New energy power system scheduling system and method based on demand response

Abstract

The invention relates to the field of power system management, and provides a new energy power system scheduling system and method based on demand response, and the method comprises the steps: firstly carrying out the data arrangement and embedded splicing of historical meteorological data and historical power generation data through employing a data processing technology based on artificial intelligence; and then sensing dynamic significant propagation representation based on the characteristic change between the meteorological-power generation data embedding splicing characteristics to intelligently predict the power generation amount of the new energy power system, and then analyzing the obtained historical power consumption data to predict a power consumption demand value. And finally, determining a scheduling strategy according to the comparison between the predicted value of the generating capacity and the predicted value of the electricity demand. Thus, the accuracy of power generation prediction can be remarkably improved, meanwhile, the future load can be estimated more accurately, and intelligent scheduling of the new energy power system can be better achieved.

Description

Technical Field

[0001] The present application relates to the field of power system management, and more specifically, to a new energy power system dispatching system and method based on demand response. Background Art

[0002] As the world pays more and more attention to environmental protection and sustainable development, the energy sector is undergoing profound changes. The extensive use of traditional fossil energy not only brings about environmental pollution problems, but also faces the risk of resource depletion. In this context, new energy represented by solar energy and wind energy has become a key force in energy transformation due to its clean and renewable characteristics. However, the intermittent and uncontrollable nature of these energy sources has brought challenges to the power grid, especially in ensuring the balance between power supply and demand. As an important means to ensure the stable operation of the power grid, the methods and strategies of power dispatching also need to be constantly adjusted with the changes in the energy structure.

[0003] Traditional power dispatch is mainly based on the predictability and controllability of conventional energy generation. Faced with the large-scale access of renewable energy power, traditional methods seem to be unable to cope with it. On the one hand, traditional dispatch models are difficult to accurately handle the uncertainty of renewable energy power generation and cannot respond to the rapid changes in power generation in a timely and effective manner. On the other hand, traditional dispatch methods focus on ensuring the reliability of power supply and pay less attention to the regulation potential on the demand side, which makes it difficult to achieve supply and demand balance through flexible strategies when renewable energy power supply fluctuates.

[0004] Therefore, a new energy power system scheduling scheme based on demand response is desired. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides a new energy power system scheduling system and method based on demand response.

[0006] According to one aspect of the present application, a new energy power system scheduling method based on demand response is provided, which includes: Obtain historical meteorological data and historical power generation data of new energy power systems; Performing data sorting and embedding coding on the historical meteorological data and the historical power generation data of the new energy power system to obtain a time series distribution of meteorological-power generation data embedding splicing coding features; The time series distribution of the meteorological-power generation data embedded in the splicing coding feature is subjected to dynamic and significant propagation of data sequences with feature change perception to obtain the dynamic and significant propagation feature of the meteorological-power generation time series, including: based on the jump degree and message propagation space span of each meteorological-power generation data embedded in the splicing coding feature in the time series distribution of the meteorological-power generation data embedded in the splicing coding feature, calculating the message propagation significant weight of each meteorological-power generation data embedded in the splicing coding feature to obtain the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature; based on the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature, weighted fusion is performed on the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to obtain the dynamic and significant propagation feature of the meteorological-power generation time series; Calculating the predicted power generation of the new energy power system based on the dynamic significant propagation characteristics of the meteorological-power generation time series; Obtain historical electricity consumption data; Calculating a power demand forecast value based on the historical power consumption data; A scheduling strategy is determined based on a comparison between the electricity demand forecast value and the power generation forecast of the new energy power system.

[0007] According to another aspect of the present application, a new energy power system dispatching system based on demand response is provided, which includes: The historical meteorological data power generation data collection module is used to obtain historical meteorological data and historical power generation data of new energy power systems; A historical meteorological data and power generation data embedded time series coding module, used for data sorting and embedding coding of the historical meteorological data and the historical power generation data of the new energy power system to obtain a time series distribution of meteorological-power generation data embedded splicing coding features; A time series distribution feature dynamic propagation module is used to perform feature change perception of the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to dynamically and significantly propagate the data sequence to obtain the meteorological-power generation time series dynamic and significant propagation feature, wherein the time series distribution feature dynamic propagation module includes: a time series distribution message propagation weight calculation unit, which is used to calculate the message propagation significant weight of each meteorological-power generation data embedded in the splicing coding feature based on the jump degree and message propagation space span of each meteorological-power generation data embedded in the splicing coding feature in the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to obtain the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature; a meteorological-power generation time series dynamic and significant propagation feature generation unit, which is used to perform weighted fusion on the time series distribution of the meteorological-power generation data embedded in the splicing coding feature based on the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature to obtain the meteorological-power generation time series dynamic and significant propagation feature; A power generation forecast result generation module, used to calculate the power generation forecast of the new energy power system based on the dynamic significant propagation characteristics of the meteorological-power generation time series; A historical electricity consumption data collection module is used to obtain historical electricity consumption data; A power demand forecast value calculation module, used to calculate the power demand forecast value based on the historical power consumption data; The scheduling strategy calculation module is used to determine the scheduling strategy based on the comparison between the power demand forecast value and the power generation forecast of the new energy power system.

[0008] This application has significant technical effects due to the adoption of the above technical solutions: The new energy power system dispatching system and method based on demand response provided by the present application adopts artificial intelligence-based data processing technology to first sort and embed historical meteorological data and historical power generation data, and then intelligently predict the power generation of the new energy power system based on the dynamic significant propagation representation of the characteristic change perception between the embedded splicing features of each meteorological-power generation data, and then analyze the acquired historical power consumption data to predict the power demand value, and finally determine the dispatching strategy based on the comparison between the power generation prediction value and the power demand prediction value. In this way, the accuracy of power generation prediction can be significantly improved, and the future load can be more accurately estimated, which is conducive to better realizing the intelligent dispatching of the new energy power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] 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.

[0010] Figure 1 It is a flow chart of a new energy power system scheduling method based on demand response according to an embodiment of the present application.

[0011] Figure 2 This is a flowchart of step S2 in the new energy power system scheduling method based on demand response according to an embodiment of the present application.

[0012] Figure 3 This is a flowchart of step S3 in the new energy power system scheduling method based on demand response according to an embodiment of the present application.

[0013] Figure 4 It is a system block diagram of a new energy power system dispatching system based on demand response according to an embodiment of the present application. DETAILED DESCRIPTION

[0014] 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.

[0015] Against the backdrop of growing global environmental awareness and the concept of sustainable development, the energy industry is undergoing a profound transformation. The large-scale use of traditional fossil fuels has not only led to a decline in environmental quality, but also accelerated the consumption of non-renewable energy. In this context, clean energy such as solar and wind energy, with its pollution-free and infinitely renewable advantages, has become the core force driving the transformation of the energy system. However, these new energy forms have brought new challenges to the stable operation of the power system due to the discontinuity and unpredictability of their output power, especially in maintaining a balance between power production and consumption. In order to ensure the stability of the power grid, power dispatching plays a vital role, and as the energy mix changes, dispatching methods and strategies also need to be adjusted accordingly.

[0016] The traditional dispatching model mainly relies on the stability and controllability of conventional energy generation. However, when a large number of renewable energy sources are connected to the power grid, this traditional method is insufficient in dealing with the volatility and uncertainty of renewable energy sources and cannot respond quickly to sudden changes in power generation. In addition, the traditional dispatching mechanism focuses more on the security of power supply and ignores the importance of demand-side management, which makes it difficult to maintain the balance of supply and demand in the power market through flexible demand response measures in the face of unstable renewable energy supply.

[0017] In response to the above technical problems, the technical concept of the present application is to obtain historical meteorological data and historical power generation data of the new energy power system, use artificial intelligence-based data analysis and embedding methods to organize and embed the historical meteorological data and the historical power generation data, and then intelligently predict the power generation of the new energy power system based on the dynamic significant propagation representation of the feature change perception between the embedded splicing features of each meteorological-power generation data. Then, the historical electricity consumption data is obtained, and the electricity demand value is predicted based on the historical electricity consumption data. In this way, the scheduling strategy is determined based on the comparison between the power generation forecast value and the power demand forecast value. Compared with the limitations of traditional power dispatching, the present application can capture the deep semantic association between meteorological and power generation data by combining historical meteorological data, thereby significantly improving the accuracy of power generation prediction. In addition, through in-depth analysis of historical electricity consumption data, the present application can more accurately estimate future loads, thereby enhancing demand-side management capabilities and realizing intelligent scheduling decisions.

[0018] Figure 1FIG. 1 is a flow chart of a new energy power system dispatching method based on demand response according to an embodiment of the present application. Figure 1 As shown, according to the demand response-based scheduling method of a new energy power system in accordance with an embodiment of the present application, the method includes: S1, acquiring historical meteorological data and historical power generation data of the new energy power system; S2, performing data sorting and embedding coding on the historical meteorological data and the historical power generation data of the new energy power system to obtain a time series distribution of meteorological-power generation data embedded in a spliced ​​coding feature; S3, performing dynamic significant propagation of a data sequence of feature change perception on the time series distribution of the meteorological-power generation data embedded in a spliced ​​coding feature to obtain a meteorological-power generation time series dynamic significant propagation feature; S4, calculating a power generation forecast of the new energy power system based on the meteorological-power generation time series dynamic significant propagation feature; S5, acquiring historical electricity consumption data; S6, calculating a power demand forecast value based on the historical electricity consumption data; S7, determining a scheduling strategy based on a comparison between the power demand forecast value and the power generation forecast of the new energy power system.

[0019] In step S1, historical meteorological data and historical power generation data of the new energy power system are obtained. It should be understood that the historical meteorological data includes light intensity data, wind speed data, temperature data, etc. at different historical time points; the historical power generation data of the new energy system includes power generation data, power generation data, etc. of different new energy power generation methods at different historical time points. Considering that new energy power generation (solar energy, wind energy) is closely related to meteorological conditions, meteorological factors are the main external factors affecting new energy power generation, and historical power generation data records the actual performance of the power generation system under past meteorological conditions. Therefore, in this application, it is first necessary to obtain historical meteorological data and historical power generation data of the new energy power system, and in subsequent steps, the correlation analysis of these two types of data is performed to establish a causal relationship model between meteorological conditions and power generation data, so as to more accurately predict the future power generation of the new energy system.

[0020] Obtaining historical meteorological data and historical power generation data of new energy power systems is not just a simple data collection step, but also involves how to effectively integrate this information from multiple channels and ensure their quality, integrity and real-time to support subsequent intelligent forecasting and scheduling decisions. The following is a detailed description of its implementation process: First, before starting to collect data, the specific parameters required must be clarified. For meteorological data, this includes but is not limited to temperature, humidity, wind speed, solar radiation and other factors closely related to renewable energy power generation; and for power generation data, it covers photovoltaic power generation, wind power output and other forms of renewable energy power generation records that may be involved. These specific requirements determine the direction and scope of data collection, thus laying the foundation for the subsequent work.

[0021] After clarifying the data requirements, the next step is to choose the appropriate data source. For meteorological data, the most direct and reliable source is the official meteorological station. Through the meteorological data published by the National Meteorological Department, historical meteorological observation data recorded by professional meteorological stations can be obtained. This type of data is usually highly accurate and reliable and is an indispensable part of building models. In addition to the data provided by ground meteorological stations, satellite remote sensing technology is also an important supplement. Especially when the target area lacks sufficient ground station coverage, sensors such as MODIS (Moderate Resolution Imaging Spectroradiometer) and VIIRS (Visible Infrared Imaging Radiometer Suite) can be used to obtain key information such as cloud cover and surface temperature. In addition, with the development of information technology, many third-party API services that provide high-resolution meteorological data worldwide have emerged on the market. With the help of these platforms, users can quickly and easily access the required meteorological data, and these services often provide a higher update frequency, which helps to capture the latest weather trends.

[0022] When it comes to historical power generation data of renewable energy power systems, it mainly relies on records in the internal systems of power companies. Detailed power generation performance indicators are stored in the SCADA (Supervisory Control and Data Acquisition) systems owned by grid operators or independent power producers. With the growth of distributed energy resources, more and more companies have installed smart meters or other forms of power generation monitoring devices. These small distributed power sources also contribute a lot of power generation data and cannot be ignored. Therefore, in the process of collecting power generation data, this part of the source should also be taken into account to cover all types of renewable energy power generation facilities as comprehensively as possible.

[0023] Once the data from different channels is obtained, it is necessary to organize and merge them. One of the tasks at this stage is to unify the data format. Since different data sources may have different formats, such as CSV files, database export files, or JSON formats returned by APIs, all collected data needs to be converted into a unified standard format to facilitate subsequent analysis tools to read and process. At the same time, special attention should be paid to the consistency of timestamps, because it is crucial to match the time dimensions of meteorological data and power generation data. Considering that there may be slight differences in the clocks of different systems, UTC (Coordinated Universal Time) should be used as the base time, and the time tags of each data set should be adjusted as needed to ensure accurate correspondence between the two.

[0024] Another important task is to identify and handle missing values ​​and outliers during the data cleaning process. In any actual scenario, completely correct data is almost non-existent. There will always be data missing at certain times or abnormal readings due to equipment failures and other reasons. For these problems, statistical methods or machine learning algorithms can be used to fill or correct them reasonably. For example, for missing data points in a continuous time period, interpolation estimates can be made based on data at adjacent times; and for outliers, it is necessary to combine business logic and technical means to identify and correct them to ensure that the data used for analysis reflects the actual situation as realistically as possible.

[0025] After completing the above preliminary processing, the next step is to establish a dedicated database to store the sorted meteorological and power generation data. A good database design can not only support efficient query operations, but also ensure the security and long-term availability of data. To this end, it is necessary to consider setting up an efficient data indexing mechanism to facilitate the rapid retrieval of records under specific conditions; at the same time, a regular backup strategy should be formulated to prevent data loss caused by unexpected situations. In addition, considering the characteristics of renewable energy power generation, that is, its power generation fluctuates greatly over time and environmental conditions, it is also necessary to ensure that the database has good scalability to cope with the possible increase in data volume in the future.

[0026] Last but not least, it is necessary to ensure the continuity and dynamic update capability of the entire data acquisition process. The characteristics of renewable energy power generation determine that it is necessary to keep up with the latest weather forecasts and actual power generation data to keep the database content up to date. To this end, an automated program can be set up to capture the latest information from various data sources on a regular basis every day, and automatically complete data cleaning and warehousing operations. At the same time, it is also necessary to regularly evaluate the quality of the stored data, check whether there are new problems or room for improvement, and make corresponding adjustments in a timely manner. Only in this way can a high-quality and reliable data foundation be truly established to provide strong support for subsequent data analysis steps, thereby promoting more intelligent and efficient new energy power system dispatching practices.

[0027] In step S2, the historical meteorological data and the historical power generation data of the new energy power system are sorted and embedded to obtain the time series distribution of the meteorological-power generation data embedded splicing coding features. Specifically, Figure 2 FIG. 1 is a flow chart of step S2 in the new energy power system dispatching method based on demand response according to an embodiment of the present application. Figure 2As shown, the step S2 includes: S21, sorting the historical meteorological data and the historical power generation data of the new energy power system according to the time dimension to obtain a time series data set of {meteorological data, power generation data}; S22, embedding and encoding each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector as the time series distribution of the meteorological-power generation data embedded splicing coding feature.

[0028] In step S21, the historical meteorological data and the historical power generation data of the new energy power system are sorted according to the time dimension to obtain a time series data set of {meteorological data, power generation data}. Accordingly, it is considered that there is a temporal correlation between the historical meteorological data and the historical power generation data. For example, the wind speed at a certain moment will directly affect the power generation of the wind power generation system at the same moment, and the sunshine intensity and photovoltaic power generation are also correlated at the same moment. At the same time, both meteorological data and power generation data have obvious time series characteristics. Based on this, in order to reflect the time series characteristics such as periodicity, trend and seasonality, and to ensure that the inherent causal relationship and correlation between the data are retained, the present application sorts the historical meteorological data and the historical power generation data of the new energy power system according to the time dimension to obtain a time series data set of {meteorological data, power generation data}.

[0029] The historical meteorological data and the historical power generation data of the new energy power system are sorted according to the time dimension to form a time series data set of {meteorological data, power generation data}. This step not only requires precise time alignment, but also requires ensuring the quality, consistency and relevance of the data. The following is a detailed description of its implementation process: Ideally, historical meteorological data and historical power generation data of new energy power systems should have corresponding observations at the same time point. However, in actual situations, due to different sampling frequencies or other reasons, incomplete matching may occur. At this time, a variety of strategies can be adopted to solve this problem. For example, when the sampling frequency of meteorological data is higher than that of power generation data, the estimated value of meteorological data at the same time as the power generation data can be generated by interpolation; on the contrary, if the sampling frequency of power generation data is higher, the meteorological data can be extrapolated within a reasonable range to fill the gap; nearest neighbor matching is also a commonly used method, which selects a valid observation closest to the target time as the data point at the corresponding time. This method is simple and direct, but it may lead to increased errors when the data fluctuates greatly. Therefore, when choosing a specific strategy, it is necessary to comprehensively consider the characteristics of the data and the requirements of the application scenario to achieve the best effect.

[0030] After the data processed in the above steps, they can be arranged in chronological order to form a complete {meteorological data, power generation data} time series data set. In this data set, each row represents a comprehensive observation result at a specific time point, including but not limited to meteorological parameters (such as temperature, humidity, wind speed, solar radiation intensity, etc.) and power generation parameters (such as photovoltaic power generation, wind power output, etc.).

[0031] Finally, in order to ensure the quality of the collated time series data set, it needs to be verified and evaluated. This includes but is not limited to the following aspects: integrity check, confirming that no data from important time periods are missed; consistency check, verifying whether the logical relationship between different variables is as expected, for example, solar power generation should theoretically increase when the temperature rises; trend analysis, observing whether the long-term trend is reasonable, and eliminating abnormal patterns caused by human errors or system failures. Through a rigorous verification process, the reliability of the data can be further improved, laying a good foundation for subsequent predictive modeling work based on this data set.

[0032] In step S22, each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} is embedded and encoded to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector as the time series distribution of the meteorological-power generation data embedded splicing coding feature. Specifically, in an embodiment of the present application, step S22 includes: using a multi-layer perceptron to embed the {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector. It should be understood that {meteorological data, power generation data} usually contains information of multiple dimensions. For example, meteorological data may contain multiple features such as temperature, humidity, wind speed, air pressure, etc., and power generation data may also contain multiple dimensions such as the power of different power generation units and the total power generation. And each {meteorological data, power generation data} contains the intrinsic connection between the two. Based on this, in the technical solution of the present application, each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} is embedded and encoded to map the high-dimensional data to a relatively low-dimensional vector space, and at the same time, the deeper intrinsic connection and mutual influence relationship between the two are excavated to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector. In particular, in a specific embodiment of the present application, a multi-layer perceptron is used to embed and encode each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector. That is, there is a complex nonlinear relationship between meteorological data and power generation data. For example, the influence of various meteorological factors such as temperature, humidity, wind speed, etc. on power generation is not a simple linear correlation, but a complex relationship of mutual interweaving and mutual influence. The multi-layer perceptron can effectively capture this complex nonlinear relationship through the nonlinear activation function (such as ReLU, Sigmoid, etc.) in the hidden layer. This enables the multi-layer perceptron to dig out the deep intrinsic connections behind the data when processing meteorological and power generation data, thereby providing richer and more accurate feature representation for embedded coding.

[0033] In step S3, the dynamic significant propagation of the data sequence of feature change perception is performed on the time series distribution of the weather-power generation data embedded splicing coding feature to obtain the weather-power generation time series dynamic significant propagation feature. Specifically, Figure 3 FIG. 1 is a flow chart of step S3 in the new energy power system dispatching method based on demand response according to an embodiment of the present application. Figure 3As shown, the step S3 includes: S31, based on the jump degree and message propagation space span of each meteorological-power generation data embedded splicing coding feature in the time series distribution of the meteorological-power generation data embedded splicing coding feature, calculating the message propagation significance weight of each meteorological-power generation data embedded splicing coding feature to obtain the time series distribution of the message propagation significance weight of the meteorological-power generation data embedded splicing coding feature; S32, based on the time series distribution of the message propagation significance weight of the meteorological-power generation data embedded splicing coding feature, weighted fusion of the time series distribution of the meteorological-power generation data embedded splicing coding feature to obtain the meteorological-power generation time series dynamic significant propagation feature.

[0034] It should be understood that there are different degrees of temporal correlation between different meteorological-power generation data embedded splicing coding features, and the contribution and significance of each feature on the entire time scale are different. Therefore, in order to capture the rapid changes and long-distance dependencies of features between different time steps, the present application performs dynamic and significant propagation of data sequences that perceive feature changes on the temporal distribution of the meteorological-power generation data embedded splicing coding features to obtain the dynamic and significant propagation characteristics of the meteorological-power generation time series. In this way, the meteorological factors that have a significant impact on the power generation situation at different time steps and their interactions can be highlighted. For example, through this processing, it is possible to obtain key feature representations that reflect how meteorological factors such as wind speed and temperature dynamically affect power generation over time, which can provide more targeted and representative information for subsequent power generation forecast calculations.

[0035] Specifically, in the embodiment of the present application, the step S31 includes: calculating the characteristic jump degree of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector to obtain the time series distribution of the meteorological-power generation data embedded splicing characteristic jump degree, and the process can be expressed as: ;in, is the time series distribution of the meteorological-power generation data embedded in the concatenated coding vector, and are the first, second, and third time series distributions of the meteorological-power generation data embedded in the concatenated coding vector. and The meteorological-power generation data are embedded in the concatenated coding vector, yes Middle The eigenvalues ​​at the positions, yes The number of eigenvalues ​​in , and They are the time series distribution of the weighted average feature value of the weather-power generation data embedding feature semantics. and The weather-power generation data are embedded with the feature semantic weighted average feature value. It is the time series distribution of the jump degree of the meteorological-power generation data embedding splicing feature. The jump degree of the meteorological-power generation data embedding splicing features; The message propagation space span of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector is calculated to obtain the time series distribution of the meteorological-power generation data embedded splicing feature message propagation space span. The process can be expressed as: ;in, and They are respectively the time series distribution of the meteorological-power generation data embedded in the concatenated coding vector. and The meteorological-power generation data are embedded in the concatenated coding vector, express arrive The number of times the message is spread, It is the temporal distribution of the spatial span of the message propagation of the meteorological-power generation data embedded in the splicing feature. The spatial span of the meteorological-power generation data embedding and concatenation feature message propagation; Based on the characteristic jump degree and message propagation space span of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector, the message propagation significance factor of each meteorological-power generation data embedded splicing coding vector is calculated to obtain the time series distribution of the meteorological-power generation data embedded splicing message propagation significance factor. This process can be expressed as: ;in, and is the modulation parameter, It is the time series distribution of the jump degree of the meteorological-power generation data embedding splicing feature. The meteorological-power generation data embedding splicing feature jump degree, It is the temporal distribution of the spatial span of the message propagation of the meteorological-power generation data embedded in the splicing feature. The spatial span of the message propagation of meteorological-power generation data embedding and concatenation features, It is the time series distribution of the significant factor of the meteorological-power generation data embedded splicing message propagation. The significant factor of the message propagation of meteorological-power generation data embedding splicing; The time series distribution of the significant factor of the meteorological-power generation data embedded splicing message propagation is gated to obtain the time series distribution of the significant weight of the meteorological-power generation data embedded splicing encoding feature message propagation. The process can be expressed as: ;in, It is the time series distribution of the significant factor of the meteorological-power generation data embedded splicing message propagation. The significant factor of the message propagation of meteorological-power generation data embedding splicing, is the normalization function, is a masking operation, is a predetermined threshold, It is the time series distribution of the significant weight of the message propagation of the meteorological-power generation data embedded in the splicing coding feature. The meteorological-power generation data embedding concatenation encodes the characteristic message propagation significant weight.

[0036] In detail, it is first necessary to calculate the characteristic jump degree of each meteorological-power generation data embedded in the splicing coding vector in the time series distribution of the meteorological-power generation data embedded in the splicing coding vector to obtain the time series distribution of the meteorological-power generation data embedded in the splicing coding vector. It should be understood that in the time series distribution of the meteorological-power generation data embedded in the splicing coding vector, the change of data may have local characteristics. For example, a sudden change in meteorological conditions (such as a strong wind or clouds quickly blocking the sun) can cause the power generation data to change in a short time. By calculating the characteristic jump degree of each meteorological-power generation data embedded in the splicing coding vector, it can help capture this local structural information, that is, it can accurately locate the time in the time series data at which the data has undergone obvious local changes. In addition, by calculating the characteristic jump degree, the degree of change of the meteorological-power generation data embedded in the splicing coding feature at each moment can be quantified, which is conducive to the model to clearly understand which moments of the meteorological-power generation data embedded in the splicing coding feature are key features, which will have an important impact on the propagation of features in the time window.

[0037] Accordingly, it is considered that the relationship between different meteorological-power generation data embedded splicing coding vectors in the time series distribution of meteorological-power generation data embedded splicing coding vectors is complex, not a simple adjacent relationship. For example, the meteorological-power generation data embedded splicing coding vector at a certain moment is not only affected by the characteristic data of the previous moment, but also associated with more distant historical characteristic data. In order to take into account the extensive connectivity between features at different moments, it is necessary to calculate the message propagation space span of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector in this application to obtain the time series distribution of the message propagation space span of the meteorological-power generation data embedded splicing feature. That is, by calculating the message propagation space span of each meteorological-power generation data embedded splicing coding vector, it can help identify the long-distance dependencies between different vectors in the time series distribution and the critical path of information flow.

[0038] Next, considering that the characteristic jump degree of each meteorological-power generation data embedded in the spliced ​​coding vector reflects the degree of change of the meteorological-power generation data embedded in the spliced ​​coding feature at each moment, and the message propagation space span of each meteorological-power generation data embedded in the spliced ​​coding vector reflects the connectivity and path influence of the features at different moments in the entire time series feature propagation process, in order to comprehensively consider these two factors, in order to more accurately evaluate the status of each moment feature in information propagation, so as to determine which moment features have a greater impact on the power generation forecast, in this application, it is necessary to calculate the message propagation significance factor of each meteorological-power generation data embedded in the spliced ​​coding vector based on the characteristic jump degree and message propagation space span of each meteorological-power generation data embedded in the spliced ​​coding vector in the time series distribution of the meteorological-power generation data embedded in the spliced ​​coding vector to obtain the time series distribution of the meteorological-power generation data embedded in the spliced ​​message propagation significance factor. The message propagation significance factor is similar to the attention mechanism, which adjusts the information exchange according to the importance of the features themselves and their neighbors at different moments, and can ensure the overall time series feature information while highlighting the features of important moments.

[0039] Subsequently, the time series distribution of the significant factor of the meteorological-power generation data embedded in the splicing message propagation is gated to obtain the time series distribution of the significant weight of the meteorological-power generation data embedded in the splicing coding feature message propagation. It should be understood that in the time series distribution of the significant factor of the meteorological-power generation data embedded in the splicing message propagation, the value of information at different times is different. Some information is crucial for power generation forecasting and scheduling decisions, while some information may cause interference. By performing gated propagation processing on it, the flow of information can be controlled according to the importance of the information, ensuring that features with key information can be highlighted, while those with redundant information or irrelevant information are weakened. In this way, the model can better adapt to the diversity and complexity of the original data, which is conducive to improving the expressiveness of the model.

[0040] Specifically, in the embodiment of the present application, the step S32 includes: weighting the position of each meteorological-power generation data embedded splicing coding feature message propagation significant weight in the time series distribution of the meteorological-power generation data embedded splicing coding feature message propagation significant weight and the time series distribution of the meteorological-power generation data embedded splicing coding vector to obtain the meteorological-power generation time series dynamic significant propagation coding vector as the meteorological-power generation time series dynamic significant propagation feature. The above process can be expressed as: ;in, is the temporal distribution of the meteorological-power generation data embedded in the concatenated coding vector. The meteorological-power generation data are embedded in the concatenated coding vector, It is the time series distribution of the significant weight of the message propagation of the meteorological-power generation data embedded in the splicing coding feature. The meteorological-power generation data embedding concatenation encoding feature message propagation significant weight, is the number of vectors in the time series distribution of the meteorological-power generation data embedded in the concatenated coding vector, is the significant propagation encoding vector of the meteorological-generation time series dynamics.

[0041] It should be understood that the time series distribution of meteorological-power generation data embedded in the splicing coding vector contains the feature vectors of the original meteorological-power generation data after encoding, which contain rich original data information, while the time series distribution of the significant weight of the meteorological-power generation data embedded in the splicing coding feature message reflects the importance of the data at each moment in the process of time series feature information propagation. By weighted fusion of the two by position, it is possible to use the original data features while adjusting them according to the importance of information to generate a meteorological-power generation time series dynamic significant propagation coding vector that can more accurately reflect the internal relationship of meteorological-power generation data.

[0042] In step S4, based on the dynamic significant propagation characteristics of the meteorological-power generation time series, the power generation forecast of the new energy power system is calculated. Specifically, in the embodiment of the present application, the step S4 includes: inputting the meteorological-power generation time series dynamic significant propagation coding vector into the power generation predictor based on the decoder to obtain the power generation decoded amount as the power generation forecast of the new energy power system. It should be understood that the meteorological-power generation time series dynamic significant propagation coding vector has deeply processed the meteorological-power generation data, integrated the complex dynamic relationship between meteorological factors and power generation data and the significant features of different time steps, and these vectors contain key information on how meteorological factors such as wind speed and temperature affect power generation over time, which is an important basis for predicting the power generation forecast. Based on this, in the technical solution of the present application, the power generation forecast of the new energy power system is calculated based on the meteorological-power generation time series dynamic significant propagation coding vector, which can enable the prediction model to focus on key information, thereby reducing prediction errors. In particular, in a specific embodiment of the present application, the weather-generation time series dynamic significant propagation coding vector is processed by a decoder-based power generation predictor to utilize the flexible structure of the decoder to gradually parse the coding vector, and the key information of the meteorological impact on power generation contained therein is converted into an understandable power generation forecast, so as to obtain the power generation decoded amount as the power generation forecast of the new energy power system. More specifically, in a specific embodiment of the present application, the weather-generation time series dynamic significant propagation coding vector is input into a decoder-based power generation predictor to obtain the power generation decoded amount, including: multiplying the decoding weight matrix of the decoder with the weather-generation time series dynamic significant propagation coding vector to obtain the decoded weather-generation time series dynamic significant propagation coding vector, and accumulating and summing all the eigenvalues ​​of the decoded weather-generation time series dynamic significant propagation coding vector to obtain the power generation decoded amount.

[0043] Here, since the time series distribution of the meteorological-power generation data embedded splicing coding vector represents the embedded splicing coding characteristics of the meteorological data and the power generation data along the time series distribution, when performing feature sequence message transmission based on the feature jump degree, the feature population attributes at different local time series scales will have differences in sequence message transmission fairness based on the feature jump degree level, thereby affecting the inclusiveness of the sequence propagation feature distribution of the meteorological-power generation time series dynamic significant propagation coding vector, and reducing the calculation accuracy of the power generation forecast of the new energy power system.

[0044] In a preferred example, calculating the power generation forecast of the new energy power system based on the meteorological-power generation time series dynamic significant propagation coding vector includes: Determine the eigenvalue mean corresponding to the meteorological-generation time series dynamic significant propagation coding vector and the standard deviation of the eigenvalues ; The first meteorological-power generation time series dynamic significant propagation encoding fair target vector is obtained by multiplying the point-subtraction vector of the meteorological-power generation time series dynamic significant propagation encoding vector and the eigenvalue mean and the eigenvalue standard deviation: ;in, represents the meteorological-generation time series dynamic significant propagation encoding vector, It means point multiplication by position. It means to subtract by position point. express The corresponding eigenvalue standard deviation is, express The corresponding feature mean is represents the first meteorological-generation time series dynamic significant propagation encoding fairness target vector; The point-wise subtraction vector of the meteorological-power generation time series dynamic significant propagation coding vector and the eigenvalue standard deviation is multiplied by the eigenvalue mean to obtain the second meteorological-power generation time series dynamic significant propagation coding fairness target vector: ;in, represents the meteorological-generation time series dynamic significant propagation encoding vector, It means point multiplication by position. It means to subtract by position point. express The corresponding eigenvalue standard deviation is, express The corresponding feature mean is represents the second meteorological-generation time series dynamic significant propagation encoding fairness target vector; After multiplying the bit-by-bit inverse of the second meteorological-power generation time series dynamic significant propagation coding fairness target vector by the first meteorological-power generation time series dynamic significant propagation coding fairness target vector, take the bit-by-bit logarithm with base 2 to obtain the meteorological-power generation time series dynamic significant propagation coding information correction vector: ;in, represents the second meteorological-generation time series dynamic significant propagation encoding fairness objective vector, Representation calculation The inverse of each eigenvalue in , It means point multiplication by position. represents the first meteorological-generation time series dynamic significant propagation encoding fairness objective vector, represents the logarithmic function value with base 2, A correction vector representing the dynamic significant propagation encoding information of the meteorological-power generation time series; The eigenvalue mean Divide by the standard deviation of the eigenvalue The square root of the quotient is multiplied by the weight hyperparameter, and then the optimized meteorological-power generation time series dynamic significant propagation coding vector is obtained by adding the meteorological-power generation time series dynamic significant propagation coding information correction vector point: ;in, represents the weight hyperparameter, express The corresponding eigenvalue standard deviation is, express The corresponding feature mean is It means point multiplication by position. represents the correction vector of the meteorological-generation time series dynamic significant propagation encoding information, A significant propagation coding vector representing the optimized meteorological-power generation time series dynamics; The power generation forecast of the new energy power system is calculated based on the optimized meteorological-power generation time series dynamic significant propagation coding vector.

[0045] That is, taking into account the attribute level fairness differences of the data population corresponding to the time series fusion characteristics of the meteorological-power generation time series dynamic significant propagation coding vector, in order to improve the propagation inclusiveness under the feature distribution diversity of the meteorological-power generation time series dynamic significant propagation coding vector, the group feature information interactive propagation of the meteorological-power generation time series dynamic significant propagation coding vector is corrected by taking the crossover probability value constraint based on the meteorological-power generation time series dynamic significant propagation coding vector as the interactive fairness target representation, and taking the unified statistical feature response interaction based on the meteorological-power generation time series dynamic significant propagation coding vector as the feature distribution multi-level fairness target bias to achieve the robust distribution fairness unified representation of the meteorological-power generation time series dynamic significant propagation coding vector, forming a fair cooperation paradigm under the feature distribution framework of the meteorological-power generation time series dynamic significant propagation coding vector, and improving the calculation accuracy of the power generation forecast of the new energy power system.

[0046] In step S5, historical electricity consumption data is obtained. It should be understood that historical electricity consumption data contains actual power consumption data at different historical time points, which is the basis for accurate electricity demand forecasting, and accurate demand forecasting is the prerequisite for implementing effective power dispatching strategies. Specifically, by analyzing historical electricity consumption data, it is possible to discover the changing patterns of electricity consumption in different time periods, including daily, weekly, and seasonal fluctuations. By understanding which periods are peak electricity consumption, targeted peak-shaving and valley-filling measures can be designed to ensure the stable operation of the power system.

[0047] The following is a detailed description of the specific implementation process of "obtaining historical electricity consumption data": First, before you start acquiring historical electricity consumption data, you must be clear about the specific parameters and data coverage you need. This includes, but is not limited to, electricity consumption by households, commercial institutions, and industrial users, data recorded at different time granularities such as hours, days, weeks, or months, and other factors that may affect electricity consumption patterns, such as seasonal changes, holiday schedules, etc. This information will determine which sources to collect data from and help you understand user behavior patterns more accurately in subsequent analysis.

[0048] The next step is to choose the right data source. For large-scale power systems, the most direct and reliable source of historical electricity consumption data is the smart meter system maintained by the grid operator. With the development of smart grid technology, more and more homes and businesses have installed smart meters that can monitor and record electricity consumption in real time. These devices can not only provide detailed electricity consumption data, but also upload them to the central database regularly through the network interface to ensure the timeliness and integrity of the data. In addition, some large commercial users and industrial enterprises may have their own energy management systems (EMS), which can also serve as important data sources. In order to cover the widest possible range of data, you should also consider integrating electricity consumption statistics of public facilities provided by third-party platforms, such as urban lighting systems, traffic lights, etc.

[0049] Once the data source is determined, a stable data transmission channel needs to be established to ensure that the data can be safely and efficiently transmitted to the central processing center. For smart meters, wireless communication technology (such as GPRS, LoRaWAN) or wired connections (such as optical fiber, power line carrier) are usually used to achieve communication with the data acquisition server. In order to ensure the security of the data, all transmission processes should follow strict security protocols, such as SSL / TLS encryption, to prevent data leakage or tampering. At the same time, considering the huge amount of data and its continuous growth, it is also necessary to deploy an efficient distributed storage architecture, such as Hadoop clusters or cloud storage services, for easy management and expansion.

[0050] After completing data collection, the next step is to clean and preprocess the raw data. In practical applications, electricity consumption data often have various quality problems, such as missing values, outliers, duplicate records, etc. In response to these problems, a series of measures need to be taken to solve them: for missing values, they can be reasonably filled by interpolation or based on machine learning models; for outliers, they need to be identified and corrected by combining business logic and technical means; and for duplicate records, it is necessary to ensure that there is only one valid observation value at each time point. In addition, special attention should be paid to the consistency of timestamps to ensure that data from different regions or systems can be compared and analyzed on the same time scale.

[0051] Finally, in order to ensure the effectiveness and reliability of the entire data acquisition process, a complete monitoring mechanism needs to be established. This includes but is not limited to setting early warning thresholds, automatically issuing alarms when data quality deteriorates or transmission is interrupted; regularly evaluating the quality of stored data to check whether there are new problems or room for improvement; and making corresponding adjustments in a timely manner, such as optimizing data transmission paths and updating hardware equipment.

[0052] Through the above steps, reliable historical electricity consumption data can be obtained to provide more accurate basic data for subsequent analysis of electricity demand forecasts.

[0053] In step S6, based on the historical electricity consumption data, the electricity demand forecast value is calculated. Specifically, in an embodiment of the present application, the step S6 includes: sorting the historical electricity consumption data according to the time dimension to obtain a time series data set of historical electricity consumption data; inputting the time series data set of the historical electricity consumption data into a time series encoder based on LSTM to obtain a time series associated feature vector of the historical electricity consumption data; inputting the time series associated feature vector of the historical electricity consumption data into a power demand forecaster based on a decoder to obtain a power demand decoding amount as the power demand forecast value. It should be understood that the historical electricity consumption data is closely related in the time dimension, and the electricity consumption at different time points does not exist in isolation. Based on this, the present application obtains a time series data set of historical electricity consumption data by sorting the historical electricity consumption data according to the time dimension to retain this time correlation, so that subsequent analysis and modeling can use this information, such as discovering the periodic law of electricity demand. Afterwards, considering the time-varying characteristics of each historical electricity consumption data in the time dimension and the dependency between each time step. Therefore, in the technical solution of the present application, the time series data set of the historical power consumption data is input into the time series encoder based on LSTM to obtain the time series associated feature vector of the historical power consumption data. Those skilled in the art should know that LSTM, as a special recurrent neural network, can effectively process such data. By introducing a gating mechanism (input gate, forget gate and output gate), it can solve the gradient vanishing or gradient explosion problems faced by traditional recurrent neural networks when processing long time series, so as to better capture the long-term dependency of historical power consumption data in the time dimension. For example, the user's power consumption habits may remain relatively stable for a long time, and LSTM can remember these long-term patterns and use them when processing the data of the current time step. Finally, the time series associated feature vector of the historical power consumption data is input into the power demand predictor based on the decoder to use the decoder through its specific structure and algorithm to parse these encoded information to convert the power consumption patterns, trends and various associations hidden in the feature vector into specific power demand numerical predictions, and obtain the power demand decoding amount as the power demand prediction value.

[0054] In step S7, a scheduling strategy is determined based on the comparison between the predicted power demand value and the predicted power generation of the new energy power system. Specifically, in the embodiment of the present application, the scheduling strategy is a peak-shaving demand response or a valley-filling demand response. It should be understood that new energy power generation is intermittent and volatile, such as solar energy is affected by light intensity and duration, wind energy depends on wind speed changes, and power generation is unstable; on the other hand, power demand is affected by various factors and fluctuates greatly in different periods. By comparing the values ​​of the two, the imbalance of power supply and demand can be clarified, and a basis can be provided for implementing peak-shaving or valley-filling demand response, thereby adjusting power supply and demand and keeping the power system stable. Specifically, when the calculated power generation forecast of the new energy power system is greater than the power demand forecast value, it means that there will be an oversupply of power in the corresponding period. In order to avoid wasting new energy power and maintain the stable operation of the power grid, a valley-filling demand response strategy needs to be adopted. At this time, valley-filling demand response can be implemented in a variety of ways. From the perspective of economic incentives, a time-of-use electricity price preferential policy can be introduced to reduce electricity prices during this period to attract residential users to increase electricity consumption. For example, for electric energy storage equipment (such as energy storage batteries in homes), users are encouraged to charge during this period; for commercial users, such as air conditioning systems in large shopping malls, the temperature setting can be appropriately lowered, the cooling power can be increased, and excess electricity can be consumed; for industrial users, electricity subsidies can be provided to encourage them to arrange some non-urgent production tasks during this period, thereby increasing electricity consumption and balancing electricity supply and demand. If the predicted power generation of the new energy power system is less than the predicted power demand, it means that the power supply during this period cannot meet the power demand, and there may be a power shortage. In order to ensure the stable operation of the power grid and avoid accidents such as power outages, a peak-cutting demand response strategy needs to be adopted. Measures that can be taken include, for industrial users, requiring some non-critical production equipment to temporarily stop production or reduce production power in accordance with a pre-signed agreement or according to a certain priority order to reduce industrial power load. For residential users, power consumption reminders can be sent to users through devices such as smart meters to inform users that they are currently in the peak power consumption period, and it is recommended to reduce the use of some non-essential electrical appliances, such as postponing the use of washing machines, dishwashers and other appliances. At the same time, the electricity price can also be appropriately increased to suppress users' power demand through economic means to achieve the purpose of peak shaving. Once the specific dispatching strategy is determined, corresponding measures need to be formulated and implemented. In addition, as the actual situation changes (such as changes in power generation caused by weather changes), it is necessary to monitor and flexibly adjust the dispatching strategy in real time to ensure that the optimal balance between power supply and demand is always maintained.

[0055] In summary, a new energy power system dispatching method based on demand response according to an embodiment of the present application is explained, which uses artificial intelligence-based data processing technology to first sort and embed historical meteorological data and historical power generation data, and then intelligently predict the power generation of the new energy power system based on the dynamic significant propagation representation of the feature change perception between the embedded splicing features of each meteorological-power generation data, and then analyze the acquired historical power consumption data to predict the power demand value, and finally determine the dispatching strategy based on the comparison between the power generation prediction value and the power demand prediction value. In this way, the accuracy of power generation prediction can be significantly improved, and at the same time, future loads can be more accurately estimated, which is conducive to better realizing the intelligent dispatching of new energy power systems.

[0056] Figure 4 : is a system block diagram of a new energy power system dispatching system based on demand response according to an embodiment of the present application. Figure 4As shown, according to the embodiment of the present application, the new energy power system dispatching system 100 based on demand response includes: a historical meteorological data power generation data collection module 110, which is used to obtain historical meteorological data and historical power generation data of the new energy power system; a historical meteorological data power generation data embedded time series coding module 120, which is used to sort and embed the historical meteorological data and the historical power generation data of the new energy power system to obtain the time series distribution of the meteorological-power generation data embedded splicing coding characteristics; a time series distribution feature dynamic propagation module 130, which is used to perform feature change perception of the time series distribution of the meteorological-power generation data embedded splicing coding characteristics to obtain the meteorological-power generation time series dynamic significant propagation characteristics; a power generation prediction result generation module 140, which is used to calculate the power generation prediction of the new energy power system based on the meteorological-power generation time series dynamic significant propagation characteristics; a historical electricity consumption data collection module 150, which is used to obtain historical electricity consumption data; a power demand forecast value calculation module 160 , used to calculate the electricity demand forecast value based on the historical electricity consumption data; a scheduling strategy calculation module 170, used to determine the scheduling strategy based on the comparison between the electricity demand forecast value and the power generation forecast of the new energy power system; wherein the time series distribution feature dynamic propagation module includes: a time series distribution message propagation weight calculation unit, used to calculate the message propagation significance weight of each meteorological-power generation data embedded splicing coding feature based on the jump degree and message propagation space span of each meteorological-power generation data embedded splicing coding feature in the time series distribution of the meteorological-power generation data embedded splicing coding feature to obtain the time series distribution of the meteorological-power generation data embedded splicing coding feature message propagation significance weight; a meteorological-power generation time series dynamic significant propagation feature generation unit, used to perform weighted fusion on the time series distribution of the meteorological-power generation data embedded splicing coding feature based on the time series distribution of the meteorological-power generation data embedded splicing coding feature message propagation significance weight to obtain the meteorological-power generation time series dynamic significant propagation feature.

[0057] Here, those skilled in the art can understand that the specific functions and operations of the various units and modules in the above-mentioned demand response-based new energy power system dispatching system 100 have been referred to above. Figures 1 to 3 The description of the new energy power system scheduling method based on demand response has been introduced in detail, and therefore, its repeated description will be omitted.

[0058] In summary, the demand-response-based new energy power system dispatching system 100 based on the embodiment of the present application is explained, which uses artificial intelligence-based data processing technology to first sort and embed historical meteorological data and historical power generation data, and then intelligently predict the power generation of the new energy power system based on the dynamic significant propagation representation of the feature change perception between the embedded splicing features of each meteorological-power generation data, and then analyze the acquired historical power consumption data to predict the power demand value, and finally determine the dispatching strategy based on the comparison between the power generation prediction value and the power demand prediction value. In this way, the accuracy of power generation prediction can be significantly improved, and at the same time, future loads can be more accurately estimated, which is conducive to better realizing the intelligent dispatching of new energy power systems.

Claims

1. A new energy power system dispatching method based on demand response, characterized in that: include: Obtain historical meteorological data and historical power generation data of new energy power systems; Performing data sorting and embedding coding on the historical meteorological data and the historical power generation data of the new energy power system to obtain a time series distribution of meteorological-power generation data embedding splicing coding features; The time series distribution of the meteorological-power generation data embedded in the splicing coding feature is subjected to dynamic and significant propagation of data sequences with feature change perception to obtain the dynamic and significant propagation feature of the meteorological-power generation time series, including: based on the jump degree and message propagation space span of each meteorological-power generation data embedded in the splicing coding feature in the time series distribution of the meteorological-power generation data embedded in the splicing coding feature, calculating the message propagation significant weight of each meteorological-power generation data embedded in the splicing coding feature to obtain the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature; based on the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature, weighted fusion is performed on the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to obtain the dynamic and significant propagation feature of the meteorological-power generation time series; Calculating the predicted power generation of the new energy power system based on the dynamic significant propagation characteristics of the meteorological-power generation time series; Obtain historical electricity consumption data; Calculating a power demand forecast value based on the historical power consumption data; A scheduling strategy is determined based on a comparison between the electricity demand forecast value and the power generation forecast of the new energy power system.

2. The method for dispatching a new energy power system based on demand response according to claim 1, characterized in that: The dispatching strategy is peak shaving demand response or valley filling demand response.

3. The method for dispatching a new energy power system based on demand response according to claim 2, characterized in that: The historical meteorological data and the historical power generation data of the new energy power system are sorted and embedded in the data to obtain a time series distribution of meteorological-power generation data embedded splicing coding features, including: Arrange the historical meteorological data and the historical power generation data of the new energy power system according to the time dimension to obtain a time series data set of {meteorological data, power generation data}; Each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} is embedded and encoded to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector as the time series distribution of the meteorological-power generation data embedded splicing coding feature.

4. The method for dispatching a new energy power system based on demand response according to claim 3 is characterized in that: Embedding and encoding each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} to obtain the time series distribution of the meteorological-power generation data embedded concatenated coding vector, including: using a multi-layer perceptron to embedding and encoding each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} to obtain the time series distribution of the meteorological-power generation data embedded concatenated coding vector.

5. The method for dispatching a new energy power system based on demand response according to claim 4 is characterized in that: Based on the jump degree and jump degree of each meteorological-power generation data embedded splicing coding feature in the time series distribution of the meteorological-power generation data embedded splicing coding feature, the message propagation significance weight of each meteorological-power generation data embedded splicing coding feature is calculated to obtain the time series distribution of the message propagation significance weight of the meteorological-power generation data embedded splicing coding feature, including: Calculating the characteristic jump degree of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector to obtain the time series distribution of the meteorological-power generation data embedded splicing characteristic jump degree; Calculating the message propagation space span of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector to obtain the time series distribution of the meteorological-power generation data embedded splicing feature message propagation space span; Based on the characteristic jump degree and message propagation space span of each meteorological-power generation data embedded splicing coding vector in the time series distribution of the meteorological-power generation data embedded splicing coding vector, calculating the message propagation significance factor of each meteorological-power generation data embedded splicing coding vector to obtain the time series distribution of the meteorological-power generation data embedded splicing message propagation significance factor; The time series distribution of the meteorological-power generation data embedded splicing message propagation significance factor is gated and propagated to obtain the time series distribution of the meteorological-power generation data embedded splicing coding feature message propagation significance weight.

6. The method for dispatching a new energy power system based on demand response according to claim 5, characterized in that: Based on the time series distribution of the significant weights of the message propagation of the meteorological-power generation data embedded in the splicing coding feature, the time series distribution of the meteorological-power generation data embedded in the splicing coding feature is weightedly fused to obtain the meteorological-power generation time series dynamic significant propagation feature, including: weighting the position of each meteorological-power generation data embedded in the splicing coding feature message propagation significant weight in the time series distribution of the meteorological-power generation data embedded in the splicing coding feature with the time series distribution of the meteorological-power generation data embedded in the splicing coding vector to obtain the meteorological-power generation time series dynamic significant propagation coding vector as the meteorological-power generation time series dynamic significant propagation feature.

7. The method for dispatching a new energy power system based on demand response according to claim 6, characterized in that: Based on the dynamic significant propagation characteristics of the meteorological-power generation time series, the power generation forecast of the new energy power system is calculated, including: inputting the meteorological-power generation time series dynamic significant propagation coding vector into a decoder-based power generation predictor to obtain the power generation decoded amount as the power generation forecast of the new energy power system.

8. The method for dispatching a new energy power system based on demand response according to claim 7, characterized in that: Calculating a power demand forecast value based on the historical power consumption data includes: Arrange the historical electricity consumption data according to the time dimension to obtain a time series data set of the historical electricity consumption data; Inputting the time series data set of the historical electricity consumption data into a time series encoder based on LSTM to obtain a time series correlation feature vector of the historical electricity consumption data; The time series associated feature vector of the historical power consumption data is input into a power demand predictor based on a decoder to obtain a power demand decoding amount as the power demand prediction value.

9. A new energy power system dispatching system based on demand response, characterized in that: include: The historical meteorological data power generation data collection module is used to obtain historical meteorological data and historical power generation data of new energy power systems; A historical meteorological data and power generation data embedded time series coding module, used for data sorting and embedding coding of the historical meteorological data and the historical power generation data of the new energy power system to obtain a time series distribution of meteorological-power generation data embedded splicing coding features; A time series distribution feature dynamic propagation module is used to perform feature change perception of the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to dynamically and significantly propagate the data sequence to obtain the meteorological-power generation time series dynamic and significant propagation feature, wherein the time series distribution feature dynamic propagation module includes: a time series distribution message propagation weight calculation unit, which is used to calculate the message propagation significant weight of each meteorological-power generation data embedded in the splicing coding feature based on the jump degree and message propagation space span of each meteorological-power generation data embedded in the splicing coding feature in the time series distribution of the meteorological-power generation data embedded in the splicing coding feature to obtain the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature; a meteorological-power generation time series dynamic and significant propagation feature generation unit, which is used to perform weighted fusion on the time series distribution of the meteorological-power generation data embedded in the splicing coding feature based on the time series distribution of the message propagation significant weight of the meteorological-power generation data embedded in the splicing coding feature to obtain the meteorological-power generation time series dynamic and significant propagation feature; A power generation forecast result generation module, used to calculate the power generation forecast of the new energy power system based on the dynamic significant propagation characteristics of the meteorological-power generation time series; A historical electricity consumption data collection module is used to obtain historical electricity consumption data; A power demand forecast value calculation module, used to calculate the power demand forecast value based on the historical power consumption data; The scheduling strategy calculation module is used to determine the scheduling strategy based on the comparison between the power demand forecast value and the power generation forecast of the new energy power system.

10. The new energy power system dispatching system based on demand response according to claim 9 is characterized in that: The historical meteorological data power generation data embedded time sequence coding module is used to Arrange the historical meteorological data and the historical power generation data of the new energy power system according to the time dimension to obtain a time series data set of {meteorological data, power generation data}; Each {meteorological data, power generation data} in the time series data set of {meteorological data, power generation data} is embedded and encoded to obtain the time series distribution of the meteorological-power generation data embedded splicing coding vector as the time series distribution of the meteorological-power generation data embedded splicing coding feature.

Images