Non-working day electricity demand forecasting method and system considering the impact of urban greening

Abstract

The present invention provides a method and system for predicting power demand on non-working days considering the impact of urban greening. The method is to obtain ground remote sensing image time series data, non-working day residential power consumption time series data and meteorological time series data of each area in the city, and to perform green space impact analysis on greening distribution time series data obtained by analyzing ground remote sensing image time series data to obtain green space impact time series data. After combining non-working day residential power consumption time series data, meteorological time series data and power consumption behavior prediction model to obtain power consumption behavior distribution time series data, the non-working day residential power consumption time series data and meteorological time series data are combined to construct a non-working day power demand prediction model for regional time period power demand prediction, and the obtained prediction results are summarized to obtain the city time period power demand prediction result. The present invention can effectively improve the adaptability, comprehensiveness and accuracy of non-working day power demand prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of electricity demand prediction, and in particular to a method and system for predicting electricity demand on non-working days taking into account the impact of urban greening. Background Art

[0002] As people's quality of life continues to improve, the concept of life on non-working days, including statutory holidays and weekends, has also changed significantly, which has also brought new challenges to the prediction of electricity demand on non-working days. Based on this, how to accurately predict electricity demand on non-working days has gradually become the decision-making basis for power supply dispatching of power grid systems.

[0003] The existing non-working day electricity demand forecast mainly improves the accuracy of non-working day electricity demand forecast by introducing weather factors during non-working days when constructing the forecast model. It does not consider that the level of urban greening helps to reduce temperature, increase humidity and improve air quality. Its regulatory effect on microclimate will affect residents' electricity consumption behavior and electricity demand. It also does not consider the impact of the unique electricity consumption pattern on non-working days on electricity demand forecast. However, in some cities with serious dust pollution, residents usually choose to use air purifiers to improve indoor air quality. If the city increases the greening area, especially planting more trees in areas with serious dust pollution, it can effectively reduce the dust content in the air and improve air quality. The demand for residents to use air purifiers will be reduced accordingly, and the frequency of residents' outdoor activities may also increase, thereby directly reducing household electricity demand; furthermore, the shift of peak electricity time on non-working days, the increase of family entertainment and housework activities, etc. will inevitably have a direct impact on electricity demand. That is, the existing non-working day electricity demand forecast only considers historical electricity demand data and weather factors, which obviously cannot guarantee the comprehensiveness and accuracy of non-working day electricity demand forecast, and it is difficult to provide a reliable analysis basis for power supply dispatching of the power grid system. Summary of the invention

[0004] The purpose of the present invention is to provide a method for predicting non-working day electricity demand taking into account the influence of urban greening. By deeply analyzing the microclimate effect of urban greening and combining the unique electricity consumption patterns of non-working days, a non-working day electricity demand prediction model with comprehensive consideration of electricity demand factors, strong adaptability and high accuracy is constructed. Based on the introduction of urban greening and analysis of the impact of special electricity consumption behaviors of residents on non-working days on electricity demand, the non-working day electricity demand is predicted in different time periods, which can effectively improve the adaptability, comprehensiveness and accuracy of non-working day electricity demand prediction, and thus provide a reliable analysis basis for power supply scheduling of the power grid system.

[0005] In order to achieve the above objectives, it is necessary to provide a method and system for predicting non-working day electricity demand taking into account the impact of urban greening in response to the above technical issues.

[0006] In a first aspect, an embodiment of the present invention provides a method for predicting power demand on non-working days considering the impact of urban greening, the method comprising the following steps:

[0007] Obtain the ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data for each preset functional area in the target city;

[0008] Perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data; the greening distribution time series data includes green space layout time series characteristics and transpiration rate distribution time series characteristics;

[0009] Performing green space impact analysis based on the greening distribution time series data to obtain corresponding green space impact time series data, and performing power consumption behavior analysis based on the greening distribution time series data, the non-working day residential power consumption time series data, the meteorological time series data and a pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data;

[0010] Based on the time series data of electricity consumption behavior distribution, greening distribution, green space impact, non-working day residential electricity consumption and meteorological time series data of all preset functional areas, a non-working day electricity demand forecasting model is constructed;

[0011] According to the non-working day electricity demand forecasting model, the electricity demand forecast for the target period is performed based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of residents on non-working days and the real-time meteorological data, and the corresponding regional period electricity demand forecast results are obtained;

[0012] The electricity demand forecast results for all regions and time periods are summarized to obtain the electricity demand forecast results for the target city time period.

[0013] Furthermore, the ground remote sensing image time series data is multispectral remote sensing image data of each historical period; the meteorological time series data includes temperature, humidity and wind speed of each historical period; the green space layout time series characteristics include green space layout characteristics of each historical period; the transpiration rate distribution time series characteristics include transpiration rate distribution characteristics of each historical period;

[0014] The step of performing greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data comprises:

[0015] Performing atmospheric correction and geometric calibration on the ground remote sensing image time series data to obtain corrected remote sensing image time series data; the corrected remote sensing image time series data includes corrected remote sensing images of various historical periods;

[0016] According to the normalized vegetation index, the vegetation coverage of all pixels in each corrected remote sensing image is calculated, and according to the comparison result between the vegetation coverage and the preset vegetation index threshold, each pixel is divided into a green space pixel and a non-green space pixel;

[0017] Compare the vegetation coverage of all green space pixels in each corrected remote sensing image with a preset green space type threshold to obtain the green space type of each green space pixel; the green space type includes grass, shrubs and trees;

[0018] Statistical analysis is performed on all green space pixels in each corrected remote sensing image to obtain the green space layout characteristics of the corresponding historical period, and the green space layout time series characteristics are obtained based on the green space layout characteristics of all historical periods; the green space layout characteristics include the proportion of the total green space area, the composition of green space types, and the area proportion and spatial distribution of different green space types;

[0019] According to the meteorological data of all historical periods and the layout characteristics of green spaces, the transpiration rates of all green space pixels in each corrected remote sensing image were obtained based on the Penman-Montes formula.

[0020] The transpiration rates of all green space pixels in each corrected remote sensing image are statistically analyzed to obtain the transpiration rate distribution characteristics of the corresponding historical period, and the transpiration rate distribution time series characteristics are obtained based on the transpiration rate distribution characteristics of all historical periods; the transpiration rate distribution characteristics include the average transpiration rate and the corresponding standard deviation.

[0021] Furthermore, the green space impact time series data includes the green space impact time series data on microclimate, the green space impact time series data on body temperature, the green space impact time series data on thermal comfort and the green space impact time series data on air purification;

[0022] The step of performing green space impact analysis according to the greening distribution time series data to obtain corresponding green space impact time series data comprises:

[0023] Input the greening distribution time series data into a pre-constructed green space microclimate impact model to perform green space microclimate impact analysis, and obtain the green space microclimate impact time series data; the green space microclimate impact time series data includes the green space temperature drop amplitude and green space humidification amplitude in each historical period;

[0024] Obtaining the time series data of perceived temperature distribution, the time series data of thermal comfort index distribution, and the time series data of air quality distribution corresponding to the greening distribution time series data;

[0025] Obtaining a body temperature distribution map corresponding to each historical period in the body temperature distribution time series data, and performing a body temperature difference analysis inside and outside the green space based on the body temperature distribution map and the corresponding green space layout map, to obtain time series data on the impact of the green space on the body temperature;

[0026] Obtain a thermal comfort distribution map corresponding to each historical period in the thermal comfort index distribution time series data, and perform a thermal comfort difference analysis inside and outside the green space based on the thermal comfort distribution map and the corresponding green space layout map to obtain the time series data of the impact of the green space on thermal comfort;

[0027] Through the inverse distance weighted interpolation method, the air quality distribution map corresponding to each historical period in the air quality distribution time series data is obtained, and the difference in pollutant concentrations inside and outside the green space is analyzed based on the air quality distribution map and the corresponding green space layout map to obtain the time series data of the impact of the green space on air purification.

[0028] Further, the thermal comfort index distribution time series data includes predicted average vote count time series data and physiological equivalent temperature time series data;

[0029] The step of obtaining the time series data of body temperature distribution, the time series data of thermal comfort index distribution and the time series data of air quality distribution corresponding to the time series data of greening distribution comprises:

[0030] According to the microclimate distribution time series data corresponding to the greening distribution time series data, the body temperature distribution time series data is obtained by a heat index method;

[0031] Acquire the surface temperature distribution time series data corresponding to the greening distribution time series data, and obtain the corresponding average radiation temperature distribution time series data and wind speed distribution time series data according to the surface temperature distribution time series data;

[0032] Input the time series data of the green space's impact on microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution into a pre-built predicted average vote number calculation model for analysis to obtain corresponding predicted average vote number time series data;

[0033] Input the time series data of the green space's impact on microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution into a pre-built physiological equivalent temperature calculation model for analysis to obtain corresponding physiological equivalent temperature time series data;

[0034] The air pollutant concentration time series data corresponding to the greening distribution time series data is obtained, and the air pollutant concentration time series data is used as the air quality distribution time series data; the air pollutant concentration time series data includes fine particulate matter concentration time series data, nitrogen dioxide concentration time series data and ozone concentration time series data.

[0035] Furthermore, the steps of constructing the green space impact model on microclimate include:

[0036] Acquire the greening distribution time series data to be analyzed, and perform Kriging interpolation processing on the green space layout map and transpiration rate distribution map corresponding to each time period in the greening distribution time series data to be analyzed, and generate the corresponding microclimate distribution time series data to be analyzed; the microclimate distribution time series data to be analyzed includes continuous temperature distribution data and relative humidity distribution data of each time period;

[0037] According to the buffer zone analysis method, the microclimate distribution time series data to be analyzed is processed by expanding the green space boundary to generate pre-processed microclimate distribution time series data with transition zones inside and outside the green space;

[0038] According to the pre-processed microclimate distribution time series data, the corresponding microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between inside and outside the green space are obtained, and the microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between inside and outside the green space are comprehensively compared and analyzed to obtain the corresponding green space microclimate impact data to be analyzed; the green space microclimate impact data to be analyzed includes the amount of temperature reduction affected by the green space and the amount of relative humidity increase affected by the green space;

[0039] According to the greening distribution time series data to be analyzed and the green space microclimate impact data to be analyzed, a model of the green space's impact on microclimate is constructed based on the random forest regression method.

[0040] Furthermore, the steps of constructing the electricity consumption behavior prediction model include:

[0041] Obtain the time series data of residential electricity consumption on non-working days to be analyzed in each preset functional area, as well as the corresponding time series data of meteorological data to be analyzed and time series data of greening distribution to be analyzed;

[0042] Based on the preset electricity consumption behavior characteristics, cluster analysis is performed on the non-working day residential electricity consumption time series data to be analyzed in each preset functional area to obtain the electricity consumption time period behavior distribution time series data; the electricity consumption time period behavior distribution time series data includes the electricity consumption behavior type of each time period; the electricity consumption behavior type includes leisure and entertainment and housework activities;

[0043] Analyze the meteorological time series data to be analyzed and the greening distribution time series data to be analyzed for each preset functional area respectively, and obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data;

[0044] The wind speed change distribution time series data and the shade area change distribution time series data of all preset functional areas are used as independent variable data, and the power consumption time period behavior distribution time series data of all preset functional areas are used as dependent variable data, and the power consumption behavior prediction model is constructed based on the generalized additive model.

[0045] Furthermore, the step of respectively analyzing the meteorological time series data to be analyzed and the greening distribution time series data to be analyzed of each preset functional area to obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data includes:

[0046] Performing fluid dynamics simulation according to the meteorological time series data to be analyzed to obtain corresponding wind environment distribution time series data, and performing differential analysis on the wind environment distribution time series data to obtain corresponding wind speed change distribution time series data;

[0047] According to the green area corresponding to the greening distribution time series data to be analyzed, the corresponding green shade effect distribution time series data is generated, and the green shade effect distribution time series data is differentially analyzed to obtain the corresponding shade area change distribution time series data.

[0048] Furthermore, the step of constructing a non-working day electricity demand prediction model according to the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data of all preset functional areas includes:

[0049] Construct a power forecast training set based on the power consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential power consumption time series data and meteorological time series data of all preset functional areas;

[0050] Normalizing the power prediction training set to obtain a corresponding normalized training set;

[0051] According to the normalized training set, the preset long short-term memory network is iteratively trained to obtain the non-working day electricity demand prediction model; the preset long short-term memory network includes two LSTM layers and one fully connected layer.

[0052] Furthermore, the method further comprises:

[0053] The degree of change of greening distribution data of all preset functional areas is monitored regularly, and when the degree of change of greening distribution data of any preset functional area exceeds a preset change degree threshold, the non-working day electricity demand prediction model is adaptively optimized and updated.

[0054] In a second aspect, an embodiment of the present invention provides a non-working day electricity demand forecasting system considering the impact of urban greening, the system comprising:

[0055] The data acquisition module is used to obtain the ground remote sensing image time series data of each preset functional area in the target city, the residential electricity consumption time series data on non-working days, and the meteorological time series data;

[0056] A greening distribution analysis module is used to perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data; the greening distribution time series data includes green space layout time series characteristics and transpiration rate distribution time series characteristics;

[0057] A greening impact analysis module, configured to perform green space impact analysis based on the greening distribution time series data to obtain corresponding green space impact time series data, and perform power consumption behavior analysis based on the greening distribution time series data, the non-working day residential power consumption time series data, the meteorological time series data and a pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data;

[0058] A prediction model building module is used to build a non-working day electricity demand prediction model based on the electricity consumption behavior distribution time series data of all preset functional areas, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data;

[0059] A regional demand forecasting module is used to forecast the electricity demand in a target period according to the non-working day electricity demand forecasting model, based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of residents on non-working days and real-time meteorological data, and obtain the corresponding regional period electricity demand forecasting results;

[0060] The city demand forecast module is used to summarize the electricity demand forecast results of all regions and time periods to obtain the electricity demand forecast results of the target city time period.

[0061] The above-mentioned application provides a method and system for predicting electricity demand on non-working days considering the impact of urban greening. The method can achieve the acquisition of ground remote sensing image time series data of each preset functional area in the target city, non-working day residential electricity consumption time series data and meteorological time series data, and perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area respectively, and obtain the corresponding greening distribution time series data including green space layout time series characteristics and transpiration rate distribution time series characteristics. After that, green space impact analysis is performed according to the greening distribution time series data to obtain the corresponding green space impact time series data, and according to the greening distribution time series data, non-working day residential electricity consumption time series data, meteorological time series data and pre-constructed electricity consumption behavior prediction system. A technical solution is provided for analyzing electricity consumption behavior for the prediction model to obtain corresponding electricity consumption behavior distribution time series data, and building a non-working day electricity demand prediction model based on the electricity consumption behavior distribution time series data of all preset functional areas, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data; and based on the non-working day electricity demand prediction model, a target period electricity demand prediction is carried out based on the real-time ground remote sensing images of each preset functional area, the real-time non-working day residential electricity consumption and real-time meteorological data to obtain the corresponding regional period electricity demand prediction results, and all regional period electricity demand prediction results are summarized to obtain the target city period electricity demand prediction results. Compared with the existing technology, the non-working day electricity demand forecasting method considering the impact of urban greening deeply analyzes the microclimate effect of urban greening, and combines the unique electricity consumption patterns of non-working days to construct a non-working day electricity demand forecasting model that takes into account comprehensive electricity demand factors, has strong adaptability and efficient prediction. Based on the introduction of urban greening and the analysis of the impact of special electricity consumption behaviors of residents on non-working days on electricity demand, the non-working day electricity demand is predicted in different time periods, which can effectively improve the adaptability, comprehensiveness and accuracy of non-working day electricity demand forecasting, and thus provide a reliable analysis basis for power supply scheduling of the power grid system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is a flow chart of a method for predicting power demand on non-working days considering the impact of urban greening in an embodiment of the present invention;

[0063] Figure 2 is another flow chart of a method for predicting power demand on non-working days considering the impact of urban greening in an embodiment of the present invention;

[0064] Figure 3 is a structural schematic diagram of a non-working day electricity demand forecasting system considering the impact of urban greening in an embodiment of the present invention;

[0065] Figure 4 2 is another structural schematic diagram of a non-working day electricity demand forecasting system taking into account the impact of urban greening in an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and beneficial effects of the present application clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Obviously, the embodiments described below are part of the embodiments of the present invention and are only used to illustrate the present invention, but are not used to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] The non-working day electricity demand forecasting method considering the influence of urban greening provided by the present invention can be understood as a technical solution for forecasting non-working day electricity demand by time period based on the fact that the existing non-working day electricity demand forecasting does not consider the regulatory effect of urban greening level on microclimate and the influence of the unique electricity consumption pattern on non-working days on electricity demand forecasting, resulting in the electricity demand forecasting being not comprehensive and accurate. Instead, a method is proposed to deeply analyze the microclimate effect of urban greening, and combine the unique electricity consumption law of non-working days to build a non-working day electricity demand forecasting model that considers electricity demand factors comprehensively, has strong adaptability and high accuracy, and introduces the analysis of the impact of urban greening and special electricity consumption behavior of residents on non-working days on electricity demand. The following embodiment will explain in detail the non-working day electricity demand forecasting method considering the influence of urban greening of the present invention.

[0068] In one embodiment, Figure 1 As shown, a method for predicting non-working day electricity demand considering the impact of urban greening is provided, including the following steps:

[0069] S11, obtaining the ground remote sensing image time series data, non-working day residential electricity time series data and meteorological time series data of each preset functional area in the target city; wherein, the preset functional area can be understood as a plurality of management areas obtained by dividing the target city according to a certain division rule, and the division rule can be determined according to the actual power analysis requirements, and is not specifically limited here; the corresponding ground remote sensing image time series data, non-working day residential electricity time series data and meteorological time series data can all be understood as regional ground image data, non-working day residential electricity data and meteorological data of a preset historical duration obtained according to a preset sampling period, and preferably, the ground remote sensing image time series data is multispectral remote sensing image data of each historical period, non-working day residential electricity data can be obtained based on the user's smart meter data, and the meteorological time series data can be obtained through public channels on the Internet, including temperature, humidity and wind speed of each historical period. It should be noted that the sampling period and preset duration of the actual data acquisition can be determined according to the actual application requirements. In order to ensure the rationality of the subsequent use of data analysis while maintaining the efficiency of data analysis, the present embodiment preferably sets the sampling period to 1 hour, that is, sampling is performed at intervals of hours.

[0070] S12. Perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area respectively to obtain corresponding greening distribution time series data; wherein, greening distribution analysis can be understood as a process of analyzing the green area ratio, vegetation coverage density and green space type of each functional area based on remote sensing image analysis technology; the corresponding greening distribution time series data include green space layout time series characteristics and transpiration rate distribution time series characteristics, wherein the green space layout time series characteristics include green space layout characteristics of each historical period; the transpiration rate distribution time series characteristics include transpiration rate distribution characteristics of each historical period.

[0071] Specifically, the step of performing greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data includes:

[0072] The ground remote sensing image time series data is subjected to atmospheric correction and geometric calibration to obtain corrected remote sensing image time series data; wherein, atmospheric correction mainly includes two types, namely, absolute atmospheric correction and relative atmospheric correction, which are used to eliminate the influence of the atmosphere on the remote sensing image so that the image can more truly reflect the surface information. The specific implementation process can be realized by referring to the relevant existing technologies; geometric calibration mainly includes rough geometric correction, fine geometric correction and orthorectification, which are used to eliminate the geometric deformation of the remote sensing image caused by factors such as sensors, earth curvature, and terrain undulations during the imaging process, so that the objects in the image can accurately reflect their actual positions. The specific implementation process can also be realized by referring to the relevant existing technologies. For example, the atmospheric influence can be eliminated by radiometric calibration and atmospheric correction, and the terrain correction can be performed by using the SRTM digital elevation model. Correspondingly, atmospheric correction and geometric calibration are performed on the ground remote sensing image data of each historical period in the ground remote sensing image time series data, so that the corrected remote sensing image time series data including the corrected remote sensing images of each historical period can be obtained.

[0073] According to the normalized vegetation index, the vegetation coverage of all pixels in each corrected remote sensing image is calculated, and according to the comparison result between the vegetation coverage and the preset vegetation index threshold, each pixel is divided into green space pixel and non-green space pixel; wherein, the normalized vegetation index NDVI is expressed as follows:

[0074]

[0075] Among them, NIR is the reflectivity of the near-infrared band, and R is the reflectivity of the red light band.

[0076] When calculating NDVI in this embodiment, the reflectance values ​​of the near-infrared band with a wavelength of 0.78-0.89μm and the red light band with a wavelength of 0.61-0.68μm are preferably selected; the preset vegetation index threshold used can be selected and set according to the actual application requirements, and is not specifically limited here. In actual applications, if the preset vegetation index threshold is set to 0.2, when the normalized vegetation index value of a certain pixel is greater than the preset vegetation index threshold, it is considered that the pixel is a pixel with vegetation coverage, that is, a green space pixel. On the contrary, if it is less than the preset vegetation index threshold, it is a non-green space pixel.

[0077] The vegetation coverage of all green space pixels in each corrected remote sensing image is compared with the preset green space type threshold to obtain the green space type of each green space pixel; wherein, the preset green space type threshold can be understood as a multi-level threshold that includes multiple green space type thresholds at the same time; the corresponding green space types include grassland, shrubs and trees. For example, if the vegetation coverage of the green space pixel calculated by the above method is less than 1.5, the green space pixel can be marked as grassland, if 1.5≤vegetation coverage<2.5, the green space pixel can be marked as shrubs, and if vegetation coverage≥2.5, the green space pixel can be marked as trees. It should be noted that the preset green space type threshold setting method given here is an exemplary description, and can be flexibly adjusted according to application requirements in actual applications.

[0078] Statistical analysis is performed on all green space pixels in each corrected remote sensing image to obtain the green space layout characteristics of the corresponding historical period, and the green space layout time series characteristics are obtained based on the green space layout characteristics of all historical periods; the green space layout characteristics include the total green space area ratio, green space type composition, and the area ratio and spatial distribution of different green space types. Among them, the total green space area ratio in the green space layout characteristics of each corrected remote sensing image can be understood as the ratio of the area occupied by all green space pixels to the total pixel area; the green space type composition can be understood as which green space types are specifically included in the corrected remote sensing image; the area ratio of different green space types can be understood as the ratio of the pixel area of ​​different green space types to the total green space pixel area, for example, the grassland area ratio is 30%, the shrub area ratio is 45%, and the tree area ratio is 25%; the spatial distribution can be understood as the green space layout map corresponding to the corrected remote sensing image marked with the positions of the pixels of different green space types, that is, the green space type distribution grid map with the corresponding green space type code assigned to each pixel.

[0079] According to the meteorological data and green space layout characteristics of all historical periods, the transpiration rate of all green space pixels in each corrected remote sensing image is obtained based on the Penman-Monteith formula; the Penman-Monteith formula is expressed as:

[0080]

[0081] Where ET represents evapotranspiration (transpiration rate); G represents soil heat flow; It represents the slope of the saturated water vapor pressure curve; represents the net radiation; Indicates the air density; It represents specific heat at constant pressure; represents the saturated water vapor pressure; is the actual water vapor pressure; represents aerodynamic impedance; represents canopy impedance; Represents the air humidity constant.

[0082] In practical applications, a vegetation transpiration model is established based on the Penman-Montes formula. The transpiration rate of each green space pixel is calculated based on the acquired meteorological data including meteorological elements such as temperature, humidity and wind speed and the green space layout characteristics (vegetation type and coverage information, etc.) obtained from the above analysis. The specific calculation process can be implemented by referring to the relevant existing technology.

[0083] The transpiration rates of all green space pixels in each corrected remote sensing image are statistically analyzed to obtain the transpiration rate distribution characteristics of the corresponding historical period, and the transpiration rate distribution time series characteristics are obtained based on the transpiration rate distribution characteristics of all historical periods; wherein, the transpiration rate distribution characteristics include the average transpiration rate and the corresponding standard deviation, based on which the corresponding transpiration rate spatial distribution heat map can be drawn, with color gradient representing transpiration intensity, red representing high transpiration areas, and blue representing low transpiration areas; finally, a greening feature report including the green space coverage rate, vegetation type composition, vegetation coverage distribution characteristics, and transpiration rate spatial pattern of each preset functional area can be generated based on the greening distribution data corresponding to the corrected remote sensing images obtained above for viewing and use.

[0084] S13. Perform green space impact analysis based on the green space distribution time series data to obtain corresponding green space impact time series data, and perform power consumption behavior analysis based on the green space distribution time series data, the non-working day residential power consumption time series data, the meteorological time series data and the pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data. Among them, green space impact analysis can be understood as a simultaneous analysis of the impact of green space on microclimate, perceived temperature, thermal comfort and air purification; the corresponding green space impact time series data include green space impact time series data on microclimate, green space impact time series data on perceived temperature, green space impact time series data on thermal comfort and green space impact time series data on air purification.

[0085] Specifically, the step of performing green space impact analysis according to the greening distribution time series data to obtain corresponding green space impact time series data includes:

[0086] Obtain the perceived temperature distribution time series data, thermal comfort index distribution time series data and air quality distribution time series data corresponding to the greening distribution time series data; wherein the thermal comfort index distribution time series data preferably includes predicted average vote number time series data and physiological equivalent temperature time series data; in this embodiment, the historical time periods corresponding to the collection of the perceived temperature distribution time series data, the thermal comfort index distribution time series data and the air quality distribution time series data must be consistent with the data time periods of the greening distribution time series data.

[0087] Specifically, the step of obtaining the time series data of perceived temperature distribution, the time series data of thermal comfort index distribution, and the time series data of air quality distribution corresponding to the greening distribution time series data includes:

[0088] The greening distribution time series data is input into a pre-constructed green space microclimate impact model to perform green space microclimate impact analysis, and the green space microclimate impact time series data is obtained; the green space microclimate impact time series data includes the green space temperature drop amplitude and green space humidification amplitude in each historical period; wherein, the steps of constructing the green space microclimate impact model include:

[0089] The greening distribution time series data to be analyzed is obtained, and the green space layout map and the transpiration rate distribution map corresponding to each time period in the greening distribution time series data to be analyzed are respectively subjected to Kriging interpolation processing, so as to generate the corresponding microclimate distribution time series data to be analyzed; wherein, the greening distribution time series data to be analyzed can be understood as the greening distribution time series data obtained by analyzing the historical data of ground remote sensing images of a preset functional area or multiple preset functional areas for a certain period of time in order to build a model. The specific acquisition method can refer to the relevant description of the greening distribution time series data obtained by performing greening distribution analysis based on the ground remote sensing image time series data, which will not be repeated here. The corresponding microclimate distribution time series data to be analyzed include continuous temperature distribution data and relative humidity distribution data for each time period. The specific acquisition process can be as follows: assuming that based on a green space layout map and a transpiration rate distribution map (transpiration rate spatial distribution heat map) with a resolution of 1km×1km, the Kriging interpolation method is used to generate a continuous temperature field and relative humidity field with a grid size of 100m×100m. Covariates such as terrain (digital elevation model) and building density data can also be introduced in the interpolation process to ensure the reliability of the acquisition of the microclimate distribution time series data to be analyzed. The specific interpolation process can be implemented by referring to the existing Kriging interpolation method and will not be described in detail here.

[0090] According to the buffer zone analysis method, the green space boundary expansion processing is performed on the microclimate distribution time series data to be analyzed to generate preprocessed microclimate distribution time series data with transition zones inside and outside the green space; wherein the buffer zone analysis method can be understood as a method of generating transition zones by boundary expansion based on the GIS buffer zone analysis tool, and the specific preprocessed microclimate distribution time series data can be understood as including green space area microclimate data, transition area microclimate data and non-green space area microclimate data at the same time, and the corresponding acquisition process can be understood as using the GIS buffer zone analysis tool to expand the green space boundary of the microclimate distribution data to be analyzed in each time period in the microclimate distribution time series data to be analyzed outward by a certain range (for example, 50 meters) as the transition zone between green space and non-green space, and then use the updated microclimate distribution time series data as the preprocessed microclimate distribution time series data.

[0091] According to the pre-processed microclimate distribution time series data, the corresponding microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between the inside and outside the green space are obtained, and the microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between the inside and outside the green space are comprehensively compared and analyzed to obtain the corresponding green space microclimate impact data to be analyzed; wherein, the comprehensive comparative analysis can be understood as first analyzing the microclimate data inside the green space with the microclimate data of the transition zone between the inside and outside the green space to obtain the first green space impact temperature reduction and the first green space impact relative humidity increase, and then analyzing the microclimate data outside the green space with the microclimate data of the transition zone between the inside and outside the green space to obtain the second green space impact temperature reduction and the second green space impact relative humidity increase, and finally averaging the first green space impact temperature reduction and the second green space impact temperature reduction, and averaging the first green space impact relative humidity increase and the second green space impact relative humidity increase to obtain the green space microclimate impact data to be analyzed including the green space impact temperature reduction and the green space impact relative humidity increase.

[0092] According to the greening distribution time series data to be analyzed and the green space microclimate impact data to be analyzed, the green space microclimate impact model is constructed based on the random forest regression method; wherein, the training process of the greening distribution time series data to be analyzed is as input variables and the green space microclimate impact data to be analyzed is as output variables, and the random forest regression model is iteratively trained using a 10-fold cross-validation method to obtain an optimal parameter model with converged loss values, and the model is used as the final greening microclimate impact model. It should be noted that in actual applications, with the update of data and the continuous expansion of data volume, the greening distribution time series data to be analyzed and the green space microclimate impact data to be analyzed can be updated based on a timed update mechanism to update and optimize the greening microclimate impact model.

[0093] According to the microclimate distribution time series data corresponding to the greening distribution time series data, the perceived temperature distribution time series data is obtained by the thermal index method; wherein, the microclimate distribution time series data includes continuous air temperature distribution data and relative humidity distribution data for each historical period, which can be obtained by performing Kriging interpolation processing on the green space layout map and transpiration rate distribution map corresponding to each historical period in the greening distribution time series data. The specific acquisition process can be found in the relevant description of the above-mentioned microclimate distribution time series data acquisition process to be analyzed, which will not be repeated here.

[0094] The acquisition process of the somatosensory temperature distribution data of each historical period in the above-mentioned somatosensory temperature distribution time series data can be understood as being based on the microclimate distribution data of the corresponding historical period in the microclimate distribution time series data, and is calculated using the following heat index calculation formula, and the specific calculation process will not be described in detail:

[0095]

[0096] Where HI is the heat index, T is the temperature, and R is the relative humidity.

[0097] Obtain the surface temperature distribution time series data corresponding to the greening distribution time series data, and obtain the corresponding average radiation temperature distribution time series data and wind speed distribution time series data based on the surface temperature distribution time series data; wherein, the process of obtaining the average radiation temperature distribution time series data can be: first analyze the surface temperature distribution time series data to obtain the corresponding surface radiation flux distribution time series data, and then estimate the corresponding average radiation temperature distribution time series data based on the surface radiation flux distribution time series data. Correspondingly, the process of obtaining the wind speed distribution time series data can be obtained by fitting a positive correlation model between surface temperature and wind speed based on the characteristics that an increase in surface temperature will cause the air to expand due to heat, thereby accelerating air flow and increasing wind speed. It should be noted that the above-mentioned average radiation temperature distribution time series data and wind speed distribution time series data can also be estimated based on the surface energy balance equation, which will not be described in detail here.

[0098] The time series data of the green space's impact on microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution are input into a pre-constructed predicted average vote calculation model for analysis to obtain corresponding predicted average vote time series data; wherein, the predicted average vote time series data is predicted average vote distribution data including multiple historical time periods, and the corresponding predicted average vote calculation model can preferably be obtained based on a multilayer perceptron neural network training including 1 input layer, 2 hidden layers, and 1 output layer. The specific training process can adopt the back propagation method and use the mean square error as the loss function to train the network, which will not be described in detail here.

[0099] The time series data of the green space's impact on the microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution are input into a pre-constructed physiological equivalent temperature calculation model for analysis to obtain corresponding physiological equivalent temperature time series data; wherein, the physiological equivalent temperature time series data can be understood as physiological equivalent temperature distribution data including multiple historical time periods, and the corresponding physiological equivalent temperature calculation model can also be obtained based on a multi-layer perceptron neural network training including 1 input layer, 2 hidden layers, and 1 output layer. The specific training process can adopt the back propagation method and use the mean square error as the loss function to train the network, which will not be described in detail here.

[0100] Acquire the air pollutant concentration time series data corresponding to the greening distribution time series data, and use the air pollutant concentration time series data as the air quality distribution time series data; wherein the air pollutant concentration time series data includes fine particulate matter concentration time series data, nitrogen dioxide concentration time series data and ozone concentration time series data.

[0101] After obtaining the relevant data for analyzing the impact of green space through the above method steps, the following method of overlaying the corresponding data with the green space distribution data based on the pixel level can be used to obtain the required green space impact results.

[0102] The somatosensory temperature distribution map corresponding to each historical period in the somatosensory temperature distribution time series data is obtained, and the somatosensory temperature difference inside and outside the green space is analyzed according to the somatosensory temperature distribution map and the corresponding green space layout map, so as to obtain the time series data of the influence of the green space on the somatosensory temperature; that is, through the raster calculator function of the geographic information system software, the somatosensory temperature distribution map corresponding to each historical period and the corresponding green space layout map are superimposed at the pixel level, and the average somatosensory temperature of all green space areas and the average somatosensory temperature of all non-green space areas are calculated, and then the difference between the two is used as the green space influence data on the somatosensory temperature of the corresponding historical period; finally, the green space influence data on the somatosensory temperature of all historical periods are summarized to obtain the time series data of the influence of the green space on the somatosensory temperature.

[0103] The thermal comfort distribution map corresponding to each historical period in the thermal comfort index distribution time series data is obtained, and the thermal comfort difference analysis inside and outside the green space is performed according to the thermal comfort distribution map and the corresponding green space layout map, so as to obtain the time series data of the influence of the green space on thermal comfort; that is, through the raster calculator function of the geographic information system software, the thermal comfort distribution map corresponding to each historical period and the corresponding green space layout map are superimposed at the pixel level, and the average thermal comfort value of all green space areas and the average thermal comfort value of all non-green space areas are calculated, and then the difference between the two is used as the green space influence data on thermal comfort in the corresponding historical period; finally, the green space influence data on thermal comfort in all historical periods are summarized to obtain the green space influence time series data on thermal comfort. It should be noted that since the thermal comfort index distribution time series data also includes the predicted average vote time series data and the physiological equivalent temperature time series data, the above-obtained green space influence time series data on thermal comfort also includes the green space influence time series data on the predicted average vote time series data and the green space influence time series data on the physiological equivalent temperature. The specific calculation method can refer to the above description.

[0104] Through the inverse distance weighted interpolation method, the air quality distribution map corresponding to each historical period in the air quality distribution time series data is obtained, and the difference in pollutant concentration inside and outside the green space is analyzed based on the air quality distribution map and the corresponding green space layout map, so as to obtain the time series data of the green space's impact on air purification; wherein, the method of obtaining the air quality distribution map can be understood as using the inverse distance weighted interpolation method (Inverse Distance Weighting, ‌IDW) to interpolate the air quality distribution data of each historical period in the air quality distribution time series data to generate an image representing the distribution of pollutant concentrations. Through the raster calculator function of the geographic information system software, the air quality distribution map corresponding to each historical period and the corresponding green space layout map are superimposed at the pixel level, and the average air quality of all green space areas and the average air quality of all non-green space areas are calculated, and then the difference between the two is used as the green space's impact on air purification data for the corresponding historical period; finally, the green space's impact on air purification data for all historical periods are summarized to obtain the green space's impact on air purification time series data.

[0105] This embodiment can effectively ensure the comprehensiveness and accuracy of the green space impact results by deeply analyzing the impact of green space levels on microclimate, perceived temperature, thermal comfort and air purification, thereby providing reliable data support for the subsequent construction of a non-working day electricity demand forecasting model based on the green space impact result data.

[0106] Considering that the electricity consumption behavior patterns of residents on non-working days are quite different from those on working days, and the regulatory effect of urban greening levels on microclimate will actually have a certain impact on the electricity consumption behavior patterns of residents, in order to ensure that the construction of the electricity demand forecasting model for non-working days considers the comprehensiveness of the factors affecting electricity demand, this embodiment preferably analyzes the electricity consumption behavior based on the greening distribution time series data, the non-working day residents' electricity consumption time series data, the meteorological time series data and the pre-constructed electricity consumption behavior forecasting model while obtaining the green space impact time series data based on the greening distribution time series data. Specifically, the steps for constructing the electricity consumption behavior forecasting model include:

[0107] Obtain the time series data of residential electricity consumption on non-working days to be analyzed for each preset functional area, as well as the corresponding meteorological time series data to be analyzed and the green distribution time series data to be analyzed; among them, the time series data of residential electricity consumption on non-working days to be analyzed, the meteorological time series data to be analyzed and the green distribution time series data to be analyzed can all be understood as historical time series data of a certain length obtained for model construction needs, and the green distribution time series data to be analyzed can be obtained based on the analysis of historical data of ground remote sensing images of the corresponding historical length, which will not be described in detail here.

[0108] Based on the preset electricity consumption behavior characteristics, cluster analysis is performed on the non-working day residential electricity consumption time series data to be analyzed in each preset functional area to obtain the time series data of electricity consumption time period behavior distribution; wherein, the preset electricity consumption behavior characteristics may include peak electricity consumption, duration, and occurrence period, etc., which are not specifically limited here. In practical applications, considering that the electricity consumption activities of residents on non-working days are mainly leisure and entertainment and housework activities, the number of clusters is set to 2, and the peak electricity consumption, duration, and occurrence period are selected as features. The K-means clustering algorithm is used to perform a preset number of iterative clustering on the non-working day residential electricity consumption time series data to be analyzed to obtain the electricity consumption behavior distribution in each period; correspondingly, the electricity consumption time period behavior distribution time series data includes the electricity consumption behavior type in each period; the electricity consumption behavior type includes leisure and entertainment and housework activities.

[0109] The meteorological time series data to be analyzed and the greening distribution time series data to be analyzed in each preset functional area are analyzed respectively to obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data; among them, the wind speed change distribution time series data and the shade area change distribution time series data can be understood as the factors affecting residents' electricity consumption behavior that are affected by the greening level, based on the consideration that the impact of urban greening on microclimate will directly affect residents' electricity consumption behavior.

[0110] Specifically, the steps of respectively analyzing the to-be-analyzed meteorological time series data and the to-be-analyzed greening distribution time series data of each preset functional area to obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data include:

[0111] Fluid dynamics simulation is performed according to the meteorological time series data to be analyzed to obtain corresponding wind environment distribution time series data, and differential analysis is performed on the wind environment distribution time series data to obtain corresponding wind speed change distribution time series data; wherein, the fluid dynamics simulation process is: constructing a 3D model in ANSYS Fluent, setting the grid size, simulation time step and boundary conditions, and setting the inlet wind speed according to the meteorological time series data to be analyzed including wind environment-related meteorological data such as wind speed and wind direction, setting the green space height and building height in combination with the corresponding ground remote sensing impact data, calculating the wind environment distribution time series data of a preset simulation time period, and performing differential analysis on the wind environment distribution time series data (the difference between the wind environment distribution data of two adjacent time periods) to analyze the wind environment changes, and obtaining the required wind speed change distribution time series data.

[0112] According to the green area corresponding to the greening distribution time series data to be analyzed, the corresponding green space shade effect distribution time series data is generated, and the green space shade effect distribution time series data is differentially analyzed to obtain the corresponding shade area change distribution time series data; wherein, the green space shade effect distribution data of each time period in the green space shade effect distribution time series data can be understood as the green space area distribution data in the corresponding greening distribution data to be analyzed; the green space shade effect distribution time series data is differentially analyzed (the difference between the green space shade effect distribution data of two adjacent time periods) to analyze the green space shade effect change and obtain the required shade area change distribution time series data.

[0113] The wind speed change distribution time series data and the shade area change distribution time series data of all preset functional areas are used as independent variable data, and the power consumption time period behavior distribution time series data of all preset functional areas are used as dependent variable data, and the power consumption behavior prediction model is constructed based on the generalized additive model; wherein the generalized additive model is a GAM (Generalized Additive Model) model, based on this model, a thin plate spline is used as a smoothing function, and the optimal parameters are selected through cross-validation to obtain the required power consumption behavior prediction model.

[0114] This embodiment constructs an electricity consumption behavior prediction model based on the impact of green space on wind environment and shade area, which can effectively capture the impact of greening level on residents' electricity consumption behavior on non-working days, thereby ensuring the effectiveness of the subsequent introduction of electricity consumption behavior to construct a non-working day electricity demand prediction model.

[0115] S14. Construct a non-working day electricity demand forecasting model based on the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data, and meteorological time series data of all preset functional areas; wherein the non-working day electricity demand forecasting model can be understood as a model that can comprehensively and reliably predict the non-working electricity demand in the future period based on the electricity demand impact related data collected in real time from each preset functional area; specifically, the step of constructing a non-working day electricity demand forecasting model based on the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data, and meteorological time series data of all preset functional areas includes:

[0116] Construct a power forecast training set based on the power consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential power consumption time series data and meteorological time series data of all preset functional areas;

[0117] The power prediction training set is normalized to obtain a corresponding normalized training set; wherein, the normalization process can be implemented in principle by any existing normalization technology, and this embodiment preferably adopts Z-score normalization processing, and the specific process of obtaining the normalized training set is not described in detail here.

[0118] According to the normalized training set, the preset long short-term memory network is iteratively trained to obtain the non-working day electricity demand forecasting model; the preset long short-term memory network includes two LSTM layers and one fully connected layer, wherein the number of hidden units in each LSTM layer and the total number of neurons in the fully connected layer can be selected based on the data in the normalized training set and the actual application requirements, and no specific limitation is made here.

[0119] This embodiment constructs a non-working day electricity demand prediction model based on the preset long short-term memory network training based on electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data. While considering the impact of meteorological data on non-working day electricity demand, the impact of urban greening and special electricity consumption behavior of non-working day residents on electricity demand is further considered, which can ensure the reliability of the construction of the non-working day electricity demand prediction model, so as to capture more comprehensive and deeper changes in electricity demand, and provide strong guarantee for the accuracy of the corresponding non-working day electricity demand prediction results.

[0120] S15. According to the non-working day electricity demand prediction model, the target period electricity demand is predicted based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of non-working day residents and the real-time meteorological data, and the corresponding regional period electricity demand prediction results are obtained; wherein, the process of obtaining the regional period electricity demand prediction results can be understood as firstly obtaining the corresponding greening distribution data, electricity consumption behavior distribution data and green space impact data based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of non-working day residents and the real-time meteorological data, and then unifying them with the real-time electricity consumption of non-working day residents and the real-time meteorological data as the input of the non-working day electricity demand prediction model, which is used to predict the non-working day residents' electricity demand for the next target period of each preset functional area, and then obtain the regional period electricity demand prediction results of each preset functional area.

[0121] S16. Summarize the electricity demand forecast results for all regional time periods to obtain the electricity demand forecast results for the target city time period; wherein, the electricity demand forecast results for the target city time period can be understood as the accumulation of the regional electricity demand forecast results for all preset functional areas in the target city. The electricity demand forecast results for the target city time period are more comprehensive and reliable non-working day residents' electricity demand forecast results obtained by introducing factors such as urban greening level and special electricity consumption behavior of residents on non-working days.

[0122] The present embodiment provides a method for obtaining the ground remote sensing image time series data, non-working day residential electricity consumption time series data, and meteorological time series data of each preset functional area in the target city, and performing greening distribution analysis on the ground remote sensing image time series data of each preset functional area respectively, and obtaining the corresponding greening distribution time series data including the green space layout time series characteristics and the transpiration rate distribution time series characteristics, and then performing green space impact analysis based on the greening distribution time series data to obtain the corresponding green space impact time series data, and performing electricity consumption behavior analysis based on the greening distribution time series data, non-working day residential electricity consumption time series data, meteorological time series data, and a pre-constructed electricity consumption behavior prediction model to obtain the corresponding electricity consumption behavior distribution time series data, and constructing a prediction model based on the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data, and meteorological time series data of all preset functional areas. A technical solution is provided for building a non-working day electricity demand prediction model, and predicting the electricity demand for a target period based on the non-working day electricity demand prediction model, based on real-time ground remote sensing images of each preset functional area, real-time electricity consumption of residents on non-working days and real-time meteorological data, to obtain corresponding regional period electricity demand prediction results, and summarizing all regional period electricity demand prediction results to obtain target city period electricity demand prediction results. This solution effectively solves the application defect that the existing non-working day electricity demand prediction method does not consider the influencing factors of greening level and non-working special electricity consumption behavior, resulting in incomplete and accurate prediction analysis. That is, through in-depth analysis of the microclimate effect of urban greening, and combining the unique electricity consumption rules of non-working days, a comprehensive and efficient non-working day electricity demand prediction model is constructed, which effectively improves the comprehensiveness and accuracy of non-working day electricity demand prediction.

[0123] In addition, considering that the level of urban greening is changing, the impact of green space on microclimate and residents' electricity consumption behavior on non-working days will also change. In order to better adapt to the changes in the actual greening level and ensure the continuous stability of the prediction effect of the non-working day electricity demand prediction model. In this embodiment, preferably, the non-working day electricity demand prediction model is adaptively and dynamically optimized by regularly monitoring the degree of change in greening distribution data; specifically, Figure 2 As shown, the method also includes:

[0124] S17. Regularly monitor the degree of change of greening distribution data of all preset functional areas, and when the degree of change of greening distribution data of any preset functional area exceeds the preset degree of change threshold, adaptively optimize and update the non-working day electricity demand forecasting model. Among them, the degree of change of greening distribution data can be understood as the difference average value of greening distribution data obtained based on two adjacent monitoring moments; when at a certain monitoring moment, the degree of change of greening distribution data of at least one preset functional area exceeds the preset degree of change threshold (which can be set according to actual needs), the training data of the currently used non-working day electricity demand forecasting model (electricity consumption behavior distribution time series data of all preset functional areas, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data) will be updated based on the latest collected relevant data of all preset functional areas, and then the currently used non-working day electricity demand forecasting model will be optimized and trained, so that the obtained non-working day electricity demand forecasting model can effectively adapt to the recent changes in urban greening levels and electricity consumption behavior, thereby ensuring the reliability and accuracy of recent electricity demand forecasting.

[0125] The embodiment of the present application obtains ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data of each preset functional area in the target city, and performs greening distribution analysis on the ground remote sensing image time series data of each preset functional area respectively, and obtains the corresponding greening distribution time series data including green space layout time series characteristics and transpiration rate distribution time series characteristics. Then, a green space impact analysis is performed based on the greening distribution time series data to obtain the corresponding green space impact time series data, and a power consumption behavior analysis is performed based on the greening distribution time series data, non-working day residential electricity consumption time series data, meteorological time series data and a pre-constructed power consumption behavior prediction model to obtain the corresponding power consumption behavior distribution time series data, and a non-working day electricity demand prediction model is constructed based on the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data of all preset functional areas, and based on the non-working day electricity demand prediction model, the real-time data of each preset functional area is generated. The target period electricity demand is predicted by ground remote sensing images, real-time electricity consumption of residents on non-working days and real-time meteorological data to obtain the corresponding regional period electricity demand forecast results, and all regional period electricity demand forecast results are summarized to obtain the target city period electricity demand forecast results. The degree of change of greening distribution data of all preset functional areas is monitored regularly. When the degree of change of greening distribution data of any preset functional area exceeds the preset change degree threshold, the non-working day electricity demand forecast model is adaptively optimized and updated. This technical solution not only effectively improves the comprehensiveness and accuracy of non-working day electricity demand forecast by introducing urban greening and special electricity consumption behavior factors of non-working day residents into electricity demand forecast analysis, but also improves the adaptability of the forecast model by regularly monitoring the greening level and electricity consumption behavior changes and dynamically optimizing and updating the forecast model, providing reliable guarantee for the continuous stability of the non-working day electricity demand forecast effect, and thus providing a reliable analysis basis for power supply dispatching of the power grid system.

[0126] It should be noted that although the steps in the above flowchart are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders.

[0127] In one embodiment, Figure 3 As shown, a non-working day electricity demand forecasting system considering the impact of urban greening is provided, and the system includes:

[0128] Data acquisition module 1 is used to obtain ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data of each preset functional area in the target city;

[0129] The greening distribution analysis module 2 is used to perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data; the greening distribution time series data includes green space layout time series characteristics and transpiration rate distribution time series characteristics;

[0130] The greening impact analysis module 3 is used to perform green space impact analysis according to the greening distribution time series data to obtain corresponding green space impact time series data, and to perform power consumption behavior analysis according to the greening distribution time series data, the non-working day residential power consumption time series data, the meteorological time series data and a pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data;

[0131] The prediction model building module 4 is used to build a non-working day electricity demand prediction model based on the electricity consumption behavior distribution time series data of all preset functional areas, the greening distribution time series data, the green space impact time series data, the non-working day residential electricity consumption time series data and the meteorological time series data;

[0132] The regional demand forecasting module 5 is used to forecast the electricity demand in the target period according to the non-working day electricity demand forecasting model, based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of residents on non-working days and the real-time meteorological data, and obtain the corresponding regional period electricity demand forecasting result;

[0133] The city demand forecasting module 6 is used to summarize the power demand forecasting results of all regions and time periods to obtain the power demand forecasting results of the target city time period.

[0134] In one embodiment, Figure 4 As shown, a non-working day electricity demand forecasting system considering the impact of urban greening is provided, and the system also includes:

[0135] The model optimization and updating module 7 is used to regularly monitor the degree of change of the greening distribution data of all preset functional areas, and when the degree of change of the greening distribution data of any preset functional area exceeds a preset change degree threshold, the non-working day electricity demand prediction model is adaptively optimized and updated.

[0136] For the specific definition of the non-working day electricity demand forecasting system considering the impact of urban greening, please refer to the definition of the non-working day electricity demand forecasting method considering the impact of urban greening above, and the corresponding technical effects can also be obtained equivalently, which will not be repeated here. Each module in the above-mentioned non-working day electricity demand forecasting system considering the impact of urban greening can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0137] In summary, an embodiment of the present invention provides a method and system for predicting non-working day electricity demand considering the impact of urban greening. The method and system for predicting non-working day electricity demand considering the impact of urban greening realizes the acquisition of ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data of each preset functional area in the target city, and performs greening distribution analysis on the ground remote sensing image time series data of each preset functional area respectively, and obtains the corresponding greening distribution time series data including green space layout time series characteristics and transpiration rate distribution time series characteristics. Then, a green space impact analysis is performed according to the greening distribution time series data to obtain the corresponding green space impact time series data, and an electricity behavior analysis is performed according to the greening distribution time series data, non-working day residential electricity consumption time series data, meteorological time series data and a pre-constructed electricity behavior prediction model to obtain the corresponding electricity behavior distribution time series data, and a non-working day electricity demand prediction model is constructed according to the electricity behavior distribution time series data of all preset functional areas, the greening distribution time series data, the green space impact time series data, the non-working day residential electricity consumption time series data and the meteorological time series data, and the The non-working day electricity demand forecasting model predicts the electricity demand for the target period based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of non-working day residents and the real-time meteorological data to obtain the corresponding regional period electricity demand forecast results, and after summarizing all regional period electricity demand forecast results to obtain the target city period electricity demand forecast results, the greening distribution data change degree of all preset functional areas is regularly monitored, and when the greening distribution data change degree of any preset functional area exceeds the preset change degree threshold, the non-working day electricity demand forecasting model is adaptively optimized and updated. The technical solution not only effectively improves the comprehensiveness and accuracy of the non-working day electricity demand forecast by introducing urban greening and special electricity consumption behavior factors of non-working day residents to the electricity demand forecast analysis, but also improves the adaptability of the forecasting model by regularly monitoring the greening level and the changes in electricity consumption behavior and dynamically optimizing and updating the forecasting model, providing reliable guarantee for the continuous stability of the non-working day electricity demand forecasting effect, and thus providing a reliable analysis basis for the power supply dispatching of the power grid system.

[0138] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above-mentioned embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0139] The above-mentioned embodiments only express several preferred implementation modes of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in the technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be based on the protection scope of the claims.

Claims

1. A method for predicting power demand on non-working days considering the impact of urban greening, characterized in that: The method comprises the following steps: According to the sampling period and the preset historical duration, the ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data of each preset functional area in the target city are obtained; Perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data; the greening distribution time series data includes green space layout time series characteristics and transpiration rate distribution time series characteristics; Performing green space impact analysis based on the greening distribution time series data to obtain corresponding green space impact time series data, and performing power consumption behavior analysis based on the greening distribution time series data, the non-working day residents' power consumption time series data, the meteorological time series data and a pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data; the green space impact time series data includes green space impact time series data on microclimate, green space impact time series data on body temperature, green space impact time series data on thermal comfort and green space impact time series data on air purification; Based on the time series data of electricity consumption behavior distribution, greening distribution, green space impact, non-working day residential electricity consumption and meteorological time series data of all preset functional areas, a non-working day electricity demand forecasting model is constructed; According to the non-working day electricity demand forecasting model, the electricity demand forecast for the target period is performed based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of residents on non-working days and the real-time meteorological data, and the corresponding regional period electricity demand forecast results are obtained; Summarize the electricity demand forecast results of all regions and time periods to obtain the electricity demand forecast results of the target city time period; The step of performing green space impact analysis according to the greening distribution time series data to obtain corresponding green space impact time series data includes: Input the greening distribution time series data into a pre-constructed green space microclimate impact model to perform green space microclimate impact analysis, and obtain the green space microclimate impact time series data; the green space microclimate impact time series data includes the green space temperature drop amplitude and green space humidification amplitude in each historical period; Obtaining the time series data of perceived temperature distribution, the time series data of thermal comfort index distribution, and the time series data of air quality distribution corresponding to the greening distribution time series data; Obtaining a body temperature distribution map corresponding to each historical period in the body temperature distribution time series data, and performing a body temperature difference analysis inside and outside the green space based on the body temperature distribution map and the corresponding green space layout map, to obtain time series data on the impact of the green space on the body temperature; Obtain a thermal comfort distribution map corresponding to each historical period in the thermal comfort index distribution time series data, and perform a thermal comfort difference analysis inside and outside the green space based on the thermal comfort distribution map and the corresponding green space layout map to obtain the time series data of the impact of the green space on thermal comfort; By using the inverse distance weighted interpolation method, an air quality distribution map corresponding to each historical period in the air quality distribution time series data is obtained, and the difference in pollutant concentration inside and outside the green space is analyzed based on the air quality distribution map and the corresponding green space layout map, so as to obtain the time series data of the impact of the green space on air purification; The steps of constructing the electricity consumption behavior prediction model include: Obtain the time series data of residential electricity consumption on non-working days to be analyzed in each preset functional area, as well as the corresponding time series data of meteorological data to be analyzed and time series data of greening distribution to be analyzed; Based on the preset electricity consumption behavior characteristics, cluster analysis is performed on the non-working day residential electricity consumption time series data to be analyzed in each preset functional area to obtain the electricity consumption time period behavior distribution time series data; the electricity consumption time period behavior distribution time series data includes the electricity consumption behavior type of each time period; the electricity consumption behavior type includes leisure and entertainment and housework activities; Analyze the meteorological time series data to be analyzed and the greening distribution time series data to be analyzed for each preset functional area respectively, and obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data; The wind speed change distribution time series data and the shade area change distribution time series data of all preset functional areas are used as independent variable data, and the power consumption time period behavior distribution time series data of all preset functional areas are used as dependent variable data, and the power consumption behavior prediction model is constructed based on the generalized additive model.

2. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The ground remote sensing image time series data is multispectral remote sensing image data of each historical period; the meteorological time series data includes temperature, humidity and wind speed of each historical period; the green space layout time series characteristics include green space layout characteristics of each historical period; The transpiration rate distribution time series characteristics include the transpiration rate distribution characteristics of each historical period; The step of performing greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data comprises: Performing atmospheric correction and geometric calibration on the ground remote sensing image time series data to obtain corrected remote sensing image time series data; the corrected remote sensing image time series data includes corrected remote sensing images of various historical periods; According to the normalized vegetation index, the vegetation coverage of all pixels in each corrected remote sensing image is calculated, and according to the comparison result between the vegetation coverage and the preset vegetation index threshold, each pixel is divided into a green space pixel and a non-green space pixel; Compare the vegetation coverage of all green space pixels in each corrected remote sensing image with a preset green space type threshold to obtain the green space type of each green space pixel; the green space type includes grass, shrubs and trees; Statistical analysis is performed on all green space pixels in each corrected remote sensing image to obtain the green space layout characteristics of the corresponding historical period, and the green space layout time series characteristics are obtained based on the green space layout characteristics of all historical periods; the green space layout characteristics include the proportion of the total green space area, the composition of green space types, and the area proportion and spatial distribution of different green space types; According to the meteorological data of all historical periods and the layout characteristics of green spaces, the transpiration rates of all green space pixels in each corrected remote sensing image were obtained based on the Penman-Montes formula. The transpiration rates of all green space pixels in each corrected remote sensing image are statistically analyzed to obtain the transpiration rate distribution characteristics of the corresponding historical period, and the transpiration rate distribution time series characteristics are obtained based on the transpiration rate distribution characteristics of all historical periods; the transpiration rate distribution characteristics include the average transpiration rate and the corresponding standard deviation.

3. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The thermal comfort index distribution time series data includes predicted average vote count time series data and physiological equivalent temperature time series data; The step of obtaining the time series data of body temperature distribution, the time series data of thermal comfort index distribution and the time series data of air quality distribution corresponding to the time series data of greening distribution comprises: According to the microclimate distribution time series data corresponding to the greening distribution time series data, the body temperature distribution time series data is obtained by a heat index method; Acquire the surface temperature distribution time series data corresponding to the greening distribution time series data, and obtain the corresponding average radiation temperature distribution time series data and wind speed distribution time series data according to the surface temperature distribution time series data; Input the time series data of the green space's impact on microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution into a pre-built predicted average vote number calculation model for analysis to obtain corresponding predicted average vote number time series data; Input the time series data of the green space's impact on microclimate, the time series data of the average radiation temperature distribution, and the time series data of the wind speed distribution into a pre-built physiological equivalent temperature calculation model for analysis to obtain corresponding physiological equivalent temperature time series data; The air pollutant concentration time series data corresponding to the greening distribution time series data is obtained, and the air pollutant concentration time series data is used as the air quality distribution time series data; the air pollutant concentration time series data includes fine particulate matter concentration time series data, nitrogen dioxide concentration time series data and ozone concentration time series data.

4. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The steps of constructing the green space impact model on microclimate include: Acquire the greening distribution time series data to be analyzed, and perform Kriging interpolation processing on the green space layout map and transpiration rate distribution map corresponding to each time period in the greening distribution time series data to be analyzed, and generate the corresponding microclimate distribution time series data to be analyzed; the microclimate distribution time series data to be analyzed includes continuous temperature distribution data and relative humidity distribution data of each time period; According to the buffer zone analysis method, the microclimate distribution time series data to be analyzed is processed by expanding the green space boundary to generate pre-processed microclimate distribution time series data with transition zones inside and outside the green space; According to the pre-processed microclimate distribution time series data, the corresponding microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between the inside and outside of the green space are obtained, and the microclimate time series data inside the green space, the microclimate time series data outside the green space, and the microclimate time series data of the transition zone between the inside and outside of the green space are comprehensively compared and analyzed to obtain the corresponding green space microclimate impact data to be analyzed; the green space microclimate impact data to be analyzed includes the amount of temperature reduction affected by the green space and the amount of relative humidity increase affected by the green space; According to the greening distribution time series data to be analyzed and the green space microclimate impact data to be analyzed, a model of the green space's impact on microclimate is constructed based on the random forest regression method.

5. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The step of respectively analyzing the to-be-analyzed meteorological time series data and the to-be-analyzed greening distribution time series data of each preset functional area to obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data comprises: Performing fluid dynamics simulation according to the meteorological time series data to be analyzed to obtain corresponding wind environment distribution time series data, and performing differential analysis on the wind environment distribution time series data to obtain corresponding wind speed change distribution time series data; According to the green area corresponding to the greening distribution time series data to be analyzed, the corresponding green shade effect distribution time series data is generated, and the green shade effect distribution time series data is differentially analyzed to obtain the corresponding shade area change distribution time series data.

6. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The step of constructing a non-working day electricity demand prediction model based on the electricity consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential electricity consumption time series data and meteorological time series data of all preset functional areas includes: Construct a power forecast training set based on the power consumption behavior distribution time series data, greening distribution time series data, green space impact time series data, non-working day residential power consumption time series data and meteorological time series data of all preset functional areas; Normalizing the power prediction training set to obtain a corresponding normalized training set; According to the normalized training set, the preset long short-term memory network is iteratively trained to obtain the non-working day electricity demand prediction model; the preset long short-term memory network includes two LSTM layers and one fully connected layer.

7. The method for predicting non-working day electricity demand considering the impact of urban greening as claimed in claim 1, characterized in that: The method further comprises: The degree of change of greening distribution data of all preset functional areas is monitored regularly, and when the degree of change of greening distribution data of any preset functional area exceeds a preset change degree threshold, the non-working day electricity demand prediction model is adaptively optimized and updated.

8. A non-working day electricity demand forecasting system considering the impact of urban greening, characterized in that: The system comprises: The data acquisition module is used to obtain the ground remote sensing image time series data, non-working day residential electricity consumption time series data and meteorological time series data of each preset functional area in the target city according to the sampling period and the preset historical duration; A greening distribution analysis module is used to perform greening distribution analysis on the ground remote sensing image time series data of each preset functional area to obtain corresponding greening distribution time series data; the greening distribution time series data includes green space layout time series characteristics and transpiration rate distribution time series characteristics; A greening impact analysis module is used to perform greening impact analysis based on the greening distribution time series data to obtain corresponding greening impact time series data, and to perform power consumption behavior analysis based on the greening distribution time series data, the non-working day residents' power consumption time series data, the meteorological time series data and a pre-built power consumption behavior prediction model to obtain corresponding power consumption behavior distribution time series data; the greening impact time series data includes greening impact time series data on microclimate, greening impact time series data on body temperature, greening impact time series data on thermal comfort and greening impact time series data on air purification; A prediction model building module is used to build a non-working day electricity demand prediction model based on the electricity consumption behavior distribution time series data of all preset functional areas, the greening distribution time series data, the green space impact time series data, the non-working day residential electricity consumption time series data and the meteorological time series data; A regional demand forecasting module is used to forecast the electricity demand in a target period according to the non-working day electricity demand forecasting model, based on the real-time ground remote sensing images of each preset functional area, the real-time electricity consumption of residents on non-working days and real-time meteorological data, and obtain the corresponding regional period electricity demand forecasting results; The city demand forecasting module is used to aggregate the power demand forecasting results of all regions and time periods to obtain the power demand forecasting results of the target city time period; The green space impact analysis is performed according to the greening distribution time series data to obtain the corresponding green space impact time series data, including: Input the greening distribution time series data into a pre-constructed green space microclimate impact model to perform green space microclimate impact analysis, and obtain the green space microclimate impact time series data; the green space microclimate impact time series data includes the green space temperature drop amplitude and green space humidification amplitude in each historical period; Obtaining the time series data of perceived temperature distribution, the time series data of thermal comfort index distribution, and the time series data of air quality distribution corresponding to the greening distribution time series data; Obtaining a body temperature distribution map corresponding to each historical period in the body temperature distribution time series data, and performing a body temperature difference analysis inside and outside the green space based on the body temperature distribution map and the corresponding green space layout map, to obtain time series data on the impact of the green space on the body temperature; Obtain a thermal comfort distribution map corresponding to each historical period in the thermal comfort index distribution time series data, and perform a thermal comfort difference analysis inside and outside the green space based on the thermal comfort distribution map and the corresponding green space layout map to obtain the time series data of the impact of the green space on thermal comfort; By using the inverse distance weighted interpolation method, an air quality distribution map corresponding to each historical period in the air quality distribution time series data is obtained, and the difference in pollutant concentration inside and outside the green space is analyzed based on the air quality distribution map and the corresponding green space layout map, so as to obtain the time series data of the impact of the green space on air purification; The construction of the electricity consumption behavior prediction model includes: Obtain the time series data of residential electricity consumption on non-working days to be analyzed in each preset functional area, as well as the corresponding time series data of meteorological data to be analyzed and time series data of greening distribution to be analyzed; Based on the preset electricity consumption behavior characteristics, cluster analysis is performed on the non-working day residential electricity consumption time series data to be analyzed in each preset functional area to obtain the electricity consumption time period behavior distribution time series data; the electricity consumption time period behavior distribution time series data includes the electricity consumption behavior type of each time period; the electricity consumption behavior type includes leisure and entertainment and housework activities; Analyze the meteorological time series data to be analyzed and the greening distribution time series data to be analyzed for each preset functional area respectively, and obtain the corresponding wind speed change distribution time series data and shade area change distribution time series data; The wind speed change distribution time series data and the shade area change distribution time series data of all preset functional areas are used as independent variable data, and the power consumption time period behavior distribution time series data of all preset functional areas are used as dependent variable data, and the power consumption behavior prediction model is constructed based on the generalized additive model.

Images