User equipment operation preference data query method, device and electronic equipment

By constructing a dual-path preference analysis model that combines a weighted local density clustering algorithm with time-period statistical analysis, the problem of insufficient judgment on the correlation between operation and environmental conditions in smart home systems is solved. This enables proactive prediction and personalized recommendations of device operations, thereby improving the automation level and user experience of smart homes.

CN123115233APending Publication Date: 2026-11-17GUANGDONG HELING CONSTRUCTION TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611514806.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing smart home systems lack the ability to dynamically determine the relationship between operation and environmental conditions, and cannot distinguish between environment-driven and time-based regular operation modes, resulting in a single preference matching path and recommendation results that do not match the user's actual habits.

Method used

A dual-path preference analysis model combining a weighted local density clustering algorithm and time-period statistical analysis is constructed. By generating multi-dimensional feature vectors, the correlation strength between user operation behavior and environmental data is identified, and environmental-driven and time-period-driven operation modes are dynamically distinguished to achieve proactive prediction and personalized recommendation of device operation.

Benefits of technology

It accurately identifies the correlation between user behavior and environmental data, dynamically distinguishes operating modes, improves the automation level and user experience of smart homes, and enables proactive prediction and personalized recommendations of device operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115233A_ABST
    Figure CN123115233A_ABST
Patent Text Reader

Abstract

This application provides a method, apparatus, and electronic device for querying user device operation preference data. The method involves obtaining a user device operation preference data query request; generating a multi-dimensional feature vector based on the current time and environmental data in the request; querying the correlation results between user device operations and the environment for a target device type from a database to determine a target matching path; wherein the database also stores preference records corresponding to environmental clustering matching paths and time-period statistical matching paths, determined by density-based incremental behavioral clustering analysis and time-period statistical analysis; calculating and sorting the matching scores of multiple preference records under the target matching path with the multi-dimensional feature vector; and making preference recommendations based on the sorted preference records. This application can accurately identify the correlation between user operations and the environment, distinguish between environment-driven and time-period-driven modes, achieve accurate recommendations of device operation preferences, and significantly improve the user experience of smart homes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of smart home and artificial intelligence technologies, and in particular to a method, device and electronic device for querying user device operation preference data. Background Technology

[0002] With the rapid development of IoT and smart home technologies, users' demand for intelligent control of home devices is increasing. However, existing systems generally suffer from the following technical defects: a lack of dynamic judgment ability regarding the correlation between operation and environmental conditions, an inability to distinguish between environment-driven operations and time-based regular operations, resulting in a single preference matching path and recommendation results that do not match users' actual habits. Summary of the Invention

[0003] The purpose of this application is to provide a method, device, and electronic device for querying user device operation preference data. By constructing a dual-path preference analysis model that combines a weighted local density clustering algorithm with time-period statistical analysis, it can accurately identify the correlation strength between user operation behavior and environmental data, dynamically distinguish between environment-driven and time-period-driven operation modes, realize proactive prediction and personalized recommendation of device operation preferences, and significantly improve the automation level and user experience of smart homes.

[0004] Firstly, this application provides a method for querying user device operation preference data. The method includes: obtaining a user device operation preference data query request; the request carries a target device type, current time, and may also include environmental data; generating a multi-dimensional feature vector based on the current time and environmental data; querying the correlation results between user device operations and the environment of the target device type from a database to determine the target matching path; wherein, the database also stores preference records corresponding to environmental clustering matching paths and time period statistical matching paths, respectively, obtained as follows: obtaining the user's historical device operation data; performing behavior operation chain aggregation and device linkage control detection on the historical device operation data to determine the aggregated operation chain data; for each operation chain... The system analyzes the correlation between user operations and the environment to determine the relationship between device operations and the environment. If the correlation is strong, it performs density-based incremental behavioral clustering analysis on the operation chain data to generate device environment preference records under the environment clustering matching path. If the correlation is weak or non-correlation, it performs time-period statistical analysis on the operation chain data to generate device environment preference records under the time-period statistical matching path. The system stores the correlation results, device environment preference records under the environment clustering matching path, and device environment preference records under the time-period statistical matching path. It calculates and sorts the matching scores of multiple preference records under the target matching path with the multi-dimensional feature vector. Based on the sorted preference records, it performs preference recommendations.

[0005] Furthermore, the steps for generating a multidimensional feature vector based on the current time and environmental data include: extracting the hour and month from the current time, encoding the hour as sine and cosine components, encoding the month as sine and cosine components to obtain a four-dimensional feature vector; the environmental data includes at least one of the following: indoor temperature, outdoor temperature, humidity, and light intensity; quantizing and normalizing the indoor temperature stepwise, quantizing and normalizing the outdoor temperature stepwise, quantizing and normalizing the humidity according to the segment boundaries, quantizing and normalizing the light intensity according to the segment boundaries, and combining the four-dimensional feature vector to obtain a multidimensional feature vector.

[0006] Furthermore, the steps described above for querying the correlation results between user equipment operations and the environment of the target device type from the database and determining the target matching path include: if the correlation result is environment-related, determining the environment clustering matching path as the target matching path; if the correlation result is environment-independent, determining the time period statistical matching path as the target matching path; if the correlation result is no correlation record, determining both the environment clustering matching path and the time period statistical matching path as the target matching path.

[0007] Furthermore, the steps described above for calculating the matching scores of multiple preference records under the target matching path with the multidimensional feature vector include: determining the device serial number and room number corresponding to the target device; querying all valid preference records corresponding to the device serial number and room number from the device environment preference records; when the target matching path is an environment clustering matching path, calculating the Euclidean distance between the cluster center vector and the multidimensional feature vector corresponding to each valid preference record; filtering out valid preference records with a distance less than a threshold; determining the comprehensive score for each filtered valid preference record based on the sum of the similarity score and the confidence score; when the target matching path is a time period statistical matching path, determining the time period based on the number of hours of the current time; filtering out target preference records whose scene tags match the time period from all valid preference records; and determining the comprehensive score of the target preference record based on the sum of the time period matching score and the confidence score.

[0008] Furthermore, the steps described above for aggregating historical equipment operation data into behavioral operation chains and detecting equipment linkage control to determine the aggregated operation chain data include: grouping historical equipment operation data by equipment serial number and room number, and sorting them by operation trigger time within each group; for the same equipment, if the time interval between two operation records is within a preset time interval, merging the attributes of the latter operation record into the former operation record; using a specified duration as a time window, merging multiple operation records within the time window into one operation chain data, and assigning different operation weights according to the control source of each operation record; for each operation chain data, taking the last valid environmental data record within the time window as an environmental snapshot of the operation chain; or further including: inputting the aggregated operation chain data into a continuous operation analyzer, which checks whether the operation chain data contains continuous control of multiple different devices, and if so, recording equipment linkage control information for the operation chain data.

[0009] Furthermore, the steps described above for performing density-based incremental behavioral clustering analysis on the operation chain data to generate device environment preference records under the environment clustering matching path include: constructing a multi-dimensional environment feature vector for each operation record in the operation chain data; calculating a weighted local density for each operation record based on the corresponding multi-dimensional environment feature vector; calculating a dynamic minimum density threshold, and determining the operation record as a core point if the weighted local density of the operation record is less than the dynamic minimum density threshold; traversing all core points, forming a cluster for each core point and all points in its neighborhood, and calculating the cluster center vector and confidence for each cluster; treating each cluster as a valid preference record to generate device environment preference records under the environment clustering matching path; each valid preference record corresponds to a cluster center vector; the method also includes: incrementally updating the cluster center, total weight, and number of samples when new data matches a historical cluster.

[0010] Furthermore, the above-mentioned step of calculating the weighted local density based on the multidimensional environmental feature vector corresponding to the operation record includes: taking the operation record as a sample point, determining the feature-weighted Euclidean distance between the sample point and other operation records according to the feature value of each feature in each dimension, the weight coefficient of each feature, and the standard deviation; taking the operation records whose feature-weighted Euclidean distance is less than a threshold as samples in the neighborhood of the sample point; calculating the time weight of each sample and summing them to obtain the weighted local density of the sample point; the time weight is determined based on the difference in days between the current analysis time and the operation trigger time and the half-life of the device type.

[0011] Furthermore, the steps described above for performing time-period statistical analysis on the operation chain data to generate device environment preference records under the time-period statistical matching path include: dividing a day into multiple time periods; for each time period, grouping and statistically analyzing the raw data of historical device operation behavior according to three dimensions; the three dimensions include: device type, device number, and attribute name; for numerical attributes, calculating the following indicators: weighted mean, weighted median, mode, and number of samples; for enumerated attributes, calculating the following indicators: most recent mode, frequency distribution, and number of samples; based on the indicators calculated for each time period, generating a user preference device operation behavior profile for each time period to describe the user's most typical device usage pattern within that time period; and forming device environment preference records under the time-period statistical matching path based on the user preference device operation behavior profiles corresponding to multiple time periods.

[0012] Secondly, this application also provides a user device operation preference data query device, the device comprising: a request acquisition module, used to acquire a user device operation preference data query request; the request carries a target device type, current time, or may also include environmental data; a feature determination module, used to generate a multi-dimensional feature vector based on the current time and environmental data; and a path determination module, used to query the correlation results between user device operation and environment of the target device type from the database, and determine the target matching path; wherein, the database also stores preference records corresponding to environmental clustering matching paths and time period statistical matching paths, respectively, and the acquisition method is as follows: acquiring the user's historical device operation data; performing behavior operation chain aggregation and device linkage control detection on the historical device operation data to determine the aggregated operation chain data; for each The system analyzes the correlation between user operations and the environment across multiple operation chains to determine the correlation between device operations and the environment. If the correlation is strong, it performs density-based incremental behavioral clustering analysis on the operation chain data to generate device environment preference records under the environment clustering matching path. If the correlation is weak or non-correlation, it performs time-period statistical analysis on the operation chain data to generate device environment preference records under the time-period statistical matching path. The system stores the correlation results, device environment preference records under the environment clustering matching path, and device environment preference records under the time-period statistical matching path. A score calculation module calculates and sorts the matching scores of multiple preference records under the target matching path against the multi-dimensional feature vector. A preference recommendation module performs preference recommendations based on the sorted preference records.

[0013] Thirdly, this application also provides an electronic device, including a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method described in the first aspect above.

[0014] Most current smart home systems remain in a passive response stage (such as voice control and timed on / off switches), lacking the ability to proactively predict and personalize services. This application provides a method, device, and electronic device for querying user device operation preference data. By constructing a dual-path preference analysis model combining a weighted local density clustering algorithm and time-period statistical analysis, it can accurately identify the correlation strength between user operation behavior and environmental data, dynamically distinguishing between environment-driven and time-period-driven operation modes. Based on an incremental clustering mechanism, the system continuously learns the evolution of user preferences, achieving proactive prediction and personalized recommendations for device operations, significantly improving the automation level and user experience of smart homes, truly achieving "understanding user needs and anticipating their needs." Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a user equipment operation preference data query method provided in this application embodiment; Figure 2 This is a schematic diagram of the data analysis process in another user equipment operation preference data query method provided in this application embodiment; Figure 3 A schematic diagram illustrating a weighted clustering analysis process provided in an embodiment of this application; Figure 4 A structural block diagram of a user equipment operation preference data query device provided in this application embodiment; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0017] The technical solutions of this application will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] Existing user operation device preference analysis systems generally suffer from the following problems: (1) The processing of user operation behavior is limited to the level of single instruction, lacking a mechanism to aggregate scattered operations into complete operation intentions, and unable to understand the linkage behavior pattern of users continuously controlling multiple devices. (2) Preference analysis methods ignore the differences in the sources of operation triggers and treat manual user operations, timed triggers, and scene linkages as the same, resulting in insufficient preference learning accuracy; (3) Lack of dynamic judgment ability on the correlation between operation and environmental conditions, unable to distinguish between environment-driven operation and time-based regular operation, resulting in a single preference matching path and recommendation results that do not match the user's actual habits; (4) The clustering results of offline analysis cannot be updated incrementally. Each time, all historical data needs to be reprocessed, which is computationally inefficient and makes it difficult to reflect the gradual evolution of user preferences in real time.

[0019] Based on this, embodiments of this application provide a method, apparatus, and electronic device for querying user device operation preference data. By constructing a dual-path preference analysis model that combines a weighted local density clustering algorithm with time-period statistical analysis, it can accurately identify the correlation strength between user operation behavior and environmental data, dynamically distinguish between environment-driven and time-period-driven operation modes, realize proactive prediction and personalized recommendation of device operation preferences, and significantly improve the automation level and user experience of smart homes.

[0020] To facilitate understanding of this embodiment, a user equipment operation preference data query method disclosed in this application embodiment will be described in detail first.

[0021] Figure 1 A flowchart illustrating a user equipment operation preference data query method provided in this application embodiment, the method specifically includes the following steps: Step S102: Obtain a user device operation preference data query request; the request carries the target device type, current time, or may also include environmental data; the environmental data includes at least one of the following: indoor temperature, outdoor temperature, humidity, and light intensity.

[0022] Step S104: Generate a multi-dimensional feature vector based on the current time and environmental data; Extract the hour and month from the current time, encode the hour into sine and cosine components, and encode the month into sine and cosine components to obtain a four-dimensional feature vector; quantize and normalize the indoor temperature step, quantize and normalize the outdoor temperature step, quantize and normalize the humidity according to the segment boundary, and quantize and normalize the light intensity according to the segment boundary, and combine the four-dimensional feature vector to obtain a multi-dimensional feature vector.

[0023] Step S106: Query the correlation results between user equipment operation and environment of the target device type from the database to determine the target matching path; wherein, the database also stores preference records corresponding to environment clustering matching path and time period statistical matching path respectively; If the correlation result indicates that the environment is relevant, the environmental clustering matching path is determined as the target matching path; if the correlation result indicates that the environment is not relevant, the time period statistical matching path is determined as the target matching path; if the correlation result indicates that there are no relevant records, both the environmental clustering matching path and the time period statistical matching path are determined as the target matching path.

[0024] The methods for obtaining preference records corresponding to environmental clustering matching paths and time period statistical matching paths are as follows: (1) Obtain the user's historical device operation data; the historical device operation data includes: multiple operation records; each operation record includes: device serial number, room number, control attribute, attribute value, control source, operation trigger time, device type, and device number; (2) Aggregate the behavior operation chain and detect the equipment linkage control of historical equipment operation data to determine the aggregated operation chain data; (3) For each operation chain data, perform a correlation analysis between user operation and environment to determine the correlation results between equipment operation and environment; (4) If the correlation results are strong, perform density-based incremental behavior clustering analysis on the operation chain data to generate device environment preference records under the environment clustering matching path; (5) If the correlation result is weak or no correlation, perform time-period statistical analysis on the operation chain data and generate equipment environment preference records under the time-period statistical matching path; (6) Store the correlation results, the device environment preference records under the environment clustering matching path, and the device environment preference records under the time period statistical matching path; Step S108: Calculate and sort the matching scores of multiple preference records under the target matching path with the multidimensional feature vector; The matching score is calculated as follows: First, determine the device serial number and room number corresponding to the target device. Second, query all valid preference records corresponding to the device serial number and room number from the device environment preference records. Third, when the target matching path is an environment clustering matching path, calculate the Euclidean distance for each valid preference record based on the cluster center vector and multidimensional feature vector. Fourth, filter out valid preference records with a distance less than a threshold. Fifth, for each filtered valid preference record, determine the comprehensive score based on the sum of the similarity score and the confidence score. Sixth, when the target matching path is a time-segment statistical matching path, determine the time period based on the current hour. Sixth, filter out target preference records whose scene tags match the time period from all valid preference records. Finally, determine the comprehensive score of the target preference record based on the sum of the time period matching score and the confidence score.

[0025] Step S110: Based on the sorted preference records, make preference recommendations.

[0026] See Figure 2 The flowchart shown is another method for querying user equipment operation preference data. This method is implemented based on the previous embodiment. The key point is that it first describes the analysis process of preference records under two paths, and then explains the user equipment operation preference data query process.

[0027] 1. Offline preference analysis of user device operation behavior.

[0028] First, data preprocessing: (1) Based on the current device serial number, room number and set time range (start time to end time), query the corresponding original operation data from the original operation record table of the algorithm library.

[0029] Each original operation log contains the following key information: Device serial number: Identifies the device gateway where the operation occurred; Room number: Identifies the room where the operation took place; Operation trigger time: The precise time when the user's operation occurred; Equipment type: The type of equipment being operated, such as air conditioners, lights, curtains, fresh air systems, etc.; Equipment number: The specific number of the equipment being operated; Control attributes: The specific attributes being manipulated, such as temperature, wind speed, brightness, on / off status, etc. Attribute value: The specific numerical value or state set for the operation; Control source: The source channel of the operation trigger, such as manual operation by the user, voice control, intelligent scene trigger, timed trigger, third-party control, etc.; (2) Behavioral data operation chain aggregation.

[0030] The raw data may contain instances where the same user performs multiple operations on the same device within a very short period (e.g., first setting it to 25 degrees Celsius, then adjusting it to 26 degrees Celsius a few seconds later). These scattered micro-operations need to be aggregated into a meaningful "single operation." A time window aggregation strategy is adopted, and the specific aggregation process is as follows: (2.1) Group the historical equipment operation data by equipment serial number and room number, and sort them by operation trigger time within each group; (2.2) For the same device, if the time interval between two operation records is within a preset time interval, the attributes of the later operation record will be merged into the earlier operation record; for example, if the time interval between two operation records is within 120 seconds, the attributes of the later record will be merged into the earlier record. This means that multiple attribute adjustments of the same device within 2 minutes are considered as different steps of the same operation intent.

[0031] (2.3) Using a specified duration as a time window, merge multiple operation records within the time window into one operation chain data, and assign different operation weights according to the control source of each operation record; For example, using a 5-minute time window, multiple operation records within the window are merged into an "operation chain." An operation chain represents a user's complete operational intent within a continuous time period, including simultaneous or continuous control of multiple devices. Different operation weights are assigned based on the control source of each operation record: direct user operation and manual control: highest weight, 1.0; device linkage trigger, voice panel control, and voice control: weight 0.9; intelligent scene trigger, third-party control, and timed trigger: weight 0.8; system recommendation trigger: lowest weight, 0.7. Operation records with higher weights have a greater influence in subsequent preference statistics.

[0032] (2.4) For each operation chain data, the last valid environmental data record in the time window is taken as the environmental snapshot of the operation chain; the environmental snapshot at the time of operation includes environmental data, including indoor temperature, outdoor temperature, humidity, light intensity, etc.

[0033] (2.5) Input the aggregated operation chain data into the continuous operation analyzer. The analyzer checks whether the operation chain data contains continuous control of multiple different devices. If so, it records the device linkage control information of the operation chain data.

[0034] Based on the aggregated data, the system checks whether the user performed continuous operations on multiple devices within a short period of time, i.e., whether a "device linkage" behavior pattern exists. Specifically, the aggregated operation data is input into a continuous operation analyzer, which checks whether the operation chain includes continuous control of multiple different devices. For example, a user might close the curtains, turn on the air conditioner, and dim the lights within two minutes, which constitutes a continuous operation of a "movie-watching mode."

[0035] The criteria for judgment are: the operation chain contains multiple different types of equipment, the time interval between operations is short enough, and it exhibits obvious linkage characteristics.

[0036] The result processing procedure is as follows: If a continuous operation mode is detected, the system saves the analysis results of the continuous operation to the database, recording the trigger time, the conditions met, and the types of devices involved; if no continuous operation mode is detected, the system records the log and skips the process, without blocking the subsequent clustering analysis process; if an anomaly occurs during the continuous operation analysis process, the system captures the anomaly and records the error log, without blocking the subsequent process.

[0037] 2. Behavioral clustering analysis after preprocessing: density-based incremental clustering method.

[0038] The process is as follows: Figure 3 As shown: (1) Construct a multidimensional environmental feature vector for each operation record in the operation chain data; For each aggregated operation chain record, extract the environmental data at the time the operation occurred, and construct an 8-dimensional environmental feature vector: The input data is the raw environmental data (time, temperature, humidity, light, etc.), and the output is an 8-dimensional normalized feature vector: V=[v1,v2,...,v8].

[0039] 1.1) Time Encoding (Dimensions 1-2): ; ; Where h ∈ [0, 23] represents the number of hours (24-hour clock).

[0040] 1.2) Seasonal coding (dimensions 3-4): ; ; Where m ∈ [1, 12] represents the month.

[0041] 1.3) Temperature normalization (dimensions 5-6): Step quantization function: ; Where: T = temperature value (°C); Tmin = 10°C (lower temperature limit); T step = 2°C (step interval); Nsteps = 28 (total number of steps); clamp(x,a,b)=max(a,min(b,x)).

[0042] The normalization formula is as follows:

[0043] 1.4) Humidity normalization (dimension 7): Segmented interval definition: Bhum = [0,20,30,40,50,60,70,80,100]; Segmented index calculation: indexhum(H) = max{i | Bhum[i] ≤ H}; Normalization formula: ; Where: H represents the original humidity value, Bhum represents the humidity segment boundary array, and index hum (H) represents the segment index number where humidity H is located.

[0044] 1.5) Illumination normalization: Segmented interval definition: Blux=[0,50,100,150,300,500,1000]; Segmented index calculation: indexlux(L) = max{i | Blux[i] ≤ L}; Normalization formula: ; Where: L represents the original light intensity value, Blux represents the light segment boundary array, and index... lux (L) represents the segment index number where the light intensity L is located.

[0045] (2) For each operation record, calculate the weighted local density based on the multidimensional environmental feature vector corresponding to the operation record; (2.1) Using operation records as sample points, determine the feature-weighted Euclidean distance between the sample points and other operation records based on the feature values ​​of each feature in each dimension, the weight coefficients of each feature, and the standard deviation of each feature. Environmental similarity between two operation records, adaptive distance based on feature importance: in Let be the standard deviation of the i-th feature, used to eliminate the influence of different feature dimensions. Let represent the feature-weighted Euclidean distance between operation records p and q, where p and q represent two operation records, and i represents the i-th dimension. This represents the i-th eigenvalue of operation record p. This represents the i-th eigenvalue of operation record p. This represents the weight coefficient of the i-th feature.

[0046] Different environmental features contribute differently to clustering, which is adjusted by the feature weight vector α.

[0047] α = [a1, a2, a3, ..., a8] ; Constraints: .

[0048] The weighting principles are shown in Table 1 below: Table 1

[0049] (2.2) Operation records whose feature-weighted Euclidean distance is less than the threshold are taken as samples in the neighborhood of the sample point; Neighborhood definition: ; Where ε is the distance threshold (default 0.5). Let represent the set of all sample points within the ε-neighborhood of sample point p. D This represents the set of operation records for the current batch. This represents the feature-weighted Euclidean distance between sample points p and q.

[0050] (2.3) Calculate the time weight of each sample and sum them to obtain the weighted local density of the sample points; the time weight is determined based on the difference in days between the current analysis time and the operation trigger time and the half-life of the device type.

[0051] The device's adaptive time weighting reflects the freshness of operation records, with more recent operations receiving higher weight. The formula is as follows: ; Among them: W t The time weight of the operation record at time t is represented by Δt = current analysis time t. now With the time t of the operation op The difference in the number of days between them, T 1 / 2 = Half-life, which is automatically adjusted according to different device types. Air conditioning and fresh air systems have a half-life of 15 days, while lighting and curtains have a half-life of 30 days.

[0052] The weighted local density of a sample point p is the sum of the time weights of all samples in its neighborhood.

[0053] Weighted local density: ; in, It is an indicator function that excludes itself. w represents the weighted local density of sample point p. (q) This represents the time weight of sample point q.

[0054] (3) Calculate the dynamic minimum density threshold. If the weighted local density of the operation record is less than the dynamic minimum density threshold, determine the operation record as the core point. Calculate the dynamic minimum density threshold MinPts=max(3,0.1×|D|) to determine whether each record is a core point. ——ρ[p]≥MinPts.

[0055] Sample points whose weighted local density exceeds the minimum density threshold.

[0056] ; Dynamic MinPts calculation: MinPts=max(3, 0.1×|D| Where |D| represents the number of unmatched records in the current batch. ρ(p) is the weighted local density of sample point p.

[0057] (4) Traverse all core points, form a cluster for each core point and all points in its neighborhood, and calculate the cluster center vector and confidence for each cluster; The algorithm iterates through all unvisited core points. For each unvisited core point, a new cluster is created, adding the core point and all its neighborhood points to the cluster. This process is repeated recursively for core points within their neighborhoods. After clustering, a post-processing phase is initiated: for each cluster, its weighted average cluster center and confidence score are calculated. For all unclassified sample points, if their time weight is greater than 0.8, they are temporarily stored in a buffer for the next batch of processing; otherwise, they are discarded. Finally, the cluster sets and noise points are output, completing the entire analysis process.

[0058] After clustering is complete, the system needs to calculate the representative centers of each cluster. Cluster centers Calculated using a weighted average method: ; Where C is the 8-dimensional feature vector of the cluster center, cluster is the set of all sample points in a certain cluster, and p is a sample point in the cluster. The time weight of sample point p at time t, V[p] is the 8-dimensional environmental feature vector of sample point p.

[0059] The advantage of weighted average is that recent operations with high time weight have a greater impact on cluster centers, enabling cluster centers to more accurately reflect users' recent preferences rather than being dominated by a large amount of historical data.

[0060] (5) Treat each cluster as a valid preference record and generate device environment preference records under the environment clustering matching path; each valid preference record corresponds to a cluster center vector; (6) When new data matches a historical cluster, the cluster center, total weight and number of samples are updated incrementally.

[0061] When new data matches historical clusters, the system needs to incrementally update the cluster centers, total weights, and sample size. The calculation of new cluster centers uses a weighted merging method to ensure smooth evolution of cluster centers when new data is introduced. The new total weight is the sum of the historical total weights and the time weights of all samples in the new data. The new sample size is the sum of the historical sample size and the sample size in the new data. This incremental update mechanism allows clustering to continuously learn from changes in user preferences without reprocessing all historical data. If the incremental clustering analysis produces new clusters (i.e., the number of new clusters is greater than 0), the new clustering results are saved to the user environment preference table.

[0062] 3. Determine the correlation between user operations and the environment.

[0063] The system does not simply treat all operations as environment-driven, but rather uses statistical analysis to determine the strength of the correlation between operations and the environment, thereby dynamically adjusting the feature weights of clustering.

[0064] Pearson correlation coefficient: Calculates the degree of linear correlation between user action attribute values ​​and environmental data. If the absolute value of the correlation coefficient is greater than 0.7, the user action is strongly correlated with the environment, and clustering analysis algorithm is used to analyze the user's actions.

[0065] Strong correlation indicates that users' device adjustment behavior is mainly triggered by environmental conditions (e.g., "turning down the air conditioner when it's hot"). Such operations directly enter the incremental clustering process, where the environmental dimension is given high weight when constructing the environmental feature vector, ensuring that similar operations under the same environment are aggregated together.

[0066] If the absolute value of the correlation coefficient is less than or equal to 0.7, it is considered that user operations are strongly uncorrelated with the environment. This means that the user's device adjustment behavior is not significantly related to environmental conditions, but is more driven by daily routines, personal habits, or fixed scenarios (e.g., "always turning on the night light before going to bed every night," regardless of the intensity of light at the time). Such operations are not included in the regular environmental clustering, but instead undergo time-based statistical analysis to specifically uncover their time-based operational patterns throughout the day.

[0067] 4. Time-period statistical analysis (for non-environment-driven operations) For operations determined to be weakly or unrelated, the system performs time-period statistical analysis to identify users' preferred device operation behaviors at different times of the day. The specific process for generating preference records is as follows: (1) Divide the day into multiple time periods; the system assigns each non-environment-driven operation to the corresponding time period based on the hour in which it occurs. The day is divided into the following six time periods, as shown in Table 2 below: Table 2

[0068] (2) For each time period, the raw data of historical equipment operation behavior are grouped and statistically analyzed according to three dimensions; the three dimensions include: equipment type, equipment number and attribute name; For numerical attributes (such as target temperature, brightness, and color temperature), the following metrics are calculated: Weighted average: The average value calculated with time decay weights, with more recent operations having higher weights; Weighted median: The median value is taken after sorting by time weight, which is more resistant to extreme values. Mode: The setting value that appears most frequently, representing the target that users choose most often; Number of samplings: The total number of historical operation records for this attribute within this time period.

[0069] For enumerated attributes, the following metrics are calculated: Recent mode: Prioritize values ​​that have appeared twice or more recently, reflecting the latest trend; Frequency distribution: the number of times each enumerated value appears and its percentage; Number of samplings: The total number of historical operation records for this attribute within this time period.

[0070] (3) Based on the statistical indicators for each time period, generate a user preference device operation behavior profile for each time period to describe the most typical device usage pattern of the user during that time period; (4) Based on the user's preferred device operation behavior profiles corresponding to multiple time periods, device environment preference records are formed under the time period statistical matching path.

[0071] Based on the above statistics, the system generates a "user preferred device operation behavior" profile for each time period, describing the user's most typical device usage patterns during that period. The user preferred device operation behaviors for different time periods are stored in a user environment preference table.

[0072] 5. Preference data query and usage.

[0073] Whether it's a room-wide query or a single device query, the underlying query service follows the same core processing logic. Before performing preference matching, the system first determines whether the current user's operation on the target device type is related to the environmental data, and selects different matching paths accordingly.

[0074] Step 1: Analyze the current time information.

[0075] The system obtains the current time and extracts the hour and month information to construct the time and seasonal components in the environmental feature vector.

[0076] Step 2: Construct the feature vector of the current environment.

[0077] If environmental data is provided in the request, the system constructs an 8-dimensional feature vector using the same encoding method as in the preference analysis phase: the hours are encoded as sine and cosine components, the months are encoded as sine and cosine components; indoor temperature is quantized and normalized in 2°C increments; outdoor temperature is quantized and normalized in 2°C increments; humidity is quantized and normalized according to segment boundaries; and light intensity is quantized and normalized according to segment boundaries. If no environmental data is provided, only the encoded components of time and season are used to construct the feature vector, and the environmental-related dimensions are left blank or use default values.

[0078] Step 3: Query the database for user environment relevance to determine the matching path.

[0079] Before performing preference matching, the system first queries the database for historical correlation analysis results of the user regarding the device type, and then decides whether to follow either the "environmental clustering" or "time-period statistical analysis" path. The query may yield three results: Result 1: Items marked as "environmentally relevant" in the database (strong or moderate correlation, with an absolute correlation coefficient greater than 0.3) indicate that the user's operation of this type of device is primarily driven by environmental conditions. The system then enters the environmental clustering matching path (steps 4A to 6A), constructing a current environmental feature vector and calculating Euclidean distance between it and the cluster centers of historical preference records for similarity matching.

[0080] Result 2: Items marked as "unrelated to the environment" in the database (weakly correlated or uncorrelated, with an absolute correlation coefficient less than 0.3 or insufficient sample size) indicate that the user's operation of this type of device is not significantly related to environmental conditions, but rather driven by regular work and rest patterns and fixed habits. The system then enters the time-period statistical matching path (steps 4B to 6B), determines the time period based on the current time, and directly searches for the user's preferred operating behaviors corresponding to that time period.

[0081] Result 3: If there are no relevant records in the database (either due to the first query or insufficient data accumulation), the system will execute two paths simultaneously by default, obtaining the environment matching score and the time period matching score respectively, and taking the higher score as the recommendation result. The selection of this match will also be recorded as reference data for subsequent relevance assessments.

[0082] A. Environment clustering matching path: Step 4A: Query historical environment preference records.

[0083] Retrieve all valid preference records corresponding to the device serial number and room number from the device environment preference table in the algorithm library. Valid records are those marked as "yes" and whose environment condition field is marked as "environment-driven" (i.e., preference records generated by environment clustering, excluding records marked as "time period-driven").

[0084] Step 5A: Calculate the environmental similarity for each item.

[0085] For each historical environment preference record, calculate the Euclidean distance between the cluster center vector stored in its environment condition field and the current environment vector. The smaller the distance, the more similar the current environment is to the environment of that historical preference.

[0086] Step 6A: Screening and scoring.

[0087] Select preference records with a distance less than the threshold (default 0.5), and calculate a comprehensive score for each selected record: Similarity score = (1 - Euclidean distance) × 0.7; Confidence score = Confidence value of the preference record × 0.3; Comprehensive score = Similarity score + Confidence score; Sort by comprehensive score from high to low.

[0088] Statistical matching path for time period B: Step 4B: Determine the current time period.

[0089] The system determines the time period based on the hour of the current time. The time period division is consistent with the offline analysis phase: early morning (6:00-9:00), morning (9:00-12:00), noon (12:00-14:00), afternoon (14:00-17:00), evening (17:00-20:00), and night (20:00-6:00 the next day).

[0090] Step 5B: Query the historical time period preference records for the matching time period.

[0091] The algorithm library queries the device environment preference table for the valid preference records corresponding to the device serial number and room number, and filters out preference records whose scene tags match the current time period and whose environment condition field is marked as "time period driven". These records are time-segmented user preference behavior profiles generated through time period statistical analysis during the offline analysis phase.

[0092] Step 6B: Time Period Matching and Scoring.

[0093] For time period preference records, since there is no concept of environmental similarity, the scoring method is adjusted as follows: Time period matching score = 1.0 (perfect match, the same time period is the full score); Confidence score = the confidence value of the time period preference record × 1.0; Overall score = Time period matching score + Confidence score; If there are multiple preference records belonging to the same time period, they are sorted from high to low confidence. The higher the confidence, the richer the preference sample and the more reliable the statistics for that time period.

[0094] Step 7: Assemble and return the results.

[0095] Regardless of which matching path is used, the final result is assembled according to the following rules: For full-room queries: group by device number, select the attribute with the highest score from each group, and assemble them into a preference list.

[0096] For single-device queries: retrieve the top 5 best matching records, each containing the attribute name and attribute statistics (weighted mean, mode, number of samples, etc.).

[0097] When executing two paths simultaneously (without relevant records): take the best results of the environment matching path and the time period matching path respectively, compare the combined scores of the two, return the one with the higher score, and attach a matching path description ("environment-driven matching" or "time period-driven matching") in the response.

[0098] If no matching record is found in either path, a "No matching preference found" message is returned, and the preference list is empty.

[0099] The user device operation preference data query method provided in this application, by constructing a dual-path preference analysis model combining a weighted local density clustering algorithm and time-period statistical analysis, can accurately identify the correlation strength between user operation behavior and environmental data, and dynamically distinguish between environment-driven and time-period-driven operation modes. Based on an incremental clustering mechanism, the system continuously learns the evolution of user preferences, enabling proactive prediction and personalized recommendations of device operations, significantly improving the automation level and user experience of smart homes, truly achieving "understanding user needs and anticipating their needs."

[0100] Based on the above method embodiments, this application also provides a user equipment operation preference data query device, see [link to relevant documentation]. Figure 4 As shown, the device includes: a request acquisition module 402, used to acquire a user device operation preference data query request; the request carries the target device type, current time, and may also include environmental data; a feature determination module 404, used to generate a multi-dimensional feature vector based on the current time and environmental data; and a path determination module 406, used to query the correlation results between user device operations and the environment of the target device type from the database to determine the target matching path; wherein, the database also stores preference records corresponding to environmental clustering matching paths and time period statistical matching paths, respectively, and the acquisition method is as follows: acquire the user's historical device operation data; perform behavior operation chain aggregation and device linkage control detection on the historical device operation data to determine the aggregated operation chain data; for each operation chain data, perform user... The system performs a correlation analysis between user operation and environment to determine the correlation between equipment operation and environment. If the correlation is strong, it performs density-based incremental behavioral clustering analysis on the operation chain data to generate equipment environment preference records under the environment clustering matching path. If the correlation is weak or non-correlated, it performs time-period statistical analysis on the operation chain data to generate equipment environment preference records under the time-period statistical matching path. The system stores the correlation results, the equipment environment preference records under the environment clustering matching path, and the equipment environment preference records under the time-period statistical matching path. The score calculation module 408 is used to calculate and sort the matching scores of multiple preference records under the target matching path with the multi-dimensional feature vector. The preference recommendation module 410 is used to recommend preferences based on the sorted preference records.

[0101] Furthermore, the aforementioned feature determination module 404 is used to extract the hour and month from the current time, encode the hour into sine and cosine components, and encode the month into sine and cosine components to obtain a four-dimensional feature vector; the environmental data includes at least one of the following: indoor temperature, outdoor temperature, humidity, and light intensity; the indoor temperature step is quantized and normalized, the outdoor temperature step is quantized and normalized, the humidity is quantized and normalized according to the segment boundary, and the light intensity is quantized and normalized according to the segment boundary, and combined with the four-dimensional feature vector, a multi-dimensional feature vector is obtained.

[0102] Furthermore, the aforementioned path determination module 406 is used to determine the environmental clustering matching path as the target matching path if the correlation result is environmentally relevant; to determine the time period statistical matching path as the target matching path if the correlation result is environmentally unrelated; and to determine both the environmental clustering matching path and the time period statistical matching path as the target matching path if the correlation result is that there are no correlation records.

[0103] Furthermore, the aforementioned score calculation module 408 is used to determine the device serial number and room number corresponding to the target device; query all valid preference records corresponding to the device serial number and room number from the device environment preference records; when the target matching path is an environment clustering matching path, for each valid preference record, calculate the Euclidean distance based on the cluster center vector and multidimensional feature vector corresponding to the valid preference record; filter out valid preference records with a distance less than a threshold; for each filtered valid preference record, determine the comprehensive score based on the sum of the similarity score and the confidence score; when the target matching path is a time period statistical matching path, determine the time period based on the number of hours of the current time; filter out target preference records whose scene tags match the time period from all valid preference records; determine the comprehensive score of the target preference record based on the sum of the time period matching score and the confidence score.

[0104] Furthermore, the aforementioned device also includes: a preference record calibration module, used to group historical device operation data by device serial number and room number, and sort them by operation trigger time within each group; for the same device, if the time interval between two operation records is within a preset time interval, the attributes of the latter operation record are merged into the former operation record; using a specified duration as a time window, multiple operation records within the time window are merged into one operation chain data, and different operation weights are assigned according to the control source of each operation record; for each operation chain data, the last valid environmental data record within the time window is taken as an environmental snapshot of the operation chain; or it may also include: inputting the aggregated operation chain data into a continuous operation analyzer, the analyzer checking whether the operation chain data contains continuous control of multiple different devices, and if so, recording device linkage control information for the operation chain data.

[0105] Furthermore, the aforementioned preference record labeling module is used to construct a multi-dimensional environmental feature vector for each operation record in the operation chain data; for each operation record, a weighted local density is calculated based on the multi-dimensional environmental feature vector corresponding to the operation record; a dynamic minimum density threshold is calculated, and if the weighted local density of the operation record is less than the dynamic minimum density threshold, the operation record is determined as a core point; all core points are traversed, and each core point and all points in its neighborhood are formed into a cluster, and a cluster center vector and confidence score are calculated for each cluster; each cluster is used as a valid preference record to generate device environment preference records under the environment clustering matching path; each valid preference record corresponds to a cluster center vector; the method also includes: when new data matches a historical cluster, the cluster center, total weight, and number of samples are incrementally updated.

[0106] Furthermore, the aforementioned preference record calibration module is used to determine the feature-weighted Euclidean distance between the sample point and other operation records, based on the feature values ​​of each feature dimension, the weight coefficient of each feature dimension, and the standard deviation of the sample point and any other operation record, using operation records as sample points. Operation records with feature-weighted Euclidean distances less than a threshold are considered as samples within the neighborhood of the sample point. The time weight of each sample is calculated and summed to obtain the weighted local density of the sample point. The time weight is determined based on the difference in days between the current analysis time and the operation trigger time, and the half-life of the device type.

[0107] Furthermore, the aforementioned preference recording and labeling module is used to divide a day into multiple time periods; for each time period, the raw data of historical device operation behavior is grouped and statistically analyzed according to three dimensions: device type, device number, and attribute name; for numerical attributes, the following indicators are statistically analyzed: weighted mean, weighted median, mode, and number of samples; for enumerated attributes, the following indicators are statistically analyzed: most recent mode, frequency distribution, and number of samples; based on the statistical indicators for each time period, a user preference device operation behavior profile is generated for each time period to describe the user's most typical device usage pattern within that time period; based on the user preference device operation behavior profiles corresponding to multiple time periods, a device environment preference record under the time period statistical matching path is formed.

[0108] The device provided in this application embodiment has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts of the device embodiment not mentioned can be referred to the corresponding content in the aforementioned method embodiment.

[0109] This application also provides an electronic device, such as... Figure 5The diagram shows the structure of the electronic device, which includes a processor 51 and a memory 50. The memory 50 stores computer-executable instructions that can be executed by the processor 51, and the processor 51 executes the computer-executable instructions to implement the above-described method.

[0110] exist Figure 5 In the illustrated embodiment, the electronic device further includes a bus 52 and a communication interface 53, wherein the processor 51, the communication interface 53, and the memory 50 are connected via the bus 52.

[0111] The memory 50 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 53 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 52 may be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 52 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0112] The processor 51 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 51 or by instructions in software form. The processor 51 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in the memory, and the processor 51 reads the information in the memory and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiment.

[0113] This application also provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions cause the processor to implement the above-described method. For specific implementation details, please refer to the foregoing method embodiments, which will not be repeated here.

[0114] The computer program products of the methods, apparatus, and electronic devices provided in the embodiments of this application include a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementations, please refer to the method embodiments, which will not be repeated here.

[0115] Unless otherwise specifically stated, the relative steps, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0116] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0117] In the description of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0118] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the technical scope disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for querying user equipment operation preference data, characterized in that, The method includes: A request to obtain user device operation preference data is submitted; the request may include the target device type, current time, or environmental data. Based on the current time and the environmental data, a multi-dimensional feature vector is generated; The database is queried to determine the correlation between user equipment operation and environment for the target device type, and the target matching path is determined. The database also stores preference records corresponding to environmental clustering matching paths and time-period statistical matching paths, obtained as follows: Historical device operation data of the user is acquired; behavioral operation chain aggregation and device linkage control detection are performed on the historical device operation data to determine the aggregated operation chain data; for each operation chain data, user operation and environment correlation analysis is performed to determine the correlation result between device operation and environment; if the correlation result is strong, density-based incremental behavioral clustering analysis is performed on the operation chain data to generate device environment preference records under the environmental clustering matching path; if the correlation result is weak or non-correlated, time-period statistical analysis is performed on the operation chain data to generate device environment preference records under the time-period statistical matching path; the correlation result, device environment preference records under the environmental clustering matching path, and device environment preference records under the time-period statistical matching path are stored. Calculate and sort the matching scores of multiple preference records under the target matching path with the multidimensional feature vector; Preference recommendations are made based on the sorted preference records.

2. The method according to claim 1, characterized in that, The step of generating a multidimensional feature vector based on the current time and the environmental data includes: Extract the hour and month from the current time, encode the hour into sine and cosine components, and encode the month into sine and cosine components to obtain a four-dimensional feature vector; The environmental data includes at least one of the following: indoor temperature, outdoor temperature, humidity, and light intensity; the indoor temperature stepwise quantization is normalized, the outdoor temperature stepwise quantization is normalized, the humidity is normalized according to segment boundaries, and the light intensity is normalized according to segment boundaries. Combined with the four-dimensional feature vector, a multi-dimensional feature vector is obtained.

3. The method according to claim 1, characterized in that, The step of querying the database for the correlation results between user equipment operation and environment of the target device type to determine the target matching path includes: If the correlation result is environmentally relevant, the environmental clustering matching path will be determined as the target matching path; If the correlation result indicates that the environment is not relevant, the statistical matching path for the specified time period will be determined as the target matching path. If the correlation result indicates that there is no correlation record, the environmental clustering matching path and the time period statistical matching path are both determined as the target matching path.

4. The method according to claim 1, characterized in that, The step of calculating the matching scores of multiple preference records under the target matching path with the multidimensional feature vector includes: Determine the device serial number and room number corresponding to the target device; Retrieve all valid preference records corresponding to the device serial number and the room number from the device environment preference record; When the target matching path is the environment clustering matching path, for each valid preference record, the Euclidean distance is calculated based on the cluster center vector corresponding to the valid preference record and the multidimensional feature vector; valid preference records with a distance less than a threshold are filtered out; for each filtered valid preference record, a comprehensive score is determined based on the sum of the similarity score and the confidence score. When the target matching path is a time period statistical matching path, the time period is determined based on the number of hours of the current time; target preference records whose scene tags match the time period are selected from all valid preference records; and the comprehensive score of the target preference record is determined based on the sum of the time period matching score and the confidence score.

5. The method according to claim 1, characterized in that, The steps of aggregating the historical equipment operation data into behavioral operation chains and detecting equipment linkage control to determine the aggregated operation chain data include: The historical device operation data is grouped by device serial number and room number, and then sorted by operation trigger time within each group; For the same device, if the time interval between two operation records is within a preset time interval, the attributes of the later operation record will be merged into the earlier operation record. Using a specified duration as a time window, multiple operation records within the time window are merged into one operation chain data, and different operation weights are assigned according to the control source of each operation record; For each operation chain data, the last valid environmental data record within the time window is taken as the environmental snapshot of the operation chain. Alternatively, it may include: inputting the aggregated operation chain data into a continuous operation analyzer, which checks whether the operation chain data contains continuous control of multiple different devices; if so, recording the device linkage control information in the operation chain data.

6. The method according to claim 1, characterized in that, The steps of performing density-based incremental behavioral clustering analysis on the operation chain data to generate device environment preference records under the environment clustering matching path include: For each operation record in the operation chain data, construct a multi-dimensional environmental feature vector; For each operation record, a weighted local density is calculated based on the multidimensional environmental feature vector corresponding to the operation record; Calculate the dynamic minimum density threshold. If the weighted local density of the operation record is less than the dynamic minimum density threshold, determine the operation record as a core point. Iterate through all core points, form a cluster for each core point and all points in its neighborhood, and calculate the cluster center vector and confidence for each cluster; Each cluster is treated as a valid preference record, generating device environment preference records under the environment clustering matching path; each valid preference record corresponds to a cluster center vector. The method further includes: when new data matches a historical cluster, incrementally updating the cluster center, total weight, and number of samples.

7. The method according to claim 6, characterized in that, The step of calculating the weighted local density based on the multidimensional environmental feature vector corresponding to the operation record includes: Using the operation record as a sample point, the feature-weighted Euclidean distance between the sample point and other operation records is determined based on the feature value of each feature in each dimension, the weight coefficient of each feature, and the standard deviation of the feature in each dimension. Operation records whose feature-weighted Euclidean distance is less than a threshold are considered as samples within the neighborhood of the sample point. Calculate the time weight of each sample and sum them to obtain the weighted local density of the sample points; the time weight is determined based on the difference in days between the current analysis time and the operation trigger time and the half-life of the device type.

8. The method according to claim 1, characterized in that, The steps of performing time-period statistical analysis on the operation chain data and generating device environment preference records under the time-period statistical matching path include: Divide the day into multiple time periods; For each time period, the raw data of historical device operation behavior is grouped and statistically analyzed according to three dimensions: device type, device number, and attribute name. For numerical attributes, the following indicators are calculated: weighted mean, weighted median, mode, and number of samples. For enumerated attributes, the following indicators are calculated: most recent mode, frequency distribution, and number of samples. Based on the statistical indicators for each time period, a user preference device operation behavior profile is generated for each time period to describe the user's most typical device usage pattern during that time period. Based on user preference device operation behavior profiles corresponding to multiple time periods, device environment preference records are formed under the time period statistical matching path.

9. A user equipment operation preference data query device, characterized in that, The device includes: The request acquisition module is used to acquire user device operation preference data query requests; the request carries the target device type, current time, and may also include environmental data; The feature determination module is used to generate a multi-dimensional feature vector based on the current time and the environmental data; The path determination module is used to query the database for the correlation results between user device operations and the environment of the target device type, and determine the target matching path. The database also stores preference records corresponding to environment clustering matching paths and time-period statistical matching paths, obtained as follows: Historical device operation data of the user is obtained; behavioral operation chain aggregation and device linkage control detection are performed on the historical device operation data to determine the aggregated operation chain data; for each operation chain data, user operation and environment correlation analysis is performed to determine the correlation result between device operation and environment; if the correlation result is strong, density-based incremental behavioral clustering analysis is performed on the operation chain data to generate device environment preference records under the environment clustering matching path; if the correlation result is weak or non-correlated, time-period statistical analysis is performed on the operation chain data to generate device environment preference records under the time-period statistical matching path; the correlation result, device environment preference records under the environment clustering matching path, and device environment preference records under the time-period statistical matching path are stored. The score calculation module is used to calculate and sort the matching scores of multiple preference records under the target matching path with the multidimensional feature vector. The preference recommendation module is used to make preference recommendations based on the sorted preference records.

10. An electronic device, characterized in that, The method includes a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 8.