Passive lock real-time monitoring and predictive maintenance management method, system and terminal device

Abstract

The present application belongs to the field of passive lock state monitoring and intelligent operation and maintenance, and relates to a passive lock real-time monitoring and predictive maintenance management and control method, system and terminal device. The method comprises the following steps: S1, state data acquisition and uploading; S2, edge preprocessing; S3, cloud storage and intelligent analysis; S4, operation and maintenance control execution. The system comprises a terminal layer, an edge layer, a platform layer and an application layer. The terminal device comprises a main control MCU, a multi-dimensional sensor node, a wireless communication module, an energy management module and an anti-disassembly detection unit. After local preprocessing and threshold alarm are performed by the edge computing gateway, the cloud platform is pushed, and combined with weighted score intelligent order allocation, spare parts estimation and three-layer composite permission control, and encryption based on the derivation of a dynamic session key of a lock unique identifier and a timestamp, a full-link management and control system from state sensing to fault early warning to closed-loop operation and maintenance is constructed.

Description

Technical Field

[0001] This invention belongs to the field of passive lock status monitoring and intelligent operation and maintenance technology, and in particular relates to a method, system and terminal device for real-time monitoring and predictive maintenance management of passive locks. It is applicable to application scenarios of multi-dimensional operation status perception, fault early warning and life prediction, and full-process intelligent operation and maintenance management of passive locks. Background Technology

[0002] Passive locks do not require a built-in main power supply and are widely used in control scenarios where external power tampering needs to be prevented. A single system can deploy thousands or even tens of thousands of locks. Existing technologies mainly focus on improving the mechanical structure of the lock body and optimizing the unlocking authentication method. The lock itself does not have the ability to communicate with the outside world or report its status. Maintenance personnel cannot obtain status parameters such as temperature, humidity, vibration, and bolt displacement, nor can they remotely check the health status of the lock.

[0003] Chinese patent document CN116798150A discloses a hardware maintenance method for a passive electronic lock. This method acquires passive power supply records and lock component operation records, extracts and fuses these records, and then inputs them into a lock hardware analysis network to obtain hardware maintenance information and execute corresponding maintenance measures. However, this approach remains a reactive maintenance method after a fault occurs. It cannot provide real-time perception of multi-dimensional states, nor does it establish a fault warning and lifespan prediction model based on historical time-series data. Potential lock faults can only be addressed reactively after they occur, resulting in a lag in maintenance.

[0004] Chinese patent document CN119647778A discloses a preventive maintenance scheduling method and platform for electricity meters driven by energy consumption analysis. It utilizes LSTM deep learning to construct a fault prediction model from historical energy consumption data, collects energy consumption data in real time to predict fault trends and assess remaining service life, and formulates maintenance scheduling plans accordingly. This solution targets metering equipment such as electricity meters, is driven by single energy consumption data, and does not involve multi-dimensional state perception of passive locks. Its maintenance scheduling and operation processes are disconnected, work order dispatch relies on manual experience, and the data transmission and operation authorization links lack dynamic security authentication mechanisms, posing risks of forward data leakage and unauthorized operation.

[0005] In summary, there is an urgent need to provide a method, system, and terminal device for real-time monitoring and predictive maintenance control of passive locks to solve the problem of lack of state awareness. Summary of the Invention

[0006] The technical problem this invention aims to solve is: by integrating multi-dimensional sensor nodes onto a passive lock and periodically encapsulating and encrypting them via a main control MCU, the data is then pushed to a cloud platform after local preprocessing and threshold alarms are performed by an edge computing gateway. The cloud platform's three-level progressive early warning model provides real-time alarms, trend warnings, and lifespan predictions based on the time-series data. Combined with weighted scoring intelligent dispatching, spare parts estimation, and three-level composite access control, as well as dynamic session key derivation encryption based on the lock's unique identifier and timestamp, a full-link management and control system is constructed, from status perception to fault warning to closed-loop operation and maintenance.

[0007] The technical problem to be solved by this invention is achieved by the following technical solution: A method for real-time monitoring and predictive maintenance control of passive locks includes the following steps: S1. Status data acquisition and upload: Deploy multi-dimensional sensor nodes on the passive lock to collect data on the lock's temperature, humidity, vibration, bolt displacement, number of unlocking attempts, and door magnetic status. The main control MCU encapsulates and encrypts the collected data before uploading it to the edge computing gateway via a wireless communication link. S2. Edge Preprocessing: The edge computing gateway receives data reported by the terminal, performs data cleaning, data compression, protocol conversion and edge alarm judgment processing on the data, pushes alarm data exceeding the local alarm threshold to the cloud platform in real time, and uploads regular data to the cloud platform through a secure link after preprocessing. S3, Cloud Storage and Intelligent Analysis: The cloud platform receives data uploaded by the edge computing gateway, writes time-series data into the time-series database, writes status data into the cache, and pushes alarm messages into the message queue; the algorithm engine pulls historical time-series data from the time-series database, extracts multi-dimensional feature parameters through feature engineering, inputs them into a three-level progressive early warning model for analysis, and generates health assessment results, early warning information and lifespan prediction results; S4. Operation and Maintenance Management Execution: Based on health assessment results, early warning information and life prediction results, operation and maintenance management actions are triggered. Operation and maintenance management actions include alarm notification push, automatic generation and intelligent dispatch of maintenance work orders, estimation of spare parts demand, and dynamic control of operation and maintenance permissions, realizing closed-loop management of the entire chain from status perception to fault early warning to intelligent operation and maintenance scheduling.

[0008] This technical solution employs a four-step end-to-end closed-loop process. It leverages multi-dimensional sensor nodes integrated into the passive lock to achieve real-time perception of the lock's physical state. Periodic wake-up of the terminal's main control MCU and vibration energy harvesting enable long-term maintenance-free operation. Data cleaning and local threshold alarm processing are performed by an edge computing gateway before uploading to the cloud platform. The cloud platform's three-tiered progressive early warning model provides real-time alarms, trend warnings, and probabilistic lifespan predictions based on historical time-series data. Combined with weighted scoring intelligent dispatching and three-tiered composite permission control, it constructs a full-link management system from state perception to fault warning to closed-loop maintenance. The entire solution achieves long-term stable operation without requiring an external power supply to the lock, endowing passive locks with multi-dimensional state perception and predictive maintenance capabilities without modifying the existing lock's mechanical structure.

[0009] The three-tiered progressive early warning model includes: a first-tier threshold alarm layer, which communicates with the rule engine and is configured to perform multi-parameter joint rule judgment on multi-dimensional sensor data to generate real-time alarm events; a second-tier trend early warning layer, which connects with the time-series database and is configured to use an LSTM encoder-decoder combined with an attention mechanism to perform reconstruction error analysis on historical time-series data, using the ratio of reconstruction error to a dynamic threshold as an anomaly score, and generating trend early warning information when the anomaly score exceeds a preset judgment threshold; and a third-tier lifetime prediction layer, which connects with the time-series database and is configured to use a method combining the Weibull distribution model and Bayesian dynamic updates, combined with particle filtering to estimate the state of accumulated wear, update the shape and scale parameters of the Weibull distribution online, and output the confidence interval of the remaining lifetime based on a given reliability target.

[0010] This technical solution constructs a three-tiered progressive early warning model, organically combining millisecond-level rule-based alarms, hourly to daily LSTM trend warnings, and weekly to monthly Weibull life predictions to form a complementary fault prediction coverage across the entire time domain. The threshold alarm layer provides second-level response to emergency faults, the trend warning layer identifies potential degradation hours to days in advance, and the life prediction layer quantifies the remaining lifespan of the locks according to probability distribution, enabling maintenance decisions to have multi-level support from immediate response to strategic spare parts procurement.

[0011] This invention also includes a dynamic session key derivation encryption step: Before each data report, the main control MCU derives a session key online based on the lock's unique identifier and the current timestamp using a key derivation algorithm. The session key is then used to symmetrically encrypt the data body, encapsulating it into a data frame containing a device identifier field, a timestamp field, an encrypted data body field, and a message authentication code field, which is then sent via a wireless communication link. After receiving the data frame, the edge computing gateway independently derives a session key based on the same key derivation algorithm and performs decryption and data integrity verification on the data frame. A different session key is used for each report to achieve forward security.

[0012] This technical solution uses a key derivation algorithm based on the lock's unique identifier and timestamp to ensure that each communication uses an independent session key. The terminal and edge sides do not need to transmit the key itself in the network, which fundamentally eliminates the risk of key interception. The key derivation operation is completed on the main control MCU side with extremely short latency, which does not affect the ultra-low power consumption setting target of the terminal device.

[0013] In step S4, the intelligent work order dispatch uses a weighted scoring algorithm to calculate distance score, skill matching score, task load score, and schedule suitability score for candidate maintenance personnel. After weighting and summing according to preset weights, the maintenance personnel with the highest comprehensive score are selected to be assigned work orders. If a work order is not processed within a preset time limit, it will be automatically upgraded to the superior management personnel, forming a closed-loop management process from work order creation to allocation, execution, acceptance and closure.

[0014] This technical solution significantly improves work order response speed and processing efficiency compared to the traditional work order dispatching method that relies on human experience, by comprehensively considering four dimensions of weighted scoring: the geographical location, professional ability, current workload, and scheduling status of maintenance personnel. The automatic escalation mechanism for work orders that have expired ensures that each work order receives a timely response, forming a complete closed-loop management process from work order creation to processing and closure.

[0015] The dynamic management of operation and maintenance permissions in step S4 adopts a three-layer superposition scheme of role-based access control, attribute-based dynamic access control and spatiotemporal fence. The spatiotemporal fence layer determines the operation based on the real-time location information of the operation and maintenance personnel and the operation time window. When all three constraints are met, the corresponding operation is authorized to be executed. The operation permission is automatically invalidated after the operation and maintenance personnel leave the restricted area.

[0016] This technical solution provides basic permission determination through role-based access control, provides fine-grained context determination through attribute-based dynamic access control, and superimposes geographical range and time window constraints through a spatiotemporal fence layer. Compared with single-role permission control, this three-layer superimposed solution can effectively prevent personnel from performing sensitive operations at unauthorized times and locations, and meet the security and compliance requirements of large-scale lock group operation and maintenance.

[0017] In step S1, the multi-dimensional sensor nodes collect data at differentiated acquisition frequencies. Vibration data is sampled at a high frequency period and uploaded at a preset reporting period. Temperature data is sampled at a low frequency period and uploaded at a preset reporting period. Humidity data and latch displacement data are uploaded at an even lower frequency period. Door magnetic status data and unlocking count data are reported in real time when an event is triggered. When an abnormal event is detected, the system automatically switches to high-frequency acquisition mode and reports all sensor data in real time.

[0018] This technical solution effectively reduces the average power consumption and wireless communication duty cycle of the terminal device by adopting a differentiated acquisition frequency strategy, while ensuring the integrity of critical status data acquisition. The event-triggered real-time reporting mechanism ensures that critical actions such as door magnetic sensors and unlocking events are not missed, and the high-frequency acquisition mode of all sensors triggered by abnormal events provides high-resolution data for subsequent root cause analysis of faults.

[0019] This invention also provides a passive lock real-time monitoring and predictive maintenance control system, comprising: The terminal layer includes multiple lock status sensing terminal devices deployed on passive locks. Each terminal device includes a multi-dimensional sensor node, a main control MCU, a wireless communication module, and an energy management module. The multi-dimensional sensor node is signal-connected to the main control MCU, the main control MCU is electrically connected to the wireless communication module, and the energy management module is electrically connected to both the main control MCU and the wireless communication module. The edge layer includes multiple edge computing gateways that communicate with the terminal layer, each edge computing gateway communicating with a group of terminal devices under its jurisdiction; The platform layer, deployed on a cloud server, communicates and connects with the edge layer, including a data access gateway, time-series database, cache, message queue, algorithm engine, and rule engine. The algorithm engine has a built-in three-level progressive early warning model. The application layer connects with the platform layer, including the work order scheduling center and the operation and maintenance management backend. It is configured to trigger alarm notifications based on health assessment results, early warning information and life prediction results, automatically generate and intelligently dispatch maintenance work orders, estimate spare parts demand, and dynamically control operation and maintenance permissions.

[0020] This technical solution adopts a modular system setup with a four-layer architecture of end-edge-cloud-application, mapping data acquisition, local preprocessing, cloud-based intelligent analysis, and operation and maintenance management closed loop to independent physical and functional layers respectively. The terminal layer achieves long-term maintenance-free operation through ultra-low power consumption and vibration energy harvesting; the edge layer maintains millisecond-level response capability in network interruption scenarios through local alarm judgment; the platform layer achieves full-time fault prediction through a three-level early warning model; and the application layer achieves operation and maintenance closed loop through work order scheduling and access control. The entire system achieves optimal synergy between data security, computing efficiency, and operation and maintenance response capability.

[0021] The platform-level algorithm engine has a built-in three-tiered progressive early warning model, including: a threshold alarm module, which communicates with the rule engine and is configured to perform multi-parameter joint rule judgment on multi-dimensional sensor data to generate real-time alarm events; a trend early warning module, which connects with the time-series database and is configured to use an LSTM encoder-decoder combined with an attention mechanism to reconstruct error analysis of historical time-series data and generate trend early warning information; and a lifespan prediction module, which connects with the time-series database and is configured to use a method combining the Weibull distribution model and Bayesian dynamic updates, combined with particle filtering, to estimate the state of accumulated wear and output the probability distribution prediction result of the remaining lifespan.

[0022] This technical solution decomposes the three-level early warning model into three independent functional modules: a threshold alarm module, a trend early warning module, and a lifespan prediction module. These modules are connected to the rule engine and the time series database, respectively, allowing each module to be deployed and expanded independently. The three modules respectively undertake early warning analysis tasks at the real-time, predictive, and strategic levels, and cooperate with each other to form a complementary fault prediction capability across the entire time domain.

[0023] The main control MCU in the terminal layer is configured to switch between deep sleep mode and active mode, and is triggered to wake up by a real-time clock. The energy management module includes a lithium manganese battery main power supply and a vibration energy harvesting auxiliary power supply. The vibration energy harvesting auxiliary power supply is configured to convert environmental vibration energy and store it in the energy storage element. When the lithium manganese battery main power supply is lower than a preset threshold, the energy storage element will supplement the system with power. The terminal device can achieve long-term maintenance-free operation without replacing the battery.

[0024] This technical solution uses a deep sleep and real-time clock wake-up mechanism to keep the terminal device in an extremely low power consumption state most of the time, only waking it up briefly during the acquisition and communication window, resulting in extremely low overall average power consumption. The vibration energy harvesting auxiliary power supply makes full use of the vibration of the on-site environment to replenish the terminal device with power, further extending the service life of the main power supply, enabling the terminal device to achieve long-term self-sustaining operation without external power supply.

[0025] The present invention also provides a passive lock real-time monitoring and predictive maintenance control terminal device, comprising: The main control MCU is configured to alternate between deep sleep mode and active mode, and is triggered to wake up by a real-time clock. The multi-dimensional sensor nodes are connected to the main control MCU and are configured to collect data on the lock's temperature, humidity, vibration, bolt displacement, door magnetic status and unlocking count. The wireless communication module is electrically connected to the main control MCU and is configured to upload the encrypted data processed by the main control MCU to the edge computing gateway through the wireless communication link. The energy management module is electrically connected to the main control MCU and the wireless communication module, and is configured to provide power to each module. The tamper detection unit is electrically connected to the main control MCU. The tamper detection unit includes a photosensitive sensor and a micro switch. When the tamper detection unit detects that the outer shell has been removed, it sends a trigger signal to the main control MCU. The main control MCU responds to the trigger signal and performs a key erasure operation.

[0026] This technical solution integrates multi-dimensional sensor nodes and anti-tamper detection units to achieve full-dimensional status perception and physical security protection of locks without relying on external power supply. The anti-tamper detection and key erasure linkage mechanism ensures that the terminal device can automatically destroy internal key materials when subjected to physical attacks, preventing the key from being illegally extracted through hardware. Combined with deep sleep and vibration energy harvesting settings, the terminal device can operate stably in the field of passive locks for a long time.

[0027] The beneficial effects of this technical solution are: (1) Filling the gap in passive lock status monitoring: By integrating six types of multi-dimensional sensor nodes, including temperature, humidity, vibration, lock tongue displacement, door magnet and unlocking count, into the passive lock, the key status parameters of the passive lock are fully and real-time perceived for the first time. This fundamentally fills the technical gap of the unknown status of the existing passive lock and provides a reliable source of multi-dimensional feature data for subsequent fault warning and life prediction.

[0028] (2) Realize full-time fault early warning and predictive maintenance: The three-level progressive early warning model organically combines millisecond-level threshold alarm, hourly to daily LSTM trend early warning and weekly to monthly Weibull life prediction, so that potential lock failures can be predicted several days to several weeks before they occur. Compared with the traditional regular maintenance strategy, it significantly reduces operation and maintenance costs and improves lock availability.

[0029] (3) Construct an end-to-end security protection system: By using a dynamic session key derivation encryption scheme based on the unique identifier and timestamp of the lock, combined with role-based access control, attribute-based dynamic access control and spatiotemporal fence three-layer composite permission control, an end-to-end security protection system from data transmission to operation authorization is constructed; the dynamic session key is different each time it is reported to achieve forward security, and the three-layer permission superposition ensures that the time, location and role constraints of operation and maintenance operations are met simultaneously before execution.

[0030] (4) Achieve ultra-low power consumption and long-term operation: Through the alternating operation mechanism of deep sleep and real-time clock wake-up of the main control MCU, combined with vibration energy harvesting to assist power supply, the terminal device is in deep sleep for most of the time, and the overall average power consumption is extremely low, so that the passive lock terminal device can achieve long-term stable operation without replacing the battery, fundamentally solving the engineering problem of frequent maintenance and battery replacement of active sensing terminals.

[0031] (5) Achieve full-process closed-loop intelligent operation and maintenance: By adopting a multi-dimensional weighted scoring intelligent dispatch algorithm to comprehensively evaluate the four dimensions of distance, skills, load and scheduling, and introducing an automatic upgrade mechanism for work order timeout, the operation and maintenance work orders are accurately and efficiently allocated and the entire process is closed-loop controlled. Compared with the traditional dispatch method that relies on human experience, the operation and maintenance response time is significantly shortened. Attached Figure Description

[0032] Figure 1 This is a block diagram illustrating the principle of the method of the present invention.

[0033] Figure 2 This is a structural diagram of the lock cylinder assembly.

[0034] Figure 3 This is one of the cross-sectional views of the lock cylinder assembly.

[0035] Figure 4 This is the second sectional view of the lock cylinder assembly.

[0036] Figure 5 This is a structural diagram of the key assembly.

[0037] Figure 6 This is one of the cross-sectional views of the key assembly.

[0038] Figure 7 This is the second cross-sectional view of the key assembly.

[0039] In the diagram: 1. Lock cylinder housing; 11. End cap; 12. Probe housing I; 13. Lock cylinder probe I; 2. Lock cylinder cover plate; 3. Permanent magnet mechanism; 31. PCB board I; 32. Shock-absorbing pad; 33. Fixed shaft; 34. Enamelled wire; 35. Coil frame; 36. Sealing ring; 4. Key probe; 41. Power probe; 42. Metal limit block; 43. Communication probe; 44. Upper cover plate; 45. Lower cover plate; 5. Charging port. Detailed Implementation

[0040] Example 1 Reference Figure 1The method of this invention adopts a four-layer architecture of end-edge-cloud-application, consisting of a terminal layer, an edge layer, a platform layer, and an application layer from bottom to top. The terminal layer includes multiple lock status sensing terminal devices deployed on passive locks. Multi-dimensional sensor nodes are connected to the main control MCU to collect multi-dimensional physical state parameters of the lock, such as temperature, humidity, vibration, bolt displacement, door magnetic status, and unlocking frequency. The main control MCU encapsulates and encrypts the collected multi-dimensional data, and then transmits it upwards to the edge computing gateway of the edge layer via a wireless communication link through a wireless communication module. The edge layer includes multiple edge computing gateways deployed in the field. Each edge computing gateway communicates with a group of terminal devices under its jurisdiction. It sequentially performs data cleaning, data compression, protocol conversion, and edge alarm judgment processing on the received data. Alarm data exceeding the local alarm threshold is pushed to the cloud platform in real time, while regular data is uploaded to the cloud platform after preprocessing. The local alarm judgment of the edge layer can maintain millisecond-level alarm response capability even during network interruptions and supports local caching to cope with network interruption scenarios.

[0041] The data flow process during the execution of the above method is as follows: At the terminal layer, multi-dimensional sensor nodes collect lock status data according to their respective acquisition strategies. The main control MCU encapsulates and encrypts the data before reporting it to the edge computing gateway via the wireless communication module. At the edge layer, the edge computing gateway performs CRC verification, data cleaning, compression encoding, and local threshold alarm determination on the received data. It pushes alarm data to the cloud platform in real time and uploads regular data to the cloud platform after protocol conversion. At the platform layer, the data access gateway completes protocol parsing and routing distribution. The algorithm engine performs feature extraction and three-level early warning analysis on the time-series data, generating alarm events, trend early warning information, and lifespan prediction results. At the application layer, the early warning results are pushed to the mobile terminals of maintenance personnel and the monitoring dashboard via WebSocket, triggering work order generation and dispatch processes. Because the edge layer undertakes data cleaning and local threshold determination, the amount of uplink data is reduced while ensuring the real-time nature of local alarms.

[0042] The specific selection and installation locations of the multi-dimensional sensor nodes are shown in Table 1:

[0043] A temperature and humidity sensor is installed on the inner wall of the lock body and connected to the main control MCU via an I2C bus to collect temperature and humidity data inside the lock body. A vibration sensor is rigidly fixed to the center of the bottom of the lock body and connected to the main control MCU via an SPI bus to collect vibration acceleration data of the lock body using high-frequency sampling. A bolt displacement sensor uses the Hall effect non-contact measurement principle and is connected to the main control MCU via an analog-to-digital converter interface to continuously measure the displacement of the permanent magnet at the tail of the bolt. A door magnetic sensor is installed between the cabinet door and the door frame and is connected to the main control MCU via a general-purpose input / output interface, triggering an event report when the door's open / closed state changes. An unlocking counter sensor counts the number of unlocking operations by detecting changes in the magnetic field of the lock cylinder's rotating shaft. It is also connected to the main control MCU via a general-purpose input / output interface and reports a count event in real time each unlocking action occurs. Each sensor uses I2C, SPI, analog-to-digital converter, and general-purpose input / output interfaces according to its signal characteristics, optimizing the peripheral resource usage of the main control MCU while ensuring data acquisition accuracy.

[0044] Before performing the three-level early warning analysis, the algorithm engine first performs feature engineering on the raw sensor data to extract multi-dimensional feature parameters. For vibration data, the following time-domain and frequency-domain features are extracted: Root mean square value of vibration signal:

[0045] In the formula: The root mean square value of the vibration signal. The number of sampling points within a sampling window. For the first The acceleration values ​​at each sampling point.

[0046] Peak factor:

[0047] In the formula: For peak factor, The peak value within the sampling window is the absolute value of the peak value. The peak factor is used to characterize the content of the impact component.

[0048] Kurtosis value:

[0049] In the formula: Kurtosis value The mean, 1 is the standard deviation, and kurtosis is used to measure the sharpness of the probability distribution of a signal.

[0050] Waveform factor:

[0051] In the formula: This is a waveform factor used to describe the waveform characteristics of a signal.

[0052] Pulse factor:

[0053] In the formula: It is the pulse factor, used to detect transient impulse components in a signal.

[0054] For frequency domain characteristics, the spectral distribution is obtained by performing a fast Fourier transform on the vibration signal:

[0055] Based on this, the frequency band energy ratio and centroid frequency are calculated: Frequency band energy percentage:

[0056] In the formula: It represents the ratio of energy within the target frequency band to the total energy.

[0057] Center of gravity frequency:

[0058] In the formula: The centroid frequency of the power spectrum reflects the concentration of vibrational energy in the frequency domain.

[0059] For temperature data, the rate of temperature change and the temperature autocorrelation coefficient are extracted as degradation features: First-order rate of change of temperature:

[0060] Second rate of change of temperature:

[0061] In the formula: The sampling interval is given by the first derivative, which reflects the rate of temperature change, and the second derivative, which reflects the acceleration of the temperature change.

[0062] Temperature series autocorrelation coefficient:

[0063] In the formula: This is the autocorrelation coefficient of the temperature series, used to determine whether there is a trend of drift in temperature changes.

[0064] Based on the above multidimensional feature parameters, the algorithm engine calculates a comprehensive wear index to quantify the overall health of the lock:

[0065] In the formula: To accumulate the comprehensive wear index, For the first The number of unlocks per statistical period. The wear coefficient per unit of unlocking operation. For the first Environmental severity correction coefficient for each cycle (taking into account the weighted effects of environmental factors such as high temperature, high humidity, and strong vibration).

[0066] The cloud platform's algorithm engine incorporates a three-tiered progressive early warning model. The first tier, the threshold alarm layer, communicates with the rule engine, receiving real-time multi-dimensional sensor data uploaded from the edge computing gateway. The rule engine then executes multi-parameter joint rule judgments, pre-configured with alarm rules for types such as excessively high or low temperature, excessively high humidity, abnormal vibration, abnormal door magnet, unauthorized unlocking, stuck lock tongue, communication interruption, low voltage, and unlocking abnormalities. When one or more parameters simultaneously trigger the rule conditions, the rule engine generates a real-time alarm event and pushes it to the application layer.

[0067] The parameter configurations for each layer of the three-tiered progressive early warning model are shown in Table 2:

[0068] The second-layer trend warning layer connects to the time-series database and employs a two-layer bidirectional LSTM encoder-decoder architecture combined with an attention mechanism to reconstruct and analyze errors in historical time-series data. The core computational units of the LSTM network include four parts: the forget gate, the input gate, the cell state update gate, and the output gate. The specific calculation formulas are as follows: Forgot Gate Output:

[0069] In the formula: This is the output vector of the forget gate. It is the sigmoid activation function. Here is the forget gate weight matrix. This is the hidden state from the previous moment. Enter the current time. This is the forget gate bias term.

[0070] Input gate output:

[0071] In the formula: The output vector of the input gate. The input gate weight matrix, This is the input gate bias term.

[0072] Candidate cell status:

[0073] In the formula: Let be the candidate cell state vector. The candidate state weight matrix, This is a bias term.

[0074] Cell status update:

[0075] In the formula: This represents the current state of the cell. This represents element-wise multiplication, and the formula enables the fusion and updating of cell states by forgetting them and integrating new information.

[0076] Output gate output:

[0077] In the formula: The output vector of the output gate. This is the output gate weight matrix. This is the output gate bias term.

[0078] Hidden state output:

[0079] In the formula: The current state is hidden and output. The cell state after being modulated by the output gate is the information passed to the next time step.

[0080] The attention mechanism assigns differentiated attention weights to the hidden states at each time step, highlighting the time steps that contribute more to anomaly detection: Attention score:

[0081] In the formula: For the first The attention score scalar for each time step. and These represent the weights and biases of the attention layer, respectively.

[0082] Normalized attention weights:

[0083] In the formula: The attention weights are normalized, and the sum of the weights for all time steps is 1.

[0084] Weighted context vector:

[0085] In the formula: The weighted context vector aggregates key information from all time steps for use by the decoder.

[0086] Anomaly detection is based on the reconstruction error of the LSTM encoder-decoder: Reconstruction error:

[0087] In the formula: For the input sequence Reconstruction error, For the reconstructed output sequence of the decoder, This represents the square of the Euclidean norm.

[0088] Dynamic threshold:

[0089] In the formula: The dynamic threshold is set to the 99.9 percentile of the reconstruction error of the validation set, and is adaptively adjusted to accommodate the normal fluctuation range of different individual locks.

[0090] Anomaly rating:

[0091] In the formula: For anomaly scoring, when When the value exceeds a preset threshold (usually 3 to 5), a trend warning message is generated.

[0092] The third-layer lifespan prediction layer connects to the time-series database and uses a Weibull distribution reliability model to model the lock failure probability. The probability density function, reliability function, and failure rate function of the Weibull distribution are defined as follows: Probability density function:

[0093] In the formula: Let be the probability density function of the failure time. For shape parameters ( ), For scale parameters ( (Also known as characteristic lifetime). When This indicates a decrease in early failure rate. Indicates random failure. This indicates the stage of consumption and loss of effectiveness.

[0094] Reliability function:

[0095] In the formula: Let be a reliability function, representing the lock's reliability over time. The probability that no failure will occur.

[0096] Failure rate function:

[0097] In the formula: The instantaneous failure rate function, or hazard function, describes the failure rate of the lock after it has been operating normally for a certain period of time. The conditional probability density of failure occurring in the next instant under the given conditions.

[0098] The Bayesian dynamic update method updates the posterior distribution of the Weibull distribution parameters online each time new running data is obtained: Prior distribution:

[0099] Likelihood function:

[0100] Posterior distribution:

[0101] In the formula: Let be the vector of parameters to be estimated. For the observed lifetime dataset, the prior distribution Based on historical statistical data of similar locks, the likelihood function is set. Based on observed data, the posterior distribution is calculated. It integrates prior knowledge with the latest observational information.

[0102] Based on the updated Weibull distribution parameters, calculate the given reliability target. Remaining useful life (RUL) below:

[0103] In the formula: This represents the current cumulative running time of the lock. For a set reliability target value (usually 0.9 or 0.95), the formula outputs a point estimate of the RUL. In practical applications, Monte Carlo sampling is used to obtain the RUL from the posterior distribution. Parameter samples are extracted from the data, and the confidence interval of RUL is calculated.

[0104] The particle filtering method is used to perform nonlinear, non-Gaussian state estimation of the cumulative wear of locks: Wear condition estimate:

[0105] In the formula: for The state estimate of wear at any given time. For the number of particles, For the first Normalized weights of individual particles, For the first The state values ​​of each particle.

[0106] Particle weight recursive update:

[0107] In the formula: particle weights are updated recursively according to the importance sampling principle. Let be the likelihood function. for The observed value at time 1. When the effective number of particles falls below a threshold, a resampling operation is performed to prevent particle degradation.

[0108] The hardware architecture of the lock status sensing terminal device includes a main control MCU (such as the STM32L4 series, Cortex-M4 core), a multi-dimensional sensor interface module, a wireless communication module (supporting NB-IoT / LoRa / BLE modes), a vibration energy harvesting unit, a lithium manganese battery main power supply (such as CR2450, 620mAh), a key derivation and encryption module (such as the ATECC608B security chip), and an anti-tamper detection unit. The main control MCU is configured to alternate between deep sleep mode and active mode. In deep sleep mode, most peripherals are powered off, with only the real-time clock running. The real-time clock triggers a wake-up interrupt periodically according to a preset acquisition cycle. The main control MCU switches from deep sleep mode to active mode to complete sensor data acquisition, data encapsulation and encryption, and wireless communication reporting before re-entering deep sleep mode. The terminal device is in deep sleep mode for the vast majority of the time, completing data acquisition and communication only within a very short active window. The overall average power consumption is extremely low, enabling the terminal device to operate for extended periods without maintenance using its built-in battery.

[0109] The platform layer employs a hybrid storage architecture for data storage. The data access gateway routes and distributes received data according to its type: sensor time-series data is written to a time-series database (such as TDengine) to support efficient writing and time-range queries of high-frequency time-series data; business data such as equipment ledgers, work order records, and permission configurations are written to a relational database (such as PostgreSQL); file and log data are stored in object storage (such as MinIO); and real-time lock status snapshots are written to a cache (such as Redis Cluster) for real-time reading by the application layer. This hybrid storage architecture ensures that different types of data are processed by the storage engine best suited to their access patterns, reducing storage costs while maintaining query efficiency.

[0110] In the dynamic session key derivation encryption process, before each data report, the terminal device derives a session key online using a key derivation algorithm based on the lock's unique identifier and the current timestamp. This session key is then used to symmetrically encrypt the collected data. The encrypted data, along with the device identifier field, timestamp field, and message authentication code field, is encapsulated into a data frame and sent to the edge computing gateway via the wireless communication module. Upon receiving the data frame, the edge computing gateway independently derives the same session key using the same key derivation algorithm, decrypts the data frame, and verifies the message authentication code to obtain the plaintext data. Because each report derives a different session key based on a different timestamp, even if the session key for one communication is intercepted, it will not lead to the decryption of historical data from other communications, thus achieving forward security.

[0111] In the intelligent work order dispatch algorithm process, when the platform layer generates an early warning message or a health assessment result triggers a maintenance request, the work order dispatch center automatically generates a maintenance work order and obtains a list of candidate maintenance personnel. The intelligent work order dispatch uses a weighted scoring algorithm, and the formula for calculating the comprehensive score is as follows:

[0112] in, For comprehensive scoring, These are the weighting coefficients for each score item, with values ​​as follows: , , , .

[0113] The various indicators and calculation methods of the weighted scoring for intelligent order dispatch are shown in Table 3:

[0114] Distance score reflects the geographical proximity of candidate maintenance personnel to the faulty equipment:

[0115] In the formula: This represents the straight-line distance between the current location of the maintenance personnel and the faulty equipment. The reference distance constant (which can be set to 5000 meters) is used. The closer the distance, the higher the score, and the score exhibits an exponential decay characteristic.

[0116] Skills matching score reflects the degree to which the qualifications of operations and maintenance personnel match the skills required for the work order:

[0117] In the formula: For the first The matching flag for each skill (1 for a match, 0 for no match). For the first The weight of each skill This represents the total number of skills required.

[0118] The load score reflects the current workload level of operations and maintenance personnel:

[0119] In the formula: This represents the number of unfinished work orders currently being processed by operations and maintenance personnel. The score is determined by the maximum number of concurrent work orders a worker can handle; the lighter the load, the higher the score.

[0120] The scheduling suitability score is directly assigned based on the current shift status of the operations and maintenance personnel: (In class); (work overtime); (rest) The maintenance personnel with the highest overall score will be selected to assign work orders. If a work order is not processed within a preset time limit, it will be automatically escalated to a higher-level administrator.

[0121] The application-layer access control module performs a three-layer stacked access check on maintenance operation requests. The first layer is role-based access control, determining whether the maintenance personnel have the basic permissions to perform the corresponding operation based on their role type. The second layer is attribute-based dynamic access control, comprehensively considering dynamic attributes such as device type, region, time, and task urgency to perform fine-grained judgment on the operation request. The third layer is spatiotemporal fence verification, judging based on the maintenance personnel's real-time location information and the work time window; verification only passes if the maintenance personnel are within the specified geographical area and within the authorized time window. Only when all three layers of verification pass will the access control module authorize the corresponding operation; operation permissions automatically expire when the maintenance personnel leave the spatiotemporal fence-defined area.

[0122] Example 2 This embodiment provides a four-layer architecture and collaborative workflow of key modules in a passive lock real-time monitoring and predictive maintenance control system. The system as a whole includes a terminal layer, an edge layer, a platform layer, and an application layer. The terminal layer consists of multiple lock status sensing terminal devices deployed on the passive lock. Each terminal device includes a multi-dimensional sensor node, a main control MCU, a wireless communication module, and an energy management module. The multi-dimensional sensor node is signal-connected to the main control MCU, the main control MCU is electrically connected to the wireless communication module, and the energy management module is electrically connected to both the main control MCU and the wireless communication module. The edge layer includes multiple edge computing gateways that communicate with the terminal layer. Each edge computing gateway communicates with a group of terminal devices under its jurisdiction, performing data preprocessing, local alarm determination, protocol conversion, and buffer forwarding processing on the received data.

[0123] The platform layer, deployed on a cloud server, communicates with the edge layer and includes a data access gateway, a time-series database, a cache, a message queue, an algorithm engine, and a rule engine. The data access gateway receives data uploaded from the edge computing gateway and routes it according to data type: time-series sensor data is written to the time-series database, status data is written to the cache, and alarm messages are pushed into the message queue. The algorithm engine retrieves historical time-series data from the time-series database, extracts multi-dimensional feature parameters such as temperature, humidity, vibration, and latch displacement through feature engineering, and inputs these parameters into a built-in three-tiered progressive early warning model for analysis, generating health assessment results, early warning information, and lifespan prediction results. The application layer connects with the platform layer, including a work order scheduling center and an operations and maintenance management backend. The work order scheduling center receives early warning information and health assessment results from the platform layer, triggering automatic maintenance work order generation and intelligent dispatch processes. The operations and maintenance management backend provides functions such as alarm notification push, spare parts demand estimation, and dynamic control of operations and maintenance permissions, achieving end-to-end operations and maintenance management from status awareness to closed-loop handling.

[0124] The main control MCU in the terminal layer is configured to alternate between deep sleep mode and active mode, and is triggered to wake up periodically by a real-time clock. The energy management module includes a lithium manganese battery main power supply and a vibration energy harvesting auxiliary power supply. The vibration energy harvesting auxiliary power supply is configured to convert environmental vibration energy and store it in an energy storage element. When the lithium manganese battery main power supply is lower than a preset threshold, the energy storage element replenishes power to the system, enabling the terminal device to operate for a long time without battery replacement. The tamper detection unit is electrically connected to the main control MCU and includes a photosensitive sensor and a micro switch. When the tamper detection unit detects that the outer casing has been removed, it sends a trigger signal to the main control MCU. The main control MCU responds to the trigger signal by performing a key erasure operation to prevent the key from being illegally extracted through hardware.

[0125] The platform's algorithm engine incorporates a three-tiered progressive early warning model: a threshold alarm module, connected to the rule engine, configured to perform multi-parameter joint rule judgments on multi-dimensional sensor data to generate real-time alarm events; a trend early warning module, connected to the time-series database, configured to use an LSTM encoder-decoder combined with an attention mechanism to reconstruct error analysis of historical time-series data and generate trend early warning information; and a lifespan prediction module, also connected to the time-series database, configured to use a combination of the Weibull distribution model and Bayesian dynamic updates, combined with particle filtering, to estimate the state of accumulated wear and output the probability distribution prediction of remaining lifespan. These three functional modules each perform their respective tasks, undertaking early warning analysis tasks at the real-time, predictive, and strategic levels, respectively, and collaborating to form a complementary fault prediction coverage across the entire time domain.

[0126] Example 3 Reference Figures 2 to 4This embodiment illustrates the specific structure of the lock cylinder assembly. The lock cylinder assembly includes a lock cylinder housing 1, a lock cylinder cover plate 2, an end cap 11, a probe housing I 12, a lock cylinder probe I 13, and a permanent magnet mechanism 3. The lock cylinder housing 1 and the lock cylinder cover plate 2 cooperate to encapsulate and form the housing of the lock cylinder assembly. The end cap 11 is disposed at the end of the lock cylinder housing 1 and is used to close the end opening of the lock cylinder housing 1. The permanent magnet mechanism 3 is disposed inside the lock cylinder assembly and includes a PCB board I 31, a shock-absorbing pad 32, a fixing shaft 33, an enameled wire 34, a coil frame 35, and a sealing ring 36. The PCB board I 31 integrates the main control MCU, wireless communication module, key derivation and encryption module, and battery management unit. A shock-absorbing pad 32 is placed between the PCB board I 31 and the lock cylinder housing 1 to buffer the impact of external vibrations on the PCB board I 31. A fixed shaft 33 is used to fix the coil frame 35 inside the lock cylinder assembly. Enamelled wire 34 is wound around the coil frame 35 to form a non-contact power coupling unit. When the key assembly is inserted, the coil frame 35 and the corresponding coil on the key side achieve wireless energy transfer through electromagnetic induction, supplementing power to the terminal device. A sealing ring 36 is placed at the mating point between the end cover 11 and the lock cylinder housing 1 to prevent external moisture from entering the lock cylinder assembly. The probe housing I 12 embeds the lock cylinder probe I 13. The lock cylinder probe I 13 is connected to the main control MCU on the PCB board I 31 via an I2C bus or a general-purpose input / output interface. It is configured to dock with the key probe 4 when the key assembly is inserted, achieving reliable contact communication and energy channel conduction.

[0127] Reference Figures 5 to 7 The key assembly includes a key probe 4 and a charging port 5. The key probe 4 includes a power probe 41, a metal limiting block 42, a communication probe 43, an upper cover plate 44, and a lower cover plate 45. The upper cover plate 44 and the lower cover plate 45 cooperate to encapsulate the internal structure of the key probe 4. The power probe 41 is located at the center of the insertion end of the key probe 4, configured to reliably connect with the lock cylinder probe I13 when the key assembly is inserted into the lock cylinder assembly, simultaneously establishing power and data paths. The communication probe 43 is symmetrically distributed around the power probe 41, configured to establish redundant communication links with the PCB board I31 to improve contact reliability. The metal limiting block 42 is located at the insertion end of the key probe 4, configured to limit the insertion depth of the key probe 4, preventing plastic deformation of the probe after long-term insertion and removal. The charging port 5 is located at the rear of the key assembly, serving as an external charging and data maintenance interface for the key assembly when not inserted.

[0128] The lock cylinder assembly and key assembly work together to form an integrated structure, embedding monitoring functions within the lock body without altering the existing mechanical operation of the lock. Energy replenishment and data communication are achieved through the docking of lock cylinder probe I13 and key probe 4. The lock cylinder assembly shell is made of PC+ABS alloy material (V-0 flame retardant) and meets the IP67 protection rating, making it suitable for long-term operation in harsh environments. The entire integrated structure retains the original mechanical operation functions of the lock while endowing it with multi-dimensional status perception and remote communication capabilities, realizing the organic integration of passive locks and IoT monitoring systems.

[0129] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, and are not intended to limit the specific implementation of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the claims of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A method for real-time monitoring and predictive maintenance control of passive locks, characterized in that, Includes the following steps: S1. Status data acquisition and upload: Deploy multi-dimensional sensor nodes on the passive lock to collect data on the lock's temperature, humidity, vibration, bolt displacement, number of unlocking attempts, and door magnetic status. The main control MCU encapsulates and encrypts the collected data before uploading it to the edge computing gateway via a wireless communication link. S2. Edge Preprocessing: The edge computing gateway receives data reported by the terminal, performs data cleaning, data compression, protocol conversion and edge alarm judgment processing on the data, pushes alarm data exceeding the local alarm threshold to the cloud platform in real time, and uploads regular data to the cloud platform through a secure link after preprocessing. S3, Cloud Storage and Intelligent Analysis: The cloud platform receives data uploaded by the edge computing gateway, writes time-series data into the time-series database, writes status data into the cache, and pushes alarm messages into the message queue; the algorithm engine pulls historical time-series data from the time-series database, extracts multi-dimensional feature parameters through feature engineering, inputs them into a three-level progressive early warning model for analysis, and generates health assessment results, early warning information and lifespan prediction results; S4. Operation and Maintenance Management Execution: Based on health assessment results, early warning information and life prediction results, operation and maintenance management actions are triggered. Operation and maintenance management actions include alarm notification push, automatic generation and intelligent dispatch of maintenance work orders, estimation of spare parts demand, and dynamic control of operation and maintenance permissions, realizing closed-loop management of the entire chain from status perception to fault early warning to intelligent operation and maintenance scheduling.

2. The method for real-time monitoring and predictive maintenance control of passive locks as described in claim 1, characterized in that: The status data acquisition and uploading in S1 also includes a dynamic session key derivation encryption step: Before each data report, the main control MCU derives a session key online based on the lock's unique identifier and the current timestamp using a key derivation algorithm. The session key is then used to symmetrically encrypt the data body, encapsulating it into a data frame containing a device identifier field, a timestamp field, an encrypted data body field, and a message authentication code field, which is then sent via a wireless communication link. After receiving the data frame, the edge computing gateway independently derives a session key based on the same key derivation algorithm and performs decryption and data integrity verification on the data frame. A different session key is used for each report to achieve forward security.

3. The method for real-time monitoring and predictive maintenance control of passive locks as described in claim 1, characterized in that: The multi-dimensional sensor nodes in S1 collect data at differentiated acquisition frequencies. Vibration data is sampled at a high frequency period and uploaded at a preset reporting period. Temperature data is sampled at a low frequency period and uploaded at a preset reporting period. Humidity data and latch displacement data are uploaded at an even lower frequency period. Door magnetic status data and unlocking count data are reported in real time when an event is triggered. When an abnormal event is detected, the system automatically switches to high-frequency acquisition mode and reports all sensor data in real time.

4. The method for real-time monitoring and predictive maintenance control of passive locks as described in claim 1, characterized in that: In S3, the three-level progressive early warning model includes: a first-level threshold alarm layer, which is connected to the rule engine and configured to perform multi-parameter joint rule judgment on multi-dimensional sensor data to generate real-time alarm events; a second-level trend early warning layer, which is connected to the time-series database and configured to use an LSTM encoder-decoder combined with an attention mechanism to perform reconstruction error analysis on historical time-series data, using the ratio of reconstruction error to a dynamic threshold as an anomaly score, and generating trend early warning information when the anomaly score exceeds a preset judgment threshold; and a third-level lifetime prediction layer, which is connected to the time-series database and configured to use a method combining the Weibull distribution model and Bayesian dynamic updates, combined with particle filtering to estimate the state of accumulated wear, update the shape and scale parameters of the Weibull distribution online, and output the confidence interval of the remaining lifetime based on a given reliability target.

5. The method for real-time monitoring and predictive maintenance control of passive locks as described in claim 1, characterized in that: The intelligent work order dispatching in S4 uses a weighted scoring algorithm to calculate distance score, skill matching score, task load score, and schedule suitability score for candidate maintenance personnel. After weighting and summing according to preset weights, the maintenance personnel with the highest comprehensive score are selected to be assigned work orders. If a work order is not processed within a preset time limit, it will be automatically upgraded to the superior management personnel, forming a closed-loop management process from work order creation to allocation, execution, acceptance and closure.

6. The method for real-time monitoring and predictive maintenance control of passive locks as described in claim 1, characterized in that: The dynamic management of operation and maintenance permissions in S4 adopts a three-layer superposition scheme of role-based access control, attribute-based dynamic access control, and spatiotemporal fence. The spatiotemporal fence layer determines the operation based on the real-time location information of the operation and maintenance personnel and the operation time window. When all three constraints are met, the corresponding operation is authorized to be executed. The operation permission is automatically invalidated after the operation and maintenance personnel leave the restricted area.

7. A passive lock real-time monitoring and predictive maintenance control system, employing the passive lock real-time monitoring and predictive maintenance control method as described in any one of claims 1-6, characterized in that, include: The terminal layer includes multiple lock status sensing terminal devices deployed on passive locks. Each terminal device includes a multi-dimensional sensor node, a main control MCU, a wireless communication module, and an energy management module. The multi-dimensional sensor node is signal-connected to the main control MCU, the main control MCU is electrically connected to the wireless communication module, and the energy management module is electrically connected to both the main control MCU and the wireless communication module. The edge layer includes multiple edge computing gateways that communicate with the terminal layer, each edge computing gateway communicating with a group of terminal devices under its jurisdiction; The platform layer, deployed on a cloud server, communicates and connects with the edge layer, including a data access gateway, time-series database, cache, message queue, algorithm engine, and rule engine. The algorithm engine has a built-in three-level progressive early warning model. The application layer connects with the platform layer, including the work order scheduling center and the operation and maintenance management backend. It is configured to trigger alarm notifications based on health assessment results, early warning information and life prediction results, automatically generate and intelligently dispatch maintenance work orders, estimate spare parts demand, and dynamically control operation and maintenance permissions.

8. The passive lock real-time monitoring and predictive maintenance control system as described in claim 7, characterized in that, The platform layer's algorithm engine incorporates a three-tiered progressive early warning model, including: a threshold alarm module, connected to the rule engine, configured to perform multi-parameter joint rule judgments on multi-dimensional sensor data to generate real-time alarm events; a trend early warning module, connected to the time-series database, configured to use an LSTM encoder-decoder combined with an attention mechanism to reconstruct error analysis of historical time-series data and generate trend early warning information; and a lifespan prediction module, also connected to the time-series database, configured to use a combination of the Weibull distribution model and Bayesian dynamic updates, combined with particle filtering, to estimate the state of accumulated wear and output the probability distribution prediction result of the remaining lifespan.

9. The passive lock real-time monitoring and predictive maintenance control system as described in claim 7, characterized in that, The main control MCU in the terminal layer is configured to switch between deep sleep mode and active mode, and is triggered to wake up by a real-time clock. The energy management module includes a lithium manganese battery main power supply and a vibration energy harvesting auxiliary power supply. The vibration energy harvesting auxiliary power supply is configured to convert environmental vibration energy and store it in an energy storage element. When the lithium manganese battery main power supply is lower than a preset threshold, the energy storage element will supplement the system with power. The terminal device can achieve long-term maintenance-free operation without replacing the battery.

10. A passive lock real-time monitoring and predictive maintenance control terminal device, applied to the passive lock real-time monitoring and predictive maintenance control system according to any one of claims 7-9, characterized in that, include: The main control MCU is configured to alternate between deep sleep mode and active mode, and is triggered to wake up periodically by a real-time clock; Multi-dimensional sensor nodes are connected to the main control MCU signal and configured to collect data on the lock's temperature, humidity, vibration, bolt displacement, door magnetic status, and number of unlocking times. The wireless communication module is electrically connected to the main control MCU and is configured to upload the encrypted data processed by the main control MCU to the edge computing gateway through the wireless communication link. The energy management module is electrically connected to the main control MCU and the wireless communication module, and is configured to provide power to each module. The tamper detection unit is electrically connected to the main control MCU. The tamper detection unit includes a photosensitive sensor and a micro switch. When the tamper detection unit detects that the outer shell has been removed, it sends a trigger signal to the main control MCU. The main control MCU responds to the trigger signal and performs a key erasure operation.

Images