Air conditioner electrical load prediction method and system based on meteorological factors

By constructing a prediction model for air conditioner power load based on meteorological factors, combining meteorological, building and air conditioning data, the problem of inaccurate prediction results in the existing technology is solved, the prediction accuracy of air conditioner power load is improved, and the operation of hydropower stations and power grids is carried out. Provides more accurate energy scheduling support.

CN119010023APending Publication Date: 2024-11-22LISHUI POWER SUPPLY COMPANY OF STATE GRID ZHEJIANG ELECTRIC POWER

Patent Information

Application Number
CN202411497491.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-25
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The prior art ignores meteorological factors and building characteristics in the prediction of air conditioner power load, resulting in inaccurate prediction results.

Method used

Using the air conditioner power load prediction method based on meteorological factors, the user side’s historical meteorological data, building configuration data and air conditioner historical operation data are obtained to build a cyclic neural network model, and combined with the key influence factors of air conditioners, building load and air conditioner operation load are constructed. Predict the model and consider the river meteorological prediction data on the power supply side of the hydropower station.

Benefits of technology

It improves the accuracy of air-conditioning power load prediction, can more accurately reflect the actual load situation in the target area on the user side, provides strong support for hydropower station power supply scheduling and air-conditioning energy consumption management, and optimizes energy utilization efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119010023A_ABST
    Figure CN119010023A_ABST
Patent Text Reader

Abstract

The invention discloses an air conditioner electrical load prediction method and system based on meteorological factors, and the method is a prediction method for the electrical load of an air conditioner at a user side, and is especially suitable for a scene where power is supplied by a hydropower station. Historical meteorological data, air conditioner basic configuration data, building configuration data and air conditioner historical operation data of a user side target area are comprehensively utilized, a recurrent neural network technology is combined, a building load prediction model is constructed and optimized, and key influence factors of an air conditioner are utilized to adjust the model, so that a building load prediction result is obtained. An air conditioner operation load prediction model is obtained; meanwhile, the influence of a hydropower station power supply side on river weather is considered, weather prediction data is used as input, and air conditioner basic configuration data is combined to accurately quantify the air conditioner electrical load of a user side target area in a future preset time period, so that the problem that the prediction result of the air conditioner electrical load of a prediction model adopted at present is inaccurate is solved, and the prediction accuracy of the air conditioner electrical load is improved. Powerful support is provided for hydropower station power supply dispatching and air conditioner energy consumption management.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method and system for predicting air-conditioning power load based on meteorological factors. Background Art

[0002] With the continuous development of machine learning technology, various prediction models formed by machine learning technology have been applied to load prediction scenarios of a large number of power equipment.

[0003] In the prior art, the prediction of air-conditioning power load measurement mostly adopts the prediction model constructed by machine learning technology to analyze the historical data of air-conditioning power load, which often ignores the meteorological factors in the area where the air-conditioning is located and the impact of the building on its power load, resulting in inaccurate prediction results.

[0004] It can be seen that how to improve the accuracy of air-conditioning power load prediction results has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0005] The present invention provides a method and system for predicting air-conditioning power load based on meteorological factors, which solves the problem of how to improve the accuracy of the prediction result of air-conditioning power load.

[0006] In order to solve the above technical problems, the first aspect of the present invention provides an air conditioning power load prediction method based on meteorological factors, which is applied to the air conditioner on the user side, and the air conditioner is powered by a hydropower station. The method includes: Acquire historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data; The historical meteorological data, the building configuration data, the historical operation mode of the air conditioner and the historical operation temperature of the air conditioner are used as training data to train a recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the target area on the user side; Determine the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjust the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; Constructing a river meteorological prediction model for the power supply side of the hydropower station, and obtaining meteorological prediction data for the target area on the user side in a preset period of time in the future output by the river meteorological prediction model; The meteorological forecast data is input into the air conditioning operation load prediction model for processing, and the air conditioning operation load prediction data is output. In combination with the air conditioning basic configuration data, the air conditioning power load of the user-side target area in the future preset period is quantified.

[0007] As one of the preferred solutions, the historical meteorological data, the building configuration data, the historical air conditioner operation mode and the historical air conditioner operation temperature are used as training data to train a recurrent neural network to obtain a trained building load prediction model, including: quantifying the building heat dissipation when each building space in the user-side target area reaches the corresponding optimal air conditioning operation temperature under different air conditioning operation modes according to the building configuration data, the air conditioning historical operation mode and the air conditioning historical operation temperature; The recurrent neural network is trained using the training data constructed by the historical meteorological data and the building heat dissipation to obtain a trained building load prediction model.

[0008] As one of the preferred solutions, the training data constructed by the historical meteorological data and the building heat dissipation is used to train the recurrent neural network to obtain a trained building load prediction model, including: Decomposing the historical meteorological data into trend items, seasonal items and random items by time series decomposition method, modeling each item separately, and constructing a building load characteristic matrix in combination with the building heat dissipation; Processing the building load characteristic matrix by a sliding time window method to obtain training input sequence data; Based on the training input sequence data, the building load prediction model is constructed through a long short-term memory network, and the historical meteorological data is used as an additional feature to train the building load prediction model to obtain a trained building load prediction model.

[0009] As one of the preferred solutions, determining the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner includes: Standardizing the air conditioner basic configuration data, the air conditioner historical fault data, and the air conditioner historical maintenance data to obtain an impact feature sequence with a unified format; The correlation between the influencing feature sequence and the corresponding building heat dissipation is calculated by the Pearson correlation coefficient method to generate a correlation coefficient matrix; Based on the correlation coefficient matrix, the influencing feature sequence is reduced in dimension and screened by principal component analysis, and the influencing feature sequence whose cumulative contribution rate reaches the contribution threshold is selected as a candidate factor; A multiple linear regression model is established with the candidate factors as independent variables and the heat dissipation of the building as the dependent variable, and the multiple linear regression model is screened by a stepwise regression method to obtain the key influencing factors.

[0010] As one of the preferred solutions, the building load prediction model is adjusted by the key influencing factors to obtain the air conditioning operation load prediction model of the user-side target area, including: Initializing a particle swarm and using the mean square error of the building load prediction model as a fitness function; wherein the position of each particle in the particle swarm is formed by encoding the initial value of the model parameter of the building load prediction model and the key influencing factor; Based on the fitness function, the positions and velocities of the particles in the particle swarm are updated by using a standard particle swarm algorithm to obtain optimal model parameters; Based on the optimal model parameters, the positions and velocities of the particles in the particle swarm are updated by a cooperative particle swarm algorithm to obtain updated optimal model parameters, and the updating steps of the cooperative particle swarm algorithm are repeated until a preset number of iterations is reached and the final optimal model parameters are output; The building load prediction model is updated based on the final optimal model parameters to obtain the air conditioning operation load prediction model of the user-side target area.

[0011] As one of the preferred solutions, the construction of the river meteorological prediction model on the power supply side of the hydropower station includes: Acquire real-time water level data of the power supply side of the hydropower station, and predict water inflow forecast data of the power supply side of the hydropower station in the future preset time period according to the real-time water level data; Real-time meteorological data of the user-side target area is acquired, and a river meteorological prediction model is constructed according to the inflow prediction data and the real-time meteorological data.

[0012] As one of the preferred solutions, the real-time water level data of the power supply side of the hydropower station is obtained, and the water inflow forecast data of the power supply side of the hydropower station in the future preset period is predicted according to the real-time water level data, including: Acquire real-time water level data collected by multiple hydrological stations located on the power supply side of the hydropower station, and query corresponding flow values ​​from a pre-established water level and flow relationship database based on the real-time water level data; The queried flow value is fitted by an autoregressive integral moving average model to obtain the water inflow forecast data of the power supply side of the hydropower station in the future preset period; Acquire historical water inflow data of the power supply side of the hydropower station, and construct a meteorological regression model based on the historical water inflow data and the historical meteorological data; Inputting the water inflow forecast data into the meteorological regression model for forecasting, and outputting the meteorological change data of the user-side target area in the future preset period; When the meteorological change data exceeds the change threshold, the real-time water level data of the power supply side of the hydropower station is reacquired to repeat the calculation steps of the water inflow forecast data of the power supply side of the hydropower station in the future preset time period, until the recalculated meteorological change data does not exceed the change threshold, and the final water inflow forecast data is output.

[0013] As one of the preferred solutions, the acquiring of real-time meteorological data of the target area on the user side and the building of a river meteorological prediction model according to the water inflow prediction data and the real-time meteorological data include: Acquire real-time meteorological data of the target area on the user side, and process the water inflow prediction data and the real-time meteorological data through a median filtering algorithm to obtain an input data set with a unified format; Based on the input data set, a river meteorological prediction model is constructed through a long short-term memory network; A sliding time window is set to update the input data set, and the river meteorology prediction model is trained with the updated input data set to obtain the trained river meteorology prediction model.

[0014] As one of the preferred solutions, the basic air conditioning configuration data includes air conditioning energy efficiency ratio; The step of inputting the meteorological forecast data into the air conditioning operation load forecasting model for processing, outputting the air conditioning operation load forecasting data, and combining the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset period of time includes: Inputting the meteorological forecast data into the air conditioning operation load forecasting model for forecasting, and outputting the air conditioning operation load forecasting data; quantifying the air conditioning power load of the building space to which the corresponding air conditioner belongs during the future preset period of time according to the air conditioning operation load prediction data and the air conditioning energy efficiency ratio of the corresponding air conditioner; The air-conditioning power load of each building space in the future preset period is weighted to obtain the air-conditioning power load of the user-side target area in the future preset period.

[0015] A second aspect of the present invention provides an air conditioning power load prediction system based on meteorological factors, which is applied to an air conditioner on the user side, the air conditioner being powered by a hydropower station, and the system comprising: A data acquisition module is used to acquire historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data; The first model building module is used to use the historical meteorological data, the building configuration data, the air conditioner historical operation mode and the air conditioner historical operation temperature as training data to train the recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the user-side target area; A second model building module is used to determine the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjust the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; The third model building module is used to build a river meteorological prediction model on the power supply side of the hydropower station, and obtain the meteorological prediction data of the user-side target area in the future preset time period output by the river meteorological prediction model; The power load prediction module is used to input the meteorological forecast data into the air conditioning operation load prediction model for processing, output the air conditioning operation load prediction data, and combine the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset time period.

[0016] Compared with the prior art, the embodiments of the present invention have the following advantages: (1) The model constructed by comprehensively considering multiple data sources (historical meteorology, building configuration, air conditioning operation, etc.) can more accurately reflect the actual situation of the target area on the user side, thereby improving the prediction accuracy of air conditioning power load; (2) The acquired data is trained through a recurrent neural network (RNN) to obtain a building load prediction model, and the model is adjusted using the key influencing factors of air conditioning to further obtain an air conditioning operation load prediction model. The training and optimization process ensures that the model can accurately reflect the actual load changes; (3) Construct a river meteorological prediction model on the power supply side where the hydropower station is located, and use its output meteorological prediction data as input. By considering the power supply side factors, the prediction results are closer to the actual power supply conditions, thus improving the practicality and accuracy of the prediction. (4) Combined with the basic air conditioning configuration data, the air conditioning power load of the target area on the user side in the future preset period can be accurately quantified to provide strong support for the power supply dispatching and air conditioning energy consumption management of the hydropower station; (5) The present invention is particularly suitable for predicting the power load of air conditioners on the user side of a power supply system supplied by a hydropower station. It has positive significance for improving energy utilization efficiency and optimizing power grid dispatching. It also provides a reference for load prediction of other types of energy supply systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the implementation mode will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 This is a flow chart of a method for predicting air conditioning power load based on meteorological factors provided by a certain embodiment of the present invention; Figure 2 This is a structural diagram of an air conditioning power load prediction system based on meteorological factors provided by a certain embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. 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.

[0020] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0021] In the description of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two components. The terms "vertical", "horizontal", "left", "right", "upper", "lower" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the system or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0022] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0023] Based on this, in one embodiment, if Figure 1 As shown, the first aspect of the present invention provides a method for predicting air conditioning power load based on meteorological factors, which is applied to an air conditioner on the user side, and the air conditioner is powered by a hydropower station. The method includes: S1. Obtain historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; Among them, meteorological data includes temperature, humidity, etc.; the basic air conditioning configuration data includes air conditioning energy efficiency ratio, installation density, installation location, power, air conditioning model, etc.; the building configuration data includes building volume, building wall area and building wall heat transfer coefficient, etc.; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data. In addition, the above configuration data and historical data can be collected by data interface, database query, on-site measurement, etc., and the collected data is cleaned, and after removing duplicate, erroneous or invalid data, the data from different sources are integrated into a unified database or data warehouse for subsequent analysis.

[0024] In addition, the user-side target area is not specifically limited in the present invention, and can be a smaller range, such as the downstream of the river where the hydropower station is located, that is, a shopping mall or a user's home on the user side. The user-side area can also be divided according to the preset area division rules to obtain several user-side target areas. After selecting a user-side target area and using the method described in the present invention to predict the air conditioning power load of the area, the user-side target area is updated, and the air conditioning power load of the updated user-side target area is predicted again until the air conditioning power load prediction results of all user-side target areas are obtained, and the prediction results are compared with the actual load results, and the output plan of the hydropower station at the power supply end is planned and scheduled according to the comparison results to make full use of resources. The model constructed by comprehensively considering multiple data sources (historical meteorology, building configuration, air conditioning operation, etc.) can more accurately reflect the actual situation of the user-side target area, thereby improving the prediction accuracy of the air conditioning power load.

[0025] S2. Using the historical meteorological data, the building configuration data, the historical operation mode of the air conditioner, and the historical operation temperature of the air conditioner as training data to train a recurrent neural network, and obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the target area on the user side; In one embodiment, step S2 includes: quantifying the building heat dissipation when each building space in the user-side target area reaches the corresponding optimal air conditioning operation temperature under different air conditioning operation modes according to the building configuration data, the air conditioning historical operation mode and the air conditioning historical operation temperature; The recurrent neural network is trained using the training data constructed by the historical meteorological data and the building heat dissipation to obtain a trained building load prediction model.

[0026] Specifically, the present invention establishes a building load prediction model through a data-driven method, and performs modeling and prediction based on the actually collected data, thereby avoiding the limitations of the traditional method that relies on empirical formulas and assumptions and improving the accuracy of the prediction; it can comprehensively consider the building's own characteristics, air-conditioning operation modes and external environmental factors, so as to more comprehensively reflect the changing laws of the building load; with the help of the ability of the recurrent neural network to process time series data, in order to capture the dynamic characteristics of the building load changing over time, the constructed building load prediction model has strong adaptability to load changes in different time scales and different seasons, and can also accurately predict future building loads based on current meteorological conditions and building heat dissipation, providing strong support for the optimized operation and energy-saving management of the air-conditioning system.

[0027] In one embodiment, quantifying the building heat dissipation when each building space in the user-side target area reaches the corresponding optimal air conditioning operation temperature under different air conditioning operation modes according to the building configuration data, the air conditioning historical operation mode and the air conditioning historical operation temperature includes: quantifying, based on the building volume, a first heat dissipation of each building space in the user-side target area when the corresponding air-conditioning operating optimal temperature is reached under different air-conditioning historical operating modes; Based on the building wall area and the building wall heat transfer coefficient, quantify the second heat dissipation of each building space in the user-side target area when the corresponding air-conditioning operation optimal temperature is reached under different air-conditioning historical operation modes; The third heat dissipation of each building space in the target area on the user side when the corresponding air conditioner reaches the optimal operating temperature under different air conditioner historical operating modes is quantified by means of a pre-constructed heat dissipation impact model; the heat dissipation impact model is used to characterize the functional relationship between the heat dissipation in the corresponding building space and its related key influencing factors; The building heat dissipation of the user-side target area is quantified based on the first heat dissipation, the second heat dissipation, and the third heat dissipation.

[0028] Specifically, during the operation of the air conditioner, there are different operation modes depending on the external environment, including at least a cooling mode and a heating mode. The building heat dissipation when each building space reaches the optimal temperature under different air conditioner historical operation modes (the optimal temperature here can be obtained from the air conditioner historical operation temperature) can also be understood as the heating capacity to be provided by the air conditioner when the building space reaches the optimal operating temperature of the air conditioner. Due to different air conditioner operation modes, the heat dissipation can also be positive or negative.

[0029] The first heat dissipation is expressed by the following formula: Q1=(T i -T b )*V*C a , where T b =T b h (Heating mode) or T b c (Cooling mode), Q1 is the first heat dissipation when each building space reaches the corresponding optimal temperature of air conditioning operation under different air conditioning historical operation modes, T i is the indoor temperature of the building space, V is the building volume, C a is the specific heat capacity of air; The second heat dissipation is expressed by the following formula: Q2= (T o -T b )*S i *C i, where Q2 is the second heat dissipation when each building space reaches the corresponding optimal temperature of air conditioning operation under different air conditioning historical operation modes. It can also be considered as the heat dissipation calculated based on the building wall area, T o is the outdoor temperature, i.e. the temperature in the historical meteorological data, S i is the building wall area, C i is the heat transfer coefficient of the building wall, i=1, 2, ..., n, n is the number of walls; The third heat dissipation is expressed by the following formula: Q3 = f(P, H, K, T i , T o ), where Q3 is the heat dissipation caused by other factors, also known as the third heat dissipation, f is the functional relationship between the heat dissipation in the corresponding building space and its related key influencing factors, by collecting relevant historical data and combining machine learning technology to fit the functional relationship between the heat dissipation of the building space and its related key influencing factors, so as to improve the accuracy of the heat dissipation value caused by additional influencing factors, P is the number of people in the building space, H is the air humidity, that is, the humidity in the historical meteorological data, K is the ventilation coefficient of the building space, and the heat dissipation influencing factors may also include climate conditions, building materials, building layout, etc., which will not be elaborated here; Then the building heat dissipation in the corresponding building space when the corresponding air-conditioning optimal operating temperature is reached under different air-conditioning historical operating modes, which is also the heating amount that the air-conditioning needs to provide when the building space is in different air-conditioning historical operating modes to reach the optimal operating temperature of the air-conditioning, can be expressed as Q=Q1+Q2+Q3; the building heat dissipation of each building space in the target area on the user side is added up to obtain the building heat dissipation of the target area on the user side.

[0030] The present invention can more comprehensively reflect multiple aspects of building heat dissipation by comprehensively considering the first heat dissipation (based on the building volume and the historical operation mode of the air conditioner), the second heat dissipation (based on the building wall area and the heat transfer coefficient) and the third heat dissipation (based on the heat dissipation influence model), thereby improving the accuracy of heat dissipation quantification. This multi-dimensional and multi-factor consideration method is helpful to capture the heat dissipation characteristics of different building spaces under different conditions, so that the quantification results are closer to the actual situation; it can flexibly respond to different historical operation modes and temperature conditions of the air conditioner, and by adjusting the parameters and models in the quantification process, it can adapt to the heat dissipation calculation under different buildings, different environments and different needs, thereby enhancing the adaptability and practicality of the solution; through quantitative analysis, it can scientifically reveal the changing law of building heat dissipation, provide strong data support for building energy-saving design, air-conditioning system operation optimization, etc., and add building and meteorological factors in the process of air-conditioning power load prediction, so as to make the prediction results more accurate.

[0031] In one embodiment, the training data constructed by the historical meteorological data and the building heat dissipation is used to train the recurrent neural network to obtain a trained building load prediction model, including: Decomposing the historical meteorological data into trend items, seasonal items and random items by time series decomposition method, modeling each item separately, and constructing a building load characteristic matrix in combination with the building heat dissipation; Processing the building load characteristic matrix by a sliding time window method to obtain training input sequence data; Based on the training input sequence data, the building load prediction model is constructed through a long short-term memory network, and the historical meteorological data is used as an additional feature to train the building load prediction model to obtain a trained building load prediction model.

[0032] Specifically, the present application considers the influence of factors such as season and type, decomposes historical meteorological data into trend items, seasonal items and random items through the time series decomposition method, models each item separately, and constructs a building load characteristic matrix in combination with the heat dissipation of the building; then, the building load characteristic matrix is ​​processed using the sliding time window method, and the window size is set to 24 hours and the step size is 1 hour to generate training input sequence data; then, based on the training input sequence data, a long short-term memory network algorithm is used to construct a regional building load prediction model, the number of hidden layer neurons is set to 128, the dropout rate is 0.2, and historical meteorological data is used as an additional feature. The model parameters are optimized through the back propagation algorithm, and the cross-validation method is used to evaluate the model performance. The 5-fold cross validation is selected, and the mean absolute percentage error (MAPE) and the root mean square error (RMSE) are calculated. The model parameters are dynamically adjusted according to the verification results to obtain a trained building load prediction model.

[0033] The present invention decomposes historical meteorological data for easy understanding and modeling, and can capture its long-term trends, seasonal fluctuations and random changes respectively, thereby improving the accuracy of predictions; and uses the LSTM feature that can effectively handle long-term dependency problems in time series data to construct a building load prediction model, so as to further improve the prediction model's sensitivity to building load changes and prediction accuracy.

[0034] S3, determining the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjusting the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; In one embodiment, determining the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner includes: Standardizing the air conditioner basic configuration data, the air conditioner historical fault data, and the air conditioner historical maintenance data to obtain an impact feature sequence with a unified format; The correlation between the influencing feature sequence and the corresponding building heat dissipation is calculated by the Pearson correlation coefficient method to generate a correlation coefficient matrix; Based on the correlation coefficient matrix, the influencing feature sequence is reduced in dimension and screened by principal component analysis, and the influencing feature sequence whose cumulative contribution rate reaches the contribution threshold is selected as a candidate factor; A multiple linear regression model is established with the candidate factors as independent variables and the building heat dissipation as dependent variable, and the multiple linear regression model is screened by a stepwise regression method to obtain the key influencing factors.

[0035] Among them, the historical fault data of the air conditioner includes the fault type (such as compressor fault, condenser fault, etc.), fault occurrence time, fault frequency, etc., and the historical maintenance data of the air conditioner includes maintenance time, maintenance content (such as replacement of parts, cleaning of filters, etc.), maintenance costs, etc.; the present invention first cleans the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, removes data with too many duplicate, erroneous or missing values, and standardizes the data to ensure that data from different sources have consistent formats and units, and obtains the influencing feature sequence; then the Pearson correlation coefficient method is used to calculate the correlation strength between the influencing feature sequence and the building heat dissipation corresponding to the data, and a correlation coefficient matrix is ​​generated, and according to the correlation coefficient matrix, the principal component analysis method is used to reduce the dimension and screen the influencing feature sequence, and the features with a cumulative contribution rate of 85% are selected as candidate factors, such as the equipment model. , rated power, energy efficiency ratio, and installation location; in addition, if the seasonal influence is considered, the above data can also be seasonally decomposed to analyze the correlation strength of the four seasons of spring, summer, autumn, and winter respectively, that is, the time series decomposition method is used to decompose the influencing characteristic sequence into three components: trend, seasonality, and residual, and the relationship between each component and its corresponding building heat dissipation is analyzed respectively, and the mutual information method is used to calculate the nonlinear correlation between the discretized influencing characteristic sequence and its corresponding building heat dissipation, to supplement the deficiencies of linear correlation analysis, and to establish a multivariate linear regression model with candidate factors as independent variables and their corresponding building heat dissipation as dependent variables, to quantitatively analyze the influence of different candidate factors on the building heat dissipation in each building space in the target area on the user side; and then the multivariate linear regression model is screened through the stepwise regression method, and the variables that contribute significantly to the model are taken as key influencing factors.

[0036] The present invention eliminates the dimensional differences between different data sources by standardizing the data, thereby improving the accuracy and reliability of data analysis; the Pearson correlation coefficient method is used to calculate the correlation relationship, which can objectively and accurately reflect the linear correlation between each influencing feature sequence and the heat dissipation of the building, providing a basis for subsequent feature screening; the principal component analysis method reduces the number of features by dimensionality reduction, while retaining the main information of the original data, improving the calculation efficiency and prediction performance of the model, setting the contribution threshold to further screen the candidate factors, and ensuring the representativeness and importance of the selected features; in addition, the multivariate linear regression model can intuitively reflect the influence of each influencing factor on the heat dissipation of the building, and gradually introduces and eliminates variables through the stepwise regression method, optimizes the model structure, and improves the prediction accuracy and interpretability of the model; the established model can quickly identify key influencing factors based on actual data to characterize the influence on the change of building load, provide a scientific basis for the optimal design, fault prediction and energy-saving management of the air-conditioning system, and has strong practical value.

[0037] In one embodiment, the building load prediction model is adjusted by the key influencing factors to obtain the air conditioning operation load prediction model of the user-side target area, including: Initializing a particle swarm and using the mean square error of the building load prediction model as a fitness function; wherein the position of each particle in the particle swarm is formed by encoding the initial value of the model parameter of the building load prediction model and the key influencing factor; Based on the fitness function, the positions and velocities of the particles in the particle swarm are updated by using a standard particle swarm algorithm to obtain optimal model parameters; Based on the optimal model parameters, the positions and velocities of the particles in the particle swarm are updated by a cooperative particle swarm algorithm to obtain updated optimal model parameters, and the updating steps of the cooperative particle swarm algorithm are repeated until a preset number of iterations is reached and the final optimal model parameters are output; The building load prediction model is updated based on the final optimal model parameters to obtain the air conditioning operation load prediction model of the user-side target area.

[0038] Specifically, the particle swarm algorithm simulates the process of bird flocks foraging. There are multiple potential solutions at the same time and they cooperate with each other. Each potential solution is called a particle. Each particle represents a potential solution. Each particle has a position, which also represents the encoding result of the parameter value and the key influencing factor. When the key influencing factors include outdoor temperature and building area, the position of the particle can be expressed as: Particle position = [model parameter 1, model parameter 2, ..., outdoor temperature coefficient, building area coefficient, ...]; Similarly, each particle also has a speed that represents the direction and magnitude of the updated encoding result. The particle speed update formula can be expressed as: Speed ​​update = [original speed update item] + [outdoor temperature impact item] + [building area impact item]; Among them, the original speed update term is the standard speed update term in the particle swarm algorithm, while the outdoor temperature influence term and the building area influence term are calculated based on the key influencing factors. The outdoor temperature influence term v_temp is defined according to the difference ΔT between the outdoor temperature and the preset benchmark temperature, which can be expressed as v_temp=k1*ΔT; the building area influence term v_area is defined according to the difference A between the building area and the preset benchmark area, which can be expressed as v_area=k2*A; and the influence coefficients k1 and k2 can be determined by experiments, experience or optimization methods to reflect the actual impact of outdoor temperature and building area on the optimization of model parameters.

[0039] These particles fly and search in the decision space of the problem in order to find the optimal position. Each particle continuously adjusts its position based on its own and other particles' search experience during the flight. That is, the standard particle swarm algorithm updates the speed and position vector of each particle as a whole. Then, the mean square error of the building load prediction model is used as the fitness function to evaluate the quality of each particle, and the position and speed of the particles in the particle swarm are updated through the standard particle swarm algorithm. In this process, each particle will remember the best position it has found so far (that is, the individual optimal solution), and the entire particle swarm will also remember the best position found so far (that is, the global optimal solution). Based on the above speed update formula, the particle's speed will be updated according to the individual optimal and global optimal positions, and the particle's position will be updated according to its speed to obtain the optimal model parameters. Then, based on the optimal model parameters, the positions of particles in the particle swarm are iteratively updated through the cooperative particle swarm algorithm. By dividing the particle swarm into several subgroups, each subgroup can search independently, and different subgroups can exchange the best position information in each iteration cycle, and each component of each particle is regarded as a separate individual. Taking the optimal gbest of the entire population as the benchmark, each component of each particle is substituted into gbest for update. If the best position obtained by substituting a component of a particle into gbest is better than the best position obtained by substituting the corresponding component of pbest of this particle into gbest, then this component is used to update the corresponding component of pbest. If the best position obtained by substituting a certain component of a particle into gbest is better than the best position obtained by retaining the corresponding component of gbest, then this component is directly used to update the corresponding component of gbest. The individual optimum and the global optimum are updated in the above manner. At the end of each iteration cycle, the information between subgroups is updated according to the set cooperation mechanism, that is, the best position information is updated. Repeat the update process of the particle position by the cooperative particle swarm algorithm until the preset number of iterations is reached and the global optimal position is output as the final optimal model parameter. Finally, the building load prediction model is updated by the final optimal model parameter to obtain the air conditioning operation load prediction model of the target area on the user side.

[0040] The present invention encodes key influencing factors and model parameters to form particle positions, thereby enhancing the adaptability of the model and being able to flexibly respond to building load prediction needs in different scenarios; by adopting a combination of standard particle swarm algorithm (PSO) and cooperative particle swarm algorithm (CPSO) to search for model parameters of the building load prediction model, the robustness of the search process is enhanced through information sharing and collaboration between particles, which helps to avoid falling into local optimal solutions and can quickly find optimal or approximately optimal weight parameters; the optimized building load prediction model can more accurately reflect the air-conditioning operating load conditions in the target area on the user side, and provide strong support for the energy-saving operation and scheduling of the air-conditioning system; it is easy to implement and has good scalability, and parameters such as the particle swarm scale and the number of iterations can be adjusted according to actual needs to adapt to building load prediction problems of different scales and complexities.

[0041] S4, constructing a river meteorological prediction model for the power supply side of the hydropower station, and obtaining meteorological prediction data for the user-side target area in a future preset period of time output by the river meteorological prediction model; In one embodiment, the step of constructing a river meteorological prediction model on the power supply side of the hydropower station includes: Acquire real-time water level data of the power supply side of the hydropower station, and predict water inflow forecast data of the power supply side of the hydropower station in the future preset time period according to the real-time water level data; Real-time meteorological data of the user-side target area is acquired, and a river meteorological prediction model is constructed according to the inflow prediction data and the real-time meteorological data.

[0042] Usually, as an electrical equipment, the power load of air conditioners is often inseparable from the energy dispatch on the power supply side. The change of water inflow on the power supply side of the hydropower station will cause the fluctuation of the environmental weather in the target area on the downstream user side, such as: when the upstream water inflow increases, the temperature and humidity of the downstream river will decrease; conversely, when the upstream water inflow decreases, the temperature and humidity of the downstream river will increase. This phenomenon will have a certain impact on the building load on the user side of the downstream river, and then affect the air conditioning power load. Therefore, when predicting the air conditioning power load of the target area on the user side, the present invention considers the impact of the river water inflow factor on the user side of the power supply side of the hydropower station and the impact of the building environment where the air conditioner is located, so that the prediction result is closer to the actual power supply conditions, so as to improve the practicality and accuracy of the air conditioning load prediction.

[0043] In one embodiment, the acquiring of real-time water level data of the power supply side of the hydropower station, and predicting water inflow forecast data of the power supply side of the hydropower station in the future preset period of time according to the real-time water level data, includes: Acquire real-time water level data collected by multiple hydrological stations located on the power supply side of the hydropower station, and query corresponding flow values ​​from a pre-established water level and flow relationship database based on the real-time water level data; The queried flow value is fitted by an autoregressive integral moving average model to obtain the water inflow forecast data of the power supply side of the hydropower station in the future preset period; Acquire historical water inflow data of the power supply side of the hydropower station, and construct a meteorological regression model based on the historical water inflow data and the historical meteorological data; Inputting the water inflow forecast data into the meteorological regression model for forecasting, and outputting the meteorological change data of the user-side target area in the future preset period; When the meteorological change data exceeds the change threshold, the real-time water level data of the power supply side of the hydropower station is reacquired to repeat the calculation steps of the water inflow forecast data of the power supply side of the hydropower station in the future preset time period, until the recalculated meteorological change data does not exceed the change threshold, and the final water inflow forecast data is output.

[0044] Specifically, the present application constructs a water level and flow relationship database by collecting historical hydrological data and its corresponding flow values, and obtains real-time water level data by setting up multiple hydrological stations on the power supply side of the hydropower station. The flow corresponding to the water level is queried from the water level and flow relationship database, and the flow data of each station is summarized to obtain the water inflow forecast data of the power supply side of the hydropower station in the future preset time period. At the same time, the moving average method is used to remove outliers and noise from the summarized water inflow forecast data; then the ARIMA model, that is, the autoregressive integrated moving average model, is used to perform time series analysis on the processed water inflow forecast data to predict the upstream water inflow in the future preset time period, that is, the water inflow forecast data, where ARIMA The optimal values ​​of the model parameters can be determined by the grid search method; then, a meteorological regression model is established based on historical data to characterize the correlation between the water inflow on the power supply side and the environmental meteorology of the target area on the user side in the same period, namely, a temperature regression model, a humidity regression model, etc., and the predicted water inflow data is used to predict the meteorological change trend of the target area on the user side in a preset future period through the meteorological regression model, and it is compared with the change threshold. When it exceeds the threshold, the real-time water level data is updated to optimize the ARIMA model, and the water inflow forecast data is recalculated using the optimized ARIMA model until the meteorological change data does not exceed the change threshold, and the final water inflow forecast data is output.

[0045] The present application combines real-time water level data with a pre-established database of the relationship between water level and flow, and can quickly and accurately obtain the flow value on the power supply side of the hydropower station. By fitting the flow value using the ARIMA model, the water inflow situation in a preset period of time in the future can be effectively predicted with high accuracy. The meteorological regression model constructed based on historical water inflow data and historical meteorological data of the target area on the user side can accurately predict the meteorological changes in the target area on the user side in a preset period of time in the future. By setting a meteorological change threshold and comparing it with the predicted meteorological changes, the dynamic adjustment mechanism of re-acquiring real-time water level data and optimizing the autoregressive integral sliding average model is triggered according to the comparison result to ensure the timeliness and accuracy of the prediction results, and can cope with changes in emergencies or extreme weather conditions, thereby improving the robustness and adaptability of the prediction system.

[0046] In one embodiment, the acquiring of real-time meteorological data of the target area on the user side and constructing a river meteorological prediction model according to the water inflow prediction data and the real-time meteorological data includes: Acquire real-time meteorological data of the target area on the user side, and process the water inflow prediction data and the real-time meteorological data through a median filtering algorithm to obtain an input data set with a unified format; Based on the input data set, a river meteorological prediction model is constructed through a long short-term memory network; A sliding time window is set to update the input data set, and the river meteorology prediction model is trained with the updated input data set to obtain the trained river meteorology prediction model.

[0047] Specifically, multiple representative cross-section locations are selected in the downstream river corresponding to the target area on the user side. According to the river terrain and hydrological characteristics, various types of equipment such as temperature sensors and humidity sensors are installed in a layered layout method to monitor the meteorological data of the target area on the user side in real time. The sampling frequency is set to once every 5 minutes. The median filtering algorithm is used to predict the water inflow on the upstream power supply side, and the multi-source heterogeneous data formed by combining historical hydrological data and meteorological data, as well as real-time meteorological data and real-time water level data of the same period are pre-processed, such as removing outliers, filling missing values ​​and standardizing, to obtain an input data set with a unified format; then, the data is used to predict the water inflow on the upstream power supply side, and the historical hydrological data and meteorological data are combined, and the real-time meteorological data and real-time water level data of the same period are pre-processed. A river meteorology prediction model was constructed using a long short-term memory (LSTM) network. The input data set was divided into a training set and a validation set in a ratio of 8:2. The network parameters of the river meteorology prediction model were adjusted using the Adam optimizer. The learning rate was set to 0.001, the batch size was set to 64, and the number of iterations was set to 1000 to improve the generalization ability and prediction accuracy of the model. Finally, the sliding time window method was used, with 24 hours as a prediction cycle, and the input data set of the model was updated every 1 hour to train the river meteorology prediction model until the model accuracy reached the preset accuracy value, thus obtaining the trained river meteorology prediction model.

[0048] This application uses the characteristics of long short-term memory networks that can capture long-term dependencies in data and effectively deal with the complexity and nonlinear characteristics of time series data to build a river meteorological forecasting model to more accurately predict the meteorological changes in the user-side target area within a preset period of time in the future; by setting a sliding time window to update the input data set, it can ensure that the forecasting model is always based on the latest data for forecasting, making the forecast results closer to the actual situation, improving the timeliness and accuracy of the forecast, and enabling the model to adapt to changes in external factors such as seasonal changes and climate fluctuations.

[0049] S5, inputting the meteorological forecast data into the air conditioning operation load forecasting model for processing, outputting the air conditioning operation load forecasting data, and combining the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset period; In one embodiment, step S5 includes: Inputting the meteorological forecast data into the air conditioning operation load forecasting model for forecasting, and outputting the air conditioning operation load forecasting data; quantifying the air conditioning power load of the building space to which the corresponding air conditioner belongs during the future preset period of time according to the air conditioning operation load prediction data and the air conditioning energy efficiency ratio of the corresponding air conditioner; The air-conditioning power load of each building space in the future preset period is weighted to obtain the air-conditioning power load of the user-side target area in the future preset period.

[0050] Specifically, the meteorological forecast data is input into the air-conditioning operation load prediction model for prediction, so as to obtain the air-conditioning operation load prediction data of the building space to which the corresponding air-conditioner belongs in the future preset time period, and the air-conditioning operation load prediction data is multiplied by the inverse of the air-conditioning energy efficiency of the corresponding air-conditioner to obtain the air-conditioning power load of the building space to which the corresponding air-conditioner belongs in the future preset time period. The air-conditioning power load of each building space in the future preset time period is weighted according to the installation density, importance or priority order, so as to obtain the air-conditioning power load of the target area on the user side in the future preset time period.

[0051] By inputting meteorological forecast data into the air conditioning operation load forecasting model, this application can take into account the direct impact of meteorological factors on the air conditioning operation load, thereby improving the accuracy of the forecast; based on the air conditioning operation load forecast data and the energy efficiency ratio of the corresponding air conditioner, the air conditioning power load of each building space in the future preset period can be quantified, which helps to identify the peak and low-peak periods of energy consumption, provide data support for energy management, realize on-demand energy allocation, and reduce unnecessary energy consumption; by weighting the air conditioning power load of each building space, the overall air conditioning power load of the user-side target area in the future preset period can be obtained. This data is of great significance to regional energy planning and power grid dispatching, and helps to achieve efficient use of energy and sustainable development.

[0052] In another embodiment, the method further comprises: Generate a power generation plan according to the operating status of the units of the hydropower station, the grid dispatching requirements and the air conditioning power load, and quantify the actual air conditioning power load according to the power generation plan; When the deviation between the actual air-conditioning power load and the air-conditioning power load exceeds the deviation threshold, the air-conditioning operation load prediction model is rebuilt, and the air-conditioning power load calculation and power generation plan generation steps are repeated until the deviation does not exceed the deviation threshold, and the air-conditioning power load and the corresponding power generation plan obtained after the final update are output.

[0053] The generating plan is generated according to the operating status of the units of the hydropower station, the grid dispatching requirements and the air conditioning power load, and the actual air conditioning power load is quantified according to the generating plan, including: Obtaining the unit operation status and grid dispatching requirements of the hydropower station, and building a comprehensive data matrix in combination with the air conditioning power load; Based on the comprehensive data matrix, a power generation plan optimization model is constructed by a particle swarm optimization algorithm, and an optimal power generation plan that meets load demand and operation constraints is obtained by iteratively calculating the power generation plan optimization model; Generate power generation instructions to execute the optimal power generation plan, obtain the actual air conditioning power load and combine it with the real-time unit status, and dynamically adjust the optimal power generation plan through a predictive control algorithm to optimize the optimal power generation plan.

[0054] Specifically, the present invention uses the SSL encrypted transmission protocol to transmit the air conditioning power load to the hydropower station power generation dispatching system in real time, and at the same time obtains the hydropower station unit operation status data and power grid dispatching demand information through the SCADA system, and builds a comprehensive data matrix including load forecasting, unit status and dispatching demand in combination with the air conditioning power load; Then, the sliding time window method is used to process the comprehensive data matrix. The window size is set to 24 hours and the step length is 1 hour. The input sequence for optimization is generated. The hydrological factors such as reservoir water level and water inflow are combined as constraints. Based on the generated input sequence for optimization, the particle swarm optimization algorithm is used to build the hydropower station power generation plan optimization model. The number of particles is set to 100, the inertia weight is 0.8, the individual learning factor is 2.0, and the social learning factor is 2.0. The advantages and disadvantages of different power generation schemes are evaluated through the objective function, and the optimal power generation plan that meets the load demand and operation constraints is obtained through iterative optimization. Finally, the model predictive control algorithm is used to dynamically adjust the optimal power generation plan. The prediction time domain is set to 4 hours and the control time domain is set to 1 hour. According to the real-time changes in air-conditioning power load and unit status, the power generation instructions are updated every 15 minutes to achieve real-time optimization and execution of the power generation plan.

[0055] This application constructs a comprehensive data matrix through the operating status of the units of the hydropower station and the grid dispatching requirements, combined with the air-conditioning power load, which not only includes the prediction results of the power load, but also takes into account the actual operating conditions of the hydropower station and the overall needs of the power grid, providing a solid foundation for the subsequent optimization of the power generation plan; with the help of the global optimization algorithm, the power generation plan optimization model is constructed to find the optimal solution under complex constraints, ensuring that the formulation of the power generation plan not only meets the load requirements of the power grid, but also complies with the operating constraints of the hydropower station units, such as water level, flow, unit efficiency, etc., thereby improving the power generation efficiency and economic benefits; the predictive control algorithm is used to fine-tune the power generation plan in real time according to the actual operating status of the system and changes in the external environment, ensuring the flexibility and adaptability of the power generation plan and improving the stability and reliability of the power system; by optimizing the power generation plan and real-time dynamic adjustment, the power generation capacity of the hydropower station is maximized and energy waste is reduced. At the same time, adjusting the power generation plan according to the real-time needs of the power grid helps to achieve a balance between energy supply and demand and improve energy utilization efficiency.

[0056] The method of iteratively calculating the power generation plan optimization model to obtain an optimal power generation plan that meets load requirements and operation constraints includes: Obtaining historical unit output and historical power generation cost corresponding to the load demand, and establishing a nonlinear relationship function through the historical unit output and the historical power generation cost; According to the upper and lower limits of the unit output, the power generation efficiency curve, the reservoir water level and storage capacity curve, the power balance requirements of the power grid and the air conditioning power load, a set of operation constraints is constructed, wherein the set of operation constraints includes unit operation constraints, water balance constraints and power balance constraints; Solving the power generation plan optimization model by a branch and bound algorithm to obtain a power generation plan, and updating the power generation plan by a rolling optimization strategy; The power generation plan optimization model is iteratively updated through the updated power generation plan, and the power generation plan solution steps are repeated through the updated power generation plan optimization model until an iteration threshold is reached, and the final power generation plan is output as the optimal power generation plan.

[0057] Specifically, the present application first establishes a nonlinear relationship model between power generation cost and unit output based on the historical operation data and unit parameters of the hydropower station, taking into account maintenance cost, start-up and shutdown cost and water resource value. For example, the power generation cost of a 100MW unit at 50% load is 0.2 yuan / kWh, which drops to 0.15 yuan / kWh at 90% load. Then, the piecewise linearization method is used to divide the output interval into 5 sub-intervals, and each sub-interval is fitted with a linear function to fit the cost curve, and the maximum fitting error is controlled within 2%; Then, we collect data such as the upper and lower limits of unit output, power generation efficiency curves, and reservoir water level and storage capacity curves, and build a set of operating constraints based on the power balance requirements of the power grid and the results of air conditioning power load. Then, the branch and bound method is used to solve the power generation plan optimization model, the branch strategy is set to the maximum score rule, and the delimitation strategy adopts depth-first search. Finally, in the iterative solution process, the calculation is stopped when the difference between the upper and lower bounds is less than 1% or the number of iterations reaches 1000 times. The 24-hour rolling optimization strategy is used in the calculation, and the model input data is updated every hour, including load forecast values, hydrological forecasts and unit status; and the parallel computing technology is used to solve the sub-problems of multiple time periods at the same time, and a global optimization is completed within 10 minutes. Finally, the power generation plan of the hydropower station that meets various constraints is obtained. After the iteration is completed, the optimal power generation plan including the unit output per hour, the reservoir water level change and the amount of abandoned water is output. In addition, 100 groups of load and hydrological scenarios can be generated by the Monte Carlo simulation method, the robustness of the optimization results can be evaluated, and the cost saving rate and load tracking accuracy under the 95% confidence interval can be calculated. It should be noted that the various preset values, thresholds, etc. described in the present invention are all determined by actual conditions, and no specific numerical limits are made here.

[0058] This application is based on a nonlinear relationship function of historical data, which can more accurately reflect the complex relationship between unit output and power generation cost, and provide a scientific basis for the formulation of power generation plans; by constructing a comprehensive set of operating constraints, it ensures that the formulation of power generation plans not only conforms to the physical characteristics of the units, but also meets the operating requirements of the power grid, thereby ensuring the feasibility and safety of the power generation plan; the branch and bound algorithm is used to solve the power generation plan optimization model, which helps to formulate high-quality power generation plans in a shorter time and improve decision-making efficiency. The power generation plan is continuously updated through a rolling optimization strategy, which can cope with the uncertainty and dynamic changes in the power system, so that the power generation plan can be adjusted in time with changes in actual conditions, thereby improving the adaptability and flexibility of the power generation plan.

[0059] In the embodiment of the present application, based on the problem of how to improve the accuracy of the prediction results of the air-conditioning power load, a method for predicting the air-conditioning power load based on meteorological factors is designed, which is applied to the air-conditioning on the user side, and the air-conditioning is powered by a hydropower station, so as to obtain the historical meteorological data, air-conditioning basic configuration data, building configuration data and air-conditioning historical operation data of the target area on the user side; the air-conditioning historical operation data includes the air-conditioning historical operation mode, the air-conditioning historical operation temperature, the air-conditioning historical fault data and the air-conditioning historical maintenance data; the historical meteorological data, the building configuration data, the air-conditioning historical operation mode and the air-conditioning historical operation temperature are used as training data to train the recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the target area on the user side; the key influencing factors of the air-conditioning are determined according to the air-conditioning basic configuration data, the air-conditioning historical fault data and the air-conditioning historical maintenance data , and adjust the building load prediction model through the key influencing factors to obtain the air conditioning operation load prediction model of the user-side target area; construct a river meteorological prediction model on the power supply side of the hydropower station, and obtain the meteorological prediction data of the user-side target area in the future preset time period output by the river meteorological prediction model; input the meteorological prediction data into the air conditioning operation load prediction model for processing, output the air conditioning operation load prediction data, and combine the air conditioning basic configuration data to quantify the technical solution of the air conditioning power load of the user-side target area in the future preset time period; the model constructed by comprehensively considering multiple data sources (historical meteorology, building configuration, air conditioning operation, etc.) can more accurately reflect the actual situation of the user-side target area, thereby improving the prediction accuracy of the air conditioning power load, and accurate load prediction helps hydropower stations and power grid operators to make energy scheduling and allocation plans in advance, optimize energy utilization efficiency, and reduce energy waste.

[0060] 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.

[0061] In another embodiment, if Figure 2 As shown, the second aspect of the present invention provides an air conditioning power load prediction system based on meteorological factors, which is applied to the air conditioning on the user side, and the air conditioning is powered by a hydropower station. The system includes: The data acquisition module 10 is used to acquire historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data; The first model building module 20 is used to use the historical meteorological data, the building configuration data, the air conditioner historical operation mode and the air conditioner historical operation temperature as training data to train the recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the user-side target area; The second model building module 30 is used to determine the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjust the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; The third model building module 40 is used to build a river meteorological prediction model for the power supply side of the hydropower station, and obtain the meteorological prediction data of the user-side target area in the future preset period output by the river meteorological prediction model; The power load prediction module 50 is used to input the meteorological forecast data into the air conditioning operation load prediction model for processing, output the air conditioning operation load prediction data, and combine the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset time period.

[0062] It should be noted that each module in the above-mentioned air-conditioning power load prediction system based on meteorological factors can be fully or partially implemented by 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 of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules. For the specific definition of an air-conditioning power load prediction system based on meteorological factors, please refer to the definition of an air-conditioning power load prediction method based on meteorological factors above. The two have the same functions and effects, which will not be repeated here.

[0063] In summary, the present invention relates to the field of information technology, and discloses a method and system for predicting air conditioning power load based on meteorological factors. The method is a method for predicting air conditioning power load on the user side, and is particularly suitable for scenarios powered by hydropower stations. It comprehensively uses historical meteorological data of the target area on the user side, basic air conditioning configuration data, building configuration data, and historical air conditioning operation data, combined with recurrent neural network technology, to construct and optimize a building load prediction model, and uses key influencing factors of air conditioning to adjust the model, thereby obtaining an air conditioning operation load prediction model; at the same time, the river meteorological forecast on the power supply side of the hydropower station is considered, and the meteorological forecast data output by it is used as input, combined with the basic air conditioning configuration data, to accurately quantify the air conditioning power load of the target area on the user side in a preset period of time in the future, which solves the problem of inaccurate prediction results of the air conditioning power load by the currently used prediction model, and provides strong support for power supply scheduling and air conditioning energy consumption management of hydropower stations.

[0064] 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.

[0065] 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 air conditioning power load based on meteorological factors, characterized in that: Applied to an air conditioner on the user side, the air conditioner is powered by a hydropower station, and the method comprises: Acquire historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data; The historical meteorological data, the building configuration data, the historical operation mode of the air conditioner and the historical operation temperature of the air conditioner are used as training data to train a recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the target area on the user side; Determine the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjust the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; Constructing a river meteorological prediction model for the power supply side of the hydropower station, and obtaining meteorological prediction data for the target area on the user side in a preset period of time in the future output by the river meteorological prediction model; The meteorological forecast data is input into the air conditioning operation load prediction model for processing, and the air conditioning operation load prediction data is output. In combination with the air conditioning basic configuration data, the air conditioning power load of the user-side target area in the future preset period is quantified.

2. The method for predicting air conditioning power load based on meteorological factors according to claim 1, characterized in that: The historical meteorological data, the building configuration data, the historical air conditioner operation mode and the historical air conditioner operation temperature are used as training data to train a recurrent neural network to obtain a trained building load prediction model, including: quantifying the building heat dissipation when each building space in the user-side target area reaches the corresponding optimal air conditioning operation temperature under different air conditioning operation modes according to the building configuration data, the air conditioning historical operation mode and the air conditioning historical operation temperature; The recurrent neural network is trained using the training data constructed by the historical meteorological data and the building heat dissipation to obtain a trained building load prediction model.

3. The method for predicting air conditioning power load based on meteorological factors according to claim 2, characterized in that: The training data constructed by the historical meteorological data and the building heat dissipation is used to train the recurrent neural network to obtain a trained building load prediction model, including: Decomposing the historical meteorological data into trend items, seasonal items and random items by time series decomposition method, modeling each item separately, and constructing a building load characteristic matrix in combination with the building heat dissipation; Processing the building load characteristic matrix by a sliding time window method to obtain training input sequence data; Based on the training input sequence data, the building load prediction model is constructed through a long short-term memory network, and the historical meteorological data is used as an additional feature to train the building load prediction model to obtain a trained building load prediction model.

4. The method for predicting air conditioning power load based on meteorological factors according to claim 1, characterized in that: The determining of the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner comprises: Standardizing the air conditioner basic configuration data, the air conditioner historical fault data, and the air conditioner historical maintenance data to obtain an impact feature sequence with a unified format; The correlation between the influencing feature sequence and the corresponding building heat dissipation is calculated by the Pearson correlation coefficient method to generate a correlation coefficient matrix; Based on the correlation coefficient matrix, the influencing feature sequence is reduced in dimension and screened by principal component analysis, and the influencing feature sequence whose cumulative contribution rate reaches the contribution threshold is selected as a candidate factor; A multiple linear regression model is established with the candidate factors as independent variables and the building heat dissipation as dependent variable, and the multiple linear regression model is screened by a stepwise regression method to obtain the key influencing factors.

5. The method for predicting air conditioning power load based on meteorological factors according to claim 1, characterized in that: The step of adjusting the building load prediction model by using the key influencing factors to obtain the air conditioning operation load prediction model of the user-side target area includes: Initializing a particle swarm, and using the mean square error of the building load prediction model as a fitness function; wherein the position of each particle in the particle swarm is formed by encoding the initial value of the model parameter of the building load prediction model and the key influencing factor; Based on the fitness function, the positions and velocities of the particles in the particle swarm are updated by using a standard particle swarm algorithm to obtain optimal model parameters; Based on the optimal model parameters, the positions and velocities of the particles in the particle swarm are updated by a cooperative particle swarm algorithm to obtain updated optimal model parameters, and the updating steps of the cooperative particle swarm algorithm are repeated until a preset number of iterations is reached and the final optimal model parameters are output; The building load prediction model is updated based on the final optimal model parameters to obtain the air conditioning operation load prediction model of the user-side target area.

6. The method for predicting air conditioning power load based on meteorological factors according to claim 1, characterized in that: The construction of the river meteorological prediction model on the power supply side of the hydropower station includes: Acquire real-time water level data of the power supply side of the hydropower station, and predict water inflow forecast data of the power supply side of the hydropower station in the future preset time period according to the real-time water level data; Real-time meteorological data of the user-side target area is acquired, and a river meteorological prediction model is constructed according to the inflow prediction data and the real-time meteorological data.

7. The method for predicting air conditioning power load based on meteorological factors according to claim 6, characterized in that: The step of acquiring the real-time water level data of the power supply side of the hydropower station, and predicting the water inflow prediction data of the power supply side of the hydropower station in the future preset time period according to the real-time water level data, includes: Acquire real-time water level data collected by multiple hydrological stations located on the power supply side of the hydropower station, and query corresponding flow values ​​from a pre-established water level and flow relationship database based on the real-time water level data; The queried flow value is fitted by an autoregressive integral moving average model to obtain the water inflow forecast data of the power supply side of the hydropower station in the future preset period; Acquire historical water inflow data of the power supply side of the hydropower station, and construct a meteorological regression model based on the historical water inflow data and the historical meteorological data; Inputting the water inflow forecast data into the meteorological regression model for forecasting, and outputting the meteorological change data of the user-side target area in the future preset period; When the meteorological change data exceeds the change threshold, the real-time water level data of the power supply side of the hydropower station is reacquired to repeat the calculation steps of the water inflow forecast data of the power supply side of the hydropower station in the future preset time period, until the recalculated meteorological change data does not exceed the change threshold, and the final water inflow forecast data is output.

8. The method for predicting air conditioning power load based on meteorological factors according to claim 6, characterized in that: The acquiring of real-time meteorological data of the target area on the user side and constructing a river meteorological prediction model according to the water inflow prediction data and the real-time meteorological data includes: Acquire real-time meteorological data of the target area on the user side, and process the water inflow prediction data and the real-time meteorological data through a median filtering algorithm to obtain an input data set with a unified format; Based on the input data set, a river meteorological prediction model is constructed through a long short-term memory network; A sliding time window is set to update the input data set, and the river meteorology prediction model is trained with the updated input data set to obtain the trained river meteorology prediction model.

9. The method for predicting air conditioning power load based on meteorological factors according to claim 1, characterized in that: The basic configuration data of the air conditioner includes the energy efficiency ratio of the air conditioner; The step of inputting the meteorological forecast data into the air conditioning operation load forecasting model for processing, outputting the air conditioning operation load forecasting data, and combining the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset period of time includes: Inputting the meteorological forecast data into the air conditioning operation load forecasting model for forecasting, and outputting the air conditioning operation load forecasting data; quantifying the air conditioning power load of the building space to which the corresponding air conditioner belongs during the future preset period of time according to the air conditioning operation load prediction data and the air conditioning energy efficiency ratio of the corresponding air conditioner; The air-conditioning power load of each building space in the future preset period is weighted to obtain the air-conditioning power load of the user-side target area in the future preset period.

10. An air conditioning power load prediction system based on meteorological factors, characterized in that: Applied to the air conditioner on the user side, the air conditioner is powered by a hydropower station, and the system includes: A data acquisition module is used to acquire historical meteorological data, air conditioning basic configuration data, building configuration data and air conditioning historical operation data of the target area on the user side; the air conditioning historical operation data includes air conditioning historical operation mode, air conditioning historical operation temperature, air conditioning historical fault data and air conditioning historical maintenance data; The first model building module is used to use the historical meteorological data, the building configuration data, the air conditioner historical operation mode and the air conditioner historical operation temperature as training data to train the recurrent neural network to obtain a trained building load prediction model; the building load prediction model reflects the building heat dissipation in the user-side target area; A second model building module is used to determine the key influencing factors of the air conditioner according to the basic configuration data of the air conditioner, the historical fault data of the air conditioner and the historical maintenance data of the air conditioner, and adjust the building load prediction model by the key influencing factors to obtain the air conditioner operation load prediction model of the target area on the user side; The third model building module is used to build a river meteorological prediction model on the power supply side of the hydropower station, and obtain the meteorological prediction data of the user-side target area in the future preset time period output by the river meteorological prediction model; The power load prediction module is used to input the meteorological forecast data into the air conditioning operation load prediction model for processing, output the air conditioning operation load prediction data, and combine the air conditioning basic configuration data to quantify the air conditioning power load of the user-side target area in the future preset time period.

Citation Information

Patent Citations

  • Intelligent control scheduling method for electric load of air conditioner based on big data analysis

    CN110864414A

  • Building electrical load baseline generation method and system considering weather factors

    CN115204946A

  • Load prediction method and device for building air conditioning system and electronic equipment

    CN116227725A

  • Air conditioner electricity consumption prediction method and device and power grid load prediction method and device

    CN116857798A

  • Air conditioner load prediction method based on similar day and long short-term memory network

    CN118427526A

Cited By

  • Building energy consumption prediction method and system based on multi-factor decomposition, terminal and medium

    CN119294616A

  • Regional power distribution load scheduling method and system

    CN119314065A

  • Energy scheduling method and system

    CN119918841A

  • Load prediction method for high-proportion new energy access power distribution network

    CN120165373A

  • Method and device for providing enterprise resource demand reference information in supply chain business, equipment, storage medium and program product

    CN120952228A