CN-122026383-A - Power grid and air conditioner load collaborative optimization scheduling method and system based on edge calculation

CN122026383ACN 122026383 ACN122026383 ACN 122026383ACN-122026383-A

Abstract

The invention discloses a grid and air conditioner load collaborative optimization scheduling method and system based on edge calculation, which relate to the technical field of intelligent grid and air conditioner load management, wherein related parameters are acquired in real time through edge calculation nodes, and the related parameters are preprocessed; the method comprises the steps of carrying out key feature extraction on the preprocessed related parameters, carrying out air conditioner energy consumption prediction by combining the key features through a preset energy consumption prediction model to obtain a prediction result, inputting the key features and the prediction result into a preset multi-target optimization model, and introducing linear inertia weight and linear learning factors to obtain a result of the multi-target optimization model as a cooperative scheduling strategy by combining an improved particle swarm algorithm. The invention combines the edge computing technology to realize the complex multi-objective optimization scheduling of the air conditioner load.

Inventors

SHI RUIJIE
LUO CONG
SHI HONGMING
WANG SHENG
LUO PENG
GONG ZHENG
LIN WUXING
OUYANG LILIN
LEI HONGWEI
HU SHI
LI CHUANG

Assignees

国网电商科技有限公司
远光能源互联网产业发展(横琴)有限公司

Dates

Publication Date: 20260512
Application Date: 20260204

Claims (10)

1. The grid and air conditioner load collaborative optimization scheduling method based on edge calculation is characterized by comprising the following steps of: collecting related parameters in real time through an edge computing node, and preprocessing the related parameters; Extracting key features from the preprocessed related parameters to obtain key features; Air conditioner energy consumption prediction is carried out by combining the key features through a preset energy consumption prediction model, and a prediction result is obtained; Inputting the key features and the prediction results into a preset multi-target optimization model, and introducing linear inertia weights and linear learning factors to obtain the results of the multi-target optimization model as a cooperative scheduling strategy by combining an improved particle swarm algorithm.
2. The grid and air conditioner load collaborative optimization scheduling method based on edge calculation according to claim 1, wherein the obtaining key features comprises extracting a grid running state, an air conditioner load characteristic and a user behavior characteristic by using a machine learning and data mining algorithm: Extracting power grid running state characteristics, modeling by adopting an autoregressive integral moving average model by using time sequence data of power grid parameters: Wherein, the Time of presentation Is used for the control of the power grid parameters, And The parameters of the model are respectively the parameters of the model, The method is characterized in that the method is used for determining an error term, c is a constant term and represents the intercept of a model to reflect the long-term average value of a time sequence, p is the order of autoregressive and is required to be determined through model selection, q is the order of moving average and is required to be determined through model selection; the method comprises the steps of extracting air conditioner load characteristics, predicting and extracting features by using historical electricity data and environment temperature data of air conditioner loads through a long-term and short-term memory network: ; Wherein, the Time of presentation Is used to determine the hidden state of the (c), The input characteristics are represented as such, And The weight and the bias are respectively given to the weight and the bias, Is an activation function; extracting user behavior characteristics, classifying by adopting a K-means clustering algorithm by using historical electricity consumption data and daily behavior data: ; Wherein, the As the number of clusters to be clustered, Is the first The number of clusters is one, Is the first The centroid of the clusters.
3. The grid and air conditioner load collaborative optimization scheduling method based on edge calculation according to claim 1, wherein the performing air conditioner energy consumption prediction specifically comprises performing energy consumption prediction by using Catboost learning method: ; Wherein, the Finding the optimal h (x) for the objective function, i.e. to minimize the loss function; Is that A strong learner obtained by iteration; the expected symbol E is usually required to be added before the loss function; In the iterative calculation process, describing the negative gradient of the iterative loss function generated by each iteration, the formula is as follows: ; In the formula, Representing the negative gradient of the sample (x, y) at the t-th iteration, wherein y is the real label of the sample, and s is the predicted value of the model on the sample x; The obtaining step The formula is as follows: 。
4. the grid and air conditioner load collaborative optimization scheduling method based on edge calculation according to claim 1, wherein the multi-objective optimization model specifically comprises: Grid stability objective function: ; Air conditioning energy efficiency objective function: ; user comfort objective function: ; Wherein, the In order to set the comfort temperature of the patient, Is the first The indoor temperature at the moment of time, Is the first The energy consumption of the air conditioning system at the moment, Is the first The grid load at the moment in time, The average load of the power grid is represented by T, and the time range is represented by T; The objective function of the multi-objective optimization model is: ; Wherein, the 、、 Respectively the weight coefficients of the stability of the power grid, the energy efficiency of the air conditioner and the comfort level of the user.
5. The grid and air conditioner load collaborative optimization scheduling method based on edge calculation according to claim 4, wherein the improved particle swarm algorithm comprises introducing linear inertia weight and linear learning factor into a basic particle swarm algorithm, and the formula is: ; ; Wherein, the 、 The current iteration times and the total iteration times are respectively; 、、、 The upper limit and the lower limit of the inertia weight and the learning factor are respectively; Replacing the individual optimum value of each particle with the arithmetic average value of the individual optimum values of the particle group, and excluding the individual optimum values which can cause local optimum in the group, wherein the formula is as follows: ; Wherein, the Is that The average value of the individual optimum values at the moment, For the number of particles to be the same, Is that Individual optimum values for each particle at a time; the speed update expression of the improved particle swarm algorithm is as follows: ; Wherein, the Is a particle In the first place The speed at which the iteration is performed, Is a particle In the first place The position at the time of the iteration, Is that The average value of the individual optimum values at the moment, As a result of the linear inertial weight, Is a particle In the first place The optimal position of the population at the time of the iteration, , The learning factors of the particle algorithm are respectively calculated, , The random numbers in [0,1], respectively.
6. An edge calculation-based grid and air conditioner load collaborative optimization scheduling system is characterized by comprising: the data processing module is used for acquiring related parameters in real time through the edge computing node and preprocessing the related parameters; The key feature extraction module is used for extracting key features of the preprocessed related parameters to obtain key features; The energy consumption prediction module is used for performing air conditioner energy consumption prediction by combining the key characteristics through a preset energy consumption prediction model to obtain a prediction result; and the strategy optimization module inputs the key features and the prediction results into a preset multi-target optimization model, and introduces linear inertia weight and linear learning factors to obtain the result of the multi-target optimization model as a collaborative scheduling strategy by combining an improved particle swarm algorithm.
7. The grid and air conditioner load collaborative optimization scheduling system based on edge computing according to claim 6, wherein the key feature extraction module is configured to extract a grid operation state, an air conditioner load characteristic and a user behavior characteristic by using a machine learning and data mining algorithm: Extracting power grid running state characteristics, modeling by adopting an autoregressive integral moving average model by using time sequence data of power grid parameters: Wherein, the Time of presentation Is used for the control of the power grid parameters, And The parameters of the model are respectively the parameters of the model, The method is characterized in that the method is used for determining an error term, c is a constant term and represents the intercept of a model to reflect the long-term average value of a time sequence, p is the order of autoregressive and is required to be determined through model selection, q is the order of moving average and is required to be determined through model selection; the method comprises the steps of extracting air conditioner load characteristics, predicting and extracting features by using historical electricity data and environment temperature data of air conditioner loads through a long-term and short-term memory network: ; Wherein, the Time of presentation Is used to determine the hidden state of the (c), The input characteristics are represented as such, And The weight and the bias are respectively given to the weight and the bias, Is an activation function; extracting user behavior characteristics, classifying by adopting a K-means clustering algorithm by using historical electricity consumption data and daily behavior data: ; Wherein, the As the number of clusters to be clustered, Is the first The number of clusters is one, Is the first The centroid of the clusters.
8. The grid and air conditioner load collaborative optimization scheduling system based on edge calculation according to claim 1, wherein the energy consumption prediction module is used for performing air conditioner energy consumption prediction specifically comprises the following steps of performing energy consumption prediction by using Catboost learning method: ; Wherein, the Finding the optimal h (x) for the objective function, i.e. to minimize the loss function; Is that A strong learner obtained by iteration; the expected symbol E is usually required to be added before the loss function; In the iterative calculation process, describing the negative gradient of the iterative loss function generated by each iteration, the formula is as follows: ; In the formula, Representing the negative gradient of the sample (x, y) at the t-th iteration, wherein y is the real label of the sample, and s is the predicted value of the model on the sample x; The obtaining step The formula is as follows: 。
9. the grid and air conditioner load collaborative optimization scheduling system based on edge calculation according to claim 1, wherein the strategy optimization module comprises the following specific multi-objective optimization model: Grid stability objective function: ; Air conditioning energy efficiency objective function: ; user comfort objective function: ; Wherein, the In order to set the comfort temperature of the patient, Is the first The indoor temperature at the moment of time, Is the first The energy consumption of the air conditioning system at the moment, Is the first The grid load at the moment in time, The average load of the power grid is represented by T, and the time range is represented by T; The objective function of the multi-objective optimization model is: ; Wherein, the 、、 Respectively the weight coefficients of the stability of the power grid, the energy efficiency of the air conditioner and the comfort level of the user.
10. The grid and air conditioner load collaborative optimization scheduling system based on edge calculation according to claim 9, wherein in the policy optimization module, the improved particle swarm algorithm comprises introducing linear inertia weight and linear learning factor into a basic particle swarm algorithm, and the formula is: ; ; Wherein, the 、 The current iteration times and the total iteration times are respectively; 、、、 The upper limit and the lower limit of the inertia weight and the learning factor are respectively; Replacing the individual optimum value of each particle with the arithmetic average value of the individual optimum values of the particle group, and excluding the individual optimum values which can cause local optimum in the group, wherein the formula is as follows: ; Wherein, the Is that The average value of the individual optimum values at the moment, For the number of particles to be the same, Is that Individual optimum values for each particle at a time; the speed update expression of the improved particle swarm algorithm is as follows: ; Wherein, the Is a particle In the first place The speed at which the iteration is performed, Is a particle In the first place The position at the time of the iteration, Is that The average value of the individual optimum values at the moment, As a result of the linear inertial weight, Is a particle In the first place The optimal position of the population at the time of the iteration, , The learning factors of the particle algorithm are respectively calculated, , The random numbers in [0,1], respectively.

Description

Power grid and air conditioner load collaborative optimization scheduling method and system based on edge calculation Technical Field The invention relates to the technical field of intelligent power grid and air conditioner load management, in particular to a power grid and air conditioner load collaborative optimization scheduling method and system based on edge calculation. Background At present, with the development of modern society and the acceleration of urban progress, the power demand has a rapidly growing trend. Particularly in hot summer, air conditioning load becomes one of the main burdens of the power system, resulting in grid load peaks and power supply shortage. The intelligent power grid utilizes advanced information and communication technology, and realizes real-time monitoring of the power grid, data acquisition and intelligent management through devices such as a sensor, an intelligent ammeter, an automatic control system and the like. And the demand response guides a user to adjust electricity consumption behavior through an excitation mechanism, so that the load pressure of the power grid is relieved. However, most of the existing smart power grids and demand response systems rely on a centralized cloud computing platform for data processing and decision making, and the centralized processing mode has the problems of large data transmission delay, slow response speed, poor data security and the like, so that the real-time and reliability requirements are difficult to meet. The edge computing is used as an emerging computing architecture, and data processing and computing are carried out on edge nodes close to a data source, so that data transmission delay can be remarkably reduced, and the response speed and the data processing efficiency of the system can be improved. By combining the edge calculation, the intelligent power grid and the demand response technology, the respective advantages can be fully exerted, more efficient, real-time and reliable power grid and air conditioner load scheduling is realized, intelligent and efficient operation of pushing energy management is realized, and although the application of the edge calculation in the intelligent power grid and the demand response has wide prospects, the prior art still faces a plurality of key challenges of how to efficiently perform data preprocessing and feature extraction, how to construct a more accurate prediction model, how to realize complex multi-objective optimal scheduling while meeting the real-time performance, how to ensure the safety and privacy protection of data and the like. Therefore, how to implement efficient and optimal scheduling of air conditioning loads in combination with edge computing technology is a problem that needs to be solved by those skilled in the art. Disclosure of Invention In view of the above, the invention provides a grid and air conditioner load collaborative optimization scheduling method and system based on edge calculation, which are used for solving the problems in the background technology. In order to achieve the above purpose, the present invention adopts the following technical scheme: a power grid and air conditioner load collaborative optimization scheduling method based on edge calculation comprises the following steps: collecting related parameters in real time through an edge computing node, and preprocessing the related parameters; Extracting key features from the preprocessed related parameters to obtain key features; Air conditioner energy consumption prediction is carried out by combining the key features through a preset energy consumption prediction model, and a prediction result is obtained; Inputting the key features and the prediction results into a preset multi-target optimization model, and introducing linear inertia weights and linear learning factors to obtain the results of the multi-target optimization model as a cooperative scheduling strategy by combining an improved particle swarm algorithm. Preferably, the acquiring key features includes extracting power grid operation state, air conditioner load characteristics and user behavior features by using machine learning and data mining algorithms: Extracting power grid running state characteristics, modeling by adopting an autoregressive integral moving average model by using time sequence data of power grid parameters: Wherein, the Time of presentationIs used for the control of the power grid parameters,AndThe parameters of the model are respectively the parameters of the model,The method comprises the steps of obtaining a model, wherein the model is a constant term, c represents the intercept of the model and reflects the long-term average value of a time sequence, p is the order of autoregressive and is required to be determined through model selection, q is the order of moving average and is required to be determined through model selection; the method comprises the steps of extracting air conditioner l