CN-122015235-A - Energy-saving control method for central air conditioner driven by machine vision

CN122015235ACN 122015235 ACN122015235 ACN 122015235ACN-122015235-A

Abstract

The invention relates to a machine vision driven energy-saving control system of a central air conditioner. The system comprises a preset data acquisition dimension, obtaining multi-dimensional original data of a central air conditioner, obtaining standardized data through preprocessing, obtaining personnel distribution and indoor environment characteristic data through target detection and environment characteristic extraction, forming a data analysis result through fusion analysis, constructing an indoor load demand assessment model based on the analysis result to obtain a real-time dynamic load demand value, obtaining a matching deviation of the load demand and the air conditioner operation by combining air conditioner operation parameter analysis, generating a multi-stage visual driving instruction, analyzing the multi-stage visual driving instruction to obtain an air conditioner operation adjustment parameter demand, optimizing calculation by combining a preset energy-saving optimization mechanism, obtaining energy-saving operation parameters, converting control signals, driving the air conditioner to perform energy-saving operation, and realizing energy-saving control of the central air conditioner by utilizing machine vision.

Inventors

WANG ENCHENG
ZHANG YUMIN

Assignees

上海云见智能科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (12)

1. The machine vision driven central air conditioner energy-saving control method is characterized by comprising the following steps of: s1, acquiring multi-dimensional original data of central air conditioner operation through a preset data acquisition dimension; S2, preprocessing the multi-dimensional original data to obtain standardized data, carrying out target detection and environment characteristic extraction based on the standardized data to obtain characteristic data of personnel distribution and indoor environment; S3, constructing an indoor load demand assessment model based on the data analysis result to obtain a real-time dynamic load demand value, analyzing by combining the operation parameters of a central air conditioner to obtain a matching deviation between the load demand and the operation of the air conditioner, and generating a multistage visual driving instruction by utilizing the matching deviation; s4, analyzing the multi-level visual driving instruction to obtain an air conditioner operation adjustment parameter demand, optimizing the parameter demand by combining a preset energy-saving optimization mechanism to obtain an energy-saving operation parameter, and converting the energy-saving operation parameter to obtain a control signal of the air conditioner energy-saving operation mechanism.
2. The system of claim 1, wherein the specific process of acquiring the multi-dimensional original data of the central air conditioner operation comprises acquiring area environment basic data, acquiring first environment data, shooting an indoor scene image by using a machine vision camera, recording personnel activity area and environment details, acquiring second vision data, acquiring operation parameters of the central air conditioner through an air conditioner controller bus interface, generating operation data of a third central air conditioner, adding a uniform time stamp to the first environment data, the second vision data and the operation data of the third central air conditioner, and acquiring the multi-dimensional original data of a time sequence mark.
3. The system of claim 1, wherein the obtaining of the standardized data comprises performing outlier rejection and normalization processing on the first environmental data in the multi-dimensional original data to obtain standardized environmental data, performing denoising, greyscale and size normalization operations on the second visual data to obtain standardized image data, performing unit unified conversion on the operation data of the third central air conditioner to obtain standardized operation parameters, and integrating the standardized environmental data, the standardized image data and the standardized operation parameters to obtain a structured standardized data set.
4. The system according to claim 1, wherein the target detection is to analyze the standardized image data frame by using a target detection rule, identify a human target in an image, acquire coordinate position, number and motion trajectory data of the human target, acquire personnel activity density based on pixel duty ratio and distance conversion of the human target, and integrate personnel distribution information by using result data of the target detection rule.
5. The system of claim 1, wherein the environmental feature extraction is performed by extracting environmental detail features of a personnel-intensive area based on the personnel distribution information, performing gray value analysis on wall surface temperature distribution in the standardized image data and a heat dissipation area of a central air conditioner to obtain indoor thermal environmental feature data, and acquiring complete indoor environmental feature data by combining non-visual dimension environmental features in the standardized environmental data.
6. The system of claim 1, wherein the specific process of performing fusion analysis on the feature data comprises performing weight distribution on corresponding data dimensions by using a weighted fusion mechanism based on the personnel distribution information, the indoor environment feature data and the standardized operation parameters, calculating three indexes of indoor comfort degree score, personnel load ratio and operation adaptation degree of a central air conditioner, and integrating the three indexes to obtain a data analysis result.
7. The system according to claim 1, wherein the acquiring of the real-time dynamic load demand value comprises constructing a dynamic load demand evaluation model based on the three indexes, taking a difference value between the real-time temperature and the real-time humidity in the standardized environment data and a set comfort threshold value as a model correction parameter, substituting the model correction parameter into operation adaptation degree data of a central air conditioner to calculate to obtain a base load demand value, and dynamically correcting the base load demand value by combining a personnel movement track and a personnel flow trend to acquire the real-time dynamic load demand value.
8. The system according to claim 1, wherein the specific process of analyzing by combining the operation parameters of the central air conditioner comprises calculating an air conditioner actual load supply value based on the standardized operation parameters, comparing the real-time dynamic load demand value with the air conditioner actual load supply value to obtain a preliminary matching deviation, analyzing the operation adaptation degree data of the central air conditioner to obtain a deviation factor tree, and correcting the preliminary matching deviation by combining the indoor thermal environment characteristic data to obtain the matching deviation of the load demand and the air conditioner operation.
9. The system of claim 1, wherein the specific process of generating the multi-stage visual driving instruction by using the matching deviation comprises presetting a high, medium and low three-stage deviation threshold, acquiring an instruction adjustment priority according to the grade of the air conditioner operation matching deviation, distributing a differential adjustment weight to a corresponding area based on the personnel distribution information to generate an area differential adjustment parameter, and integrating the instruction adjustment priority, the area differential adjustment parameter and a load demand to acquire the multi-stage visual driving instruction.
10. The system according to claim 1, wherein the specific process of obtaining the air conditioner operation adjustment parameter requirement comprises analyzing adjustment priority and region differentiation adjustment parameters in the multi-stage visual driving instruction, extracting a load correction direction in the multi-stage visual driving instruction, calculating a basic adjustment amplitude by combining the real-time dynamic load requirement value, and performing safety threshold constraint on the basic adjustment amplitude based on the operation adaptation degree data of the central air conditioner to obtain the air conditioner operation adjustment parameter requirement.
11. The system according to claim 1, wherein the optimizing calculation of the parameter requirements includes using a preset particle swarm optimization algorithm, taking the air conditioner operation adjustment parameter requirements as an optimization target, inputting a safe operation parameter range of equipment as a constraint condition, and constructing an energy-saving-comfort dual-objective optimization function by combining an indoor comfort level scoring threshold value to obtain energy-saving operation parameters of energy-saving efficiency and comfort level.
12. The system of claim 1, wherein the specific process of obtaining the control signal of the energy-saving operation mechanism of the air conditioner comprises format conversion of the energy-saving operation parameters of the energy-saving efficiency and the comfort level, adaptation of a communication protocol of an execution mechanism of the central air conditioner, weight adjustment based on regional differentiation, and decomposition of the parameters into independent control parameters of the execution mechanisms of the corresponding regions, so as to generate the energy-saving operation control signal of the central air conditioner.

Description

Energy-saving control method for central air conditioner driven by machine vision Technical Field The invention belongs to the technical field of automatic control and artificial intelligence, and particularly relates to a machine vision driven energy-saving control method for a central air conditioner. Background At present, the energy-saving control aspect of the central air conditioner still has the following to be improved: Under the background of global energy transformation and double-carbon target propulsion, building energy consumption is taken as the core field of energy consumption, the energy saving potential of the building energy consumption is paid attention to, and a central air conditioning system is taken as core energy equipment in places such as commercial buildings, office parks, large-scale venues and the like, and the energy consumption is larger than the total energy consumption of the building, so that the building energy consumption is a key break of building energy saving. However, the current energy-saving control technology of the central air conditioner still has a plurality of outstanding bottlenecks, and the actual requirements of both accurate energy saving and comfortable experience are difficult to meet: First, the perception dimension is single, and data acquisition is incomplete. The traditional control system depends on a small number of physical sensors such as temperature and humidity, only captures a single environmental index, lacks effective perception on key load influence factors such as dynamic distribution, activity track and intensity of indoor personnel, and simultaneously does not fully integrate operation parameters and environmental detail characteristics of a central air conditioner, so that data dimension is incomplete, and dynamic changes of actual load demands cannot be comprehensively reflected. Secondly, the load assessment is static and has serious inadequacy. The existing load evaluation model is constructed based on building fixed parameters, historical energy consumption data or static algorithm, and is difficult to respond to dynamic scenes such as personnel flow, temporary gatherings, environmental mutation and the like. For example, the difference of the personnel density between office area working peak and noon break period is obvious, the traditional model can not adjust the load evaluation result in real time, so that the air conditioner load supply and the actual demand are in serious mismatch, namely, the personnel-intensive area can be refrigerated/heated insufficiently, and the unmanned area or the low-load area still maintains high-power operation, so that a large amount of energy sources are wasted. Thirdly, the control strategy lacks differentiation and has weak collaborative optimization capability. Most systems adopt a unified control mode of one cut, and the system is not subjected to differential regulation and control according to personnel density, environmental characteristics and functional requirements of different areas, so that accurate energy supply cannot be realized. Meanwhile, the prior art often fails to balance energy-saving efficiency and indoor comfort, or excessively reduces operation load for pursuing energy saving, so that indoor temperature and humidity deviate from a comfortable range to influence personnel experience, or maintains high-load operation for ensuring comfort, neglects energy consumption, and falls into the dilemma of 'energy saving and comfort can not be achieved' at the same time. Fourth, the connection between instruction generation and execution is not smooth, and the control precision is low. The existing control instructions are mostly generated based on a single deviation signal, scientific priority division and regional pertinence are lacked, and the parameter optimization process is not fully combined with the equipment safe operation constraint, so that the response of the control instructions is lagged, and the adjustment amplitude is too large or too small. In addition, the communication protocol difference and the area suitability of the central air conditioner actuating mechanism are not considered in part of the system, the equipment operation is difficult to precisely drive after the parameter conversion, the control effect is further reduced, and the balance between high-efficiency energy conservation and stable operation cannot be realized. These problems not only result in low energy utilization efficiency of the central air conditioning system and increase the running cost of users, but also are contrary to the current development concept of green and low carbon, so that an intelligent control method integrating multidimensional sensing, dynamic load evaluation, differential control and collaborative optimization is needed, the bottleneck of the prior art is broken, and the energy-saving control of the central air conditioner is promoted to be upgraded to be accurate, intellige