CN-121993883-A - Air conditioner fresh air control method and system based on multi-source perception and scene self-learning

CN121993883ACN 121993883 ACN121993883 ACN 121993883ACN-121993883-A

Abstract

The invention relates to the technical field of intelligent air conditioning, and provides an air conditioner fresh air control method and system based on multi-source perception and scene self-learning. When the air quality exceeds the standard, the indoor and outdoor comprehensive feature vectors are constructed, and the knowledge base is searched to match the optimal operation mode. And for a non-matching scene, calculating the air quality improvement rate through mode polling, and autonomously optimizing and warehousing. And monitoring the actual fresh air efficiency in real time in operation, and triggering a relearning mechanism to update the knowledge base entry if the deviation between the actual fresh air efficiency and the recorded value reaches a threshold value. The invention realizes the transition of fresh air control from passive response to active early warning, autonomous optimizing and efficiency self-compensating, and remarkably improves the accuracy and efficiency of indoor air purification in complex environments.

Inventors

SHENG YUJUN
FENG TIAN
LI QIANG

Assignees

广东三华钒音科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260304

Claims (9)

1. The fresh air control method of the air conditioner based on multi-source perception and scene self-learning is characterized by being applied to the air conditioner comprising a multi-parameter perception module, a weather forecast access module and a fresh air executing mechanism, and comprises the following steps: acquiring indoor pollutant parameters in real time through the multi-parameter sensing module, and calculating a comprehensive air quality index corresponding to the pollutant parameters by the main control module according to preset weights; The weather forecast access module is used for acquiring external environment forecast information, and the main control module is used for controlling the fresh air executing mechanism to switch between a wind-rush mode and an internal circulation mode according to the external environment forecast information and the judgment result of the comprehensive air quality index; When the comprehensive air quality index exceeds a set threshold value, acquiring current outdoor environment parameters by the main control module, constructing indoor and outdoor comprehensive feature vectors by combining indoor pollutant parameters, searching a pre-stored knowledge base according to the indoor and outdoor comprehensive feature vectors, and matching a corresponding optimal operation mode; If the knowledge base has no matching mode, the main control module acquires air quality improvement rates in different modes through mode polling, determines an optimal operation mode in the current scene according to the air quality improvement rate, and stores the optimal operation mode in the knowledge base in a correlation manner; And continuously acquiring actual fresh air efficiency by the main control module in the operation process, and triggering a relearning mechanism to update the optimal operation mode in the knowledge base according to the deviation value of the actual fresh air efficiency and the recording efficiency in the knowledge base.
2. The intelligent control method for fresh air of an air conditioner based on multi-source perception and scene self-learning according to claim 1, wherein calculating the comprehensive air quality index corresponding to the pollutant parameter according to the preset weight comprises: Acquiring actual measurement values of pollutant parameters acquired by the multi-parameter sensing module, and respectively establishing a ratio relation between each actual measurement value and a preset safety threshold value to generate a normalized parameter matrix corresponding to each pollutant parameter one by one; Retrieving and calling a weight coefficient set matched with the current environment scene from a weight database based on the timestamp information of the current environment and the indoor and outdoor meteorological parameter characteristics, wherein the weight coefficient set comprises weight ratios for different pollutant parameters; And carrying out weighted fusion logic operation on each normalization parameter in the normalization parameter matrix and the weight proportion corresponding to the normalization parameter in the weight coefficient set to obtain the comprehensive air quality index for representing the indoor air condition.
3. The intelligent control method for fresh air of an air conditioner based on multi-source perception and scene self-learning according to claim 1, wherein controlling the fresh air actuator to switch between the air-rush mode and the internal circulation mode according to the external environment forecast information and the determination result of the comprehensive air quality index comprises: If the external environment forecast information indicates that a target pollution event exists in a future preset period, the main control module judges that a reserve condition is met currently, controls the fresh air executing mechanism to open an external circulation air valve and adjust the fan to a first preset power operation, and enters an air-robbing mode; The main control module calculates early warning countdown of the arrival of the target pollution event in real time, and controls the fresh air executing mechanism to close the external circulation air valve and open an internal circulation purification mode before the early warning countdown reaches a preset switching threshold value; And during the operation of the internal circulation purification mode, the main control module adjusts the purification operation frequency of the fresh air executing mechanism in real time according to the comprehensive air quality index until the external environment forecast information indicates that the target pollution event is ended.
4. The method for controlling fresh air of an air conditioner based on multi-source perception and scene self-learning according to claim 1, wherein controlling the fresh air actuator to switch between the air-rush mode and the internal circulation mode according to the external environment forecast information and the determination result of the comprehensive air quality index comprises: The main control module obtains rated ventilation efficiency and indoor space volume of the fresh air executing mechanism, and calculates preset air-taking time required for reaching an air reserve target according to the difference between the current comprehensive air quality index and a target cleaning index; Determining a wind-robbing trigger time by combining the preset wind-robbing time based on the predicted starting time of the target pollution event in the external environment forecast information; when the real-time moment reaches the wind-robbing triggering moment and the current outdoor environment quality parameter is better than a preset admission threshold value, controlling the fresh air executing mechanism to enter the wind-robbing mode; and when the early warning countdown is smaller than or equal to a reserved switching allowance or the currently monitored indoor air reserve quantity reaches a saturation threshold, controlling the fresh air executing mechanism to execute switching action from external circulation to internal circulation.
5. The fresh air control method of an air conditioner based on multi-source perception and scene self-learning according to claim 1, wherein searching a pre-stored knowledge base according to indoor and outdoor comprehensive feature vectors, matching the corresponding optimal operation mode comprises: Packaging the outdoor environment parameters into outdoor feature vectors, and fusing the outdoor feature vectors with the indoor pollutant parameters, indoor and outdoor temperature and humidity gradients, season classification identifiers and 24-hour time stamps to construct indoor and outdoor comprehensive feature vectors representing the current indoor and outdoor comprehensive environment states; Acquiring a target pollutant with the largest normalized concentration value in the indoor pollutant parameters, and according to the type of the target pollutant, calling contribution degree weights corresponding to each dimension characteristic from a weight database, and mapping the indoor and outdoor comprehensive characteristic vectors to a weighted characteristic space; Locking a candidate historical scene set in the pre-stored knowledge base by taking the seasonal classification identifier and the timestamp in the weighted feature space as primary indexes, and calculating weighted Euclidean distance between the indoor and outdoor comprehensive feature vectors and each candidate historical scene feature vector after mapping; Judging whether the minimum weighted Euclidean distance is smaller than a preset scene confidence threshold, if so, judging that the current scene is successfully matched, and extracting a historical operation mode corresponding to the minimum weighted Euclidean distance from the candidate historical scene set to be used as the optimal operation mode in the current environment scene.
6. The method for controlling fresh air of an air conditioner based on multi-source perception and scene self-learning according to claim 5, wherein obtaining the air quality improvement rate in different modes by mode polling comprises: If the minimum weighted Euclidean distance is greater than or equal to the preset scene confidence threshold, starting a mode polling flow by the main control module, and controlling the fresh air executing mechanism to be sequentially switched to an operation state corresponding to a wind-robbing mode and an internal circulation mode; denoising the comprehensive air quality index acquired in real time by using a moving average filtering algorithm to generate a smoothed index evolution sequence; And carrying out first-order differential operation on the index evolution sequence, obtaining a dynamic change gradient of the comprehensive air quality index in unit sampling time, and determining a statistical average value of the dynamic change gradient as the air quality improvement rate in a corresponding mode.
7. The method for controlling fresh air of an air conditioner based on multi-source perception and scene self-learning as claimed in claim 6, wherein determining an optimal operation mode in a current scene according to the magnitude of the air quality improvement rate comprises: The latest air quality improvement rate of the air-rush mode and the latest air quality improvement rate of the internal circulation mode are obtained, and the mode corresponding to the larger value of the air-rush mode and the internal circulation mode is used as the optimal operation mode in the current scene; and storing the redetermined optimal operation mode and the corresponding air quality improvement rate in the pre-stored knowledge base in a correlated mode so as to finish the coverage updating of the original failure entries.
8. The method of claim 6, wherein triggering a re-learning mechanism to update the best operation mode in the knowledge base according to the deviation value between the actual fresh air efficiency and the recording efficiency in the knowledge base comprises: During the execution of the optimal operation mode, the dynamic change gradient is used as a real-time index for evaluating the actual fresh air efficiency; Calculating the reduction proportion of the actual fresh air efficiency relative to the recording efficiency in the knowledge base; And if the reduction ratio exceeds a preset attenuation threshold, judging that the performance of the current execution mode is lower than expected, and re-triggering mode polling by the main control module.
9. Air conditioner new trend control system based on multisource perception and scene are from study, its characterized in that includes the air conditioner of multiparameter perception module, weather forecast access module and new trend actuating mechanism, still includes: the early warning switching module is used for acquiring indoor pollutant parameters in real time through the multi-parameter sensing module, and calculating a comprehensive air quality index corresponding to the pollutant parameters according to a preset weight by the main control module; The weather forecast access module is used for acquiring external environment forecast information, and the main control module is used for controlling the fresh air executing mechanism to switch between a wind-rush mode and an internal circulation mode according to the external environment forecast information and the judgment result of the comprehensive air quality index; The scene retrieval module is used for acquiring current outdoor environment parameters by the main control module when the comprehensive air quality index exceeds a set threshold value, constructing indoor and outdoor comprehensive feature vectors by combining indoor pollutant parameters, retrieving a pre-stored knowledge base according to the indoor and outdoor comprehensive feature vectors and matching a corresponding optimal operation mode; The polling learning module is used for acquiring air quality improvement rates in different modes through mode polling by the main control module if no matching mode exists in the knowledge base, determining an optimal operation mode in the current scene according to the air quality improvement rate, and storing the optimal operation mode in the knowledge base in a correlation manner; And the re-learning module is used for continuously acquiring the actual fresh air efficiency by the main control module in the operation process, and triggering a re-learning mechanism to update the optimal operation mode in the knowledge base according to the deviation value of the actual fresh air efficiency and the recording efficiency in the knowledge base.

Description

Air conditioner fresh air control method and system based on multi-source perception and scene self-learning Technical Field The invention relates to the technical field of intelligent air conditioning, in particular to an air conditioner fresh air control method and system based on multi-source perception and scene self-learning. Background The control strategy of the existing air conditioner fresh air system has the technical defects that in a perception level, the existing scheme is mainly monitored by adopting a single or few sensors, control decision is carried out only by relying on isolated parameters such as carbon dioxide or PM2.5, and indoor air quality deterioration is often the combined action result of various pollutants, the one-sided perception mode leads to slow response and even failure of the system under a complex pollution scene, in a control logic level, the existing system basically belongs to passive reaction type control, fresh air intervention is started only after the parameter is detected to be out of standard, sudden change of the environment cannot be dealt with, when severe events such as sand storm or severe haze are about to happen outside, the system can only be passively resisted after pollution invasion, at the moment, fresh air is started instead, indoor pollution is aggravated, the service life of a filter screen is rapidly consumed, in a strategy optimization level, traditional control logic is based on a preset fixed threshold value and an operation mode, the capability of autonomous learning from historical operation data is not available, the optimal fresh air mode under different application scenes is not identified, control efficiency is low, the energy consumption is high, and individual comfort requirements are difficult to meet. Therefore, an intelligent fresh air solution capable of fully sensing multidimensional pollution parameters, actively pre-judging external risks and continuously optimizing a control strategy by oneself is urgently needed in the prior art. Disclosure of Invention Aiming at the defects, the invention aims to provide an air conditioner fresh air control method and system based on multi-source perception and scene self-learning, which aims to realize prospective wind defense control by constructing a multi-parameter fusion comprehensive air quality evaluation system and combining weather forecast information, and continuously optimize operation mode selection by establishing a scene self-learning mechanism, so that the fresh air system has intelligent decision making capability of comprehensive perception, active prejudgment and continuous evolution, thereby remarkably improving the accuracy, timeliness and self-adaption level of indoor air quality control. To achieve the purpose, the invention adopts the following technical scheme: the fresh air control method based on multi-source perception and scene self-learning is applied to an air conditioner comprising a multi-parameter perception module, a weather forecast access module and a fresh air executing mechanism, and comprises the following steps: acquiring indoor pollutant parameters in real time through the multi-parameter sensing module, and calculating a comprehensive air quality index corresponding to the pollutant parameters by the main control module according to preset weights; The weather forecast access module is used for acquiring external environment forecast information, and the main control module is used for controlling the fresh air executing mechanism to switch between a wind-rush mode and an internal circulation mode according to the external environment forecast information and the judgment result of the comprehensive air quality index; When the comprehensive air quality index exceeds a set threshold value, acquiring current outdoor environment parameters by the main control module, constructing indoor and outdoor comprehensive feature vectors by combining indoor pollutant parameters, searching a pre-stored knowledge base according to the indoor and outdoor comprehensive feature vectors, and matching a corresponding optimal operation mode; If the knowledge base has no matching mode, the main control module acquires air quality improvement rates in different modes through mode polling, determines an optimal operation mode in the current scene according to the air quality improvement rate, and stores the optimal operation mode in the knowledge base in a correlation manner; And continuously acquiring actual fresh air efficiency by the main control module in the operation process, and triggering a relearning mechanism to update the optimal operation mode in the knowledge base according to the deviation value of the actual fresh air efficiency and the recording efficiency in the knowledge base. Preferably, calculating the comprehensive air quality index corresponding to the pollutant parameter according to the preset weight includes: Acquiring actual measurement values of po