Multimodal access control method and system for IoT smart home appliances

Abstract

This application relates to a multimodal access control method and system for IoT smart home appliances, belonging to the field of IoT access control technology. The method includes: acquiring call parameter sequences of multiple home appliances controlled by IoT within a preset time range; identifying user activity patterns based on the multiple call parameter sequences to obtain identified activity patterns; calculating the similarity between the multiple call parameter sequences and the standard call parameter sequences of the multiple home appliances under the identified activity patterns to obtain pattern confidence; and optimizing the access methods of the multiple home appliances based on the pattern confidence to obtain an optimized access control scheme including multiple optimized access methods. This invention solves the problems in traditional IoT smart home appliance access control methods, where relying solely on manual user control leads to excessive device power consumption, impacting user experience, and unstable communication quality due to the inability to adaptively adjust the access method according to actual scenarios.

Description

Technical Field

[0001] This application relates to the field of Internet of Things (IoT) access control, and in particular to a multimodal access control method and system for IoT smart home appliances. Background Technology

[0002] With the development of IoT smart home appliance technology, the proper control of access methods is crucial for ensuring the efficient and stable operation of devices, becoming a key technical challenge in the IoT smart home appliance field. Currently, traditional access control relies on manual user operation, which is difficult to adapt to the differences in access requirements caused by the characteristics of different home appliances and the dynamic changes in usage scenarios. This easily leads to a mismatch between the access method and actual needs, which not only increases the power consumption of the devices and affects the user experience, but also reduces the stability and energy efficiency of smart home appliance communication. Summary of the Invention

[0003] To address the aforementioned technical issues, this application provides a multimodal access control method and system for IoT smart home appliances, which improves upon the current situation in traditional access control where reliance on manual user operation and the inability to adaptively adjust leads to excessive device power consumption, negative user experience, and unstable communication quality.

[0004] The embodiments of this application disclose the following technical solutions:

[0005] In a first aspect, embodiments of this application provide a multimodal access control method for IoT smart home appliances, the method comprising:

[0006] Obtain the call parameter sequence of multiple home appliances controlled by IoT within a preset time range in the past, wherein multiple home appliances are connected to IoT through a first access method or a second access method;

[0007] Based on multiple call parameter sequences, user activity pattern recognition is performed to obtain the recognized activity pattern;

[0008] Calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode to obtain the pattern confidence.

[0009] Based on the pattern confidence level, the access methods of multiple home appliances are optimized to obtain an optimized access control scheme that includes multiple optimized access methods, wherein the optimization step size is configured according to the pattern confidence level.

[0010] Secondly, embodiments of this application provide a multimodal access control system for IoT smart home appliances, the system comprising:

[0011] The parameter sequence acquisition module is used to acquire the call parameter sequence of multiple home appliances controlled by IoT within a preset time range in the past, wherein the multiple home appliances are connected to IoT through a first access method or a second access method.

[0012] The activity pattern recognition module is used to identify the user's activity pattern based on multiple call parameter sequences and obtain the identified activity pattern.

[0013] The pattern confidence calculation module is used to calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode, and obtain the pattern confidence.

[0014] The access method optimization module is used to optimize the access methods of multiple home appliances based on the mode confidence level, and obtain an optimized access control scheme including multiple optimized access methods, wherein the optimization step size is configured according to the mode confidence level.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0016] This application proposes a multimodal access control method and system for IoT smart home appliances. Through collaborative operations involving acquiring the call parameter sequences of home appliances, identifying user activity patterns, calculating pattern confidence, and optimizing access methods based on pattern confidence, adaptive control and efficient management of smart home appliance access are achieved. First, the call parameter sequences of multiple IoT-controlled home appliances within a preset time range are acquired. These appliances access the system via a first or second access method. Next, the call frequency is calculated based on the call parameter sequences and sent to a cloud server for an activity pattern recognizer to identify user activity patterns. Then, the similarity between the actual call parameter sequences and the standard call parameter sequences under the identified activity patterns is calculated to obtain the pattern confidence. Finally, the pattern deviation is calculated based on the pattern confidence as the optimization step size. The step size is determined by considering the number of devices. Using the current access method as the initial access control scheme, a corresponding number of devices are randomly selected to switch access methods to form an adjusted access control scheme. Simultaneously, the access fitness is calculated and iterative optimization is performed to obtain a converged optimized access control scheme.

[0017] This technical solution addresses the problems of poor communication quality and high energy consumption in traditional IoT smart home appliance access caused by the inability of fixed connection methods to flexibly cope with interference and bandwidth changes, through steps such as the collection and analysis of device call parameters, cloud server-based activity pattern recognition, pattern confidence-driven access method optimization, and multi-dimensional access adaptability evaluation. It achieves dynamic adaptation and optimization of access control methods, providing technical support for improving the communication quality and energy efficiency of smart home appliances. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating the multimodal access control method for IoT smart home appliances provided in this application embodiment;

[0020] Figure 2 This is a schematic diagram of the structure of a multimodal access control system for IoT smart home appliances provided in an embodiment of this application.

[0021] The components represented by each number in the attached diagram are explained below:

[0022] Parameter sequence acquisition module 01, activity pattern recognition module 02, pattern confidence calculation module 03, access method optimization module 04. Detailed Implementation

[0023] This application provides a multimodal access control method and system for IoT smart home appliances, which solves the technical problems in the prior art that rely on manual user control of the access method, resulting in excessive power consumption and affecting use in some cases, and the inability to adaptively adjust the access method according to device characteristics and usage scenarios, thus causing unstable communication quality and low energy efficiency.

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and 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.

[0025] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0026] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0027] Example 1, as shown in the appendix Figure 1 As shown, this application provides a multimodal access control method for IoT smart home appliances, the method comprising the following steps:

[0028] S110: Obtain the call parameter sequence of multiple home appliances controlled by the IoT within a preset time range in the past, wherein the multiple home appliances are connected to the IoT through the first access method or the second access method;

[0029] In this embodiment of the application, in the scenario of IoT smart home appliance access control, in order to accurately capture the calling patterns of home appliances to support the subsequent optimization of access methods, it is necessary to continuously monitor and record key parameters during the operation of the devices in real time.

[0030] Specifically, we must first clarify the differences in characteristics between access methods. The first access method (such as Wi-Fi) is characterized by high power consumption but low latency, while the second access method (such as Bluetooth) is characterized by low power consumption but high latency. Many home appliances initially connect to the IoT through one of these two methods.

[0031] Furthermore, during the operation of home appliances, various call parameters are monitored and recorded in real time. These call parameters not only include the operation instructions of the home appliances, but also cover the key call timestamps, thereby accurately marking the time node of each device call.

[0032] Furthermore, all call parameters within a preset time range (e.g., the past half hour) are sorted and integrated in chronological order to form a call parameter sequence that corresponds one-to-one with each home appliance, so as to fully present the call trajectory of the device within that time period.

[0033] This step, through real-time monitoring and time-series recording, provides raw data support for subsequent user activity pattern recognition and access method optimization, ensuring the timeliness and accuracy of the analysis process.

[0034] Step S110 in the method provided in this application embodiment includes:

[0035] During the operation of multiple home appliances controlled by IoT, call parameters are continuously monitored and recorded. Among them, multiple home appliances are connected to IoT through a first access method or a second access method. The energy consumption of the first access method is greater than that of the second access method, and the latency of the first access method is less than that of the second access method.

[0036] Record the call parameters of multiple home appliances within a preset time range in chronological order to obtain multiple call parameter sequences, where each call parameter includes a call timestamp.

[0037] In this embodiment of the application, in order to accurately capture the calling patterns of home appliances and provide reliable data support for subsequent activity pattern recognition and access method optimization, it is necessary to conduct comprehensive monitoring and recording of calling parameters to adapt to the operating characteristics of devices under different access methods.

[0038] Specifically, during the operation of multiple home appliances controlled by IoT, each home appliance is connected to IoT through either the first or second access method.

[0039] The first access method is through Wi-Fi. This method has relatively high energy consumption, but low data transmission latency, making it suitable for scenarios with high response speed requirements, such as the transmission of high-definition content from smart TVs and the uploading of real-time monitoring data from smart cameras.

[0040] Secondly, the second access method is through Bluetooth (such as Bluetooth Mesh), which has lower power consumption but relatively higher transmission latency. It is more suitable for devices that are sensitive to power consumption and do not have strict requirements for timely response, such as status feedback of smart door locks and data reporting of temperature and humidity sensors.

[0041] Furthermore, real-time monitoring will be automatically initiated when home appliances are in operation to continuously record the call parameters of each home appliance.

[0042] These call parameters not only cover the operation type of home appliances (such as on, off, adjustment mode, etc.), but also include key call timestamps, accurate to the second, so as to accurately mark the specific time when each home appliance is called.

[0043] For example, if a smart lighting device is connected via Wi-Fi and turned on at 8:30:25, the timestamp and operation type will be fully recorded; if a smart curtain is connected via Bluetooth and turned off at 10:15:40, the relevant information will also be stored in the monitoring database.

[0044] Based on this, all call parameters within a preset time range (such as the most recent half hour) are sorted and arranged in chronological order to form a call parameter sequence that corresponds one-to-one with each home appliance.

[0045] For example, for a smart air conditioner, if it is invoked at 9:05:10 (Wi-Fi access, temperature adjusted to 26℃), 9:15:30 (Wi-Fi access, dehumidification mode activated), and 9:25:45 (Wi-Fi access, device turned off) during the period from 9:00 to 9:30, these three timestamped call parameters will be arranged in chronological order to form the call parameter sequence of the smart air conditioner during that time period.

[0046] Meanwhile, to ensure the integrity and accuracy of the data, a redundant backup method is adopted when monitoring home appliances in real time. That is, the recorded call parameters are cached locally and synchronized with the cloud every 5 minutes to avoid data loss due to sudden power outages or network fluctuations.

[0047] In addition, an automatic verification and correction function is set up for abnormal data (such as duplicate timestamps, incorrect parameter formats, etc.). For example, when there is an obvious logical contradiction between the timestamp of a record and the records before and after it (such as time reversal), automatic verification will be triggered. The data will be corrected or marked as suspicious data in combination with the operation records of home appliances, and then included in the call parameter sequence after manual review.

[0048] In addition, the preset time range can be flexibly adjusted according to the actual application scenario. In addition to the default half hour, it can also be set to 1 hour, 2 hours, etc., to adapt to the usage habits of different users and the calling frequency of devices.

[0049] For example, for kitchen appliances that are frequently used (such as microwave ovens and coffee makers), a shorter preset time range (such as 15 minutes) can be set to capture their usage patterns more promptly; for devices that are used less frequently (such as smart wardrobes and air purifiers), a longer time range (such as 1 hour) can be set to ensure that enough sample data can be collected.

[0050] Ultimately, through the above steps of continuous monitoring, timestamp recording, time-series sorting, and data verification, multiple call parameter sequences were formed, which fully presented the call trajectory of each home appliance within the preset time. This laid a solid foundation for subsequent user activity pattern recognition, pattern confidence calculation, and access method optimization based on this data, ensuring the scientific nature and reliability of the entire access control optimization process.

[0051] S120: Based on multiple call parameter sequences, identify the user's activity pattern and obtain the identified activity pattern;

[0052] In this embodiment of the application, in the scenario of IoT smart home appliance access control, in order to determine the current activity status of the home appliance by the user's usage pattern, and thus provide a scenario basis for optimizing the access method, it is necessary to perform activity pattern recognition based on the device call parameter sequence.

[0053] Specifically, based on the obtained call parameter sequences of multiple home appliances, the number of times each appliance is called within a unit of time is calculated, thus obtaining multiple call frequencies. This call frequency data is then transmitted to a cloud server via the Internet of Things (IoT) to serve as input for activity pattern recognition.

[0054] Furthermore, a pre-trained activity pattern recognizer is deployed in the cloud server. This recognizer can analyze and match the frequency of input calls to accurately associate them with the user's actual activity status.

[0055] Based on this, when multiple call frequencies are input into the activity pattern recognizer, the recognizer outputs the activity pattern that best matches the current call pattern through feature matching and probability calculation, so as to achieve accurate prediction of user behavior.

[0056] Finally, the cloud server feeds back the identified activity patterns to the local system, completing the process of identifying user activity patterns.

[0057] This step uses the cloud server's pattern recognizer to accurately judge user behavior, providing key scenario information for subsequent optimization of access methods based on activity patterns, ensuring that the access strategy is adapted to the user's actual needs.

[0058] Step S120 in the method provided in this application embodiment includes:

[0059] Based on the multiple call parameter sequences, multiple call frequencies are calculated and sent to the cloud server via the Internet of Things;

[0060] The cloud server receives the identified activity pattern, which is obtained by inputting the multiple call frequencies into the activity pattern recognizer within the cloud server.

[0061] In this embodiment of the application, in order to accurately identify the user's current activity pattern through the calling patterns of home appliances and provide a scenario-based basis for subsequent access method optimization, it is necessary to carry out systematic activity pattern recognition based on the calling parameter sequence, covering calling frequency calculation, cloud recognition and result feedback.

[0062] Specifically, based on the obtained call parameter sequences of multiple home appliances, the call frequency of each home appliance within a unit of time is calculated.

[0063] The calculation of the call frequency is based on a preset time range. For example, for the call parameter sequence of the past half hour, the total number of calls for each home appliance in that time period is counted and then divided by the time length (30 minutes) to obtain the call frequency in "times / minute".

[0064] For example, if a smart TV is called 6 times in the past half hour, its call frequency is 0.2 times / minute (6 / 30=0.2); if a rice cooker is called 2 times in the past half hour, its call frequency is 0.067 times / minute (2 / 30≈0.067). In this way, the call parameter sequence of each home appliance is converted into a quantifiable call frequency, forming a dataset composed of multiple call frequencies.

[0065] Furthermore, the calculated call frequencies are transmitted to the cloud server via the Internet of Things.

[0066] In particular, encryption protocols are used during data transmission to ensure the security of transmitted data, and the data format is standardized (such as uniformly encapsulating device ID, call frequency value, data generation time and other information in JSON format) so that the cloud server can parse it efficiently.

[0067] For example, a certain transmitted data may contain structured data such as "Home appliance ID: TV-001, call frequency: 0.2 times / minute; Home appliance ID: RiceCooker-002, call frequency: 0.067 times / minute".

[0068] Based on this, the cloud server processes the received call frequency and completes pattern recognition through a pre-configured activity pattern recognizer.

[0069] The method provided in this application embodiment includes the following configuration steps for the "activity pattern recognizer":

[0070] Based on the user's home appliance call data over a historical period, multiple sample call frequency sets are collected, and the user's activity patterns under different sample call frequency sets are labeled to obtain a sample activity pattern set.

[0071] Based on machine learning, an activity pattern recognizer is constructed and trained using the multiple sample call frequency sets and sample activity pattern sets.

[0072] Configure the activity pattern recognizer on the cloud server.

[0073] In this embodiment of the application, in order to achieve accurate identification of user activity patterns, it is necessary to build and train an activity pattern recognizer based on the user's historical home appliance call data, and learn the correlation features between call frequency and activity pattern through machine learning algorithms in order to adapt to the different usage habits and scenario differences of different users.

[0074] Specifically, the first step is to collect and label sample data. Based on users' appliance usage data over a historical period (e.g., the past 3 months), the sequence of usage parameters for different time periods is extracted, and multiple sets of sample usage frequencies are calculated.

[0075] Each sample call frequency set contains the call frequency of various home appliances within that time period, such as "smart TV: 0.3 times / minute, audio equipment: 0.25 times / minute, humidifier: 0.05 times / minute" and other data.

[0076] Meanwhile, through manual annotation, the user activity patterns corresponding to each sample call frequency set are determined, forming a sample activity pattern set.

[0077] For example, in one sample call frequency set, "gas stove: 0.2 times / minute, range hood: 0.2 times / minute, rice cooker: 0.15 times / minute" was manually identified as the "cooking" mode; in another sample call frequency set, "treadmill: 0.4 times / minute, smart bracelet: 0.3 times / minute, air conditioner: 0.1 times / minute" was manually identified as the "exercise" mode.

[0078] In addition, in another set of sample call frequencies, "Smart TV: 0.5 times / minute, Projector: 0.4 times / minute, Sofa Massager: 0.2 times / minute" were manually analyzed and labeled as "watching TV" mode. By covering multiple scenarios including daily living, leisure and entertainment, and housework, the richness and representativeness of the sample activity pattern set are ensured.

[0079] Based on this, the random forest algorithm is selected as the basic framework to build an activity pattern recognizer, so as to learn and process the mapping relationship between multiple device call frequencies and multiple activity patterns.

[0080] Specifically, the call frequency of each household appliance in the sample call frequency set is first used as the input feature, and the data is standardized (that is, the call frequency is normalized to the [0, 1] interval) to eliminate the impact of the difference in the numerical range of call frequency of different devices on model training.

[0081] For example, “0.3 times / minute” is standardized to 0.6 (assuming the device’s historical maximum call frequency is 0.5 times / minute) to ensure the rationality of feature weight calculation.

[0082] Secondly, a random forest model containing 100 decision trees is constructed, with a maximum depth of 10 for each decision tree and a minimum number of samples per leaf node of 5. The model's generalization ability is enhanced by randomly selecting a subset of features (choosing the optimal splitting feature from the square root of the total number of features at each split) and simultaneously employing bootstrap sampling to select a subset of samples.

[0083] Furthermore, the sample call frequency set and the sample activity pattern set are divided in an 8:2 ratio. For example, 8,000 samples are selected from 10,000 samples as the training set and 2,000 samples as the validation set. The training set is used for model parameter learning, and the validation set is used for real-time monitoring of the model's recognition performance.

[0084] Meanwhile, classification accuracy was used as the core evaluation metric, and a grid search method was employed to optimize the model's hyperparameters. The initial learning rate was set to 0.01, and the hyperparameter combination (such as the number of decision trees, maximum depth, etc.) was adjusted every 20 iterations.

[0085] During training, model error is monitored in real time using a validation set. The model is considered converged when the accuracy improvement on the validation set is less than 0.5% for 15 consecutive iterations. For example, after 500 iterations, the model achieves 96% accuracy on the validation set, which meets the needs of practical applications.

[0086] During the testing phase, a reserved test set (10% of the total sample) is input into the trained model. If the activity pattern recognition accuracy is not less than 95%, and the recall rate for the main activity patterns (such as "cooking" and "watching TV") is higher than 90%, the activity pattern recognizer is considered qualified. For example, if there are 100 "exercise" pattern samples in the test set, and the model correctly recognizes 96 of them, the recall rate is 96%, which meets the evaluation criteria.

[0087] In addition, a model update mechanism was established, which involves incremental training every two weeks based on newly generated user call data and activity pattern annotations to update the activity pattern recognizer parameters and adapt to dynamic changes in user habits. For example, if a user adds a "remote work" mode (manifested as frequent calls to computers, desk lamps, and printers), the activity pattern recognizer gradually learns the features of this mode through incremental training, and after multiple iterations, the recognition accuracy improves to 97%.

[0088] Finally, the trained activity pattern recognizer is deployed to a cloud server, enabling it to receive real-time call frequency data and output activity pattern recognition.

[0089] For example, when the cloud server receives real-time call frequency data from a user's household: "Smart TV: 0.4 times / minute, Audio: 0.35 times / minute, Lights: 0.2 times / minute," the activity pattern recognizer performs feature matching and probability calculation on this data. By comparing it with the "watching TV" pattern features learned during training, it outputs a matching probability of 95% for this pattern and feeds "watching TV" as the recognized activity pattern back to the local system, providing a basis for subsequently optimizing the home appliance access method based on this pattern.

[0090] Meanwhile, if the call frequency data of "gas stove: 0.25 times / minute, range hood: 0.25 times / minute, microwave oven: 0.15 times / minute" is received, the activity pattern recognizer will match the "cooking" mode and output a matching probability of 96% to ensure that the recognition result is highly consistent with the user's actual activity scenario.

[0091] Ultimately, the local system completes the closed loop of identifying user activity patterns by receiving the activity patterns sent by the cloud server, thus providing a direct basis for optimizing home appliance access methods based on activity patterns.

[0092] S130: Calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode to obtain the pattern confidence.

[0093] In this embodiment of the application, in the scenario of IoT smart home appliance access control, in order to quantify the matching degree between the current device calling pattern and the identified activity pattern, and to provide an adjustment basis for the optimization of the access method, it is necessary to obtain the pattern confidence by calculating the similarity, so as to achieve accurate measurement of the adaptability between the current device operating status and the user activity pattern.

[0094] Specifically, the process begins by acquiring multiple standard call frequencies for various home appliances under different activity modes. These standard call frequencies are pre-set based on a large amount of historical data and together constitute the core indicators of the standard call parameter sequence corresponding to the activity mode.

[0095] Furthermore, based on the acquired multiple call parameter sequences, the actual call frequency of the current multiple home appliances is calculated using the same time unit as the calculation standard call frequency, in order to form key quantitative data of the current call parameter sequence.

[0096] Based on this, the similarity between multiple actual call frequencies and corresponding multiple standard call frequencies is calculated, and this similarity is used as the pattern confidence score, which reflects the degree of matching between the current device call pattern and the identified activity pattern.

[0097] Specifically, the cosine similarity algorithm in the existing technology is adopted to determine the degree of matching by measuring the cosine value of the angle between two call frequency vectors. The value range is [0, 1], and the closer the value is to 1, the higher the similarity between the two.

[0098] Ultimately, the calculated pattern confidence score intuitively reflects the degree of fit between the current actual call situation and the identified activity pattern. This provides a quantitative reference for configuring optimization step size and adjusting access methods based on the confidence score, ensuring that the access method optimization strategy is more in line with the actual scenario requirements.

[0099] Step S130 in the method provided in this application embodiment includes:

[0100] Obtain multiple standard call frequencies of multiple home appliances under the identified activity mode;

[0101] Multiple call frequencies are calculated based on multiple call parameter sequences;

[0102] Calculate the similarity between the multiple call frequencies and the multiple standard call frequencies to obtain the pattern confidence score.

[0103] In this embodiment of the application, in order to quantify the matching degree between the current home appliance call frequency and the identified activity pattern, and to provide a precise adjustment basis for the optimization of the access method, it is necessary to calculate the similarity to obtain the pattern confidence, so as to realize the quantitative evaluation of the adaptability between the current device operating status and the user activity pattern.

[0104] Specifically, the system first acquires multiple standard call frequencies for various home appliances under different activity modes. These standard call frequencies are pre-set baseline values ​​based on a large amount of historical data and scene characteristics, and can reflect the typical usage intensity of each device under specific activity modes.

[0105] For example, in "watching TV" mode, the standard call frequency of the smart TV is set to 0.3 times / minute, the sound system to 0.25 times / minute, and the lights to 0.15 times / minute; in "cooking" mode, the standard call frequency of the gas stove is 0.2 times / minute, the range hood to 0.2 times / minute, and the rice cooker to 0.1 times / minute.

[0106] Meanwhile, the standard call frequency will be configured according to the different usage habits of different user groups. For example, for the "cooking" mode of elderly users, the standard call frequency of the microwave oven may be appropriately increased to 0.15 times / minute to adapt to their usage characteristics.

[0107] Furthermore, based on the obtained multiple call parameter sequences, using the same calculation method as the standard call frequency calculation, and taking a preset time range (such as the past half hour) as the benchmark, the total number of calls for each home appliance within that time period is counted, and then divided by the time length to obtain the actual call frequency per unit time.

[0108] For example, in a scenario where the activity mode is identified as "watching TV," statistics from the current call parameter sequence show that: the smart TV was called 8 times in the past half hour, with an actual call frequency of 0.27 times / minute (8 / 30≈0.27); the speaker was called 7 times, with an actual call frequency of 0.23 times / minute (7 / 30≈0.23); and the lights were called 5 times, with an actual call frequency of 0.17 times / minute (5 / 30≈0.17). This calculation method can transform the current call parameter sequence of electrical appliances into quantitative data that can be compared with standard values.

[0109] Based on this, the similarity between multiple actual call frequencies and corresponding multiple standard call frequencies is calculated, and this similarity value is used as the pattern confidence score.

[0110] Specifically, the similarity calculation adopts the cosine similarity algorithm in the existing technology. It constructs the actual call frequency vector and the standard call frequency vector, and calculates the cosine value of the angle between the two to measure the degree of matching. This similarity value is used as the pattern confidence.

[0111] The confidence level of this pattern ranges from [0, 1]. The closer the value is to 1, the higher the match between the actual call frequency of the current home appliance and the standard call frequency under the identified activity pattern, indicating that the user's current behavior is more consistent with the identified activity pattern. Conversely, the lower the match between the actual call frequency and the standard call frequency, the more obvious the deviation between the user's current behavior and the identified activity pattern.

[0112] For example, in the "cooking" mode, the standard call frequency vector is (gas stove: 0.2 times / minute, range hood: 0.2 times / minute, rice cooker: 0.1 times / minute). If the current actual call frequency vector is (gas stove: 0.19 times / minute, range hood: 0.21 times / minute, rice cooker: 0.09 times / minute), the cosine similarity between the two calculated by the cosine similarity algorithm is 0.99, that is, the pattern confidence is 0.99, indicating that the matching degree between the two is extremely high.

[0113] Conversely, if the actual call frequency vector is (gas stove: 0.05 times / minute, range hood: 0.06 times / minute, rice cooker: 0.03 times / minute), and the cosine similarity calculated by the cosine similarity algorithm is 0.6, that is, the pattern confidence is 0.6, it indicates that the matching degree between the two is low, and the current call situation is significantly different from the "cooking" pattern.

[0114] S140: Based on the pattern confidence level, optimize the access methods of multiple home appliances to obtain an optimized access control scheme that includes multiple optimized access methods, wherein the optimization step size is configured according to the pattern confidence level.

[0115] In this embodiment of the application, in the scenario of IoT smart home appliance access control, in order to optimize the access method in a targeted manner based on the matching degree between the current device call frequency and the identification activity mode in order to reduce power consumption and latency, it is necessary to configure the optimization step size based on the mode confidence and achieve the optimization of the access scheme through iterative adjustment.

[0116] Specifically, the pattern bias is first calculated based on the pattern confidence score and then used as the optimization step size. The pattern bias score represents the degree of deviation between the current identified activity pattern and the standard pattern, reflecting the uniqueness of the current scene.

[0117] The higher the pattern confidence, the smaller the pattern bias, indicating that the current recognized activity pattern is more in line with typical rules, and the optimization step size can be set to be smaller; the lower the pattern confidence, the larger the pattern bias, indicating that the current scene is more special, and a larger optimization step size needs to be set to accommodate the pattern bias.

[0118] At the same time, the step size is determined based on the total number of home appliances, that is, the number of devices adjusted in each iteration, to ensure that the optimization process can cover a sufficient number of home appliances to achieve overall performance improvement, while avoiding system fluctuations caused by adjusting too many devices at once.

[0119] Furthermore, the current access methods of multiple home appliances are used as the initial access control scheme, and this is used as a benchmark for gradual iterative optimization.

[0120] Based on this, home appliances with a step size equal to the number of devices randomly selected from the initial access control scheme are switched in their access mode (e.g., the first access mode is changed to the second access mode, or vice versa) to form an adjusted access control scheme.

[0121] Furthermore, based on the device call frequency within multiple call parameter sequences, the access adaptability of the adjusted access control scheme is calculated.

[0122] Similarly, the access control scheme is iteratively optimized according to a determined step size. In each iteration, a number of devices with a step size are randomly selected to switch access methods, the access fitness of the new scheme is calculated, and the fitness of the new scheme is compared with that of the previous scheme.

[0123] If the new scheme has a higher access fitness, the adjustment is retained; if it is lower, it is discarded and other combinations are tried. This process is repeated until the access fitness no longer improves significantly after several iterations (i.e., convergence is reached). The scheme corresponding to the maximum access fitness at this point is the optimized access control scheme.

[0124] This step transforms pattern confidence into optimization step size, enabling the adjustment range of the access method to be adapted to the specific characteristics of the scenario. This avoids resource waste caused by over-optimization and ensures access performance in special scenarios, ultimately achieving the goals of reducing power consumption and latency, and providing support for the efficient and stable operation of smart home appliances.

[0125] Step S140 in the method provided in this application embodiment includes:

[0126] Based on the pattern confidence, the pattern bias is calculated and used as the optimization step size. The number of steps is determined by combining the number of multiple home appliances.

[0127] Use the current connection methods of multiple home appliances as the initial access control scheme;

[0128] The access methods of the number of home appliances randomly selected within the initial access control scheme are adjusted to obtain an adjusted access control scheme.

[0129] The access adaptability of the adjusted access control scheme is calculated based on the multiple call frequencies of multiple home appliances within multiple call parameter sequences.

[0130] Continue to iteratively optimize the access control scheme according to the stated step size to obtain the optimized access control scheme corresponding to the maximum converged access fitness value.

[0131] In this embodiment of the application, in order to optimize the access method in a targeted manner based on the matching degree between the current device call frequency and the identified activity mode in order to reduce power consumption and latency, it is necessary to configure the optimization step size based on the mode confidence and achieve the optimization of the access scheme through multi-step iterative adjustment to ensure that the access strategy is highly adapted to the actual needs of users.

[0132] Specifically, the mode bias is first calculated based on the obtained mode confidence using the formula "mode bias = 1 - mode confidence", and then used as the optimization step size.

[0133] Among them, the pattern deviation degree and the pattern confidence degree are negatively correlated. That is, the higher the pattern confidence degree, the higher the matching degree between the current calling frequency and the recognition activity pattern. The smaller the pattern deviation degree, the smaller the optimization step size should be set accordingly to avoid excessive adjustment and interference to the stable operation of the equipment. The lower the pattern confidence degree, the greater the deviation between the current scene and the typical pattern. The greater the pattern deviation degree, the larger the optimization step size should be set to quickly adapt to the special scene.

[0134] Based on this, the step size (i.e. the number of home appliances adjusted in each iteration) is determined by the total number of home appliances. For example, when the total number of home appliances is 20 and the step size ratio corresponding to the mode deviation is 10%, the step size is 2 (20×10%=2) to ensure that each adjustment can cover a certain range of home appliances while controlling the adjustment range to maintain equipment stability.

[0135] Furthermore, the current access methods of multiple home appliances are used as the initial access control scheme. This initial access control scheme reflects the current actual connection status of the home appliances. For example, smart TVs and cameras are connected via the first access method (Wi-Fi), while smart door locks and temperature and humidity sensors are connected via the second access method (Bluetooth). Subsequent optimization work is carried out based on this.

[0136] Based on this, home appliances with a step size equal to the number of devices randomly selected from the initial access control scheme are switched in their access mode, i.e., by changing the original first access mode to the second access mode, or changing the original second access mode to the first access mode, so as to form an adjusted access control scheme.

[0137] For example, when the step size is 2, a smart TV and a temperature and humidity sensor are randomly selected, the smart TV is switched to Bluetooth access, and the temperature and humidity sensor is switched to Wi-Fi access, thus obtaining the adjusted access control scheme.

[0138] By randomly selecting devices, the influence of subjective preferences on the optimization results can be effectively avoided, thus ensuring the objectivity of the access control scheme.

[0139] Furthermore, based on the device call frequency within multiple call parameter sequences, the access adaptability of the adjusted access control scheme is calculated to comprehensively evaluate whether the adjusted access control scheme performs better in terms of energy consumption and latency.

[0140] The method provided in this application embodiment includes the step of "calculating the access adaptability of the adjusted access control scheme based on the multiple calling frequencies of multiple home appliances within multiple calling parameter sequences":

[0141] Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the unit energy consumption of each home appliance under different access methods, the adjustment access energy consumption is calculated.

[0142] Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the interaction latency under different access methods for each home appliance, the adjustment access latency is calculated.

[0143] Based on the mode deviation, adjusted access energy consumption, and adjusted access delay, the access adaptability of the adjusted access control scheme is calculated.

[0144] In this embodiment of the application, in order to comprehensively measure the overall performance of the access control scheme in terms of energy consumption and latency, and to provide a quantitative basis for the iterative optimization of the access method, it is necessary to calculate the energy consumption, latency and overall access adaptability to achieve an accurate assessment of the adaptability of the access control scheme.

[0145] Specifically, the adjustment access energy consumption is first calculated based on the adjustment access methods of multiple home appliances within the adjustment access control scheme, and combined with the unit energy consumption of each home appliance under different access methods.

[0146] Unit energy consumption refers to the energy consumption of home appliances per unit time under a specific access method. It is obtained through long-term monitoring and statistical analysis of the operating data of home appliances under different access methods. For example, the unit energy consumption of a smart TV is 8Wh when connected via Wi-Fi and 3Wh when connected via Bluetooth; the unit energy consumption of a humidifier is 5Wh when connected via Wi-Fi and 2Wh when connected via Bluetooth.

[0147] Furthermore, when calculating and adjusting access energy consumption, it is necessary to take into account the calling frequency of home appliances and perform weighted calculation of unit energy consumption. The specific calculation formula can be expressed as "Adjusted access energy consumption = Σ (unit energy consumption of a certain home appliance × calling frequency of the home appliance)".

[0148] For example, if the access control scheme is adjusted so that the smart TV accesses via Bluetooth (unit energy consumption 3Wh) and the call frequency is 0.3 times / minute, and the humidifier accesses via Wi-Fi (unit energy consumption 5Wh) and the call frequency is 0.1 times / minute, then the access energy consumption adjustment = (3Wh × 0.3 times / minute) + (5Wh × 0.1 times / minute) = 1.4 (Wh·time) / minute, thus clearly showing the comprehensive energy consumption of different devices after adjusting the access method.

[0149] Furthermore, based on the adjusted access methods of multiple home appliances within the adjusted access control scheme, and combined with the interaction latency of each home appliance under different access methods, the adjusted access latency is calculated.

[0150] Interaction latency refers to the time interval between a device sending a command and receiving a response. For example, the interaction latency of a smart camera is 10ms when connected via Wi-Fi and 40ms when connected via Bluetooth; the interaction latency of a smart door lock is 15ms when connected via Wi-Fi and 30ms when connected via Bluetooth.

[0151] Similarly, just like the method for calculating the energy consumption of the adjusted access, the calculation of the adjusted access latency also needs to be weighted in combination with the calling frequency of home appliances, and is calculated by the formula "Adjusted access latency = Σ (interaction latency of a certain home appliance × calling frequency of the home appliance)".

[0152] For example, if the smart camera in the adjustment scheme accesses via Wi-Fi (interaction latency 10ms) and the call frequency is 0.2 times / minute; and the smart door lock accesses via Bluetooth (interaction latency 30ms) and the call frequency is 0.05 times / minute, then the adjusted access latency = (10ms × 0.2 times / minute) + (30ms × 0.05 times / minute) = 3.5 (ms·times) / minute, which intuitively reflects the response speed performance of the adjusted access control scheme.

[0153] Based on this, the access adaptability of the adjusted access control scheme is calculated according to the obtained mode deviation degree, adjusted access energy consumption and adjusted access delay.

[0154] The method provided in this application embodiment includes the step of "calculating the access adaptability of the adjusted access control scheme based on the mode deviation, adjusted access energy consumption, and adjusted access latency" as follows:

[0155] The mode deviation is used as the energy consumption weight, and the time delay weight is calculated.

[0156] Obtain the standard access energy consumption and standard access latency of the identified activity pattern;

[0157] The access fitness is obtained by weighting the ratio of standard access energy consumption to adjusted access energy consumption and the ratio of standard access latency to adjusted access latency using the energy consumption weight and latency weight.

[0158] In this embodiment of the application, in order to comprehensively evaluate the performance of the adjusted access control scheme in terms of energy consumption and latency, it is necessary to obtain the access fitness through weight allocation and weighted calculation, so as to quantify the degree of fit between the adjusted access control scheme and the ideal state under the identification activity mode.

[0159] Specifically, the mode deviation is first used as the energy consumption weight, and the time delay weight is calculated using the formula "time delay weight = 1 - energy consumption weight".

[0160] Among them, the mode deviation reflects the degree of deviation between the current activity mode and the standard mode. When the mode deviation is large (i.e. the current scenario is more specific), the energy consumption weight increases accordingly, which means that more emphasis should be placed on optimizing energy consumption. Conversely, when the mode deviation is small, the latency weight increases, and more emphasis should be placed on reducing latency to ensure response speed.

[0161] Furthermore, the standard access energy consumption and standard access latency for the identified activity mode are obtained. These standard values ​​are benchmark parameters pre-set based on typical access methods of home appliances under this identified activity mode, and can reflect the ideal state of the access solution in this scenario.

[0162] For example, in the "watch TV" mode, considering that smart TVs and speakers require low latency to ensure smooth playback, their typical access method is Wi-Fi, while lights have lower latency requirements and use Bluetooth access. The standard access energy consumption is set to 60 (Wh·times) / minute (the total energy consumption of TVs and speakers accessed via Wi-Fi and lights accessed via Bluetooth), and the standard access latency is 30ms (mainly determined by the latency of Wi-Fi access devices).

[0163] Furthermore, in "Nighttime Rest" mode, the smart curtains and nightlights prioritize low energy consumption, both using Bluetooth connectivity. The standard access energy consumption is set at 15 (Wh·times) / minute, and the standard access latency is 80ms, balancing energy consumption with basic response requirements. These standard values ​​are dynamically adjusted based on the core needs of different activity modes to ensure a high degree of match with scene characteristics.

[0164] Based on this, the ratio of standard access energy consumption to adjusted access energy consumption and the ratio of standard access latency to adjusted access latency are weighted and calculated according to the obtained energy consumption weight and latency weight, so as to obtain the access fitness.

[0165] Specifically, it can be calculated using the formula "Access adaptability = (standard access energy consumption / adjusted access energy consumption) × energy consumption weight + (standard access latency / adjusted access latency) × latency weight".

[0166] The ratio of standard access energy consumption to adjusted access energy consumption reflects the degree of optimization of energy consumption after adjustment relative to the standard state. The larger the ratio, the more significant the energy consumption optimization effect. The ratio of standard access latency to adjusted access latency reflects the improvement of latency after adjustment relative to the standard state. The larger the ratio, the better the latency optimization effect.

[0167] For example, in "watching TV" mode, if the mode deviation is 0.3 (i.e., energy consumption weight is 0.3 and latency weight is 0.7), the standard access energy consumption is 60 (Wh·times) / minute, and the standard access latency is 30ms. In a certain adjusted access control scheme, the smart TV and speakers still access via Wi-Fi, while the lights switch to Wi-Fi access. The calculated adjusted access energy consumption is 70 (Wh·times) / minute, and the adjusted access latency is 25ms. Therefore, the ratio of standard access energy consumption to adjusted access energy consumption is 60 / 70≈0.86, the ratio of standard access latency to adjusted access latency is 30 / 25=1.2, and the access adaptability is 0.86×0.3+1.2×0.7≈1.1. This indicates that the adjusted access control scheme is significantly effective in latency optimization and has good overall adaptability.

[0168] In another access control scheme, the smart TV and speakers switch to Bluetooth access, while the lights remain connected via Bluetooth. The access power consumption is adjusted to 45 Wh / minute, and the access latency is 40 ms. The ratio of standard access power consumption to adjusted access power consumption is 60 / 45 ≈ 1.33, and the ratio of standard access latency to adjusted access latency is 30 / 40 = 0.75. Therefore, the access adaptability is 1.33 × 0.3 + 0.75 × 0.7 ≈ 0.93. This indicates that this scheme significantly optimizes energy consumption, but the increased latency results in slightly lower overall adaptability compared to the previous scheme.

[0169] Furthermore, the access control scheme is iteratively optimized according to the step size to obtain the optimized access control scheme corresponding to the maximum convergent access fitness.

[0170] Similarly, in each iteration, a number of home appliances equal to the step size are randomly selected from the current adjustment access control scheme, and their access methods are switched (i.e., the first access method and the second access method are interchanged) to form a new adjustment access control scheme.

[0171] Subsequently, following the same calculation method, based on the adjusted access control method, unit energy consumption, interaction delay, and call frequency of electrical equipment in the new adjusted access control scheme, the adjusted access energy consumption and adjusted access delay are recalculated, and finally the new access adaptability is obtained.

[0172] Meanwhile, the access adaptability of the newly adjusted access control scheme is compared with that of the previous adjusted access control scheme. If the access adaptability of the new scheme is higher, the new scheme is retained as the current optimal scheme; if the access adaptability of the new scheme is lower, the new scheme is discarded and the original scheme is still used as the current optimal scheme.

[0173] Based on this, the process of "selecting a device - switching access methods - calculating access fitness - comparing and retaining" is repeated for continuous iterative optimization until, after multiple (e.g., 10) iterations, the change in access fitness is less than a preset threshold (e.g., 0.01). At this point, access fitness is considered to have reached a convergence state. The adjustment access control scheme corresponding to the maximum access fitness value in the convergence state is the final optimized access control scheme.

[0174] For example, in the "cooking" mode, the initial access control scheme has an access fitness of 0.85. After 20 iterations of optimization, with a step size of 2, the adjusted access control scheme obtained in the 20th iteration allows the gas stove and range hood to access via the first access method (Wi-Fi), and the rice cooker and microwave oven to access via the second access method (Bluetooth), achieving an access fitness of 1.32. In the subsequent 10 iterations, the access fitness remains stable at around 1.32, with a change of less than 0.01. At this point, convergence is determined, and this access control scheme is the optimized access control scheme, which effectively reduces overall energy consumption while meeting the device response speed requirements of the "cooking" scenario.

[0175] Ultimately, the optimized access control scheme was applied to the actual IoT smart home appliance access control, enabling multiple home appliances to connect to the gateway according to the optimized access method. Through adaptive access method adjustments, it can flexibly respond to changes in electromagnetic interference and bandwidth requirements in different home environment scenarios, ensuring stable network operation while significantly improving device communication quality and energy efficiency, making the use of smart home appliances more aligned with users' actual activity patterns and needs.

[0176] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0177] This application proposes a multimodal access control method for IoT smart home appliances. First, it obtains the call parameter sequences of multiple home appliances controlled by the IoT within a preset time range. These appliances access the IoT through either a first access method (high energy consumption, low latency) or a second access method (low energy consumption, high latency), and the call parameters include information such as timestamps. Next, it calculates multiple call frequencies based on the call parameter sequences and sends them to a cloud server. An activity pattern recognizer within the cloud server identifies the user's activity pattern, trained based on the user's historical call data. Then, it calculates the similarity between the multiple call parameter sequences and the standard call parameter sequences under the identified activity pattern to obtain pattern confidence. Finally, it calculates the pattern deviation degree based on the pattern confidence as the optimization step size, determines the step size based on the number of devices, uses the current access method as the initial access control scheme, randomly selects a corresponding number of devices to switch access methods to form an adjusted access control scheme, calculates the access fitness, and iteratively optimizes to obtain a converged optimized access control scheme, which is then applied to a practical IoT smart home appliance system.

[0178] The method provided in this application, through the technical solution of "parameter acquisition - pattern recognition - confidence calculation - access optimization", solves the problem of poor communication quality and high energy consumption caused by the inability of fixed connection methods to flexibly adapt to changes in scenarios in traditional IoT smart home appliance access. It realizes dynamic adjustment and optimization of access methods, improves device communication quality and energy efficiency, and provides reliable technical support for the efficient and stable operation of smart home appliances.

[0179] Example 2, as shown in the appendix Figure 2 As shown, based on the inventive concept of the multimodal access control method for IoT smart home appliances provided in Embodiment 1, this application also provides a multimodal access control system for IoT smart home appliances, specifically including:

[0180] The parameter sequence acquisition module 01 is used to acquire the call parameter sequence of multiple home appliances controlled by the IoT within a preset time range in the past, wherein the multiple home appliances are connected to the IoT through a first access method or a second access method.

[0181] The activity pattern recognition module 02 is used to identify the user's activity pattern based on multiple call parameter sequences and obtain the identified activity pattern.

[0182] The pattern confidence calculation module 03 is used to calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode, and obtain the pattern confidence.

[0183] The access method optimization module 04 is used to optimize the access methods of multiple home appliances based on the mode confidence level, and obtain an optimized access control scheme including multiple optimized access methods, wherein the optimization step size is configured according to the mode confidence level.

[0184] In one embodiment, the parameter sequence acquisition module 01 is further configured to:

[0185] During the operation of multiple home appliances controlled by IoT, call parameters are continuously monitored and recorded. Among them, multiple home appliances are connected to IoT through a first access method or a second access method. The energy consumption of the first access method is greater than that of the second access method, and the latency of the first access method is less than that of the second access method.

[0186] Record the call parameters of multiple home appliances within a preset time range in chronological order to obtain multiple call parameter sequences, where each call parameter includes a call timestamp.

[0187] In one embodiment, the activity pattern recognition module 02 is further configured to:

[0188] Based on the multiple call parameter sequences, multiple call frequencies are calculated and sent to the cloud server via the Internet of Things;

[0189] The cloud server receives the identified activity pattern, which is obtained by inputting the multiple call frequencies into the activity pattern recognizer within the cloud server.

[0190] In one embodiment, the pattern confidence calculation module 03 is further configured to:

[0191] Obtain multiple standard call frequencies of multiple home appliances under the identified activity mode;

[0192] Multiple call frequencies are calculated based on multiple call parameter sequences;

[0193] Calculate the similarity between the multiple call frequencies and the multiple standard call frequencies to obtain the pattern confidence score.

[0194] In one embodiment, the access method optimization module 04 is further configured to:

[0195] Based on the pattern confidence, the pattern bias is calculated and used as the optimization step size. The number of steps is determined by combining the number of multiple home appliances.

[0196] Use the current connection methods of multiple home appliances as the initial access control scheme;

[0197] The access methods of the number of home appliances randomly selected within the initial access control scheme are adjusted to obtain an adjusted access control scheme.

[0198] The access adaptability of the adjusted access control scheme is calculated based on the multiple call frequencies of multiple home appliances within multiple call parameter sequences.

[0199] Continue to iteratively optimize the access control scheme according to the stated step size to obtain the optimized access control scheme corresponding to the maximum converged access fitness value.

[0200] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0201] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0202] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A multimodal access control method for IoT smart home appliances, characterized in that, The method includes: Obtain the call parameter sequence of multiple home appliances controlled by IoT within a preset time range in the past, wherein multiple home appliances are connected to IoT through a first access method or a second access method; Based on multiple call parameter sequences, user activity pattern recognition is performed to obtain the recognized activity pattern; Calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode to obtain the pattern confidence. The standard call parameter sequence is the standard call frequency of multiple home appliances under the activity mode, which is preset based on historical data. Based on the pattern confidence level, the access methods of multiple home appliances are optimized to obtain an optimized access control scheme that includes multiple optimized access methods. The optimization is performed by configuring an optimization step size based on the pattern confidence level, including: Based on the pattern confidence, the pattern bias is calculated and used as the optimization step size. The number of steps is determined by combining the number of multiple home appliances. Use the current connection methods of multiple home appliances as the initial access control scheme; The access methods of the number of home appliances randomly selected within the initial access control scheme are adjusted to obtain an adjusted access control scheme. Based on the multiple calling frequencies of multiple home appliances within multiple calling parameter sequences, the access adaptability of the adjusted access control scheme is calculated, including: Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the unit energy consumption of each home appliance under different access methods, the adjustment access energy consumption is calculated. Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the interaction latency under different access methods for each home appliance, the adjustment access latency is calculated. Based on the mode deviation, adjusted access energy consumption, and adjusted access latency, the access fitness of the adjusted access control scheme is calculated, including: The mode deviation is used as the energy consumption weight, and the time delay weight is calculated. Obtain the standard access energy consumption and standard access latency of the identified activity pattern; The access fitness is obtained by weighting the ratio of standard access energy consumption to adjusted access energy consumption and the ratio of standard access latency to adjusted access latency using the energy consumption weight and latency weight. Continue to iteratively optimize the access control scheme according to the stated step size to obtain the optimized access control scheme corresponding to the maximum converged access fitness value.

2. The multimodal access control method for IoT smart home appliances according to claim 1, characterized in that, Retrieve the call parameter sequence of multiple home appliances controlled by IoT within a preset time range in the past, including: During the operation of multiple home appliances controlled by IoT, call parameters are continuously monitored and recorded. Among them, multiple home appliances are connected to IoT through a first access method or a second access method. The energy consumption of the first access method is greater than that of the second access method, and the latency of the first access method is less than that of the second access method. Record the call parameters of multiple home appliances within a preset time range in chronological order to obtain multiple call parameter sequences, where each call parameter includes a call timestamp.

3. The multimodal access control method for IoT smart home appliances according to claim 1, characterized in that, Based on multiple call parameter sequences, user activity pattern recognition is performed to obtain the identified activity patterns, including: Based on the multiple call parameter sequences, multiple call frequencies are calculated and sent to the cloud server via the Internet of Things; The cloud server receives the identified activity pattern, which is obtained by inputting the multiple call frequencies into the activity pattern recognizer within the cloud server.

4. The multimodal access control method for IoT smart home appliances according to claim 3, characterized in that, The configuration steps for the activity pattern recognizer include: Based on the user's home appliance call data over a historical period, multiple sample call frequency sets are collected, and the user's activity patterns under different sample call frequency sets are labeled to obtain a sample activity pattern set. Based on machine learning, an activity pattern recognizer is constructed and trained using the multiple sample call frequency sets and sample activity pattern sets. Configure the activity pattern recognizer on the cloud server.

5. The multimodal access control method for IoT smart home appliances according to claim 1, characterized in that, Calculate the similarity between multiple call parameter sequences and standard call parameter sequences of multiple home appliances under the identified activity pattern to obtain pattern confidence, including: Obtain multiple standard call frequencies of multiple home appliances under the identified activity mode; Multiple call frequencies are calculated based on multiple call parameter sequences; Calculate the similarity between the multiple call frequencies and the multiple standard call frequencies to obtain the pattern confidence score.

6. A multimodal access control system for IoT smart home appliances, characterized in that, The system is used to execute the multimodal access control method for IoT smart home appliances as described in any one of claims 1-5, the system comprising: The parameter sequence acquisition module is used to acquire the call parameter sequence of multiple home appliances controlled by IoT within a preset time range in the past, wherein the multiple home appliances are connected to IoT through a first access method or a second access method. The activity pattern recognition module is used to identify the user's activity pattern based on multiple call parameter sequences and obtain the identified activity pattern. The pattern confidence calculation module is used to calculate the similarity between multiple call parameter sequences and the standard call parameter sequences of multiple home appliances under the identified activity mode, and obtain the pattern confidence. The standard call parameter sequences are the standard call frequencies of multiple home appliances under the activity mode, which are preset based on historical data. The access method optimization module is used to optimize the access methods of multiple home appliances based on the pattern confidence level, thereby obtaining an optimized access control scheme that includes multiple optimized access methods. The optimization is performed by configuring an optimization step size based on the pattern confidence level, including: Based on the pattern confidence, the pattern bias is calculated and used as the optimization step size. The number of steps is determined by combining the number of multiple home appliances. Use the current connection methods of multiple home appliances as the initial access control scheme; The access methods of the number of home appliances randomly selected within the initial access control scheme are adjusted to obtain an adjusted access control scheme. Based on the multiple calling frequencies of multiple home appliances within multiple calling parameter sequences, the access adaptability of the adjusted access control scheme is calculated, including: Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the unit energy consumption of each home appliance under different access methods, the adjustment access energy consumption is calculated. Based on the adjustment access methods of multiple home appliances within the aforementioned adjustment access control scheme, and combined with the interaction latency under different access methods for each home appliance, the adjustment access latency is calculated. Based on the mode deviation, adjusted access energy consumption, and adjusted access latency, the access fitness of the adjusted access control scheme is calculated, including: The mode deviation is used as the energy consumption weight, and the time delay weight is calculated. Obtain the standard access energy consumption and standard access latency of the identified activity pattern; The access fitness is obtained by weighting the ratio of standard access energy consumption to adjusted access energy consumption and the ratio of standard access latency to adjusted access latency using the energy consumption weight and latency weight. Continue to iteratively optimize the access control scheme according to the stated step size to obtain the optimized access control scheme corresponding to the maximum converged access fitness value.

Images