Air conditioner load real-time monitoring and intelligent regulation and control system

Abstract

The invention provides an air conditioner load real-time monitoring and intelligent regulation and control system which comprises a water chilling unit located in a control room, fan coils located in all areas of the room, a temperature sensor, a people flow detector, a power detector and an illumination intensity detector and further comprises computing equipment, and a prediction model is operated in the computing equipment. The calculation module is used for calculating first correlation among the temperature, the heat production data and the valve opening in the time dimension; extracting data in the time dimension according to the first correlation, and calculating the first correlation among the temperature, the heat production data and the valve opening in the space dimension; the extraction module extracts data in a spatial dimension according to the first correlation, and is used for performing feature extraction on the extracted data to obtain feature data; and the regression module is used for predicting the feature data to obtain the valve opening at the next moment. The data is analyzed from the dimensions of time and space, prediction analysis is carried out by considering various data, and the prediction accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a real-time monitoring and intelligent control system for air-conditioning load. Background Art

[0002] With the expansion of building scale and people's increasing requirements for living environment, air conditioning systems have become an indispensable infrastructure for most public buildings and residences. The high energy consumption of air conditioning systems has also become increasingly prominent, becoming one of the main components of building energy consumption.

[0003] Traditional air conditioning control methods often perform simple on / off control or timing control based on fixed set values, lacking the ability to dynamically respond to environmental changes, resulting in energy waste and poor indoor environmental comfort. With the development of artificial intelligence, by collecting environmental data of the target space, analyzing and predicting the environmental data of the target space, determining the air conditioning load size, and controlling the air conditioning operation according to the predicted air conditioning load size, the air conditioning system can be operated on demand.

[0004] However, when analyzing environmental data, only analyzing from the time dimension results in a single data dimension and inaccurate prediction results. Summary of the invention

[0005] The present application provides a real-time monitoring and intelligent control system for air conditioning load. By analyzing and predicting data from two dimensions, time and space, in addition to temperature and heat production data, the valve opening of the fan coil unit is also taken into account for prediction, thereby improving the accuracy of the prediction results.

[0006] In a first aspect, the present application provides a real-time monitoring and intelligent control system for air conditioning load, wherein the room is divided into multiple areas, and the system comprises: a chiller located in a control room and fan coils located in each area of ​​the room, wherein the chiller provides chilled water to each fan coil;

[0007] The system also includes temperature sensors located in each area, a flow rate detector in each area, a power detector in each area, and a light intensity detector in each area;

[0008] The system also includes a computing device, which is used to obtain the temperature of each area at multiple times, the heat generation data of each area at multiple times, and the valve opening of each fan coil at multiple times;

[0009] The prediction model is run in the computing device, and the prediction model includes a computing module, an extraction module, and a regression module;

[0010] The calculation module is used to calculate the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment; extract data from the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment according to the first correlation to obtain first extracted data; and calculate the second correlation between the temperature of each area, the heat generation data of each area, and the valve opening of each area, and extract data from the temperature of each area, the heat generation data of each area, and the valve opening of each area according to the second correlation to obtain second extracted data;

[0011] The extraction module is used to perform feature extraction on the first extracted data to obtain first feature data; and perform feature extraction on the second extracted data to obtain second feature data; and fuse the first feature data and the second feature data to obtain a fused feature;

[0012] The regression module is used to predict the fused features to obtain the valve opening at the next moment.

[0013] In the above technical solution, the real-time monitoring and intelligent control system of air conditioning load includes temperature sensors, human flow detectors, power detectors and light intensity detectors located in each area, so as to obtain the temperature, human flow, power consumption and light intensity of each area at multiple times, calculate the heat generation data of each area at multiple times based on human flow, power consumption and light intensity, calculate the correlation between the temperature, heat generation data and valve opening at each time, and select corresponding data from the temperature, heat generation data and valve opening at each time according to the correlation, so as to obtain input data from the time dimension. Calculate the correlation between the temperature, heat generation data and valve opening of each area, and select corresponding data from the temperature, heat generation data and valve opening of each area according to the correlation, so as to obtain input data from the regional dimension. After feature extraction, feature fusion is performed on the input data of two dimensions, and the fused data is input into the regression module to predict the valve opening at the next moment. On the one hand, the data is analyzed and predicted from the two dimensions of time and space. On the other hand, in addition to temperature and heat production data, the valve opening of the fan coil unit is also taken into account. By calculating the correlation between the above data, the three types of data are combined for prediction. Data are selected from the temperature, heat production data and valve opening of each area at multiple times according to the correlation, which can better retain the information in the original data and improve the accuracy of the prediction results.

[0014] In a possible implementation manner, the calculation module is further used for:

[0015] The temperatures of each region at multiple times are converted into a temperature matrix, where each row represents the temperature of a region at multiple times, and each column represents the temperature of multiple regions at one time;

[0016] The heat generation data of each region at multiple times are converted into a heat generation matrix, wherein each row of the heat generation matrix represents the heat generation data of one region at multiple times, and each column represents the heat generation data of multiple regions at one time;

[0017] The valve openings of each region at multiple times are converted into a valve opening matrix; each row in the valve opening matrix represents the valve openings of a region at multiple times, and each column represents the valve openings of multiple regions at one time;

[0018] The temperature matrix and the heat generation matrix are spliced ​​together, and the correlation between the spliced ​​matrix and the valve opening matrix is ​​calculated to obtain the first correlation and the second correlation.

[0019] In the above technical scheme, by converting the temperature, heat production data and valve opening conversion matrix of each area at multiple times, splicing the temperature matrix and the heat production matrix, calculating the correlation between the spliced ​​matrix and the valve opening matrix, and converting the data into matrix calculation, the data processing efficiency can be improved and the data processing time can be shortened.

[0020] In a possible implementation manner, the calculation module is further used for:

[0021] Randomly generate a third weight parameter and a fourth weight parameter, convert the third weight parameter into a third parameter vector, and convert the fourth weight parameter into a fourth parameter vector;

[0022] The heat generation matrix and the valve opening matrix are spliced ​​on the columns to obtain a second spliced ​​matrix; the transpose of the third parameter vector and the product of the second spliced ​​matrix are calculated to obtain a third calculation vector; the transpose of the fourth parameter vector and the product of the valve opening matrix are calculated to obtain a fourth calculation vector;

[0023] Calculate the second correlation between the third calculation vector and the fourth calculation vector. If the second correlation is greater than the first preset threshold, extract data from the second splicing matrix according to the third weight parameter; and extract data from the valve opening matrix according to the fourth weight parameter to obtain second extracted data.

[0024] In the above technical solution, when performing matrix conversion, each row represents the temperature, heat generation data or valve opening corresponding to a region, and each column represents the temperature, heat generation data and valve opening corresponding to a moment. In this way, the third weight parameter and the fourth weight parameter are randomly generated, the third weight parameter is converted into a third parameter vector, the fourth weight parameter is converted into a fourth parameter vector, the inner product of the transpose of the third parameter vector and the concatenation matrix is ​​calculated, and the inner product of the transpose of the fourth parameter vector and the valve opening matrix is ​​calculated to achieve data superposition in the regional dimension. After calculating and obtaining the correlation, data can be extracted from the temperature and heat generation data of multiple regions based on the third weight parameter, and data can be extracted from the valve opening of multiple regions based on the fourth weight parameter, so as to achieve data screening from the spatial dimension.

[0025] In one possible implementation, the computing device is further configured to:

[0026] Acquire multiple sets of candidate data, including the temperature of each area at multiple historical moments, the power consumption at multiple historical moments, the flow of people at multiple historical moments, the light intensity at multiple historical moments, the valve opening of the fan coil at multiple historical moments, and the power of the chiller at multiple historical moments;

[0027] For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirements, and determine whether the power of the chiller in the candidate data meets the minimum power requirements. If both are met, the candidate data is used as training data; the training data is used to train the prediction model.

[0028] In the above technical solution, training data is selected from the alternative data based on the constant temperature adjustment requirements and the minimum power requirements, and the training data is used to train the prediction model. While ensuring the comfort of the indoor environment, the energy consumption of the air-conditioning system is significantly reduced, achieving the goal of energy conservation and emission reduction.

[0029] In one possible implementation, the computing device is further configured to:

[0030] For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirement. If so, use the candidate data as intermediate data;

[0031] According to the total energy consumption of each group of intermediate data, they are sorted in ascending order, and the first n groups of intermediate data are selected as training data.

[0032] In the above technical solution, the constant temperature adjustment requirement is set according to the user's temperature adjustment demand, and then the alternative data that meet the constant temperature adjustment requirement is screened out, and the alternative data with the minimum energy consumption is selected from the screened alternative data as training data. In this way, the training data that meets both the constant temperature adjustment requirement and the minimum power requirement is screened. The predicted model is trained using the screened data, which significantly reduces the energy consumption of the air-conditioning system while ensuring the comfort of the indoor environment, thereby achieving the goal of energy conservation and emission reduction.

[0033] In one possible implementation, the computing device is further configured to:

[0034] In each set of candidate data, the temperature of each area at all historical moments is counted to obtain the temperature variance and temperature mean; if the temperature variance is less than the preset variance threshold and the temperature mean is within the preset temperature range, the area is determined to be the target area; if the number of target areas is greater than the preset number threshold, the candidate data is used as intermediate data.

[0035] In the above technical solution, by performing statistical analysis on the temperature data of each area, it is determined whether each area meets the constant temperature regulation requirements within the area. The areas that meet the constant temperature regulation requirements within the area are in the majority, so it is determined whether the alternative data meets the constant temperature regulation requirements.

[0036] In one possible implementation, the computing device is further configured to:

[0037] The valves of the fan coil units are controlled according to the predicted valve opening to obtain the actual temperatures of each area at multiple times. The actual temperatures of each area at all times are counted to obtain the actual temperature variance and the actual temperature mean. If the actual temperature variance is less than the preset variance threshold and the actual temperature mean is within the preset temperature range, the area is determined to be a statistical area. If the number of statistical areas is less than the preset number threshold, the prediction model continues to be trained.

[0038] In the above technical solution, after using the prediction model to predict the valve opening at the next moment, the actual temperature of each area is monitored and statistical analysis is performed on the actual temperature to determine whether the valve opening predicted by the prediction model can meet the constant temperature regulation. If it cannot be met, the model is continued to be trained to improve the prediction accuracy of the prediction model through continued training.

[0039] In a second aspect, the present application further provides a method for real-time monitoring and intelligent control of air conditioning load, characterized in that it is applied to a real-time monitoring and intelligent control system for air conditioning load, the system comprising: a chiller located in a control room and fan coils located in various areas of the room, the chiller providing chilled water to each fan coil;

[0040] The system also includes temperature sensors located in each area, a flow rate detector in each area, a power detector in each area, and a light intensity detector in each area;

[0041] The system further includes a computing device, the computing device being used to obtain the temperature of each area at multiple times, the heat generation data of each area at multiple times, and the valve opening of each fan coil at multiple times;

[0042] The prediction model is run in the computing device, and the prediction model includes a calculation module, an extraction module and a regression module;

[0043] The calculation module is used to calculate the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment; extract data from the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment according to the first correlation to obtain first extracted data; and calculate the second correlation between the temperature of each area, the heat generation data of each area, and the valve opening of each area, and extract data from the temperature of each area, the heat generation data of each area, and the valve opening of each area according to the second correlation to obtain second extracted data;

[0044] The extraction module is used to perform feature extraction on the first extracted data to obtain first feature data; and perform feature extraction on the second extracted data to obtain second feature data; and fuse the first feature data and the second feature data to obtain fused features;

[0045] The regression module is used to perform prediction processing on the fused features to obtain the valve opening at the next moment.

[0046] In a third aspect, the present application provides an electronic device, including:

[0047] processor; and,

[0048] A memory, configured to store executable instructions of the processor;

[0049] Wherein, the processor is configured to perform the method of the second aspect by executing the executable instructions.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to implement the method of the second aspect when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0052] Figure 1 A schematic diagram of the structure of a real-time monitoring and intelligent control system for air conditioning load provided in some embodiments of the present application;

[0053] Figure 2 A schematic diagram of the architecture of the prediction model provided in some embodiments of the present application;

[0054] Figure 3 A schematic diagram of the structure of a computing device provided for some embodiments of the present application.

[0055] Reference numerals:

[0056] 100. Chiller; 200. Fan coil unit; 300. Computing equipment; 410. Light intensity detector; 420. People flow detector; 430. Power detector; 440. Temperature sensor.

[0057] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0058] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0059] Traditional air conditioning control methods often perform simple on / off control or timing control based on fixed set values, lacking the ability to dynamically respond to environmental changes, resulting in energy waste and poor indoor environmental comfort. With the development of artificial intelligence, by collecting environmental data of the target space, analyzing and predicting the environmental data of the target space, determining the air conditioning load size, and controlling the air conditioning operation according to the predicted air conditioning load size, the air conditioning system can be operated on demand.

[0060] However, when analyzing environmental data, only analyzing from the time dimension results in a single data dimension and inaccurate prediction results.

[0061] The embodiment of the present application provides a real-time monitoring and intelligent control system for air-conditioning load. By analyzing and predicting data from two dimensions, time and space, in addition to temperature and heat production data, the valve opening of the fan coil unit 200 is also taken into consideration for prediction, thereby improving the accuracy of the prediction results.

[0062] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0063] Figure 1 The structural diagram of the air conditioning load real-time monitoring and intelligent control system provided in this application is as follows: Figure 1 As shown, the system includes a chiller 100 and a fan coil 200. The chiller 100 is located in the control room, the fan coil 200 is located in the room, and the chiller 100 is connected to the fan coil 200 through a pipeline. The chiller 100 provides chilled water to each fan coil 200.

[0064] In order to more accurately adjust the temperature in the room, the room is divided into multiple areas. A fan coil unit 200 is arranged in each area to adjust the temperature of the area. Each area is provided with a light intensity detector 410, which is used to detect the light intensity of each area at multiple times. A human flow detector 420 is arranged in each area, which is used to detect the human flow in each area at multiple times. The human flow detector 420 can be a detector based on machine vision. Each area is provided with a power detector 430, which is used to detect the power consumption of each area at multiple times. By arranging power detectors in the lines of each area, the power consumption of each area can be detected. Each area is arranged with a temperature sensor 440, which is used to detect the temperature of each area at multiple times.

[0065] The system further includes a computing device 300 for acquiring the temperature of each zone at multiple times, the heat generation data of each zone at multiple times, and the valve opening of each fan coil unit 200 at multiple times.

[0066] The prediction model is run in the computing device 300, such as Figure 2 As shown, the prediction model includes a calculation module, an extraction module and a regression module.

[0067] The calculation module is used to calculate the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment. According to the first correlation, data is extracted from the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment to obtain first extracted data; and, the second correlation between the temperature of each area, the heat generation data of each area, and the valve opening of each area is calculated, and data is extracted from the temperature of each area, the heat generation data of each area, and the valve opening of each area according to the second correlation to obtain second extracted data.

[0068] The extraction module is used to extract features from the first extracted data to obtain first feature data; and to extract features from the second extracted data to obtain second feature data. The first feature data and the second feature data are fused to obtain fused features. The regression module is used to perform prediction processing on the fused features to obtain the valve opening at the next moment.

[0069] More specifically, the computing device 300 is connected to the light intensity detector 410, the human flow detector 420 and the power detector 430 of each area, and obtains the human flow, light intensity and power consumption of each area at multiple times, and calculates the heat generation data of each area at multiple times according to the human flow, light intensity and power consumption of each area at multiple times. The computing device 300 is connected to the temperature sensor 440 of each area, and obtains the temperature collected by the temperature sensor 440 of each area at multiple times. The computing device 300 also obtains the valve opening of the fan coil 200 of each area at multiple times.

[0070] In the time dimension, the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment is calculated. At least one temperature at each moment is selected from the temperatures at each moment according to the first correlation. At least one heat generation data at each moment is selected from the heat generation data at each moment according to the first correlation. At least one valve opening at each moment is selected from the valve opening at each moment according to the first correlation, so as to obtain the first extracted data.

[0071] In the spatial dimension, a second correlation is calculated between the temperature of each region, the heat generation data of each region, and the valve opening of each region. The temperature of at least one region is selected from the temperatures of each region according to the second correlation. The heat generation data of at least one region is selected from the heat generation data of each region according to the second correlation. The valve opening of at least one region is selected from the valve opening of each region according to the second correlation, so as to obtain the second extracted data.

[0072] The temperature, heat generation data and valve opening extracted from the time dimension are normalized, and the normalized data are spliced ​​into a matrix. The spliced ​​matrix is ​​convolved using a multi-layer convolution layer to obtain the first feature data.

[0073] The temperature, heat generation data and valve opening extracted from the spatial dimension are normalized, and the normalized data are spliced ​​into a matrix. The spliced ​​matrix is ​​convolved using a multi-layer convolution layer to obtain the second feature data.

[0074] The first characteristic data and the second characteristic data are fused and input into a regression module, and the regression module predicts the valve opening at the next moment.

[0075] When the first feature data and the second feature data are fused, the first feature data and the second feature data are converted into a matrix representation. The fused feature matrix is ​​obtained by calculating the product of the first feature matrix, the fusion matrix and the second feature data. The fused feature matrix is ​​input into the regression module, and the regression module predicts the valve opening at the next moment. The parameters in the fusion matrix can be obtained through model training.

[0076] In the above technical solution, the real-time monitoring and intelligent control system of air conditioning load includes temperature sensors, human flow detectors, power detectors and light intensity detectors located in each area, so as to obtain the temperature, human flow, power consumption and light intensity of each area at multiple times, calculate the heat generation data of each area at multiple times based on human flow, power consumption and light intensity, calculate the correlation between the temperature, heat generation data and valve opening at each time, and select corresponding data from the temperature, heat generation data and valve opening at each time according to the correlation, so as to obtain input data from the time dimension. Calculate the correlation between the temperature, heat generation data and valve opening of each area, and select corresponding data from the temperature, heat generation data and valve opening of each area according to the correlation, so as to obtain input data from the regional dimension. After feature extraction, feature fusion is performed on the input data of two dimensions, and the fused data is input into the regression module to predict the valve opening at the next moment. On the one hand, the data is analyzed and predicted from the two dimensions of time and space. On the other hand, in addition to temperature and heat production data, the valve opening of the fan coil unit is also taken into account. By calculating the correlation between the above data, the three types of data are combined for prediction. Data are selected from the temperature, heat production data and valve opening of each area at multiple times according to the correlation, which can better retain the information in the original data and improve the accuracy of the prediction results.

[0077] Specifically, the system in the above technical solution can obtain multi-dimensional data such as temperature, human flow, power consumption and light intensity in each area in real time through temperature sensors, human flow detectors, power detectors and light intensity detectors deployed in various areas. These data provide a solid foundation for subsequent load forecasting and intelligent control. In particular, the heat generation data calculated based on human flow, power consumption and light intensity further enriches the data dimension, enabling the system to more comprehensively understand the changes in heat load in the room. Through the calculation module, the system calculates the first correlation between the temperature, heat generation data and valve opening at each moment, and the second correlation between the temperature, heat generation data and valve opening in each area. This step realizes the analysis of the correlation of data from the time dimension and the regional dimension, and extracts key information from the original data based on the correlation. This data fusion strategy not only retains the effective information in the original data, but also reduces the redundancy of the data, providing more accurate data input for subsequent feature extraction and prediction processing. The extraction module extracts features from the data extracted from the time dimension and the regional dimension, and obtains the first feature data and the second feature data. Subsequently, the system fuses the feature data of these two dimensions to obtain the fused features. This step further refines the key information in the data through feature extraction and data fusion technology, providing more comprehensive and accurate feature input for subsequent prediction processing. The regression module uses the fused feature data to obtain the valve opening at the next moment through prediction processing. This step realizes the intelligent prediction of air-conditioning load and provides a basis for the dynamic regulation of the system. By predicting the valve opening at the next moment, the system can adjust the working state of the fan coil in advance, thereby realizing accurate control of the air-conditioning load. This intelligent prediction and dynamic control strategy not only improves the response speed of the system, but also significantly improves the energy efficiency ratio of the air-conditioning system and the comfort of the indoor environment. By analyzing and predicting data from the two dimensions of time and space, and combining temperature, heat generation data and valve opening for prediction, the system can more comprehensively understand the changing law of air-conditioning load. In addition, by calculating the correlation between data and extracting key information from the original data, the system can better retain the useful information in the original data, thereby improving the accuracy of the prediction results. This highly accurate prediction result provides strong support for the intelligent regulation of the system, enabling the system to automatically adjust the operating parameters of the air-conditioning system according to real-time load changes, achieving the goals of energy saving and emission reduction and optimizing the comfort of the indoor environment.

[0078] In some embodiments, the computing module is further configured to:

[0079] The temperatures of each region at multiple moments are converted into a temperature matrix, in which each row represents the temperature of a region at multiple moments, and each column represents the temperature of multiple regions at one moment.

[0080] The heat generation data of each region at multiple times are converted into a heat generation matrix, in which each row represents the heat generation data of one region at multiple times, and each column represents the heat generation data of multiple regions at one time.

[0081] The valve openings of each region at multiple times are converted into a valve opening matrix. In the valve opening matrix, each row represents the valve openings of a region at multiple times, and each column represents the valve openings of multiple regions at one time.

[0082] The temperature matrix and the heat generation matrix are spliced ​​together, and the correlation between the spliced ​​matrix and the valve opening matrix is ​​calculated to obtain the first correlation and the second correlation.

[0083] More specifically, after obtaining the temperature of each area at multiple times, the heat generation data of each area at multiple times, and the valve opening of each area at multiple times, the above data are converted into matrices. That is, the temperature matrix, the heat generation matrix, and the valve opening matrix are obtained. Subsequently, the temperature matrix and the heat generation matrix are spliced, and the correlation between the spliced ​​matrix and the valve opening matrix is ​​calculated.

[0084] In the above technical scheme, by converting the temperature, heat production data and valve opening conversion matrix of each area at multiple times, splicing the temperature matrix and the heat production matrix, calculating the correlation between the spliced ​​matrix and the valve opening matrix, and converting the data into matrix calculation, the data processing efficiency can be improved and the data processing time can be shortened.

[0085] Specifically, the temperature, heat generation data and valve opening are converted into matrix form respectively, so that the originally scattered and disordered data become structured and ordered. This conversion not only facilitates subsequent data processing and analysis, but also enhances the readability and comprehensibility of the data, providing a solid foundation for further model construction. By converting the data into matrix form and utilizing the efficiency of matrix operations, the speed and efficiency of data processing can be significantly improved. Matrix operations are usually faster than point-by-point operations, especially when processing large-scale data sets. Therefore, this technical solution can effectively shorten the data processing time and meet the needs of real-time or quasi-real-time data analysis. By splicing the temperature matrix and the heat generation matrix, and calculating the correlation (first correlation and second correlation) between the spliced ​​matrix and the valve opening matrix, the intrinsic connection and law between the data can be deeply explored. This correlation analysis helps to understand the impact of temperature and heat generation changes in different regions and at different times on the valve opening, and provides a scientific basis for optimizing system control strategies and improving energy efficiency.

[0086] In some embodiments, the computing module is further configured to:

[0087] A first weight parameter and a second weight parameter are randomly generated, the first weight parameter is converted into a first parameter vector, and the second weight parameter is converted into a second parameter vector.

[0088] The temperature matrix and the heat generation matrix are concatenated row by row to obtain a first concatenated matrix, and the product of the first concatenated matrix and the first parameter vector is calculated to obtain a first calculation vector.

[0089] The product of the valve opening matrix and the second parameter vector is calculated to obtain a second calculation vector.

[0090] A first correlation between the first calculation vector and the second calculation vector is calculated. If the absolute value of the first correlation is greater than a first preset threshold, data is extracted from the first splicing matrix according to the first weight parameter, and data is extracted from the valve opening matrix according to the second weight parameter to obtain first extracted data.

[0091] More specifically, a first weight parameter and a second weight parameter are randomly generated, and the first weight parameter is used to process the first splicing matrix to obtain a first calculation vector, thereby converting the first splicing matrix into a vector. The second weight parameter is used to process the valve opening matrix to obtain a second calculation vector, thereby converting the valve opening matrix into a vector. Subsequently, the correlation between the two vectors is calculated to determine whether the absolute value of the correlation is greater than a first preset threshold value. If so, data is extracted from the first splicing matrix according to the first weight parameter, and data is extracted from the valve opening matrix according to the second weight parameter.

[0092] If not, the first weight parameter and the second weight parameter are adjusted to obtain a new first weight parameter and a new second weight parameter. The first concatenated matrix is ​​processed using the new first weight parameter to obtain a new first calculation vector, and the valve opening matrix is ​​processed using the new second weight parameter to obtain a new second calculation vector.

[0093] Calculate the correlation between the new first calculation vector and the new second calculation vector. For example, the Pearson correlation coefficient of the two calculation vectors can be calculated. If the absolute value of the correlation is greater than a first preset threshold, extract data from the first splicing matrix according to the new first weight parameter, and extract data from the valve opening matrix according to the new second weight parameter to obtain first extracted data.

[0094] The temperature matrix T is expressed as:

[0095]

[0096] t11 represents the temperature of the first zone at the first moment, and t1p represents the temperature of the first zone at the pth moment. tq1 represents the temperature of the qth zone at the first moment, and tqp represents the temperature of the qth zone at the pth moment. q and p are positive integers, q represents the total number of zones, and p represents the total number of moments.

[0097] The heat generation matrix H is expressed as:

[0098]

[0099] h11 represents the heat generation data of the first zone at the first moment, h1p represents the heat generation data of the first zone at the pth moment, hq1 represents the heat generation data of the qth zone at the first moment, and hqp represents the heat generation data of the qth zone at the pth moment.

[0100] The valve opening matrix V is expressed as:

[0101]

[0102] v11 represents the valve opening of the first zone at the first moment, v1p represents the valve opening of the first zone at the pth moment. vq1 represents the valve opening of the qth zone at the first moment, and vqp represents the valve opening of the qth zone at the pth moment.

[0103] The temperature matrix T and the heat generation matrix H are concatenated row by row to obtain a first concatenated matrix B1.

[0104]

[0105] Randomly generate 2p first weight parameters and convert them into the first parameter vector λ:

[0106] λ=[λ1 … λ2p] T

[0107] Randomly generate p second weight parameters and convert the p second weight parameters into the second parameter vector μ:

[0108] μ=[μ1 … μp] T

[0109] The size of the first splicing matrix B1 is q×2p, and the size of the first parameter vector λ is 2p×1. The first calculation vector α is obtained by multiplying the first splicing matrix B1 and the first parameter vector λ:

[0110]

[0111] The size of the valve opening matrix V is q×p, and the size of the second parameter vector μ is p×1. The valve opening matrix V and the second parameter vector μ are multiplied to obtain the second calculation vector β:

[0112]

[0113] Calculate the correlation between the first calculation vector α and the second calculation vector β, and determine whether the absolute value of the correlation is greater than or equal to a first preset threshold. If so, extract data from the first splicing matrix according to the first weight parameter, and extract data from the valve opening matrix according to the second weight parameter.

[0114] In the above technical solution, when performing matrix conversion, each row represents the temperature, heat generation data or valve opening corresponding to a region, and each column represents the temperature, heat generation data and valve opening corresponding to a moment. In this way, 2p first weight parameters and p second weight parameters are randomly generated, and the 2p first weight parameters are converted into first parameter vectors, and the p second weight parameters are converted into second parameter vectors. The inner product is calculated using the splicing matrix and the first parameter vector, and the inner product is calculated using the valve opening matrix and the second parameter vector to achieve data superposition in the time dimension. After calculating and obtaining the correlation, data can be extracted from the temperature and heat generation data at multiple moments based on the first weight parameter, and data can be extracted from the valve opening at multiple moments based on the second weight parameter, so as to achieve data screening from the time dimension.

[0115] Specifically, by splicing the temperature matrix and the heat generation matrix by row, and calculating the inner product with the first parameter vector, and calculating the inner product of the valve opening matrix and the second parameter vector at the same time, data superposition in the time dimension is achieved. This processing method enables the system to comprehensively consider the data at different times, so as to understand the system dynamics more comprehensively. In addition, according to the comparison result of the absolute value of the calculated first correlation and the first preset threshold, the system can intelligently extract key data from the temperature, heat generation data and valve opening at multiple times, realize data screening in the time dimension, and help to analyze the system behavior more accurately. By randomly generating the first weight parameter and the second weight parameter and converting them into a parameter vector, the system can flexibly adjust the weight parameter according to different analysis requirements and data characteristics, so as to realize customized processing of the data. This flexibility enables the system to adapt to the data analysis requirements in different scenarios, and improves the applicability and practicality of the system. By calculating the first correlation between the first calculation vector and the second calculation vector, the system can quantify the degree of correlation between the temperature, heat generation data and the valve opening. This correlation analysis not only helps to reveal the interaction relationship between the variables within the system, but also provides a scientific basis for system optimization and decision-making. At the same time, setting the first preset threshold as a screening condition further ensures the accuracy and reliability of the correlation analysis. Based on the key data screened from the time dimension, the system can more accurately evaluate the relationship between the temperature, heat production and valve opening in different areas and at different times, so as to formulate a more accurate and effective control strategy. This helps to reduce system energy consumption, improve energy efficiency, and improve the comfort of the indoor environment. By converting data into matrix form and using vector operations for calculations, the system can significantly improve the speed and efficiency of data processing. This efficient data processing method enables the system to respond to external changes more quickly, realize real-time or quasi-real-time data analysis and decision support, and help improve the overall performance of the system and user experience.

[0116] In some embodiments, the calculation module is also used to: sort the 1st to pth first weight parameters in the first parameter vector in order of size, select the first w first weight parameters, and extract the columns corresponding to the first w first weight parameters from the 1st to pth columns of the first splicing matrix as the extracted temperature. w is a positive integer, for example: w is set to 4, and p is 10. The first 4 first weight parameters ranked first are the 7th first weight parameter data, the 3rd first weight parameter data, the 4th first weight parameter data, and the 10th first weight parameter data, then the 7th column data, the 3rd column data, the 4th column data, and the 10th column data in the first splicing matrix are selected. That is, the temperature at the 3rd moment, the temperature at the 4th moment, the temperature at the 7th moment, and the temperature at the 10th moment are extracted.

[0117] Sort the p+1th weight parameter to the 2pth weight parameter in the first parameter vector in order of size, select the first x first weight parameters, and extract the columns corresponding to the first x first weight parameters from the p+1th to 2pth columns of the first splicing matrix as the extracted heat production data. x is a positive integer. For example: x is set to 3, and p is 10. The first three first weight parameters are the 18th first weight parameter data, the 15th first weight parameter, and the 17th first weight parameter, then the 18th column data, the 15th column data, and the 17th column data in the first splicing matrix are selected. That is, the extracted heat production data is the heat production data at the 5th moment, the heat production data at the 7th moment, and the heat production data at the 8th moment.

[0118] Sort each second weight parameter in the second parameter vector from large to small, select the first y second weight parameters, and extract the columns corresponding to the first y second weight parameters from the valve opening matrix as the extracted valve opening. y is a positive integer. For example: y is set to 3, and p is 10. The top 3 second weight parameters are the 4th second weight parameter, the 3rd second weight parameter, and the 9th second weight parameter, then select the 4th column data, the 3rd column data, and the 9th column data in the valve opening matrix. That is, the extracted valve opening is the valve opening at the 3rd moment, the valve opening at the 4th moment, and the valve opening at the 9th moment.

[0119] In the above technical solution, if the currently generated first weight parameter and second weight parameter make the absolute value of the correlation between the first calculation vector and the second calculation vector greater than the first preset threshold, the larger weight parameter corresponds to more information in the data, so selecting a larger weight parameter to correspond to the data output can better retain the information in the original data and improve the prediction accuracy.

[0120] Specifically, by sorting the first parameter vector and the second parameter vector in the order of the weight parameters, and selecting the first w, first x, and first y larger weight parameters, the calculation module can extract the data of the corresponding columns from the temperature matrix, the heat generation matrix, and the valve opening matrix. Since a larger weight parameter usually means that the corresponding data has a higher importance or information content in the overall data set, this extraction method can better retain the key information in the original data, which helps to improve the accuracy of subsequent data analysis and prediction. In the data set, there are often a large amount of redundant or highly correlated data, which is of no substantial help in improving the accuracy and efficiency of the analysis. By screening the data corresponding to the larger weight parameters, the calculation module can reduce data redundancy to a certain extent, so that the subsequent processing can focus more on key information, thereby improving the overall data processing efficiency. In the field of machine learning and data analysis, the quality of data directly affects the prediction performance of the model. By retaining and enhancing key information and reducing data redundancy, this technical solution helps to improve the accuracy of the prediction model built based on these data. In practical applications, this means that the system can more accurately predict the future temperature, heat generation, and valve opening trends, providing a more reliable basis for intelligent control. Since this technical solution can automatically adjust the selection of weight parameters according to the information distribution in different data sets, it has certain adaptability and robustness. This means that the system can cope with data changes under different environmental conditions and operating conditions and maintain stable performance output. In the process of data processing and analysis, the allocation of computing resources and storage resources is an important consideration. By screening key information and reducing data redundancy, this technical solution helps to optimize resource allocation, enabling the system to achieve higher processing efficiency and better performance under limited resource conditions.

[0121] In some embodiments, the computing module is further configured to:

[0122] A third weight parameter and a fourth weight parameter are randomly generated, the third weight parameter is converted into a third parameter vector, and the fourth weight parameter is converted into a fourth parameter vector.

[0123] The heat generation matrix and the valve opening matrix are spliced ​​on the columns to obtain a second spliced ​​matrix. The transpose of the third parameter vector and the product of the second spliced ​​matrix are calculated to obtain a third calculation vector. The transpose of the fourth parameter vector and the product of the valve opening matrix are calculated to obtain a fourth calculation vector.

[0124] Calculate the second correlation between the third calculation vector and the fourth calculation vector. If the second correlation is greater than the first preset threshold, extract data from the second splicing matrix according to the third weight parameter; and extract data from the valve opening matrix according to the fourth weight parameter to obtain second extracted data.

[0125] More specifically, the temperature matrix T and the heat generation matrix H are spliced ​​by columns to obtain a second spliced ​​matrix B2.

[0126]

[0127] Randomly generate 2q third weight parameters, convert the q third weight parameters into a third parameter vector θ, and the transpose of the third parameter vector is θ T :

[0128] θ T =[θ1 … θ2q]

[0129] Randomly generate q fourth weight parameters, convert the q fourth weight parameters into a fourth parameter vector γ, and the transpose of the fourth parameter vector is γ T :

[0130] γ T =[γ1 … γq]

[0131] The size of the second concatenation matrix B is 2q×p, the size of the third parameter vector θ is 1×2q, and the transpose θ of the third parameter vector is T The third calculation vector τ is obtained by calculating the product with the second concatenated matrix B2:

[0132]

[0133] The size of the valve opening matrix V is q×p, the size of the fourth parameter vector γ is 1×q, and the transpose γ of the fourth parameter vector is calculated. T The fourth calculation vector ρ is obtained by multiplying it with the valve opening matrix V:

[0134]

[0135] Calculate the correlation between the third calculation vector τ and the fourth calculation vector ρ, for example, calculate the Pearson correlation coefficient of the two calculation vectors, determine whether the absolute value of the correlation is greater than or equal to the first preset threshold, and if so, extract the heat production data from the second splicing matrix according to the third weight parameter. Extract the valve opening from the valve opening matrix according to the fourth weight parameter.

[0136] In the above technical solution, when performing matrix conversion, each row represents the temperature, heat generation data or valve opening corresponding to a region, and each column represents the temperature, heat generation data and valve opening corresponding to a moment. In this way, 2p third weight parameters and p fourth weight parameters are randomly generated, and the 2p third weight parameters are converted into third parameter vectors, and the p fourth weight parameters are converted into fourth parameter vectors. The inner product of the transpose of the third parameter vector and the concatenation matrix is ​​calculated, and the inner product of the transpose of the fourth parameter vector and the valve opening matrix is ​​calculated to achieve data superposition in the regional dimension. After calculating and obtaining the correlation, data can be extracted from the temperature and heat generation data of multiple regions based on the third weight parameter, and data can be extracted from the valve opening of multiple regions based on the fourth weight parameter, so as to achieve data screening from the spatial dimension.

[0137] Specifically, by splicing the heat generation matrix and the valve opening matrix on the column, and calculating the inner product with the transpose of the third parameter vector, and calculating the inner product of the transpose of the fourth parameter vector and the valve opening matrix, the calculation module realizes data superposition in the regional dimension. This processing method enables the system to comprehensively consider the data of different regions, so as to more comprehensively understand the operating conditions of the system in different regions. In addition, according to the comparison result of the calculated second correlation with the first preset threshold, the system can intelligently extract key data from the temperature and heat generation data of multiple regions, and extract data from the valve opening of multiple regions, so as to realize data screening in the spatial dimension, which is helpful to more accurately analyze the performance differences of the system in different regions. By introducing the third and fourth weight parameters, the calculation module can process and analyze the data in a wider dimension. This processing method not only increases the depth of data processing, so that the system can dig out deeper data associations and laws; it also broadens the breadth of data processing, so that the system can cover more regions and data points, and improves the comprehensiveness and accuracy of data processing. The method of randomly generating weight parameters and converting them into parameter vectors provides higher flexibility for data analysis. The system can flexibly adjust the settings of weight parameters according to different analysis requirements and data characteristics, thereby realizing customized processing of data. This flexibility not only helps to improve the accuracy of data analysis, but also meets the analysis requirements in different scenarios and improves the applicability and practicality of the system. By superimposing and screening data in the regional dimension, the calculation module can more accurately evaluate the operating status of the system in different areas, thereby formulating more accurate and effective control strategies. This helps to reduce system energy consumption, improve energy utilization efficiency, and improve the overall performance of the system. At the same time, since the system can focus on data in key areas for analysis and processing, it can reduce the waste of computing resources to a certain extent and improve resource utilization efficiency.

[0138] In some embodiments, the calculation module is also used to: sort the 1st third weight parameter to the qth third weight parameter in the third parameter vector in order of size, select the first s third weight parameters, and extract the rows corresponding to the first s third weight parameters from the 1st to the qth rows of the second splicing matrix as the extracted temperature. s is a positive integer. For example: s is set to 4, and q is 8. The first four third weight parameters are the 2nd third weight parameter data, the 5th third weight parameter data, the 6th third weight parameter, and the 3rd third weight parameter, then the 2nd row of data, the 3rd row of data, the 5th row of data, and the 6th row of data in the second splicing matrix are selected. That is, the extracted temperature is the temperature of the 2nd area, the temperature of the 3rd area, the temperature of the 5th area, and the temperature of the 6th area.

[0139] Sort the q+1th weight parameter to the 2qth weight parameter in the third parameter vector in order of size, select the first u third weight parameters, and select the rows corresponding to the u third weight parameters from the q+1th to 2qth rows of the second splicing matrix as the extracted heat generation data. u is a positive integer. For example: u is set to 4, and q is 8. The first three third weight parameters are the 1st third weight parameter data, the 4th third weight parameter data, and the 6th third weight parameter, then select the 1st row of data, the 4th row of data, and the 6th row of data in the second splicing matrix. That is, the extracted temperatures are the temperature of the 1st area, the temperature of the 4th area, and the temperature of the 6th area.

[0140] Sort each fourth weight parameter in the fourth parameter vector in descending order, select the first z fourth weight parameters, and extract the rows corresponding to the first z fourth weight parameters from the valve opening matrix as the extracted valve opening. z is a positive integer. For example: z is set to 3, and q is 8. The first three fourth weight parameters are the third fourth weight parameter data, the second fourth weight parameter data, and the fourth fourth weight parameter, then select the second row data, the third row data, and the fourth row data in the valve opening matrix. That is, the extracted valve opening is the valve opening of the second area, the valve opening of the third area, and the valve opening of the fourth area.

[0141] In the above technical solution, if the currently generated third weight parameter and fourth weight parameter make the absolute value of the correlation between the third calculation vector and the fourth calculation vector greater than the first preset threshold, the larger weight parameter corresponds to more information in the data, so selecting a larger weight parameter to correspond to the data output can better retain the information in the original data and improve the prediction accuracy.

[0142] In some embodiments, the computing module is further configured to:

[0143] Obtain the illuminated area of ​​each area, the power consumption of each area at multiple times, the light intensity of each area at multiple times, and the flow of people in each area at multiple times.

[0144] In each region and at each moment, the product of the electric power, the first heat conversion coefficient and the unit time is calculated to obtain the first heat production of each region at multiple moments. More specifically, the electric power of the i-th region at the j-th moment is obtained, the electric power of the i-th region at the j-th moment, the product of the first heat conversion coefficient and the unit time is calculated to obtain the first heat production of the i-th region at the j-th moment, and i is allowed to traverse q and j is allowed to traverse p to obtain the first heat production of each region at multiple moments.

[0145] In each area and at each moment, the product of the illumination intensity, illumination area, second heat conversion coefficient and unit time is calculated to obtain the second heat production of each area at multiple moments. More specifically, the illumination intensity of the i-th area at the j-th moment and the illumination area of ​​the i-th area are obtained, the power consumption, illumination area, second heat conversion coefficient and unit time of the i-th area at the j-th moment are calculated, the second heat production of the i-th area at the j-th moment is obtained, and i is traversed through q and j is traversed through p to obtain the second heat production of each area at multiple moments.

[0146] In each area and at each moment, the product of the flow of people, unit heat production and unit time is calculated to obtain the third heat production of each area at multiple moments. More specifically, the flow of people in the i-th area at the j-th moment is obtained, the product of the flow of people, unit heat production and unit time in the i-th area at the j-th moment is calculated, the third heat production of the i-th area at the j-th moment is obtained, and i is traversed through q and j is traversed through p to obtain the third heat production of each area at multiple moments.

[0147] For each time of each area, the sum of the first heat generation, the second heat generation and the third heat generation of the area at a time is calculated to obtain the heat generation data of the area at a time. More specifically, the sum of the first heat generation, the second heat generation and the third heat generation of the i-th area at the j-th time is calculated to obtain the heat generation data of the i-th area at the j-th time. Let i traverse q and j traverse p to obtain the heat generation data of each area at multiple times.

[0148] In the above technical solution, by converting the collected light intensity, pedestrian flow and electricity power in each area into heat production, the above data are integrated, which can reduce the amount of input data of the prediction model, improve the data processing efficiency of the prediction model, and shorten the processing time.

[0149] Specifically, by converting data of different dimensions such as light intensity, human flow, and power consumption into a unified unit of measurement, the calculation module successfully achieved deep data integration. This processing method not only simplifies the data model, but also enables data of different sources and types to be compared and analyzed under the same framework, which facilitates subsequent data processing and prediction model construction. Since multiple data are integrated into a single heat production data, the amount of input data for the prediction model is significantly reduced. This not only reduces the complexity of the model, but also improves data processing efficiency and shortens processing time. This advantage is particularly obvious in scenarios with high real-time requirements, which helps the system respond to environmental changes more quickly and make corresponding adjustments. By incorporating key factors such as light intensity, human flow, and power consumption into the calculation of heat production, the system can more comprehensively consider various factors affecting air conditioning load. This comprehensive consideration helps to improve the accuracy of the prediction model, enabling it to more accurately predict the future trend of air conditioning load changes and provide a more reliable basis for intelligent control. By accurately calculating and analyzing the heat production in each area and time, the system can more reasonably allocate resources and optimize energy efficiency management. For example, in areas or times with high heat production, the system can automatically adjust the operating parameters of the air conditioning system to reduce energy consumption and improve energy efficiency. This intelligent resource allocation method helps achieve the goal of energy conservation and emission reduction, and improves the comfort of the indoor environment. This technical solution enables the system to flexibly process data from different areas, different times and different sources, and dynamically adjust according to actual conditions. This flexibility not only improves the adaptability of the system, enabling it to cope with various complex and changing environmental conditions; it also enhances the scalability of the system, providing possibilities for subsequent functional upgrades and optimizations.

[0150] In some embodiments, the computing device is further configured to:

[0151] Acquire multiple sets of alternative data, including the temperature of each area at multiple historical moments, the power consumption at multiple historical moments, the flow of people at multiple historical moments, the light intensity at multiple historical moments, the valve opening of the fan coil unit 200 at multiple historical moments, and the power of the chiller 100 at multiple historical moments.

[0152] For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirements, and determine whether the power of the chiller 100 in the candidate data meets the minimum power requirements. If both are met, the candidate data is used as training data; the training data is used to train the prediction model.

[0153] More specifically, the valve opening of the fan coil unit 200 at the last historical moment is selected as the target value from the valve openings of the fan coil unit 200 at multiple historical moments in each area, and the other moments are the moments except the last historical moment in the historical moments. The temperature of each area at other moments, the power consumption at other moments, the flow of people at other moments, the light intensity at other moments, and the valve opening at other moments are input into the prediction model to obtain the predicted value of the fan coil unit 200 at the last moment, and the difference between the valve opening of the fan coil unit 200 at the last moment and the predicted value is calculated as the loss value, and the parameters in the prediction model are optimized according to the loss value, so as to realize the training of the prediction model.

[0154] In the above technical solution, training data is selected from the alternative data based on the constant temperature adjustment requirements and the minimum power requirements, and the training data is used to train the prediction model. While ensuring the comfort of the indoor environment, the energy consumption of the air-conditioning system is significantly reduced, achieving the goal of energy conservation and emission reduction.

[0155] Specifically, by selecting training data that meets the thermostatic regulation requirements and the minimum power requirements from multiple sets of candidate data, the computing device ensures the quality of the training data. These selected data are closer to the actual operating conditions and can more accurately reflect the operating status and load changes of the air-conditioning system. Therefore, the prediction model trained using these data will have higher prediction accuracy, can more accurately predict the changing trend of the air-conditioning load in the future, and provide a reliable basis for intelligent control. This technical solution emphasizes that the power of the chiller meets the minimum power requirement under the premise of meeting the thermostatic regulation requirements. This screening condition helps to reduce the energy consumption of the air-conditioning system while ensuring the comfort of the indoor environment. By training the prediction model based on these conditions, the system can optimize the operating power of the chiller while ensuring the stability of the indoor temperature, avoid unnecessary energy waste, and thus achieve the goal of energy conservation and emission reduction. By selecting training data that meets specific conditions, the prediction model can learn under what conditions the air-conditioning system can maintain the comfort of the indoor environment with the lowest energy consumption. This will help the system to automatically adjust operating parameters such as temperature setting, wind speed, air supply mode, and the power of the chiller in actual operation to seek the operation plan with the lowest energy consumption. In the long run, this will significantly improve the energy efficiency of the system and reduce operating costs. This technical solution introduces constant temperature regulation requirements and minimum power requirements as screening conditions, so that the prediction model can better adapt to load changes under different environmental conditions and user behaviors. As the system operates and data accumulates, the prediction model will continue to learn and optimize, improving its adaptability and intelligence. This will enable the system to respond more flexibly to various complex and changing situations and provide more personalized intelligent control solutions.

[0156] In some embodiments, the computing device is further configured to:

[0157] For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirement. If so, use the candidate data as intermediate data.

[0158] According to the total energy consumption of each group of intermediate data, the order is from small to large, and the first n groups of intermediate data are selected as training data.

[0159] More specifically, the constant temperature adjustment requirement can be set according to the user's needs. After the intermediate data that meets the constant temperature adjustment requirement is screened out according to the temperature in each set of candidate data, the total energy consumption of each set of intermediate data is obtained according to the power calculation of the chiller 100 in each area at multiple historical moments. The total energy consumption of each set of intermediate data is sorted in ascending order, and the first n sets of intermediate data are selected as training data, so that the data with the minimum energy consumption can be selected as training data.

[0160] In the above technical solution, the constant temperature adjustment requirement is set according to the user's temperature adjustment demand, and then the alternative data that meet the constant temperature adjustment requirement is screened out, and the alternative data with the minimum energy consumption is selected from the screened alternative data as training data. In this way, the training data that meets both the constant temperature adjustment requirement and the minimum power requirement is screened. The predicted model is trained using the screened data, which significantly reduces the energy consumption of the air-conditioning system while ensuring the comfort of the indoor environment, thereby achieving the goal of energy conservation and emission reduction.

[0161] Specifically, by setting the constant temperature adjustment requirements, the computing device can ensure that the selected candidate data meets the actual needs of the user in terms of temperature. This means that the prediction model trained based on these training data will be able to more accurately reflect the user's preference for indoor temperature, thereby achieving more accurate temperature control in the intelligent control process and improving the user's comfort experience. After preliminarily screening the candidate data that meet the constant temperature adjustment requirements, the computing device further sorts them in order from small to large in terms of total energy consumption, and selects the top n groups of data with the lowest energy consumption as training data. This step ensures that the trained prediction model can learn how to minimize energy consumption while meeting the user's temperature adjustment requirements. This will help to reduce the operating energy consumption of the air-conditioning system through intelligent control strategies in practical applications and achieve the goal of energy conservation and emission reduction. Since the training data is selected under the conditions of meeting the constant temperature adjustment requirements and minimum energy consumption, the prediction model trained based on these data will have a higher energy efficiency ratio. This means that the prediction model can more accurately predict the energy consumption of the air-conditioning system under different working conditions, and provide more reasonable operating parameter suggestions for intelligent control, thereby further reducing energy consumption while ensuring the comfort of the indoor environment. This technical solution achieves double screening of training data by comprehensively considering user temperature control needs and system energy consumption. This helps improve the predictive model's adaptability to complex and changing working conditions, enabling it to automatically adjust control strategies according to environmental changes and user behavior in actual applications, and achieve more intelligent and adaptive air conditioning load management.

[0162] In some embodiments, the computing device is further configured to:

[0163] In each set of candidate data, the temperature of each area at all historical moments is counted to obtain the temperature variance and temperature mean. If the temperature variance is less than the preset variance threshold and the temperature mean is within the preset temperature range, the area is determined to be the target area. If the number of target areas is greater than the preset number threshold, the candidate data is used as intermediate data.

[0164] For example: for the temperature data of the i-th region at p moments, the average value and variance are calculated for the temperature data of the i-th region at the 1st moment, the temperature data of the i-th region at the 2nd moment, and ... the temperature data of the i-th region at the pth moment, and the temperature average value and temperature variance of the i-th region are obtained. It is determined whether the temperature average value of the i-th region is within the preset temperature range, and whether the temperature variance of the i-th region is less than the preset variance threshold. If so, the i-th region is the target region. The preset temperature range and the preset variance threshold can be set according to user needs, for example: the temperature is between 26℃ and 28℃, and the variance threshold is 1℃. By traversing q regions, m target regions are selected from q regions. If m is greater than the preset number threshold, for example: m>0.8q, it is determined that the alternative data meets the constant temperature adjustment requirements, and the alternative data is used as intermediate data.

[0165] In the above technical solution, by performing statistical analysis on the temperature data of each area, it is determined whether each area meets the constant temperature regulation requirements within the area. The areas that meet the constant temperature regulation requirements within the area are in the majority, so it is determined whether the alternative data meets the constant temperature regulation requirements.

[0166] Specifically, by calculating the temperature variance and mean of each area and comparing them with the preset variance threshold and temperature range, the computing device can accurately identify which areas meet the constant temperature regulation requirements. The temperature variance less than the preset variance threshold indicates that the temperature fluctuation in the area is small, while the temperature mean within the preset temperature range ensures that the temperature in the area is within the comfort range expected by the user. This dual condition judgment effectively improves the accuracy of area screening. The technical solution not only considers the temperature characteristics of a single area, but also further screens the candidate data by counting the number of target areas and comparing them with the preset number threshold. Only when most areas meet the constant temperature regulation requirements will the candidate data be confirmed as intermediate data. This statistical data-based screening method improves the scientificity and rationality of data screening and ensures the data quality of subsequent prediction model training. By screening the candidate data that meets the constant temperature regulation requirements as training data, the prediction model can learn under what conditions the air-conditioning system can maintain the stability of the indoor temperature. This will help the model to more accurately predict the changing trend of air-conditioning load in practical applications and formulate corresponding control strategies, so as to reduce energy consumption and improve system energy efficiency while ensuring the comfort of the indoor environment. Since this technical solution can identify and filter out areas and data that can meet the requirements of constant temperature regulation in most cases, the prediction model trained based on this data will be more adaptive and robust. This means that the model can better adapt to load changes under different environmental conditions and user behaviors, and provide a more stable and reliable intelligent control solution. By ensuring that the air-conditioning system can maintain the stability of indoor temperature in most cases, this technical solution helps to improve user comfort and satisfaction. Stable indoor temperature not only helps to improve the quality of users' living and working environment, but also reduces the discomfort caused by temperature fluctuations, thereby enhancing users' trust and reliance on intelligent air-conditioning systems.

[0167] In one possible implementation, the computing device is further configured to:

[0168] The valves of the fan coil units are controlled according to the predicted valve opening, and the actual temperatures of each area at multiple times are obtained. The actual temperatures of each area at all times are counted to obtain the actual temperature variance and the actual temperature mean. If the actual temperature variance is less than the preset variance threshold and the actual temperature mean is within the preset temperature range, the area is determined to be a statistical area. If the number of statistical areas is less than the preset number threshold, the prediction model continues to be trained.

[0169] More specifically, the data detected by the sensor is input into the prediction model, which predicts the valve opening at the next moment, and controls the valve of the fan coil according to the predicted valve opening to obtain the actual temperature of each area at the next moment. By repeating the above process, the actual temperature of each area at multiple moments can be obtained. For each area, a statistical analysis is performed on the actual temperature of the area at all times to determine whether the temperature regulation of the area meets the constant temperature regulation. After analyzing the temperature of each area, the areas that meet the constant temperature regulation are counted. If the number of areas is less than the preset threshold, it is determined that the accuracy of the prediction model is insufficient and the prediction model needs to be further trained.

[0170] In the above technical solution, after using the prediction model to predict the valve opening at the next moment, the actual temperature of each area is monitored and statistical analysis is performed on the actual temperature to determine whether the valve opening predicted by the prediction model can meet the constant temperature regulation. If it cannot be met, the model is continued to be trained to improve the prediction accuracy of the prediction model through continued training.

[0171] Specifically, by monitoring the actual temperature of each area at multiple times, the computing device can verify in real time whether the valve opening predicted by the prediction model is accurate. The statistical results of the actual temperature variance and the actual temperature mean directly reflect the stability and comfort of the indoor temperature. If the actual temperature variance is small and the actual temperature mean is within the preset temperature range, it means that the prediction effect of the prediction model is good and can meet the constant temperature regulation requirements. When the number of statistical areas is less than the preset number threshold, it means that the prediction effect of the prediction model fails to meet expectations, and the computing device will continue to train the prediction model. This dynamic adjustment mechanism ensures that the prediction model can be continuously optimized with the change of environmental conditions and the adjustment of user behavior, thereby improving its prediction accuracy and robustness. By accurately controlling the valve opening of the fan coil unit and combining the feedback of the actual temperature for model training, the computing device can ensure that the air-conditioning system can meet the constant temperature regulation requirements while minimizing energy consumption. By real-time monitoring and statistical analysis of the actual temperature, the computing device can ensure that the air-conditioning system can maintain a stable indoor temperature in most areas, thereby improving the comfort and satisfaction of users. Stable indoor temperature helps to reduce the discomfort caused by temperature fluctuations and improve the quality of the user's living and working environment.

[0172] Figure 3 Schematic diagram of the structure of the computing device 300 provided in this application. Figure 3 As shown, the computing device 30 provided in this embodiment includes: at least one processor 301 and a memory 302. Optionally, the device 30 also includes a communication component 303. The processor 301, the memory 302 and the communication component 303 are connected via a bus.

[0173] In a specific implementation process, at least one processor 301 executes the computer-executable instructions stored in the memory 302, so that at least one processor 301 executes the above method.

[0174] The specific implementation process of the processor 301 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.

[0175] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0176] The memory may include a high-speed memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-volatile Memory, NVM), such as at least one disk memory.

[0177] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.

[0178] The present application also provides a computer program product, including a computer program, which implements the above method when executed by a processor.

[0179] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above method is implemented.

[0180] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.

[0181] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.

[0182] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0183] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0184] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0185] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0186] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.

[0187] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A real-time monitoring and intelligent control system for air conditioning load, characterized in that: The system includes: a chiller located in the control room and fan coils located in various areas of the room, the chiller providing chilled water to each fan coil; The system also includes temperature sensors located in each area, a flow rate detector in each area, a power detector in each area, and a light intensity detector in each area; The system further includes a computing device, the computing device being used to obtain the temperature of each area at multiple times, the heat generation data of each area at multiple times, and the valve opening of each fan coil at multiple times; The prediction model is run in the computing device, and the prediction model includes a calculation module, an extraction module and a regression module; The calculation module is used to calculate the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment; extract data from the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment according to the first correlation to obtain first extracted data; and calculate the second correlation between the temperature of each area, the heat generation data of each area, and the valve opening of each area, and extract data from the temperature of each area, the heat generation data of each area, and the valve opening of each area according to the second correlation to obtain second extracted data; The extraction module is used to perform feature extraction on the first extracted data to obtain first feature data; and perform feature extraction on the second extracted data to obtain second feature data; and fuse the first feature data and the second feature data to obtain fused features; The regression module is used to perform prediction processing on the fused features to obtain the valve opening at the next moment.

2. The air conditioning load real-time monitoring and intelligent control system according to claim 1 is characterized in that: The calculation module is also used for: Converting the temperatures of each region at multiple moments into a temperature matrix, wherein each row of the temperature matrix represents the temperatures of a region at multiple moments, and each column represents the temperatures of multiple regions at one moment; Converting the heat generation data of each region at multiple times into a heat generation matrix, wherein each row of the heat generation matrix represents the heat generation data of one region at multiple times, and each column represents the heat generation data of multiple regions at one time; The valve openings of each region at multiple times are converted into a valve opening matrix; each row in the valve opening matrix represents the valve openings of a region at multiple times, and each column represents the valve openings of multiple regions at one time; The temperature matrix and the heat generation matrix are spliced ​​together, and the correlation between the spliced ​​matrix and the valve opening matrix is ​​calculated to obtain the first correlation and the second correlation.

3. The air conditioning load real-time monitoring and intelligent control system according to claim 2 is characterized in that: The calculation module is also used for: Randomly generate a third weight parameter and a fourth weight parameter, convert the third weight parameter into a third parameter vector, and convert the fourth weight parameter into a fourth parameter vector; The heat generation matrix and the valve opening matrix are spliced ​​on the columns to obtain a second spliced ​​matrix; the product of the transpose of the third parameter vector and the second spliced ​​matrix is ​​calculated to obtain a third calculation vector; the product of the transpose of the fourth parameter vector and the valve opening matrix is ​​calculated to obtain a fourth calculation vector; Calculate a second correlation between the third calculation vector and the fourth calculation vector. If the second correlation is greater than a first preset threshold, extract data from the second splicing matrix according to the third weight parameter; and extract data from the valve opening matrix according to the fourth weight parameter to obtain the second extracted data.

4. The air conditioning load real-time monitoring and intelligent control system according to any one of claims 1 to 3, characterized in that: The computing device is also used to: Acquire multiple sets of candidate data, wherein the candidate data include the temperature of each area at multiple historical moments, the power consumption at multiple historical moments, the flow of people at multiple historical moments, the light intensity at multiple historical moments, the valve opening of the fan coil at multiple historical moments, and the power of the chiller at multiple historical moments; For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirement, and determine whether the power of the chiller in the candidate data meets the minimum power requirement. If both are satisfied, use the candidate data as training data; The training data is used to train the prediction model.

5. The air conditioning load real-time monitoring and intelligent control system according to claim 4 is characterized in that: The computing device is also used to: For each set of candidate data, determine whether the temperature in the candidate data meets the constant temperature regulation requirement, and if so, use the candidate data as intermediate data; The total energy consumption of each group of intermediate data is sorted in ascending order, and the first n groups of intermediate data are selected as the training data.

6. The air conditioning load real-time monitoring and intelligent control system according to claim 5 is characterized in that: The computing device is also used to: In each group of the alternative data, the temperatures of each area at all historical moments are counted to obtain the temperature variance and the temperature mean; if the temperature variance is less than a preset variance threshold and the temperature mean is within a preset temperature range, the area is determined to be a target area; if the number of the target areas is greater than a preset number threshold, the alternative data is used as intermediate data.

7. The air conditioning load real-time monitoring and intelligent control system according to any one of claims 1 to 3, characterized in that: The computing device is also used to: Control the valve of the fan coil unit according to the predicted valve opening to obtain the actual temperature of each area at multiple times; count the actual temperature of each area at all times to obtain the actual temperature variance and the actual temperature mean; If the actual temperature variance is less than the preset variance threshold and the actual temperature mean is within the preset temperature range, the area is determined to be a statistical area; if the number of the statistical areas is less than the preset number threshold, the prediction model continues to be trained.

8. A method for real-time monitoring and intelligent control of air conditioning load, characterized in that: Applied to the real-time monitoring and intelligent control system of air conditioning load, the system comprises: a chiller located in the control room and fan coils located in various areas of the room, the chiller provides chilled water to each fan coil; The system also includes temperature sensors located in each area, a flow rate detector in each area, a power detector in each area, and a light intensity detector in each area; The system further includes a computing device, the computing device being used to obtain the temperature of each area at multiple times, the heat generation data of each area at multiple times, and the valve opening of each fan coil at multiple times; The prediction model is run in the computing device, and the prediction model includes a calculation module, an extraction module and a regression module; The calculation module is used to calculate the first correlation between the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment; extract data from the temperature at each moment, the heat generation data at each moment, and the valve opening at each moment according to the first correlation to obtain first extracted data; and calculate the second correlation between the temperature of each area, the heat generation data of each area, and the valve opening of each area, and extract data from the temperature of each area, the heat generation data of each area, and the valve opening of each area according to the second correlation to obtain second extracted data; The extraction module is used to perform feature extraction on the first extracted data to obtain first feature data; and perform feature extraction on the second extracted data to obtain second feature data; and fuse the first feature data and the second feature data to obtain fused features; The regression module is used to perform prediction processing on the fused features to obtain the valve opening at the next moment.

9. An electronic device, characterized in that: include: processor; as well as, A memory, configured to store executable instructions of the processor; Wherein, the processor is configured to perform the method of claim 8 by executing the executable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to claim 8 when executed by a processor.

Images