Cloud platform monitoring method for multi-data parallel processing of intelligent lock

Abstract

The invention discloses a cloud platform monitoring method for intelligent lock multi-data parallel processing, and relates to the field of intelligent lock monitoring, S1, data acquisition and transmission: an intelligent lock end collects data of an intelligent lock through a sensor and a communication module built in the intelligent lock; s2, transmission monitoring; s3, the state of the intelligent lock is monitored, specifically, deep analysis is conducted on biological recognition feedback, and the micro state in the intelligent lock is monitored; s4, evaluating the monitoring quality by using a machine learning algorithm; and S5, alarming and notifying. According to the cloud platform monitoring method for multi-data parallel processing of the intelligent lock, multiple states of the intelligent lock can be monitored, meanwhile, different intelligent lock data can be processed based on the cloud platform, when the multiple state data of the intelligent lock are transmitted, the data integrity can be improved, and the intelligent lock multi-data parallel processing efficiency is improved. And errors and malicious tampering in the data transmission process can be avoided, so that the monitoring precision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of monitoring of smart locks, and in particular to a cloud platform monitoring method for parallel processing of multiple data of smart locks. Background Art

[0002] Smart door locks are improved upon traditional mechanical locks and are more intelligent and simple in terms of user security, identification, and manageability. Smart door locks are the executive components of door locking in access control systems. Smart door locks are different from traditional mechanical locks and are composite locks that combine security, convenience, and advanced technology.

[0003] When monitoring the current status of a smart lock, the monitoring range is limited, and large-scale monitoring and processing cannot be performed based on the data that the smart lock can provide. When a problem occurs inside the smart lock, it is necessary to find a professional technician to find the problem and then repair it. This process is time-consuming and labor-intensive, and users cannot directly understand the current status of the smart lock based on the monitored content, which will have a certain impact on its handling and maintenance.

[0004] Therefore, it is necessary to propose a cloud platform monitoring method for parallel processing of multi-data of smart locks to solve the above problems. Summary of the invention

[0005] The main purpose of the present invention is to provide a cloud platform monitoring method for parallel processing of multiple data of smart locks, which can effectively solve the problems in the background technology.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] A cloud platform monitoring method for parallel processing of multiple data of smart locks includes the following steps:

[0008] S1: Data collection and transmission. The smart lock collects data about the smart lock through the built-in sensors and communication modules, including unlocking time, locking time, unlocking method, power information, lock status, and user operation records, and transmits the data to the cloud platform through the MQTT protocol.

[0009] S2: Transmission monitoring, including connection status monitoring and data integrity detection. The connection status monitoring includes: continuously checking the communication connection between the smart lock and the cloud platform, regularly sending heartbeat packets through the MQTT protocol to confirm whether the smart lock is online. If no heartbeat response is received within the specified time, the smart lock is marked as offline and the offline time is recorded;

[0010] Data integrity detection is used to determine whether data has been tampered with or has not been erroneous during transmission;

[0011] S3: Monitor the status of the smart lock, including in-depth analysis of biometric feedback and internal microscopic status monitoring of the smart lock;

[0012] S4: Use machine learning algorithms to evaluate monitoring quality, collect the above content, use it as a data set, train it, learn the corresponding feature representation, and output its quality level;

[0013] S5: Alarm and notification. When the smart lock is monitored to be in an abnormal state, different alarm levels are set according to the severity of the abnormal situation, and the user, property and relevant security departments are notified.

[0014] Preferably, in S1, the unlocking methods include fingerprint, face, card, password, mechanical, and remote unlocking; the lock body status includes open, closed, slight abnormal state, medium abnormal state, and severe abnormal state; the user operation record includes user setting operation record, user management operation record, and abnormal operation record;

[0015] The cloud platform includes a data receiving interface, the number of which is consistent with the number of received smart lock data. The cloud platform preprocesses the received data, including data format unification, data cleaning and data verification. The cloud platform converts the data sent by the smart lock into a unified format of the cloud platform through data format unification;

[0016] Data cleaning is used to remove noise and erroneous data in the data, including but not limited to filtering out duplicate and incomplete unlocking records caused by network fluctuations;

[0017] Data verification is used to check the current information of the data, including whether the power information is within the normal power range and the current status of the smart lock;

[0018] The cloud platform divides different smart lock data into different partitions, and the data of each partition is analyzed on an independent computing node.

[0019] Preferably, the data integrity detection in S2 is verified by using a hash value, and the verification includes the following steps:

[0020] S201: Classify the data collected in S1 and form separate data blocks, and calculate a 256-bit hash value using the SHA-256 algorithm;

[0021] S202: Packing the generated hash value and the data block connected thereto, and sending them to the cloud platform through an encrypted channel. After receiving the data, the cloud platform parses the packed data and separates the generated hash value and the data block again.

[0022] S203: recalculate the hash value of the received data block using the same SHA-256 algorithm, and compare the recalculated hash value with the hash value received from the smart lock. If the two hash values ​​are exactly the same, it is determined that the data has not been tampered with during transmission. If the two hash values ​​are different, it is determined that the data has been tampered with or has been tampered with maliciously during transmission.

[0023] S204: When it is determined that the data is erroneous or maliciously tampered with during transmission, the smart lock is requested to resend the data packet and its hash value and re-verify them. When it is still determined that the data is tampered with during transmission, the user is notified of the data anomaly through the cloud platform.

[0024] Preferably, in S3, the in-depth analysis of biometric feedback includes dynamic fingerprint quality assessment and multi-dimensional analysis of facial recognition, wherein the dynamic fingerprint quality assessment includes the following steps:

[0025] S30101: Fingerprint texture clarity analysis, obtain the fingerprint texture image, calculate the gradient according to the coordinates of the fingerprint texture image, and evaluate the fingerprint quality based on the calculation results. The following is

[0026] The x-direction gradient G x The calculation formula is:

[0027]

[0028] The following is the y-direction gradient G y The calculation formula is:

[0029]

[0030] The gradient information in the x and y directions is used to calculate the gradient magnitude and direction, where the magnitude is calculated as:

[0031]

[0032] in is the amplitude, I is the pixel gray value, i, j are the coordinates of the image pixels;

[0033] S30102: Set the threshold of the gradient amplitude of the fingerprint texture image, set the threshold T = 10, and calculate Statistics are performed to determine the proportion P of pixels with gradient amplitudes greater than 10. When P>70%, the fingerprint texture clarity is judged to be up to standard. After meeting the standard, the fingerprint owner can pass monitoring and open the smart lock. When 30%≤P≥70%, it is judged that the fingerprint owner's finger skin condition is abnormal, the sensor is dirty and damaged; when P<30%, the fingerprint texture clarity is judged to be poor, and the fingerprint owner cannot open the smart lock.

[0034] Preferably, in the facial recognition multi-dimensional analysis, the smart lock includes a camera with a wide-angle function and an automatically adjustable viewing angle, the camera is used to collect facial images from the front and side, and is used to obtain comprehensive facial contours and spatial position information of facial features. The smart lock also includes a light sensor for collecting facial images under different light intensities and color temperatures, including the following operating steps:

[0035] S30201: Static feature in-depth analysis: further analyze facial texture details based on the acquired facial contour and spatial position information of the facial features. Based on the image analysis algorithm, identify and model the tiny texture of the skin, and analyze the color characteristics of the face, including subtle differences in skin color and color characteristics of local areas.

[0036] S30202: Dynamic feature integration analysis, introducing dynamic facial expression analysis. When a user is identified in front of a smart lock, the system observes the amplitude, frequency and coordination of the user's facial movements, and combines dynamic features with static features to form an independent personal facial recognition mode;

[0037] S30203: Multimodal fusion algorithm, using convolutional neural network to extract multi-angle images, images with different illumination, static and dynamic features, using recurrent neural network to model dynamic feature sequences, and integrating all features through the fusion layer;

[0038] S30204: Analyze the current user's expression and action based on the above features, including observing the user's blinking frequency and smiling degree, and establish a database to set a blinking frequency and smiling time range. If the blinking frequency and smiling time range are too high or too low during the recognition process, it is judged as an abnormal situation and the smart lock does not unlock.

[0039] S30205: Based on the above features, the facial muscle features of the current user are analyzed, and the deep learning model is used to analyze the changes in facial muscle tension and identify different muscle movement patterns. When excessive muscle tension and unnatural contraction patterns are detected, the user is judged to be in a tense state. At this time, the smart lock needs further verification. If further verification fails, it will refuse to unlock.

[0040] Preferably, in the internal microscopic state monitoring of the smart lock, the smart lock includes a pressure sensor, an environmental sensor, and a stress color-changing layer detection. The pressure sensor is used to judge whether the smart lock is a normal door opening collision or a malicious external force impact through the data it monitors; the environmental sensor is an environmental sensor array, which is installed inside the smart lock and is used to detect the gas composition, concentration changes and humidity changes in the internal space of the lock; the detection of the stress color-changing layer includes a stress color-changing layer and an optical detection device for detecting the stress color-changing layer, and the stress color-changing layer is applied to the lock core inside the smart lock.

[0041] Preferably, the pressure sensor is installed on both sides and the top of the smart lock tongue, the data acquisition frequency of the pressure sensor is set to 10-100 times per second, and the collected data is stored in the cloud platform, which specifically includes the following steps:

[0042] S30301: Data cleaning, using median filtering to remove noise and outliers in the data collected by the pressure sensor;

[0043] S30303: Feature extraction, including pressure peak features: Extract the peak features in the pressure data, perform empirical mode decomposition on the pressure data sequence P, obtain n intrinsic mode functions IMF, and for each IMF i Perform Hilbert transform to obtain its instantaneous amplitude A i (t) and instantaneous frequency ω i (t), in each A i (t) Find the maximum value point, the pressure data position corresponding to the maximum value point here is recorded as the peak point, and it is recorded as the normal peak point. Based on this, it is compared with the peak point when the current smart lock is unlocked. When the difference between the current peak point and the normal peak point is within 15%, the current pressure peak feature is judged to be normal, otherwise it is judged to be abnormal;

[0044] S30304: Pressure change rate feature, calculates the change rate of pressure data, the formula is:

[0045]

[0046] where r i is the pressure data change rate sequence; p i is the pressure data; Δt is a constant; where Δt = t i+1 -t i ;t i is time, p i With t i corresponding;

[0047] The value of the pressure data change rate sequence is averaged and identified as a threshold. Based on the identified threshold, the pressure data change rate sequence value calculated is judged. When it is greater than the threshold by 15% and lower than the threshold by 15%, the current pressure change rate characteristic is judged to be abnormal, otherwise it is judged to be normal.

[0048] S30305: Prioritize the extracted features, set the pressure peak feature as the first priority, and set the pressure change rate feature as the second priority. When the extracted pressure peak feature is judged to be normal, the smart lock status is directly judged to be normal. When the extracted pressure peak feature is judged to be abnormal, continue to judge through the pressure change rate feature of the second priority. When it is judged to be normal, the smart lock works normally and records the pressure peak feature extracted at this time. When it is judged to be abnormal, it is judged as a serious abnormality and the smart lock cancels subsequent work.

[0049] Preferably, the environmental sensor includes a gas sensor and a humidity sensor. The gas sensor is used to monitor whether the circuit board inside the smart lock generates special gas due to overheating. When special gas is detected, it is determined that the circuit board inside the smart lock is faulty, and it is judged as a serious abnormality. The humidity sensor is used to monitor changes in humidity to determine whether liquid has entered the smart lock. When it is determined that there is liquid, it is determined that the circuit board of the smart lock is damp, and it is judged as a serious abnormality.

[0050] Preferably, in the detection of the stress chromic layer, an optical detection device is used to monitor whether the stress chromic coating applied to the lock core of the smart lock changes color. When the stress chromic coating changes color, it is judged that the state of the smart lock is abnormal, and it is also judged as a serious abnormality.

[0051] Preferably, in S5, the alarm levels include minor abnormalities, medium abnormalities, and severe abnormalities, wherein minor abnormalities include insufficient power, online problems, addition and reduction of management personnel, password modification, addition of face, password, fingerprint, and card swiping information; medium abnormalities include multiple errors in unlocking methods; severe abnormalities include changes detected by the smart lock pressure, environment, and optical detection device;

[0052] Minor anomalies are pushed directly to users through the cloud platform; moderate anomalies are notified to users via text messages; and serious anomalies are notified to users, property management and relevant security departments at the same time.

[0053] Compared with the prior art, the present invention provides a cloud platform monitoring method for parallel processing of multiple data of smart locks, which has the following beneficial effects:

[0054] 1. The cloud platform monitoring method for parallel processing of multiple data of smart locks can monitor the status of smart locks. At the same time, different smart lock data can be processed based on the cloud platform. When transmitting various status data of smart locks, the integrity of the data can be improved, and errors and malicious tampering during data transmission can be avoided, thereby effectively improving the monitoring accuracy.

[0055] 2. The cloud platform monitoring method for parallel processing of multiple data of smart locks can dynamically evaluate the fingerprint quality of users and conduct multi-dimensional analysis of their facial recognition by conducting in-depth analysis of biometric feedback of smart locks. The current user and smart lock status can be judged according to the results of dynamic fingerprint quality evaluation, and corresponding smart lock operations and notification operations can be performed according to the judgment results. It can also be judged whether the user is in an abnormal state according to the multi-dimensional analysis results of facial recognition. Based on this, the environment in which the current user is located can be judged, which can improve their safety and prevent them from opening the smart lock and entering the room due to coercion.

[0056] 3. The cloud platform monitoring method for parallel processing of multiple data of smart locks can further monitor the status of the smart lock through the detection of pressure sensors, environmental sensors, and stress discoloration layers through the microscopic status monitoring inside the smart lock. The pressure sensor can determine whether the smart lock is a normal door opening collision or a malicious external force impact. The environmental sensor can monitor the status of the circuit board inside the smart lock. The detection of the stress discoloration layer can detect whether the smart lock cylinder has inserted a key that does not belong to the current smart lock through an optical detection device. Based on this, the monitoring efficiency of the smart lock can be further improved.

[0057] 4. The cloud platform monitoring method for parallel processing of multiple data of smart locks can also alarm and notify according to the current monitored status of the smart lock, and select different notification methods according to different status. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0059] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0060] Embodiment 1:

[0061] like Figure 1 As shown, a cloud platform monitoring method for parallel processing of multiple data of smart locks includes the following steps:

[0062] S1: Data collection and transmission. The smart lock collects data about the smart lock through the built-in sensors and communication modules, including unlocking time, locking time, unlocking method, power information, lock status, and user operation records, and transmits the data to the cloud platform through the MQTT protocol.

[0063] S2: Transmission monitoring, including connection status monitoring and data integrity detection. The connection status monitoring includes: continuously checking the communication connection between the smart lock and the cloud platform, regularly sending heartbeat packets through the MQTT protocol to confirm whether the smart lock is online. If no heartbeat response is received within the specified time, the smart lock is marked as offline and the offline time is recorded;

[0064] Data integrity detection is used to determine whether data has been tampered with or has not been erroneous during transmission;

[0065] S3: Monitor the status of the smart lock, including in-depth analysis of biometric feedback and internal microscopic status monitoring of the smart lock;

[0066] S4: Use machine learning algorithms to evaluate monitoring quality, collect the above content, use it as a data set, train it, learn the corresponding feature representation, and output its quality level;

[0067] S5: Alarm and notification. When the smart lock is monitored to be in an abnormal state, different alarm levels are set according to the severity of the abnormal situation, and the user, property and relevant security departments are notified.

[0068] Embodiment 2:

[0069] A cloud platform monitoring method for parallel processing of multiple data of smart locks, data collection and transmission. The smart lock end collects the data of the smart lock through the built-in sensors and communication modules of the smart lock, including unlocking time, locking time, unlocking method, power information, lock body status, user operation records, and transmits the data to the cloud platform through the MQTT protocol;

[0070] Unlocking methods include fingerprint, face, card, password, mechanical, and remote unlocking; lock status includes open, closed, slight abnormal state, medium abnormal state, and severe abnormal state; user operation records include user setting operation records, user management operation records, and abnormal operation records;

[0071] The cloud platform includes data receiving interfaces. The number of data receiving interfaces is consistent with the number of smart lock data received. The cloud platform pre-processes the received data, including data format unification, data cleaning and data verification. The cloud platform converts the data sent by the smart lock into the unified format of the cloud platform through data format unification.

[0072] Data cleaning is used to remove noise and erroneous data in the data, including but not limited to filtering out duplicate and incomplete unlocking records caused by network fluctuations;

[0073] Data verification is used to check the current information of the data, including whether the power information is within the normal power range and the current status of the smart lock;

[0074] The cloud platform divides different smart lock data into different partitions, and the data of each partition is analyzed on an independent computing node.

[0075] Embodiment three:

[0076] A cloud platform monitoring method for parallel processing of multiple data for smart locks, transmission monitoring, including connection status monitoring and data integrity detection, wherein the connection status monitoring includes: continuously checking the communication connection between the smart lock and the cloud platform, regularly sending heartbeat packets through the MQTT protocol to confirm whether the smart lock is online, and if no heartbeat response is received within the specified time, marking the smart lock as offline and recording the offline time.

[0077] Data integrity detection is used to determine whether data has been tampered with or has not been erroneous during transmission;

[0078] Data integrity detection uses hash values ​​for verification, and the verification includes the following steps:

[0079] S201: Classify the data collected in S1 and form separate data blocks, and calculate a 256-bit hash value using the SHA-256 algorithm;

[0080] S202: Packing the generated hash value and the data block connected thereto, and sending them to the cloud platform through an encrypted channel. After receiving the data, the cloud platform parses the packed data and separates the generated hash value and the data block again.

[0081] S203: recalculate the hash value of the received data block using the same SHA-256 algorithm, and compare the recalculated hash value with the hash value received from the smart lock. If the two hash values ​​are exactly the same, it is determined that the data has not been tampered with during transmission. If the two hash values ​​are different, it is determined that the data has been tampered with or has been tampered with maliciously during transmission.

[0082] S204: When it is determined that the data is erroneous or maliciously tampered with during transmission, the smart lock is requested to resend the data packet and its hash value and re-verify them. When it is still determined that the data is tampered with during transmission, the user is notified of the data anomaly through the cloud platform.

[0083] Embodiment 4:

[0084] A cloud platform monitoring method for parallel processing of multiple data of smart locks, monitoring the status of smart locks, including in-depth analysis of biometric feedback and internal microscopic status monitoring of smart locks;

[0085] In-depth analysis of biometric feedback includes dynamic fingerprint quality assessment and multi-dimensional analysis of facial recognition. The dynamic fingerprint quality assessment includes the following steps:

[0086] S30101: Fingerprint texture clarity analysis, obtain the fingerprint texture image, calculate the gradient according to the coordinates of the fingerprint texture image, and evaluate the fingerprint quality based on the calculation results. The following is

[0087] The x-direction gradient G x The calculation formula is:

[0088]

[0089] The following is the y-direction gradient G y The calculation formula is:

[0090]

[0091] The gradient information in the x and y directions is used to calculate the gradient magnitude and direction, where the magnitude is calculated as:

[0092]

[0093] in is the amplitude, I is the pixel gray value, i, j are the coordinates of the image pixels;

[0094] S30102: Set the threshold of the gradient amplitude of the fingerprint texture image, set the threshold T = 10, and calculate The proportion of pixels with gradient amplitude greater than 10 is counted. When P>70%, the fingerprint texture clarity is judged to be up to standard. After reaching the standard, the fingerprint owner can pass the monitoring and open the smart lock. When 30%≤P≥70%, the fingerprint owner's finger skin condition is abnormal, the sensor is dirty and damaged; when P<30%, the fingerprint texture clarity is judged to be poor, and the fingerprint owner cannot open the smart lock.

[0095] In the multi-dimensional analysis of facial recognition, the smart lock includes a camera with a wide-angle function and can automatically adjust the viewing angle. The camera is used to collect facial images from the front and side to obtain comprehensive facial contours and spatial position information of the facial features. The smart lock also includes a light sensor to collect facial images under different light intensities and color temperatures, including the following steps:

[0096] S30201: Static feature in-depth analysis: further analyze facial texture details based on the acquired facial contour and spatial position information of the facial features. Based on the image analysis algorithm, identify and model the tiny texture of the skin, and analyze the color characteristics of the face, including subtle differences in skin color and color characteristics of local areas.

[0097] S30202: Dynamic feature integration analysis, introducing dynamic facial expression analysis. When a user is identified in front of a smart lock, the system observes the amplitude, frequency and coordination of the user's facial movements, and combines dynamic features with static features to form an independent personal facial recognition mode;

[0098] S30203: Multimodal fusion algorithm, using convolutional neural network to extract multi-angle images, images with different illumination, static and dynamic features, using recurrent neural network to model dynamic feature sequences, and integrating all features through the fusion layer;

[0099] S30204: Analyze the current user's expression and action based on the above features, including observing the user's blinking frequency and smiling degree, and establish a database to set a blinking frequency and smiling time range. If the blinking frequency and smiling time range are too high or too low during the recognition process, it is judged as an abnormal situation and the smart lock does not unlock.

[0100] S30205: Based on the above features, the facial muscle features of the current user are analyzed, and the deep learning model is used to analyze the changes in facial muscle tension and identify different muscle movement patterns. When excessive muscle tension and unnatural contraction patterns are detected, the user is judged to be in a tense state. At this time, the smart lock needs further verification. If further verification fails, it will refuse to unlock.

[0101] In the internal microscopic state monitoring of the smart lock, the smart lock includes pressure sensors, environmental sensors, and stress color layer detection. The pressure sensor is used to judge whether the smart lock is a normal door opening collision or a malicious external force impact through its monitored data; the environmental sensor is an environmental sensor array installed inside the smart lock to detect the gas composition, concentration changes and humidity changes in the internal space of the lock; the stress color layer detection includes a stress color layer and an optical detection device for detecting the stress color layer. The stress color layer is applied to the lock core inside the smart lock;

[0102] The pressure sensor is installed on both sides and the top of the smart lock tongue. The data collection frequency of the pressure sensor is set to 10-100 times per second, and the collected data is stored in the cloud platform. The specific operation steps include the following:

[0103] S30301: Data cleaning, using median filtering to remove noise and outliers in the data collected by the pressure sensor;

[0104] S30303: Feature extraction, including pressure peak features: Extract the peak features in the pressure data, perform empirical mode decomposition on the pressure data sequence P, obtain n intrinsic mode functions IMF, and for each IMF i Perform Hilbert transform to obtain its instantaneous amplitude A i(t) and instantaneous frequency ω i (t), in each A i (t) Find the maximum value point, the pressure data position corresponding to the maximum value point here is recorded as the peak point, and it is recorded as the normal peak point. Based on this, it is compared with the peak point when the current smart lock is unlocked. When the difference between the current peak point and the normal peak point is within 15%, the current pressure peak feature is judged to be normal, otherwise it is judged to be abnormal;

[0105] S30304: Pressure change rate feature, calculates the change rate of pressure data, the formula is:

[0106]

[0107] where r i is the pressure data change rate sequence; p i is the pressure data; Δt is a constant; where Δt = t i+1 -t i ;t i is time, p i With t i corresponding;

[0108] The value of the pressure data change rate sequence is averaged and identified as a threshold. Based on the identified threshold, the pressure data change rate sequence value calculated is judged. When it is greater than the threshold by 15% and lower than the threshold by 15%, the current pressure change rate characteristic is judged to be abnormal, otherwise it is judged to be normal.

[0109] S30305: Prioritize the extracted features, set the pressure peak feature as the first priority, and set the pressure change rate feature as the second priority. When the extracted pressure peak feature is judged to be normal, the smart lock status is directly judged to be normal. When the extracted pressure peak feature is judged to be abnormal, continue to judge through the pressure change rate feature of the second priority. When it is judged to be normal, the smart lock works normally and records the pressure peak feature extracted at this time. When it is judged to be abnormal, it is judged as a serious abnormality and the smart lock cancels subsequent work.

[0110] Environmental sensors include gas sensors and humidity sensors. The gas sensor is used to monitor whether the circuit board inside the smart lock generates special gas due to overheating. When special gas is detected, it is judged that the circuit board inside the smart lock is faulty and it is judged as a serious abnormality. The humidity sensor is used to monitor the change of humidity to determine whether liquid has entered the smart lock. When it is determined that there is liquid, it is judged that the circuit board of the smart lock is damp and it is judged as a serious abnormality.

[0111] The special gases are based on the constituent materials of the circuit board, including but not limited to gases generated by the decomposition of insulating materials, gases generated by the volatilization of solder, gases generated by plastic parts, and gases generated by circuit and component coatings.

[0112] In the detection of the stress-chromic coating, the optical detection device is used to monitor whether the stress-chromic coating applied to the lock core of the smart lock changes color. When the stress-chromic coating changes color, it is judged that the state of the smart lock is abnormal, and it is also judged as a serious abnormality;

[0113] The stress chromic layer uses a pressure-sensitive pigment type stress chromic layer. The optical detection device records the initial color of the pressure-sensitive pigment type stress chromic layer. When the color change of the pressure-sensitive pigment type stress chromic layer is detected, a serious abnormality can be judged.

[0114] Embodiment five:

[0115] A cloud platform monitoring method for parallel processing of multiple data for smart locks, using a machine learning algorithm to evaluate the monitoring quality, collecting the above-obtained content, using it as a data set, training it so that it can learn the corresponding feature representation, and output its quality level;

[0116] The quality levels include high, medium and low. Taking fingerprint quality as an example, a fingerprint with clear texture, complete detail feature points and no obvious noise is marked as high, and its score is marked as 8-10 points; a fingerprint with fuzzy texture and fewer feature points is marked as low, and its score is marked as 0-3 points; a fingerprint between the two is medium, and its score is marked as 4-7.

[0117] Embodiment six:

[0118] A cloud platform monitoring method for parallel processing of multiple data for smart locks, alarm and notification. When the smart lock is monitored to be in an abnormal state, different alarm levels are set according to the severity of the abnormal situation, and users, property and relevant security departments are notified;

[0119] The alarm levels include minor abnormalities, medium abnormalities, and severe abnormalities. Minor abnormalities include low battery, online problems, increase and decrease of management personnel, password modification, addition of face, password, fingerprint, and card swiping information; medium abnormalities include multiple incorrect unlocking methods; severe abnormalities include changes detected by the smart lock pressure, environment, and optical detection device;

[0120] Minor anomalies are pushed directly to users through the cloud platform; moderate anomalies are notified to users via text messages; and serious anomalies are notified to users, property management and relevant security departments at the same time.

[0121] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A cloud platform monitoring method for parallel processing of multiple data of smart locks, characterized in that: The steps include: S1: Data collection and transmission. The smart lock collects data about the smart lock through the built-in sensors and communication modules, including unlocking time, locking time, unlocking method, power information, lock status, and user operation records, and transmits the data to the cloud platform through the MQTT protocol. S2: Transmission monitoring, including connection status monitoring and data integrity detection. The connection status monitoring includes: continuously checking the communication connection between the smart lock and the cloud platform, regularly sending heartbeat packets through the MQTT protocol to confirm whether the smart lock is online. If no heartbeat response is received within the specified time, the smart lock is marked as offline and the offline time is recorded; Data integrity detection is used to determine whether data has been tampered with or has not been erroneous during transmission; S3: Monitor the status of the smart lock, including in-depth analysis of biometric feedback and internal microscopic status monitoring of the smart lock; S4: Use machine learning algorithms to evaluate monitoring quality, collect the above content, use it as a data set, train it, learn the corresponding feature representation, and output its quality level; S5: Alarm and notification. When the smart lock is monitored to be in an abnormal state, different alarm levels are set according to the severity of the abnormal situation, and the user, property and relevant security departments are notified.

2. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 1, characterized in that: In S1, the unlocking methods include fingerprint, face, card, password, mechanical, and remote unlocking; the lock body status includes open, closed, slight abnormal state, medium abnormal state, and severe abnormal state; the user operation record includes user setting operation record, user management operation record, and abnormal operation record; The cloud platform includes a data receiving interface, the number of which is consistent with the number of received smart lock data. The cloud platform preprocesses the received data, including data format unification, data cleaning and data verification. The cloud platform converts the data sent by the smart lock into a unified format of the cloud platform through data format unification; Data cleaning is used to remove noise and erroneous data in the data, including but not limited to filtering out duplicate and incomplete unlocking records caused by network fluctuations; Data verification is used to check the current information of the data, including whether the power information is within the normal power range and the current status of the smart lock; The cloud platform divides different smart lock data into different partitions, and the data of each partition is analyzed on an independent computing node.

3. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 1, characterized in that: The data integrity check in S2 is verified by using a hash value, and the verification includes the following steps: S201: Classify the data collected in S1 and form separate data blocks, and calculate a 256-bit hash value using the SHA-256 algorithm; S202: Packing the generated hash value and the data block connected thereto, and sending them to the cloud platform through an encrypted channel. After receiving the data, the cloud platform parses the packed data and separates the generated hash value and the data block again. S203: recalculate the hash value of the received data block using the same SHA-256 algorithm, and compare the recalculated hash value with the hash value received from the smart lock. If the two hash values ​​are exactly the same, it is determined that the data has not been tampered with during transmission. If the two hash values ​​are different, it is determined that the data has been tampered with or has been tampered with maliciously during transmission. S204: When it is determined that the data is erroneous or maliciously tampered with during transmission, the smart lock is requested to resend the data packet and its hash value and re-verify them. When it is still determined that the data is tampered with during transmission, the user is notified of the data anomaly through the cloud platform.

4. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 1, characterized in that: In S3, the in-depth analysis of biometric feedback includes dynamic fingerprint quality assessment and multi-dimensional analysis of facial recognition, wherein the dynamic fingerprint quality assessment includes the following steps: S30101: Fingerprint texture clarity analysis, obtain the fingerprint texture image, calculate the gradient according to the coordinates of the fingerprint texture image, and evaluate the quality of the fingerprint based on the calculation results. The following is the x-direction gradient G x The calculation formula is: The following is the y-direction gradient G y The calculation formula is: The gradient information in the x and y directions is used to calculate the gradient magnitude and direction, where the magnitude is calculated as: in is the amplitude, I is the pixel gray value, i, j are the coordinates of the image pixels; S30102: Set the threshold of the gradient amplitude of the fingerprint texture image, set the threshold T = 10, and calculate Statistics are performed to determine the proportion P of pixels with gradient amplitudes greater than 10. When P>70%, the fingerprint texture clarity is judged to be up to standard. After meeting the standard, the fingerprint owner can pass monitoring and open the smart lock. When 30%≤P≥70%, it is judged that the fingerprint owner's finger skin condition is abnormal, the sensor is dirty and damaged; when P<30%, the fingerprint texture clarity is judged to be poor, and the fingerprint owner cannot open the smart lock.

5. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 4, characterized in that: In the multi-dimensional analysis of facial recognition, the smart lock includes a camera with a wide-angle function and an automatically adjustable viewing angle. The camera is used to collect facial images from the front and side to obtain comprehensive facial contours and spatial position information of facial features. The smart lock also includes a light sensor for collecting facial images under different light intensities and color temperatures, including the following steps: S30201: Static feature in-depth analysis: further analyze facial texture details based on the acquired facial contour and spatial position information of the facial features. Based on the image analysis algorithm, identify and model the tiny texture of the skin, and analyze the color characteristics of the face, including subtle differences in skin color and color characteristics of local areas. S30202: Dynamic feature integration analysis, introducing dynamic facial expression analysis. When a user is identified in front of a smart lock, the system observes the amplitude, frequency and coordination of the user's facial movements, and combines dynamic features with static features to form an independent personal facial recognition mode; S30203: Multimodal fusion algorithm, using convolutional neural network to extract multi-angle images, images with different illumination, static and dynamic features, using recurrent neural network to model dynamic feature sequences, and integrating all features through the fusion layer; S30204: Analyze the current user's expression and action based on the above features, including observing the user's blinking frequency and smiling degree, and establish a database to set a blinking frequency and smiling time range. If the blinking frequency and smiling time range are too high or too low during the recognition process, it is judged as an abnormal situation and the smart lock does not unlock. S30205: Based on the above features, the facial muscle features of the current user are analyzed, and the deep learning model is used to analyze the changes in facial muscle tension and identify different muscle movement patterns. When excessive muscle tension and unnatural contraction patterns are detected, the user is judged to be in a tense state. At this time, the smart lock needs further verification. If further verification fails, it will refuse to unlock.

6. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 1, characterized in that: In the internal microscopic state monitoring of the smart lock, the smart lock includes a pressure sensor, an environmental sensor, and a stress color-changing layer detection. The pressure sensor is used to judge whether the smart lock is a normal door opening collision or a malicious external force impact through the data it monitors; the environmental sensor is an environmental sensor array, which is installed inside the smart lock and is used to detect the gas composition, concentration changes and humidity changes in the internal space of the lock; the detection of the stress color-changing layer includes a stress color-changing layer and an optical detection device for detecting the stress color-changing layer, and the stress color-changing layer is applied to the lock core inside the smart lock.

7. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 6, characterized in that: The pressure sensor is installed on both sides and the top of the smart lock tongue. The data collection frequency of the pressure sensor is set to 10-100 times per second, and the collected data is stored in the cloud platform. The specific operation steps include: S30301: Data cleaning, using median filtering to remove noise and outliers in the data collected by the pressure sensor; S30303: Feature extraction, including pressure peak features: Extract the peak features in the pressure data, perform empirical mode decomposition on the pressure data sequence P, obtain n intrinsic mode functions IMF, and for each IMF i Perform Hilbert transform to obtain its instantaneous amplitude A i (t) and instantaneous frequency ω i (t), in each A i (t) Find the maximum value point, the pressure data position corresponding to the maximum value point here is recorded as the peak point, and it is recorded as the normal peak point. Based on this, it is compared with the peak point when the current smart lock is unlocked. When the difference between the current peak point and the normal peak point is within 15%, the current pressure peak feature is judged to be normal, otherwise it is judged to be abnormal; S30304: Pressure change rate feature, calculates the change rate of pressure data, the formula is: where r i is the pressure data change rate sequence; p i is the pressure data; Δt is a constant; where Δt = t i+1 -t i ;t i is time, p i With t i corresponding; The value of the pressure data change rate sequence is averaged and identified as a threshold. Based on the identified threshold, the pressure data change rate sequence value calculated is judged. When it is greater than the threshold by 15% and lower than the threshold by 15%, the current pressure change rate characteristic is judged to be abnormal, otherwise it is judged to be normal. S30305: Prioritize the extracted features, set the pressure peak feature as the first priority, and set the pressure change rate feature as the second priority. When the extracted pressure peak feature is judged to be normal, the smart lock status is directly judged to be normal. When the extracted pressure peak feature is judged to be abnormal, continue to judge through the pressure change rate feature of the second priority. When it is judged to be normal, the smart lock works normally and records the pressure peak feature extracted at this time. When it is judged to be abnormal, it is judged as a serious abnormality and the smart lock cancels subsequent work.

8. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 6, characterized in that: The environmental sensor includes a gas sensor and a humidity sensor. The gas sensor is used to monitor whether the circuit board inside the smart lock generates special gas due to overheating. When special gas is detected, it is judged that the circuit board inside the smart lock is faulty and it is judged as a serious abnormality. The humidity sensor is used to monitor changes in humidity to determine whether liquid has entered the smart lock. When it is determined that there is liquid, it is judged that the circuit board of the smart lock is damp and it is judged as a serious abnormality.

9. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 6, characterized in that: In the detection of the stress-chromic layer, the optical detection device is used to monitor whether the stress-chromic coating applied to the lock core of the smart lock changes color. When the stress-chromic coating changes color, it is judged that the state of the smart lock is abnormal, and it is also judged as a serious abnormality.

10. A cloud platform monitoring method for parallel processing of multiple data for smart locks according to claim 1, characterized in that: In S5, the alarm levels include minor abnormalities, medium abnormalities, and severe abnormalities. Minor abnormalities include insufficient power, online problems, increase and decrease of management personnel, password modification, addition of face, password, fingerprint, and card swiping information; medium abnormalities include multiple errors in unlocking methods; severe abnormalities include changes detected by the smart lock pressure, environment, and optical detection device; Minor anomalies are pushed directly to users through the cloud platform; moderate anomalies are notified to users via text messages; and serious anomalies are notified to users, property management and relevant security departments at the same time.

Images