Whole-province power load prediction method combining meteorological and economic data

Abstract

The invention belongs to the technical field of power prediction, and provides a whole-province power load prediction method combining meteorological and economic data. The method comprises the steps of collecting historical meteorological data and historical economic data related to a target province, determining first meteorological prediction data and first meteorological response big data in a future prediction period, and associating second meteorological prediction data and second meteorological response big data of the province; the first depth prediction model predicts the historical meteorological data and the historical economic data to obtain a first whole province power load prediction value corresponding to a future prediction time period; and using a second depth prediction model to perform depth prediction on the first weather prediction data and the first weather response big data and the second weather prediction data and the second weather response big data to obtain a first influence coefficient and a second influence coefficient, and correcting the first whole province power load prediction value to obtain a second whole province power load prediction value. The method can significantly improve the accuracy of the prediction result of the power load of the whole province.

Description

Technical Field

[0001] The present invention relates to the technical field of power forecasting, and in particular to a method for forecasting provincial power load by combining meteorological and economic data. Background Art

[0002] In today's power system management, accurate power load forecasting is essential to ensure the stability and reliability of power supply. The change of power load is affected by many factors, among which meteorological conditions and economic activities are two key factors. Meteorological conditions, such as temperature, humidity, wind speed, etc., will directly affect the power demand of residential and industrial users. For example, in the hot summer, the extensive use of air conditioners will cause a sharp increase in power load; while in the cold winter, the operation of heating equipment will also have a significant impact on the power load. The level and structure of economic activities will also have an important impact on the power load. Economic growth is usually accompanied by an increase in power demand, and the power consumption characteristics of different industries are also different.

[0003] At present, when conducting power load forecasting across the province, the existing methods are mainly based on historical meteorological data and economic data to establish forecasting models. These methods usually collect meteorological data (such as temperature, humidity, wind speed, etc.) and economic data (such as GDP, industrial output value, consumer spending, etc.) over a period of time in the past, and use statistical analysis or machine learning algorithms to find the relationship between these data and power load, and then predict future power load. However, these existing methods have some obvious limitations, such as the lack of consideration of the province's ability to respond to abnormal weather in the future period targeted by the power load forecast, and the lack of consideration of the impact on economic activities in neighboring provinces.

[0004] Therefore, in order to improve the accuracy of power load forecasting in the province and better meet the operation and management needs of the power system, a new power load forecasting method that combines meteorological and economic data is urgently needed. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a method for predicting the province-wide power load by combining meteorological and economic data, an electronic device, a computer storage medium and a computer program product.

[0006] The present invention provides a method for predicting the province's power load by combining meteorological and economic data, comprising the following steps: Collect historical meteorological data and historical economic data related to the target province, and determine first meteorological forecast data and first meteorological response big data for a future forecast period corresponding to the target province, and determine second meteorological forecast data and second meteorological response big data for associated provinces corresponding to the target province; Use the first deep prediction model to predict the historical meteorological data and the historical economic data to obtain a first province-wide power load prediction value corresponding to the future prediction period; Use a second deep prediction model to perform deep prediction on the first meteorological prediction data and the first meteorological response big data, the second meteorological prediction data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively; The first province-wide power load forecast value is corrected based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value.

[0007] In some embodiments, using a second deep prediction model to perform deep prediction on the first meteorological prediction data and the first meteorological response big data, the second meteorological prediction data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively, including: Using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological forecast data according to the first clustering index to obtain the first meteorological key semantic information; using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological response big data according to the first clustering index to obtain the first meteorological response key semantic information; Using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological forecast data according to the second clustering index to obtain second meteorological key semantic information; using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological response big data according to the second clustering index to obtain second meteorological response key semantic information; Based on the first meteorological key semantic information and the first meteorological response key semantic information, the second meteorological key semantic information and the second meteorological response key semantic information, respectively, a response capability question is asked to the third semantic analysis model to obtain a first response capability and a second response capability; wherein the third semantic analysis model is constructed based on a general large model; The first coping capability, the second coping capability and a coping capability-influence coefficient comparison table are compared to obtain the first influence coefficient and the second influence coefficient respectively.

[0008] In some embodiments, the clustering distances of the first clustering indicator and the second clustering indicator are different, and the clustering distance of the first clustering indicator is greater than the clustering distance of the second clustering indicator; wherein the clustering distance is the minimum value indicating that two data points or two clusters are similar data that can be clustered.

[0009] In some embodiments, the first clustering index is determined by: Retrieving abnormal weather loss record data of the target province for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the first loss characteristic data, and the first Euclidean distance between the second loss characteristic data of the most recent year after vectorization processing and the first loss characteristic data is calculated. The clustering distance of the first clustering indicator is determined based on the first Euclidean distance and the standard clustering distance.

[0010] In some embodiments, the second clustering index is determined by: Retrieving abnormal weather loss record data of related provinces for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the third loss characteristic data, and the second Euclidean distance between the fourth loss characteristic data of the most recent year after vectorization processing and the third loss characteristic data is calculated. The clustering distance of the second clustering indicator is determined based on the second Euclidean distance and the standard clustering distance.

[0011] In some embodiments, if the clustering distance of the first clustering indicator is smaller than the clustering distance of the second clustering indicator, the method further includes: The clustering distance of the first clustering indicator is increased at least once using a preset adjustment coefficient to make it greater than the clustering distance of the second clustering indicator.

[0012] In some embodiments, the correcting the first province-wide power load forecast value based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value includes: The first influence coefficient and the second influence coefficient are multiplied by the first province-wide power load forecast value at the same time to obtain a revised second province-wide power load forecast value.

[0013] The present invention also provides an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method as described in any of the preceding items.

[0014] The present invention also provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any of the above items is executed.

[0015] The present invention also provides a computer program product, which includes a computer program stored in a computer storage medium, and when the computer program is executed by a processor of an electronic device, it implements the method described in any of the above items.

[0016] The beneficial effects of the present invention are: The power load forecasting method of the present invention takes into account the meteorological response capabilities of the target province in the future forecast period, and also takes into account the meteorological response capabilities of the associated provinces. By accurately evaluating the impact of multiple factors on the load, the accuracy of power load forecasting can be effectively improved, providing a reliable basis for power system planning and scheduling, helping to rationally allocate resources and reduce supply risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. Figure 1 It is a flow chart of a method for predicting provincial power load by combining meteorological and economic data disclosed in an embodiment of the present invention; Figure 2 It is a schematic diagram of the structure of the second depth prediction model disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] like Figure 1 As shown, the embodiment of the present invention discloses a method for predicting the province's power load by combining meteorological and economic data, comprising the following steps: S10, collecting historical meteorological data and historical economic data related to the target province, and determining the first meteorological forecast data and the first meteorological response big data for the future forecast period corresponding to the target province, and determining the second meteorological forecast data and the second meteorological response big data for the associated provinces corresponding to the target province.

[0020] Among them, historical meteorological data contains information such as temperature, humidity, and wind speed, which will affect the electricity consumption of residents and industrial users. For example, in a certain target province, during the high temperatures in the past summer, the electricity consumption of residents for air conditioning increased significantly. By collecting temperature data and corresponding power load data in the past many years of summer, it can provide a basis for subsequent analysis. Historical economic data such as GDP, industrial output value, and consumer spending reflect the demand for electricity in the province's economic activities. For example, when industrial output value increases, the electricity consumption of factories will increase accordingly.

[0021] Next, determine the first meteorological forecast data and the first meteorological response big data (including multiple data) for the future forecast period (such as the next month, a quarter, half a year, etc.) corresponding to the target province. The first meteorological forecast data is an estimate of the meteorological conditions of the target province in the future forecast period, such as the average temperature, number of rainy days, probability and degree of flood disasters in the target province in the next month. The first meteorological response big data involves data related to the province's ability to respond to meteorological changes, such as the number of power generation equipment reserved by the province to respond to extreme weather (such as floods caused by heavy rains), the construction and maintenance of emergency transmission lines, the repair of rivers and reservoirs, the construction of flood control facilities (completed, planned), and the purchase of flood control equipment. For example, if the target province may face high temperature weather in the future forecast period, the first meteorological forecast data will reflect the number of high temperature days and temperature range, and the first meteorological response big data may show that the province has prepared enough raw materials for cooling equipment to respond to high temperatures. These data reflect the ability to respond to abnormal weather, which is related to the impact of abnormal weather on production enterprises in the region, and the impact is related to power load. For example, when the target province has a poor ability to respond to abnormal weather, more enterprises in the region will stop production due to abnormal weather, and the electricity load in the target province will be lower than expected.

[0022] At the same time, the second meteorological forecast data and second meteorological response big data of the associated provinces corresponding to the target province are also determined. Associated provinces refer to provinces that are economically related or dependent on the target province, especially neighboring provinces. The meteorological and response data of these provinces also have an impact on the power load of the target province. For example, if the second meteorological forecast data of an associated province predicts that there will be severe rainstorms in the future, and the second meteorological response big data shows that the province’s material reserves for responding to the rainstorm are insufficient, it may be necessary to obtain materials from the target province. The production of these materials will cause the factories in the target province to increase electricity consumption, thereby affecting the power load of the target province.

[0023] S20, using a first deep prediction model to predict the historical meteorological data and the historical economic data, to obtain a first provincial power load forecast value corresponding to the future forecast period.

[0024] Among them, the first deep prediction model is a model based on deep learning algorithms (such as CNN, Transformer, etc.), which can use the collected historical meteorological and economic data of the target province to predict the power load of the target province in the future forecast period. For example, the first deep prediction model learned that in the past few years, when GDP grew by 10% and the average summer temperature rose by 2°C, the power load would increase by 15%. Based on the expected GDP growth and temperature changes in the future forecast period, the first deep prediction model predicts the first provincial power load forecast value.

[0025] However, the above-mentioned first provincial power load forecast value does not take into account the meteorological response capabilities and the impact of related provinces, so it is a preliminary forecast result.

[0026] S30, using a second deep prediction model to perform deep prediction on the first meteorological forecast data and the first meteorological response big data, the second meteorological forecast data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively.

[0027] Among them, the second deep prediction model (such as CNN, Transformer, etc.) is used to analyze the matching degree between the meteorological data and the response capability of the province. The second deep prediction model first analyzes the first meteorological forecast data and the first meteorological response big data of the target province to obtain the first impact coefficient. For example, when the first impact coefficient is 0.8, it means that the target province is not well prepared for abnormal weather, which may cause it to be greatly affected by floods and waterlogging disasters in the future forecast period, resulting in interruption of enterprise production, which will cause the actual power load of the target province to be lower than expected.

[0028] Similar to the above, the second deep prediction model also analyzes the second meteorological forecast data and the second meteorological response big data of the associated province to obtain the second impact coefficient. For example, if the associated province predicts a serious meteorological disaster, its response capacity is insufficient, and it needs to purchase a large amount of materials from the target province, the second impact coefficient is, for example, 1.2. Obviously, the second impact coefficient at this time will cause the actual power load of the target province to be higher than expected.

[0029] S40: Correct the first province-wide power load forecast value based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value.

[0030] Since the first province-wide power load forecast value obtained in the aforementioned steps does not take into account the target province's own meteorological response capability and the impact of related provinces, it is corrected by the first influence coefficient and the second influence coefficient.

[0031] For example, the first provincial power load forecast value is 1 million kWh, the first impact coefficient indicates that the power load is reduced by 10% due to the insufficient meteorological response capability of the target province itself, and the second impact coefficient indicates that the power load needs to be increased by 5% due to the demand of related provinces. Then the final revised second provincial power load forecast value is 100×(1-10%)×(1+5%)=945,000 kWh. The forecast value obtained in this way more accurately considers the impact of multiple factors on the power load, making up for the limitations of the existing method that lacks consideration of the impact of meteorological response capabilities and economic activities in related provinces.

[0032] The power load forecasting method of the present invention takes into account the meteorological response capabilities of the target province in the future forecast period, and also takes into account the meteorological response capabilities of the associated provinces. By accurately evaluating the impact of multiple factors on the load, the accuracy of power load forecasting can be effectively improved, providing a reliable basis for power system planning and scheduling, helping to rationally allocate resources and reduce supply risks.

[0033] In some embodiments, using a second deep prediction model to perform deep prediction on the first meteorological prediction data and the first meteorological response big data, the second meteorological prediction data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively, including: Using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological forecast data according to the first clustering index to obtain the first meteorological key semantic information; using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological response big data according to the first clustering index to obtain the first meteorological response key semantic information; Using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological forecast data according to the second clustering index to obtain second meteorological key semantic information; using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological response big data according to the second clustering index to obtain second meteorological response key semantic information; Based on the first meteorological key semantic information and the first meteorological response key semantic information, the second meteorological key semantic information and the second meteorological response key semantic information, respectively, a response capability question is asked to the third semantic analysis model to obtain a first response capability and a second response capability; wherein the third semantic analysis model is constructed based on a general large model; The first coping capability, the second coping capability and a coping capability-influence coefficient comparison table are compared to obtain the first influence coefficient and the second influence coefficient respectively.

[0034] In this embodiment, if Figure 2As shown, the second depth prediction model of the present invention includes three semantic analysis models, namely the first semantic analysis model, the second semantic analysis model and the third semantic analysis model. Among them, the first semantic analysis model and the second semantic analysis model are constructed by hierarchical clustering algorithm, convolutional neural network (CNN), etc.

[0035] Among them, the hierarchical clustering algorithm can perform hierarchical clustering analysis on text data. In meteorological semantic analysis, agglomerative hierarchical clustering can start from a single word or phrase, and gradually merge according to semantic similarity to form semantic categories at different levels; divisive hierarchical clustering is the opposite, starting from the entire text set and gradually splitting into smaller semantic subsets. Through the hierarchical clustering algorithm, the semantic information in the meteorological forecast data and the meteorological response big data can be classified according to the degree of similarity, so as to filter out key semantic information, such as clustering abnormal meteorological data of the same type and degree into one category, and clustering semantic information related to meteorological disaster response into one category. Then, the data with a meteorological abnormality higher than the preset abnormality and a capability to respond to abnormal meteorology higher than the preset response capability are filtered out, so as to obtain the above-mentioned first / second meteorological key semantic information and first / second meteorological response key semantic information.

[0036] The third semantic analysis model is built based on a general large model, such as DeepSeek, GPT, Wenxinyiyan, etc. The two types of key semantic information obtained above can be input into the third semantic analysis model respectively, and questions can be asked to it, such as "the additional semantic information is the meteorological forecast data and meteorological response capability data of a certain province in a certain period of time in the future. Please evaluate the degree of match between the response capability of the target province and the above meteorological forecast data, that is, the response capability level". After in-depth thinking, the third semantic analysis model outputs the above-mentioned first response capability and second response capability. A response capability-impact coefficient comparison table is pre-constructed. By querying the comparison table, the above-mentioned first impact coefficient and second impact coefficient can be quickly determined, providing a quantitative basis for correcting the power load forecast value.

[0037] In some embodiments, the clustering distances of the first clustering indicator and the second clustering indicator are different, and the clustering distance of the first clustering indicator is greater than the clustering distance of the second clustering indicator; wherein the clustering distance is the minimum value indicating that two data points or two clusters are similar data that can be clustered.

[0038] In this embodiment, when the first semantic analysis model performs semantic clustering and screening on meteorological data, cluster distance is an important indicator, which plays a key role in judging the similarity between data points (meteorological forecast data such as heavy rain, meteorological response data such as flood control measures preparation) and determining the clustering results. It should be noted that the clustering in the present invention is for similar data, such as clustering meteorological data of different temperatures, clustering meteorological data of different rainfall, or clustering the maintenance of emergency transmission lines and the repair data of rivers and reservoirs (maintenance category).

[0039] The cluster distance is the minimum value that indicates that two data points or two clusters are similar data that can be clustered. For example, if the similarity between the two meteorological data "high temperature 30 degrees Celsius" and "hot climate 35 degrees Celsius" is higher than the above cluster distance, the two will be clustered into one category, such as "high temperature"; while the similarity between "high temperature" and "heavy rainfall" is lower than the above cluster distance, the two will not be clustered into one category. By using different cluster distances, different degrees of clustering can be obtained for meteorological forecast data and meteorological response big data, thereby obtaining different meteorological key semantic information and meteorological response key semantic information.

[0040] Since the scheme of the present invention is to predict the power load of the target province, the meteorological response capability of the province should be the main factor, and the meteorological response capability of the associated provinces should be the auxiliary factor. Therefore, the clustering distance of the first clustering index set by the present invention is greater than the clustering distance of the second clustering index. This arrangement allows the first meteorological key semantic information and the first meteorological response key semantic information to retain more detailed features than the second meteorological key semantic information and the second meteorological response key semantic information, which is conducive to the third semantic analysis model to analyze more accurate response capabilities. In addition, the degree of influence of the associated provinces is usually not too large, so the accuracy requirement for the second influence coefficient is set to be lower. Correspondingly, the clustering distance of the second clustering index is set to be smaller, that is, the clustered second meteorological key semantic information and the second meteorological response key semantic information retain fewer detailed features, which can significantly improve the analysis rate of the third semantic analysis model for the second response capability.

[0041] In some embodiments, the first clustering index is determined by: Retrieving abnormal weather loss record data of the target province for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the first loss characteristic data, and the first Euclidean distance between the second loss characteristic data of the most recent year after vectorization processing and the first loss characteristic data is calculated. The clustering distance of the first clustering indicator is determined based on the first Euclidean distance and the standard clustering distance.

[0042] In some embodiments, the second clustering index is determined by: Retrieving abnormal weather loss record data of related provinces for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the third loss characteristic data, and the second Euclidean distance between the fourth loss characteristic data of the most recent year after vectorization processing and the third loss characteristic data is calculated. The clustering distance of the second clustering indicator is determined based on the second Euclidean distance and the standard clustering distance.

[0043] In this embodiment, the method of determining the clustering distance of the first clustering index is firstly taken as an example: Retrieve the target province’s record data on abnormal weather losses in the past several years. These data record the various losses caused by abnormal weather (such as heavy rain, typhoons, high temperatures and other extreme weather) in the target province in different years in the past, which may include economic losses (such as losses from disasters on crops, repair costs for damaged infrastructure, etc.), casualties and other related information.

[0044] Then, each abnormal weather loss record data is vectorized, that is, these non-numerical data (such as the loss situation described in text) are converted into numerical vectors. For example, the text description is converted into a vector using technologies such as word embedding. Each dimension in the vector represents a certain feature of the data.

[0045] The abnormal weather loss record data of non-recent years (e.g., 2024 is the most recent year, and years before 2024 are non-recent years) after vectorization processing are integrated into the first loss feature data. The integration here includes statistics, aggregation and other operations on the data of each non-recent year, such as calculating the average value, sum, etc. of the same type of losses in each year, to form a comprehensive feature vector to represent the overall abnormal weather loss characteristics of non-recent years. For the abnormal weather loss record data of the most recent year after vectorization processing, the second loss feature data is formed separately, which represents the feature vector of the abnormal weather loss situation in the most recent year.

[0046] Calculate the Euclidean distance between the second loss feature data and the first loss feature data. The Euclidean distance represents the similarity between the two vectors. Obviously, the larger the Euclidean distance between the two, the lower the similarity between the two. Determine the clustering distance of the first clustering indicator based on the calculated Euclidean distance and the pre-set standard clustering distance (a reference distance value set based on experience, historical data or specific business needs). Specifically, determine the corresponding coefficient value based on the Euclidean distance, and then multiply the coefficient value by the standard clustering distance. For example, the coefficient value corresponding to a certain Euclidean distance is 1.2, and the standard clustering distance is 65% (similarity threshold, that is, the above-mentioned minimum value, namely the threshold). The clustering distance of the first clustering indicator is calculated to be 1.2*65%=78%. For example, when the similarity between two meteorological forecast data reaches 78%, the two are clustered into one category, for example, both are classified as "high temperature".

[0047] It should be noted that the Euclidean distance should be positively correlated with the above clustering distance. That is, the larger the Euclidean distance, the lower the similarity between the abnormal meteorological losses in the recent year and the earlier year, that is, the more abnormal it is. At this time, the larger the coefficient value is set (for example, 1.2), the larger the clustering distance obtained, so that more detailed features are retained in the meteorological key semantic information and the meteorological response key semantic information, so as to improve the accuracy of the influence coefficient for correcting the forecast value of the power load in the whole province; conversely, the smaller the coefficient value is set (for example, 1.1), the smaller the clustering distance is, so that fewer detailed features are retained in the meteorological key semantic information and the meteorological response key semantic information, that is, so many detailed features are not needed.

[0048] The method for determining the clustering distance of the second clustering index is the same and will not be repeated here.

[0049] In some embodiments, if the clustering distance of the first clustering indicator is smaller than the clustering distance of the second clustering indicator, the method further includes: The clustering distance of the first clustering indicator is increased at least once using a preset adjustment coefficient to make it greater than the clustering distance of the second clustering indicator.

[0050] In this embodiment, since the clustering distances of the first clustering index and the second clustering index are independently calculated based on the abnormal meteorological loss record data of the target province and the associated province, the clustering distance of the first clustering index may be smaller than the clustering distance of the second clustering index. At this time, it is necessary to use a preset adjustment coefficient to gradually increase the clustering distance of the first clustering index until it is greater than the clustering distance of the second clustering index. The preset adjustment coefficient is, for example, 1.1 or 1.2, and the present invention does not specifically limit this.

[0051] In some embodiments, the correcting the first province-wide power load forecast value based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value includes: The first influence coefficient and the second influence coefficient are multiplied by the first province-wide power load forecast value at the same time to obtain a revised second province-wide power load forecast value.

[0052] An embodiment of the present invention further discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method as described in any of the preceding items.

[0053] The embodiment of the present invention further discloses a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any of the above items is executed.

[0054] An embodiment of the present invention further discloses a computer program product, which includes a computer program stored in a computer storage medium, and when the computer program is executed by a processor of an electronic device, any of the above methods is implemented.

[0055] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0056] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0057] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0058] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of this disclosure can be achieved, and this document is not limited here.

[0059] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for predicting provincial power load by combining meteorological and economic data, characterized in that: The steps include: Collect historical meteorological data and historical economic data related to the target province, and determine first meteorological forecast data and first meteorological response big data for a future forecast period corresponding to the target province, and determine second meteorological forecast data and second meteorological response big data for associated provinces corresponding to the target province; Use the first deep prediction model to predict the historical meteorological data and the historical economic data to obtain a first province-wide power load prediction value corresponding to the future prediction period; Use a second deep prediction model to perform deep prediction on the first meteorological prediction data and the first meteorological response big data, the second meteorological prediction data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively; The first province-wide power load forecast value is corrected based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value.

2. A provincial power load forecasting method combining meteorological and economic data according to claim 1, characterized in that: Use the second deep prediction model to perform deep prediction on the first meteorological prediction data and the first meteorological response big data, the second meteorological prediction data and the second meteorological response big data, respectively, to obtain a first influence coefficient and a second influence coefficient, respectively, including: Using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological forecast data according to the first clustering index to obtain the first meteorological key semantic information; using the first semantic analysis model to perform semantic clustering and screening processing on the first meteorological response big data according to the first clustering index to obtain the first meteorological response key semantic information; Using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological forecast data according to the second clustering index to obtain second meteorological key semantic information; using the second semantic analysis model to perform semantic clustering and screening processing on the second meteorological response big data according to the second clustering index to obtain second meteorological response key semantic information; Based on the first meteorological key semantic information and the first meteorological response key semantic information, the second meteorological key semantic information and the second meteorological response key semantic information, respectively, a response capability question is asked to the third semantic analysis model to obtain a first response capability and a second response capability; wherein the third semantic analysis model is constructed based on a general large model; The first coping capability, the second coping capability and a coping capability-influence coefficient comparison table are compared to obtain the first influence coefficient and the second influence coefficient respectively.

3. A method for predicting provincial power load by combining meteorological and economic data according to claim 2, characterized in that: The clustering distances of the first clustering indicator and the second clustering indicator are different, and the clustering distance of the first clustering indicator is greater than the clustering distance of the second clustering indicator; wherein the clustering distance is the minimum value indicating that two data points or two clusters are similar data that can be clustered.

4. A method for predicting provincial power load by combining meteorological and economic data according to claim 3, characterized in that: The first clustering index is determined by: Retrieving abnormal weather loss record data of the target province for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the first loss characteristic data, and the first Euclidean distance between the second loss characteristic data of the most recent year after vectorization processing and the first loss characteristic data is calculated. The clustering distance of the first clustering indicator is determined based on the first Euclidean distance and the standard clustering distance.

5. The method for predicting the province's power load by combining meteorological and economic data according to claim 3 is characterized by: The second clustering index is determined by: Retrieving abnormal weather loss record data of related provinces for several years, and vectorizing each abnormal weather loss record data; The abnormal meteorological loss record data of non-recent years after vectorization processing are integrated into the third loss characteristic data, and the second Euclidean distance between the fourth loss characteristic data of the most recent year after vectorization processing and the third loss characteristic data is calculated. The clustering distance of the second clustering indicator is determined based on the second Euclidean distance and the standard clustering distance.

6. A method for predicting provincial power load by combining meteorological and economic data according to claim 5, characterized in that: If the clustering distance of the first clustering indicator is smaller than the clustering distance of the second clustering indicator, the method further includes: The clustering distance of the first clustering indicator is increased at least once using a preset adjustment coefficient to make it greater than the clustering distance of the second clustering indicator.

7. A method for predicting provincial power load by combining meteorological and economic data according to claim 1, characterized in that: The step of correcting the first province-wide power load forecast value based on the first influence coefficient and the second influence coefficient to obtain a second province-wide power load forecast value includes: The first influence coefficient and the second influence coefficient are multiplied by the first province-wide power load forecast value at the same time to obtain a revised second province-wide power load forecast value.

Images