Photovoltaic load prediction method and system based on multi-source data deep learning

Abstract

According to the photovoltaic load prediction method and system based on multi-source data deep learning provided by the invention, the first photovoltaic load description content list is determined, and then the to-be-predicted photovoltaic load description content and each matched photovoltaic load description content are subjected to load prediction processing; obtaining a load prediction result of each matched photovoltaic load description content; and finally, performing important content derivation processing on the load prediction result of each matched photovoltaic load description content to obtain the load prediction result of each matched photovoltaic load description content subjected to important content derivation, so that in the embodiment of the invention, after the primary load prediction processing is performed on the multiple groups of photovoltaic load description contents, the load prediction result of each matched photovoltaic load description content subjected to important content derivation is obtained. And important content derivation processing can be carried out on the photovoltaic load description content after the first-order load prediction processing is carried out, so that the load prediction precision can be more effectively improved.

Description

Technical Field

[0001] The present application relates to the field of data prediction technology, and specifically, to a photovoltaic load prediction method and system based on deep learning of multi-source data. Background Art

[0002] Photovoltaic load refers to the power load borne by a photovoltaic power station, including internal power consumption and external power supply. The load of a photovoltaic power station directly determines the power generation efficiency and economic benefits of the power station, so it is very important to correctly judge the load of a photovoltaic power station.

[0003] At present, with the continuous development of photovoltaic technology, photovoltaic technology is being applied in more and more places. Photovoltaic load is very important because it can avoid damage to electrical appliances and equipment. However, how to accurately predict photovoltaic load is a technical problem that is difficult to solve at present. Summary of the invention

[0004] In view of this, the present application provides a photovoltaic load prediction method and system based on deep learning of multi-source data.

[0005] In a first aspect, a photovoltaic load forecasting method based on multi-source data deep learning is provided, which is applied to a network security forecasting system, and the method at least comprises: Determine a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; perform load prediction processing on the photovoltaic load description content to be predicted and each matching photovoltaic load description content, respectively, to obtain load prediction results of each matching photovoltaic load description content; Based on the photovoltaic load description content to be predicted, important content derivation processing is performed on the load prediction results of each matching photovoltaic load description content to obtain the load prediction results after the important content is derived from each matching photovoltaic load description content; based on the load prediction results derived from the important content, the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted is determined; wherein the target photovoltaic load description content represents the load description content of the photovoltaic power generation system converting solar energy into electrical energy and transmitting electrical energy through solar panels.

[0006] It should be understood that when executing the above content, the embodiment of the present disclosure determines a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; then the photovoltaic load description content to be predicted is subjected to load prediction processing with each matching photovoltaic load description content respectively, to obtain load prediction results of each matching photovoltaic load description content; finally, based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing, to obtain load prediction results of each matching photovoltaic load description content after completing important content derivation, wherein the load prediction results after completing important content derivation are used to determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted. In this way, in the embodiment of the present disclosure, after performing primary load prediction processing on multiple groups of photovoltaic load description contents, important content derivation processing can also be performed on the photovoltaic load description content after the primary load prediction processing is performed, so as to more effectively improve the accuracy of load prediction.

[0007] In this application, the load forecasting process includes the following steps: The description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content are associated and analyzed to obtain a load prediction result of the matching photovoltaic load description content, wherein both the first key load description feature and the second key load description feature contain description contents with at least one different feature analysis degree, and the feature analysis degree of the load prediction result of the matching photovoltaic load description content is the same as the feature analysis degree of the photovoltaic load description content to be predicted.

[0008] It should be understood that when executing the above content, the accuracy of the load forecasting results can be improved by performing multi-dimensional processing on the description content.

[0009] In the present application, the first key load description feature of the photovoltaic load description content to be predicted and the description content in the second key load description feature matching the photovoltaic load description content are associated and analyzed to obtain the load prediction result of the matching photovoltaic load description content, including: Performing the first feature expansion processing on the description content with the lowest feature analysis degree in the first key load description feature and the description content with the lowest feature analysis degree in the second key load description feature to obtain a basic feature evaluation index; Based on the description content in the first key load description feature, the description content in the second key load description feature and the basic feature evaluation index, the association process and the abnormality analysis process are implemented one by one in a loop until a processed load prediction result having the same feature analysis degree as the feature analysis degree of the photovoltaic load description content to be predicted is determined, and the processed load prediction result is the result of the association process; The processed load forecasting result having the same characteristic analysis degree as the characteristic analysis degree of the photovoltaic load description content to be forecasted is used as the load forecasting result.

[0010] It should be understood that when executing the above content, when the description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature that matches the photovoltaic load description content are associated and analyzed, the problem of inaccurate basic feature evaluation indicators is improved, so that the load prediction result of the matching photovoltaic load description content can be obtained.

[0011] In this application, the steps of the association process include: Performing feature translation processing on the feature evaluation index with the highest feature analysis degree to obtain the feature evaluation index after feature translation; Determine, from the second key load description feature, a first description content having the same feature analysis degree as the feature evaluation indicator after the feature translation; The first description content and the feature evaluation index after feature translation are bucketed to obtain a processed load forecast result. The processed load forecast result has the same feature analysis degree as the feature evaluation index after feature translation. The processed load forecast result is used to implement the abnormality analysis processing to obtain a feature evaluation index with the same feature analysis degree as the processed load forecast result.

[0012] It should be understood that when executing the above content, the credibility of the feature evaluation index can be effectively improved through feature translation processing, so that the processed load forecasting result can be accurately obtained.

[0013] In this application, the steps of abnormal analysis and processing include: Determining, from the first key load description feature, a second description content having the same feature analysis degree as the feature evaluation indicator after the feature translation; Performing a first feature expansion process on the processed load forecast result and the second description content; Based on the result of the first feature expansion processing and the feature evaluation index after the feature translation, a candidate feature evaluation index is obtained, and the candidate feature evaluation index is used for the next implementation of the association processing to obtain a candidate processed load forecast result.

[0014] It should be understood that, when executing the above content, the credibility of the candidate feature evaluation index can be effectively improved by performing the first feature expansion process on the second description content.

[0015] In this application, the important content derivation process includes the following steps: Determine key evaluation data of a load prediction result matching the photovoltaic load description content based on the photovoltaic load description content to be predicted, wherein the key evaluation data includes a first load level ranking of the load prediction result matching the photovoltaic load description content and a second load level ranking of the highest characteristic analysis degree description content of the photovoltaic load description content to be predicted, wherein the characteristic analysis degree of the highest characteristic analysis degree description content is the same as the characteristic analysis degree of the load prediction result matching the photovoltaic load description content; Determine the load forecast result matching the photovoltaic load description content and the first total forecast result of the first load level sorting; Determine a second overall prediction result of the highest feature analysis level description content and the second load level ranking; Based on the first all prediction results and the second all prediction results, a load prediction result derived from important contents matching the photovoltaic load description contents is obtained.

[0016] It should be understood that when executing the above content, predictions are made from two perspectives, so that the first complete prediction result and the second complete prediction result can be accurately obtained, and comparison and prediction processing are performed based on the two complete prediction results, thereby improving the accuracy of the load prediction results.

[0017] In the present application, the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted is determined based on the load forecasting result derived from the completed important content, including: using a second AI network model to adjust the load forecasting results derived from the important content of each matching photovoltaic load description content to obtain the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted.

[0018] It should be understood that when executing the above content, when the load prediction result derived from the important content is completed, the problem of defects in each matching photovoltaic load description content is improved, so that the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted can be accurately determined.

[0019] In a second aspect, a photovoltaic load forecasting system based on deep learning of multi-source data is provided, comprising a processor and a memory that communicate with each other, wherein the processor is used to retrieve a computer program from the memory and implement the above method by running the computer program.

[0020] The photovoltaic load prediction method and system based on multi-source data deep learning provided in the embodiment of the present application, by determining a first photovoltaic load description content list, the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; then the photovoltaic load description content to be predicted is respectively subjected to load prediction processing with each matching photovoltaic load description content to obtain the load prediction results of each matching photovoltaic load description content; finally, based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing to obtain the load prediction results of each matching photovoltaic load description content after completing the important content derivation, wherein the load prediction results after completing the important content derivation are used to determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted. In this way, in the embodiment of the present disclosure, after performing primary load prediction processing on multiple groups of photovoltaic load description contents, important content derivation processing can also be performed on the photovoltaic load description content after the primary load prediction processing is implemented, which can more effectively improve the accuracy of load prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A flowchart of a photovoltaic load prediction method based on deep learning of multi-source data provided in an embodiment of the present application.

[0023] Figure 2 A block diagram of a photovoltaic load prediction device based on deep learning of multi-source data provided in an embodiment of the present application.

[0024] Figure 3 An architectural diagram of a photovoltaic load forecasting system based on deep learning of multi-source data provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0026] See also Figure 1 , shows a photovoltaic load forecasting method based on deep learning of multi-source data, which may include the technical solutions described in the following steps 101-104.

[0027] 101 ; determining a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted.

[0028] In this embodiment, the matching photovoltaic load description content may be photovoltaic load description content that is continuous with the photovoltaic load description content to be predicted, or may be photovoltaic load description content that is separated from the photovoltaic load description content to be predicted by one or more groups.

[0029] 102; The photovoltaic load description content to be predicted may be processed with each matching photovoltaic load description content to obtain load prediction results of each matching photovoltaic load description content.

[0030] In the process of processing the photovoltaic load description content, it is necessary to select at least one photovoltaic load description content as a reference group for matching processing, and the remaining photovoltaic load description contents are matched to the reference group. It is further understood that in the embodiment of the present disclosure, the above-mentioned reference group can be used as the photovoltaic load description content to be predicted, and the photovoltaic load description content to be predicted and at least one photovoltaic load description content that has a matching relationship with the above-mentioned photovoltaic load description content to be predicted constitute the above-mentioned photovoltaic load description content list.

[0031] Furthermore, in a specific implementation scenario, the above-mentioned photovoltaic load description content to be predicted can be separately processed with each matching photovoltaic load description content for load forecasting, or the photovoltaic load description content to be predicted can be processed with an example for load forecasting, that is, the photovoltaic load description content example to be predicted can also be considered as a matching photovoltaic load description content, thereby obtaining multiple load forecasting results.

[0032] In a possible implementation example, before the above-mentioned photovoltaic load description content to be predicted can be respectively processed with each matching photovoltaic load description content for load forecasting, the photovoltaic load description content to be predicted and at least one matching photovoltaic load description content can be network selected according to the above-mentioned content to obtain a first key load description feature of the above-mentioned photovoltaic load description content to be predicted and at least one second key load description feature of the matching photovoltaic load description content in turn.

[0033] Furthermore, the above-mentioned description content is obtained after network selection of the photovoltaic load description content in the above-mentioned event list. Therefore, the above-mentioned photovoltaic load description content list can be subjected to network selection of multiple feature analysis degrees, so as to obtain description content with different feature analysis degrees. Then, the description content with different feature analysis degrees of the above-mentioned photovoltaic load description content to be predicted can be regarded as the above-mentioned first key load description feature, and the description content of multiple feature analysis degrees of each matching photovoltaic load description content can be regarded as the second key load description feature of the matching photovoltaic load description content. After obtaining the above-mentioned first key load description feature and the above-mentioned second key load description feature, the description content in the above-mentioned first key load description feature and the above-mentioned second key load description feature is used to perform association processing and abnormal analysis processing, so as to obtain the load prediction result of the matching photovoltaic load description content corresponding to the above-mentioned second key load description feature.

[0034] Exemplarily, a feature determination method may be used to perform network selection on the photovoltaic load description content in the above event list to obtain description content with different feature analysis degrees.

[0035] Furthermore, the first AI network model can be used to perform network selection on the photovoltaic load description content to be predicted to obtain the first key load description feature of the flat structure, and the first AI network model can be used to perform network selection on the matching photovoltaic load description content to obtain the second key load description feature of the flat structure.

[0036] For example, the first AI network model can be used to perform network selection on the photovoltaic load description content to be predicted and the matching photovoltaic load description content to obtain two description contents with different characteristic analysis degrees of the photovoltaic load description content to be predicted, and obtain two description contents with different characteristic analysis degrees of the matching photovoltaic load description content. Then, the description contents with two different characteristic analysis degrees of the photovoltaic load description content to be predicted are considered as the first key load description feature, and the description contents with two different characteristic analysis degrees of the matching photovoltaic load description content can be considered as the second key load description feature. Among them, the two characteristic analysis degrees of the description contents with two different characteristic analysis degrees of the photovoltaic load description content to be predicted and the two characteristic analysis degrees of the description contents with two different characteristic analysis degrees of the matching photovoltaic load description content correspond to the same. Furthermore, after network selection, no less than three description contents with different characteristic analysis degrees can also be obtained, and the embodiments of the present disclosure are not limited one by one.

[0037] In a possible implementation example, the load forecasting process includes the following steps: The description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content are subjected to association processing and abnormal analysis processing to obtain the load prediction result of the matching photovoltaic load description content. Among them, the description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content are subjected to association processing and abnormal analysis processing to obtain the load prediction result of the matching photovoltaic load description content, which may specifically include the following contents: the description contents with the lowest feature analysis degree in the first key load description feature and the description contents with the lowest feature analysis degree in the second key load description feature are subjected to the first feature expansion processing to obtain the basic feature evaluation index; then, based on the description contents in the first key load description feature and the second key load description feature and the basic feature evaluation index, the association processing and abnormal analysis processing are cyclically implemented one by one until the processed load prediction result with the same feature analysis degree as the feature analysis degree of the photovoltaic load description content to be predicted is determined. The processed load forecast result is the result of the association processing; finally, the processed load forecast result having the same characteristic analysis degree as the characteristic analysis degree of the photovoltaic load description content to be predicted is used as the load forecast result.

[0038] In the embodiment of the present disclosure, when executing the load forecasting process, it is necessary to implement the above-mentioned association processing and the above-mentioned abnormal analysis processing one by one in a loop, and the stopping requirement in the process of the loop implementation can refer to the following steps: the characteristic analysis degree of the result of executing the association processing (it should be understood that it is the characteristic analysis degree of the processed load forecasting result) is the same as the characteristic analysis degree of the above-mentioned photovoltaic load description content to be predicted.

[0039] In this embodiment, the steps of the above-mentioned association processing may specifically include the following: performing feature translation processing on the feature evaluation index with the highest feature analysis degree to obtain the feature evaluation index after feature translation; determining the first description content with the same feature analysis degree as the feature evaluation index after feature translation from the above-mentioned second key load description feature; performing a compensation operation on the above-mentioned first description content and the feature evaluation index after feature translation to obtain a processed load forecast result. Among them, the processed load forecast result has the same feature analysis degree as the feature evaluation index after feature translation, and the processed load forecast result is used to implement the abnormal analysis processing to obtain a feature evaluation index with the same feature analysis degree as the processed load forecast result.

[0040] In the embodiment of the present disclosure, the feature evaluation index may represent the error amount of the description content of the photovoltaic load description content to be predicted and the description content of the matching event at each location relative to the description content of the photovoltaic load description content to be predicted for the same feature analysis degree.

[0041] In this embodiment, the above-mentioned abnormal analysis processing step may specifically include the following contents: determining a processed load forecast result having the same feature analysis degree as the feature evaluation index after the feature translation, and determining a second description content having the same feature analysis degree as the feature evaluation index after the feature translation from the above-mentioned first key load description feature; performing a first feature expansion processing on the processed load forecast result and the second description content; obtaining a candidate feature evaluation index based on the result of the first feature expansion processing and the feature evaluation index after the feature translation, and the candidate feature evaluation index is used for the next implementation of the association processing to obtain a candidate processed load forecast result. Among them, the feature evaluation index after the feature translation may represent the feature evaluation index obtained after feature translation of the feature evaluation index with the highest feature analysis degree currently available.

[0042] In the disclosed embodiment, the processed load forecast result is obtained by the feature evaluation index with the highest feature analysis degree and the description content matching the photovoltaic load description content. The feature evaluation index is obtained by abnormal analysis based on the processed load forecast result and the existing feature evaluation index. However, when the load forecast processing is started, there is neither the processed load forecast result nor the feature evaluation index. Therefore, it can be set that there is a first feature evaluation index, that is, the basic feature evaluation index, and the quantitative results of each positioning in the feature evaluation index are all M. The feature analysis degree of the first feature evaluation index after feature translation is the same as the feature analysis degree of the description content with the lowest feature analysis degree in the above-mentioned first key load description feature or the second key content set. Then, the description content with the same feature analysis degree as the standardized feature evaluation index in the first feature evaluation index and the second key load description feature is associated to obtain the load forecast result with the same feature analysis degree as the description content with the lowest feature analysis degree, and then the load forecast result and the first feature evaluation index are abnormally analyzed to obtain the basic feature evaluation index with the same feature analysis degree as the description content with the lowest feature analysis degree. The above association processing and the above abnormal analysis processing are continuously and cyclically implemented one by one according to the obtained basic characteristic evaluation index until a load prediction result with the same characteristic analysis degree as the above photovoltaic load description content to be predicted is obtained. It is assumed that there are three description contents in the first key load description feature and three description contents in the second key load description feature, and the characteristic analysis degree is gradually increased.

[0043] 103; Based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing to obtain the load prediction results of each matching photovoltaic load description content after completing the important content derivation.

[0044] In the disclosed embodiment, the above-mentioned important content derivation processing may mean using the description content with the highest feature analysis degree of the above-mentioned photovoltaic load description content to be predicted (or the photovoltaic load description content template to be predicted) to optimize the abnormal or erroneous information in the above-mentioned load forecasting result. For example, when the local constraint requirements in the photovoltaic load description content to be predicted are unclear or disordered in the above-mentioned matching photovoltaic load description content, the data of the local constraint requirements in the above-mentioned load forecasting result will be affected to a certain extent, so it is necessary to optimize the above-mentioned load forecasting result through the information of the local constraint requirements in the photovoltaic load description content to be predicted, so as to obtain the load forecasting result that completes the derivation of important content.

[0045] Furthermore, after obtaining the load prediction results (the characteristic analysis degree is the same as the characteristic analysis degree of the above-mentioned photovoltaic load description content to be predicted) of each matching photovoltaic load description content (which may include the photovoltaic load description content template to be predicted), the load prediction results of each matching photovoltaic load description content can be optimized using the description content with the highest characteristic analysis degree of the above-mentioned photovoltaic load description content to be predicted (or the photovoltaic load description content template to be predicted) to obtain the load prediction results derived from the completed important contents of each matching photovoltaic load description content, so as to facilitate the use of the load prediction results derived from the completed important contents of each matching photovoltaic load description content to determine the target photovoltaic load description content corresponding to the above-mentioned photovoltaic load description content to be predicted.

[0046] In a possible implementation example, the above-mentioned important content derivation processing may specifically include the following: determining the key evaluation data of the load prediction result matching the photovoltaic load description content based on the photovoltaic load description content to be predicted, the above-mentioned key evaluation data includes the first load level ranking of the load prediction result matching the photovoltaic load description content and the second load level ranking of the description content with the highest characteristic analysis degree of the photovoltaic load description content to be predicted, the characteristic analysis degree of the description content with the highest characteristic analysis degree is the same as the characteristic analysis degree of the load prediction result matching the photovoltaic load description content; determining the first total prediction result of the load prediction result matching the photovoltaic load description content and the first load level ranking; determining the second total prediction result of the description content with the highest characteristic analysis degree and the second load level ranking; based on the first total prediction result and the second total prediction result, obtaining the load prediction result matching the photovoltaic load description content to complete the derivation of important content. Among them, the above-mentioned first load level ranking represents the importance evaluation of each location in the load prediction result matching the photovoltaic load description content. The above-mentioned second load level ranking represents the importance evaluation of each location in the description content with the highest characteristic analysis degree of the photovoltaic load description content to be predicted (or the photovoltaic load description content to be predicted).

[0047] In this embodiment, the key evaluation data of the load forecasting result that matches the photovoltaic load description content based on the photovoltaic load description content to be predicted may specifically include the following: the load forecasting result that matches the photovoltaic load description content and the description content with the highest characteristic analysis degree may be subjected to a second feature expansion process, and then the first load level ranking may be obtained using the first AI network; the load forecasting result that matches the photovoltaic load description content and the description content with the highest characteristic analysis degree may be subjected to a third feature expansion process, and then the second load level ranking may be obtained using the second AI network. Then, the first load level ranking and the first total forecasting result of the load forecasting result that matches the photovoltaic load description content are determined, and the second total forecasting result of the description content with the highest characteristic analysis degree and the second load level ranking are determined; finally, the first total forecasting result and the second total forecasting result may be weighted to obtain the load forecasting result derived from the important content of the matching photovoltaic load description content.

[0048] The embodiment of the present disclosure determines a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; then the photovoltaic load description content to be predicted can be subjected to load prediction processing with each matching photovoltaic load description content respectively, to obtain load prediction results of each matching photovoltaic load description content; finally, based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing, to obtain load prediction results of each matching photovoltaic load description content after completing important content derivation, wherein the load prediction results after completing important content derivation are used to determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted. It should be understood that in the embodiment of the present disclosure, after performing primary load prediction processing on multiple groups of photovoltaic load description contents, it is also possible to perform important content derivation processing on the photovoltaic load description content after the primary load prediction processing is performed, so as to more effectively improve the accuracy of load prediction.

[0049] Furthermore, before the above photovoltaic load description contents to be predicted are respectively processed with each matching photovoltaic load description content for load prediction, the above method may further include the following contents: optimizing the photovoltaic load description contents in the above first photovoltaic load description content list.

[0050] 104; Determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted based on the load forecasting result derived from the completed important content, wherein the target photovoltaic load description content represents the load description content of the photovoltaic power generation system converting solar energy into electrical energy and transmitting electrical energy through solar panels.

[0051] In the disclosed embodiment, event adjustment can be performed through the load forecasting result derived from the above-mentioned important content to obtain the target photovoltaic load description content corresponding to the above-mentioned photovoltaic load description content to be predicted, and a target photovoltaic load description content with higher quality can be selected.

[0052] Furthermore, the event adjustment step of the load forecasting results derived from the above-mentioned important contents may include the following contents: using a second AI network model to adjust the load forecasting results derived from the important contents of each of the above-mentioned matching photovoltaic load description contents, and obtaining the target photovoltaic load description content corresponding to the above-mentioned photovoltaic load description content to be predicted.

[0053] In a possible implementation, under the premise that the accuracy of the photovoltaic load description content list in the first information set collected by the network security prediction system is not greater than the judgment value, each photovoltaic load description content in the above photovoltaic load description content list can be considered as the above photovoltaic load description content to be predicted and processed to obtain a processed photovoltaic load description content list through the steps in the photovoltaic load prediction method based on multi-source data deep learning of the embodiment of the present disclosure; the above AI network is configured using an information set containing multiple reference photovoltaic load description content list pairs, the above reference photovoltaic load description content pairs contain multiple first reference photovoltaic load description content lists and second reference photovoltaic load description content lists, and the above first reference photovoltaic load description content list is a photovoltaic load description content list obtained by compressing the above second reference photovoltaic load description content list and having an accuracy rate less than the above second reference photovoltaic load description content list.

[0054] The configured AI network can complete the steps of loading a list of photovoltaic load description contents, outputting load forecast results derived from important contents, and processing events that can determine the above target photovoltaic load description contents.

[0055] In the disclosed embodiment, in the above load prediction process, for any two features loaded (the description content of the photovoltaic load description content to be predicted and the description content of the matching photovoltaic load description content), the most preferred method is to match the matching photovoltaic load description content to the photovoltaic load description content to be predicted.

[0056] Further, a first photovoltaic load description content list is determined, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; the photovoltaic load description content to be predicted can be subjected to load prediction processing with each matching photovoltaic load description content respectively to obtain load prediction results of each matching photovoltaic load description content; based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing to obtain load prediction results derived from important contents of each matching photovoltaic load description content; based on the load prediction results derived from the completed important contents, a target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted is determined, wherein the target photovoltaic load description content represents the load description content of the photovoltaic power generation system converting solar energy into electrical energy and transmitting electrical energy through solar panels.

[0057] The embodiment of the present disclosure determines a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; then the photovoltaic load description content to be predicted can be subjected to load prediction processing with each matching photovoltaic load description content respectively, to obtain load prediction results of each matching photovoltaic load description content; finally, based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing, to obtain load prediction results of each matching photovoltaic load description content after completing important content derivation, wherein the load prediction results after completing important content derivation are used to determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted. It should be understood that in the embodiment of the present disclosure, after performing primary load prediction processing on multiple groups of photovoltaic load description contents, it is also possible to perform important content derivation processing on the photovoltaic load description content after the primary load prediction processing is performed, so as to more effectively improve the accuracy of load prediction.

[0058] In a possible implementation embodiment, the load prediction processing step may specifically include the following contents: performing association processing and abnormal analysis processing on the description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content to obtain the load prediction result of the matching photovoltaic load description content, the first key load description feature and the second key load description feature both contain description contents with at least one different feature analysis degree, and the feature analysis degree of the load prediction result of the matching photovoltaic load description content is the same as the feature analysis degree of the photovoltaic load description content to be predicted.

[0059] In a possible implementation example, the step of performing association processing and abnormal analysis processing on the first key load description feature of the photovoltaic load description content to be predicted and the description content in the second key load description feature that matches the photovoltaic load description content to obtain the load forecast result of the matching photovoltaic load description content may specifically include the following steps: performing the first feature expansion processing on the description content with the lowest feature analysis degree in the first key load description feature and the description content with the lowest feature analysis degree in the second key load description feature to obtain basic feature evaluation indicators; based on the description content in the first key load description feature and the description content in the second key load description feature and the basic feature evaluation indicators, cyclically performing the association processing and the abnormal analysis processing one by one until a processed load forecast result having the same feature analysis degree as that of the photovoltaic load description content to be predicted is determined, and the processed load forecast result is the result of the association processing; and taking the processed load forecast result having the same feature analysis degree as that of the photovoltaic load description content to be predicted as the load forecast result.

[0060] In a possible implementation embodiment, the association processing step may specifically include the following contents: performing feature translation processing on the feature evaluation indicator with the highest feature analysis degree to obtain the feature evaluation indicator after feature translation; and determining the first description content with the same feature analysis degree as the feature evaluation indicator after feature translation from the second key load description feature; and performing bucketing processing on the first description content and the feature evaluation indicator after feature translation to obtain a processed load forecast result, the processed load forecast result has the same feature analysis degree as the feature evaluation indicator after feature translation, and the processed load forecast result is used to implement the abnormality analysis processing to obtain the feature evaluation indicator with the same feature analysis degree as the processed load forecast result.

[0061] In a possible implementation embodiment, the above-mentioned abnormal analysis processing step may specifically include the following contents: determining a processed load forecasting result having the same feature analysis degree as the feature evaluation indicator after the above-mentioned feature translation and determining a second description content having the same feature analysis degree as the feature evaluation indicator after the above-mentioned feature translation from the above-mentioned first key load description feature; and performing a first feature expansion processing on the processed load forecasting result and the second description content; obtaining a candidate feature evaluation indicator based on the result of the first feature expansion processing and the feature evaluation indicator after the above-mentioned feature translation, and the candidate feature evaluation indicator is used for the next implementation of the association processing to obtain a candidate processed load forecasting result.

[0062] In a possible implementation embodiment, the important content derivation processing step may specifically include the following contents: determining key evaluation data of the load prediction result matching the photovoltaic load description content based on the photovoltaic load description content to be predicted, the key evaluation data including a first load level ranking of the load prediction result matching the photovoltaic load description content and a second load level ranking of the highest feature analysis degree description content of the photovoltaic load description content to be predicted, the feature analysis degree of the highest feature analysis degree description content being the same as the feature analysis degree of the load prediction result matching the photovoltaic load description content; and determining a first total prediction result of the load prediction result matching the photovoltaic load description content and the first load level ranking; and determining a second total prediction result of the highest feature analysis degree description content and the second load level ranking; and obtaining the load prediction result matching the photovoltaic load description content to complete the derivation of important content based on the first total prediction result and the second total prediction result.

[0063] In a possible implementation example, the importance evaluation step of determining the load prediction result matching the photovoltaic load description content based on the photovoltaic load description content to be predicted may specifically include the following contents: the load prediction result matching the photovoltaic load description content and the description content with the highest feature analysis degree may be subjected to a second feature expansion process, and then the first load level ranking may be obtained using a first AI network; and the load prediction result matching the photovoltaic load description content and the description content with the highest feature analysis degree may be subjected to a third feature expansion process, and then the second load level ranking may be obtained using a second AI network.

[0064] In a possible implementation example, before the above-mentioned photovoltaic load description content to be predicted can be respectively compared with each matching photovoltaic load description content for load forecasting processing, the following contents may be specifically included: using the first AI network model to perform network selection on the above-mentioned first photovoltaic load description content list to obtain at least one description content with a different characteristic analysis degree of the above-mentioned photovoltaic load description content to be predicted, and at least one description content with a different characteristic analysis degree of the above-mentioned matching photovoltaic load description content.

[0065] In a possible implementation example, the step of determining the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted based on the load forecasting result derived from the completed important content may specifically include the following contents: using a second AI network model to adjust the load forecasting results derived from the important contents of each matching photovoltaic load description content to obtain the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted.

[0066] In one possible implementation embodiment, the network security prediction system is implemented based on an AI network; the AI ​​network is configured using an information set including multiple reference photovoltaic load description content list pairs, the reference photovoltaic load description content pairs including multiple first reference photovoltaic load description content lists and second reference photovoltaic load description content lists, the first reference photovoltaic load description content list is a photovoltaic load description content list obtained by compressing the second reference photovoltaic load description content list, and the accuracy of the list is lower than that of the second reference photovoltaic load description content list.

[0067] In a possible implementation example, before the above photovoltaic load description contents to be predicted are respectively processed with each matching photovoltaic load description content for load prediction, the following contents may be specifically included: optimizing the photovoltaic load description contents in the above first photovoltaic load description content list.

[0068] Based on the above, please refer to Figure 2 , a photovoltaic load prediction device 200 based on multi-source data deep learning is provided, which is applied to a photovoltaic load prediction system based on multi-source data deep learning, and the device comprises: The load prediction module 210 is used to determine a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; perform load prediction processing on the photovoltaic load description content to be predicted and each matching photovoltaic load description content, respectively, to obtain load prediction results of each matching photovoltaic load description content; The content acquisition module 220 is used to perform important content derivation processing on the load prediction results of each matching photovoltaic load description content based on the photovoltaic load description content to be predicted, so as to obtain the load prediction results after the important content is derived from each matching photovoltaic load description content; based on the load prediction results derived from the important content, the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted is determined; wherein the target photovoltaic load description content represents the load description content of the photovoltaic power generation system converting solar energy into electrical energy and transmitting electrical energy through solar panels.

[0069] Based on the above, please refer to Figure 3 , shows a photovoltaic load forecasting system 300 based on deep learning of multi-source data, including a processor 310 and a memory 320 that communicate with each other, and the processor 310 is used to read and execute a computer program from the memory 320 to implement the above method.

[0070] Based on the above, a computer-readable storage medium is also provided, on which a computer program stored implements the above method when running.

[0071] In summary, based on the above scheme, the embodiment of the present disclosure determines a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; then the photovoltaic load description content to be predicted is subjected to load prediction processing with each matching photovoltaic load description content respectively, to obtain load prediction results of each matching photovoltaic load description content; finally, based on the photovoltaic load description content to be predicted, the load prediction results of each matching photovoltaic load description content are subjected to important content derivation processing, to obtain load prediction results of each matching photovoltaic load description content after completing important content derivation, wherein the load prediction results after completing important content derivation are used to determine the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted. In this way, in the embodiment of the present disclosure, after performing primary load prediction processing on multiple groups of photovoltaic load description contents, important content derivation processing can also be performed on the photovoltaic load description content after the primary load prediction processing is performed, so as to more effectively improve the accuracy of load prediction.

[0072] It should be understood that the system and its modules shown above can be implemented in various ways. For example, in some embodiments, the system and its modules can be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or a dedicated design hardware. It should be understood by those skilled in the art that the above methods and systems can be implemented using computer executable instructions and / or included in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The system and its modules of the present application can not only be implemented by hardware circuits such as ultra-large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., but can also be implemented by software such as executed by various types of processors, and can also be implemented by a combination of the above hardware circuits and software (e.g., firmware).

[0073] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or any other possible beneficial effects.

[0074] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of the present application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements and amendments to the present application. Such modifications, improvements and amendments are suggested in the present application, so such modifications, improvements and amendments still belong to the spirit and scope of the exemplary embodiments of the present application.

[0075] At the same time, the present application uses specific words to describe the embodiments of the present application. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0076] In addition, it should be understood by those skilled in the art that various aspects of the present application can be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application can be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0077] A computer storage medium may include a propagated data signal containing computer program code, for example, in baseband or as part of a carrier wave. The propagated signal may be in a variety of forms, including electromagnetic, optical, etc., or a suitable combination. A computer storage medium may be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, device or apparatus to communicate, propagate or transmit the program for use. The program code on the computer storage medium may be transmitted via any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of the above media.

[0078] The computer program codes required for the operation of each part of the present application can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python, etc., conventional procedural programming languages ​​such as C language, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages ​​such as Python, Ruby and Groovy, or other programming languages, etc. The program code can be run entirely on the user's computer, or run on the user's computer as an independent software package, or run partially on the user's computer and partially on a remote computer, or run entirely on a remote computer or server. In the latter case, the remote computer can be connected to the user's computer through any network form, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as software as a service (SaaS).

[0079] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this application, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this application. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this application. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0080] Similarly, it should be noted that in order to simplify the description of the disclosure of this application and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this application, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the object of this application are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0081] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used for the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers allow adaptive changes. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which can be changed according to the required characteristics of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining the digits. Although the numerical domains and parameters used to confirm the breadth of the range in some embodiments of the present application are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0082] Each patent, patent application, patent application disclosure, and other materials, such as articles, books, instructions, publications, documents, etc., cited in this application are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the content of this application, documents that limit the broadest scope of the claims of this application (currently or later attached to this application) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the attached materials of this application are inconsistent with or conflicting with the content described in this application, the descriptions, definitions, and / or use of terms in this application shall prevail.

[0083] Finally, it should be understood that the embodiments described in this application are only used to illustrate the principles of the embodiments of the present application. Other variations may also fall within the scope of the present application. Therefore, as an example and not a limitation, the alternative configurations of the embodiments of the present application may be considered to be the same as the teachings of the present application. Accordingly, the embodiments of the present application are not limited to the embodiments explicitly introduced and described in the present application.

[0084] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A photovoltaic load forecasting method based on multi-source data deep learning, characterized in that: The method at least comprises: Determine a first photovoltaic load description content list, wherein the first photovoltaic load description content list includes photovoltaic load description content to be predicted and at least one matching photovoltaic load description content that has a matching relationship with the photovoltaic load description content to be predicted; perform load prediction processing on the photovoltaic load description content to be predicted and each matching photovoltaic load description content, respectively, to obtain load prediction results of each matching photovoltaic load description content; Based on the photovoltaic load description content to be predicted, important content derivation processing is performed on the load prediction results of each matching photovoltaic load description content to obtain the load prediction results after the important content is derived from each matching photovoltaic load description content; based on the load prediction results derived from the important content, the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted is determined; wherein the target photovoltaic load description content represents the load description content of the photovoltaic power generation system converting solar energy into electrical energy and transmitting electrical energy through solar panels.

2. The photovoltaic load forecasting method based on multi-source data deep learning according to claim 1 is characterized in that: The load forecasting process comprises the following steps: The description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content are associated and analyzed to obtain a load prediction result of the matching photovoltaic load description content, wherein both the first key load description feature and the second key load description feature contain description contents with at least one different feature analysis degree, and the feature analysis degree of the load prediction result of the matching photovoltaic load description content is the same as the feature analysis degree of the photovoltaic load description content to be predicted.

3. The photovoltaic load forecasting method based on multi-source data deep learning according to claim 2 is characterized in that: The associating and analyzing the description contents in the first key load description feature of the photovoltaic load description content to be predicted and the second key load description feature matching the photovoltaic load description content to obtain the load prediction result matching the photovoltaic load description content includes: Performing the first feature expansion processing on the description content with the lowest feature analysis degree in the first key load description feature and the description content with the lowest feature analysis degree in the second key load description feature to obtain a basic feature evaluation index; Based on the description content in the first key load description feature, the description content in the second key load description feature and the basic feature evaluation index, the association process and the abnormality analysis process are implemented one by one in a loop until a processed load prediction result having the same feature analysis degree as the feature analysis degree of the photovoltaic load description content to be predicted is determined, and the processed load prediction result is the result of the association process; The processed load forecasting result having the same characteristic analysis degree as the characteristic analysis degree of the photovoltaic load description content to be forecasted is used as the load forecasting result.

4. The photovoltaic load forecasting method based on multi-source data deep learning according to claim 3 is characterized in that: The steps of the association process include: Performing feature translation processing on the feature evaluation index with the highest feature analysis degree to obtain the feature evaluation index after feature translation; Determine, from the second key load description feature, a first description content having the same feature analysis degree as the feature evaluation indicator after the feature translation; The first description content and the feature evaluation index after feature translation are bucketed to obtain a processed load forecast result. The processed load forecast result has the same feature analysis degree as the feature evaluation index after feature translation. The processed load forecast result is used to implement the abnormality analysis processing to obtain a feature evaluation index with the same feature analysis degree as the processed load forecast result.

5. The photovoltaic load forecasting method based on multi-source data deep learning according to claim 4 is characterized in that: The steps of the abnormal analysis process include: Determining, from the first key load description feature, a second description content having the same feature analysis degree as the feature evaluation indicator after the feature translation; Performing a first feature expansion process on the processed load forecast result and the second description content; Based on the result of the first feature expansion processing and the feature evaluation index after the feature translation, a candidate feature evaluation index is obtained, and the candidate feature evaluation index is used for the next implementation of the association processing to obtain a candidate processed load forecast result.

6. The photovoltaic load forecasting method based on multi-source data deep learning according to any one of claims 1 to 5, characterized in that: The important content derivation process comprises the following steps: Determine key evaluation data of a load prediction result matching the photovoltaic load description content based on the photovoltaic load description content to be predicted, wherein the key evaluation data includes a first load level ranking of the load prediction result matching the photovoltaic load description content and a second load level ranking of the highest characteristic analysis degree description content of the photovoltaic load description content to be predicted, wherein the characteristic analysis degree of the highest characteristic analysis degree description content is the same as the characteristic analysis degree of the load prediction result matching the photovoltaic load description content; Determine the load forecast result matching the photovoltaic load description content and the first total forecast result of the first load level sorting; Determine a second overall prediction result of the highest feature analysis level description content and the second load level ranking; Based on the first all prediction results and the second all prediction results, a load prediction result derived from important contents matching the photovoltaic load description contents is obtained.

7. The photovoltaic load forecasting method based on multi-source data deep learning according to claim 6 is characterized in that: The determining of the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted based on the load forecasting result derived from the completed important content includes: using a second AI network model to adjust the load forecasting results derived from the important content of each matching photovoltaic load description content to obtain the target photovoltaic load description content corresponding to the photovoltaic load description content to be predicted.

8. A photovoltaic load forecasting system based on multi-source data deep learning, characterized in that: It comprises a processor and a memory communicating with each other, the processor is used to call a computer program from the memory, and implement the method according to any one of claims 1 to 7 by running the computer program.

Images