An adaptive intelligent door lock system based on biometrics and behavior analysis

Abstract

The present invention provides an adaptive smart door lock system based on biometrics and behavioral analysis. The system includes a biometric module, a behavioral analysis module, an adaptive learning module and a control module. The biometric module is used to collect and compare the user's biometric data, such as fingerprint, facial recognition, iris scanning or voiceprint recognition. The behavioral analysis module records the user's door lock usage frequency, time, movement trajectory and strength and other behavioral interaction data, and calculates the user's behavioral pattern feature value. The adaptive learning module analyzes historical behavioral data, adjusts the biometric template and behavioral pattern standards, and enhances the system's adaptability to changes in user characteristics. The control module is responsible for comprehensively processing the biometric verification signal and the behavioral pattern verification signal, and determines whether to issue an unlocking command or trigger a security alarm. By combining biometrics and behavioral analysis, the present invention provides an efficient and flexible intelligent door lock system, which significantly improves security and user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of smart door locks, and in particular to an adaptive smart door lock system based on biometrics and behavior analysis. Background Art

[0002] Existing smart door lock systems usually rely on a single biometric technology, such as fingerprint recognition, facial recognition or voiceprint recognition, to verify the user's identity. With the continuous development of technology, more and more smart door lock systems are beginning to integrate multiple biometric methods, improving the accuracy and convenience of user identity verification.

[0003] However, the biometric systems in the prior art face several major problems. First, the fixed nature of the biometric template makes it difficult for the system to cope with subtle changes in the user's biological characteristics, resulting in recognition failures or misjudgments. In addition, smart door lock systems that rely solely on biometrics have certain limitations when dealing with abnormal user behaviors or complex usage scenarios, lacking flexibility and multi-level verification mechanisms. Smart door locks in the prior art are usually unable to make adaptive adjustments based on the user's usage habits or behavior patterns, reducing overall security and user experience.

[0004] Therefore, it is necessary to develop a new intelligent door lock system. Summary of the invention

[0005] The present application provides an adaptive smart door lock system based on biometrics and behavioral analysis to improve the security and user experience of the door lock.

[0006] The present application provides an adaptive smart door lock system based on biometrics and behavior analysis, including:

[0007] A biometric module, for collecting biometric data of a user, wherein the biometric data includes fingerprint, facial recognition, iris scan or voiceprint recognition; comparing the collected biometric data with a pre-stored user biometric template; and outputting a biometric verification pass signal to the control module when the comparison result meets a preset matching threshold;

[0008] A behavior analysis module is used to record the current behavior interaction data between the user and the door lock, wherein the behavior interaction data includes the user's door lock usage frequency, time, motion trajectory data approaching the door lock, and force data of the unlocking operation; based on the recorded current behavior interaction data, calculate the user's current behavior pattern feature value; compare the calculated current behavior pattern feature value with the preset behavior pattern standard feature value; when the comparison result meets the preset comparison threshold, output a behavior pattern verification pass signal to the control module;

[0009] An adaptive learning module, communicating with the biometric recognition module and the behavior analysis module, is used to dynamically adjust the user's biometric template according to the user's multiple biometric recognition operations to adapt to the subtle changes in the user's biometric characteristics; dynamically adjust the user's behavior pattern standard feature value according to the user's behavior interaction data; and adjust the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data;

[0010] The control module communicates with the biometric module and the behavior analysis module, and is used to receive a biometric verification pass signal from the biometric module; receive a behavior pattern verification pass signal from the behavior analysis module; if a biometric verification pass signal and a behavior pattern verification pass signal are received, send an unlocking command to the door lock actuator, otherwise, keep the door lock locked or trigger a security alarm.

[0011] Furthermore, the adaptive learning module adjusts the unlocking conditions or sensitivity of the door lock based on the user's historical behavioral interaction data, including:

[0012] Calculate the statistical characteristics of user historical behavior interaction data, including average, standard deviation, peak and valley values;

[0013] Based on the statistical features, dynamically adjust at least one of the following parameters:

[0014] Matching threshold of the biometric module;

[0015] Comparison thresholds for the behavior analysis module;

[0016] The unlocking decision weight of the control module is used to balance the importance of biometric verification and behavioral pattern verification in the unlocking decision.

[0017] Furthermore, the adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavioral interaction data, and also includes:

[0018] Establish a normal range model for user behavior. When a behavior that significantly deviates from the model is detected, increase the system's vigilance and require additional verification steps. Depending on the degree of deviation and the current security level, select one or more of the following additional verification steps to perform:

[0019] (a) Ask users to answer pre-set identity verification questions;

[0020] (b) sending a verification code to the user's pre-registered mobile device and requiring the user to enter the verification code;

[0021] (c) Activate the real-time video call function and require the user to confirm their identity via video;

[0022] (d) reduce the tolerance of biometric matching thresholds and require higher matching accuracy;

[0023] (e) Triggering secondary biometrics, requiring the user to provide two or more biometric data;

[0024] (f) Start geolocation verification to confirm whether the user’s current location is as expected;

[0025] (g) Require users to perform additional behavioral verification through pre-set gestures or action sequences.

[0026] Furthermore, the adaptive learning module is also used for:

[0027] Based on the door lock usage frequency and time period provided by the user's historical behavioral interaction data, different security modes are automatically switched, wherein the security mode includes a high security mode, a regular security mode and a low security mode, wherein the high security mode requires at least one additional verification step to be performed on the basis of satisfying biometric verification and behavioral pattern verification; the regular security mode requires satisfying biometric verification and behavioral pattern verification; the low security mode requires satisfying one of the biometric verification and behavioral pattern verification requirements of the regular security mode.

[0028] Furthermore, the adaptive smart door lock system also includes a security policy module, which is used to pre-configure verification strategies under different security levels according to user-set preferences; automatically select additional verification steps according to the current security level when abnormal behavior is detected; record the results of each additional verification and user feedback for optimizing future verification strategies; and allow authorized users to remotely adjust security strategies and verification step priorities through mobile applications.

[0029] Furthermore, the biometric module includes a liveness detection unit for detecting whether the user is alive when collecting biometric data to prevent fraudulent unlocking.

[0030] Furthermore, the behavior analysis module includes a three-dimensional space perception sensor for capturing the three-dimensional motion trajectory and posture information of the user when approaching the door lock.

[0031] Furthermore, the adaptive smart door lock system also includes an environmental perception module for detecting environmental parameters around the door lock, including light intensity, temperature, humidity and noise level; the adaptive learning module dynamically adjusts the working parameters of the biometric module and the behavior analysis module based on the detected environmental parameters to adapt to different environmental conditions.

[0032] Furthermore, the adaptive smart door lock system also includes an energy management module for optimizing energy consumption according to the user's usage habits and the system's working status; when it is detected that it has not been used for a long time or is in a low-risk time period, the system enters a low-power standby mode to extend battery life.

[0033] Furthermore, the adaptive smart door lock system also includes a remote communication module, which is connected to the user's mobile terminal or cloud server via a wireless network. The user can monitor the door lock status in real time, receive abnormal alarm information, and remotely adjust the system's security settings or unlocking conditions.

[0034] The beneficial effects of the present application mainly include: (1) Through the combination of the biometric module and the behavior analysis module, identity authentication is not only based on the user's biometrics, but also double verification is performed by analyzing the user's behavior pattern. When the user's biometrics and behavior patterns match at the same time, the unlocking operation is allowed, which effectively reduces potential safety hazards. (2) The adaptive learning module can dynamically adjust the biometric template and behavior pattern standards based on the user's multiple biometric operations. In this way, the verification conditions can be automatically updated with subtle changes in the user's biometrics (such as fingerprint wear, facial feature changes, etc.) and usage behavior changes, reducing the possibility of recognition failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of an adaptive smart door lock system based on biometrics and behavioral analysis provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0036] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0037] The first embodiment of the present application provides an adaptive smart door lock system based on biometrics and behavior analysis. Figure 1 , which is a schematic diagram of the first embodiment of the present application. Figure 1 The first embodiment of the present application provides an adaptive smart door lock system based on biometrics and behavior analysis for detailed description.

[0038] The adaptive smart door lock system based on biometrics and behavior analysis includes a biometrics module 101 , a behavior analysis module 102 , an adaptive learning module 103 and a control module 104 .

[0039] The biometric module 101 is used to collect the user's biometric data, wherein the biometric data includes fingerprint, facial recognition, iris scanning or voiceprint recognition; compare the collected biometric data with a pre-stored user biometric feature template; when the comparison result meets a preset matching threshold, output a biometric verification pass signal to the control module.

[0040] The biometric module 101 is one of the core components of the adaptive smart door lock system, which is used to accurately verify the identity of the user. Its functions include collecting and processing a variety of biometric data, and specific data types include but are not limited to fingerprints, facial recognition, iris scanning and voiceprint recognition. The biometric module 101 collects these biometric features through embedded sensors, and the sensors can be configured according to different application scenarios. For example, for fingerprint recognition, the system can use capacitive or optical sensors to obtain the user's fingerprint image and generate a corresponding digital template; for facial recognition, the module 101 can capture the user's facial features through a camera, combine deep learning or image processing algorithms, extract key facial feature points (such as eyes, nose, lips, etc.), and compare them with the stored facial feature templates. Iris scanning can obtain the user's iris features through near-infrared imaging technology, while voiceprint recognition is based on the voice data collected by the microphone, and generates the user's voiceprint template through a voiceprint feature extraction algorithm.

[0041] The collected biometric data will be preprocessed to remove noise and enhance the quality of the image or signal to ensure the accuracy of subsequent comparison. For fingerprint data, preprocessing may include edge enhancement, image smoothing and binarization to improve the resolution of fingerprint details; for facial recognition data, the system may apply illumination compensation and geometric normalization to cope with different lighting conditions and facial posture changes. The processing of iris and voiceprint data will also be optimized according to their respective characteristics to ensure that the data quality is sufficient to support high-precision recognition.

[0042] The biometric module 101 compares the pre-processed biometric data with the user's biometric template pre-stored in the system. The comparison process usually uses a pattern matching algorithm, such as Euclidean distance, cosine similarity, or a neural network-based deep learning algorithm, depending on the type of biometric data collected. For fingerprint recognition, the system may use a minutiae matching algorithm to match the user's fingerprint image with minutiae points in the template (such as bifurcation points and termination points); for facial recognition, the system may compare based on the similarity of facial feature vectors to ensure that the user's identity matches the stored template to a preset threshold.

[0043] During the comparison process, the biometric module 101 will output a verification signal based on the matching result. When the matching result meets the preset matching threshold, the module 101 outputs a "biometric verification passed" signal to the control module 104, indicating that the user's identity verification is successful; if the comparison result is lower than the threshold, the system will not output a pass signal, and the user needs to re-identify or the system triggers other security measures.

[0044] The design of the biometric module 101 fully considers the scalability and adaptability of the system. Different biometric technologies can be used in different application scenarios or a combination of multiple identification methods to improve overall security. For example, fingerprint recognition and facial recognition can be used in combination. In some high-security situations, the system may require both the user's fingerprint and facial recognition to pass before the unlocking operation can be triggered. In addition, the module can dynamically adapt to subtle changes in the user's biometrics. For example, over time, the user's fingerprint may be worn or the facial features may change slightly. The module 101 maintains a high recognition accuracy by adaptively adjusting the biometric template.

[0045] The module is also highly scalable and compatible with future biometric technologies, such as palm print recognition, gait recognition and other emerging technologies, ensuring that the system can maintain efficient identity authentication functions in the long term.

[0046] Furthermore, the biometric module includes a liveness detection unit for detecting whether the user is alive when collecting biometric data to prevent fraudulent unlocking.

[0047] Specifically, the liveness detection unit performs liveness detection before or simultaneously with biometrics, using various sensor technologies to determine whether the user is truly live. For fingerprint recognition, the system can use optical sensors, capacitive sensors, or ultrasonic sensors to detect fingerprint characteristics, such as sweat gland activity or the capacitance characteristics of the skin. These sensors can distinguish between live fingerprints and forged fingerprints, preventing the use of fake fingerprints made of silicone or other materials to unlock the phone.

[0048] In facial recognition, liveness detection can be achieved in a variety of ways. For example, the system can use infrared sensors to detect temperature changes, blood flow, or skin reflectivity characteristics of the user's face. In addition, the system can also require the user to perform specific actions during the recognition process, such as blinking, turning the head, etc., to ensure that the object of facial recognition is a living person, rather than a static photo or video forgery. These detection methods can effectively improve the security of facial recognition and prevent fraudulent unlocking through photos or videos on the screen.

[0049] For iris recognition, the liveness detection unit can determine whether the object to be recognized is alive by detecting subtle changes in the reflection of the iris or the natural movement of the eye (such as the contraction or expansion of the pupil). These biological features are difficult to accurately simulate in static images or artificial imitations, ensuring that the object recognized by the system is a real user.

[0050] During voiceprint recognition, the liveness detection unit can determine whether the voice is live by analyzing the frequency of the user's voice, the natural vibration of the sound waves, and the subtle changes in the airflow during the voice. In addition, the system can ask the user to say a randomly generated phrase or a specific command to prevent fraudulent unlocking by recording or simulating voices.

[0051] Through the liveness detection unit, the system can add an extra layer of security during the identification process to prevent potential fraudulent unlocking attempts. This detection is imperceptible and does not require additional user operation, which can greatly improve the security of the system without affecting the user experience. This module design ensures that only real and present users can successfully unlock through biometric identification, avoiding illegal unlocking attempts using forged biometric features.

[0052] The implementation of the liveness detection unit can be based on a variety of existing sensor technologies and algorithms, and the specific technology selection can be adjusted according to different biometric types and system requirements.

[0053] The behavior analysis module 102 is used to record the current behavior interaction data between the user and the door lock, wherein the behavior interaction data includes the user's door lock usage frequency, time, motion trajectory data approaching the door lock, and force data of the unlocking operation; based on the recorded current behavior interaction data, calculate the user's current behavior pattern feature value; compare the calculated current behavior pattern feature value with the preset behavior pattern standard feature value; when the comparison result meets the preset comparison threshold, output a behavior pattern verification pass signal to the control module.

[0054] The behavior analysis module 102 is one of the key components of the adaptive smart door lock system, which is specifically used to collect and analyze the user's behavior interaction data to assist the biometric recognition module in providing a double verification mechanism to improve overall security and adaptability. The behavior analysis module collects the user's behavior data when interacting with the door lock in real time through a variety of sensors. These data include but are not limited to the user's door lock usage frequency, unlocking time, movement trajectory when approaching the door lock, and the strength of the unlocking operation.

[0055] Regarding the frequency of door lock use by the user, the behavior analysis module can record the number of times the user uses the door lock every day or every week. By counting these frequency data, the system can identify the user's usage habits, such as whether the user uses the door lock at a fixed time every day or whether there is a significant frequency change. If the frequency suddenly increases or decreases, the module 102 will mark these abnormal behaviors as potential security threats.

[0056] For the recording of unlocking time, module 102 can accurately record the time spent by the user each time the door is unlocked. This data usually includes the time period from the user approaching the door lock to the completion of the entire unlocking process. The system can establish a characteristic pattern of the user's unlocking time by analyzing the user's historical data. If the current unlocking time is significantly shorter or longer than the average unlocking time of the user, the system can determine whether the current behavior is abnormal. For example, if the user usually completes the unlocking operation within 5 seconds, and the current operation exceeds 15 seconds, the system may think that the user has encountered difficulties or that an unexpected user has attempted the operation.

[0057] The behavior analysis module can also record the movement trajectory of the user approaching the door lock. By using a distance sensor or an acceleration sensor, the module can monitor the movement path and speed of the user from approaching the door lock to performing the unlocking operation. For example, module 102 can record whether the user regularly accelerates or decelerates when approaching the door lock, or whether there is a sudden change of direction during the approach process. These movement patterns are stored by the system as user-specific behavioral features. If the current movement trajectory deviates significantly from the user's historical trajectory, such as approaching the door lock at an abnormally fast speed or the movement path changes significantly, the system will determine that the current behavior may be abnormal and trigger further security measures.

[0058] The force data of the unlocking operation is also collected by the module 102 through the force sensor installed on the door lock. The system can analyze the user's operation characteristic pattern according to the change in the force applied by the user in each unlocking operation. The unlocking force of each user may be different. The module 102 forms the force characteristic value range of the user by recording the force characteristics of multiple unlocking operations. If the force of a certain unlocking operation deviates from the range, the system will think that the operation may not be performed by the user himself, and then judge it as abnormal behavior.

[0059] After the data collection is completed, the behavior analysis module will calculate and analyze these behavior interaction data. First, the system will calculate the user's behavior pattern feature value based on the current behavior data collected in real time. For example, the user's unlocking time, operation force, movement trajectory and other behavior data will be integrated into a feature vector, which represents the current user's operation characteristics. Subsequently, the system will compare the feature vector with the preset standard feature value of the behavior pattern. The preset standard feature value is usually generated by statistics and analysis of the user's historical behavior data. For example, the system can calculate statistical parameters such as the mean and standard deviation based on the user's behavior data for the past month as a benchmark for judging whether it is abnormal.

[0060] The comparison process is usually implemented using a variety of algorithms, such as Euclidean distance, cosine similarity, or a classification algorithm based on machine learning. If the difference between the current behavior feature and the standard feature value is within a preset threshold range, the behavior analysis module will output a signal to the control module 104 indicating that the behavior pattern has been verified. If the behavior feature deviates too much from the standard feature value, the system will consider the current operation to be abnormal and may refuse to unlock or trigger other safety measures.

[0061] The flexibility of the behavior analysis module is reflected in its ability to continuously make adaptive adjustments based on the user's usage behavior. By accumulating and analyzing the user's multiple behavioral data, the module can dynamically adjust the user's behavior pattern standards. For example, if the user's behavioral characteristics change slightly in a specific season or different time period (such as wearing heavy clothes in winter, which increases the operating force), the system can automatically adapt to these changes to avoid misjudgments due to normal behavioral deviations. In addition, the behavior analysis module can also work in conjunction with other system functions to perform multi-modal data fusion, such as working together with the biometric module to combine the user's biological characteristics and behavioral characteristics in certain situations to further improve the accuracy and security of the system.

[0062] The behavior analysis module can also cooperate with other smart devices or systems, such as smart phones and smart watches, to collect user behavior feature data using the built-in accelerometers, gyroscopes, GPS, etc. in these devices, further enhancing the richness and accuracy of the data. In home automation systems, the behavior analysis module can even be combined with indoor surveillance cameras and other sensors to fully understand user behavior patterns and achieve more intelligent security management.

[0063] In short, the behavior analysis module 102 greatly improves the security and user experience of the smart door lock system through comprehensive analysis and adaptive adjustment of user behavior. Based on the combination of biometric technology, the behavior analysis module provides more multi-level security protection by monitoring the user's daily behavior.

[0064] Furthermore, the behavior analysis module includes a three-dimensional space perception sensor for capturing the three-dimensional motion trajectory and posture information of the user when approaching the door lock.

[0065] Three-dimensional space perception sensors can be implemented through a variety of technologies, such as infrared sensors, LiDAR, or optical-based depth cameras. Through these sensors, the system can capture the user's movement trajectory in three-dimensional space as the user approaches the door lock. Specifically, the sensor can sense the changes in the user's path, speed, and direction when approaching the door lock from a long distance to a close distance, and record and analyze this data as one of the user's behavioral characteristics.

[0066] The user's three-dimensional motion trajectory includes the user's walking route, acceleration or deceleration behavior when approaching the door lock, the body's tilt angle, the direction of the head, etc. This information helps the system determine whether the user's behavior is consistent with their regular unlocking mode. For example, the user may be accustomed to approaching the door lock from a fixed direction and at a certain speed. If the system detects that the user's motion trajectory has deviated significantly, such as approaching the door lock quickly from an unusual direction, it may be considered that the behavior is abnormal, thereby triggering additional verification steps to ensure the safety of the system.

[0067] In addition to motion trajectories, the 3D space perception sensor can also capture the user's posture information. For example, the system can determine whether it is a normal unlocking behavior by sensing the changes in the user's body posture. If the user's body posture when approaching the door lock is significantly different from his normal pattern, such as excessive bending, squatting, or moving too fast, these abnormal posture information will be recorded as potential risk signals. By analyzing this data, the system can further determine whether the current operation is performed by a legitimate user.

[0068] This 3D motion trajectory and gesture capture enables the system to obtain the user's dynamic behavior data in real time, and combined with other static biometric information (such as fingerprint or facial recognition), it provides double or multiple security protection. In this way, the system can effectively prevent abnormal behavior or forced entry attempts, such as someone trying to quickly approach and operate the door lock, or trying to bypass the biometric mechanism with an abnormal posture.

[0069] To ensure the efficient operation of the system, the 3D spatial perception sensor is closely integrated with the behavior analysis module, which can process the captured spatial data in real time and compare it with the user's historical behavior patterns. This process is automated, and users can enjoy a higher level of security without additional operations. At the same time, the system will gradually adjust and optimize the recognition criteria for different user behaviors through the adaptive learning module to ensure that the system can accurately adapt to the behavior habits of individual users.

[0070] The adaptive learning module 103 communicates with the biometric module and the behavior analysis module, and is used to dynamically adjust the user's biometric template according to the user's multiple biometric operations to adapt to subtle changes in the user's biometrics; dynamically adjust the user's behavior pattern standard feature value according to the user's behavior interaction data; and adjust the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data.

[0071] The adaptive learning module 103 is an important component for realizing dynamic adjustment and adaptive optimization in the adaptive smart door lock system. Its core function is to continuously analyze the user's multiple biometric identification and behavioral interaction data, dynamically adjust the biometric template and behavioral pattern feature values, and ensure that the system can adapt to the subtle changes in the user's biometric characteristics and the evolution of behavioral habits, thereby improving the accuracy and security of identification.

[0072] First, regarding the dynamic adjustment of the biometric template, the adaptive learning module 103 works in conjunction with the biometric module 101. The user's biometric features, such as fingerprints, facial features, irises, or voiceprints, may change slightly over time. For example, fingerprints become blurred due to skin wear, facial features may change slightly due to age, expression, or other factors, and even voiceprints may vary due to changes in mood or physical condition. The adaptive learning module can dynamically adjust the user's biometric template to adapt to these changes through data changes in multiple recognition operations. Specifically, after each successful biometric operation, the module updates the existing biometric template, merges the latest feature data with the previous data, and generates a more accurate template. In this way, the system can not only improve the accuracy of the current recognition, but also reduce recognition failures caused by changes in microbial features, and enhance the adaptability of the system.

[0073] In terms of adaptive adjustment of behavior patterns, the adaptive learning module works closely with the behavior analysis module 102 to analyze the user's behavior interaction data in real time. The user's behavior patterns when using door locks, such as frequency of use, unlocking time, trajectory of approaching the door lock, and unlocking force, all have certain regularities. However, these rules may change with time and changes in the usage environment. The adaptive learning module can dynamically adjust the standard characteristic values ​​of the behavior pattern by analyzing multiple user behavior data. For example, users wearing heavy clothing in winter may cause the unlocking force to increase, or the frequency of door lock use on weekdays and weekends is different. The adaptive learning module will adjust the behavior standards according to the user's actual usage, so that the system can still effectively judge whether the user's behavior is normal under changing circumstances, and reduce the possibility of false alarms or misjudgments.

[0074] In addition, the adaptive learning module not only relies on current behavior and biometric data, but also performs more complex analysis based on historical behavior interaction data. Through long-term accumulated historical data, the module is able to identify potential changing trends in user behavior patterns. For example, if a user exhibits different behavior patterns during a certain time period (such as holidays or seasonal changes), the system can predict these changes and adjust the unlocking conditions or sensitivity in advance to adapt to these trends. With the support of historical data, the module can not only adapt to short-term changes in a timely manner, but also identify long-term trends and provide more personalized security.

[0075] Another important function of the adaptive learning module is to adjust the unlocking conditions and sensitivity of the door lock according to changes in user behavior. In some cases, the user's behavior pattern may temporarily deviate from its usual behavior characteristics due to changes in the external environment (such as weather, physical state, etc.). In order to prevent such short-term changes from causing false alarms or refusing to unlock, the adaptive learning module can adjust the sensitivity of the unlocking conditions based on the user's historical behavior data. For example, in some cases, the system can relax the matching threshold of the behavior pattern to allow a larger range of behavior changes, while in other cases, the system can improve security and tighten the unlocking conditions. In addition, the adaptive learning module can dynamically adjust the response speed or verification times of the door lock according to the user's behavior frequency or security level requirements. For normal operation mode, the system can reduce the verification steps to improve the user experience; when anomalies or potential risks are detected, the system will increase the complexity of security verification to ensure the safety of the door lock.

[0076] The design of the adaptive learning module also allows for expansion to different application scenarios. For example, in an enterprise environment, the user's behavioral characteristics may vary as job responsibilities or working hours change, and the module can dynamically adjust the unlocking conditions of the door lock according to the user's identity or role. For senior managers, the system can require more stringent biometric and behavioral pattern verification, while for ordinary employees, the system may relax the requirements to ensure a balance between security and convenience. In addition, in a home environment, the adaptive learning module can learn and distinguish the behavioral patterns of family members, thereby providing a personalized unlocking experience for each member.

[0077] The adaptive learning module can provide higher security and flexibility without interfering with the normal use of users through its efficient adaptive capabilities. Its dynamic adjustment mechanism based on multiple biometric and behavioral data of users can minimize the occurrence of misidentification and denial of service while preventing potential security risks. The functions of this module can be realized through machine learning techniques and algorithms, such as weighted averaging, pattern matching, time series analysis, and dynamic threshold adjustment. These technologies have a wide range of applications in the processing of behavioral data and biometric data, and can ensure the effective implementation of the adaptive learning module.

[0078] Furthermore, the adaptive learning module adjusts the unlocking conditions or sensitivity of the door lock based on the user's historical behavioral interaction data, including:

[0079] Calculate the statistical characteristics of user historical behavior interaction data, including average, standard deviation, peak and valley values;

[0080] Based on the statistical features, dynamically adjust at least one of the following parameters:

[0081] Matching threshold of the biometric module;

[0082] Comparison thresholds for the behavior analysis module;

[0083] The adaptive learning module in the adaptive smart door lock system adjusts the unlocking conditions or sensitivity of the door lock by analyzing the user's historical behavioral interaction data. The core task of this module is to identify the statistical characteristics of the user's past operation patterns, including the average, standard deviation, peak and valley values, and then dynamically adjust multiple key parameters of the system to adapt to changes in individual user behavior and improve the system's security and user experience.

[0084] First, the adaptive learning module needs to comprehensively collect and store the user's historical behavior data. This data comes from each interaction between the user and the door lock, including frequency of use, unlocking time, unlocking force, and movement trajectory approaching the door lock. This data is obtained through the behavior analysis module and accumulated in the adaptive learning module for subsequent analysis and calculation. Through a sufficiently long period of data accumulation, the system can generate a complete data set that reflects the user's operating habits, based on which the system calculates statistical features.

[0085] For these historical data, the system first calculates the average of user behavior. For example, how users usually approach door locks within a specific time period, or unlock the door within a certain speed range, these averages reflect the typical behavior patterns of users. In addition, the system also calculates the standard deviation to understand the volatility of user behavior, that is, the degree of deviation of user behavior from the average. For some users, the behavior pattern may be relatively stable, while for other users, the behavior fluctuates greatly, and the system can judge these changes through the standard deviation. The calculation of peaks and valleys helps to identify the occurrence of extreme situations in user operations, such as using too much or too little force when unlocking, or unlocking for too long or too short a time. By identifying these extreme situations, the system can better distinguish abnormal behavior.

[0086] Based on these statistical features, the adaptive learning module can dynamically adjust multiple key parameters of the door lock system. An important parameter is the matching threshold of the biometric module. Usually, the biometric module compares the user's current biometric features (such as fingerprints, facial recognition, irises, etc.) with the stored templates and determines whether the verification is passed based on the preset matching threshold. If the threshold is set too low, the system may allow some inaccurate matches, thereby reducing security; if it is set too high, the system may frequently reject the user's legitimate unlocking operations, affecting the user experience. Through adaptive learning, the system can automatically adjust the matching threshold based on the user's historical data. For example, for users whose biometric features may change slightly over time (such as fingerprint wear or slight changes in facial features), the system can appropriately lower the matching threshold to increase fault tolerance; while for users whose biometric features are relatively stable, the system can maintain a higher matching threshold to enhance security.

[0087] Another parameter that can be adjusted dynamically is the comparison threshold of the behavior analysis module. This threshold is used to determine whether the difference between the user's current behavior and his historical behavior pattern is within a reasonable range. The behavior analysis module generates a user's behavior pattern template by recording the user's unlocking time, strength, and way of approaching the door lock. During the unlocking process, the system determines whether the user's behavior is normal by comparing the current behavior with this template. The adaptive learning module can adjust this threshold based on the volatility of the user's historical behavior. If the user's behavior is relatively stable, the system can set a lower comparison threshold to enhance security; for users with large behavioral changes, the system can appropriately relax the threshold to reduce unnecessary unlocking failures.

[0088] Finally, the adaptive learning module can also adjust the unlocking decision weight of the control module. The control module is responsible for combining the results of biometric verification and behavioral pattern verification to decide whether to allow unlocking. The weight distribution of biometric verification and behavioral verification in unlocking decisions may vary for different users or different scenarios. In some cases, the accuracy of biometrics may be more important, while in other cases, the user's behavioral pattern may provide more valuable verification information. For example, in certain situations with a higher security level, the system can set a higher weight for biometric verification to ensure that only highly matched biometric features can unlock; while in some scenarios that focus more on convenience, the system can reduce the weight of biometrics and increase reliance on behavioral verification. This dynamic adjustment of weights can be adaptively adjusted according to the actual needs and security level of the user, so that the system can provide sufficient security while ensuring a balance between convenience and user experience.

[0089] Through the above steps, the adaptive learning module enables the door lock system to not only adapt to changes in user behavior and biometrics in real time, but also dynamically adjust the key parameters of the system and optimize the unlocking decision process. The implementation of this module relies on the long-term accumulation and statistical analysis of user data, combined with appropriate algorithms such as weighted average and standard deviation analysis, to ensure that the system can operate accurately and efficiently in different environments and scenarios. This module is highly flexible and scalable, and can be customized according to specific needs to meet the security needs and usage preferences of different users.

[0090] Furthermore, the adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavioral interaction data, and also includes:

[0091] Establish a normal range model for user behavior. When a behavior that significantly deviates from the model is detected, increase the system's vigilance and require additional verification steps. Depending on the degree of deviation and the current security level, select one or more of the following additional verification steps to perform:

[0092] (a) Ask users to answer pre-set identity verification questions;

[0093] (b) sending a verification code to the user's pre-registered mobile device and requiring the user to enter the verification code;

[0094] (c) Activate the real-time video call function and require the user to confirm their identity via video;

[0095] (d) reduce the tolerance of biometric matching thresholds and require higher matching accuracy;

[0096] (e) Triggering secondary biometrics, requiring the user to provide two or more biometric data;

[0097] (f) Start geolocation verification to confirm whether the user’s current location is as expected;

[0098] (g) Require users to perform additional behavioral verification through pre-set gestures or action sequences.

[0099] One of the core functions of this module is to establish a normal range model of user behavior, identify the user's typical behavior patterns, and when it is detected that the user's behavior deviates significantly from this model, the system will increase its vigilance and trigger additional security verification steps.

[0100] First, the adaptive learning module generates a user's "normal behavior range" model through long-term accumulation of user behavior data. This model usually includes common user behavior characteristics, such as unlocking time, speed when approaching the door lock, unlocking force, etc. Through these historical data, the system can identify the user's behavior pattern and set a reasonable fluctuation range, allowing the user's daily behavior to vary within this range. Any behavior that obviously deviates from this normal range will be marked as abnormal, triggering further security verification.

[0101] When the system detects that user behavior deviates from the preset range of this model, the system's vigilance will increase, and it will decide what additional verification measures to take based on the degree of deviation and the current security level. The security level may be based on the external environment or user-defined security settings. For example, in the case of high security requirements, the system will more strictly detect behavioral deviations and trigger more verification steps. In scenarios with lower security levels, the system may only require simple additional verification to ensure convenience.

[0102] The specific additional verification steps are selected based on the severity of the abnormal behavior and the security requirements of the system, providing a series of flexible verification mechanisms. The first possible step is to ask the user to answer preset identity verification questions (step a). Such questions can be personal questions set by the user in advance, such as birth city, pet name, etc. The system will prompt the user to enter the answer. Since these questions are personalized, they can effectively verify the true identity of the user.

[0103] Another optional verification step is to send a verification code to the user's pre-registered mobile device (step b) and ask the user to enter the verification code in the system. This dual verification method is usually performed via SMS or app push, which can ensure that the current operation is initiated by the user himself. The effectiveness of verification code verification lies in its timeliness and uniqueness, which can effectively prevent unauthorized unlocking.

[0104] In cases where security requirements are higher, the system can also choose to start the real-time video call function (step c) and require the user to confirm his identity through a video call. Through the camera, the system can perform facial recognition and real-time interaction to verify the user's identity. This method is suitable for scenarios that require high security, such as important areas of a company or places where valuables are stored in the home.

[0105] In addition, the system can also choose to reduce the tolerance of the biometric matching threshold (step d) to increase the accuracy of the verification of the user's biometric features. In normal operation, the system may allow a certain range of matching errors to improve the user experience, and when abnormal behavior is detected, the system will lower the tolerance rate to ensure a higher accuracy biometric match.

[0106] In more complex cases, the system can also choose to trigger secondary biometrics (step e), requiring users to provide two or more biometric data. This means that users not only need to enter fingerprints, but may also need to provide additional verification methods such as facial recognition or voiceprint recognition to ensure multiple protections of the system at a high security level.

[0107] The system can also perform geolocation verification (step f), by detecting the GPS data or other positioning information of the user's mobile phone to verify whether the user's current actual location is consistent with the expected. This verification method can prevent remote attacks or illegal operations in unexpected locations. For example, when the system detects that the user is trying to unlock the door in another city, it can immediately refuse the operation or require additional verification.

[0108] Finally, the system can ask the user to verify through a preset gesture or action sequence (step g). The system may ask the user to perform a series of specific gestures through a touch screen or motion sensor, similar to the gesture password to unlock the phone screen. This verification method is relatively simple and suitable for scenarios with medium security requirements.

[0109] Overall, these verification steps provide different levels of security for the system, and can select the appropriate verification method according to the degree of deviation detected and the current security requirements. This flexible security strategy enables the adaptive smart door lock system to provide high security and flexibility in various scenarios, which can effectively deal with high-risk situations and provide a convenient user experience in daily use.

[0110] Furthermore, the adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data, specifically including:

[0111] A Gaussian mixture model is established to fit the probability distribution of user behavior interaction data. The probability distribution function P(x) of the Gaussian mixture model adopts the following formula 1:

[0112]

[0113] Among them, x represents the user's behavior interaction data, and the user's behavior interaction data x=[x1,x2,…x i ,…,x n ], where x i represents the behavior interaction data for the i-th dimension, n represents the total number of dimensions of the behavior interaction data; π k (x) is the weight function of the kth Gaussian distribution; represents the probability distribution function of the kth Gaussian distribution, μ k is the mean of the kth Gaussian distribution; Σ k is the covariance matrix of the kth Gaussian distribution; K represents the total number of dimensions of the behavioral interaction data; where the weight function π k (x) is calculated according to the following formula 2:

[0114]

[0115] Among them, α is the sensitivity coefficient used to control dynamic adjustment; D k (x) is the mean μ of the user's behavior interaction data x and the kth Gaussian distribution k The Mahalanobis distance between k The calculation formula of (x) is shown in the following formula 3:

[0116]

[0117] Estimate the parameters μ of the Gaussian mixture model using the expectation-maximization algorithm k ,Σ k ;

[0118] According to the estimated parameters of the Gaussian mixture model, the current behavior interaction data x is calculated using formula 1 new The probability P(x new ); when the probability P(x new ) is less than the preset threshold T P When the current behavior deviates from the normal range;

[0119] The predicted value of the user's behavior interaction data at the current time t is obtained through the autoregressive time series model provided by the following formula 4

[0120]

[0121] where φ i is the i-th autoregressive coefficient; ∈ t is the noise term; p represents the lag order of historical data; x t―irepresents the actual value of the behavioral interaction data obtained at time t-i;

[0122] According to the following formula 5, the predicted value is calculated The actual value x of the behavior interaction data at the current time t t The difference between the data Δx:

[0123]

[0124] When P(x new ) <T P And Δx>T AR When the system determines that the user's behavior is abnormal, it performs additional verification, including:

[0125] When T P1 <P(x new )≤T P2 When , perform additional verification steps a or b;

[0126] When T P2 <P(x new )≤T P3 When , perform additional verification steps c or d;

[0127] When T P3 <P(x new )≤T P , perform additional verification steps e, f, or g.

[0128] Among them, T P , T P1 , T P2 and T P3 is a pre-determined threshold.

[0129] The adaptive learning module of the adaptive smart door lock system is based on the user's historical behavioral interaction data. It builds a Gaussian mixture model to fit the user's behavioral data distribution, thereby dynamically adjusting the unlocking conditions and sensitivity of the door lock. The system determines whether the current user behavior conforms to the historical behavior pattern through multi-dimensional data analysis and processing, and triggers additional security verification in abnormal situations.

[0130] The first step of the adaptive learning module is to build a Gaussian mixture model (GMM) based on the user's historical behavior interaction data to represent the probability distribution of the user's behavior pattern. The user's behavior interaction data x includes data of multiple dimensions, such as the frequency, time, force, and movement trajectory when approaching the door lock. Specifically, the user behavior interaction data x is represented as a vector x = [x1, x2, ... x i ,…,x n ], where x iis the i-th dimension of the behavioral interaction data, and n represents the total number of data dimensions. By processing multidimensional data, the system can capture complex patterns of user behavior.

[0131] The probability distribution function P(x) of the Gaussian mixture model is used to describe the distribution of these behavioral data, which is specifically expressed by the following formula:

[0132]

[0133] P(x) in formula (1) is the overall probability distribution function of user behavior data x, which indicates the probability of user behavior occurring in a specific mode. Here K represents the number of different Gaussian components, each of which represents a potential behavior pattern. For each k component, π k (x) represents the weight of the kth Gaussian distribution, reflecting the importance of the behavior pattern in the current behavior. The weight is adjusted dynamically and calculated by formula (2). The mean μ of each Gaussian distribution is k represents the center point of the kth behavior mode, and the covariance matrix Σ k It reflects the data distribution in the model, that is, the relationship and distribution range between different dimensions in the model.

[0134] Next, the system uses formula (2) to dynamically calculate the weight function π k (x):

[0135]

[0136] Among them, α is a sensitivity control parameter used to control the dynamic adjustment of weights. k (x) is calculated based on the distance between the user’s current behavior and the center of each Gaussian component. This distance is calculated using the Mahalanobis distance D k (x) is measured and is calculated by formula (3):

[0137]

[0138] Mahalanobis distance D k (x) measures the user's current behavior x and the center μ of the kth Gaussian distribution k The distance between them, while taking into account the covariance Σ between the dimensions of the data k Unlike the Euclidean distance, the Mahalanobis distance can effectively consider the correlation between different dimensions, allowing the system to more accurately measure the deviation between the behavior pattern and the current data. The smaller the distance, the closer the current behavior is to the specific pattern, so the weight of the pattern is π k (x) will be larger.

[0139] Through the expectation maximization algorithm (EM algorithm), the adaptive learning module can estimate the mean μ of each Gaussian distribution k and the covariance matrix Σ k , in order to fit the user's behavior distribution. Based on these parameters, the system can use formula (1) to calculate the current user behavior data x new The probability P(x new ). If the calculated probability P(x new ) is less than the preset threshold T P , it means that the user's current behavior has obviously deviated from the normal range, and the system determines that the current behavior may be abnormal.

[0140] On the basis of anomaly detection, the system also combines the autoregressive time series model to predict the user's future behavior. Specifically, the predicted value of the user's behavior interaction data at the current time t is calculated using formula (4):

[0141]

[0142] In formula (4), φ i is the coefficient of the autoregressive model, which is used to capture the impact of historical behavior data on current behavior. t―i represents the behavioral interaction data recorded at time t-i, and the lag order p represents the time length of the historical data considered by the system. The noise term ∈ t It is used to represent random disturbances or errors that cannot be explained by the model. Through the autoregressive model, the system can predict the user's future behavior and compare it with the actual behavior.

[0143] To measure the difference between predicted behavior and actual behavior, the system calculates the behavior difference Δx using formula (5):

[0144]

[0145] In formula (5), Δx represents the predicted behavior of the user at the current time t and actual behavior x t By calculating the difference between the two, the system can determine whether the current behavior is normal. If Δx exceeds the preset difference threshold T AR , then the behavior is abnormal.

[0146] When P(x new ) <T P And Δx>T ARWhen the system determines that the user's behavior is abnormal, it will perform additional verification steps. Based on the degree of current behavioral deviation, the system selects appropriate verification steps for further identity confirmation. The system's verification steps include: asking the user to answer preset identity verification questions, sending a verification code to the user's mobile device, and starting the video call function. These steps provide different levels of verification depending on the severity of the behavioral deviation.

[0147] Ultimately, through this series of complex calculations and verification mechanisms, the system can accurately detect and handle abnormal user behavior, ensuring a balance between security and convenience.

[0148] Furthermore, the adaptive learning module is also used for:

[0149] Based on the door lock usage frequency and time period provided by the user's historical behavioral interaction data, different security modes are automatically switched, wherein the security mode includes a high security mode, a regular security mode and a low security mode, wherein the high security mode requires at least one additional verification step to be performed on the basis of satisfying biometric verification and behavioral pattern verification; the regular security mode requires satisfying biometric verification and behavioral pattern verification; the low security mode requires satisfying one of the biometric verification and behavioral pattern verification requirements of the regular security mode.

[0150] First, the adaptive learning module analyzes the user's historical behavioral interaction data, including the frequency of door lock use and the time period of use. This data includes the number of times the user unlocks the door every day or week, the time pattern of using the door lock (such as morning, night, etc.), and the user's behavioral characteristics during these time periods. For example, a user may frequently use the door lock between 7 and 9 in the morning every day, while the number of times it is used at night is significantly reduced. Through the long-term accumulation of this data, the adaptive learning module can identify the user's typical usage patterns and automatically adjust the system's security settings based on these patterns.

[0151] The system switches to different security modes based on the user's frequency of use and behavioral characteristics. High security mode is mainly used in scenarios that require high security, such as periods when users rarely use door locks but have high security requirements. In this mode, the system not only requires biometric verification and behavioral pattern verification, but also adds an additional verification step. The choice of additional verification steps can include sending a verification code to the user's mobile phone, requiring the user to answer preset security questions, or even starting a real-time video call function to confirm the user's identity. In this mode, the unlocking conditions of the door lock are very strict, ensuring that only legitimate and multi-verified users can unlock the door.

[0152] Conventional security mode is suitable for scenarios where users use door locks on a daily basis, especially during time periods when user behavior patterns are relatively fixed and risks are low. In this mode, the system requires that the user's biometric verification and behavior pattern verification meet preset standards, such as through fingerprint or facial recognition, and the user's behavior characteristics (such as unlocking time and strength) must also meet the standard characteristics in their historical behavior data. In this mode, the system provides verification requirements that balance security and convenience, ensuring a smooth experience during daily use while maintaining sufficient security.

[0153] Low security mode is suitable for scenarios where users frequently use door locks and have high requirements for convenience. In this mode, the system can simplify the verification process and only require one of biometric verification or behavioral pattern verification. For example, if the user's fingerprint verification passes, the system may directly allow unlocking without further analyzing the behavioral pattern. This mode is suitable for use in low-risk environments, such as homes or offices that users frequently use, to ensure fast and convenient unlocking operations without affecting the user experience due to complex verification steps.

[0154] The adaptive learning module automatically switches between these different security modes based on the frequency and time period of door lock use. This switching is based on long-term analysis of user historical behavior rather than a simple fixed mode. For example, if the system detects that the user's usage frequency is low at night and the security needs are high, it can automatically set the night to high security mode, while during the day it may switch to regular security mode based on the user's frequent use. In low security mode, the system will also dynamically adjust according to real-time conditions. If abnormal behavior is detected or the frequency of door lock use suddenly increases, the system can instantly upgrade the security mode and switch to regular or high security mode to add additional verification steps.

[0155] In addition, the switching of these security modes is highly flexible. The adaptive learning module can not only automatically adjust the mode based on historical data, but also adjust it according to real-time data. For example, if the system detects that the user's behavior pattern suddenly deviates from the historical data (such as unusual door lock operation at night), it can temporarily switch to high-security mode and add additional verification steps to ensure security.

[0156] In short, the adaptive smart door lock system can provide multi-level security protection according to the actual usage of users and the needs of different time periods through the automatic switching security mode function of the adaptive learning module. This mode switching ensures that the system provides strong security protection during high-risk periods, while providing sufficient convenience in daily use to meet the needs of different users and scenarios.

[0157] Furthermore, the adaptive smart door lock system also includes a security policy module, which is used to pre-configure verification strategies under different security levels according to user-set preferences; automatically select additional verification steps according to the current security level when abnormal behavior is detected; record the results of each additional verification and user feedback for optimizing future verification strategies; and allow authorized users to remotely adjust security strategies and verification step priorities through mobile applications.

[0158] First, the security policy module allows users to pre-set verification strategies at different security levels according to their own security needs and living habits. Users can choose verification methods at high, medium, and low security levels through the system interface or mobile application. For example, at a high security level, users may require not only biometric and behavioral pattern verification every time they unlock the door, but also secondary verification, such as sending a verification code to the user's mobile phone or confirming the identity through a real-time video call. At a low security level, users may only need to complete fingerprint recognition to complete the unlocking operation. The system can pre-configure different verification combinations based on these settings, thereby achieving flexible and personalized security management.

[0159] When the system detects that the user's behavior does not match the preset behavior pattern, the security policy module will automatically evaluate the current security level and select additional verification steps based on the set policy. For example, in high-security mode, the system may trigger more complex verification steps, such as requiring the user to enter a verification code or answer security questions. If the system is in normal or low-security mode and the user's behavior deviation is small, the system may only need to add a simple biometric verification to complete the operation. This process is fully automated, ensuring that security policies can be adjusted quickly in different scenarios, which not only guarantees user experience but also provides adequate security protection.

[0160] In addition, the security policy module also has a recording function. After each additional verification, the system will record the verification results and user feedback. For example, the system can record whether the user has passed the additional verification, or whether the user expressed satisfaction or dissatisfaction with a certain verification method. Through these records, the system can continuously optimize future verification strategies. For example, if a user is troubled by repeated secondary verification failures at night, the system may automatically reduce the additional verification steps at night, or increase the weight of biometrics while ensuring safety, thereby reducing unnecessary trouble for users. This feedback mechanism enables the system to learn and adapt to user needs and continuously improve user experience.

[0161] The security policy module also allows authorized users to remotely adjust security policies and the priority of verification steps through a mobile application. Users do not need to operate directly on the door lock system, but can modify the system's security policy through the mobile application. For example, users can remotely set their home door locks to high security mode when they are on a business trip or vacation, or switch the door locks to low security mode when they need more convenience. In addition, users can also adjust the priority of different verification steps. For example, users can set biometrics as the preferred verification method and verification code verification as the secondary method to ensure that the system executes the optimal verification strategy in different occasions. This flexible remote management function not only improves the operability of the system, but also provides users with a high degree of security control.

[0162] In summary, the security policy module ensures that the adaptive smart door lock system can provide users with personalized security management in various scenarios by configuring verification policies according to user preferences, automatically selecting verification steps, recording and optimizing verification results, and remotely adjusting security policies. Through this module, the system can not only adapt to different security needs, but also continuously optimize based on user feedback to provide a more intelligent and flexible door lock management experience.

[0163] The control module 104 communicates with the biometric module and the behavior analysis module, and is used to receive a biometric verification pass signal from the biometric module; receive a behavior pattern verification pass signal from the behavior analysis module; if a biometric verification pass signal and a behavior pattern verification pass signal are received, send an unlocking command to the door lock actuator, otherwise, keep the door lock locked or trigger a security alarm.

[0164] The control module 104 is the core of the entire adaptive intelligent door lock system, responsible for receiving and processing verification signals from other modules, and making the final decision to unlock or keep the door locked based on these signals. It works closely with the biometric module 101, the behavior analysis module 102 and the adaptive learning module 103 to ensure that the door lock system can achieve flexible and efficient door lock management while ensuring safety.

[0165] First, the control module 104 receives a biometric verification pass signal from the biometric module 101. The biometric module is responsible for collecting the user's biometric data, such as fingerprints, facial features, irises or voiceprints, and comparing it with the pre-stored user biometric template. When the comparison result meets the preset matching threshold, the biometric module generates a "verification passed" signal and sends it to the control module. The control module immediately receives and stores the signal as one of the basis for the next decision.

[0166] At the same time, the control module 104 also communicates with the behavior analysis module 102 to receive a verification signal about the user's current behavior interaction. The behavior analysis module calculates the characteristic value of a user behavior pattern by recording the user's usage behavior (such as usage frequency, unlocking time, movement trajectory approaching the door lock, and unlocking force), and compares it with the behavior pattern standard preset in the system. If the user's current behavior matches its standard pattern, the behavior analysis module will generate a "behavior verification passed" signal and transmit it to the control module. The control module will receive signals from the behavior analysis module and the biometric recognition module at the same time as an important basis for comprehensively judging whether unlocking is allowed.

[0167] The main function of the control module is to make a comprehensive judgment based on these two verification signals. If the system receives both the biometric verification pass signal and the behavior pattern verification pass signal, the control module will identify the current user as a legitimate user and send an unlock command to the door lock actuator. The door lock actuator is responsible for the actual unlocking action. After receiving the command from the control module, it performs the corresponding mechanical operation to allow the user to enter. If either of the two signals fails, the control module will not issue an unlock command and keep the door lock in a locked state.

[0168] In addition to performing unlocking operations, the control module 104 also has a security protection function. In the event of a user identity verification failure, the system will not only keep the door locked, but may also trigger a further security alarm mechanism. For example, the control module can be configured to trigger an alarm device after multiple consecutive authentication failures, send a security alarm to the user's mobile device, or notify a preset security service system. This mechanism can effectively prevent repeated attempts by illegal users and improve the security of the overall system.

[0169] The design of the control module also allows for flexible expansion to cope with different usage scenarios. By working in collaboration with the adaptive learning module 103, the control module can dynamically adjust the unlocking conditions and sensitivity based on the user's historical behavior data and usage habits. In some cases, such as slight changes in the user's biometrics or behavior patterns, the system may automatically relax the matching threshold and allow errors within a certain range to avoid unlocking failures due to natural changes in the user. In scenarios with high security requirements, the control module can set more stringent standards, requiring both biometrics and behavior verification to achieve a high degree of match before unlocking, thereby improving the system's protection level.

[0170] The control module also has efficient communication and processing capabilities. It can not only quickly transmit information between multiple modules, but also make decisions quickly after receiving the verification signal to ensure the immediacy of the unlocking operation. In order to ensure the efficiency of the system, the control module usually adopts a multi-threaded processing architecture, which enables it to process signals from the biometric module and the behavior analysis module at the same time and draw conclusions in a very short time. This design not only improves the user experience, but also reduces waiting time and ensures a smooth unlocking process.

[0171] The scalability of the control module also enables it to integrate more security features. For example, it can work with other smart home systems to further enhance security based on data from sensors, cameras or other smart devices. For specific high-security scenarios, the control module can also be linked with Internet of Things (IoT) devices to further enhance the system's security protection level through remote control and real-time monitoring. In this way, when the door lock system detects abnormal behavior, users can monitor the door lock status in real time through a mobile phone application, or implement emergency closure or alarm through remote operation.

[0172] In general, the control module 104 plays a crucial role in decision-making in the smart door lock system. By receiving and comprehensively analyzing biometric and behavioral pattern verification signals, it can ensure that the door lock is unlocked quickly when operated by a legitimate user, and remains locked or sounds an alarm when there is illegal or abnormal behavior. This module is highly flexible and scalable, can adapt to a variety of usage scenarios, and improve the security of the overall system while ensuring user experience.

[0173] Furthermore, the adaptive smart door lock system also includes an environmental perception module for detecting environmental parameters around the door lock, including light intensity, temperature, humidity and noise level; the adaptive learning module dynamically adjusts the working parameters of the biometric module and the behavior analysis module based on the detected environmental parameters to adapt to different environmental conditions.

[0174] The environmental perception module can be implemented through multiple sensors installed on the door lock. For example, the light sensor is used to detect changes in light intensity around the door lock. This is particularly critical in facial recognition or other biometric processes that rely on optical sensors. If the ambient light is too strong or too weak, the system can automatically adjust the camera exposure or use infrared light sources for assistance to ensure that the biometric module can accurately capture the user's face or other biometric features. In addition, light perception can also be used to detect whether there are abnormal conditions in the environment. For example, sudden changes in light may indicate that someone is trying to interfere with the biometric device using tools such as flashlights.

[0175] The temperature sensor is used to monitor changes in ambient temperature. Too high or too low a temperature may affect the accuracy of biometrics, especially in fingerprint recognition or voiceprint recognition, where extreme temperatures may affect the performance of the skin or vocal cords. By detecting the ambient temperature, the adaptive learning module can adjust the sensitivity or matching threshold of the biometric module. For example, in a low temperature environment, the fingerprint sensor may need to increase sensitivity to ensure that the user's fingerprint can still be successfully recognized when the finger skin shrinks.

[0176] Humidity sensors are used to sense the humidity level in the environment, which is particularly important for fingerprint recognition and other biometric technologies that require direct contact. Excessive humidity may cause moisture or sweat on the surface of the finger, thus affecting the fingerprint recognition effect. By detecting the ambient humidity, the system can dynamically adjust the working parameters of the fingerprint sensor, such as reducing the sensitivity of the contact area to avoid moisture interfering with the recognition result.

[0177] Noise sensors are used to detect the noise level of the surrounding environment, which is especially important when the system uses voiceprint recognition. Excessive noise levels may interfere with the system's collection and recognition of the user's voiceprint. By detecting the noise level in real time, the system can choose to improve the anti-interference ability of voiceprint recognition in a high-noise environment, or suggest that the user perform voiceprint recognition in a quiet environment. In addition, noise perception can also be used as part of security precautions. For example, when abnormal noise (such as knocking or impact) is detected, the system can trigger additional alarms or verification steps.

[0178] After the environmental perception module detects these environmental parameters, the adaptive learning module adjusts the working parameters of the biometric module and the behavioral analysis module based on real-time data. For example, if the ambient light is weak, the adaptive learning module may instruct the facial recognition module to enable infrared fill light to ensure that insufficient light during the facial recognition process does not affect recognition accuracy. If the door lock is detected in a high humidity environment, the adaptive learning module may adjust the matching threshold of fingerprint recognition to reduce recognition errors caused by moisture or sweat.

[0179] This dynamic adjustment mechanism based on environmental perception enables the smart door lock system to flexibly respond to changes under different environmental conditions, ensuring the stability and reliability of the system.

[0180] Furthermore, the adaptive smart door lock system also includes an energy management module for optimizing energy consumption according to the user's usage habits and the system's working status; when it is detected that it has not been used for a long time or is in a low-risk time period, the system enters a low-power standby mode to extend battery life.

[0181] The main function of the energy management module is to determine the frequency of door lock usage in different time periods by learning and analyzing the user's usage behavior. By recording the time and number of times the user unlocks the door, as well as the usage pattern during the day, the system can identify which time periods are high-frequency or high-risk periods, and which time periods are less frequently used or have lower risks. Based on this data, the energy management module can optimize the energy consumption of the system.

[0182] When the energy management module detects that the door lock has not been used for a long time or is in a low-risk period, it will automatically switch the system to low-power standby mode. In low-power mode, the various components of the system will enter a dormant state or operate at the lowest energy consumption, retaining only necessary functions, such as waiting for the receipt of unlocking commands or regularly checking sensor data. At this time, the biometric module, behavior analysis module, and other functions related to identification and control will enter a dormant state until the system detects the user's approach or operation.

[0183] For example, if the system recognizes that the user rarely uses the door lock between 11 pm and 7 am every day, the energy management module can enter a low-power standby mode during this period. By reducing unnecessary energy consumption, the system can significantly extend the battery life when it is not in use for a long time.

[0184] In addition, the energy management module can also make real-time adjustments based on the system's operating status. For example, when the system is in high frequency of use or detects potential security threats (such as abnormal activities around the door lock), the energy management module can switch the system from low power mode back to normal working mode to ensure that all security functions of the door lock can run at full power when necessary. This dynamic energy management enables the system to maximize battery life while ensuring safety, especially for smart door lock devices that rely on battery power.

[0185] Through intelligent energy optimization, the energy management module ensures that the system can be adaptively adjusted according to the user's actual usage habits and environmental conditions.

[0186] Furthermore, the adaptive smart door lock system also includes a remote communication module, which is connected to the user's mobile terminal or cloud server via a wireless network. The user can monitor the door lock status in real time, receive abnormal alarm information, and remotely adjust the system's security settings or unlocking conditions.

[0187] One of the core functions of the remote communication module is to monitor the door lock status in real time. Through the wireless network connection, the current status of the door lock (such as unlocked, locked, low-power standby, etc.) will be sent to the user's mobile terminal or cloud platform in real time. Users can view the status information of the door lock through the mobile phone application, and no matter where they are, they can understand whether the door lock is used correctly or there are any abnormalities. For example, when the door lock is opened, the system will immediately push this information to the user's mobile phone, allowing the user to keep track of the use of the home or office door at any time.

[0188] In addition, the remote communication module also has an abnormal alarm function. When the system detects potential security threats or abnormal behaviors (such as continuous failed unlocking attempts, unusual door lock use at night, abnormal environmental changes around the door lock, etc.), the system will immediately send an alarm message to the user via the wireless network. This alarm can not only be sent through a mobile terminal, but can also be pushed to the user's preset email address or other cloud service platform, allowing the user to take timely action. In this way, no matter where the user is, they can immediately know the potential security risks of the door lock.

[0189] The remote communication module also allows users to remotely adjust the security settings or unlocking conditions of the door lock system. For example, users can dynamically modify the system's security mode in different situations through the mobile phone application. If the user is away on vacation, the system can be set to high security mode, which requires more stringent verification steps to unlock; in normal situations, users can choose convenient regular mode or low security mode to reduce the complexity of unlocking. Users can also adjust the matching threshold of biometrics, the sensitivity of behavioral verification and other settings according to actual needs, so as to flexibly respond to different scenarios and needs.

[0190] The module can communicate through a variety of wireless network technologies, including Wi-Fi, cellular data networks (such as 4G, 5G) or other IoT protocols (such as Zigbee, Z-Wave, etc.). Through the connection with the cloud server, the door lock's historical usage records, environmental data and security logs can be stored for a long time, allowing users to review or analyze the use of the door lock at any time. This not only helps with security management, but also provides more data support for the system's adaptive learning module to further optimize the door lock's behavior patterns and security strategies.

[0191] Through the integration of the remote communication module, it is possible to combine the existing wireless network technology and mobile application platform to achieve real-time monitoring, remote control and security management of the door lock system. The design of this module enables users to control the status of the door lock at any time, ensuring that the door lock system can be safely operated and adjusted at any location, improving the overall user experience and security.

[0192] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. An adaptive intelligent door lock system based on biometrics and behavior analysis, characterized in that: include: A biometric module, for collecting biometric data of a user, wherein the biometric data includes fingerprint, facial recognition, iris scan or voiceprint recognition; comparing the collected biometric data with a pre-stored user biometric template; and outputting a biometric verification pass signal to the control module when the comparison result meets a preset matching threshold; A behavior analysis module is used to record the current behavior interaction data between the user and the door lock, wherein the behavior interaction data includes the user's door lock usage frequency, time, motion trajectory data approaching the door lock, and force data of the unlocking operation; based on the recorded current behavior interaction data, calculate the user's current behavior pattern feature value; compare the calculated current behavior pattern feature value with the preset behavior pattern standard feature value; when the comparison result meets the preset comparison threshold, output a behavior pattern verification pass signal to the control module; An adaptive learning module, communicating with the biometric recognition module and the behavior analysis module, is used to dynamically adjust the user's biometric template according to the user's multiple biometric recognition operations to adapt to the subtle changes in the user's biometric characteristics; dynamically adjust the user's behavior pattern standard feature value according to the user's behavior interaction data; and adjust the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data; A control module, communicating with the biometric module and the behavior analysis module, for receiving a biometric verification pass signal from the biometric module; receiving a behavior pattern verification pass signal from the behavior analysis module; if the biometric verification pass signal and the behavior pattern verification pass signal are received, sending an unlocking instruction to the door lock actuator, otherwise, keeping the door lock locked or triggering a security alarm; The adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data, specifically including: Calculate the statistical characteristics of user historical behavior interaction data, including average, standard deviation, peak and valley values; Based on the statistical features, dynamically adjust at least one of the following parameters: Matching threshold of the biometric module; Comparison thresholds for the behavior analysis module; The unlocking decision weight of the control module, wherein the unlocking decision weight is used to balance the importance of biometric verification and behavioral pattern verification in the unlocking decision; The adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data, and also includes: Establish a normal range model for user behavior. When a behavior that significantly deviates from the model is detected, increase the system's vigilance and require additional verification steps. Depending on the degree of deviation and the current security level, select one or more of the following additional verification steps to perform: (a) Ask users to answer pre-set identity verification questions; (b) sending a verification code to the user's pre-registered mobile device and requiring the user to enter the verification code; (c) Activate the real-time video call function and require the user to confirm their identity via video; (d) reduce the tolerance of biometric matching thresholds and require higher matching accuracy; (e) Triggering secondary biometrics, requiring the user to provide two or more biometric data; (f) Start geolocation verification to confirm whether the user’s current location is as expected; (g) requiring users to perform additional behavioral verification through pre-set gestures or action sequences; The adaptive learning module adjusts the unlocking condition or sensitivity of the door lock based on the user's historical behavior interaction data, and also includes: A Gaussian mixture model is established to fit the probability distribution of user behavior interaction data. The probability distribution function P(x) of the Gaussian mixture model adopts the following formula 1: Among them, x represents the user's behavior interaction data, and the user's behavior interaction data x=[x1,x2,…x i ,…,x n ], where x i represents the behavior interaction data for the i-th dimension, n represents the total number of dimensions of the behavior interaction data; π k (x) is the weight function of the kth Gaussian distribution; represents the probability distribution function of the kth Gaussian distribution, μ k is the mean of the kth Gaussian distribution; Σ k is the covariance matrix of the kth Gaussian distribution; K represents the total number of dimensions of the behavioral interaction data; where the weight function π k (x) is calculated according to the following formula 2: Among them, α is the sensitivity coefficient used to control dynamic adjustment; D k (x) is the mean μ of the user's behavior interaction data x and the kth Gaussian distribution k The Mahalanobis distance between k The calculation formula of (x) is shown in the following formula 3: Estimate the parameters μ of the Gaussian mixture model using the expectation-maximization algorithm k ,Σ k ; According to the estimated parameters of the Gaussian mixture model, the current behavior interaction data x is calculated using formula 1 new The probability P(x new ); when the probability P(x new ) is less than the preset threshold T P When the current behavior deviates from the normal range; The predicted value of the user's behavior interaction data at the current time t is obtained through the autoregressive time series model provided by the following formula 4 where φ i is the i-th autoregressive coefficient; ∈ t is the noise term; p represents the lag order of historical data; x t-i represents the actual value of the behavioral interaction data obtained at time ti; According to the following formula 5, the predicted value is calculated The actual value x of the behavior interaction data at the current time t t The difference between the data Δx: When P(x new ) <T P And Δx>T AR When the system determines that the user's behavior is abnormal, it performs additional verification, including: When T P1 <P(x new )≤T P2 When , perform additional verification steps a or b; When T P2 <P(x new )≤T P3 When , perform additional verification steps c or d; When T P3 <P(x new )≤T P When , perform additional verification steps e, f or g; Among them, T P , T P1 , T P2 and T P3 is a pre-determined threshold.

2. The adaptive intelligent door lock system according to claim 1, characterized in that: The adaptive learning module is also used to: Based on the door lock usage frequency and time period provided by the user's historical behavioral interaction data, different security modes are automatically switched, wherein the security mode includes a high security mode, a regular security mode and a low security mode, wherein the high security mode requires at least one additional verification step to be performed on the basis of satisfying biometric verification and behavioral pattern verification; the regular security mode requires satisfying biometric verification and behavioral pattern verification; the low security mode requires satisfying one of the biometric verification and behavioral pattern verification requirements of the regular security mode.

3. The adaptive intelligent door lock system according to claim 2, characterized in that: It also includes a security policy module for pre-configuring authentication policies at different security levels based on user-defined preferences; when abnormal behavior is detected, it automatically selects additional authentication steps based on the current security level; Record the results and user feedback of each additional verification for optimizing future verification strategies; allow authorized users to remotely adjust security policies and priorities of verification steps through the mobile application.

4. The adaptive intelligent door lock system according to claim 1, characterized in that: The biometric module includes a liveness detection unit, which is used to detect whether the user is alive when collecting biometric data to prevent fraudulent unlocking.

5. The adaptive intelligent door lock system according to claim 1, characterized in that: The behavior analysis module includes a three-dimensional space perception sensor for capturing the three-dimensional motion trajectory and posture information of the user when approaching the door lock.

6. The adaptive intelligent door lock system according to claim 1, characterized in that: It also includes an environmental perception module for detecting environmental parameters around the door lock, including light intensity, temperature, humidity and noise level; the adaptive learning module dynamically adjusts the working parameters of the biometric module and the behavior analysis module based on the detected environmental parameters to adapt to different environmental conditions.

7. The adaptive intelligent door lock system according to claim 1, characterized in that: It also includes an energy management module for optimizing energy consumption based on user usage habits and system operating status; when it detects that the system has not been used for a long time or is in a low-risk time period, the system enters a low-power standby mode to extend battery life.

8. The adaptive intelligent door lock system according to claim 1, characterized in that: It also includes a remote communication module, which is connected to the user's mobile terminal or cloud server via a wireless network. The user can monitor the door lock status in real time, receive abnormal alarm information, and remotely adjust the system's security settings or unlocking conditions.

Images