Regional power grid load prediction method and system based on deep learning

Abstract

The invention provides a regional power grid load prediction method and system based on deep learning. The regional power grid load prediction method comprises the following steps: integrating real-time power consumption data and a historical power consumption database to construct a dynamic data set; using a time sequence analysis model to extract current period power consumption characteristics and performing similarity judgment on the current period power consumption characteristics and historical same period data; if not, acquiring first position information, and judging whether the current region meets a first electricity charge condition or not; matching historical similar weather modes in combination with weather forecast, and extracting power utilization characteristic data and power generation characteristic data in a second period and an associated period in the future; judging whether the current region meets the load requirement in the second period in the future; if not, electric quantity allocation of the next region is needed; if yes, daily monitoring continues to be carried out, and electricity borrowing early warning is sent to an adjacent area in advance; if yes, the load condition of the current region is stable. According to the invention, the intelligent degree of power grid load prediction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid load forecasting, and in particular to a method and system for regional power grid load forecasting based on deep learning. Background Art

[0002] With the development of science and technology, the daily living standards of residents are also improving. As living standards improve, most scenarios require electricity, and various factories also require large amounts of electricity. Therefore, this puts a test on the load capacity of regional power grids. Regional power grid load forecasting refers to predicting the power grid load in a certain area over a period of time in the future by collecting and analyzing various data. This forecast is crucial for the planning and operation of power systems, helping power companies better manage power supply and demand and ensure the stable operation of the power grid. However, the current power grid load forecasting methods are not intelligent enough. Summary of the Invention

[0003] Therefore, the embodiments of the present invention provide a method and system for regional power grid load forecasting based on deep learning, which improves the intelligence level of power grid load forecasting.

[0004] To solve the above problems, the present invention provides a regional power grid load forecasting method based on deep learning, including: integrating real-time electricity consumption data and historical electricity consumption database to construct a dynamic data set; according to the dynamic data set, using a time series analysis model to extract the electricity consumption characteristics of the current period, and make a similarity judgment with the historical data of the same period; if it is judged to be dissimilar, obtaining the geographical location of the current area to obtain first location information, and judging whether the current area meets the first electricity fee condition according to the first location information to obtain a first judgment result; combining meteorological forecasts to match historical similar weather patterns, extracting electricity consumption characteristic data and power generation characteristic data in the second period in the future and the associated period; judging whether the current area meets the load requirements in the second period in the future according to the electricity consumption characteristic data, power generation characteristic data and the first judgment result; if it is judged to be not satisfied, it is necessary to allocate electricity from the neighboring area; if it is judged to be satisfied, continuing daily monitoring, and issuing a power borrowing warning to the neighboring area in advance; if it is judged to be similar, it means that the load situation in the current area is stable.

[0005] Compared with the existing technology, the technical effect achieved by adopting this technical solution is as follows: by integrating real-time electricity consumption data with historical electricity consumption databases and constructing dynamic data sets, subsequent electricity query is more convenient. At the same time, the time series analysis model extracts the electricity consumption characteristics of the current period and makes similar judgments with historical data of the same period, which can facilitate subsequent more reasonable judgments on whether the requirements are met. At the same time, the geographical location is used to determine whether the first electricity fee condition is met, and then the weather conditions are obtained according to the weather forecast. By combining the weather conditions and electricity fee conditions, a more comprehensive judgment can be made on the electricity consumption of the region. The electricity fee reflects the overall electricity consumption of the region, while the weather conditions affect the power generation and the power generation capacity and electricity consumption of the region in response to sudden weather. Therefore, based on the comprehensive judgment of the regional electricity consumption, the subsequent regional load judgment can be better. At the same time, the electricity fee, weather conditions and historical electricity consumption are comprehensively judged to determine whether the load requirements are met, thereby making the prediction of the regional load more reasonable and intelligent. The comprehensive multiple factors can improve the accuracy of the prediction results, thereby making a more accurate judgment on the regional electricity load, better meeting the current regional electricity supply guarantee, and providing more stable guarantee for the electricity supply of residents in the region.

[0006] In one example of the present invention, based on a dynamic data set, a time series analysis model is used to extract the electricity consumption characteristics of the current cycle, and a similarity judgment is made with the historical data of the same period, which also includes: obtaining the electricity consumption in the first cycle of the current area from the historical electricity consumption database, recorded as the first electricity consumption data; obtaining the electricity consumption in the historical first cycle corresponding to the first cycle of the current area from the historical electricity consumption database, recorded as the historical first electricity consumption data; comparing the first electricity consumption data with the historical first electricity consumption data to determine whether the first electricity consumption data and the historical first electricity consumption data are similar.

[0007] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by comparing the electricity consumption in the current first cycle with the historical first electricity consumption data in the corresponding historical first cycle to see whether they are similar, a clear understanding of the electricity consumption in the first cycle can be obtained, which is convenient for subsequent judgment.

[0008] In one example of the present invention, the first electricity consumption data is compared with the historical first electricity consumption data, and whether the first electricity consumption data and the historical first electricity consumption data are similar is determined, including: obtaining the electricity consumption of multiple historical first periods before the first period, adding and taking the average value, and obtaining the historical average electricity consumption; subtracting the historical average electricity consumption from the electricity consumption of each historical first period to obtain multiple first difference values, and comparing the multiple first differences; subtracting the electricity consumption of the historical first period corresponding to the largest first difference from the historical average electricity consumption to obtain a second difference value; and judging whether the first electricity consumption data and the historical first electricity consumption data are similar based on the second difference value.

[0009] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by setting to obtain the average power consumption of multiple historical first periods, and subtracting it from the power consumption of each period, and taking the power consumption of the historical first period corresponding to the maximum difference to subtract from the average power consumption, the similarity is judged by the difference, so that the judgment of the historical similarity value of the current first power consumption data is more accurate, and the setting of the average power consumption enables better comprehensive feedback on the power consumption in multiple historical periods, thereby making the data more effective and scientific, and can better provide more reasonable data support for subsequent similarity judgments, thereby improving the accuracy of judgments.

[0010] In one example of the present invention, judging whether the first electricity usage data and the historical first electricity usage data are similar based on the second difference also includes: setting the similarity range to be the range of the positive and negative second difference with the similarity threshold as the center point; if the first electricity usage data falls within the similarity range, it is judged to be similar; if the first electricity usage data falls outside the similarity range, it is judged to be dissimilar.

[0011] Compared with the existing technology, the technical effect achieved by adopting this technical solution is as follows: by setting a similarity range with a similarity threshold as the center point and a positive and negative second difference value to judge whether the first electricity consumption data is similar, the similarity judgment of the first electricity consumption data is made more accurate, and by combining the similarity threshold and the second difference value, the similarity range is made more in line with the actual situation, thereby better reflecting the historical electricity consumption situation in the current area, facilitating the subsequent judgment and prediction of the electricity load, and improving the accuracy and intelligence of the prediction.

[0012] In one example of the present invention, whether the current area meets the first electricity fee condition is determined based on the first location information, and obtaining the first judgment result also includes: obtaining the total electricity fee of the current area in each same quarter in history based on the first location information; calculating the electricity fee coefficient based on the total electricity fee of each same quarter in history; and determining whether the first electricity fee condition is met based on the electricity fee coefficient.

[0013] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by obtaining the total electricity charges of the same historical quarter to calculate the electricity charge coefficient, and then judging whether the first electricity charge condition is met based on the electricity charge coefficient. The total electricity charge can intuitively reflect the electricity consumption situation in the current area, and thus can serve as data support to support subsequent load forecasting, thereby better improving the accuracy of load forecasting.

[0014] In one example of the present invention, determining whether the first electricity fee condition is met based on the electricity fee coefficient also includes: subtracting the total electricity fee of each same quarter in history pairwise to obtain multiple electricity fee difference values; calculating the average of the multiple electricity fee difference values to obtain the electricity fee coefficient; if the electricity fee coefficient is greater than or equal to a set threshold, it is determined that the first electricity fee condition is met; if the electricity fee coefficient is less than the set threshold, it is determined that the first electricity fee condition is not met.

[0015] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by subtracting the total electricity charges of each same quarter in history step by step and calculating the average, the obtained electricity charge coefficient can be more in line with the actual situation, and can also reflect the growth of electricity consumption in the current area, so as to better predict the subsequent load and improve the accuracy of the prediction.

[0016] In one example of the present invention, combining weather forecasts to match historical similar weather patterns, extracting electricity consumption characteristic data and power generation characteristic data in the future second period and associated periods includes: obtaining the weather forecast for the future second period in the current area to obtain the first weather condition; querying the historical first weather condition similar to the first weather condition in the meteorological system, and obtaining the electricity consumption of the historical second period corresponding to the historical first weather condition, obtaining the historical second electricity consumption data and obtaining the historical first power generation data corresponding to the historical first weather condition; the matching method of the historical first power generation data and the historical second electricity consumption data is the same as that of the historical first weather condition; obtaining the number of power generation weathers existing in the first weather condition, recorded as the first weather number, obtaining the number of severe weathers existing in the first weather condition, recorded as the second weather number; searching for the corresponding historical first weather condition based on the first weather number and the second weather number; wherein, power generation weather is sunny and cloudy, and severe weather is all weather conditions except power generation weather.

[0017] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by obtaining the number of power generation weather and severe weather in the first weather condition to judge and find the corresponding historical first weather condition, this method makes the search for the historical first weather condition faster, and the historical first weather condition found is more consistent with the actual situation. At the same time, it is also convenient to understand the weather conditions within the cycle, so as to consider the electricity consumption and power generation conditions more comprehensively, and thus make the subsequent load forecast more accurate.

[0018] In one example of the present invention, searching for the corresponding historical first weather condition based on the first weather number and the second weather number also includes: if there is no historical first weather condition that matches the first weather number and the second weather number, reducing the first weather number by a set number for matching; if there is a matching historical first weather condition, determining whether there are multiple historical first weather conditions; if there are multiple historical first weather conditions, taking the average of the electricity consumption in the multiple historical first weather conditions as the historical second electricity consumption data, and taking the average of the power generation in the multiple historical first weather conditions as the historical first power generation data.

[0019] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by setting a setting that reduces a set number for matching when there is no matching number of historical first weather conditions, thereby making the matching range larger, and then obtaining similar weather conditions as reference data, so as to facilitate the subsequent power load forecast under the weather conditions, thereby making the power forecast more accurate, and at the same time, by setting it to determine whether there are multiple historical first weather conditions when there is a matching situation, and then when there are multiple situations, taking the average as the historical first power generation data and power consumption data, in this way, the obtained historical first power generation data and historical first power consumption data are more in line with the current situation, thereby making subsequent predictions more accurate.

[0020] In one example of the present invention, judging whether the current region meets the load requirements in the second future period based on the electricity consumption characteristic data, the power generation characteristic data and the first judgment result also includes: if the historical second electricity consumption data is greater than or equal to the historical first power generation data, then regardless of whether the first electricity fee condition is met, it is judged that the load requirements are not met; if the historical second electricity consumption data is less than the historical first power generation data and meets the first electricity fee condition, then it is judged that the load requirements are not met; if the historical second electricity consumption data is less than the historical first power generation data and does not meet the first electricity fee condition, then it is judged that the load requirements are met.

[0021] Compared with the existing technology, the technical effect achieved by adopting this technical solution is: by setting up the load requirements for judgment under different circumstances, the judgment method can cover a variety of different situations, and thus the load can be predicted more comprehensively and comprehensively in actual use, so that the obtained prediction results are more accurate and can better reflect the actual situation. When the second historical electricity consumption data is greater than or equal to the first historical power generation data, it means that the current electricity consumption is too large and it is difficult for the region to bear the current electricity consumption, so the judgment is not met. When the first electricity fee condition is met, it means that the current electricity consumption in the region is growing rapidly, so there is a possibility of load exceeding, which requires attention.

[0022] The present invention also provides a regional power grid load forecasting system based on deep learning, which is used to implement the regional power grid load forecasting method as described above. The regional power grid load forecasting system includes: an acquisition module, which is used to obtain regional power consumption, power generation, first location information and first weather conditions; a judgment and calculation module, which is used to judge and calculate whether the current region meets the load requirements; and a control module, which is used to allocate and control power.

[0023] Compared with the existing technology, the technical effect achieved by adopting this technical solution is as follows: by setting up an acquisition module to obtain regional electricity consumption, power generation and other information, the acquisition is faster and convenient for subsequent calculations, and then the load requirements are quickly judged through the judgment and calculation module, which improves the convenience and accuracy of the judgment, thereby making the system more intelligent, and finally by setting up a control module to control the allocation of electricity, making it more convenient to use.

[0024] After adopting the technical solution of the present invention, the following technical effects can be achieved: (1) By integrating real-time electricity consumption data with historical electricity consumption database and constructing a dynamic data set, it is more convenient to query electricity consumption in the future. At the same time, the time series analysis model extracts the electricity consumption characteristics of the current period and makes similar judgments with historical data of the same period, which can facilitate the subsequent more reasonable judgment of whether it meets the requirements. At the same time, the geographical location is used to judge whether the first electricity fee condition is met, and then the weather conditions are obtained according to the weather forecast. By combining the weather conditions and electricity fee conditions, a more comprehensive judgment can be made on the electricity consumption of the region. The electricity fee situation reflects the overall electricity consumption of the region, while the weather conditions affect the power generation situation and the power generation capacity and electricity consumption of the region in response to sudden weather. Therefore, judging the electricity consumption of the region based on the two can better judge the subsequent regional load. At the same time, the electricity fee situation, weather conditions and historical electricity consumption are comprehensively judged to determine whether the load requirements are met, making the prediction of the regional load more reasonable and intelligent. The comprehensive combination of multiple factors can improve the accuracy of the prediction results, so that the regional electricity load can be more accurately judged, which can better meet the current regional electricity security, thereby providing more stable security for the residents' electricity consumption in the region. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings to be used in describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive efforts. Figure 1A flowchart of a regional power grid load forecasting method based on deep learning provided by an embodiment of the present invention; Figure 2 A module diagram of a regional power grid load forecasting system based on deep learning provided in an embodiment of the present invention.

[0026] Description of reference numerals: 100. Regional power grid load forecasting system; 110. Control module; 120. Judgment and calculation module; 130. Control module. DETAILED DESCRIPTION

[0027] To make the above-mentioned objectives, features, and advantages of the present invention more clearly understood, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0028] [First embodiment] See also Figure 1 The present invention provides a regional power grid load forecasting method based on deep learning, comprising: Step S100: Integrate real-time electricity consumption data with the historical electricity consumption database to construct a dynamic data set; Step S200: Based on the dynamic data set, a time series analysis model is used to extract the current cycle power consumption characteristics and perform similarity judgment with historical data of the same period; Step S210: If it is determined that the locations are not similar, the geographical location of the current area is obtained to obtain first location information, and whether the current area meets the first electricity fee condition is determined based on the first location information to obtain a first determination result; Step S300: extracting power consumption characteristic data and power generation characteristic data for the second future period and the associated period by combining the weather forecast with matching historical similar weather patterns; and determining whether the current region meets the load requirements for the second future period based on the power consumption characteristic data, the power generation characteristic data, and the first judgment result. Step S310: If the requirement is not met, power allocation to the neighboring area is required; Step S320: If the conditions are met, continue routine monitoring and issue a power borrowing warning to neighboring areas in advance; Step S220: If the judgment is similar, it means that the load situation in the current area is stable Specifically, in daily use, the power generation and power consumption of each region are uploaded to the power grid system for record preservation, and the regional power consumption is used as the first data, and a historical power consumption database is established based on all historical data in the power grid system.

[0029] Furthermore, when load forecasting is required, all electricity consumption within the first cycle of the current region is first obtained. The first cycle is a manually set value, which can be one week, two weeks, or one month, etc., and can be adjusted according to actual needs. After the acquisition is completed, the electricity consumption corresponding to the historical first cycle is searched in the historical electricity consumption database and recorded as the historical first electricity consumption cycle. The historical first cycle is the date segment corresponding to the first cycle pushed forward. For example, if the first cycle is from the 10th to the 15th of January, the historical first cycle can be from the 10th to the 15th of January of the previous year, and from the 10th to the 15th of January of the previous year, and so on.

[0030] Furthermore, the average of the electricity consumption of multiple historical first periods is taken to obtain the historical average electricity consumption, and the difference is made from the electricity consumption of each historical first period to obtain multiple first difference values, and the multiple first difference values are compared to obtain a maximum value, and the electricity consumption of the historical first period corresponding to the maximum value is then subtracted from the historical average electricity consumption to obtain a second difference value, and then a similarity threshold is set. The similarity threshold is a manually set value and can be changed according to actual conditions. The similarity range is set to the range of the positive and negative second difference values with the similarity threshold as the center point, that is, when the similarity threshold is 1.6 billion kWh and the second difference is 200 million kWh, the similarity range is [14,18], and the first electricity consumption data is judged based on the similarity range. When the judgment falls outside the similarity range, the weather conditions and electricity consumption conditions are obtained based on the first location information.

[0031] Furthermore, the total electricity charges for each quarter in the history of the current region are obtained to calculate the electricity charge coefficient, and the total electricity charges for each same quarter in history are subtracted pairwise to obtain multiple electricity charge differences, and the multiple electricity charge differences are averaged to obtain the electricity charge coefficient, which is compared with the set threshold value, where the set threshold value is a manually set value and can be adjusted according to actual conditions to obtain the result of whether the first electricity charge condition is met.

[0032] Furthermore, by combining weather forecasts with matching historically similar weather patterns, electricity consumption and power generation characteristic data for the second future cycle and associated cycles are extracted. That is, the first weather condition for the second future cycle in the current region is obtained, and historically similar weather conditions are searched for based on the number of power generation weather conditions and severe weather conditions. If there is no historical first weather condition that matches the first and second weather conditions, the first weather condition is reduced by a set number for matching. The set number is a manually set value that can be adjusted based on actual conditions, and can specifically be set to 1. If there is a matching historical first weather condition, the average of the electricity consumption in multiple historical first weather conditions is taken as the historical second electricity consumption data, and the average of the power generation in multiple historical first weather conditions is taken as the historical first power generation data.

[0033] Furthermore, a deep neural network is used to fuse multidimensional data for load forecasting, where the multidimensional data includes electricity consumption characteristic data, power generation characteristic data and the first judgment result, that is, the historical second electricity consumption data, the historical first power generation data and the first judgment result are used to judge whether the load in the region in the second period in the future meets the requirements.

[0034] Preferably, this method is applied to a deep learning model for training, and historical data is substituted into the model for measurement, thereby making the model prediction results more accurate, so that the subsequent model can be put into use after it is successfully established.

[0035] Preferably, a historical electricity consumption database is established by acquiring daily electricity consumption and historical electricity consumption, so that subsequent electricity consumption inquiries are more convenient. At the same time, by comparing the electricity consumption in the existing first period with the historical first electricity consumption data in the corresponding historical first period to see whether they are similar, a clear understanding of the electricity consumption in the first period is obtained, which can facilitate subsequent more reasonable judgment on whether the requirements are met. At the same time, whether the first electricity fee condition is met is judged by the geographical location, and the weather conditions are obtained according to the weather forecast. By combining the weather conditions and the electricity fee conditions, a more comprehensive judgment can be made on the electricity consumption in the region, and the electricity fee conditions It reflects the overall electricity consumption of the region, while weather conditions affect the power generation situation and the region's power generation capacity and electricity consumption in response to sudden weather events. Therefore, comprehensively judging the regional electricity consumption based on the two can better judge the subsequent regional load. At the same time, the electricity price, weather conditions and historical electricity consumption can be comprehensively judged to see whether the load requirements are met, making the prediction of the regional load more reasonable and intelligent. The combination of multiple factors can improve the accuracy of the prediction results, thereby making it possible to have a more accurate judgment on the regional electricity load, better meet the current regional electricity security, and provide more stable protection for the electricity consumption of residents in the region.

[0036] Specifically, based on the dynamic data set, a time series analysis model is used to extract the electricity consumption characteristics of the current cycle, and a similarity judgment is made with the historical data of the same period. It also includes: obtaining the electricity consumption in the first cycle of the current area from the historical electricity consumption database, recorded as the first electricity consumption data; obtaining the electricity consumption in the historical first cycle corresponding to the first cycle of the current area from the historical electricity consumption database, recorded as the historical first electricity consumption data; comparing the first electricity consumption data with the historical first electricity consumption data to determine whether the first electricity consumption data and the historical first electricity consumption data are similar.

[0037] Preferably, by comparing the electricity consumption in the current first cycle with the historical first electricity consumption data in the corresponding historical first cycle to determine whether they are similar, a clear understanding of the electricity consumption in the first cycle is obtained, which facilitates subsequent judgment.

[0038] Specifically, comparing the first electricity consumption data with the historical first electricity consumption data, and determining whether the first electricity consumption data and the historical first electricity consumption data are similar includes: obtaining the electricity consumption of multiple historical first periods before the first period, adding and taking the average value, and obtaining the historical average electricity consumption; subtracting the historical average electricity consumption from the electricity consumption of each historical first period to obtain multiple first difference values, and comparing the multiple first differences; subtracting the electricity consumption of the historical first period corresponding to the largest first difference from the historical average electricity consumption to obtain a second difference value; and determining whether the first electricity consumption data and the historical first electricity consumption data are similar based on the second difference value.

[0039] Preferably, the average power consumption of multiple historical first periods is obtained and subtracted from the power consumption of each period, and the power consumption of the historical first period corresponding to the maximum difference is taken to be subtracted from the average power consumption. The degree of similarity is judged by the difference, so that the judgment of the historical similarity value of the current first power consumption data is more accurate, and the setting of the average power consumption enables better comprehensive feedback on the power consumption in multiple historical periods, thereby making the data more effective and scientific, and can better provide more reasonable data support for subsequent similarity judgments, so as to improve the accuracy of judgment.

[0040] Specifically, judging whether the first electricity usage data and the historical first electricity usage data are similar based on the second difference also includes: setting the similarity range to be the range of the positive and negative second difference with the similarity threshold as the center point; if the first electricity usage data falls within the similarity range, it is judged to be similar; if the first electricity usage data falls outside the similarity range, it is judged to be dissimilar.

[0041] Preferably, by setting a similarity range with a similarity threshold as the center point and a positive and negative second difference value to judge whether the first electricity consumption data is similar, the similarity judgment of the first electricity consumption data is made more accurate, and by combining the similarity threshold and the second difference value, the similarity range is made more in line with the actual situation, so that it can better reflect the historical electricity consumption situation in the current area, facilitate subsequent judgment and prediction of electricity load, and improve the accuracy and intelligence of the prediction.

[0042] Specifically, judging whether the current region meets the first electricity fee condition based on the first location information, obtaining the first judgment result also includes: obtaining the total electricity fee of the current region in each same quarter in history based on the first location information; calculating the electricity fee coefficient based on the total electricity fee of each same quarter in history; judging whether the first electricity fee condition is met based on the electricity fee coefficient.

[0043] Preferably, the electricity cost coefficient is calculated by obtaining the total electricity bill of the same historical quarter, and then judging whether the first electricity cost condition is met based on the electricity cost coefficient. The total electricity bill can intuitively reflect the electricity consumption situation in the current area, and thus can serve as data support to support subsequent load forecasting, thereby better improving the accuracy of load forecasting.

[0044] Specifically, judging whether the first electricity fee condition is met based on the electricity fee coefficient also includes: subtracting the total electricity fee of each same quarter in history pairwise to obtain multiple electricity fee difference values; calculating the average of the multiple electricity fee difference values to obtain the electricity fee coefficient; if the electricity fee coefficient is greater than or equal to the set threshold, it is judged that the first electricity fee condition is met; if the electricity fee coefficient is less than the set threshold, it is judged that the first electricity fee condition is not met.

[0045] Preferably, by taking the difference between the total electricity charges of each same quarter in history and calculating the average, the obtained electricity cost coefficient can be more in line with the actual situation and can also reflect the growth of electricity consumption in the current area, so as to better predict the subsequent load and improve the accuracy of the prediction.

[0046] Specifically, querying the meteorological system for the historical first weather condition that is similar to the first weather condition, and obtaining the power consumption of the historical second period corresponding to the historical first weather condition, obtaining the historical second power consumption data, and obtaining the historical first power generation data corresponding to the historical first weather condition also includes: the matching method of the historical first power generation data and the historical second power consumption data is the same as that of the historical first weather condition; obtaining the number of power generation weathers existing in the first weather condition, recorded as the first weather number, obtaining the number of severe weathers existing in the first weather condition, recorded as the second weather number; searching for the corresponding historical first weather condition according to the first weather number and the second weather number; wherein, power generation weather is sunny and cloudy, and severe weather is all weather conditions except power generation weather.

[0047] Preferably, the corresponding historical first weather condition is determined by obtaining the number of power-generating weather and severe weather in the first weather condition. This method makes the search for the historical first weather condition faster, and the historical first weather condition found is more consistent with the actual situation. At the same time, it is also convenient to understand the weather conditions within the period, so as to consider the electricity consumption and power generation conditions more comprehensively, and thus make the subsequent load forecast more accurate.

[0048] Specifically, searching for the corresponding historical first weather condition based on the first weather number and the second weather number also includes: if there is no historical first weather condition that matches the first weather number and the second weather number, reducing the first weather number by a set number for matching; if there is a matching historical first weather condition, determining whether there are multiple historical first weather conditions; if there are multiple historical first weather conditions, taking the average of the electricity consumption in the multiple historical first weather conditions as the historical second electricity consumption data, and taking the average of the power generation in the multiple historical first weather conditions as the historical first power generation data.

[0049] Preferably, by setting it to reduce a set number for matching when there is no matching number of historical first weather conditions, the matching range is made larger, and similar weather conditions can be obtained as reference data, which is convenient for subsequent power load forecasting under the weather conditions, thereby making the power forecast more accurate. At the same time, it is also set to determine whether there are multiple historical first weather conditions when there is a matching situation, and then when there are multiple situations, take the average as the historical first power generation data and power consumption data. In this way, the obtained historical first power generation data and historical first power consumption data are more in line with the current situation, thereby making subsequent predictions more accurate.

[0050] Specifically, judging whether the current region meets the load requirements in the future second period based on the historical second electricity consumption data, the historical first power generation data and the first judgment result also includes: if the historical second electricity consumption data is greater than or equal to the historical first power generation data, then regardless of whether the first electricity fee condition is met, it is judged that the load requirements are not met; if the historical second electricity consumption data is less than the historical first power generation data and meets the first electricity fee condition, then it is judged that the load requirements are not met; if the historical second electricity consumption data is less than the historical first power generation data and does not meet the first electricity fee condition, then it is judged that the load requirements are met.

[0051] Preferably, by setting up load requirements for judgment under different circumstances, the judgment method can cover a variety of different situations, and thus the load can be predicted more comprehensively and integratedly in actual use, so that the obtained prediction results are more accurate and can better reflect the actual situation. When the second historical electricity consumption data is greater than or equal to the first historical power generation data, it means that the current electricity consumption is too large and it is difficult for the region to bear the current electricity consumption, so the judgment is not met. When the first electricity fee condition is met, it means that the current electricity consumption in the region is growing rapidly, so there is a possibility of load exceeding, which requires attention.

[0052] The present invention also provides a regional power grid load forecasting system 100 based on deep learning. The regional power grid load forecasting system 100 is used to implement any of the regional power grid load forecasting methods mentioned above. The regional power grid load forecasting system 100 includes: an acquisition module 110, the acquisition module 110 is used to obtain regional power consumption, power generation, first location information and first weather conditions; a judgment and calculation module 120, the judgment and calculation module 120 is used to judge and calculate whether the current region meets the load requirements; and a control module 130, the control module 130 is used to allocate and control power.

[0053] Preferably, an acquisition module 110 is set up to obtain information such as regional electricity consumption and power generation, thereby making the acquisition faster and facilitating subsequent calculations. The load requirements are then quickly judged through the judgment and calculation module 120, thereby improving the convenience and accuracy of the judgment, thereby making the system more intelligent. Finally, a control module 130 is set up to control the allocation of electricity, making it more convenient to use.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A regional power grid load forecasting method based on deep learning, characterized in that: include: Integrate real-time electricity consumption data with historical electricity consumption database to build a dynamic data set; Based on the dynamic data set, a time series analysis model is used to extract the current cycle power consumption characteristics and perform similarity judgment with historical data of the same period; If it is determined that they are not similar, obtaining the geographical location of the current area to obtain first location information, and determining whether the current area meets the first electricity fee condition based on the first location information to obtain a first determination result; Combined with weather forecasts to match historical similar weather patterns, extract electricity consumption and power generation characteristic data for the second cycle in the future and related cycles; determining whether the current region meets the load requirement in a future second period according to the power consumption characteristic data, the power generation characteristic data, and the first judgment result; If the judgment is not satisfied, the power allocation in the neighboring area is required; If the judgment is satisfied, daily monitoring will continue and an early warning of power borrowing will be issued to the neighboring area; If the judgment is similar, it means that the load situation in the current area is stable.

2. The regional power grid load forecasting method based on deep learning according to claim 1 is characterized in that: The extracting of the current cycle electricity consumption characteristics by using a time series analysis model based on the dynamic data set and performing similarity judgment with historical data for the same period also includes: Obtaining the power consumption in the first period of the current region from the historical power consumption database, and recording it as first power consumption data; Obtaining the electricity consumption in the first historical period corresponding to the first period in the current region from the historical electricity consumption database, and recording the data as the first historical electricity consumption data; The first electricity usage data is compared with the historical first electricity usage data to determine whether the first electricity usage data and the historical first electricity usage data are similar.

3. The regional power grid load forecasting method based on deep learning according to claim 2, characterized in that: The comparing the first electricity usage data with the first historical electricity usage data to determine whether the first electricity usage data and the first historical electricity usage data are similar includes: Obtaining the power consumption of multiple historical first cycles before the first cycle, adding and averaging the power consumption to obtain a historical average power consumption; Subtracting the historical average power consumption from the power consumption of each historical first cycle to obtain a plurality of first difference values, and comparing the plurality of first differences; Subtract the power consumption of the historical first period corresponding to the largest first difference from the historical average power consumption to obtain a second difference; It is determined whether the first electricity usage data and the historical first electricity usage data are similar based on the second difference.

4. The regional power grid load forecasting method based on deep learning according to claim 3 is characterized in that: The determining, based on the second difference, whether the first electricity usage data and the historical first electricity usage data are similar further includes: Setting the similarity range to be a range of the second difference value plus or minus the similarity threshold value as the center point; If the first electricity usage data falls within the similarity range, then it is determined to be similar; If the first electricity usage data falls outside the similarity range, it is determined to be dissimilar.

5. The regional power grid load forecasting method based on deep learning according to claim 1, characterized in that: The determining, based on the first location information, whether the current area meets the first electricity fee condition, and obtaining a first determination result further includes: Acquire the total electricity charges of the current region in each same quarter in history according to the first location information; Calculate the electricity cost coefficient based on the total electricity cost for each identical quarter in the history; It is determined whether a first electricity rate condition is met according to the electricity rate coefficient.

6. The regional power grid load forecasting method based on deep learning according to claim 5 is characterized in that: The determining whether the first electricity fee condition is met according to the electricity fee coefficient further includes: The total electricity charges for each same quarter in the history are subtracted from each other step by step to obtain multiple electricity charge difference values; Calculate the average of the multiple electricity fee differences to obtain the electricity fee coefficient; If the electricity cost coefficient is greater than or equal to the set threshold, it is determined that the first electricity cost condition is met; If the electricity rate coefficient is less than the set threshold, it is determined that the first electricity rate condition is not met.

7. The method for regional power grid load forecasting based on deep learning according to claim 6, characterized in that: The method of combining weather forecasts with matching historical similar weather patterns to extract electricity consumption characteristic data and power generation characteristic data for the second period and associated periods in the future includes: Obtain a weather forecast for the second period in the future for the current region to obtain a first weather condition; query a meteorological system for a historical first weather condition similar to the first weather condition, and obtain power consumption for a historical second period corresponding to the historical first weather condition to obtain historical second power consumption data; and obtain historical first power generation data corresponding to the historical first weather condition: The matching method of the first historical weather condition of the first historical power generation data is the same as the matching method of the first historical weather condition of the second historical power consumption data; Obtain the number of power-generating weather days in the first weather condition, which is recorded as the first weather number; obtain the number of severe weather days in the first weather condition, which is recorded as the second weather number; searching for the corresponding first historical weather condition according to the first weather number and the second weather number; The power generation weather refers to sunny and cloudy days, and the severe weather refers to all weather conditions except the power generation weather.

8. The method for regional power grid load forecasting based on deep learning according to claim 7, characterized in that: Searching for the corresponding first historical weather condition according to the first weather number and the second weather number further includes: If there is no historical first weather condition that matches the first weather number and the second weather number, reducing the first weather number by a set number for matching; If there is a matching first historical weather condition, determining whether there are multiple first historical weather conditions; If there are multiple historical first weather conditions, the average of the electricity consumption in the multiple historical first weather conditions is taken as the historical second electricity consumption data, and the average of the power generation in the multiple historical first weather conditions is taken as the historical first power generation data.

9. The method for regional power grid load forecasting based on deep learning according to claim 8, characterized in that: The determining whether the current region meets the load requirement in the second period in the future according to the power consumption characteristic data, the power generation characteristic data and the first determination result further includes: If the second historical electricity consumption data is greater than or equal to the first historical power generation data, then regardless of whether the first electricity fee condition is met, it is determined that the load requirement is not met; If the second historical electricity consumption data is less than the first historical power generation data and satisfies the first electricity fee condition, it is determined that the load requirement is not met; If the second historical electricity consumption data is less than the first historical power generation data and does not meet the first electricity fee condition, it is determined that the load requirement is met.

10. A regional power grid load forecasting system based on deep learning, wherein the regional power grid load forecasting system is used to implement the regional power grid load forecasting method according to any one of claims 1 to 9, characterized in that: The regional power grid load forecasting system includes: an acquisition module, the acquisition module being configured to acquire the regional power consumption, power generation, the first location information, and the first weather condition; A judgment and calculation module, the judgment and calculation module is used to judge and calculate whether the current area meets the load requirement; A control module is used to allocate and control power.

Images