Monitoring system and method for electricity market transaction
By combining quantum computing and multimodal data analysis with graph neural networks and convolutional long short-term memory networks, real-time risk management and automatic trading regulation in the power market have been achieved. This solves the problems of inaccurate market risk assessment and high operation and maintenance costs in existing systems, and improves market responsiveness and the intelligence level of trading strategies.
Patent Information
- Application Number
- CN202511036940.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-25
- Publication Date
- 2025-11-18
AI Technical Summary
Existing power market monitoring systems lack real-time dynamic monitoring and intelligent analysis capabilities, making them unable to effectively cope with complex market changes. This leads to inaccurate market risk assessments, increased operation and maintenance costs, and missed trading opportunities.
It employs a quantum computing engine to accelerate data processing, and combines a multimodal data fusion module, an adaptive quantitative risk assessment model, and a quantum reinforcement learning control module to achieve real-time risk identification and automatic trading control. It also uses graph neural networks and convolutional long short-term memory networks for market volatility prediction and risk management.
It improves the accuracy and responsiveness of market risk management, reduces human error, ensures market stability and decision support, and provides immediate and efficient market response capabilities.
Smart Images
Figure CN120975820A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of power market, in particular to a monitoring system and method for power market transaction. BACKGROUND
[0002] Currently, power market transactions are mainly conducted through power trading platforms, and major power market regulatory agencies, power production enterprises, consumers participating in the market, and transaction intermediaries also rely on information systems to support their transaction operations. However, most existing power market monitoring systems are based on traditional data information collection and management, with relatively single functions, mainly focusing on real-time display of market prices and transaction volumes and historical data queries. Most monitoring systems have a narrow application scope and lack in-depth analysis of complex market behavior, and cannot effectively predict and alarm potential market risks in real time. Existing monitoring systems are mostly manually operated, relying on human experience for judgment and data processing. This manual dependence not only increases the system operation and maintenance cost, but also has the risk of incorrect judgment and delayed response. Some systems provide automated data collection and processing functions, but still lack an effective market risk assessment system and lack in-depth analysis of transaction abnormal behavior, price fluctuations, and market liquidity, which cannot provide timely support for decision-making.
[0003] The biggest defect of the prior art is the lack of real-time dynamic monitoring and intelligent analysis capability of the power market. Most current systems cannot effectively respond to complex market changes, and when the power market experiences sharp fluctuations, traditional monitoring systems cannot timely capture market abnormal behavior and provide early warnings. Lack of intelligent data analysis and processing results in an inability to accurately judge risk factors and potential transaction errors in the market. Therefore, market participants cannot quickly respond to market changes, missing the best transaction opportunities or incurring unforeseen losses. SUMMARY
[0004] To address the deficiencies of the prior art, the present application provides a monitoring system and method for power market transactions, which solves the technical problem of how to achieve efficient and intelligent market risk management through real-time analysis of multi-modal data and dynamic optimization of risk factor weights.
[0005] To achieve the above objectives, the present application is implemented by the following technical solutions: a monitoring system and method for power market transactions, comprising: a quantum computing engine for accelerating real-time data processing of the power market; a multi-modal data fusion module for integrating multi-modal data from the power market; an adaptive quantitative risk assessment model for automatically identifying potential risks in the market based on real-time data streams; The intelligent early warning and automatic transaction regulation system is used for automatically issuing early warning signals and adjusting market transaction behavior according to preset rules according to real-time risk assessment results. The quantum reinforcement learning regulation module is used for self-learning and optimizing transaction strategies in the electricity market through quantum reinforcement learning technology, providing dynamic adjustment suggestions and real-time response to market fluctuations.
[0006] Preferably, the quantum computing engine adopts quantum parallel computing method, which accelerates the transaction data analysis process of the electricity market through quantum algorithm in multi-dimensional data processing, significantly improving the speed and accuracy of data processing.
[0007] Preferably, the multi-modal data includes price fluctuations, transaction volume, external economic factors, weather data, macroeconomic indicators, and public opinion information in social media.
[0008] A monitoring method for electricity market transactions, comprising: S1. Utilize quantum computing engine for multi-dimensional parallel data acceleration processing in real-time data stream of electricity market; S2. Use graph neural network to perform real-time fusion of multi-modal data, generate a comprehensive market situation map to predict market fluctuations and capture the correlation between multiple dimensions; S3. Based on the adaptive quantitative risk assessment model, use convolutional long short-term memory network to predict market fluctuations, automatically adjust risk assessment strategies, and improve the response speed and accuracy of market risk management; S4. Trigger market regulation mechanism automatically through intelligent early warning and automatic transaction regulation system, optimize transaction behavior and market strategy, automatically adjust high-risk transaction varieties, transaction volume, and ensure the consistency and compliance of strategy execution; S5. Use quantum reinforcement learning algorithm to adaptively optimize transaction strategies in the electricity market, automatically adjust decisions through real-time simulation of multiple market scenarios, thereby providing immediate and efficient market response capability, maximizing profits and reducing potential risks caused by market fluctuations.
[0009] Preferably, the graph neural network uses the following formula for data fusion and map generation: , Wherein: represents the layer node feature matrix of the graph, is the input initial feature, represents the normalized adjacency matrix, which represents the relationship between nodes in the graph structure, wherein is the adjacency matrix, is the unit matrix, which represents self-connection, is the first is the weight matrix of the layer, used for the transformation of node features, is the activation function, usually ReLU function, enhancing the non-linear relationship in graph generation, is the output feature matrix after graph convolution, used to update the feature representation of each node.
[0010] Preferably, the automatic adjustment of the risk assessment strategy specifically comprises the following steps: S3.1. inputting multi-modal data of the electricity market; S3.2. extracting spatial features through a convolution layer; S3.3. learning time dependence using the convolutional long short-term memory network; S3.4. combining convolutional features and time series features for market fluctuation prediction; S3.5. automatically adjusting the risk assessment strategy of the quantitative risk assessment model according to the prediction result; S3.6. finally outputting real-time risk assessment results and responding to risks.
[0011] Preferably, the convolution layer performs convolution operation on the input multi-modal data through a convolution kernel, converts local features of the multi-modal data into global features, and enables the model to better capture the spatial dependence of price fluctuations or trading volume.
[0012] Preferably, the automatic adjustment of the risk assessment strategy sets the following fluctuation thresholds: high fluctuation threshold, low fluctuation threshold and medium fluctuation threshold.
[0013] The present application provides a monitoring system and method for electricity market transactions. It has the following advantages: The monitoring system and method for electricity market transactions effectively capture the spatial and temporal dependence of data in the electricity market by combining convolutional long short-term memory network and graph neural network. By extracting spatial features through a convolution layer and combining them with time series features, the system can accurately predict market fluctuations and identify potential factors of market risk. Compared with traditional single time series analysis or spatial feature extraction, this method can provide more comprehensive market prediction ability, improving the prediction accuracy and real-time response ability of market fluctuations.
[0014] This scheme automatically adjusts the risk factor weight through an adaptive quantitative risk assessment model, dynamically optimizes the risk assessment strategy according to real-time fluctuation prediction, so that the system can flexibly respond to different market states. In high or low fluctuation state, the system can intelligently regulate the trading strategy to avoid excessive intervention or lack of response measures, thereby ensuring market stability and decision support for participants. This automatic adjustment mechanism greatly reduces the error of human operation, improves the efficiency and accuracy of risk management. BRIEF DESCRIPTION OF DRAWINGS
[0015] Figure 1 is a schematic diagram of an internal structure for implementing the invention. DETAILED DESCRIPTION
[0016] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.
[0017] As shown in Figure 1 , the embodiments of the present application provide a monitoring system and method for power market transactions, including a quantum computing engine for accelerating real-time data processing of the power market. The quantum computing engine uses quantum parallel computing methods to accelerate the transaction data analysis process of the power market through quantum algorithms when processing multi-dimensional data, significantly improving the speed and accuracy of data processing.
[0018] A multi-modal data fusion module is used to integrate multi-modal data from the power market, including price fluctuations, transaction volumes, external economic factors, weather data, macroeconomic indicators, and public opinion information in social media.
[0019] An adaptive quantitative risk assessment model is used to automatically identify potential risks in the market based on real-time data streams.
[0020] An intelligent early warning and automatic transaction regulation system is used to automatically issue warning signals and adjust market transaction behavior according to preset rules based on real-time risk assessment results.
[0021] A quantum reinforcement learning regulation module is used to self-learn and optimize transaction strategies in the power market through quantum reinforcement learning technology, providing dynamic adjustment suggestions and real-time responses to market fluctuations.
[0022] A monitoring method for power market transactions is characterized by comprising: S1. Utilize the quantum computing engine to perform multi-dimensional parallel data acceleration processing in the real-time data stream of the power market.
[0023] S2. Use a graph neural network to perform real-time fusion of multi-modal data to generate a comprehensive market situation map to predict market fluctuations and capture correlations between multiple dimensions. The graph neural network uses the following formula for data fusion and map generation: , Where: represents the Layer node feature matrix, is the initial feature of the input, represents the normalized adjacency matrix, representing the relationship between nodes in the graph structure, wherein is the adjacency matrix, is the identity matrix, representing self-connection, is the weight matrix of the first layer, used for transformation of node features, is the activation function, usually ReLU function, enhancing the nonlinear relationship in graph generation, is the output feature matrix after graph convolution, used to update the feature representation of each node.
[0024] S3. Based on the adaptive quantization risk assessment model, the market fluctuation is predicted by using convolutional long short-term memory network, the risk assessment strategy is automatically adjusted, the reaction speed and accuracy of market risk management are improved, and the automatic adjustment of risk assessment strategy is as follows: S3.1. Input the multi-modal data of the electricity market.
[0025] S3.2. Extract spatial features through convolutional layers, which convert local features of multi-modal data into global features through convolution kernel, so that the model can better capture the spatial dependence of price fluctuations or trading volume.
[0026] S3.3. Learn time dependence by using convolutional long short-term memory network.
[0027] S3.4. Combine convolutional features and time series features to predict market fluctuations.
[0028] S3.5. According to the prediction result, the risk assessment strategy of the quantization risk assessment model is automatically adjusted.
[0029] S3.6. Finally output the real-time risk assessment result and carry out risk response.
[0030] S4. Trigger the market regulation mechanism automatically through the intelligent early warning and automatic transaction regulation system, optimize the transaction behavior and market strategy, automatically adjust the high-risk transaction varieties and trading volume, and ensure the consistency and compliance of strategy execution.
[0031] S5. Utilize quantum reinforcement learning algorithm to adaptively optimize trading strategies in the electricity market, automatically adjust decisions through real-time simulation of multiple market scenarios, thereby providing immediate and efficient market response capabilities, maximizing profits and reducing potential risks brought by market fluctuations, and automatically adjusting risk assessment strategy settings. The following fluctuation thresholds are set: high fluctuation threshold, low fluctuation threshold and medium fluctuation threshold. The low fluctuation threshold is set to 20% of the market historical fluctuation range, the medium fluctuation threshold is based on the dynamic calculation value of the actual market, and is set to the median value between the low fluctuation threshold and the high fluctuation threshold, and the high fluctuation threshold is set to the 95th percentile of the market historical fluctuation range.
[0032] Embodiment Two
[0033] In this implementation, a graph neural network is used to analyze and integrate multi-modal data in the electricity market, and generate a comprehensive market scenario graph. This graph is used to predict market fluctuations and capture the relationships between different market dimensions.
[0034] Step 1: Construct market scenario graph
[0035] Each node in the graph structure represents a market scenario, and the relationship between nodes is represented by edges. The multi-modal data of the electricity market (including price fluctuations, transaction volume, external economic factors, weather data, etc.) is mapped into the graph structure, forming a multi-dimensional market scenario graph.
[0036] Node features: Each node contains a feature vector of the market. These features can be historical prices, transaction volumes, demand volumes, weather data, and other related market data.
[0037] For example, node represents the market price fluctuation at a certain time point, and node represents the transaction volume at that time point.
[0038] Edge definition: The edges in the graph represent the relationship between market scenarios. For example, market price fluctuations may be related to a certain weather condition, macroeconomic factor or policy change. Through the graph neural network, these relationships are revealed and learned.
[0039] Step 2: Graph neural network data fusion
[0040] The convolution operation in the graph neural network is used to integrate multi-modal data in the market. Specifically, the convolution operation propagates the relationship between nodes through the adjacency matrix. In each layer of graph convolution, the features of the nodes are affected by their neighbor nodes and are transformed through the weight matrix.
[0041] 1. Adjacency matrix : The adjacency matrix represents the relationship between nodes in a graph, containing the weight of the edge between each pair of nodes. To avoid excessive computational complexity, the adjacency matrix is usually normalized and represented as where is the original adjacency matrix, is the identity matrix, representing the connection of a node to itself (self-connection).
[0042] 2. Node Feature Matrix : The node feature matrix of each layer represents the feature representation of each node in the network at the th layer. The initial feature matrix is generated from the data of the electricity market (price, transaction volume, etc.).
[0043] For the 1st layer, the initial node feature is the original input feature of the market: , Step 3: Graph Convolution Operation The convolution operation in the graph neural network updates the node features through the adjacency matrix. In the th layer, the update formula for the node features is: , where: is the node feature matrix of the th layer.
[0044] is the weight matrix of the th layer, used for the transformation of node features.
[0045] is the normalized adjacency matrix, used to capture the relationship between nodes.
[0046] is the activation function, usually using the ReLU (Rectified Linear Unit) activation function to enhance the non-linear features in the network.
[0047] Step 4: Multi-layer Graph Convolution and Feature Learning
[0048] Through multi-layer graph convolution, the graph neural network can gradually extract higher-order node features. In each layer of graph convolution, the feature representation of a node is not only determined by its own features, but also influenced by the features of its neighbor nodes. The graph convolution operation can help the network learn the multi-dimensional correlation of market fluctuations, such as: The relationship between price fluctuations and transaction volume: the influence of price fluctuations on transaction volume, and the feedback effect of transaction volume fluctuations on price.
[0049] Weather data and market prices: How changes in weather conditions affect electricity demand and market prices.
[0050] Macroeconomic and market behavior: How macroeconomic changes affect market prices and demand.
[0051] Through multi-layer convolution, the network can capture more and more complex market fluctuation patterns and correlation features.
[0052] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A monitoring system for electricity market transactions, characterized in that, include: A quantum computing engine for accelerating real-time data processing in the electricity market; A multimodal data fusion module is used to integrate multimodal data from the electricity market; An adaptive quantitative risk assessment model is used to automatically identify potential market risks based on real-time data streams. The intelligent early warning and automatic trading control system is used to automatically issue early warning signals based on real-time risk assessment results and adjust market trading behavior according to preset rules. The quantum reinforcement learning control module is used to learn and optimize trading strategies in the electricity market through quantum reinforcement learning technology, providing dynamic adjustment suggestions and responding to market fluctuations in real time.
2. The monitoring system for electricity market transactions according to claim 1, characterized in that: The quantum computing engine employs quantum parallel computing methods to accelerate the analysis of transaction data in the electricity market through quantum algorithms during multi-dimensional data processing.
3. The monitoring system for electricity market transactions according to claim 1, characterized in that: The multimodal data includes: price fluctuations, trading volume, external economic factors, weather data, macroeconomic indicators, and public opinion information on social media.
4. A monitoring method for electricity market transactions, characterized in that, include: S1. Utilize a quantum computing engine to perform multi-dimensional parallel data acceleration processing on real-time data streams in the electricity market; S2. A graph neural network is used to fuse multimodal data in real time to generate a comprehensive market scenario map to predict market fluctuations and capture the correlation between multiple dimensions; S3. Based on the adaptive quantitative risk assessment model, a convolutional long short-term memory network is used to predict market fluctuations and automatically adjust the risk assessment strategy. S4. The market regulation mechanism is automatically triggered through the intelligent early warning and automatic trading control system; S5. Using quantum reinforcement learning algorithms, the trading strategies of the electricity market are adaptively optimized, and decisions are automatically adjusted by simulating multiple market scenarios in real time.
5. A monitoring method for electricity market transactions according to claim 4, characterized in that: The graph neural network uses the following formula for data fusion and graph generation: ， in: The first part of the graph represents the second part. Layer node feature matrix, The initial features are the input. The normalized adjacency matrix represents the relationships between nodes in a graph structure. ,in It is an adjacency matrix. Let be the identity matrix, representing a self-join. It is the first The weight matrix of the layer, The activation function, typically the ReLU function, enhances the nonlinear relationships in spectrum generation. This is the output feature matrix after graph convolution.
6. A monitoring method for electricity market transactions according to claim 5, characterized in that: The automatic adjustment risk assessment strategy specifically includes the following steps: S3.
1. Input multimodal data from the electricity market; S3.
2. Extract spatial features through convolutional layers; S3.
3. Utilize the convolutional long short-term memory network to learn time dependence; S3.
4. Market volatility prediction by combining convolutional features and temporal features; S3.
5. Automatically adjust the risk assessment strategy of the quantitative risk assessment model based on the prediction results; S3.
6. Finally, output the real-time risk assessment results and take risk response measures.
7. A monitoring method for electricity market transactions according to claim 6, characterized in that: The convolutional layer performs convolution operations on the input multimodal data through convolution kernels, transforming the local features of the multimodal data into global features.
8. A monitoring method for electricity market transactions according to claim 7, characterized in that: The automatic risk assessment strategy sets the following volatility thresholds: high volatility threshold, low volatility threshold, and medium volatility threshold.