Air conditioner controller individual upgrading push method based on user behavior portrait

Abstract

This invention relates to the field of smart home appliance technology and proposes a method for personalized upgrade push for air conditioner controllers based on user behavior profiles. The method includes collecting and storing operational data from several air conditioner controllers, including environmental parameters and user operation data; constructing a group operation habit profile for different geographical regions based on geographical location information in the operational data; constructing an individual operation habit model for a target air conditioner controller based on historical operational data; comparing the individual operation habit model of the target air conditioner controller with the group operation habit profile of the geographical region to which the target air conditioner controller belongs, and generating at least one demand tag to describe personalized needs based on the differences found; matching the corresponding personalized upgrade package from an upgrade package library according to the demand tag, and pushing the personalized upgrade package to the target air conditioner controller.

Description

Technical Field

[0001] This invention relates to the field of smart home appliance technology, and in particular to a method for personalized upgrade push for air conditioner controllers based on user behavior profiles. Background Technology

[0002] Currently, air conditioner controllers generally support OTA (Over-The-Air) firmware upgrades. Existing technologies mainly employ two push modes: a unified push mode, where manufacturers release a new version and push the same upgrade package indiscriminately to all online devices or specific models via the cloud; and a problem-driven mode, where patches are only pushed out in emergencies or security vulnerabilities. However, these "one-size-fits-all" upgrade strategies have significant drawbacks: First, they ignore personalized user needs, failing to consider differentiated functional expectations based on geographical environment (e.g., dry in the north vs. humid in the south), usage habits (e.g., young people prefer strong cooling while older people dislike drafts), and performance preferences (energy saving or rapid cooling priority). Second, poor upgrade experiences can even lead to new problems; for example, firmware designed to optimize energy saving may reduce compressor peak power, causing users seeking rapid cooling to perceive a "slower air conditioner," thus increasing complaints and after-sales costs. Finally, there is a serious disconnect between R&D and user needs; the technical department mainly relies on laboratory data and limited market feedback, lacking large-scale, refined data on real user needs, resulting in unclear development directions for new features and difficulty in maximizing user satisfaction through resource allocation. Therefore, there is an urgent need for an intelligent upgrade solution that can accurately identify individual needs and achieve customized upgrades. Summary of the Invention

[0003] To address the aforementioned shortcomings, the present invention aims to propose a personalized upgrade push method for air conditioner controllers based on user behavior profiles. This method collects and analyzes the operating data of air conditioner controllers to construct a profile of group operating habits and an individual operating habit model. By comparing the differences between the two, personalized demand tags are generated, enabling precise push of upgrade packages that match individual needs. This solves the problems of a uniform push mode ignoring personalized user needs, poor upgrade experience leading to new problems, and a disconnect between research and development and user needs.

[0004] To achieve this objective, the present invention adopts the following technical solution:

[0005] A method for personalized upgrade push notifications for air conditioner controllers based on user behavior profiles includes the following steps:

[0006] S1: Collect and store the operating data of several air conditioner controllers, including environmental parameters and user operation data;

[0007] S2: Based on the geographical location information in the operational data, construct a profile of the group's operational habits in different geographical regions;

[0008] S3: For the target air conditioner controller, build an individual operating habit model based on historical operating data;

[0009] S4: Compare the individual operation habit model of the target air conditioner controller with the group operation habit profile of the geographical area to which the target air conditioner controller belongs, and generate at least one demand label to describe personalized needs based on the differences found in the comparison.

[0010] S5: Match the corresponding personalized upgrade package from the upgrade package library according to the demand tag, and push the personalized upgrade package to the target air conditioner controller.

[0011] Preferably, the environmental parameters include indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and geographical location information;

[0012] The user operation data includes temperature setting, fan speed selection records, operating mode usage records, timed on / off records, and energy-saving mode activation records.

[0013] Preferably, step S2 includes:

[0014] Based on geographical location and climate characteristics, several air conditioner controllers are divided into first-level groups, and a first-level group operation habit profile is constructed based on the corresponding operation data.

[0015] From the operational data on which the first-level group operation habit profile is based, a first data subset that conforms to the dimensions of living environment and user attributes is selected, and a second-level division is carried out based on the first data subset to construct a second-level group operation habit profile.

[0016] A second subset of data that matches the core operational habit preference dimension is selected from the first subset of data, and a third-level division is performed based on the second subset of data to construct a third-level group operational habit profile;

[0017] A comparison priority is set for the group operation habit profiles of the first, second and third levels, wherein the group operation habit profile of the third level has the highest priority and the group operation habit profile of the first level has the lowest priority.

[0018] Preferably, step S3 includes:

[0019] Clean the historical operating data of the target air conditioner controller, including removing abnormal operation data and abnormal equipment operating parameters, and completing the missing data;

[0020] Perform correlation analysis or time series analysis on the cleaned historical operation data to extract the linkage features between environmental parameters and user operation data. The linkage features include features that reflect the correlation rules or time series patterns between environmental conditions and user operation behavior.

[0021] The user operation data in the cleaned historical operation data is processed, including at least one of statistical distribution analysis, behavior classification and sequence pattern mining, to extract quantitative features. The quantitative features include features that reflect the user's statistical preferences or classification results for set parameters, operation modes and operation sequences.

[0022] By integrating the linkage features and the quantitative features, the individual operation habit model is constructed and optimized. The individual operation habit model represents the stable operation pattern and environmental adaptation preference of the target air conditioner controller.

[0023] Preferably, step S4 includes:

[0024] The priority of the group operation habit profiles at each level that match the target air conditioner controller is determined, and the group operation habit profile with the highest priority is used as the core comparison benchmark.

[0025] The individual operation habit model is compared with the core comparison benchmark by linkage feature comparison, including: extracting the association rules or time sequence patterns between environmental parameters of individuals and groups and user operation data, and calculating the degree of difference between the extraction results in triggering conditions and operation response.

[0026] The individual operation habit model is compared with the core comparison benchmark by quantitative feature comparison, including: calculating the degree of deviation between individual statistical values ​​and group statistical intervals or distribution characteristics based on statistical indicators of set parameter preferences, usage ratio of operation mode and operation sequence.

[0027] Perform auxiliary verification and comparison to compare the individual operation habit model with at least one other group operation habit profile other than the core comparison benchmark, in order to verify whether the degree of difference and deviation is consistent at the corresponding group level.

[0028] Based on the degree of difference and the degree of deviation, valid differences are selected from the comparison results. The selection criteria for valid differences include that the degree of difference or the degree of deviation continuously exceeds a preset threshold for a first preset duration, and that the frequency of occurrence of the difference exceeds a second preset threshold in a specific use case associated with the difference.

[0029] According to the preset mapping rules, each effective difference is converted into a requirement label describing personalized functional needs. The mapping rules map the difference features into standardized labels that include functional directions and usage scenarios or preference keywords.

[0030] Preferably, step S5 includes:

[0031] Based on the requirement tags, at least one corresponding upgrade package is matched from the upgrade package library, wherein the upgrade packages in the upgrade package library are pre-associated with core requirement tags and at least one auxiliary tag;

[0032] Perform compatibility checks on the matched upgrade packages to detect whether there are parameter conflicts between the core functional algorithms of different upgrade packages;

[0033] If a conflict exists, conflict handling is performed, which includes generating a merged upgrade package or providing the user with selection information; if no conflict exists, the upgrade package to be pushed is determined.

[0034] Based on the historical operating data of the target air conditioner controller, periods of low device usage frequency are identified as candidate periods for upgrade push.

[0035] Generate and send upgrade prompt information to the target air conditioner controller, the upgrade prompt information including a description of upgrade optimization content based on the difference between the requirement tag and the effective difference;

[0036] Upon receiving the confirmation upgrade instruction, the system sends the upgrade package or integrated upgrade package to be pushed to the target air conditioner controller and controls the target air conditioner controller to complete the installation.

[0037] Preferably, the compatibility verification of the matched upgrade package includes:

[0038] When an upgrade package is matched, extract the core functional algorithm parameters and triggering conditions of each upgrade package;

[0039] Simulate a scenario where the upgrade package operates collaboratively on the target air conditioning controller;

[0040] Based on the simulation results, determine whether there is mutual exclusion in functional logic or parameter range.

[0041] Preferably, the degree of deviation between individual statistical values ​​and population statistical intervals or distribution characteristics is calculated. The following relation is satisfied:

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] in, Indicates the overall degree of deviation. Indicates the first The weighted deviation component of each quantitative feature This represents the first [item] extracted from the individual operational habit model. Individual statistical values ​​of a quantitative characteristic, This indicates the first [item] extracted from the group's operational habit profile, which serves as the core comparison benchmark. The population statistical mean of a quantitative characteristic Indicates the first The standard deviation of the statistical distribution of a quantitative characteristic in a population Indicates the first Normalized weights of each quantized feature, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. For indicator functions, Indicates the first The statistical interval or distribution range of a quantitative characteristic. This represents the total number of quantized features involved in the calculation. Indicates the first Standardized individual statistical values ​​of a quantitative characteristic, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. Indicates the first The statistical interval or distribution range of a quantitative characteristic. Indicates the first in the group The first sample The corresponding feature values ​​of each quantized feature, Indicates the first in the group The first sample The corresponding feature value of each quantized feature.

[0047] One of the above technical solutions has the following advantages or beneficial effects:

[0048] This invention continuously collects and stores environmental parameters and user operation data, transforming real-world usage behavior into structured data assets. This shifts technology development from reliance on laboratory simulations to large-scale user scenarios, preventing a disconnect between research and real-world applications. It constructs user profiles based on geographic location information, establishing a benchmark reflecting common characteristics across different climate zones and user types, providing a comparative coordinate system for identifying individual uniqueness. By building individual user modeling for target devices, it meticulously depicts individual user behaviors such as temperature preferences, wind speed selection, and mode usage frequency, achieving a leap from vague perception to precise modeling of individual needs. By comparing individual models with user profiles of the same region, it transforms differentiated features into functional tags such as strong cooling needs, rapid cooling needs, and strong dehumidification needs, making implicit user preferences explicit, identifiable, categorizable, and matchable. Finally, by matching corresponding personalized upgrade packages from the upgrade package library based on these demand tags and pushing them to users, each user receives not standardized firmware, but functional optimizations tailored to their unique usage scenarios. This forms a complete chain of "data collection - profile building - difference identification - tag generation - precise push", which not only ends the experience mismatch and complaint risks caused by "one-size-fits-all" upgrades, but also builds a data closed loop that allows user needs to reach R&D directly. Ultimately, without increasing the user's understanding cost, the product performance can proactively adapt to individual habits, achieving the dual effect of improved satisfaction and reduced operation and maintenance costs. Attached Figure Description

[0049] 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.

[0050] Figure 1 This is a flowchart of a personalized upgrade push method for air conditioner controllers based on user behavior profiles provided in an embodiment of the present invention. Detailed Implementation

[0051] 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.

[0052] 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.

[0053] A method for personalized upgrade push notifications for air conditioner controllers based on user behavior profiles, such as... Figure 1 As shown, a preferred embodiment of the present invention includes:

[0054] S1: Collect and store the operating data of several air conditioner controllers, including environmental parameters and user operation data;

[0055] It should be noted that operational data refers to the multi-dimensional information collection generated by the air conditioner controller during actual operation. Environmental parameters include indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and geographical location information. These parameters are obtained through the air conditioner's built-in temperature and humidity sensors, as well as GPS modules or IP address positioning technology, and are used to reflect the objective environmental conditions during air conditioner operation. User operation data includes set temperature, fan speed selection records, operating mode usage records, timed on / off records, and energy-saving mode activation records. This data is collected through the air conditioner controller's operation log interface and is used to characterize the user's subjective usage preferences. The data collection process can employ real-time streaming technology, transmitting data from the device to the cloud server via the MQTT protocol, or batch collection, uploading data in packages during device idle periods. Storage can utilize distributed file systems such as HDFS or time-series databases such as InfluxDB to ensure high data availability and fast retrieval capabilities.

[0056] S2: Based on the geographical location information in the operational data, construct a profile of the group's operational habits in different geographical regions;

[0057] It should be noted that geographic location information refers to coded location data that identifies the geographical area where the air conditioner controller is located. This can be obtained through reverse lookup of the device's IP address, address information filled in during user registration, or the built-in positioning module. Its function is to serve as a primary dimension for group segmentation. Group operation habit profiles refer to the characteristic descriptions formed after aggregating and analyzing the operational data of all users within a specific geographical area using machine learning algorithms. Clustering algorithms such as K-means or DBSCA can be used to group users according to behavioral similarity, and then the parameter distribution characteristics of each group can be statistically analyzed. The construction process can adopt an offline batch processing method, calculating and updating the full amount of data from the previous day every morning, or an online incremental update method, incorporating newly collected data points in real time.

[0058] Understandably, step S2 elevates fragmented individual behavioral data into group characteristics that reflect regional commonalities, forming a behavioral reference system for users in different climate zones and cities. For example, users in the hot and humid South China region may generally exhibit high dehumidification mode usage and a preference for high-speed cooling, while users in the dry and cold North may exhibit high heating mode usage and a preference for low-speed cooling. This group profiling provides a comparative benchmark for subsequently identifying unique individual needs, making the difference analysis reasonable in terms of geographical and climatic dimensions, and avoiding benchmark distortion caused by global averaging.

[0059] S3: For the target air conditioner controller, build an individual operating habit model based on historical operating data;

[0060] It should be noted that the target air conditioner controller refers to the specific device requiring personalized upgrade push, which can be located in the system through the device's unique identifier. The individual operating habit model refers to a mathematical description of a single user's long-term behavioral patterns abstracted through data analysis techniques. This can be stored as a rule base to store deterministic operating rules, or as a weight matrix to quantify the importance of each behavioral feature. Historical operating data refers to the collection of operating data accumulated by the device over a period of time, which can span 6 months, 12 months, or longer, depending on the model's need for a balance between stability and flexibility. The construction process can include three sub-stages: data cleaning, feature extraction, and model training. Data cleaning uses the Isolation Forest algorithm to identify outliers; feature extraction uses association rule mining or time series analysis techniques; and model training can be completed in the cloud using a lightweight machine learning framework such as Scikit-learn.

[0061] Understandably, step S3 involves systematically and structurally modeling the behavioral patterns of specific users, making implicit usage habits explicit as calculable and comparable model parameters. By cleaning and removing noisy data and outliers, the model is ensured to be built based on genuine and valid user intent. Correlation analysis uncovers the linkage between environment and operation, revealing users' adaptive strategies under different environmental conditions. Feature quantification transforms operational preferences into numerical weights, forming a precise measure of the intensity of user needs. The final individual model accurately reflects the user's stable operational patterns and environmental adaptation preferences, providing standardized input for subsequent refined comparisons with group profiles, ensuring that difference identification is based on reliable individual behavioral characterization.

[0062] S4: Compare the individual operation habit model of the target air conditioner controller with the group operation habit profile of the geographical area to which the target air conditioner controller belongs, and generate at least one demand label to describe personalized needs based on the differences found in the comparison.

[0063] It should be noted that comparison refers to the analytical process of identifying the degree of deviation by calculating the similarity or difference between an individual model and a group profile. This can be quantified using cosine similarity, Euclidean distance, or a custom weighted deviation formula. Difference refers to the inconsistency between individual behavioral characteristics and group baseline characteristics in terms of statistical distribution, association rules, or temporal patterns. Identifying differences requires setting a threshold to determine significance. Demand tags are standardized text identifiers used to describe users' personalized functional needs. They can use a structured naming convention of "functional keywords + scenario / preference keywords," such as "strong dehumidification - low fan speed linkage demand." The generation process can use rule mapping to automatically convert specific types of differences into corresponding tags, or it can use classification algorithms to automatically label the difference feature vectors.

[0064] Understandably, step S4 involves transforming individual differences into comprehensible and categorizable demand descriptions through comparative analysis. This means revealing the uniqueness of user operational logic through linked feature comparison; for example, a user's insistence on low-speed dehumidification in high-humidity environments, rather than the commonly used medium-speed strategy, reflects a specific need for a balance between comfort and effectiveness. Quantitative feature comparison identifies the strength of parameter preferences; for example, a consistently low temperature setting compared to the group average with a stable deviation indicates a clear preference for cooling or energy saving. Auxiliary verification comparison ensures the robustness of the differences, eliminating interference from occasional fluctuations. The resulting demand tags explicitly express implicit behavioral differences as directions for product function optimization, enabling the matching of upgrade packages to have a clear semantic correspondence, achieving a precise transformation from behavioral analysis to demand definition.

[0065] S5: Match the corresponding personalized upgrade package from the upgrade package library according to the demand tag, and push the personalized upgrade package to the target air conditioner controller.

[0066] It's important to note that the upgrade package library is a collection of personalized upgrade programs stored on a cloud server. Each upgrade package includes executable feature optimization code and is pre-associated with core requirement tags and auxiliary tags. The matching process can employ precise tag matching, using string comparison or hash value comparison to find upgrade packages that perfectly match the requirement tags, or semantic similarity matching, which calculates the semantic distance between tags using a word vector model to match the upgrade package with the highest relevance. The push process can be initiated by the device itself, triggering an upgrade check request during periods of low device usage, or by the cloud itself, issuing an upgrade command after determining that the device is in standby mode.

[0067] Understandably, step S5 involves transforming the identified requirement tags into actual functional upgrades and ensuring a smooth, uninterrupted upgrade process. Tag matching achieves precise alignment between requirements and solutions, ensuring each upgrade package corresponds to a clear functional optimization direction; compatibility checks prevent potential algorithmic conflicts when multiple tags coexist, ensuring collaborative work among functional modules; push notifications during low-usage periods minimize disruption to normal user experience; and generating contextualized prompts enhances users' perception of the upgrade's value, increasing upgrade confirmation rates. Ultimately, the upgrade package is no longer a uniform, standardized firmware, but a customized functional optimization highly tailored to individual needs, completing a closed loop from data analysis to service delivery.

[0068] Preferably, the environmental parameters include indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and geographical location information;

[0069] The user operation data includes temperature setting, fan speed selection records, operating mode usage records, timed on / off records, and energy-saving mode activation records.

[0070] It should be noted that indoor temperature refers to the real-time indoor ambient air temperature value measured by the temperature sensor built into the air conditioner controller, expressed in degrees Celsius. It reflects the actual thermal environment of the user's space and serves as a benchmark for setting user temperature preferences in subsequent modeling. Outdoor temperature refers to the external ambient temperature value obtained through the outdoor unit sensor or external meteorological data interface. It characterizes the impact of the external environment on the air conditioning load and serves as a precondition variable for user operation responses in linkage feature analysis. Indoor humidity refers to the indoor relative humidity value measured by the humidity sensor built into the air conditioner controller. This parameter directly affects the accuracy of identifying the need for dehumidification mode in user behavior analysis in humid areas. Outdoor humidity refers to external air humidity data, used to comprehensively assess regional climate characteristics and help determine the urgency of users' dehumidification needs. Geographic location information refers to the geographic coordinates or administrative region code obtained through device IP address resolution, GPS module positioning, or user registration address backfilling. It serves as a primary dimension for group profile construction, ensuring accurate matching between users and regional climate types. Set temperature refers to the target temperature value actively set by the user through the remote control or APP. It is the most direct behavioral data characterizing user temperature preferences, and its statistical distribution constitutes the core quantitative feature of the individual model. The fan speed selection log refers to the user's operation log for each adjustment of the air conditioner's fan speed level (1-5 or automatic). The log includes the level value and operation timestamp, used to analyze the user's sensitivity to fan speed and their preferred airflow. The operating mode usage log refers to the user's operation sequence when switching between cooling, heating, dehumidifying, and ventilation modes, reflecting the user's actual demand intensity and usage scenarios for different functions. The timed on / off log refers to the user's set timed on / off time parameters; rules can be used to extract the user's work-rest patterns and time preference characteristics. The energy-saving mode activation log refers to the operation log of the user actively activating energy-saving functions. This data directly reflects the user's energy efficiency preferences and is a key basis for identifying energy-saving needs. All these parameters are collected through the air conditioner controller's sensor network and operation log interface, transmitted to the cloud using MQTT or HTTPS protocols, and stored in a time-series database or distributed file system, providing structured data support for subsequent profile construction and model training.

[0071] Understandably, the combined collection of five environmental parameters—indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and geographical location information—can comprehensively depict the objective environmental conditions under which the air conditioner operates. The difference between indoor and outdoor temperature and humidity reflects the air conditioner's load demand, while geographical location information correlates with regional climate types, providing a foundational input for constructing a group profile reflecting regional climate characteristics. The combined collection of five types of user operation data—set temperature, fan speed selection records, operating mode usage records, timed on / off records, and energy-saving mode activation records—can fully record users' subjective usage preferences. Set temperature and fan speed selections reflect users' quantitative requirements for comfort, operating mode usage records reveal users' actual needs for different functions, timed on / off records reflect users' work-rest patterns, and energy-saving mode activation records directly indicate users' energy efficiency preferences. The mechanism of synchronously collecting environmental parameters and user operation data enables the system to perform environment-operation linkage analysis, uncovering user operational response patterns under specific environmental conditions. For example, the combination of "high outdoor temperature + low set temperature" can identify a strong cooling demand, and the combination of "high indoor humidity + frequent dehumidification mode" can identify a strong dehumidification demand.

[0072] Preferably, step S2 includes:

[0073] Based on geographical location and climate characteristics, several air conditioner controllers are divided into first-level groups, and a first-level group operation habit profile is constructed based on the corresponding operation data.

[0074] From the operational data on which the first-level group operation habit profile is based, a first data subset that conforms to the dimensions of living environment and user attributes is selected, and a second-level division is carried out based on the first data subset to construct a second-level group operation habit profile.

[0075] A second subset of data that matches the core operational habit preference dimension is selected from the first subset of data, and a third-level division is performed based on the second subset of data to construct a third-level group operational habit profile;

[0076] A comparison priority is set for the group operation habit profiles of the first, second and third levels, wherein the group operation habit profile of the third level has the highest priority and the group operation habit profile of the first level has the lowest priority.

[0077] It should be noted that the first-level segmentation refers to the technical operation of macroscopically grouping a massive number of air conditioning controller devices based on geographical location and climate characteristics. Geographical location information can be obtained through IP address resolution, GPS positioning, or user registration address. Climate characteristic data can be obtained by connecting to publicly available climate zoning data from the China Meteorological Administration or by using the Köppen climate classification method. This segmentation uses clustering algorithms such as DBSCAN or K-means to divide the devices into 6-8 primary groups, such as the high-temperature and high-humidity region in South China, the dry and cold region in North China, and the temperate monsoon region in East China. Each group includes several devices, and its purpose is to establish the most macroscopic regional behavioral benchmark and capture the fundamental impact of climate type on air conditioning usage patterns. The first-level group operation habit profile refers to the statistical characteristic description formed by aggregating and analyzing the operating data of the device groups after the first-level segmentation. The construction process uses an offline batch processing computing framework such as Apache Spark. Statistical indicators include climate-correlated features such as cooling season duration, average daily usage time of dehumidification mode, and high-temperature period operation rate. This profile serves as the original data foundation for subsequent segmentation. The living environment and user attribute dimensions refer to two types of filtering conditions used for the second-level segmentation. Living environment includes building type (high-rise apartments, low-rise villas, rural residences), floor height, and unit area, etc. This information can be obtained through text analysis of the device installation address during user registration or through property data interfaces. User attributes include family structure (families with children, elderly families, single young adults), user type (family users, business users), etc., which can be supplemented through user profile tagging systems or questionnaire survey data. The role of this dimension is to achieve behavioral segmentation from macro-regional to meso-level scenario. The first data subset refers to the set of records that meet specific living environment and user attribute conditions extracted from the full operational data on which the first-level profile is built, through SQL filtering or data warehouse layering techniques. For example, from 1200 records in the high-temperature and high-humidity area of ​​South China, 350 records of high-rise apartments in Guangzhou with user attribute identification as families with children are filtered out to form the input data for the second-level segmentation. The second-level segmentation refers to the further grouping of devices based on the first data subset according to living environment and user attributes. Decision tree classification or logistic regression models are used to further break down devices within the same climate zone based on different living environments and user combinations, forming more than 20 secondary groups, such as families with children in high-rise apartments in Guangzhou and commercial users in low-rise villas in Shenzhen. Each group includes dozens of devices, and its purpose is to capture the meso-level impact of building characteristics and family structure on user behavior. The second-level group operation habit profile refers to the feature description formed by in-depth analysis of the second-level group data. The calculation process uses a distributed computing engine to extract features strongly correlated with living environment and user attributes, such as the proportion of low-speed wind speed usage, sleep mode activation rate, and frequency of anti-direct-blowing mode usage. This profile serves as input for the third-level segmentation.The core operating habit preference dimension refers to the filtering criteria used for the third-level segmentation. It is defined based on the quantitative performance of users' core demands such as energy saving, quiet operation, rapid cooling, and comfort. For example, energy saving preference can be judged by the activation rate of energy-saving mode and the set temperature range, while quiet operation preference can be judged by the usage time at low fan speed and the noise level at night. The role of this dimension is to achieve behavioral refinement from meso-level scenarios to micro-level habits. The second data subset refers to further filtering records from the first data subset that meet specific core habit preferences. For example, in 350 records of families with children in high-rise apartments, 86 records with set temperatures in the 26-28℃ range and an energy-saving mode activation rate of over 90% are selected to form the input data for the third-level segmentation. The third-level segmentation refers to the final grouping operation based on the core operating habit preferences on the basis of the second data subset. Using a Gaussian mixture model or hierarchical clustering algorithm, devices within the same living environment and user attribute combination are subdivided according to habit preferences to form 5-8 tertiary groups, such as energy-saving and low-noise preference type and rapid cooling and gentle fan preference type. Each group can include dozens of devices, and its role is to accurately identify the user's core value demands. The third-level group operation habit profile refers to the feature description formed by refined analysis of the third-level group data. The calculation process uses machine learning feature engineering technology to extract highly detailed features such as the set temperature concentration range, the activation rate of variable frequency energy-saving mode, the proportion of low-noise operation time at night, and the frequency of use of specific mode combinations. This profile serves as a direct reference benchmark for individual comparison. The comparison priority refers to the calling order weight set for the multi-level profiles. The system assigns a priority score to each profile level, with the third-level profile having a priority score of 100, the second-level profile having a priority score of 70, and the first-level profile having a priority score of 30. In subsequent individual comparisons, the system prioritizes calling the profile with the highest priority as the core comparison benchmark. If the third-level profile data is insufficient, the second-level or first-level profile will be used instead. This mechanism ensures that individual comparisons are always based on the most closely related group benchmark.

[0078] Understandably, the first-level segmentation is based on geographical location and climate characteristics, aggregating massive amounts of devices by climate zone to construct a primary profile reflecting the commonalities of regional climates. This profile provides a basic environmental adaptation benchmark. For example, the primary profile of the high-temperature and high-humidity region of South China reveals that users in this area generally have high dehumidification needs and a preference for high-speed cooling, providing a climatic background reference for subsequent segmentation. The second-level segmentation filters out a first data subset from the raw data of the primary profile that matches specific living environments and user attributes. It further breaks down devices of different building types and family structures within the same climate zone to construct a secondary profile reflecting scenario characteristics. This profile captures behavioral differences at the meso-level. For example, the secondary profile of families with children in high-rise apartments shows a high demand for low-speed and anti-direct-blow functions, avoiding the submergence of niche characteristics caused by the averaging effect of the primary profile. The third level of segmentation further filters from the first data subset to select a second data subset that aligns with core operational habits and preferences. This deeply subdivides devices with different value orientations within the same residential scenario, constructing a three-level profile reflecting core needs. This profile extracts refined features at the micro level; for example, the three-level profile for energy-saving and low-noise preferences clearly defines the user's set temperature range and the proportion of nighttime low-noise operation time, ensuring that the group benchmark closely matches the direct reference needs for individual comparisons. Comparison priority settings ensure that the system prioritizes the most refined three-level profile as the core benchmark during individual comparisons. If the three-level profile has low confidence due to insufficient data or low activity, it automatically downgrades to using a second-level or first-level profile as a supplement. This dynamic priority mechanism maximizes comparison accuracy while avoiding comparison bias caused by sparse data in subdivided groups. Through three-level segmentation and priority settings, the system can simultaneously consider statistical significance and feature accuracy, allowing subsequent individual comparisons to reference both the stable characteristics of large-scale groups and the precise characteristics of highly subdivided groups, thereby improving the accuracy of difference identification and the consistency of demand transformation.

[0079] Preferably, step S3 includes:

[0080] Clean the historical operating data of the target air conditioner controller, including removing abnormal operation data and abnormal equipment operating parameters, and completing the missing data;

[0081] Perform correlation analysis or time series analysis on the cleaned historical operation data to extract the linkage features between environmental parameters and user operation data. The linkage features include features that reflect the correlation rules or time series patterns between environmental conditions and user operation behavior.

[0082] The user operation data in the cleaned historical operation data is processed, including at least one of statistical distribution analysis, behavior classification and sequence pattern mining, to extract quantitative features. The quantitative features include features that reflect the user's statistical preferences or classification results for set parameters, operation modes and operation sequences.

[0083] By integrating the linkage features and the quantitative features, the individual operation habit model is constructed and optimized. The model represents the stable operation pattern and environmental adaptation preference of the target air conditioner controller.

[0084] It should be noted that cleaning refers to the preprocessing operation of raw operating data to improve its quality. Algorithms are used to identify and remove noise, errors, and invalid records, ensuring that subsequent modeling is based on genuine user intent. Historical operating data refers to the collection of all operating records accumulated by the target air conditioner controller over past periods, covering 6-24 months. This data is stored in a cloud-based time-series database or distributed file system and serves as the raw input for individual model construction. Abnormal operation data refers to non-intent-driven operation records caused by user accidental touches, children playing, or signal interference. Examples include temperature adjustments lasting less than 30 seconds or multiple mode switches within one second. This type of data can be identified and removed using the Isolation Forest algorithm or threshold judgment based on the operation interval. Abnormal equipment operating parameters refer to distorted data generated when the equipment is in a faulty state or the sensors are malfunctioning. Examples include power fluctuations due to frequent compressor start-stops or constant abnormal readings caused by temperature sensor short circuits. This type of data is filtered through equipment health status marking or parameter physical rationality verification. Missing data refers to data gaps caused by network interruptions, storage failures, or sensor offline status. Examples include blank indoor humidity values ​​or lost geographic location information for a certain period. This type of data needs to be supplemented to ensure the integrity of the time series. Completion refers to the technique of using mathematical methods to fill in missing values. For missing environmental parameters, linear interpolation or average values ​​of preceding and following time periods can be used. For missing user operations, the mode of similar scenarios can be used or the user can be marked as a special value. The completion algorithm is uniformly implemented in the data cleaning module. Association analysis refers to algorithms that mine the dependency relationship between environmental parameters and user operations. For example, the Apriori association rule mining algorithm or the FP-Growth algorithm can be used to extract strong rules such as "indoor humidity ≥ 60% → dehumidification mode activated" from transactional operation logs. Support and confidence are used as quantitative indicators of rule strength. Temporal analysis refers to algorithms that identify the temporal pattern of environmental changes and operational behaviors. Time series decomposition or dynamic time warping (DTW) technology can be used to capture the periodic response pattern of "outdoor temperature ≥ 32℃ → 17:00-23:00 power-on" to reveal the user's environmental adaptation strategy. Linkage features refer to the structured feature vectors output by association analysis and temporal analysis that characterize the coupling relationship between environment and operation. Each feature includes a triplet of trigger condition, response operation, and intensity score, which serves as the core input of environmental adaptation preferences in individual models. Association rules refer to environment-operation dependencies expressed in the form of conditional probabilities, such as a dehumidifier start-up rule with 83% support and 91% confidence. This rule is transformed into feature weights to participate in model construction. Temporal patterns refer to environment-operation response patterns expressed in the form of time delays or periods, such as the pattern of starting up one hour earlier during high-temperature periods. This pattern is transformed into temporal feature parameters.Statistical distribution analysis refers to algorithms that fit the probability distribution of user operation parameters, such as kernel density estimation (KDE) or Gaussian mixture model (GMM), to analyze the distribution of parameters such as set temperature and wind speed, extracting statistical quantities such as mode, quantiles, and probability density peaks to quantify the strength of user preference for specific parameters. Behavioral classification refers to algorithms that discretize and categorize user operation modes, using K-means clustering or decision tree classification to classify wind speed selection behavior into low wind speed preference, medium wind speed preference, and high wind speed preference, and mode switching sequences into rapid cooling-oriented, energy-saving-oriented, and comfort-oriented types, assigning discrete labels to each user based on the classification results. Sequence pattern mining refers to algorithms that discover frequent sequences from the operation event stream, such as using SPADE or PrefixSpan algorithms to extract frequent operation sequences like "sleep mode → anti-direct blowing mode → low wind speed," revealing users' contextualized usage habits. Quantitative features refer to the set of numerical or label-type features output from statistical distribution analysis, behavioral classification, and sequence pattern mining, including set temperature preference ranges, wind speed category proportions, and pattern sequence support. These features constitute the computable parameters of the individual model. Set parameters refer to numerical parameters such as temperature and wind speed that users directly adjust; their statistical distribution reflects the user's quantitative preferences. Operating modes refer to functional modes such as cooling, heating, dehumidification, and ventilation; their usage frequency and switching sequences reflect the user's functional needs. Operation sequences refer to combinations of operations such as mode switching and parameter adjustment performed by the user in chronological order; their frequent patterns reveal the user's contextual usage logic. Fusion refers to the technical operation of integrating linkage features and quantitative features, using feature concatenation or weighted averaging to integrate environmental adaptation features and user preference features into a unified feature vector. A multi-task learning framework can also be used to optimize the two types of features in parallel. The individual operation habit model refers to the mathematical representation of the behavioral patterns of a single user formed after fusion. It can adopt a hybrid structure of rule base + weight matrix. The rule base stores deterministic rules for the transformation of linkage features, and the weight matrix stores the importance scores of quantitative features. This model can be continuously optimized through incremental learning. Stable operating patterns refer to repetitive behavioral patterns that remain unchanged over a long period, such as a consistent preference for setting the temperature to 26°C year-round or always using sleep mode at night. This pattern is validated through statistical stability testing of long-term data (12-24 months). Environmental adaptation preferences refer to users' tendency to adjust their operations based on environmental changes, such as automatically activating dehumidification in high humidity or turning on the device earlier in high temperatures. This preference is quantified using the confidence and support metrics of linked features.

[0085] Specifically, step S3 can be implemented using a pipeline-style data processing architecture: First, the data cleaning module reads the target device's raw operating data for the past 12 months from the cloud-based time-series database, calls the isolated forest algorithm to identify outliers, marks temperature adjustments with a single operation duration of less than 30 seconds as accidental touch data and removes them, marks periods with compressor start-stop frequency exceeding 5 times per minute as fault data and removes them, uses linear interpolation to fill in periods with missing environmental parameters for more than 2 hours, and fills in missing operation records with the mode of similar operations in the preceding and following periods. The 1860 cleaned valid records are then written into an intermediate data table. The second step involves the linkage feature extraction module extracting environmental parameters and user operation data pairs from the cleaned data. It then uses the Apriori algorithm with a minimum support of 60% and a minimum confidence of 80% to identify the rules "indoor humidity ≥ 60% → dehumidification mode activated" (support 83%, confidence 91%) and "outdoor temperature ≥ 32℃ → power on from 17:00 to 23:00" (support 76%, confidence 85%). A time series decomposition algorithm is used to identify the periodic pattern of "low wind speed operation from 22:00 to 7:00 at night" (period intensity 0.88), and these three linkage features are encoded into feature vectors. The third step involves the quantitative feature extraction module performing multi-dimensional processing on user operation data. It uses the KDE algorithm to analyze the set temperature distribution, identifying the probability density peak (peak height 0.92) in the 26-27℃ range. The K-means algorithm then clusters wind speed levels into low wind speed (levels 1-2), medium wind speed (level 3), and high wind speed (levels 4-5), with the low wind speed category accounting for 85%. The PrefixSpan algorithm is used to mine operation sequences, discovering a frequent sequence "sleep mode → anti-direct-blow mode" (support 68%). The extracted quantitative features include temperature preference range, wind speed category proportion, and mode sequence support. The fourth step involves the feature fusion module concatenating the linked feature vector and the quantitative feature vector into a 128-dimensional unified feature vector. A dual-tower model structure is used: the left tower is a rule base structure storing three linked rules, and the right tower is a weight matrix structure assigning weights to the 128-dimensional features. The weights are optimized using a backpropagation algorithm, achieving a prediction accuracy of 92% on the validation set. The constructed individual operation habit model is persistently stored as a PMML format file for subsequent comparison and retrieval. The four modules are executed in a pipeline, with the output of the preceding module serving as the input of the following module. The modules are decoupled using message queues, and the parallel processing of data from multiple devices is supported.

[0086] In summary, by combining data cleaning, correlation analysis, time series analysis, statistical distribution analysis, behavior classification, sequence pattern mining, and feature fusion techniques, a robust and representative individual operation habit model was constructed, solving the technical problems of low quality of raw operating data, single feature dimension, and lack of environmental context.

[0087] Preferably, step S4 includes:

[0088] The priority of the group operation habit profiles at each level that match the target air conditioner controller is determined, and the group operation habit profile with the highest priority is used as the core comparison benchmark.

[0089] The individual operation habit model is compared with the core comparison benchmark by linkage feature comparison, including: extracting the association rules or time sequence patterns between environmental parameters of individuals and groups and user operation data, and calculating the degree of difference between the extraction results in triggering conditions and operation response.

[0090] The individual operation habit model is compared with the core comparison benchmark by quantitative feature comparison, including: calculating the degree of deviation between individual statistical values ​​and group statistical intervals or distribution characteristics based on statistical indicators of set parameter preferences, usage ratio of operation mode and operation sequence.

[0091] Perform auxiliary verification and comparison to compare the individual operation habit model with at least one other group operation habit profile other than the core comparison benchmark, in order to verify whether the degree of difference and deviation is consistent at the corresponding group level.

[0092] Based on the degree of difference and the degree of deviation, valid differences are selected from the comparison results. The selection criteria for valid differences include that the degree of difference or the degree of deviation continuously exceeds a preset threshold for a first preset duration, and that the frequency of occurrence of the difference exceeds a second preset threshold in a specific use case associated with the difference.

[0093] According to the preset mapping rules, each effective difference is converted into a requirement label describing personalized functional needs. The mapping rules map the difference features into standardized labels that include functional directions and usage scenarios or preference keywords.

[0094] It should be noted that each group's operational habit profile refers to a group behavior benchmark system constructed using a three-tiered progressive architecture. This includes a first-level profile based on geographical location and climate characteristics, a second-level profile based on living environment and user attributes, and a third-level profile based on core operational habit preferences. Each profile level corresponds to a priority field value in the profile metadata table: third-level profile priority = 3, second-level profile priority = 2, and first-level profile priority = 1. This priority value is automatically assigned by step S3 during profile construction based on the level of granularity. The core comparison benchmark refers to the group profile selected as the primary reference object during individual comparison. The system sorts all profiles matching the target device in descending order of priority and selects the profile with the highest priority value as the core comparison benchmark. If multiple profiles have the same priority, the profile with the highest data activity is selected. This benchmark determines the main reference coordinate system for difference identification. Linkage feature comparison refers to the technical process of aligning and comparing the environment-operation linkage features in an individual's operational habit model with the linkage features in a group profile. It extracts the differences between individuals and groups in association rules and temporal patterns. A weighted distance formula is used to calculate the difference, with a weight of 0.6 for association rules and 0.4 for temporal patterns. A difference score exceeding 60 is considered significant. Association rules refer to the dependency relationship between environmental parameters and user operations expressed in conditional probability form. This can be represented as a rule structure of "IF environmental conditions THEN operation response," including two strength indicators: support and confidence. Support reflects the frequency of rule occurrence in the data, and confidence reflects the reliability of the rule's validity. Temporal patterns refer to the time dependency pattern between environmental changes and operational responses expressed in the form of time delay or periodicity. Examples include the delay pattern of "powering on within 1 hour after the outdoor temperature exceeds 32℃" or the periodic pattern of "automatically entering sleep mode at 22:00 every night." Temporal patterns are extracted using time series analysis algorithms and include periodic strength and phase shift parameters. The difference score is a numerical indicator that quantifies the degree of inconsistency between individual linkage characteristics and group linkage characteristics. For differences in association rules, it calculates the weighted absolute difference between the coverage difference of trigger conditions and the confidence difference of response intensity. For differences in time-series patterns, it calculates the Euclidean distance between the periodic intensity difference and the phase shift difference. A higher difference score indicates that the individual's environmental response strategy deviates more from the group's commonality. Quantitative feature comparison refers to the technical process of calculating the numerical deviation between the quantitative features in an individual's operational habit model and the statistical intervals or distribution features in a group profile. It calculates the percentage deviation between the individual's probability density peak and the group's mode for set parameter preferences, calculates the absolute difference between the individual's frequency and the group's frequency using the proportion of operating modes, and calculates the deviation between the ratio of individual support and the group's support for operational sequence statistical indicators. The comprehensive score of the deviation degree is the weighted square root value of the deviation components of each dimension.Deviation degree refers to a numerical indicator that quantifies the extent of deviation between an individual's characteristics and the group benchmark. It is derived through four mathematical steps: standardization, weight calculation, segmented weighting, and multi-dimensional aggregation. A greater deviation indicates that the individual's operational habits deviate more from the group benchmark and that their needs are more unique. Standardization involves converting raw feature values ​​into dimensionless standard scores, achieved by subtracting the group mean from the individual's statistical value and then dividing by the group standard deviation, thus avoiding differences in the dimensions of different features. Weight calculation assigns importance based on the representativeness of a feature within the group, calculated as the proportion of samples falling within the feature's statistical interval to the total number of samples across all feature intervals. Segmented weighting uses differentiated penalties based on the direction of deviation. For example, when an individual's feature is below the group mean, only the baseline deviation is calculated; when an individual's feature is above the group mean, an additional penalty is applied to the baseline deviation, strengthening the identification of positive deviations. Multi-dimensional aggregation combines the weighted deviation components of each feature into a single value, integrating information from various dimensions to derive an overall quantitative indicator. Auxiliary verification comparison refers to the technical process of cross-validating an individual model with profiles at other levels besides the core comparison benchmark. For example, when the core benchmark is a level 3 profile, the individual model is compared with the corresponding level 2 and level 1 profiles respectively to verify whether the differences identified at the level 3 profile level also show a consistent trend at a more macro level. If the level 3 profile shows a difference in "low wind speed preference", and the level 2 profile also shows a strengthening trend of "low wind speed usage ratio higher than the group average", then the difference is confirmed to have cross-level consistency, excluding random fluctuations caused by insufficient data in the level 3 profile. The auxiliary verification comparison uses the same difference calculation framework, but the weight allocation is tilted towards macro features. Valid differences refer to differences that are confirmed to be authentic and stable after continuous verification and scenario frequency verification. The screening criteria include both time-duration and scenario-relevance requirements. The time-duration requirement means that the degree of difference or deviation is above a threshold for a continuous six-month period to avoid misjudgments caused by short-term behavioral changes. The scenario-relevance requirement means that the frequency of the difference in the specific usage scenario associated with the difference exceeds 60% of the total frequency of the scenario. For example, the frequency of dehumidification-related differences during the rainy season must exceed 60% of the total operating days in that season, ensuring that the difference is strongly linked to the user's actual usage scenario. The first preset duration refers to the length of the time window used to test the persistence of the difference, which can be set to six months. This duration is determined by analyzing the seasonal cycle of user behavior habits, covering the summer cooling season and the winter heating season to ensure the difference has cross-seasonal stability. The second preset threshold refers to the proportion threshold used to test the scenario frequency, which can be set to 60%. This threshold is determined by statistically analyzing the baseline frequency of user scenario switching. Exceeding this threshold indicates that the difference is not accidental but a necessary behavior driven by the scenario.Mapping rules refer to transformation functions that convert effective difference features into standardized demand labels. This is implemented using a rule engine or decision tree. The input is a triplet of difference type, difference intensity, and associated scenario; the output is a structured label string. The rule base predefines the correspondence between difference features and demand labels. For example, "dehumidification mode activation rate difference > 20% and low fan speed matching difference > 50 points" maps to the label "strong dehumidification - low fan speed linkage demand". Demand labels are standardized text identifiers used to describe users' personalized functional needs, using a hyphenated structure of "functional keyword - scenario / preference keyword", such as "strong dehumidification - low fan speed linkage demand", "ultimate energy saving - 26℃ setting optimization", and "sleep mode + anti-direct blowing depth optimization". After generation, the labels are stored in the device metadata table for matching and calling in subsequent upgrade packages.

[0095] Understandably, establishing a core comparison benchmark and using the highest-priority profile as a reference ensures that difference identification is based on the most refined and relevant group benchmark. If the target device belongs to the energy-saving and low-noise preference level three group of families with children in high-rise apartments in Guangzhou, then this level three profile, rather than the level one profile of the high-temperature and high-humidity area of ​​South China, is used as the core benchmark to avoid the macro-average effect masking micro-individual characteristics. Linkage feature comparison reveals the uniqueness of user environmental response strategies by extracting and comparing differences between individuals and groups in association rules and temporal patterns. For example, an individual might use a low fan speed for dehumidification startup while the group uses a medium fan speed; this difference in linkage logic directly points to the user's personalized demand for a balance between comfort and effectiveness. Quantitative feature comparison transforms abstract behavioral differences into numerical deviation scores by calculating the degree of deviation in set parameters, operating modes, and operation sequences. For example, a 35% deviation in set temperature from the group's energy-saving range or a 62% deviation in the proportion of sleep mode usage provides a comparable and ranking basis for demand intensity assessment. The auxiliary verification and comparison, through cross-validation with the secondary and primary profiles, ensures that the differences identified at the tertiary profile level also show a consistent trend at a more macro level, avoiding accidental misjudgments caused by insufficient data in the tertiary profile. If the low wind speed preference shown in the tertiary profile also shows a strengthening trend in the secondary profile, then the difference is confirmed to have cross-level robustness. Effective difference screening uses dual constraints of time persistence and scenario relevance conditions to filter out short-term fluctuations and scenario-irrelevant pseudo-differences. Differences must persist for more than 6 months and occur more than 60% of the time in a specific scenario to ensure that retained differences are supported by genuine user habits. Demand tag conversion standardizes the verified difference features into machine-readable structured tags through mapping rules, making implicit usage preferences explicit into semantic identifiers that can be matched with upgrade packages. The comparison and screening mechanism ensures that the demand identification process has the rigor of multi-level reference, multi-dimensional quantification, and multi-condition verification. The final output demand tags accurately describe user functional needs and can directly drive subsequent upgrade package matching, achieving a reliable transformation from behavioral differences to demand definitions.

[0096] Preferably, step S5 includes:

[0097] Based on the requirement tags, at least one corresponding upgrade package is matched from the upgrade package library, wherein the upgrade packages in the upgrade package library are pre-associated with core requirement tags and at least one auxiliary tag;

[0098] Perform compatibility checks on the matched upgrade packages to detect whether there are parameter conflicts between the core functional algorithms of different upgrade packages;

[0099] If a conflict exists, conflict handling is performed, which includes generating a merged upgrade package or providing the user with selection information; if no conflict exists, the upgrade package to be pushed is determined.

[0100] Based on the historical operating data of the target air conditioner controller, periods of low device usage frequency are identified as candidate periods for upgrade push.

[0101] Generate and send upgrade prompt information to the target air conditioner controller, the upgrade prompt information including a description of upgrade optimization content based on the difference between the requirement tag and the effective difference;

[0102] Upon receiving the confirmation upgrade instruction, the system sends the upgrade package or integrated upgrade package to be pushed to the target air conditioner controller and controls the target air conditioner controller to complete the installation.

[0103] It should be noted that the requirement tag refers to the standardized text identifier generated in step S4 that describes the user's personalized functional requirements. It adopts a "functional keyword - scenario / preference keyword" structure, such as "strong dehumidification - low fan speed linkage requirement." This tag serves as the primary key for matching upgrade packages and is indexed in the upgrade package library. The upgrade package library refers to a collection of personalized upgrade programs stored on a cloud server, using a relational database or document database. Each upgrade package record includes fields such as upgrade package number, core requirement tag, auxiliary tag list, description of functional optimization direction, algorithm parameter configuration file, version number, and range of compatible device models. The library structure supports quick retrieval by tag and version management. The core requirement tag refers to the identifier of the main functional requirement that the upgrade package addresses. Each upgrade package must be associated with a core requirement tag when entering the library to ensure accurate response to the user's main requirements during matching. Auxiliary tags refer to the compatibility tags attached to the upgrade package, including the compatibility group level (e.g., three-level group compatibility), device attributes (e.g., new and old device compatibility), and usage scenario (e.g., nighttime scenario compatibility). Auxiliary tags are used for secondary verification during matching to ensure compatibility between the upgrade package and the device's usage environment. Compatibility verification refers to the technical process of detecting algorithm conflicts among multiple matched upgrade packages. By extracting the core functional algorithm parameters of each upgrade package and simulating collaborative operation scenarios on the target device, it determines whether there are mutual exclusions in functional logic or parameter ranges such as power allocation, fan speed control, and mode switching. Verification is implemented using a conflict detection engine, supporting both static code analysis and dynamic sandbox simulation. Parameter conflict refers to contradictory parameter settings used by different upgrade packages when controlling the same device component. For example, one upgrade package requires the compressor to maintain high power operation in dehumidification mode to improve dehumidification efficiency, while another upgrade package requires the compressor to reduce power in dehumidification mode to save energy. These two parameter settings cannot be satisfied simultaneously, thus constituting a parameter conflict. Conflict handling refers to the resolution strategies adopted when parameter conflicts are detected. The first strategy is to generate a fusion upgrade package, calling the fusion algorithm module to re-optimize and allocate conflicting parameters, generating a customized upgrade package that meets multiple needs. The second strategy is to provide users with selection information, explaining the characteristics of each upgrade package through pop-ups or app push notifications, allowing users to decide the priority order. The integrated upgrade package refers to a new type of upgrade package generated through conflict resolution, integrating multiple requirement tag functions. It adopts a modular architecture design, breaking down the core algorithms of each upgrade package into independent modules. An adaptation layer coordinates parameter calls between modules, achieving functional overlay rather than simple merging. Low-usage-frequency periods refer to the time intervals within a 24-hour period when the target air conditioner controller is in low-load operation or standby mode. This can be achieved by analyzing historical operating data to statistically analyze indicators such as the device's start-up probability, operating power, and operation frequency during different time periods. Periods with a start-up probability below 20% and an operation frequency below the average level are identified as low-usage-frequency periods. Pushing upgrades during these periods minimizes interference with normal user operation.Candidate time slots refer to specific time points selected from low-frequency usage periods that are suitable for pushing upgrades. The system can select a period with a continuous duration of more than 30 minutes, a stable network connection, and normal device temperature as the final push time to ensure a stable and reliable upgrade process. Upgrade prompt information refers to explanatory text displayed to users before the upgrade is pushed, simultaneously delivered through the device display panel and mobile app. The content includes information such as the basis for matching needs, upgrade optimization points, estimated time consumption, and operation options. The prompt information is generated using template engine technology, dynamically filling preset templates with need tags and difference characteristics. Upgrade optimization content description refers to the part of the prompt information that specifically explains the functional improvements after the upgrade. It is generated based on the effective differences identified in step S4, such as "Your dehumidification mode usage time in the past month was 3 times the regional average. After the upgrade, low-speed dehumidification efficiency is improved by 15%, and nighttime noise is reduced by 8%." This description directly cites the difference data, enhancing the user's perception of the upgrade's value. The upgrade confirmation command refers to the upgrade authorization signal initiated by the user by clicking the "Upgrade Now" button or replying with a confirmation SMS. This command is transmitted to the cloud through an encrypted channel, triggering the upgrade package delivery process. The system records the user's confirmation time and the source of the command for subsequent upgrade effect analysis. The delivery refers to the data transmission process of sending the upgrade package to the target device. Breakpoint resumption technology is used to ensure transmission integrity during network fluctuations. The transmission protocol can use HTTPS, and data transmission is fragmented, with each data segment accompanied by a checksum to ensure transmission accuracy. Installation refers to the firmware update process performed by the device after receiving the upgrade package. The device-side upgrade daemon writes the firmware sequentially in the order of "core functional modules → linkage adaptation modules → configuration parameters." A watchdog is set up to monitor the installation process; if the installation fails, an automatic rollback mechanism is triggered to ensure the device is not bricked due to upgrade failure.

[0104] Understandably, the tag matching process retrieves corresponding upgrade packages from the upgrade package library based on demand tags. Precise matching using core demand tags ensures consistent functional direction, while auxiliary tag verification ensures compatibility between the upgrade package and the device group level and usage scenario, avoiding the push of incompatible upgrade packages. The compatibility verification process performs algorithm conflict detection on multiple matched upgrade packages. By extracting the core functional algorithm parameters of each upgrade package and simulating collaborative operation scenarios, it identifies potential risks of mutual exclusion in parameters such as power allocation and wind speed control in advance, preventing equipment malfunctions or functional failures caused by the simultaneous installation of multiple upgrade packages. The conflict resolution process provides two solutions: when conflicts are reconcilable, the fusion algorithm module is invoked to re-optimize parameter allocation, generating a fused upgrade package that accommodates multiple needs, allowing users to obtain multiple optimizations at once without selection; when conflicts are irreconcilable, the characteristics of each upgrade package are clearly explained to the user, allowing the user to decide the priority based on their own preferences, ensuring user autonomy. The process of determining low-usage periods is based on statistical analysis of historical equipment operation data, including power-on probability and operation frequency. Standby or low-load operating times are selected as candidate upgrade push notifications, such as during weekdays when users are away. This ensures that the upgrade process does not consume the equipment's normal cooling and heating resources and does not affect user comfort. The upgrade notification message generation process dynamically constructs personalized wording based on demand tags and effective differences, directly citing difference data to illustrate the upgrade's value. For example, "Your dehumidification usage time in the past month has been three times the regional average; after the upgrade, dehumidification efficiency will increase by 15%." This explanation based on user-specific data is more persuasive than general function descriptions, significantly increasing user upgrade confirmation rates. The upgrade execution process employs breakpoint resume technology to ensure transmission stability, modular installation sequence, and watchdog monitoring to ensure installation reliability. A success notification is sent to the user upon installation, forming a closed loop. In summary, this transforms the upgrade process from standardized, indiscriminate pushes to personalized, precise pushes; from forced installation to user self-confirmation; from functional descriptions to value perception; and from availability risks to security and controllability, ultimately achieving the dual goals of increased user satisfaction and reduced maintenance costs.

[0105] Preferably, the compatibility verification of the matched upgrade package includes:

[0106] When an upgrade package is matched, extract the core functional algorithm parameters and triggering conditions of each upgrade package;

[0107] Simulate a scenario where the upgrade package operates collaboratively on the target air conditioning controller;

[0108] Based on the simulation results, determine whether there is mutual exclusion in functional logic or parameter range.

[0109] It should be noted that the core function algorithm parameters refer to the key configuration values ​​in the upgrade package used to regulate the behavior of the air conditioner hardware, including the compressor power distribution curve, fan speed mapping table, temperature control PID coefficient, mode switching threshold, dehumidification duration limit, etc. These parameters are stored in the upgrade package's configuration file in JSON or XML format. By parsing the configuration file, the name, value range, default value, and effective conditions of each parameter can be extracted. These parameters directly determine the functional performance of the upgrade package. Trigger conditions refer to the context conditions for activating specific functions of the upgrade package, including environmental parameter thresholds (such as indoor humidity ≥60%), time windows (such as 22:00-7:00), user operation sequences (such as lowering the temperature 3 times consecutively), etc. Trigger conditions are defined in the form of Boolean expressions or rule engine scripts. After parsing, the judgment logic and dependent variables of the conditions can be obtained. Simulation refers to the process of predicting the running effect of the upgrade package through software simulation technology before the actual equipment is installed. Two methods are used: static code analysis and dynamic sandbox execution. Static analysis scans for numerical overlaps and logical contradictions between algorithm parameters, while dynamic sandbox actually executes the upgrade package code in a virtualized device environment, monitoring the operation logs and changes in intermediate variables. Collaborative operation scenarios refer to the composite operating state when multiple upgrade package functions are activated simultaneously. Simulations must consider situations where the triggering conditions of each upgrade package may be met simultaneously. For example, a dehumidification optimization package might be triggered simultaneously in a high-humidity environment, or a sleep optimization package might be triggered simultaneously during nighttime. The simulation engine needs to load the algorithm modules of multiple upgrade packages in parallel and schedule their execution order and parameter interactions. Functional logic mutual exclusion refers to different upgrade packages issuing conflicting instructions to the same device component. For example, one upgrade package requires the compressor to operate continuously at high frequency for rapid cooling, while another requires the compressor to operate intermittently for energy saving. These two logics cannot coexist in the same operating cycle. Parameter range mutual exclusion refers to different upgrade packages setting non-overlapping or conflicting value ranges for the same parameter. For example, one upgrade package limits the dehumidification power range to 60-80%, while another limits it to 40-60%. The two ranges intersect at 60%, but their optimization objectives differ. In actual operation, slight fluctuations may cause frequent switching. Mutual exclusion judgment requires detecting the overlap of parameter ranges and the consistency of objectives.

[0110] Understandably, extracting the core functional algorithm parameters and triggering conditions of each upgrade package provides clear input data for the verification process, transforming the black-box upgrade package into an analyzable and comparable set of structured parameters, and providing quantitative evidence for subsequent conflict detection. Simulating the scenario of upgrade packages running collaboratively on the target device, and reproducing the composite state of multiple functions being activated simultaneously through software simulation, allows for the discovery of dynamic conflicts that static analysis cannot capture before actual deployment, such as two upgrade packages competing for the same memory resource or causing deadlocks. Based on the simulation results, mutual exclusion of functional logic or parameter ranges is determined, refining conflict types into two categories: logical contradictions and parameter range conflicts. Logical contradictions refer to contradictory instructions for the same component, while parameter range conflicts refer to incompatible value ranges for the same parameter. This classification allows for differentiated conflict handling strategies. In summary, verification makes compatibility testing systematic, predictable, and verifiable, ensuring that the pushed upgrade package combination is functionally synergistic and parameter-compatible, eliminating the risk of device instability or user complaints due to conflicts.

[0111] Preferably, the degree of deviation between individual statistical values ​​and population statistical intervals or distribution characteristics is calculated. The following relation is satisfied:

[0112] ;

[0113] ;

[0114] ;

[0115] ;

[0116] in, Indicates the overall degree of deviation. Indicates the first The weighted deviation component of each quantitative feature This represents the first [item] extracted from the individual operational habit model. Individual statistical values ​​of a quantitative characteristic, This indicates the first [item] extracted from the group's operational habit profile, which serves as the core comparison benchmark. The population statistical mean of a quantitative characteristic Indicates the first The standard deviation of the statistical distribution of a quantitative characteristic in a population Indicates the first Normalized weights of each quantized feature, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. For indicator functions, Indicates the first The statistical interval or distribution range of a quantitative characteristic. This represents the total number of quantized features involved in the calculation. Indicates the first Standardized individual statistical values ​​of a quantitative characteristic, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. Indicates the first The statistical interval or distribution range of a quantitative characteristic. Indicates the first in the group The first sample The corresponding feature values ​​of each quantized feature, Indicates the first in the group The first sample The corresponding feature value of each quantized feature.

[0117] It should be noted that, This refers to a specific numerical value of a quantitative feature extracted from an individual's operational habit model. For example, the average preference of individual users for a set temperature is 26.5℃. This value reflects the central tendency of the user's actual operational behavior and serves as the molecular input for deviation calculation. This refers to the statistical mean of the same quantitative feature extracted from the group's operational habits profile. For example, the average preference value of the target group for a set temperature is 27.0℃, which represents the common benchmark of the group. This refers to the standard deviation of the statistical distribution of the quantified feature within the population. For example, if the standard deviation of the temperature distribution in a population is set to 1.2℃, this value quantifies the degree of dispersion within the population and is used for standardization to eliminate the influence of differences in the dimensions of different features. Standardized individual statistical values ​​are those that convert the original deviation values ​​into dimensionless standard scores, making features of different dimensions (such as temperature in °C and wind speed) comparable. A standardized value greater than 0 indicates that the individual feature is higher than the group mean, and a value less than 0 indicates that it is lower than the group mean. Refers to the first The normalized weight of the quantified feature is calculated using a formula, where the numerator is the number of individuals in the population belonging to the quantified feature. Number of samples in each characteristic statistical interval The denominator is the total number of samples across all feature intervals. This weight reflects the representativeness of the feature in the population. The more samples a feature covers, the higher its weight, ensuring that universal features contribute more to the overall deviation. The first in the group The number of samples within a quantitative characteristic statistical interval is obtained by counting the number of device records that meet a specific interval. For example, the number of devices in the energy-saving range of 26-28℃ is set to 86. This is an indicator function whose output is either 1 or 0. Belongs to the characteristic statistical interval or Belongs to the characteristic statistical interval It returns 1 if the condition is met, otherwise it returns 0. This function is used to accurately count the number of samples that fall within the target interval and is the key logical unit for weight calculation. Indicates the first A statistical interval or distribution range of a quantitative characteristic, for example, the energy-saving range of a set temperature can be defined as... This interval is determined by the statistical distribution of the group profile and serves as the basis for judging sample affiliation. Similarly... Indicates the first The statistical interval or distribution range of a quantitative characteristic. This represents the total number of quantified features involved in the calculation, such as when simultaneously calculating three features: set temperature, sleep mode activation rate, and low fan speed usage percentage. This parameter controls the number of dimensions for the overall deviation. The calculation uses a piecewise function design: when When (the individual statistical value is lower than the group statistical mean), At this point, only the basic deviation is calculated; when When (the individual statistical value is higher than the group statistical mean), That is, adding an extra amount on top of the basic deviation. The penalty item strengthens the identification of positive deviations. D refers to the overall degree of deviation, which comprehensively reflects the individual's overall deviation level across multiple dimensions. The larger the D value, the more the individual's operating habits deviate from the group benchmark, and the more unique their needs.

[0118] 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.

[0119] 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 personalized upgrade push notifications for air conditioner controllers based on user behavior profiles, characterized in that, Includes the following steps: S1: Collect and store the operating data of several air conditioner controllers, including environmental parameters and user operation data; S2: Based on the geographical location information in the operational data, construct a profile of the group's operational habits in different geographical regions; S3: For the target air conditioner controller, build an individual operating habit model based on historical operating data; S4: Compare the individual operation habit model of the target air conditioner controller with the group operation habit profile of the geographical area to which the target air conditioner controller belongs, and generate at least one demand label to describe personalized needs based on the differences found in the comparison. S5: Match the corresponding personalized upgrade package from the upgrade package library according to the demand tag, and push the personalized upgrade package to the target air conditioner controller; Step S2 includes: Based on geographical location and climate characteristics, several air conditioner controllers are divided into first-level groups, and a first-level group operation habit profile is constructed based on the corresponding operation data. From the operational data on which the first-level group operation habit profile is based, a first data subset that conforms to the dimensions of living environment and user attributes is selected, and a second-level division is carried out based on the first data subset to construct a second-level group operation habit profile. A second subset of data that matches the core operational habit preference dimension is selected from the first subset of data, and a third-level division is performed based on the second subset of data to construct a third-level group operational habit profile; A comparison priority is set for the group operation habit profiles of the first level, the second level and the third level, wherein the group operation habit profile of the third level has the highest priority and the group operation habit profile of the first level has the lowest priority. Step S4 includes: The priority of the group operation habit profiles at each level that match the target air conditioner controller is determined, and the group operation habit profile with the highest priority is used as the core comparison benchmark. The individual operation habit model is compared with the core comparison benchmark by linkage feature comparison, including: extracting the association rules or time sequence patterns between environmental parameters of individuals and groups and user operation data, and calculating the degree of difference between the extraction results in triggering conditions and operation response. The individual operation habit model is compared with the core comparison benchmark by quantitative feature comparison, including: calculating the degree of deviation between individual statistical values ​​and group statistical intervals or distribution characteristics based on statistical indicators of set parameter preferences, usage ratio of operation mode and operation sequence. Perform auxiliary verification and comparison to compare the individual operation habit model with at least one other group operation habit profile other than the core comparison benchmark, in order to verify whether the degree of difference and deviation is consistent at the corresponding group level. Based on the degree of difference and the degree of deviation, valid differences are selected from the comparison results. The selection criteria for valid differences include that the degree of difference or the degree of deviation continuously exceeds a preset threshold for a first preset duration, and that the frequency of occurrence of the difference exceeds a second preset threshold in a specific use case associated with the difference. According to the preset mapping rules, each effective difference is converted into a requirement label describing personalized functional needs. The mapping rules map the difference features into standardized labels that include functional directions and usage scenarios or preference keywords.

2. The personalized upgrade push method for air conditioner controllers based on user behavior profiles according to claim 1, characterized in that, The environmental parameters include indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and geographical location information; The user operation data includes temperature setting, fan speed selection records, operating mode usage records, timed on / off records, and energy-saving mode activation records.

3. The personalized upgrade push method for air conditioner controllers based on user behavior profiles according to claim 1, characterized in that, Step S3 includes: Clean the historical operating data of the target air conditioner controller, including removing abnormal operation data and abnormal equipment operating parameters, and completing the missing data; Perform correlation analysis or time series analysis on the cleaned historical operation data to extract the linkage features between environmental parameters and user operation data. The linkage features include features that reflect the correlation rules or time series patterns between environmental conditions and user operation behavior. The user operation data in the cleaned historical operation data is processed, including at least one of statistical distribution analysis, behavior classification and sequence pattern mining, to extract quantitative features. The quantitative features include features that reflect the user's statistical preferences or classification results for set parameters, operation modes and operation sequences. By integrating the linkage features and the quantitative features, the individual operation habit model is constructed and optimized. The individual operation habit model represents the stable operation pattern and environmental adaptation preference of the target air conditioner controller.

4. The method for personalized upgrade push of air conditioner controller based on user behavior profile as described in claim 1, characterized in that, Step S5 includes: Based on the requirement tags, at least one corresponding upgrade package is matched from the upgrade package library, wherein the upgrade packages in the upgrade package library are pre-associated with core requirement tags and at least one auxiliary tag; Perform compatibility checks on the matched upgrade packages to detect whether there are parameter conflicts between the core functional algorithms of different upgrade packages; If a conflict exists, conflict handling is performed, which includes generating a merged upgrade package or providing the user with selection information; if no conflict exists, the upgrade package to be pushed is determined. Based on the historical operating data of the target air conditioner controller, periods of low device usage frequency are identified as candidate periods for upgrade push. Generate and send upgrade prompt information to the target air conditioner controller, the upgrade prompt information including a description of upgrade optimization content based on the difference between the requirement tag and the effective difference; Upon receiving the confirmation upgrade instruction, the system sends the upgrade package or integrated upgrade package to be pushed to the target air conditioner controller and controls the target air conditioner controller to complete the installation.

5. The personalized upgrade push method for air conditioner controllers based on user behavior profiles according to claim 4, characterized in that, The compatibility check for the matched upgrade package includes: When an upgrade package is matched, extract the core functional algorithm parameters and triggering conditions of each upgrade package; Simulate a scenario where the upgrade package operates collaboratively on the target air conditioning controller; Based on the simulation results, determine whether there is mutual exclusion in functional logic or parameter range.

6. The method for personalized upgrade push of air conditioner controller based on user behavior profile as described in claim 1, characterized in that, Calculate the degree of deviation between individual statistical values ​​and population statistical intervals or distribution characteristics. The following relation is satisfied: ； ； ； ； in, Indicates the overall degree of deviation. Indicates the first The weighted deviation component of each quantitative feature This represents the first [item] extracted from the individual operational habit model. Individual statistical values ​​of a quantitative characteristic, This indicates the first [item] extracted from the group's operational habit profile, which serves as the core comparison benchmark. The population statistical mean of a quantitative characteristic Indicates the first The standard deviation of the statistical distribution of a quantitative characteristic in a population Indicates the first Normalized weights of each quantized feature, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. For indicator functions, Indicates the first The statistical interval or distribution range of a quantitative characteristic. This represents the total number of quantized features involved in the calculation. Indicates the first Standardized individual statistical values ​​of a quantitative characteristic, Indicates that the group belongs to the first The number of samples in each quantitative feature statistical interval. Indicates the first The statistical interval or distribution range of a quantitative characteristic. Indicates the first in the group The first sample The corresponding feature values ​​of each quantized feature, Indicates the first in the group The first sample The corresponding feature value of each quantized feature.

Images