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

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. And the air conditioner is driven to be switched between a wind grabbing mode and an internal circulation mode in combination with weather forecast information. And when the air quality exceeds the standard, indoor and outdoor comprehensive feature vectors are constructed, and the knowledge base is retrieved to match the optimal operation mode. And for a non-matching scene, the air quality improvement rate is calculated through mode polling, and autonomous optimization and warehousing are performed. The actual fresh air efficiency is monitored in real time during operation, and if the deviation between the actual fresh air efficiency and the recorded value reaches a threshold value, a re-learning mechanism is triggered to update knowledge base entries. According to the method, the conversion of fresh air control from passive response to active early warning, autonomous optimization and efficiency self-compensation is realized, and the precision and efficiency of indoor air purification in a complex environment are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent air conditioning technology, and in particular to an air conditioning fresh air control method and system based on multi-source sensing and scene self-learning. Background Technology

[0002] Existing air conditioning fresh air systems generally suffer from the following technical deficiencies: At the sensing level, current solutions often rely on single or a few sensors for monitoring, depending solely on isolated parameters such as carbon dioxide or PM2.5 for control decisions. However, indoor air quality deterioration is often the result of multiple pollutants acting together, and this one-sided sensing approach leads to slow response or even system failure in complex pollution scenarios. At the control logic level, existing systems are essentially passive, reactive controls, only initiating fresh air intervention after parameters exceed limits. This makes them unable to cope with sudden environmental changes. When severe events such as sandstorms or heavy smog are imminent, the system can only passively defend against pollution intrusion; activating fresh air at this time may exacerbate indoor pollution and rapidly deplete filter life. At the strategy optimization level, traditional control logic is based on preset fixed thresholds and operating modes, lacking the ability to learn autonomously from historical operating data. It cannot identify the optimal fresh air mode for different application scenarios, resulting in low control efficiency, high energy consumption, and difficulty in meeting personalized comfort needs. Therefore, there is an urgent need for an intelligent fresh air solution that can comprehensively sense multi-dimensional pollution parameters, proactively predict external risks, and continuously optimize its control strategy. Summary of the Invention

[0003] To address the aforementioned shortcomings, the present invention aims to propose an air conditioning fresh air control method and system based on multi-source perception and scenario self-learning. This method aims to construct a comprehensive air quality assessment system integrating multiple parameters, combine weather forecast information to achieve proactive wind defense control, and establish a scenario self-learning mechanism to continuously optimize the selection of operating modes. This enables the fresh air system to possess intelligent decision-making capabilities for comprehensive perception, proactive prediction, and continuous evolution, thereby significantly improving the accuracy, timeliness, and adaptability of indoor air quality control.

[0004] To achieve this objective, the present invention adopts the following technical solution: A method for controlling fresh air in air conditioning systems based on multi-source sensing and scene self-learning is applied to an air conditioning system that includes a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator. The method includes: The multi-parameter sensing module acquires indoor pollutant parameters in real time, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to preset weights. The weather forecast access module obtains external environmental forecast information, and the main control module controls the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. When the comprehensive air quality index exceeds the set threshold, the main control module obtains the current outdoor environmental parameters and constructs an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, it searches the pre-stored knowledge base and matches the corresponding optimal operating mode. If there is no matching pattern in the knowledge base, the main control module obtains the air quality improvement rate under different modes through mode polling, determines the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and stores it in the knowledge base. During operation, the main control module continuously acquires the actual fresh air efficiency and triggers a relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

[0005] Preferably, calculating the comprehensive air quality index corresponding to the pollutant parameters according to preset weights includes: The measured values ​​of pollutant parameters collected by the multi-parameter sensing module are obtained, and the ratio relationship between each measured value and the preset safety threshold is established to generate a normalized parameter matrix that corresponds one-to-one with each pollutant parameter. Based on the timestamp information of the current environment and the characteristics of indoor and outdoor meteorological parameters, a set of weight coefficients matching the current environmental scenario is retrieved from the weight database, wherein the set of weight coefficients includes the weight ratio for different pollutant parameters; The normalized parameter in the normalized parameter matrix is ​​weighted and fused logically operated with its corresponding weight ratio in the weight coefficient set to obtain the comprehensive air quality index that characterizes the indoor air condition.

[0006] Preferably, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode based on the external environment forecast information and the determination result of the comprehensive air quality index includes: If the external environment forecast information indicates that a target pollution event will occur within a preset time period in the future, the main control module determines that the current reserve conditions are met, controls the fresh air actuator to open the external circulation valve and adjusts the fan to operate at the first preset power, and enters the wind grabbing mode; The main control module calculates the warning countdown for the arrival of the target pollution event in real time, and controls the fresh air actuator to close the external circulation valve and start the internal circulation purification mode before the warning countdown reaches the preset switching threshold. During the operation of the internal circulation purification mode, the main control module adjusts the purification operation frequency of the fresh air actuator in real time according to the comprehensive air quality index until the external environmental forecast information indicates that the target pollution event has ended.

[0007] Preferably, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode based on the external environment forecast information and the determination result of the comprehensive air quality index includes: The main control module obtains the rated air exchange efficiency and indoor space volume of the fresh air actuator, and calculates the preset air-saving time required to reach the air reserve target based on the difference between the current comprehensive air quality index and the target cleanliness index. Based on the predicted start time of the target pollution event in the external environment forecast information, the wind-fighting trigger time is determined in combination with the preset wind-fighting duration; When the real-time moment reaches the wind-fighting trigger moment and the current outdoor environmental quality parameters are better than the preset access threshold, the fresh air actuator is controlled to enter the wind-fighting mode. When the warning countdown is less than or equal to the reserved switching margin, or when the currently monitored indoor air reserve reaches the saturation threshold, the fresh air actuator is controlled to perform a switching action from external circulation to internal circulation.

[0008] Preferably, the optimal operating mode is matched by retrieving a pre-stored knowledge base based on the comprehensive indoor and outdoor feature vectors, including: The outdoor environmental parameters are encapsulated into an outdoor feature vector, and the outdoor feature vector is fused with the indoor pollutant parameters, indoor and outdoor temperature and humidity gradients, seasonal classification identifiers and 24-hour timestamps to construct an indoor and outdoor comprehensive feature vector that characterizes the current comprehensive indoor and outdoor environmental status. The target pollutant with the largest normalized concentration value among the indoor pollutant parameters is obtained, and the contribution weights corresponding to each dimension feature are retrieved from the weight database according to the type of the target pollutant. The indoor and outdoor comprehensive feature vectors are then mapped to the weighted feature space. Using the seasonal classification identifier and the timestamp in the weighted feature space as a first-level index, the candidate historical scene set is locked in the pre-stored knowledge base, and the weighted Euclidean distance between the mapped indoor and outdoor integrated feature vector and the feature vector of each candidate historical scene is calculated. Determine whether the minimum weighted Euclidean distance is less than a preset scene confidence threshold. If it is, determine that the current scene is successfully matched, and extract the historical operating mode corresponding to the minimum weighted Euclidean distance from the candidate historical scene set as the best operating mode in the current environment scene.

[0009] Preferably, obtaining the air quality improvement rate under different modes through mode polling includes: If the minimum weighted Euclidean distance is greater than or equal to the preset scenario confidence threshold, the main control module will initiate a mode polling process to control the fresh air actuator to switch to the corresponding operating states of the wind-grabbing mode and the internal circulation mode in sequence. The real-time collected comprehensive air quality index is denoised using a moving average filtering algorithm to generate a smoothed index evolution sequence. Perform a first-order difference operation on the index evolution sequence to obtain the dynamic change gradient of the comprehensive air quality index within a unit sampling time, and determine the statistical average value of the dynamic change gradient as the air quality improvement rate under the corresponding mode.

[0010] Preferably, determining the optimal operating mode for the current scenario based on the air quality improvement rate includes: The latest air quality improvement rate of the wind-catching mode and the internal circulation mode is obtained, and the mode corresponding to the larger value of the two is taken as the best operating mode in the current scenario. The newly determined optimal operating mode and its corresponding air quality improvement rate are associated and stored in the pre-stored knowledge base to complete the overwrite update of the original failed entries.

[0011] Preferably, triggering a relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base includes: During the execution of the optimal operating mode, the dynamically changing gradient is used as a real-time indicator to evaluate the actual fresh air efficiency. Calculate the percentage decrease in the actual fresh air efficiency relative to the efficiency recorded in the knowledge base; If the decrease ratio exceeds the preset attenuation threshold, it is determined that the performance of the current execution mode is lower than expected, and the main control module will re-trigger mode polling.

[0012] An air conditioning and fresh air control system based on multi-source sensing and scene self-learning, including a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator, also includes: The early warning switching module is used to acquire indoor pollutant parameters in real time through the multi-parameter sensing module, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to the preset weights. The weather forecast access module obtains external environmental forecast information, and the main control module controls the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. The scene retrieval module is used to obtain the current outdoor environmental parameters by the main control module when the comprehensive air quality index exceeds a set threshold, and to construct an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, the module retrieves the pre-stored knowledge base and matches the corresponding optimal operating mode. The polling learning module is used to obtain the air quality improvement rate under different modes by the main control module through mode polling if there is no matching mode in the knowledge base, and determine the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and store it in the knowledge base. The relearning module is used to continuously acquire the actual fresh air efficiency from the main control module during operation, and trigger the relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

[0013] One of the above technical solutions has the following advantages or beneficial effects: This invention overcomes the limitations of single-parameter assessment by using a multi-parameter sensing module to collect real-time indoor pollutant parameters and calculate the comprehensive air quality index based on preset weights. This allows the system to accurately reflect complex indoor pollution conditions and achieve a more scientific and reasonable air quality assessment. By acquiring external environmental forecast information through a weather forecast access module and combining it with the comprehensive air quality index results, the system controls the fresh air actuator to switch between pre-emptive airflow mode and internal circulation mode. This enables the system to predict future external pollution events, allowing it to store clean air in advance when external air quality is acceptable and proactively close off the system before pollution occurs, thus transforming passive response into proactive defense, effectively avoiding the risk of external pollution intrusion and extending filter lifespan. When the comprehensive air quality index exceeds a set threshold, the system... By acquiring current outdoor environmental parameters and combining them with indoor pollutant parameters, a comprehensive indoor and outdoor feature vector is constructed. Based on this feature vector, a pre-stored knowledge base is retrieved to match the corresponding optimal operating mode, enabling the system to quickly call upon historical experience to achieve scenario-based precise control. If no matching mode is found in the knowledge base, the air quality improvement rate under different modes is obtained through mode polling, and the optimal operating mode for the current scenario is determined accordingly. The results are then associated and stored in the knowledge base, enabling the system to learn from scratch and accumulate scenario knowledge. During operation, the actual fresh air efficiency is continuously acquired, and a relearning mechanism is triggered based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base to update the optimal operating mode in the knowledge base. This allows the system to adapt to environmental changes and equipment performance degradation, achieving continuous optimization of the control strategy. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0015] Figure 1 This is a flowchart of the air conditioning fresh air control method based on multi-source sensing and scene self-learning provided in the embodiments of the present invention; Figure 2 This is a schematic diagram of the structure of the air conditioning fresh air control system based on multi-source perception and scene self-learning provided in the embodiment of the present invention. Detailed Implementation

[0016] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0017] In this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0018] Air conditioning fresh air control method based on multi-source sensing and scene self-learning, such as Figure 1 As shown, a preferred embodiment of the present invention is applied to an air conditioner including a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator. The air conditioner fresh air control method includes: Step S1: The multi-parameter sensing module acquires indoor pollutant parameters in real time, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to preset weights; It should be noted that a multi-parameter sensing module refers to a hardware acquisition unit that integrates multiple sensors, including... Sensors include: a carbon dioxide sensor (to detect carbon dioxide concentration, reflecting indoor occupancy density and ventilation needs), a PM2.5 sensor (to detect fine particulate matter concentration, reflecting dust pollution levels), a formaldehyde (HCHO) sensor (to detect the concentration of pollutants from renovations), a total volatile organic compound (TVOC) sensor (to detect the overall pollution level of organic compounds), an odor sensor (such as a MOS gas sensor, used to detect kitchen fumes, pet odors, etc.), and a temperature and humidity sensor (to detect indoor thermal environment parameters). Indoor pollutant parameters refer to the digital quantities obtained after analog-to-digital conversion of the raw electrical signals collected in real time by the above sensors, representing the current concentration level of a specific pollutant. The main control module refers to an embedded microcontroller (MCU) or processor unit, responsible for performing data calculations, logical judgments, and generating control instructions. Preset weights refer to the influence coefficients pre-set for different pollutant types and application scenarios, used to quantify the contribution of each pollutant to overall air quality. The Comprehensive Air Quality Index (CAQI) is a single dimensionless index calculated by normalizing multiple pollutant parameters through a weighted fusion algorithm, used to uniformly represent the overall quality of indoor air.

[0019] Understandably, step S1 aims to overcome the biased assessment problem caused by single-parameter sensing in existing technologies. By deploying multiple types of sensors to construct a multi-dimensional sensing network, the system can simultaneously capture pollution information from different dimensions. By introducing a weighting mechanism, the system can dynamically adjust the assessment weight of each pollutant based on building characteristics, seasonal changes, or user preferences. For example, it can increase the weight of formaldehyde and TVOC in newly renovated scenarios, and increase the weight of formaldehyde and TVOC in densely populated conference room scenarios. The weights are calculated; the comprehensive air quality index is obtained through weighted fusion calculation, which transforms multidimensional and heterogeneous pollution data into a unified comparable indicator, enabling the system to truly reflect the complex indoor pollution situation and avoid ignoring other potential pollution risks due to the compliance of a single parameter. This provides a comprehensive and accurate input basis for subsequent control decisions, achieving the technical effect of moving from partial perception to comprehensive evaluation.

[0020] Step S2: Obtain external environmental forecast information through the weather forecast access module, and control the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. It should be noted that the weather forecast access module refers to a hardware unit with network communication capabilities. It connects to the internet via Wi-Fi, 4G / 5G, or other communication methods, and calls third-party meteorological service API interfaces (such as the China Meteorological Administration API or commercial meteorological service provider interfaces) to obtain weather and air quality forecast data for a specific future period. External environment forecast information refers to a data package containing predicted data on weather phenomena (such as sandstorms, haze, and rainfall), Air Quality Index (AQI), PM2.5 concentration, and pollen concentration for the next 2-6 hours. The "rush wind mode" refers to the operating state where, when the external air quality is acceptable, the fresh air actuator opens the external circulation valve and runs the fan at high power, introducing a large amount of clean outdoor air into the room for air replacement and storage. The internal circulation mode refers to the operating state where, after closing the external circulation valve, only indoor air is circulated and purified (if an air purification function is available) or kept closed. The fresh air actuator refers to an electromechanical actuator unit that includes a fan (used to drive airflow), a valve (used to switch between internal and external circulation ducts), and a filter (used to filter particulate matter).

[0021] Understandably, step S2 aims to overcome the lag problem of passive reactive control in existing technologies. By accessing weather forecast information, the system gains the ability to predict future external environmental conditions. By jointly determining the forecast information with the current comprehensive air quality index, the system can proactively activate a pre-emptive airflow mode to reserve clean air when current air quality is acceptable but a pollution event is imminent, avoiding the forced introduction of polluted air during peak pollution periods. By switching to internal circulation mode before pollution arrives, the system can actively close off and protect itself, blocking the intrusion paths of external pollutants. This proactive defensive control logic transforms the system from responding after pollution occurs to being prepared before pollution arrives, effectively mitigating external pollution risks, reducing filter wear under harsh conditions, and extending equipment maintenance cycles. This, in turn, ensures indoor air quality while reducing operating costs, achieving a shift from passive response to proactive defense.

[0022] Step S3: When the comprehensive air quality index exceeds the set threshold, the main control module obtains the current outdoor environmental parameters and constructs an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, the module retrieves the pre-stored knowledge base and matches the corresponding optimal operating mode. It should be noted that outdoor environmental parameters refer to the current external air quality data (such as PM2.5) obtained through outdoor sensors or weather forecast access modules. The data includes concentration and meteorological data (such as temperature and humidity). The integrated indoor and outdoor feature vector refers to a feature descriptor formed by fusing multi-dimensional information such as indoor pollutant parameters, outdoor environmental parameters, indoor and outdoor temperature and humidity gradients (the difference between indoor and outdoor temperature and humidity), seasonal classification identifiers (such as spring, summer, autumn, and winter), and 24-hour timestamps (such as morning, afternoon, and night). It is used to uniquely represent the current environmental scenario state. The pre-stored knowledge base refers to a structured database stored in non-volatile memory (such as Flash or EEPROM), containing historical scenario records and their corresponding optimal operating modes and fresh air efficiency data. The optimal operating mode refers to the operating state that has been historically verified to have the best air quality improvement effect under a specific scenario, including the wind-boosting mode or the internal circulation mode and its specific operating parameters (such as fan speed and operating time).

[0023] Understandably, step S3 aims to overcome the problem that fixed control strategies in existing technologies cannot adapt to diverse scenarios. When indoor air quality exceeds the standard, the system first constructs a vector representation that comprehensively characterizes the current environment, transforming complex scenario information into a computable data structure. By matching and retrieving the current feature vector with historical records in the knowledge base, the system can quickly reuse control strategies that have been validated in similar scenarios, avoiding the need to re-explore the optimal solution each time. This knowledge-reuse-based decision-making mechanism enables the system to adapt to different scenarios and address various pollution sources (such as those caused by gatherings of people). It automatically selects the most suitable operating mode according to different seasonal climate conditions and time periods, thereby shortening the response time and improving the stability of the control effect, realizing the technical effect of moving from mechanical solidification to scene adaptation.

[0024] Step S4: If there is no matching pattern in the knowledge base, the main control module obtains the air quality improvement rate under different modes through mode polling, determines the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and stores it in the knowledge base. It should be noted that mode polling refers to an exploratory process in which the main control module actively controls the fresh air actuator to sequentially switch to different operating modes (such as first the preemptive air mode and then the internal circulation mode), and runs in each mode for a fixed duration to test the effect. Air quality improvement rate refers to the decrease in the comprehensive air quality index per unit time, characterizing the system's ability to purify air under a specific mode; a higher rate indicates higher efficiency of that mode in the current scenario. Association storage refers to writing the current scenario feature vector, the determined optimal operating mode, and the air quality improvement rate under that mode as a complete record into the knowledge base, establishing a mapping relationship between scenarios and strategies.

[0025] Understandably, step S4 addresses the issues of cold start of the knowledge base and insufficient scenario coverage. When the system encounters a new, unseen scenario, it cannot rely on historical experience for decision-making. Instead, it proactively explores the actual effects of different strategies through pattern polling, objectively evaluating their performance by comparing the air quality improvement rates of each mode. The best-performing mode is then identified as the optimal solution for the current scenario and recorded in the knowledge base. This exploration-evaluation-recording learning mechanism enables the system to accumulate knowledge from scratch, eliminating the need for manual pre-setting of control strategies for all scenarios. As the system runs, the knowledge base expands, the types of scenarios it can handle become increasingly diverse, and decision-making gradually shifts from exploratory to reusable, continuously improving operational efficiency and achieving the technical effect of moving from reliance on manual configuration to autonomous experience accumulation.

[0026] Step S5: During operation, the main control module continuously acquires the actual fresh air efficiency and triggers a relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

[0027] It should be noted that the actual fresh air efficiency refers to the rate of air quality improvement monitored in real time by the system during the execution of the optimal operating mode, reflecting the current actual operating effect. The efficiency recorded in the knowledge base refers to the historically stored benchmark value of air quality improvement rate for this scenario. The deviation value refers to the percentage decrease or absolute difference between the actual fresh air efficiency and the recorded efficiency, used to measure whether the current mode is still suitable for the current scenario. The relearning mechanism refers to the system's automatic re-triggering of the mode polling process when the deviation value exceeds a preset attenuation threshold, re-exploring and determining a new optimal operating mode, and updating the corresponding record in the knowledge base as a self-optimization process.

[0028] Understandably, the purpose of step S5 is to address the strategy failure caused by dynamic environmental changes and equipment performance degradation. Building usage evolves over time (e.g., new furniture releases formaldehyde, newly built roads increase dust), and filters experience reduced ventilation efficiency due to dust accumulation. These factors gradually render historically validated optimal modes ineffective. By continuously monitoring actual operating efficiency and comparing it with historical benchmarks, the system can promptly detect strategy performance degradation. By triggering a relearning mechanism to re-explore the optimal strategy, the system can adapt to environmental changes, ensuring that control effectiveness remains at a high level. This closed-loop optimization mechanism of monitoring-diagnosis-relearning enables the system to evolve over the long term, preventing performance degradation due to knowledge base solidification, and achieving the technical effect of moving from static configuration to dynamic adaptation.

[0029] Preferably, calculating the comprehensive air quality index corresponding to the pollutant parameters according to preset weights includes: The measured values ​​of pollutant parameters collected by the multi-parameter sensing module are obtained, and the ratio relationship between each measured value and the preset safety threshold is established to generate a normalized parameter matrix that corresponds one-to-one with each pollutant parameter. Based on the timestamp information of the current environment and the characteristics of indoor and outdoor meteorological parameters, a set of weight coefficients matching the current environmental scenario is retrieved from the weight database, wherein the set of weight coefficients includes the weight ratio for different pollutant parameters; The normalized parameter in the normalized parameter matrix is ​​weighted and fused logically operated with its corresponding weight ratio in the weight coefficient set to obtain the comprehensive air quality index that characterizes the indoor air condition.

[0030] It should be noted that the measured value refers to the digital quantity output by each sensor in the multi-parameter sensing module after signal conditioning and analog-to-digital conversion, representing the actual concentration level of pollutants at the current moment. The preset safety threshold refers to the upper limit reference value of various pollutant concentrations set according to national standards (such as GB / T18883 "Indoor Air Quality Standard") or industry specifications, used to convert measured values ​​of different dimensions and orders of magnitude into comparable dimensionless relative values. The normalized parameter matrix is ​​a vector data structure formed by dividing the measured value of each pollutant by its corresponding safety threshold; each element in the matrix represents the degree of exceedance or safety margin of the pollutant relative to the safety limit. The timestamp information refers to the current date and time data provided by the system clock, including year, month, day, hour, and minute information, used to identify seasonal characteristics and diurnal rhythms. Indoor and outdoor meteorological parameter characteristics refer to the temperature and humidity values ​​and their differences obtained through indoor and outdoor temperature and humidity sensors, reflecting the current thermal environment state and ventilation potential. The weight database refers to a structured data table pre-burned into memory, containing records of weight coefficient sets configured for different scenario labels (such as season, time period, building type). The weight coefficient set refers to the set of weight values ​​assigned to each pollutant parameter under a specific scenario. The weight ratio refers to the relative proportion between the weight values, used to quantify the contribution of different pollutants to the comprehensive air quality assessment in the current scenario. The weighted fusion logic operation refers to the mathematical operation process of multiplying the normalized parameters by their corresponding weights and then summing the results, integrating multidimensional information through a linear weighting method.

[0031] Understandably, to address the issue of overly simplistic preset weight calculations failing to adapt to changing scenarios, a normalized parameter matrix is ​​generated by establishing a ratio between measured values ​​and safety thresholds. This system avoids the incomparability caused by differences in the dimensions of different pollutants, thus enabling... PM2.5 concentration Concentration, formaldehyde Concentration can be assessed on a uniform scale; by introducing timestamp information and indoor and outdoor meteorological parameter characteristics as search conditions, the system can retrieve a set of weight coefficients that highly match the current environmental scenario from the weight database, achieving dynamic configuration of weights rather than fixed weights; through weighted fusion logic operations, the normalized parameters are combined with scenario-based weights, and the comprehensive air quality index generated by the system reflects both the objective degree of pollutant exceedance and the subjective hazard weights of various pollutants in the current scenario, thus making the assessment results more in line with actual application needs and avoiding the bias caused by factors such as crowds in conference rooms. The technology aims to achieve a shift from static assessment to dynamic scenario adaptation, avoiding situations where ventilation needs are neglected due to excessively low formaldehyde weighting or the risk of indoor pollution in newly renovated homes.

[0032] For example, in a high-rise residential building application scenario near a road, the multi-parameter sensing module collects data... Measured value 800ppm (safe threshold 1000ppm, normalized value 0.8), PM2.5 measured value (Safety threshold) (Normalized value 1.857), measured formaldehyde value (Safety threshold) (Normalized value 0.75), forming a normalized parameter matrix. The system's current timestamp is for a winter evening, and the outdoor temperature is... ,humidity Indoor temperature ,humidity The set of weight coefficients for winter night scenes was retrieved from the weight database: , , (Due to the high incidence of smog in winter and the closed nature of doors and windows, the weight of PM2.5 and formaldehyde increases). Considering the significant temperature difference between indoors and outdoors... (Exceed (Threshold), the main control module will The weights were fine-tuned to 0.20. The weighted fusion calculation yielded: If the value exceeds the set threshold of 1.0, the system determines that fresh air intervention needs to be initiated. Since PM2.5 has the highest normalized value and the largest weight, PM2.5 is identified as the dominant pollutant.

[0033] Preferably, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode based on the external environment forecast information and the determination result of the comprehensive air quality index includes: If the external environment forecast information indicates that a target pollution event will occur within a preset time period in the future, the main control module determines that the current reserve conditions are met, controls the fresh air actuator to open the external circulation valve and adjusts the fan to operate at the first preset power, and enters the wind grabbing mode; The main control module calculates the warning countdown for the arrival of the target pollution event in real time, and controls the fresh air actuator to close the external circulation valve and start the internal circulation purification mode before the warning countdown reaches the preset switching threshold. During the operation of the internal circulation purification mode, the main control module adjusts the purification operation frequency of the fresh air actuator in real time according to the comprehensive air quality index until the external environmental forecast information indicates that the target pollution event has ended.

[0034] It should be noted that the future preset time period refers to a fixed time window calculated from the current moment, such as 2 hours, 4 hours, or 6 hours, set by the system based on the update frequency of weather forecast data and the typical duration of pollution events. The target pollution event refers to an abnormal external environmental condition that threatens indoor air quality, including specific weather phenomena such as sandstorms, heavy smog (AQI greater than 200), high concentrations of pollen, and industrial exhaust emissions. Reserve conditions refer to the trigger state where the system determines the need to reserve clean air in advance, requiring the simultaneous fulfillment of two sub-conditions: the current external air quality is acceptable and it is expected to deteriorate in the future. The first preset power refers to the fan operating at a higher speed to maximize ventilation efficiency and introduce clean air into the room; the power level is higher than that of daily maintenance ventilation. The warning countdown refers to the time difference calculated by the main control module based on the expected start time of the target pollution event and the current moment, representing the remaining time until the pollution arrives. The preset switching threshold refers to the safety margin set to allow time for damper operation and air balancing, such as 30 minutes or 15 minutes before the pollution arrives. Purification operation frequency refers to the periodic characteristics of the fan operating intermittently or in shifting modes under internal circulation mode. The higher the frequency, the greater the purification intensity.

[0035] Understandably, by setting a preset time period and detecting target pollution events within that period, the system specifies the timeframe for forward-looking defense, avoiding overreaction to forecasts that are too far in the future or insufficient response to impending threats. By determining reserve conditions and controlling the entry into emergency ventilation mode, the system uses a high-intensity air exchange at a first preset power when the outside air is still clean, quickly replacing indoor polluted air and reserving sufficient clean air. By calculating the warning countdown in real time and actively switching to internal circulation mode before reaching the preset switching threshold, the system allows sufficient time for the air valve to close and for indoor pressure to balance, ensuring that the system is closed before the pollution front arrives, avoiding the influx of pollutants due to late switching. By adjusting the purification operation frequency in real time according to the comprehensive air quality index during internal circulation, the system dynamically responds to changes in indoor air quality during periods of pollution, avoiding both insufficient purification leading to an increase in the index and excessive purification wasting energy, until the external pollution event ends and then returning to normal. The above time-segmented and intensity-dynamically adjusted control logic upgrades the system from simple mode switching to refined management throughout the entire process, achieving the effect of moving from extensive control to precise time-series management.

[0036] For example, the weather forecast access module receives a warning that a sandstorm is expected within the next two hours. The current external AQI is 80 (good), and the CAQI is 0.7 (not exceeding the standard). The main control module determines that the reserve conditions are met. The system controls the fresh air actuator to open the external circulation valve and adjust the fan to the first preset power (high level), entering the wind-fighting mode. The main control module calculates the warning countdown in real time as 110 minutes and continues wind-fighting operation. When the warning countdown reaches 30 minutes (preset switching threshold), the main control module controls the external circulation valve to close and the internal circulation purification mode to start. During the sandstorm (approximately 4 hours), the internal circulation purification mode dynamically adjusts according to the CAQI: when the index rises to 1.2, the purification frequency is increased to a high frequency; when the index drops to 0.8, it is reduced to a medium frequency; and when the index stabilizes below 0.6, it remains at a low frequency until the weather forecast indicates the sandstorm has ended, at which point the system returns to normal operation.

[0037] Preferably, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode based on the external environment forecast information and the determination result of the comprehensive air quality index includes: The main control module obtains the rated air exchange efficiency and indoor space volume of the fresh air actuator, and calculates the preset air-saving time required to reach the air reserve target based on the difference between the current comprehensive air quality index and the target cleanliness index. Based on the predicted start time of the target pollution event in the external environment forecast information, the wind-fighting trigger time is determined in combination with the preset wind-fighting duration; When the real-time moment reaches the wind-fighting trigger moment and the current outdoor environmental quality parameters are better than the preset access threshold, the fresh air actuator is controlled to enter the wind-fighting mode. When the warning countdown is less than or equal to the reserved switching margin, or when the currently monitored indoor air reserve reaches the saturation threshold, the fresh air actuator is controlled to perform a switching action from external circulation to internal circulation.

[0038] It should be noted that rated ventilation efficiency refers to the volume of indoor air that the fresh air actuator can replace per unit time under standard operating conditions, usually measured in cubic meters per hour, and determined by the fan performance curve and duct resistance characteristics. Indoor space volume refers to the geometric volume of the building space requiring ventilation, calculated by multiplying the room area by the floor height, representing the total amount of air that needs to be replaced. Target cleanliness index refers to the comprehensive air quality index target value that the system expects to achieve, usually set at a level below the safety threshold, representing the required cleanliness of the reserve air. Preset pre-emptive air intake duration refers to the theoretical pre-emptive air intake operation time calculated based on the ventilation efficiency, space volume, and index difference, representing the shortest time required to achieve the air reserve target. Expected start time refers to the specific time point in the weather forecast information when the target pollution event begins to occur. Pre-emptive air intake trigger time refers to the pre-emptive air intake start time point determined by the main control module after working backward from the expected start time and calculating the preset pre-emptive air intake duration, representing the latest time when pre-emptive air intake must begin. Preset access threshold refers to the upper limit of outdoor air quality that allows the external air circulation to be activated; external air is only allowed to be introduced when the outdoor environment is better than this threshold. Reserved switching margin refers to the time buffer reserved for damper operation and airflow stabilization, usually measured in minutes. Indoor air reserve capacity refers to the reserve completion indicator assessed by monitoring changes in the comprehensive air quality index or calculating the volume of clean air introduced. Saturation threshold refers to the upper limit of the reserve capacity reaching the expected target; continuing to increase airflow will not bring additional benefits.

[0039] Understandably, by introducing rated ventilation efficiency and indoor space volume parameters, the system transforms the pre-emptive ventilation duration from an experience-based setting to a calculation based on a physical model, enabling the preset pre-emptive ventilation duration to reflect actual ventilation capacity and space requirements. By combining the expected start time of the target pollution event with the preset pre-emptive ventilation duration to determine the pre-emptive ventilation trigger time, the system precisely correlates the pre-emptive ventilation initiation timing with the pollution arrival time, ensuring that the reserve task is completed within a limited time window. By setting a preset access threshold as an additional condition for pre-emptive ventilation initiation, the system avoids erroneously opening the external circulation when the outdoor environment is already polluted. By setting a warning countdown and reserved switching margin as one of the switching conditions, the system ensures the lead time for the closure action. By introducing indoor air reserve monitoring and saturation thresholds, the system adds reserve completion as a basis for switching, avoiding excessive pre-emptive ventilation that wastes energy or insufficient reserves that affect the protective effect. The above control logic based on physical calculation and multiple condition judgments upgrades the pre-emptive ventilation strategy from qualitative description to quantitative optimization, achieving the effect of moving from experience-driven to model-driven.

[0040] For example, the rated air exchange efficiency of the fresh air actuator is 300 cubic meters per hour, the indoor space volume is 150 cubic meters (area 50 square meters, floor height 3 meters), the current comprehensive air quality index is 1.5, and the target clean air index is set to 0.8. The main control module calculates the preset pre-emptive air exchange time using the first implementation method: Preset pre-emptive air exchange time = Minutes. The weather forecast indicates that the target pollution event (severe smog) is expected to begin at 14:00, so the emergency air recirculation trigger time is determined to be 13:41. When the real-time time reaches 13:41 and the current outdoor AQI is 90 (better than the preset threshold of 120), the system control enters the emergency air recirculation mode. During the emergency air recirculation operation, the main control module continuously monitors the warning countdown and indoor air reserve (estimated in real time through the comprehensive air quality index). When the warning countdown is less than or equal to 15 minutes (with reserved switching margin) or the reserve is detected to reach the saturation threshold (comprehensive air quality index drops below 0.8 and remains stable), the control executes the switching action from external circulation to internal circulation.

[0041] Preferably, the optimal operating mode is matched by retrieving a pre-stored knowledge base based on the comprehensive indoor and outdoor feature vectors, including: The outdoor environmental parameters are encapsulated into an outdoor feature vector, and the outdoor feature vector is fused with the indoor pollutant parameters, indoor and outdoor temperature and humidity gradients, seasonal classification identifiers and 24-hour timestamps to construct an indoor and outdoor comprehensive feature vector that characterizes the current comprehensive indoor and outdoor environmental status. The target pollutant with the largest normalized concentration value among the indoor pollutant parameters is obtained, and the contribution weights corresponding to each dimension feature are retrieved from the weight database according to the type of the target pollutant. The indoor and outdoor comprehensive feature vectors are then mapped to the weighted feature space. Using the seasonal classification identifier and the timestamp in the weighted feature space as a first-level index, the candidate historical scene set is locked in the pre-stored knowledge base, and the weighted Euclidean distance between the mapped indoor and outdoor integrated feature vector and the feature vector of each candidate historical scene is calculated. Determine whether the minimum weighted Euclidean distance is less than a preset scene confidence threshold. If it is, determine that the current scene is successfully matched, and extract the historical operating mode corresponding to the minimum weighted Euclidean distance from the candidate historical scene set as the best operating mode in the current environment scene.

[0042] It should be noted that the outdoor feature vector refers to the vector containing outdoor temperature, outdoor humidity, outdoor PM2.5 concentration, and outdoor humidity. Outdoor environmental parameters, such as concentration, are arranged in a fixed order to form a numerical vector, used to characterize the current state of the external environment. The indoor and outdoor temperature and humidity gradient refers to a binary tuple composed of the difference between indoor and outdoor temperature (temperature gradient) and the difference between indoor and outdoor humidity (humidity gradient), reflecting the potential impact of indoor and outdoor thermal and humidity differences on ventilation efficiency. Seasonal classification identifiers refer to the seasonal category (spring, summer, autumn, winter) determined based on the current date, used to capture seasonal environmental patterns. 24-hour timestamps refer to category identifiers after dividing the day into several time periods (e.g., 6-9 am, 9-12 am, 12-2 pm, 2-6 pm, 6-10 pm, 10-6 am the next day), used to capture diurnal behavior patterns. The indoor and outdoor comprehensive feature vector refers to a high-dimensional feature descriptor formed by splicing together multi-dimensional information such as outdoor feature vectors, indoor pollutant parameters, temperature and humidity gradients, seasonal identifiers, and timestamps in a predetermined format, comprehensively depicting the current environmental scenario. The target pollutant refers to the pollutant type with the highest value among the normalized indoor pollutant parameters, representing the most significant current source of pollution. Contribution weight refers to the weighting of various dimensions of characteristics (such as the effect of temperature and humidity gradients) for different target pollutant types. The importance coefficients in similarity calculations are as follows: Scene is more important (outdoor PM2.5 is more important than particulate matter scenes). The weighted feature space refers to the multi-dimensional mathematical space containing feature vectors adjusted for contribution weights, where similar scenes are geometrically closer. The first-level index refers to the combination of primary key fields used in the knowledge base for quickly locating candidate records. The candidate historical scene set refers to a subset of potentially similar historical records filtered through the first-level index. Weighted Euclidean distance is a measure of the geometric distance between two feature vectors in the weighted feature space; the smaller the distance, the higher the scene similarity. The scene confidence threshold is the upper limit for determining a successful match; if the distance is below this threshold, the scenes are considered sufficiently similar, and historical experience can be directly reused.

[0043] Understandably, by encapsulating outdoor environmental parameters into outdoor feature vectors and fusing them with indoor parameters, gradient information, and temporal identifiers, the system constructs a comprehensive indoor-outdoor feature vector that encompasses the internal and external environmental states, dynamic differences, and spatiotemporal background, forming a comprehensive representation of the scene. By identifying target pollutants and retrieving their corresponding contribution weights, the system can dynamically adjust the importance of each dimension for different pollution types, avoiding over-reliance on external pollutants. The system avoids over-focusing on outdoor PM2.5 in dominant scenarios or over-focusing on temperature and humidity gradients in particulate matter-dominated scenarios. By mapping feature vectors to a weighted feature space, the system calculates distances within a specific similarity metric framework, adapting the matching criteria to the scenario type. By using seasons and timestamps as primary indexes to quickly lock the candidate set, the system avoids costly distance calculations across the entire database, improving retrieval efficiency. Through joint determination of weighted Euclidean distance and scenario confidence thresholds, the system can both quantify the degree of scenario similarity and set an acceptable lower limit for similarity, ensuring that reused historical experience is indeed comparable to the current scenario. The above hierarchical retrieval, dynamic weighting, and threshold determination matching mechanism upgrades knowledge base querying from simple comparison to intelligent similarity retrieval.

[0044] For example, the outdoor environmental parameter is temperature. Humidity 50%, PM2.5 concentration The data is encapsulated as an outdoor feature vector [18, 50, 120]; indoor pollutant parameters show a normalized formaldehyde value of 1.8. With a normalized value of 0.9 and a normalized value of 0.6 for PM2.5, formaldehyde was identified as the target pollutant. The indoor and outdoor temperature and humidity gradient was determined by temperature rise. (indoor The humidity rise is 5% (55% indoors); the season classification is spring, and the timestamp is evening. An indoor / outdoor integrated feature vector is constructed [18, 50, 120, 1.8, 0.9, 0.6, 4, 5, spring code, evening code]. Based on the formaldehyde dominance type, contribution weights are retrieved from the weight database: time weight 0.4, season weight 0.3, temperature and humidity gradient weight 0.2, and outdoor feature weight 0.1. The vector is then mapped to a weighted feature space. Using spring and evening as primary indices, a candidate historical scene set (containing 20 records for that time period) is locked in the knowledge base. The weighted Euclidean distance between the current vector and each candidate vector is calculated. The minimum distance is 0.15, which is less than the preset scene confidence threshold of 0.25, indicating a successful match. The corresponding historical operating mode is extracted as the internal circulation mode (because external PM2.5 is higher in spring evenings and formaldehyde release peaks at night due to temperature influences, historical experience shows that internal circulation combined with purification has a better effect), and this is used as the current optimal operating mode.

[0045] Preferably, obtaining the air quality improvement rate under different modes through mode polling includes: If the minimum weighted Euclidean distance is greater than or equal to the preset scenario confidence threshold, the main control module will initiate a mode polling process to control the fresh air actuator to switch to the corresponding operating states of the wind-grabbing mode and the internal circulation mode in sequence. The real-time collected comprehensive air quality index is denoised using a moving average filtering algorithm to generate a smoothed index evolution sequence. Perform a first-order difference operation on the index evolution sequence to obtain the dynamic change gradient of the comprehensive air quality index within a unit sampling time, and determine the statistical average value of the dynamic change gradient as the air quality improvement rate under the corresponding mode.

[0046] It should be noted that the pattern polling process refers to the system's proactive exploration of different operating modes' effects when knowledge base matching fails. This process accumulates new scenario knowledge by sequentially running candidate modes and recording their performance. The moving average filtering algorithm is a digital signal processing method that performs local averaging on time-series data. It effectively suppresses random noise interference by calculating the arithmetic mean of several sampling points before and after the current moment as the output. The exponential evolution sequence refers to a smoothed data sequence of the composite air quality index changing over time after filtering, reflecting the true trend of air quality changes rather than instantaneous fluctuations. The first-order difference operation is a mathematical operation that calculates the difference between adjacent sampling points in a discrete time series, i.e., subtracting the previous moment's value from the current moment's value, used to obtain the instantaneous rate of change of the sequence. The dynamic gradient refers to the change in the composite air quality index within a unit sampling time. A positive value indicates an increase in the index (deteriorating air quality), a negative value indicates a decrease in the index (improving air quality), and a larger absolute value indicates a more drastic change. The statistical average refers to calculating the arithmetic mean or weighted average of multiple dynamic gradient values ​​over a period of time, used to smooth random fluctuations and obtain a stable rate estimate.

[0047] Understandably, by using the comparison of the minimum weighted Euclidean distance with the scene confidence threshold as the trigger condition for pattern polling, the system ensures that the exploration process is only initiated when the knowledge base truly cannot provide reliable experience, avoiding unnecessary test runs. By controlling the fresh air actuator to sequentially switch between the rush mode and the internal circulation mode, the system conducts comparative experiments on the two main strategies, providing comparable data for subsequent decision-making. By using a moving average filtering algorithm to denoise the original exponential data, the system eliminates random interference such as sensor measurement noise and airflow fluctuations, ensuring that the exponential evolution sequence truly reflects the trend of air quality changes. By performing first-order difference operations on the smoothed sequence, the system transforms the cumulative exponential changes into instantaneous gradient changes, capturing the real-time dynamics of the improvement effect. By determining the statistical average of the dynamic gradient changes as the air quality improvement rate, the system obtains a stable estimate of the purification capacity under this mode while suppressing residual noise. The above rate calculation mechanism with noise suppression and trend extraction upgrades the model evaluation from direct calculation of raw data to robust estimation enhanced by signal processing, achieving a technical effect from being susceptible to interference to being stable and reliable.

[0048] For example, in a newly renovated residential application scenario, if knowledge base matching fails (minimum weighted Euclidean distance 0.32, greater than the scenario confidence threshold of 0.25), the main control module initiates a mode polling process. First, the fresh air actuator switches to "rush air" mode for 15 minutes, during which the raw value of the comprehensive air quality index is collected every minute. After passing through a moving average filter (window length 5 minutes), a smoothed index evolution sequence is generated, showing the index gradually decreasing from 1.5 to 1.1. A first-order difference operation is performed on the smoothed sequence to obtain the dynamic gradient of each minute (all negative values, indicating improvement). The statistical average is calculated, yielding an air quality improvement rate of -0.027 (per index per minute) in the rush air mode. Subsequently, the system switches to internal circulation mode for 15 minutes, and similarly undergoes filtering, differencing, and averaging, resulting in an air quality improvement rate of -0.015 in the internal circulation mode. Comparing the rates of the two modes, the rush air mode shows faster improvement (larger absolute rate value).

[0049] Preferably, determining the optimal operating mode for the current scenario based on the air quality improvement rate includes: The latest air quality improvement rate of the wind-catching mode and the internal circulation mode is obtained, and the mode corresponding to the larger value of the two is taken as the best operating mode in the current scenario. The newly determined optimal operating mode and its corresponding air quality improvement rate are associated and stored in the pre-stored knowledge base to complete the overwrite update of the original failed entries.

[0050] It should be noted that the larger value refers to the rate of air quality improvement between the wind-catching mode and the internal circulation mode, with the larger absolute value (since the improvement rate is a negative value, a larger absolute value indicates faster improvement), corresponding to the operating mode with higher purification efficiency in this scenario. "Original invalid entries" refer to historical records in the knowledge base that have the same or similar characteristics as the current scenario but whose recorded optimal operating mode is no longer optimal after this polling verification, or logically considered invalid vacancies when there are no records for the current scenario in the knowledge base. "Overwrite update" means replacing the original records in the knowledge base with the newly determined optimal operating mode and rate data from this polling, or adding new records when none exist, ensuring that the knowledge base always retains the latest verified optimal strategy for this scenario.

[0051] Understandably, by acquiring the latest air quality improvement rates of the two modes and comparing them numerically, the system uses objective performance indicators as the basis for decision-making, determining the mode with faster improvement (larger absolute rate) as the optimal operating mode for the current scenario, avoiding suboptimal choices caused by subjective judgment or fixed preferences. By associating the newly determined optimal operating mode and its corresponding air quality improvement rate into a pre-stored knowledge base, the system completes the recording or updating of knowledge for this scenario, transforming the results of this exploration into reusable experience. Through a clear overlay and update mechanism, the system ensures that the records in the knowledge base always reflect the latest verified results, rather than outdated historical data. When environmental changes cause the original strategy to fail, it can be corrected in a timely manner through relearning. The above decision-making and real-time update mechanism based on performance comparison upgrades knowledge accumulation from simple data recording to continuous optimization of experience, achieving the effect of dynamic evolution from static storage.

[0052] For example, the main control module obtains the air quality improvement rate as -0.027 for the "rushing wind" mode and -0.015 for the "internal circulation" mode. Comparing the absolute values, 0.027 is greater than 0.015, and the "rushing wind" mode is determined to be the optimal operating mode for the current scenario. The main control module associates the scenario feature vector, the optimal operating mode identifier "rushing wind," and the improvement rate of -0.027 into a pre-stored knowledge base. If the knowledge base already contains historical records for this scenario (spring, evening, formaldehyde-dominated) showing that the optimal mode is internal circulation, then this record overwrites the original invalid entry and updates it to "rushing wind" mode, reflecting the current optimal strategy where the "rushing wind" effect is better than internal circulation due to the relatively good external air quality (AQI 80).

[0053] Preferably, triggering a relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base includes: During the execution of the optimal operating mode, the dynamically changing gradient is used as a real-time indicator to evaluate the actual fresh air efficiency. Calculate the percentage decrease in the actual fresh air efficiency relative to the efficiency recorded in the knowledge base; If the decrease ratio exceeds the preset attenuation threshold, it is determined that the performance of the current execution mode is lower than expected, and the main control module will re-trigger mode polling.

[0054] It should be noted that the actual fresh air efficiency refers to the current rate of improvement calculated by real-time monitoring of changes in the comprehensive air quality index during the actual execution of the optimal operating mode, reflecting the true performance of the mode under current actual conditions. Recorded efficiency refers to the baseline value of the historically verified air quality improvement rate stored in the knowledge base for this scenario, representing the performance level of the mode at a certain point in the past. Deviation value refers to the degree of difference between the actual fresh air efficiency and the recorded efficiency; in this embodiment, it is specifically quantified as the percentage decrease in actual efficiency relative to the recorded efficiency. The preset attenuation threshold is the critical percentage for determining whether performance has significantly degraded; when the percentage decrease exceeds this threshold, the original strategy is considered no longer suitable for the current situation.

[0055] Understandably, by continuously monitoring the system's actual performance with dynamically changing gradients as real-time indicators during the execution of the optimal operating mode, the system obtains immediate feedback on the effectiveness of the strategy; by calculating the percentage decrease in actual fresh air efficiency relative to recorded efficiency, the system quantifies the degree of performance degradation, transforming absolute changes into relative deviations to facilitate the setting of unified judgment criteria; by comparing the percentage decrease with a preset degradation threshold, the system establishes clear performance degradation judgment criteria, avoiding overreaction to minor fluctuations or ignoring significant degradation; when the performance is determined to be lower than expected, the mode polling is re-triggered, and the system proactively initiates a re-exploration process to find a new optimal operating mode to replace the old, ineffective strategy.

[0056] An air conditioning fresh air control system based on multi-source sensing and scene self-learning, such as Figure 2 As shown, the air conditioner includes a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator, and also includes: The early warning switching module is used to acquire indoor pollutant parameters in real time through the multi-parameter sensing module, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to the preset weights. The weather forecast access module obtains external environmental forecast information, and the main control module controls the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. The scene retrieval module is used to obtain the current outdoor environmental parameters by the main control module when the comprehensive air quality index exceeds a set threshold, and to construct an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, the module retrieves the pre-stored knowledge base and matches the corresponding optimal operating mode. The polling learning module is used to obtain the air quality improvement rate under different modes by the main control module through mode polling if there is no matching mode in the knowledge base, and determine the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and store it in the knowledge base. The relearning module is used to continuously acquire the actual fresh air efficiency from the main control module during operation, and trigger the relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

[0057] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0058] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for controlling fresh air in air conditioning based on multi-source sensing and scene self-learning, characterized in that, The air conditioning fresh air control method, applied to an air conditioner including a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator, includes: The multi-parameter sensing module acquires indoor pollutant parameters in real time, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to preset weights. The weather forecast access module obtains external environmental forecast information, and the main control module controls the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. When the comprehensive air quality index exceeds the set threshold, the main control module obtains the current outdoor environmental parameters and constructs an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, it searches the pre-stored knowledge base and matches the corresponding optimal operating mode. If there is no matching pattern in the knowledge base, the main control module obtains the air quality improvement rate under different modes through mode polling, determines the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and stores it in the knowledge base. During operation, the main control module continuously acquires the actual fresh air efficiency and triggers a relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

2. The intelligent control method for air conditioning fresh air based on multi-source sensing and scene self-learning as described in claim 1, characterized in that, The comprehensive air quality index corresponding to the pollutant parameters is calculated based on preset weights, including: The measured values ​​of pollutant parameters collected by the multi-parameter sensing module are obtained, and the ratio relationship between each measured value and the preset safety threshold is established to generate a normalized parameter matrix that corresponds one-to-one with each pollutant parameter. Based on the timestamp information of the current environment and the characteristics of indoor and outdoor meteorological parameters, a set of weight coefficients matching the current environmental scenario is retrieved from the weight database, wherein the set of weight coefficients includes the weight ratio for different pollutant parameters; The normalized parameter in the normalized parameter matrix is ​​weighted and fused logically operated with its corresponding weight ratio in the weight coefficient set to obtain the comprehensive air quality index that characterizes the indoor air condition.

3. The intelligent control method for air conditioning fresh air based on multi-source sensing and scene self-learning according to claim 1, characterized in that, Based on the external environment forecast information and the determination result of the comprehensive air quality index, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode includes: If the external environment forecast information indicates that a target pollution event will occur within a preset time period in the future, the main control module determines that the current reserve conditions are met, controls the fresh air actuator to open the external circulation valve and adjusts the fan to operate at the first preset power, and enters the wind grabbing mode; The main control module calculates the warning countdown for the arrival of the target pollution event in real time, and controls the fresh air actuator to close the external circulation valve and start the internal circulation purification mode before the warning countdown reaches the preset switching threshold. During the operation of the internal circulation purification mode, the main control module adjusts the purification operation frequency of the fresh air actuator in real time according to the comprehensive air quality index until the external environmental forecast information indicates that the target pollution event has ended.

4. The air conditioning fresh air control method based on multi-source sensing and scene self-learning according to claim 1, characterized in that, Based on the external environment forecast information and the determination result of the comprehensive air quality index, controlling the fresh air actuator to switch between the rush mode and the internal circulation mode includes: The main control module obtains the rated air exchange efficiency and indoor space volume of the fresh air actuator, and calculates the preset air-saving time required to reach the air reserve target based on the difference between the current comprehensive air quality index and the target cleanliness index. Based on the predicted start time of the target pollution event in the external environment forecast information, the wind-fighting trigger time is determined in combination with the preset wind-fighting duration; When the real-time moment reaches the wind-fighting trigger moment and the current outdoor environmental quality parameters are better than the preset access threshold, the fresh air actuator is controlled to enter the wind-fighting mode. When the warning countdown is less than or equal to the reserved switching margin, or when the currently monitored indoor air reserve reaches the saturation threshold, the fresh air actuator is controlled to perform a switching action from external circulation to internal circulation.

5. The air conditioning fresh air control method based on multi-source sensing and scene self-learning according to claim 1, characterized in that, Based on the comprehensive indoor and outdoor feature vectors, a pre-stored knowledge base is retrieved to match the corresponding optimal operating modes, including: The outdoor environmental parameters are encapsulated into an outdoor feature vector, and the outdoor feature vector is fused with the indoor pollutant parameters, indoor and outdoor temperature and humidity gradients, seasonal classification identifiers and 24-hour timestamps to construct an indoor and outdoor comprehensive feature vector that characterizes the current comprehensive indoor and outdoor environmental status. The target pollutant with the largest normalized concentration value among the indoor pollutant parameters is obtained, and the contribution weights corresponding to each dimension feature are retrieved from the weight database according to the type of the target pollutant. The indoor and outdoor comprehensive feature vectors are then mapped to the weighted feature space. Using the seasonal classification identifier and the timestamp in the weighted feature space as a first-level index, the candidate historical scene set is locked in the pre-stored knowledge base, and the weighted Euclidean distance between the mapped indoor and outdoor integrated feature vector and the feature vector of each candidate historical scene is calculated. Determine whether the minimum weighted Euclidean distance is less than a preset scene confidence threshold. If it is, determine that the current scene is successfully matched, and extract the historical operating mode corresponding to the minimum weighted Euclidean distance from the candidate historical scene set as the best operating mode in the current environment scene.

6. The air conditioning fresh air control method based on multi-source sensing and scene self-learning according to claim 5, characterized in that, The air quality improvement rate under different modes is obtained through mode polling, including: If the minimum weighted Euclidean distance is greater than or equal to the preset scenario confidence threshold, the main control module will initiate a mode polling process to control the fresh air actuator to switch to the corresponding operating states of the wind-grabbing mode and the internal circulation mode in sequence. The real-time collected comprehensive air quality index is denoised using a moving average filtering algorithm to generate a smoothed index evolution sequence. Perform a first-order difference operation on the index evolution sequence to obtain the dynamic change gradient of the comprehensive air quality index within a unit sampling time, and determine the statistical average value of the dynamic change gradient as the air quality improvement rate under the corresponding mode.

7. The air conditioning fresh air control method based on multi-source sensing and scene self-learning according to claim 6, characterized in that, Determining the optimal operating mode for the current scenario based on the air quality improvement rate includes: The latest air quality improvement rate of the wind-catching mode and the internal circulation mode is obtained, and the mode corresponding to the larger value of the two is taken as the best operating mode in the current scenario. The newly determined optimal operating mode and its corresponding air quality improvement rate are associated and stored in the pre-stored knowledge base to complete the overwrite update of the original failed entries.

8. The air conditioning fresh air control method based on multi-source sensing and scene self-learning according to claim 6, characterized in that, Based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base, a relearning mechanism is triggered to update the optimal operating mode in the knowledge base, including: During the execution of the optimal operating mode, the dynamically changing gradient is used as a real-time indicator to evaluate the actual fresh air efficiency. Calculate the percentage decrease in the actual fresh air efficiency relative to the efficiency recorded in the knowledge base; If the decrease ratio exceeds the preset attenuation threshold, it is determined that the performance of the current execution mode is lower than expected, and the main control module will re-trigger mode polling.

9. An air conditioning fresh air control system based on multi-source sensing and scene self-learning, characterized in that, The air conditioning system includes a multi-parameter sensing module, a weather forecast access module, and a fresh air actuator, and also includes: The early warning switching module is used to acquire indoor pollutant parameters in real time through the multi-parameter sensing module, and the main control module calculates the comprehensive air quality index corresponding to the pollutant parameters according to the preset weights. The weather forecast access module obtains external environmental forecast information, and the main control module controls the fresh air actuator to switch between wind-catching mode and internal circulation mode based on the external environmental forecast information and the determination result of the comprehensive air quality index. The scene retrieval module is used to obtain the current outdoor environmental parameters by the main control module when the comprehensive air quality index exceeds a set threshold, and to construct an indoor and outdoor comprehensive feature vector by combining the indoor pollutant parameters. Based on the indoor and outdoor comprehensive feature vector, the module retrieves the pre-stored knowledge base and matches the corresponding optimal operating mode. The polling learning module is used to obtain the air quality improvement rate under different modes by the main control module through mode polling if there is no matching mode in the knowledge base, and determine the best operating mode in the current scenario based on the magnitude of the air quality improvement rate, and store it in the knowledge base. The relearning module is used to continuously acquire the actual fresh air efficiency from the main control module during operation, and trigger the relearning mechanism to update the optimal operating mode in the knowledge base based on the deviation between the actual fresh air efficiency and the efficiency recorded in the knowledge base.

Images