market factor information into a low-dimensional, easy-to-process form. Through this conversion, market factor data can be represented as vectors with semantic meaning, which not only retain the key information in the original data, but also enhance the model's ability to understand the data. Specifically, the embedding layer of the transformer-based market factor context encoder is used to convert the market factor data into market factor embedding vectors to obtain a sequence of market factor embedding vectors; the transformer-based Bert model of the transformer-based market factor context encoder is used to perform global context semantic encoding on the sequence of market factor embedding vectors and cascade to obtain market factor semantic embedding feature vectors.

[0041] Furthermore, the semantic embedding feature vector of market factors is passed through the market factor semantic encoder based on the long short-term memory (LSTM) neural network model to obtain the semantic feature vector of market factors of average electricity price change, in order to more effectively capture the long-term dependence of market factors on electricity price changes and enhance the prediction ability and explanatory power of the model. The semantic embedding feature vector of market factors represents the deep semantic information of market data and provides a detailed understanding of market factors. However, the impact of market factors on the average electricity price is often not instantaneous, but has time continuity. Therefore, using only static semantic embedding vectors cannot fully reflect the dynamic impact of market factors on electricity price changes. The market factor semantic encoder based on LSTM can effectively process such time series data and capture the long-term dependence of market factors on electricity price changes. LSTM network is particularly suitable for processing time series data because it has a gating mechanism that can remember and forget important information. The "memory unit" and "forget gate" of LSTM can help the network effectively maintain long-term dependent information while ignoring irrelevant short-term fluctuations. By inputting the semantic embedding feature vector of market factors into the LSTM model, the encoder can extract key information that has a lasting impact on the changes in the average electricity price from historical market factors, thereby generating a more predictive semantic feature vector. The LSTM model can handle complex time series patterns and has good stability and generalization capabilities. In electricity price forecasting, market factors often exhibit periodic and nonlinear characteristics. LSTM can capture these complex time dependencies through its gating mechanism and multi-layer structure. Passing the semantic embedding feature vector of market factors to the LSTM model can better reveal these complex dynamic characteristics and improve the prediction accuracy of electricity price changes.

[0042] In a specific embodiment of the present application, the user real-time electricity consumption feature extraction unit 124 includes: passing the user real-time electricity consumption behavior data through a user real-time electricity consumption behavior word embedding encoder to obtain a user real-time electricity consumption behavior word embedding vector sequence; passing the user real-time electricity consumption behavior word embedding vector sequence through a user real-time electricity consumption behavior semantic extractor based on a Taming model to obtain the user real-time electricity consumption behavior semantic feature vector.

[0043] It should be understood that the real-time electricity consumption behavior data of users usually includes time series data of electricity consumption, which presents complex time series characteristics, such as peak electricity consumption periods, daily electricity consumption patterns, and seasonal changes. Directly processing these high-dimensional data may not fully reveal their inherent semantics and patterns. Through the word embedding encoder, the original electricity consumption behavior data can be converted into a sequence of word embedding vectors, which can better capture and express the key characteristics and laws of the user's electricity consumption behavior. Specifically, the user's real-time electricity consumption behavior data is segmented to obtain a sequence of electricity consumption behavior words; the embedding layer of the user's real-time electricity consumption behavior word embedding encoder is used to map each electricity consumption behavior word in the electricity consumption behavior word sequence to a word embedding vector to obtain a sequence of user real-time electricity consumption behavior word embedding vectors.

[0044] Furthermore, the user's real-time electricity consumption behavior word embedding vector sequence is processed by the user's real-time electricity consumption behavior semantic extractor based on the Taming model in order to extract meaningful and semantic information from the high-dimensional embedding vector of the user's electricity consumption behavior, thereby improving the understanding and prediction capabilities of the user's electricity consumption pattern. After the user's real-time electricity consumption behavior data is converted into a vector sequence through the word embedding encoder, it has represented the user's electricity consumption behavior in a high-dimensional and dense form. Although these vectors contain rich semantic information, they may still be too complex and lengthy to be directly used for subsequent analysis and model training. The semantic extractor based on the Taming model can effectively extract key semantic features from these word embedding vectors, reduce the dimension of the data, and retain the most representative electricity consumption behavior pattern. Specifically, the user's real-time electricity consumption behavior word embedding vector sequence is encoded based on global context semantics using the converter-based Bert model of the user's real-time electricity consumption behavior semantic extractor based on the Taming model to obtain multiple electricity consumption behavior feature vectors; and the multiple electricity consumption behavior feature vectors are cascaded to obtain the user's real-time electricity consumption behavior semantic feature vector.

[0045] In the above-mentioned average power selling price abnormality identification system 100 based on big data, the average power selling price abnormality warning judgment module 130 is used to determine whether to issue an average power selling price abnormality warning based on the multimodal correlation feature vector of the average power selling price change and the semantic feature vector of the user's real-time power consumption behavior. It should be understood that the multimodal correlation feature vector of the average power selling price change covers information from different data sources and modes, such as historical power selling prices, market demand fluctuations, climate factors, power supply conditions, etc. These feature vectors can reveal the complex factors and trends of power selling price changes, and help detect possible price anomalies by capturing the patterns and anomalies of price changes. Fusion of these multimodal data into a feature vector can more comprehensively reflect the overall picture of market price changes and provide a multi-dimensional analysis basis. The semantic feature vector of the user's real-time power consumption behavior provides in-depth information on the user's power consumption behavior, including power consumption patterns, peak hours, consumption habits, etc. By analyzing these feature vectors, the user's sensitivity and response to power price changes can be understood, thereby identifying whether there is abnormal power consumption behavior associated with price fluctuations. Combining these two feature vectors for judgment can help to comprehensively analyze whether the change in the average electricity price truly reflects the normal fluctuation of the market or is caused by some abnormal situation. The comprehensive judgment based on these two feature vectors can improve the accuracy and reliability of the early warning system. By integrating multi-modal market data and user behavior data, it can reduce misjudgments caused by a single data source or feature.

[0046] Figure 4 FIG. 1 is a block diagram of an abnormal warning module for the average electricity price in the average electricity price abnormality identification system based on big data according to an embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, the average electricity selling price abnormal warning judgment module 130 includes: an average electricity selling price feature fusion unit 131, which is used to perform multi-source data fusion based on perspective projection coordination on the multimodal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior to obtain an average electricity selling price abnormal warning classification feature vector; an average electricity selling price abnormal classification judgment unit 132, which is used to pass the average electricity selling price abnormal warning classification feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue an average electricity selling price abnormal warning.

[0047] It should be understood that the fusion of these two types of feature vectors can evaluate the abnormal situation of the average electricity price in a more comprehensive context and effectively classify it. The fused classification feature vector helps to reveal the potential causes of price anomalies and improve the accuracy of the early warning system by combining multi-dimensional information of market and user data. Doing so can not only reduce misjudgments caused by a single data source, but also better adapt to the complexity of market dynamics and user behavior. Through this comprehensive analysis, the early warning system can monitor and predict price anomalies in real time, issue early warnings in a timely manner, and prevent market fluctuations from adversely affecting users.

[0048] In particular, in the technical solution of the present application, considering that the multimodal correlation feature vector of the average electricity price change is extracted from the historical average electricity price data and market factor data, the main focus is on the historical changes in the average electricity price and the impact of market factors. The semantic feature vector of the user's real-time electricity behavior is extracted from the user's real-time electricity behavior data, focusing on the user's electricity behavior pattern and real-time electricity consumption. The multimodal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity behavior are extracted from different data sources and processing paths, respectively, and they represent different aspects of the average electricity price change and the user's electricity behavior. Due to the essential differences in these features and the different extraction methods, these two feature vectors have significant feature heterogeneity. This heterogeneity may cause the following problems in the feature fusion process: First, the multimodal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity behavior may exist on different manifolds, which means that they have different distribution laws in their respective spaces. Directly splicing them together may result in the final feature vector not having a clear manifold in the new space, that is, the feature manifold smoothness is insufficient. Second, since the two feature vectors come from different data sources and processing paths, they may not be well mapped to each other in the fused feature space, that is, one type of feature may not accurately reflect the feature changes of another type, resulting in increased uncertainty in feature expression. In order to solve the above technical problems, in the technical solution of this application, multi-source data fusion based on perspective projection coordination is performed on the multimodal correlation feature vector of the average price change of electricity sales and the semantic feature vector of the user's real-time electricity consumption behavior.

[0049] Specifically, in an embodiment of the present application, a multi-source data fusion based on perspective projection coordination is performed on the multi-modal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior to obtain a classification feature vector of abnormal warning of the average electricity selling price, including: constructing a distance topological matrix of the average electricity selling price change-user real-time electricity consumption behavior between the multi-modal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior; constructing a granularity correlation matrix of the average electricity selling price change-user real-time electricity consumption behavior feature value between the multi-modal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior; and classifying the average electricity selling price based on the average electricity selling price change-user real-time electricity consumption behavior distance topological matrix. The change-user real-time electricity consumption behavior feature value granularity association matrix is ​​topologically associated modulated to obtain the average electricity selling price change-user real-time electricity consumption behavior topological modulation association matrix; the average electricity selling price change-user real-time electricity consumption behavior topological modulation association matrix is ​​used as the perspective co-projection space, and the average electricity selling price change multimodal association feature vector and the user real-time electricity consumption behavior semantic feature vector are projected into the perspective co-projection space to obtain the perspective-modulated average electricity selling price change multimodal association feature vector and the perspective-modulated user real-time electricity consumption behavior semantic feature vector; the perspective-modulated average electricity selling price change multimodal association feature vector and the perspective-modulated user real-time electricity consumption behavior semantic feature vector are fused to obtain the average electricity selling price abnormal warning classification feature vector.

[0050] In the embodiment of the present application, specifically, another implementation expression of performing multi-source data fusion based on perspective projection coordination on the multi-modal correlation feature vector of the average power selling price change and the semantic feature vector of the user's real-time power consumption behavior to obtain the average power selling price abnormal warning classification feature vector can be: the multi-modal correlation feature vector of the average power selling price change and the semantic feature vector of the user's real-time power consumption behavior are processed by the following formula to obtain the average power selling price abnormal warning classification feature vector; wherein, the formula is:

[0051] M t =D(V1,V2)

[0052]

[0053] V c =α·V1′+β·V2′

[0054] Wherein, V1 represents the multimodal correlation feature vector of the average electricity price change, V2 represents the semantic feature vector of the user's real-time electricity consumption behavior, and D(V1, V2) is the distance topological matrix between the multimodal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior, that is, d i,j =d(v 1i ,v 2j ), di,j is the eigenvalue of the position (i, j) in the distance topology matrix, d(v 1i ,v 2j ) represents the distance between the eigenvalue of the ith position of the multimodal correlation feature vector of the average electricity price change and the eigenvalue of the jth position of the semantic feature vector of the user's real-time electricity consumption behavior, and V1 and V2 are both column vectors, M t represents the distance topology matrix, T represents the transpose of the vector, Represents matrix multiplication, M v represents the eigenvalue granularity correlation matrix, exp represents the natural exponential function, M c represents the topological modulation association matrix, V1′ represents the multimodal association feature vector of the average price change of electricity sales from the perspective modulation, V2′ represents the semantic feature vector of the user's real-time electricity consumption behavior from the perspective modulation, α and β represent weighted hyperparameters, V c Represents the classification feature vector of abnormal warning of average electricity price.

[0055] That is, in the technical solution of the present application, a multi-source data fusion based on perspective projection coordination is performed on the multi-modal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior. First, a distance topological matrix of the average electricity price change-user's real-time electricity consumption behavior is constructed between the multi-modal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior. This step utilizes the idea of ​​graph theory, taking the eigenvalues ​​of each position in the multi-modal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior as nodes, and using the distance metric function to determine the low-dimensional embedding expression of the edges between nodes, so as to construct a feature expression that can quantify the topological correlation relationship between the multi-modal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior. This step is crucial to understanding the spatial distribution of data points and their mutual connections, and also provides basic structural information for subsequent steps.

[0056] Next, a granularity association matrix of the average electricity price change-user real-time electricity behavior feature value between the multimodal association feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity behavior is constructed. This step focuses on exploring the correlation between the internal attributes of the feature vectors. In a specific example, the product between the multimodal association feature vector of the average electricity price change and the transposed vector of the semantic feature vector of the user's real-time electricity behavior is calculated to obtain the granularity association matrix of the average electricity price change-user real-time electricity behavior feature value. Furthermore, based on the average electricity price change-user real-time electricity behavior distance topological matrix, the average electricity price change-user real-time electricity behavior feature value granularity association matrix is ​​topologically associated modulated to obtain the average electricity price change-user real-time electricity behavior topological modulation association matrix. That is, the eigenvalue granularity association information of the multimodal association feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior is driven to wander on the distance topology graph, so as to construct a method for mapping the multimodal association feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior to a low-dimensional modulated perspective co-projection space domain with continuous high-dimensional regression space properties.

[0057] Next, the average price change of electricity sales-user real-time electricity consumption behavior topological modulation association matrix is ​​used as the perspective co-projection space, and the multimodal association feature vector of the average price change of electricity sales and the semantic feature vector of the user's real-time electricity consumption behavior are projected into the perspective co-projection space to obtain the perspective modulation average price change multimodal association feature vector and the perspective modulation user real-time electricity consumption behavior semantic feature vector. In other words, the key to this technical solution is to find an implicit third space with modulation capability, in which as much information of the original high-dimensional data as possible is retained, so that the converted new feature vector not only reduces the dimension, but also maintains the relative position relationship between the original features.

[0058] Finally, the multimodal correlation feature vector of the perspective-modulated average electricity price change and the semantic feature vector of the perspective-modulated user's real-time electricity consumption behavior are fused to obtain the average electricity price abnormal warning classification feature vector. Fusion can be achieved in a variety of forms, ranging from simple arithmetic averages to complex ensemble learning algorithms, depending on the application scenario and the goals pursued. The final average electricity price abnormal warning classification feature vector combines the main features of the two sets of input data and provides a more comprehensive data representation, which is not only conducive to improving the performance of machine learning models, but also promotes the ability to analyze cross-modal or multi-source data, thereby significantly enhancing the understanding and utilization efficiency of complex data structures.

[0059] Furthermore, by processing the classification feature vector of abnormal warning of average electricity price by the classifier, the complex data analysis process can be simplified into classification decision. For example, the classifier can map the classification feature vector of abnormal warning of average electricity price to the category of "normal" or "abnormal", so as to clarify whether there is price abnormality. The annotation information in the historical data is used in the training process of the classifier, so that it can learn the relationship between various market and user behaviors and price abnormalities, so that it can make accurate judgments when processing the classification feature vector of abnormal warning of average electricity price. The final classification result directly determines whether it is necessary to issue an abnormal warning of average electricity price. The purpose of this warning mechanism is to remind relevant parties to take necessary measures in time when the price changes exceed the normal range, so as to avoid market instability and unfair treatment of users. The classification result uses a clear "abnormal" label, so that the warning system can respond quickly and issue warning information to protect market stability and user interests.

[0060] In summary, the embodiment of the present application first obtains historical average electricity selling price data, market factor data and user real-time electricity consumption behavior data, and then uses deep learning technology to perform feature extraction and correlation analysis on the three, and finally uses a classifier to determine whether to issue an abnormal warning of the average electricity selling price, thereby helping power sales companies to formulate more effective pricing strategies and risk management measures in the highly competitive market to adapt to market changes.

[0061] As described above, the average power price anomaly identification system 100 based on big data according to the embodiment of the present application can be implemented in various terminal devices. In one example, the average power price anomaly identification system 100 based on big data can be integrated into the terminal device as a software module and / or hardware module. For example, the average power price anomaly identification system 100 based on big data can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the average power price anomaly identification system 100 based on big data can also be one of the many hardware modules of the terminal device.

[0062] Alternatively, in another example, the big data-based average electricity price anomaly identification system 100 and the terminal device may also be separate devices, and the big data-based average electricity price anomaly identification system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0063] Figure 5 Flow chart of a method for identifying abnormal average electricity price based on big data according to an embodiment of the present application. Figure 5As shown, according to the method for identifying abnormal average electricity selling price based on big data according to the embodiment of the present application, it includes: S110, acquiring historical average electricity selling price data, market factor data and user real-time electricity consumption behavior data; S120, extracting the multimodal correlation feature vector of average electricity selling price change and the semantic feature vector of user real-time electricity consumption behavior from the historical average electricity selling price data, the market factor data and the user real-time electricity consumption behavior data; S130, judging whether to issue an abnormal average electricity selling price warning based on the multimodal correlation feature vector of average electricity selling price change and the semantic feature vector of user real-time electricity consumption behavior.

[0064] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned method for identifying abnormal average electricity selling price based on big data have been referred to above. Figures 1 to 4 It has been introduced in detail in the description of the average electricity price anomaly identification system based on big data, and therefore, its repeated description will be omitted.

[0065] Below, reference Figure 6 To describe an electronic device according to an embodiment of the present application.

[0066] See also Figure 6 , Figure 6 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0067] The processor 11 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solution provided in the embodiment of the present application; the memory 12 can be implemented in the form of a read-only memory (Read Only Memory, ROM), a static storage device, a dynamic storage device or a random access memory (Random Access Memory, RAM), etc.

[0068] The memory 12 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 12, and the processor 11 calls and executes the method for identifying abnormal average electricity selling price based on big data in the embodiment of this application;

[0069] Input / output interface 13, used to implement information input and output;

[0070] The communication interface 14 is used to realize the communication interaction between the device and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0071] A bus 15 that transmits information between the various components of the device (e.g., the processor 11, the memory 12, the input / output interface 13, and the communication interface 14);

[0072] The processor 11 , the memory 12 , the input / output interface 13 and the communication interface 14 are connected to each other in communication within the device via a bus 15 .

[0073] An embodiment of the present application also provides a storage medium, which is a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for identifying abnormal average electricity selling price based on big data is implemented.

[0074] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0075] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0076] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0077] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0078] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0079] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0080] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0081] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0082] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0083] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method of each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs. The preferred embodiment of the embodiment of the present application is described above with reference to the accompanying drawings, and the scope of rights of the embodiment of the present application is not limited thereto. Any modifications, equivalent substitutions and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of rights of the embodiments of the present application.

Claims

1. A system for identifying abnormal average electricity prices based on big data, characterized in that: include: The module for acquiring the average electricity price data is used to acquire the historical average electricity price data, market factor data and real-time electricity consumption behavior data of users; The average electricity selling price data extraction module is used to extract the average electricity selling price change multimodal correlation feature vector and the user's real-time electricity consumption behavior semantic feature vector from the historical average electricity selling price data, the market factor data and the user's real-time electricity consumption behavior data; The average electricity selling price abnormality warning judgment module is used to determine whether to issue an average electricity selling price abnormality warning based on the multimodal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior.

2. The system for identifying abnormal average electricity selling price based on big data according to claim 1 is characterized in that: The electricity sales average price data extraction module includes: A historical average electricity selling price feature extraction unit, used to extract features from the historical average electricity selling price data to obtain a semantically associated feature vector of the historical average electricity selling price; A market factor data feature extraction unit, used to extract features from the market factor data to obtain a semantic feature vector of the market factor of the average electricity selling price change; The average power selling price change multimodal feature association unit is used to associate the semantic association feature vector of the historical average power selling price with the semantic feature vector of the market factor of the average power selling price change to obtain the multimodal association feature vector of the average power selling price change; The user real-time electricity usage feature extraction unit is used to extract features from the user real-time electricity usage behavior data to obtain a semantic feature vector of the user real-time electricity usage behavior.

3. The system for identifying abnormal average electricity selling price based on big data according to claim 2 is characterized in that: The historical average electricity selling price feature extraction unit includes: A historical average electricity selling price data preprocessing subunit, used for preprocessing the historical average electricity selling price data to obtain a historical average electricity selling price text correlation feature matrix; The historical average electricity selling price convolutional encoding subunit is used to pass the historical average electricity selling price text association feature matrix through the historical average electricity selling price text association convolutional neural network to obtain the historical average electricity selling price semantic association feature vector.

4. The system for identifying abnormal average electricity selling price based on big data according to claim 3 is characterized in that: The historical average electricity selling price data preprocessing subunit includes: Converting the historical average electricity selling price data into multiple historical average electricity selling price text vectors; The plurality of historical average electricity selling price text vectors are respectively subjected to historical average electricity selling price feature encoding through a plurality of fully connected layers to obtain a plurality of historical average electricity selling price text feature vectors; The plurality of historical average electricity selling price text feature vectors are arranged in two dimensions to obtain the historical average electricity selling price text associated feature matrix.

5. The system for identifying abnormal average electricity selling price based on big data according to claim 4 is characterized in that: The market factor data feature extraction unit comprises: Passing the market factor data through a converter-based market factor context encoder to obtain a market factor semantic embedding feature vector; The market factor semantics are embedded into the feature vector and passed through a market factor semantic encoder based on a long short-term memory neural network model to obtain the semantic feature vector of the market factor of the change in the average electricity selling price.

6. The system for identifying abnormal average electricity selling price based on big data according to claim 5 is characterized in that: The user real-time electricity consumption feature extraction unit includes: The user's real-time electricity usage behavior data is passed through a user's real-time electricity usage behavior word embedding encoder to obtain a user's real-time electricity usage behavior word embedding vector sequence; The user's real-time electricity usage behavior word embedding vector sequence is passed through a user's real-time electricity usage behavior semantic extractor based on a Taming model to obtain a user's real-time electricity usage behavior semantic feature vector.

7. The system for identifying abnormal average electricity selling price based on big data according to claim 6 is characterized in that: The power selling average price abnormality early warning judgment module includes: The average power selling price feature fusion unit is used to perform multi-source data fusion based on perspective projection coordination on the multi-modal correlation feature vector of the average power selling price change and the semantic feature vector of the user's real-time power consumption behavior to obtain an abnormal warning classification feature vector of the average power selling price; The average electricity selling price abnormality classification judgment unit is used to pass the average electricity selling price abnormality warning classification feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether to issue an average electricity selling price abnormality warning.

8. The system for identifying abnormal average electricity selling price based on big data according to claim 7 is characterized in that: The electricity selling average price feature fusion unit is used to: Constructing a distance topological matrix between the average electricity price change multimodal association feature vector and the user's real-time electricity consumption behavior semantic feature vector; Constructing a granular correlation matrix of average electricity price change-user real-time electricity consumption behavior feature value between the multimodal correlation feature vector of the average electricity price change and the semantic feature vector of the user's real-time electricity consumption behavior; Based on the average electricity price change-user real-time electricity consumption behavior distance topological matrix, the average electricity price change-user real-time electricity consumption behavior characteristic value granularity association matrix is ​​topologically associated modulated to obtain the average electricity price change-user real-time electricity consumption behavior topological modulation association matrix; Taking the average electricity price change-user real-time electricity consumption behavior topological modulation association matrix as the perspective co-projection space, projecting the average electricity price change multimodal association feature vector and the user real-time electricity consumption behavior semantic feature vector into the perspective co-projection space to obtain the perspective modulation average electricity price change multimodal association feature vector and the perspective modulation user real-time electricity consumption behavior semantic feature vector; The multimodal correlation feature vector of the perspective-modulated average electricity selling price change and the perspective-modulated user real-time electricity consumption behavior semantic feature vector are integrated to obtain the average electricity selling price abnormality warning classification feature vector.

9. A method for identifying abnormal average electricity sales price based on big data, characterized in that: include: Obtain historical average electricity sales price data, market factor data, and real-time electricity consumption behavior data of users; Extracting a multimodal correlation feature vector of average electricity selling price change and a semantic feature vector of user real-time electricity consumption behavior from the historical average electricity selling price data, the market factor data and the user real-time electricity consumption behavior data; Based on the multimodal correlation feature vector of the average electricity selling price change and the semantic feature vector of the user's real-time electricity consumption behavior, it is determined whether to issue an abnormal warning of the average electricity selling price.

10. The method for identifying abnormal average electricity selling price based on big data according to claim 9, characterized in that: Extracting a multimodal correlation feature vector of average electricity selling price change and a semantic feature vector of user real-time electricity consumption behavior from the historical average electricity selling price data, the market factor data and the user real-time electricity consumption behavior data, including: Performing feature extraction on the historical average electricity selling price data to obtain a semantically associated feature vector of the historical average electricity selling price; Performing feature extraction on the market factor data to obtain a semantic feature vector of the market factor of the average electricity selling price change; Associating the semantic association feature vector of the historical average electricity selling price with the semantic feature vector of the market factor of the average electricity selling price change to obtain the multimodal association feature vector of the average electricity selling price change; Feature extraction is performed on the user's real-time electricity usage behavior data to obtain a semantic feature vector of the user's real-time electricity usage behavior.

Images