Smart home equipment control method and system based on smart door lock

Abstract

The invention belongs to the technical field of smart home, and particularly relates to a smart home equipment control method and system based on a smart door lock, and the method comprises the steps: obtaining a plurality of historical data slices containing a leading event sequence; constructing a user behavior map according to the historical data, and calculating the mode goodness of fit of the current behavior; meanwhile, learning user environment preferences according to historical data, and constructing a probability density model to calculate the unbalance degree of the current environment; and finally, fusing the behavior pattern goodness of fit and the environmental unbalance degree into a decision factor through an adaptive nonlinear function, and using the decision factor to perform dynamic matching in a historical successful scene library and drive final equipment control. According to the invention, adaptive, personalized and continuously optimized smart home control is realized, and the user experience can be improved.

Description

Technical Field

[0001] This invention relates to the field of smart home technology. More specifically, this invention relates to a smart home device control method and system based on a smart door lock. Background Technology

[0002] In modern smart home systems, smart door locks are not only the first line of defense for home security, but also a key node for users to interact with their home environment. By unlocking and locking smart door locks, corresponding scene linkages are triggered. For example, the "coming home" mode automatically turns on the lights, air conditioner, and curtains, which has become an important way to improve the living experience.

[0003] To achieve this kind of scene linkage, most existing technologies adopt control strategies based on fixed rules, such as pre-setting instructions to execute a certain set of devices if the door is unlocked within a certain time period. The advantage of this method is its simplicity and intuitiveness, but its core drawback is its lack of personalization. Therefore, existing technologies also introduce machine learning algorithms to analyze users' historical behavior and assist in personalized automatic control of home devices. For example, Chinese patent document CN114237072B discloses an integrated smart home door lock control system and method using cloud control technology. It discloses that by monitoring and analyzing the indoor environment, it predicts the indoor environment conditions, and based on the prediction results, it pre-determines the control status of the door lock and executes the corresponding judgment.

[0004] However, existing technologies have the following limitations in achieving the above control: the rules are static and cannot adapt to users' changing habits or atypical behaviors, resulting in low accuracy of scenario execution. Furthermore, as the number of devices increases, the configuration and maintenance of the rules become extremely complicated. In addition, they cannot deeply explore the complex and non-linear behavioral patterns of users in different time and environmental contexts, and they do not have the ability to self-optimize based on subsequent user feedback. Summary of the Invention

[0005] To address the aforementioned technical problems of insufficient personalization and poor control performance in smart home device control, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for controlling smart home devices based on a smart door lock, comprising: The system acquires multiple historical data slices containing preceding event sequences; constructs a user behavior graph based on the event types in the historical data slices, the user behavior graph containing multiple event nodes and directed edges connecting the event nodes; acquires the current event sequence of the current data slice, calculates the behavior pattern fit of the current event sequence, the behavior pattern fit is positively correlated with the average edge weight of the corresponding path in the user behavior graph and the average centrality of the corresponding node; constructs a preference probability density model for each internal environment dimension based on the internal environment vector in the historical data slices and user manual adjustment records; acquires the current internal environment vector of the current data slice, calculates the current environment imbalance, the imbalance is positively correlated with the weighted sum of the negative log-likelihoods of each dimension of the current internal environment vector under the corresponding preference probability density model, the weight of the weighted sum being the user's historical adjustment frequency for each internal environment dimension; multiplies the behavior pattern fit by the environment imbalance to obtain a decision factor, and selects and executes control commands from a historical successful scenario library based on the decision factor.

[0007] This invention constructs a user behavior graph to understand users' dynamically changing behavioral habits, and simultaneously constructs a preference probability density model to learn users' personalized environmental comfort. The decision factor is obtained by multiplying the degree of consistency of behavioral patterns representing intentions with the degree of environmental imbalance representing needs. This ensures that a scenario is triggered only when user behavior conforms to habits and the environment actually needs adjustment. This dual-condition judgment mechanism greatly improves the accuracy of scenario triggering, effectively avoids misoperation caused by atypical behaviors, and achieves true adaptive and personalized intelligent control.

[0008] Preferably, obtaining the weights of the directed edges includes: Get connection event nodes and directed edges The interaction type in the historical data slice is obtained by arrive The set of event transitions, denoted as ; Timestamp of the current data slice and The Middle The difference in timestamps corresponding to the historical data slices of the event transformation, multiplied by the time decay constant, yields... The Middle The freshness of the event transition occurring for the first time; the freshness of all historical occurrences of the event transition is summed to obtain the freshness from the event node. arrive The weight of the directed edges.

[0009] This invention introduces a freshness concept based on timestamp differences and time decay constants, and accumulates these values ​​to give higher weight to recent and more frequent behavioral transitions in the graph, dynamically reflecting the evolution of user habits. As a result, the system can more accurately identify the user's current core behavioral patterns, rather than being misled by outdated habits, thereby improving the accuracy of behavioral pattern matching.

[0010] Preferably, the time decay constant is positively correlated with the reciprocal of the average interval duration of adjacent timestamps in all historical data slices.

[0011] Preferably, the centrality of the node is calculated using a graph centrality algorithm, specifically the PageRank algorithm.

[0012] Preferably, the preference probability density model is a KDE model based on kernel density estimation.

[0013] Preferably, the decision factor is obtained by multiplying the behavioral pattern fit degree by the environmental imbalance degree, including: performing a first nonlinear mapping on the behavioral pattern fit degree to obtain behavioral intention potential energy; performing a second nonlinear mapping on the environmental imbalance degree to obtain environmental demand potential energy; and multiplying the behavioral intention potential energy by the environmental demand potential energy to obtain the decision factor.

[0014] This invention first converts behavioral fit and environmental imbalance into behavioral intention potential energy and environmental demand potential energy respectively through nonlinear mapping, and then multiplies them to achieve AND logic. Only when the potential energy of both behavioral intention and environmental demand is high will the final decision factor increase significantly. This ensures the prudence of the system decision and avoids triggering scene linkage when there is only a weak signal in any dimension, making the decision-making process more robust.

[0015] Preferably, the shape of the function curve of the first nonlinear mapping is determined by the mean and variance of the fit of all historical behavior patterns; the shape of the function curve of the second nonlinear mapping is determined by the mean and variance of the imbalance of all historical environments.

[0016] This invention optimizes the nonlinear mapping function so that its shape is dynamically determined by the historical mean and variance of the corresponding input indicators, namely behavioral fit or environmental imbalance. This makes the judgment threshold no longer fixed, but dynamically adjusted according to the unique data distribution of each user, thereby improving the system's sensitivity and adaptability to changes in the state of individual users.

[0017] Preferably, the internal environment vector includes at least two of the following: indoor temperature; indoor air quality index; and the presence of a human being as sensed by millimeter-wave radar.

[0018] Preferably, it further includes: after executing the control command, monitoring whether the user has any manual intervention behavior within a preset time window; if there is manual intervention behavior, storing the result of the manual intervention behavior as a new successful scenario in the historical successful scenario library, and triggering an online update of the user behavior map and preference probability density model.

[0019] This invention treats the user's manual intervention after the system automatically executes instructions as a feedback signal. This intervention is not only stored as a new success scenario, but also triggers online updates to the behavior graph and preference model, enabling the system to continuously learn and self-correct. It can learn from mistakes and its decisions will become more and more in line with the user's real needs as the usage time increases.

[0020] Secondly, the present invention provides a smart home device control system based on a smart door lock, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned smart home device control method based on a smart door lock is implemented.

[0021] By adopting the above technical solution, a computer program is generated from the above-mentioned smart home device control method based on smart door lock and stored in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0022] The beneficial effects of this invention are as follows: (1) By constructing a preference probability density model based on user history manual adjustment for each environmental dimension, this invention can accurately capture each user's unique comfort range for environmental factors such as temperature and brightness, and quantify the user's sensitivity to different dimensions, so that device control is no longer executing general instructions, but aims to meet the personalized comfort needs of specific users, which significantly improves the living experience. (2) The user behavior graph constructed by this invention can dynamically reflect the evolution of user habits by introducing a weight mechanism based on time decay, and give higher weight to recent behaviors. Combined with the graph centrality algorithm to evaluate the importance of events, the behavior pattern conformity can accurately determine whether the current behavior conforms to the user's recent core life routine, thereby effectively distinguishing different intentions such as regular homecoming and temporary outings, and avoiding accidental triggering. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a smart home device control method based on a smart door lock according to the present invention; Figure 2 This is a schematic diagram illustrating the probability density distribution model of air quality index preferences; Figure 3This is a schematic diagram illustrating the probability density distribution model of temperature preference. Detailed Implementation

[0024] This invention discloses a method for controlling smart home devices based on smart door locks, referring to... Figure 1 This includes steps S1-S4: S1: Obtain multiple historical data slices of the smart door lock, wherein the data slices include interaction data, internal environment vector, external environment vector, and user status data.

[0025] It's important to note that to optimize the control effect of smart home devices controlled by smart locks, it's crucial to first fully acquire user behavior data, including lock interaction data and related data on home device control. User actions, such as returning home or leaving home, are not isolated, instantaneous movements but rather processes comprised of a series of sequentially related events. For instance, returning home might begin with the phone entering the home's geographic area before the lock is opened. Therefore, simply capturing a snapshot of the lock interaction is insufficient; it's essential to simultaneously capture the sequence of preceding events that led to the interaction. This invention therefore provides unified management of all relevant events recorded by the system and uses smart lock interaction events as anchor points to construct richer historical data slices.

[0026] Specifically, an event log is established in the door lock system to continuously record all events generated from all sources since the smart door lock began to be used. These sources include the smart door lock, various sensors, and user personal devices. A timestamp is assigned to each event to form an event log sorted by time.

[0027] The system acquires interaction data for door lock interaction events. The interaction data includes a timestamp, user identification, interaction type, and authentication method. The interaction type includes at least unlocking, locking, and deadbolting. The authentication method includes at least fingerprint, password, and NFC.

[0028] Obtain the internal environment vector corresponding to the timestamp of the door lock interaction event. The data source of the internal environment is the smart home device that transmits data with the smart door lock. The internal environment includes, for example, temperature, air quality index, and the human presence status sensed by millimeter-wave radar. The human presence status uses 0 to represent no one, 1 to represent stillness, and 2 to represent activity. All internal environments constitute the internal environment vector.

[0029] Obtain the external environment vector corresponding to the timestamp of the door lock interaction event. The data source of the external environment is third-party meteorological and environmental service data obtained in real time through network API based on home geographical location information. The external environment includes, for example, weather type, outdoor temperature, ultraviolet index, and outdoor air quality index.

[0030] The internal environment vector and the external environment vector together constitute a snapshot of the scene corresponding to the door lock interaction event.

[0031] The user status data is obtained from information obtained from the user's smart device or cloud service after authorization by the user. It includes the mobile phone geofence status and event tags that overlap with the timestamps of the interaction data. The mobile phone geofence status is, for example, entering the home area or being in the home area. The event tags are, for example, working or on vacation.

[0032] Using the timestamp of each door lock interaction data as the center moment, events within a preset time window to the left of the center moment are acquired and arranged in chronological order to form a sequence of preceding events for an interaction event. For example, the preset time window is 10 minutes.

[0033] The timestamps of historical data slices, along with the corresponding door lock interaction data, context snapshots, and preceding event sequences, together constitute a complete historical data slice.

[0034] At this point, multiple historical data slices of the smart lock have been obtained. These historical data slices include timestamps, lock interaction data, context snapshots, and preceding event sequences. The set of historical data slices is denoted as... ,in For the current data slice, This is the Nth historical data slice.

[0035] S2: Construct a user behavior graph based on the interaction data in the historical data slices; obtain the degree of consistency of the behavior pattern based on the graph and the current data slice.

[0036] It's important to note that a user's home life is not a random combination of isolated events, but rather follows internal logic and patterns. For example, when a user returns home from get off work, a series of sequential events occur, such as the phone's geofence status changing to indicate entry into the home area and the interaction type being unlocking the door. These event sequences collectively constitute the user's behavioral pattern. Therefore, this invention constructs a weighted directed graph from all door lock interaction data and preceding event sequences in historical data slices, expressing these implicit behavioral patterns in a structured form. User behavior is dynamic, and therefore, behavioral patterns are also dynamically changing. Recent behavioral patterns are more representative of a user's current habits than long-term patterns, and the higher the frequency of a behavioral sequence, the stronger the habit. Therefore, when constructing the user behavior graph, this invention introduces a time decay mechanism to the weights of the edges connecting events, ensuring that the edge weights simultaneously reflect the frequency and relevance of the sequence.

[0037] It's important to further clarify that assessing whether current behavior aligns with habits requires consideration from two dimensions. First, does the current behavioral sequence match high-weighted paths in the user behavior graph, i.e., strong habitual paths? Second, are the events themselves that constitute the behavioral sequence core and pivotal in the user's daily activities, reflecting the importance of the behavioral nodes? Solid habits are typically manifested both as fixed behavioral paths and as paths composed of high-frequency, core event nodes. Therefore, this invention establishes a behavioral pattern conformity score, comprehensively considering path strength and node importance to quantitatively assess the compliance of current behavior.

[0038] Specifically, using the event types contained in all door lock interaction data and preceding event sequences in the historical data slices as nodes, and the chronological order of event occurrence as directed edges, a user behavior graph is constructed: an empty graph is initialized. Iterate through each historical data slice in the historical data slice set, and assemble the preceding event sequence and events in the door lock interaction data into a complete event chain in chronological order; for each event type in the event chain, check whether it already exists as a node. If it does not exist, then in Create the node in the event chain; for each pair of adjacent events in the event chain, in Create a directed edge between the two corresponding nodes. If the edge already exists, increment its frequency count; this yields the user behavior graph. .

[0039] Preferably, the weights of the edges connecting two event nodes in the user behavior graph are obtained based on the frequency and relevance of events in historical data slices, including: Get connection event nodes and directed edges The interaction type in the historical data slice is obtained by arrive The set of event transitions, denoted as .

[0040] The time decay constant is denoted as the reciprocal of the average interval length of all adjacent timestamps in the historical data slice set. It should be noted that this time decay constant is dynamically determined by the statistical characteristics of the timestamp interval sequence of all the user's historical data slices and can be used as a coefficient to adjust the differences in edge weights.

[0041] The weights satisfy the expression: ; In the formula, Indicates from the event node arrive The weight of the directed edge; express The Middle The timestamp corresponding to the historical data slice that the event transformation occurs for the first time; Indicates the timestamp of the current data slice; This represents the time decay constant.

[0042] In the formula, express The Middle The freshness of the event transition occurring for the first time, this value varies with time difference. The value increases and decreases exponentially, indicating that events from a more distant time period contribute less to current habits; This represents the sum of the freshness of all historical occurrences of the event described; the higher the value, the fresher the event. arrive The more frequent and closer the transition is to the current state, the higher the corresponding habit strength and edge weight. The larger.

[0043] Preferably, the behavioral pattern fit is obtained based on the path strength and node importance of the current data slice and neighboring historical data slices in the user behavior graph, including: Construct the current event sequence from the preceding event sequence and door lock interaction data of the current data slice in chronological order. .

[0044] It should be noted that not all events are equally important in a user's behavior patterns. The unlocking event, as a hub connecting the state inside and outside the home, is more important than events such as turning on the lights. Graph centrality algorithms, such as PageRank, can calculate the influence of a node in the network. A node with a high centrality score means that many important behavioral paths point to it or originate from it.

[0045] The graph centrality algorithm is used to calculate the average importance of all nodes in the current event sequence within the user behavior graph: the graph centrality algorithm is run on all nodes in the user behavior graph to calculate a centrality score for each node; The arithmetic mean of the centrality scores of each node is calculated to obtain the average importance of all nodes in the current event sequence in the user behavior graph. The graph centrality algorithm is an existing technology, such as PageRank, and will not be elaborated here.

[0046] It should be noted that different users exhibit different behavioral structures. Some users' behaviors are highly modularized; for example, cooking-related events form a close-knit community with little overlap with audio-visual entertainment communities. Other users' behaviors tend to be more linear and procedural long chains. The average clustering coefficient is an indicator that measures the degree of community or modularity of the entire graph. By analyzing the overall structure of user behavior, the evaluation strategy can be dynamically adjusted to determine whether to emphasize the fixedness of the path or the criticality of the nodes.

[0047] Calculate the local clustering coefficient of each node in the user behavior graph, then take the arithmetic mean of the local clustering coefficients of all nodes to obtain the average clustering coefficient of the user behavior graph, which is denoted as the balance factor.

[0048] The behavioral pattern fit satisfies the expression: ; In the formula, Indicates the degree of consistency in behavioral patterns; This represents the average edge weight of the corresponding path in the user behavior graph for the current event sequence; This represents the average importance of all nodes in the current event sequence within the user behavior graph; Indicates the balance factor; This represents the normalization function.

[0049] In the formula, This represents the path strength of the corresponding path in the user behavior graph for the current event sequence. This indicates the overall importance of nodes in the current event sequence. This indicates a weighted fusion of the path strength and node importance. The larger the value, the more the current event sequence conforms to a strong habitual path composed of important events.

[0050] At this point, the behavioral pattern consistency was obtained.

[0051] S3: Obtain environmental imbalance based on contextual snapshots of historical data slices and user-manually adjusted records.

[0052] It's important to note that smart home device linkage shouldn't be solely triggered by user actions; the actual needs of the environment must also be considered. Device control is only meaningful and beneficial for improving the user experience when the environment deviates from the user's comfort zone—that is, when the environment is in an unbalanced state. Furthermore, the user's comfort zone is not a fixed value. For example, a user might feel most comfortable at 23°C, but also tolerate 22°C or 24°C, while experiencing significant discomfort at 18°C ​​or 28°C. This preference characteristic can be precisely described by a probability distribution, with the center being the user's optimal preference point, gradually decreasing towards both sides. Therefore, this invention calculates the degree of imbalance in the environmental state based on its probability within the user's historical preference probability distribution. The lower the probability, the further the environmental state deviates from the comfort zone, and the higher the degree of imbalance.

[0053] It should be further explained that users exhibit individual differences in their sensitivity to different environmental dimensions. For example, some people are highly sensitive to temperature changes and require frequent adjustments to the air conditioner, while others are more sensitive to changes in light and need to adjust the lighting according to different indoor brightness levels. This difference in sensitivity is directly reflected in the frequency of manual intervention with various devices in the user's history. Therefore, when calculating the overall environmental imbalance, the deviation of different internal environmental dimensions should be weighted, with the weight proportional to the user's historical sensitivity to that internal environmental dimension. Thus, this invention establishes an environmental imbalance score, which obtains the overall degree of improvement needed for the current environment by weighted summing of the deviations of each internal environmental dimension from its corresponding preference probability distribution.

[0054] Specifically, for each internal environment dimension, based on the internal environment vector in the historical data slices and the associated user manual adjustment records, a preference probability density distribution model is learned and constructed: the historical data slice set is filtered, and if no environment-related manual device adjustment event occurs within a preset time range after the timestamp of any historical data slice, the historical data slice is marked as a type of slice; for any internal environment dimension, the values ​​of the internal environment dimension in the context snapshots of all type-one slices are extracted to form the type-one dataset of the internal environment dimension.

[0055] The KDE method is used to process the dataset of the internal environment dimension to generate a preference probability density distribution model for the internal environment dimension. The KDE method is a kernel density estimation method, which is existing technology and will not be elaborated upon here. Figure 2 This is a schematic diagram of the probability density distribution model for air quality index preferences. Figure 3 This is a schematic diagram of the probability density distribution model for temperature preference.

[0056] Preferably, the environmental imbalance is obtained based on the deviation of the current value of each internal environmental dimension from its preference probability density distribution model and the user's sensitivity to each dimension, including: Get the first element of the internal environment vector in the current data slice. Current value of each dimension .

[0057] Adjusting the device in the user's history The ratio of the number of times the value was adjusted to the total number of adjustments across all devices is denoted as the user-dimensional sensitivity weight.

[0058] The degree of environmental imbalance satisfies the following expression: ; In the formula, Indicates the degree of environmental imbalance; Indicates the current value In the Probability density values ​​under the preference probability density distribution model of each internal environmental dimension; Indicates the sensitivity weight of the user dimension; Represents the normalization function; For example, the minimum value is... .

[0059] In the formula, Indicates the current value The negative log-likelihood is used as the deviation in that dimension. If If it is located in the center of the user's comfort zone, then A larger value indicates a smaller deviation; conversely, a smaller value indicates a smaller deviation. Moving away from your comfort zone, A smaller value indicates a larger deviation.

[0060] This indicates that the deviation from this dimension is weighted. This indicates that the weighted deviation of all environmental dimensions is summed to obtain the total degree of deviation. The larger the value, the further the overall environment deviates from the user's comfort zone, and the higher its degree of imbalance.

[0061] Thus, the degree of environmental imbalance was obtained.

[0062] S4: Based on the fit of behavioral patterns and the imbalance of the environment, generate scene activation potential and perform scene matching and decision execution.

[0063] It's important to note that ideally, smart home device control decisions should only be triggered when both conditions are largely met: the user's behavior aligns with their established habits, and the home environment genuinely requires adjustment. However, the historical distribution of user behavior habits and environmental imbalances varies from person to person. For example, users with more regular lifestyles tend to have higher historical values ​​for behavioral pattern consistency, meaning even small decreases should be considered abnormal. Conversely, users with more casual lifestyles may experience significant fluctuations in their behavioral pattern consistency. Therefore, a fixed mapping function cannot accommodate these individual differences.

[0064] It should be further explained that, to achieve more intelligent integration, this invention designs two adaptive nonlinear mapping functions. Before fusing behavioral pattern fit and environmental imbalance, they are processed separately through their respective nonlinear mapping functions. The shapes of these two nonlinear mapping functions are dynamically determined by the distribution of values ​​calculated from historical data slices of the corresponding input indicators. This allows the mapping process to automatically adapt to the personalized characteristics of each user. Therefore, this invention establishes a decision factor as the final decision-making basis for triggering scene linkage.

[0065] Specifically, decision factors are obtained based on the distribution of behavioral pattern fit and environmental imbalance across historical data slices, including: The decision factors satisfy the expression: ; In the formula, This indicates the potential energy for scene activation; Indicates the degree of consistency in behavioral patterns; Indicates the degree of environmental imbalance; , It is a nonlinear mapping function corresponding to the degree of fit between behavioral patterns and the degree of environmental imbalance. It is usually an sigmoid function, and the nonlinear mapping function is a Gaussian function.

[0066] In the formula, Potential energy representing the intention to act. This represents the potential energy of environmental demand. Multiplying the two achieves logical consistency; only when both the behavioral intention and the potential energy of the environmental demand are high will the final result be achieved. Only then will it be high, thus powerfully triggering subsequent scenes.

[0067] It should be noted that, with For example, the key parameters of its mapping function, such as the center point and shape parameters, are dynamically determined by the historical mean and historical variance of C calculated from all historical data slices. If the values ​​are generally high and concentrated, then... The curve will be quite steep, so even a small decrease in value will lead to a sharp drop in output potential energy.

[0068] Preferably, based on decision factors, scenario matching and decision execution are performed, including: Will As a dynamic search threshold, in the historical success scenario library In the process, retrieve all context snapshots whose similarity to the context snapshot of the current data slice is greater than 1. Based on historical success stories, a set of candidate scenarios is formed.

[0069] If the candidate scenario set is not empty, select the scenario that is most similar to the current situation and execute its corresponding device control instruction set. If the candidate scenario set is empty, do not perform any operation.

[0070] After executing the instruction, monitor whether the user manually intervenes within a preset time window. If intervention occurs, the user's action result is stored as a new successful scenario sample. This triggers online updates to the user behavior graph of S2 and the preference probability density model of S3.

[0071] This completes the control of smart home devices based on smart door locks.

[0072] This invention also discloses a smart home device control system based on a smart door lock, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a smart home device control method based on a smart door lock according to the present invention is implemented.

[0073] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0074] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A method for controlling smart home devices based on a smart door lock, characterized in that, include: Obtain multiple historical data slices containing a sequence of preceding events; A user behavior graph is constructed based on the event types in the historical data slices. The user behavior graph contains multiple event nodes and directed edges connecting the event nodes. Obtain the current event sequence of the current data slice, calculate the behavior pattern matching degree of the current event sequence, and the behavior pattern matching degree is positively correlated with the average edge weight of the corresponding path and the average centrality of the corresponding node in the user behavior graph of the current event sequence; Based on the internal environment vectors in the historical data slices and the user's manual adjustment records, a preference probability density model is constructed for each internal environment dimension; the current internal environment vector of the current data slice is obtained, and the imbalance of the current environment is calculated. The imbalance is positively correlated with the weighted sum of the negative log-likelihoods of the values ​​of each dimension of the current internal environment vector under the corresponding preference probability density model. The weight of the weighted sum is the frequency of the user's historical adjustment to each internal environment dimension. The decision factor is obtained by multiplying the behavioral pattern fit degree by the environmental imbalance degree, and control instructions are selected and executed from the historical successful scenario library based on the decision factor.

2. The smart home device control method based on a smart door lock according to claim 1, characterized in that, Obtaining the weights of the directed edges includes: Get connection event nodes and Directed edge The interaction type in the historical data slice is obtained by arrive The set of event transitions, denoted as ; Timestamp of the current data slice and The Middle The difference in timestamps corresponding to the historical data slices of the event transformation, multiplied by the time decay constant, yields... The Middle The freshness of the event transition occurring for the first time; the freshness of all historical occurrences of the event transition is summed to obtain the freshness from the event node. arrive The weight of the directed edges.

3. The smart home device control method based on a smart door lock according to claim 2, characterized in that, The time decay constant is positively correlated with the reciprocal of the average interval between adjacent timestamps in all historical data slices.

4. The smart home device control method based on a smart door lock according to claim 1, characterized in that, The centrality of the node is calculated using a graph centrality algorithm, specifically the PageRank algorithm.

5. A smart home device control method based on a smart door lock according to claim 1, characterized in that, The preference probability density model is a KDE model based on kernel density estimation.

6. A smart home device control method based on a smart door lock according to claim 1, characterized in that, The decision factor is obtained by multiplying the behavioral pattern fit degree by the environmental imbalance degree, including: performing a first nonlinear mapping on the behavioral pattern fit degree to obtain behavioral intention potential energy; performing a second nonlinear mapping on the environmental imbalance degree to obtain environmental demand potential energy; and multiplying the behavioral intention potential energy by the environmental demand potential energy to obtain the decision factor.

7. A smart home device control method based on a smart door lock according to claim 6, characterized in that, The shape of the function curve of the first nonlinear mapping is determined by the mean and variance of the fit of all historical behavior patterns; the shape of the function curve of the second nonlinear mapping is determined by the mean and variance of the imbalance of all historical environments.

8. A smart home device control method based on a smart door lock according to claim 1, characterized in that, The internal environment vector includes at least the following two items: indoor temperature; indoor air quality index; and the presence of the human body as sensed by millimeter-wave radar.

9. A smart home device control method based on a smart door lock according to claim 1, characterized in that, Also includes: After executing the control command, monitor whether the user engages in any manual intervention within a preset time window; If manual intervention occurs, the result of the manual intervention is stored as a new successful scenario in the historical successful scenario library, and an online update of the user behavior graph and preference probability density model is triggered.

10. A smart home device control system based on a smart door lock, characterized in that, include: A processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement a smart home device control method based on a smart door lock according to any one of claims 1-9.

Images