Room air conditioner heat and humidity load abnormity monitoring and intelligent control method and system

Abstract

The invention discloses a room air conditioner heat and humidity load abnormity monitoring and intelligent control method and system, and belongs to the technical field of air conditioner intelligent control. The method comprises the steps that indoor and outdoor environment parameters and air conditioner operation data are collected through a sensor, and a data set with labels is constructed; cloud training is carried out based on an improved lightweight Informer model, and the model is rapidly migrated to a target room in combination with a migration learning adaptation module; and the training model is deployed to an air conditioner controller, abnormal working conditions such as windowing, multi-person gathering and high heat sources are recognized in real time, and control strategies such as an energy-saving mode and fresh air adjustment are executed. The system comprises a multi-source environmental parameter sensor module, a data processing module, an abnormity identification module, a control execution module and a communication feedback module. Through time sequence modeling and transfer learning, the problem that a traditional air conditioning system cannot dynamically sense user behaviors is solved, and the energy consumption efficiency and comfort are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and room air conditioner control technology, and in particular to a room air conditioner heat and humidity load abnormality monitoring and intelligent control method and system. Background Art

[0002] With the development of smart home technology, room air conditioners, as the main source of building energy consumption, are gradually evolving from traditional fixed set value control to intelligent, data-driven prediction and adjustment methods. During the use of room air conditioners, due to users' frequent window opening, door opening, gathering activities, and activation of additional heat or humidity sources, the indoor heat and humidity load will suddenly change, causing the system energy consumption to increase, the equipment operation efficiency to decrease, and even adversely affecting the indoor air quality and comfort. Traditional air conditioning control systems are mostly based on PID control logic with fixed set values ​​(such as Chinese patent document CN119713504A), and the compressor frequency or air valve opening is adjusted by the deviation between the return air temperature and the set value. However, this method has significant defects: first, it cannot dynamically perceive the load disturbance caused by user behavior (such as opening windows, gathering of many people) or sudden changes in the environment (such as high heat source, high humidity source), resulting in energy waste and reduced comfort; second, it relies on manual experience to adjust the set value and lacks data-driven adaptive capabilities.

[0003] In addition, traditional air conditioning control systems mainly rely on simple temperature and humidity closed-loop control logic, which cannot accurately identify and respond to the above abnormal conditions, especially in unattended or energy-saving scenarios, and it is difficult to effectively perceive and adaptively adjust user behavior. Some studies have begun to try to use machine learning models for condition identification, but most of them rely on static features or short-term behavior patterns, which makes it difficult to capture the temporal evolution characteristics of user behavior.

[0004] In recent years, long-term series modeling methods, especially the Informer (Information Aggregator for Long-Term Series Forecasting) model based on the Transformer architecture, have shown superior performance in the fields of power load forecasting and meteorological change modeling. However, the original Informer model has a large number of structural parameters and relies on large-scale data training, which is not suitable for resource-constrained edge control devices. At the same time, different rooms have structural differences and usage behavior differences, which makes it difficult for a single model to be generalized to multiple scenarios.

[0005] Therefore, how to build a lightweight, portable anomaly recognition model that supports multi-room personalized modeling and work in conjunction with the on-site air-conditioning operation system has become a key issue that needs to be solved urgently. Summary of the invention

[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and to provide a method and system for monitoring and intelligently controlling abnormal heat and humidity loads of room air conditioners.

[0007] The purpose of the present invention is achieved by at least one of the following technical solutions.

[0008] A room air conditioner heat and humidity load abnormality monitoring and intelligent control method, comprising the following steps: (1) Collect indoor and outdoor environmental parameters of typical rooms and air conditioning operation data, build a labeled abnormal operating condition data set, and upload it to the cloud server; (2) Based on the abnormal working condition data set, an improved Informer model is used for cloud-based modeling training to obtain a lightweight working condition classification model; (3) Based on the working condition classification model, a personalized model is obtained through transfer learning adaptation module and deployed to the air conditioning intelligent controller of the target room to predict the abnormal working condition type in real time; (4) Execute the corresponding air-conditioning operation mode control strategy according to the prediction results. The cloud server iteratively optimizes the adaptation module based on user feedback data to achieve model optimization.

[0009] Furthermore, in step (1), the environmental parameters include indoor temperature (return air temperature) and relative humidity (return air relative humidity), CO 2 concentration , Outdoor temperature and relative humidity and infrared information of human activities ; Air conditioning operation data including air conditioner power , Air conditioner set temperature 、Air conditioner fan speed setting and timestamp ; The abnormal operating conditions include open windows, open doors, high heat sources, high humidity sources, and large gatherings of people.

[0010] Furthermore, in step (2), the encoder of the improved Informer model is a single-layer structure, and the number of attention heads is set to 2-4 (preferably 2); and the multi-layer decoder is removed, and only the predicted label of the current moment or the next moment is output.

[0011] Furthermore, in step (2), the improved Informer model is trained using a weighted cross entropy loss function to solve the problem of imbalanced samples in the working condition category. The loss function is: ; in, is the total number of samples for batch training, is the total number of categories ( , 7 categories in total: 0 to 6), For the The weight of the class, , For the The true labels of samples, For the model Samples are predicted as categories probability.

[0012] Furthermore, in step (3), the adaptation module is used to realize room-level model migration and personalized recognition based on the shared Informer model backbone, and adapt the trained working condition classification model to the target room; the adaptation module structure is: ; in, is the down-projection weight matrix, is the up-projection weight matrix, the activation function is ReLU, For input, For the adapter module; and The random initialization of uses Gaussian distribution, ; ; is the standard deviation of the Gaussian distribution.

[0013] Furthermore, in step (3), the step of obtaining the personalized model and deploying it to the air conditioning intelligent controller of the target room specifically includes: For the working condition classification model obtained in step (2), the core parameters of the model encoder are set to the shared frozen layer, and the initialized adaptation module is inserted into the output layer of the encoder to obtain the basic model; Deploy the basic model to the air-conditioning intelligent controller of the target room, activate only the adaptation module parameters, and freeze all other layers; collect the indoor and outdoor environmental parameters of the target room and the air-conditioning operation data to build a small-scale training set; The adaptation module is trained by transferring initialization and fine-tuning parameters on the constructed small-scale training set; After training is completed, the adaptation module and the basic model are merged into a personalized model, and the personalized model is deployed to the air-conditioning intelligent controller.

[0014] Furthermore, in step (4), the control strategy includes: When the prediction is that the window or door is open, if the continuous detection time exceeds the set threshold, it will switch to energy-saving mode; When a large gathering of people is predicted, increase the fresh air volume or the air conditioning speed; When a high heat source or a high humidity source is predicted, the set temperature is adjusted or the dehumidification function is turned on respectively.

[0015] Furthermore, in step (4), the feedback data includes the recognition results and control logs annotated by the user, and the model optimization is achieved by regularly updating the adaptation module parameters through the cloud server.

[0016] The present invention also provides a room air-conditioning heat and humidity load abnormality monitoring and intelligent control system, which comprises: Multi-source environmental parameter sensor module, used to collect indoor and outdoor environmental parameters and air conditioning operation data; Data processing module, which filters, normalizes and constructs features of data; Cloud server, used for lightweight working condition classification model training, adaptation module optimization and data management; An anomaly recognition module, embedded with the adaptation module and the personalized model, is deployed through edge computing devices to support real-time reasoning and control decisions; The control execution module adjusts the air-conditioning operation mode according to the abnormality recognition result; Communication and feedback module, used for data uploading, model updating and user interaction.

[0017] Compared with the prior art, the present invention has the following advantages and technical effects: (1) Significantly reduce computing resource requirements by simplifying the multi-layer encoder of the original Informer model to a single layer and reducing the number of attention heads from 8 to 2, meeting the real-time requirements of edge devices (such as air conditioning controllers).

[0018] (2) Improve the accuracy of abnormal operating condition identification and introduce the weighted cross entropy loss function to solve the problem of sample imbalance.

[0019] (3) Optimize the input feature design (such as time difference and sliding window statistics) to enhance the model's sensitivity to sudden changes in signals (such as a sudden drop in temperature caused by window opening), thereby achieving higher recognition accuracy.

[0020] (4) Rapid migration and adaptation across scenarios is achieved by inserting a lightweight adaptation module. Only the adaptation module parameters need to be fine-tuned in the target room, which shortens the training time and improves the efficiency and accuracy of cross-room deployment.

[0021] (5) Closed-loop control and energy efficiency optimization. After real-time identification of abnormal operating conditions, adaptive control strategies are dynamically executed (such as switching to energy-saving mode when windows are open and increasing fresh air volume when many people gather), thereby reducing overall energy consumption while maintaining indoor temperature and humidity comfort.

[0022] (6) Enhance system robustness and scalability, remove the decoder module from the model, simplify the model structure, reduce redundant calculations, and improve system stability; (7) Support user feedback and cloud-based iterative updates, and adapt to long-term environmental changes and user behavior evolution by continuously optimizing adaptation module parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of a method for abnormal monitoring and intelligent control of heat and humidity load of a room air conditioner in an embodiment of the present invention.

[0024] Figure 2 It is a cloud modeling flow chart in the embodiment.

[0025] Figure 3 Schematic diagram of the structure of the abnormal monitoring and intelligent control system of room air conditioning heat and humidity load in the embodiment. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical scheme and advantages of the present invention clearer and more concise, the present invention is further described in detail below in conjunction with the accompanying drawings and an embodiment. In the following description, for the purpose of illustration rather than limitation, specific details such as specific system structures, technologies and the like are proposed so as to thoroughly understand the embodiments of the present application. However, it should be clear to those skilled in the art that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits and methods are omitted to prevent unnecessary details from hindering the description of the present application.

[0027] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0028] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0029] like Figure 1 , which is a process of a method for abnormal monitoring and intelligent control of heat and humidity loads of a room air conditioner based on transfer learning and an improved lightweight Informer model proposed in this embodiment, is specifically as follows.

[0030] 1. Typical room experimental data collection Collect infrared information including human activities (infrared sensor or human monitoring equipment), indoor temperature (return air temperature) and relative humidity (Relative humidity of return air), CO2 concentration , Outdoor temperature and relative humidity , Air conditioner power , Air conditioner set temperature and air conditioner fan speed settings , Timestamp And other data, build a data set containing abnormal working condition labels.

[0031] As an example, an air conditioner, a multi-source environmental parameter sensor module, and an air conditioner intelligent controller can be deployed in a typical room. By simulating operating conditions such as window opening, door opening, high heat source, high humidity source, and multiple people gathering, the environment inside and outside the room and the air conditioner operation data are collected to construct a time series data set containing labels. The specific steps are as follows: (1) Experimental environment setting: Select representative air-conditioning application environments (such as north-south orientation, different floors, etc.), and artificially set up experimental scenarios to simulate different user behaviors under different indoor and outdoor temperature and humidity backgrounds.

[0032] (2) The data collection cycle is Minutes, collection parameters include infrared information of human activities (infrared sensor or human monitoring equipment), indoor temperature (return air temperature) and relative humidity (return air relative humidity), CO 2 concentration , Outdoor temperature and relative humidity , Air conditioner power , Air conditioner set temperature and air conditioner fan speed settings , Timestamp wait. As a simulation of abnormal working conditions, the following experimental scenarios are carried out to enrich data labels.

[0033] Opening windows / doors: simulate opening angles of different degrees to observe thermal and humidity disturbances; High heat source / high humidity source: Start high heat or high humidity equipment such as electric heaters or humidifiers; Crowds: Simulate people entering, increase heat load and CO 2 concentration; Normal operation: continuous operation without interference; Through the above experiments, a labeled working condition data set was constructed. Abnormal working conditions include: opening windows (working condition 1), opening doors (working condition 2), high heat sources (working condition 3), high humidity sources (working condition 4), large crowds of people (working condition 5), equipment abnormalities (working condition 6), and normal operation is recorded as (working condition 0).

[0034] 2. Data Preprocessing and Cloud Modeling Upload the collected data to the cloud server for data cleaning, label arrangement and feature construction (feature engineering).

[0035] (1) De-noise the data and use the SG filtering method to smooth the data. The filter form is: ; Where: After filtering Time data; is the window size; is the filter coefficient; For the past Time data, ; Standardize the denoised data. Use the Z-Score method to normalize the features of each dimension. The formula is as follows: ; Where: for Time data; is the mean of the data; is the standard deviation of the data. is the normalized value.

[0036] (2) Perform feature engineering again. This embodiment constructs the following features.

[0037] 1) Time difference feature (Δ), the following are indoor temperature (return air temperature) and relative humidity (return air relative humidity), CO 2 concentration , Air conditioner power The corresponding time difference features: ; ; ; ; in, , They are Moment and The indoor temperature at the moment; , They are Moment and Relative humidity at the time; , They are Moment and CO at the moment 2 concentration; , They are Moment and Air conditioner power at the moment .

[0038] 2) Cross-parameter difference features Indoor and outdoor temperature difference: ; Relative humidity difference: ;in Outdoor temperature at all times ,and Relative humidity at all times .

[0039] 3) Sliding window statistical features Through the length of The time window (which may be 3 minutes in this embodiment) is used to count the changing trends and improve the perception of slow signals.

[0040] Average indoor temperature: ; Standard deviation of indoor temperature: ; 5) One-hot time cycle characteristics Encodes time features such as hour (0-23) and day of the week (0-6).

[0041] (3) Using the processed data, the improved lightweight Informer model is trained in the cloud to output the working condition classification model. Figure 2 , which is the cloud modeling process, which specifically includes: Step 1: Construct input samples. Construct sequence input: , ; Step 2: Improve the lightweight informer model structure, which has the following key features: a) The encoder is a single-layer structure, and the number of attention heads is set to 2; b) Remove multiple layers of decoders and only output the predicted labels for the current or next moment; Step 3: Introduce category weight optimization loss function. The weighted multi-classification cross entropy loss function is used to solve the sample imbalance: ; in, is the total number of samples for batch training, is the total number of categories ( , 7 categories in total: corresponding to working conditions 0 to 6), For the The weight of the class, , For the The true labels of samples, For the model Samples are predicted as categories probability.

[0042] Step 4: Model evaluation. Use accuracy and F1-Score to evaluate model performance, and prioritize high energy consumption anomalies to have a higher recall rate.

[0043] Accuracy

[0044] 1 is the indicator function, which is 1 if the prediction is correct and 0 otherwise; F1-Score is calculated using weighted average: ； For Category F1-score; :category The actual number of samples; ; in, , indicating that the model predicts the category In the sample, the real proportion; , indicating that the real belongs to the category The proportion of samples that are correctly predicted by the model. (Real example): The real And the prediction is The number of samples; (False positive example): True is not But the prediction is The number of samples; (False negative): True is But the prediction is not The number of samples.

[0045] 3. Transfer Learning Adaptation Module Deploy the basic model in the target room (new room) and insert the adaptation module. Only train the adaptation module to achieve rapid model migration.

[0046] The steps for fine-tuning the target room transfer learning model are as follows: Step 1. Basic model construction. In the complete Informer model trained with typical room data, the core parameters of the Encoder are set to the shared frozen layer, and the initialized adaptation module is inserted into the output layer of the Encoder to obtain the basic model; Step 2. Initialize the target room model. Deploy the basic model to the target room air conditioner intelligent controller, activate only the adaptation module parameters, and freeze all other layers; collect a small amount of new room data to build a small-scale training set.

[0047] Step 3. Local training and fast convergence. Train the adaptation module on the target room data. Since the total number of parameters is small, the training time is short. If there is not enough data, transfer initialization + parameter fine-tuning can be used.

[0048] Step 4. Model fusion and deployment. After training is completed, the new room adaptation module is merged with the basic model into a personalized model.

[0049] In order to solve the problem of insufficient initial data for new rooms, this embodiment designs a transfer learning strategy, which uses parameter sharing and adaptation modules to perform migration fine-tuning on the new room, and quickly adapts the trained typical room model to the target room.

[0050] The adaptation module proposed in this embodiment is used to realize room-level model migration and personalized recognition based on the shared Informer model backbone. The adaptation module realizes low-dimensional mapping and residual connection with a bottleneck structure (two-layer fully connected neural network).

[0051] Adaptation module structure: , in, is the down-projection weight matrix, is the up-projection weight matrix, the activation function is ReLU, , For input, For the adapter module.

[0052] 4. Model deployment and prediction The trained room personalization model is deployed to the air-conditioning intelligent controller to collect real-time data, input it into the Informer model, and output the current working condition prediction result. This embodiment supports continuous update and incremental training mechanism. If the system detects a drop in recognition rate or misjudgment of user feedback, the adaptation module can be retrained.

[0053] 5. Abnormal Identification and Intelligent Control The operating condition type is determined based on the recognition results, including open windows, open doors, high heat sources, high humidity sources, and large gatherings of people, and the corresponding air conditioning operation mode adjustment strategy is executed.

[0054] As an embodiment, the operation mode adjustment strategy includes: When the prediction is that the window or door is open, if the continuous detection time exceeds the set threshold, it will switch to energy-saving mode; When a large gathering of people is predicted, increase the fresh air volume or the air conditioning speed; When a high heat source or a high humidity source is predicted, the set temperature is adjusted or the dehumidification function is turned on respectively.

[0055] 6. Control feedback and remote synchronization Record control decision logs and upload them to the cloud server to support user feedback, corrections and model updates.

[0056] As an example, Figure 3 , a system for abnormal monitoring and intelligent control of heat and humidity load of room air conditioner, mainly comprising: Cloud server, used for lightweight working condition classification model training, adaptation module optimization and data management; The air conditioner intelligent control module may include: a data processing module, which is used to realize on-site data collection, data processing and operation control, and data processing includes filtering, normalization and feature construction; an abnormality recognition module, which is embedded with the adaptation module and the personalized model, and is deployed through edge computing devices to support real-time reasoning and control decisions; a control execution module, which adjusts the air conditioner operation mode according to the abnormality recognition results; a communication and feedback module, which is used for data uploading, model updating and user interaction.

[0057] User APP, used to receive recognition results, provide feedback and remotely control the air conditioner.

[0058] In this embodiment, it also includes a multi-source environmental parameter sensor module for collecting indoor and outdoor environmental parameters and air conditioning operation data; in this embodiment, the multi-source environmental parameter sensor module may include an indoor temperature sensor, an indoor humidity sensor, an outdoor temperature sensor, a human infrared temperature sensor, a carbon dioxide sensor, a power sensor, etc.

[0059] As an embodiment, the air-conditioning intelligent control module is deployed in the air-conditioning intelligent controller. The air-conditioning intelligent controller of this embodiment refers to an air-conditioning controller with core computing resources, storage and communication functions. An existing air-conditioning controller can be used or the chip on the existing air-conditioning controller can be upgraded to meet the requirements of running the relevant modules of the present invention. The specific working steps are as follows: Step 1: Real-time data collection from multiple sources. Various sensors deployed in the room periodically collect the aforementioned environmental and equipment parameters. The recommended data collection period is 1 minute. All raw data enter the next stage of processing module. Step 2: Data preprocessing and feature construction. Perform filtering and denoising (SG filtering) and normalization (Z-score standardization) on the collected multidimensional data; construct differential features and cross features; use sliding windows to construct input sequences; Step 3: Abnormal operating condition identification and reasoning. The processed input feature sequence is input into the Informer lightweight model deployed in the air-conditioning intelligent controller. The model performs reasoning and outputs the current operating condition prediction result: Step 4: Intelligent control of operation mode. According to the recognition results, the system matches the preset control strategy library and performs the corresponding air conditioning operation mode adjustment, for example: a) Open window or door (operating conditions 1 and 2): If the condition persists for ≥ 6 minutes (2 sampling cycles), the energy-saving operation mode will be automatically switched; b) Many people gather (condition 3): Increase the fresh air volume or wind speed to control CO 2 concentration; c) High heat source (condition 4): set the temperature to increase by 1°C to 2°C and switch to energy-saving mode; d) High humidity source (condition 5): turn on the dehumidification mode and appropriately lower the humidity setting; e) Equipment abnormality (condition 6): Suspend operation and send cloud alarm; f) Normal (condition 0): Restore the operating status set by the user.

[0060] Step 5: Control feedback and remote synchronization. The controller stores the recognition results and control actions in the form of logs and uploads them regularly. If the user notification function is enabled, the system pushes messages to the user's APP through the cloud. Users can confirm, adjust or mark the recognition results through the feedback button for continuous optimization of the subsequent model.

[0061] Step 6: Model and strategy update mechanism. The air conditioner intelligent controller regularly evaluates the model recognition accuracy and user feedback consistency; if the model performance degrades or the environment changes significantly, the cloud update process is triggered. In this embodiment, the model is trained with typical room data and then migrated to the target room (new room) through the adaptation module. The air conditioner intelligent controller identifies abnormal working conditions based on real-time data and executes control strategies. Users can correct model parameters through APP feedback.

[0062] It should be understood that the present invention is not limited to the above-mentioned specific embodiments. Any modification, substitution, combination, simplification, etc. made by any technician familiar with the technical field within the technical scope disclosed by the present invention in accordance with the spirit and principle of the present invention are equivalent replacement methods and should be included in the protection scope of the present invention.

Claims

1. A method for abnormal monitoring and intelligent control of heat and humidity load of room air conditioner, characterized in that: The following steps are involved: (1) Collect indoor and outdoor environmental parameters of typical rooms and air conditioning operation data, build a labeled abnormal operating condition data set, and upload it to the cloud server; (2) Based on the abnormal working condition data set, the Informer model is used to perform cloud modeling training to obtain a lightweight working condition classification model; (3) Based on the working condition classification model, a personalized model is obtained through transfer learning adaptation module and deployed to the air conditioning intelligent controller of the target room to predict the abnormal working condition type in real time; (4) Execute the corresponding air-conditioning operation mode control strategy according to the prediction results. The cloud server iteratively optimizes the adaptation module based on the user feedback data, and then optimizes the personalized model.

2. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (1), the environmental parameters include indoor temperature and relative humidity, CO2 concentration, outdoor temperature and relative humidity, and infrared information of human activities; the air conditioning operation data includes air conditioner power, air conditioner set temperature, air conditioner set wind speed and timestamp; the abnormal operating conditions include open windows, open doors, high heat sources, high humidity sources, and large crowds.

3. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (2), the Informer model is an improved Informer model; the encoder of the improved Informer model is a single-layer structure, the number of attention heads is 2-4, and the multi-layer decoder is removed, and only the predicted label of the current moment or the next moment is output.

4. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (2), the improved Informer model is trained using a weighted cross entropy loss function to solve the problem of unbalanced samples of working condition categories. The loss function is: ； in, is the total number of samples for batch training, is the total number of categories, For the The weight of the class, , For the The true labels of samples, For the model Samples are predicted as categories probability.

5. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (3), the adaptation module is used to realize room-level model migration and personalized recognition based on the shared Informer model backbone, and adapt the trained working condition classification model to the target room; the adaptation module structure is: ； in, is the down-projection weight matrix, is the up-projection weight matrix, the activation function is ReLU, For input, For the adapter module; and The random initialization of adopts Gaussian distribution.

6. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (3), the step of obtaining the personalized model and deploying it to the air conditioning intelligent controller of the target room specifically includes: For the working condition classification model obtained in step (2), the core parameters of the model encoder are set to the shared frozen layer, and the initialized adaptation module is inserted into the output layer of the encoder to obtain the basic model; Deploy the basic model to the target room air conditioner intelligent controller, activate only the adaptation module parameters, and freeze all other layers; collect the indoor and outdoor environmental parameters of the target room and the air conditioner operation data to build a small-scale training set; The adaptation module is trained by transferring initialization and fine-tuning parameters on the constructed small-scale training set; After training is completed, the adaptation module and the basic model are merged into a personalized model, and the personalized model is deployed to the air-conditioning intelligent controller.

7. A room air conditioning heat and humidity load abnormal monitoring and intelligent control method according to claim 1, characterized in that: In step (4), the control strategy includes: When the prediction is that the window or door is open, if the continuous detection time exceeds the set threshold, it will switch to energy-saving mode; When a large gathering of people is predicted, increase the fresh air volume or the air conditioning speed; When a high heat source or a high humidity source is predicted, the set temperature is adjusted or the dehumidification function is turned on respectively.

8. A room air conditioning heat and humidity load abnormality monitoring and intelligent control method according to claim 1, characterized in that: In step (4), the feedback data includes the recognition results and control logs annotated by the user, and the cloud server regularly updates the adaptation module parameters.

9. A system for implementing the method for abnormal monitoring and intelligent control of heat and humidity load of a room air conditioner according to any one of claims 1 to 8, characterized in that: include: Multi-source environmental parameter sensor module, used to collect indoor and outdoor environmental parameters and air conditioning operation data; Data processing module, used for filtering, normalizing and feature construction of data; Cloud server, used for lightweight working condition classification model training, adaptation module optimization and data management; An anomaly recognition module, which is used to embed the adaptation module and the personalized model, and is deployed through edge computing devices to support real-time reasoning and control decisions; A control execution module, used for adjusting the air-conditioning operation mode according to the abnormality recognition result; Communication and feedback module, used for data uploading, model updating and user interaction.

10. The system according to claim 9, characterized in that The multi-source environmental parameter sensor module includes an indoor temperature sensor, an indoor humidity sensor, an outdoor temperature sensor, a human infrared temperature sensor, a carbon dioxide sensor and a power sensor.

Images