A multi-source perception container global safety intelligent risk control system and method

Abstract

The application relates to the technical field of data acquisition and monitoring control, and discloses a multi-source perception container global security intelligent risk control system and method, which comprises a vehicle-mounted or ship-mounted perception edge node, a container global risk control terminal, a cloud monitoring control platform and a local protection module; the risk control terminal comprises a space-time alignment and data fusion module, which is used for eliminating clock drift to generate a state tensor by using an interpolation algorithm; an adaptive sampling rate control module adjusts a sampling frequency according to a state tensor change rate and shrinks the upper limit of the frequency when the voltage is low; a local risk state machine control module drives the local protection module to trigger action when communication is interrupted and fluctuation is out of limit; and a channel hedging module controls the edge node to cut into a cache mode when packet loss is out of limit and reconstructs the time sequence after recovery, so that the application reduces data distortion, considers sampling energy consumption and risk capture demand, and realizes local protection control.

Description

Technical Field

[0001] This invention belongs to the field of data acquisition and monitoring control technology, and in particular relates to a multi-source sensing container full-domain safety intelligent risk control system and method. Background Technology

[0002] In container multimodal transport, data acquisition and monitoring control systems are typically used to continuously monitor container status and perform risk control. Existing systems generally install environmental temperature and humidity modules, three-axis attitude sensing modules, and status response seal modules on the container body. They collect relevant data according to a set cycle and then upload it to a cloud monitoring and control platform via a wireless communication network. The cloud monitoring and control platform performs risk assessment and issues control commands. When containers are running under different conditions such as ocean shipping, port loading and unloading, and land transportation, the hardware crystal oscillator accuracy of each sensing module varies. Timing deviations accumulate gradually over time. Moreover, the sampling frequencies of different types of sensing modules are not consistent. When data asynchronously collected by each sensing module is combined into data for status monitoring, different data at the same moment may not correspond accurately, which can easily affect the risk assessment results and even cause misjudgments.

[0003] Container monitoring terminals typically rely on built-in batteries for power, making it difficult to replenish power promptly during transportation. While fixed-period low-frequency sampling can reduce power consumption, it can easily miss short-lived mechanical shocks, unauthorized opening of containers, and other emergencies. Continuous high-frequency sampling accelerates battery consumption, potentially causing the terminal to lose its monitoring capability in the latter half of the journey. Simply increasing battery capacity or continuously raising the sampling frequency not only increases the terminal's weight and hardware cost but also generates a large amount of invalid data, increasing the burden on wireless transmission. It is still difficult to balance battery life and the ability to detect sudden risks. When containers are stacked at the bottom of the ship or obstructed by metal racks, wireless signals are easily affected by multipath attenuation, which may lead to long communication delays or interruptions. Existing systems rely heavily on cloud-based monitoring and control platforms for risk assessment and control. When wireless communication is lost, cloud control commands cannot reach the container monitoring terminal in a timely manner, making it difficult for the terminal to perform protection and recovery operations locally.

[0004] Therefore, how to reduce the deviation of multi-source sensing data in sampling time, and enable the system to adjust the data acquisition process according to changes in container status under limited communication and power conditions, while completing monitoring, judgment and control locally, remains a problem that existing container safety monitoring systems need to solve. Summary of the Invention

[0005] To address the problems in the background art, the technical solution of the present invention is as follows: A multi-source sensing intelligent risk control system for container full-domain safety, comprising: The vehicle / shipborne sensing edge node includes a three-axis attitude sensing module, a status response sealing module, and an environmental temperature and humidity module, and is connected to the container full-domain risk control terminal. The container-based full-domain risk control terminal includes: The spatiotemporal alignment and data fusion module is used to receive sampling data from vehicle / shipborne sensing edge nodes, and uses a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor. The adaptive sampling rate control module is used to calculate the fluctuation index based on the Euclidean distance change rate of the state tensor, and adjust the sampling frequency of the vehicle / shipborne sensing edge node in a stepwise manner based on the fluctuation index; when the power supply voltage is lower than the set threshold, the adaptive sampling rate control module shrinks the upper limit of the highest sampling frequency based on the ratio of the power supply voltage to the set cutoff voltage. The local risk state machine control module is used to match state transition conditions based on state tensors and drive the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than a preset threshold. The channel offset module is used to control the vehicle / shipborne sensing edge nodes to switch to buffer mode when the communication packet loss rate exceeds a preset threshold, and to complete the interpolation recovery using a linear predictive coding algorithm after recovery.

[0006] Preferably, the spatiotemporal alignment and data fusion module establishes a unified reference time axis and maps the asynchronous sampling data of vehicle / shipborne sensing edge nodes onto the reference time axis based on the third-order Lagrange polynomial interpolation formula, generating a state tensor containing triaxial acceleration, temperature, humidity, and the state response seal module loop impedance state; wherein, when the jitter variance of the sampling timestamp sequence exceeds the preset jitter threshold, the spatiotemporal alignment and data fusion module switches from the third-order Lagrange polynomial interpolation downgrade to the first-order linear extrapolation mode.

[0007] Preferably, the adaptive sampling rate control module calculates the 2-norm as the fluctuation index based on the Hadamard product of the difference value of the continuous state tensor and the weight vector; the adaptive sampling rate control module calculates the sampling frequency based on the fluctuation index using the following formula. : ,in, The updated sampling frequency, The pre-set base sampling frequency, The preset maximum sampling frequency, This represents the volatility index at the current moment. The first sampling adjustment threshold is preset. The second sampling adjustment threshold is preset; in the volatility index Greater than or equal to the second sampling adjustment threshold At that time, the sampling frequency Set to the highest sampling frequency In the volatility index Less than or equal to the first sampling adjustment threshold At that time, the sampling frequency Set to the preset minimum sampling frequency .

[0008] Preferably, the local risk state machine control module includes a normal monitoring state, an environmental early warning state, a physical intrusion state, and a severe accident state; the state transition logic combines the fluctuation index, the environmental temperature and humidity limit, and the state response seal module loop impedance state to establish a composite adjudication matrix; the state transition condition requires a continuously preset number of discrete time sequence numbers to satisfy the composite adjudication matrix, and when the state recovers to a low risk level, a hysteresis threshold is superimposed.

[0009] Preferably, when the state transitions to the physical intrusion state or the severe accident state, the local risk state machine control module directly drives the container full-domain risk control terminal to lock the local data log, activate the high-frequency audible and visual alarm module, and control the short-range / long-range wireless communication module to switch to the emergency sudden communication mode to send alarm messages with location information.

[0010] Preferably, the adaptive sampling rate control module further includes a vibration and noise mitigation module; the vibration and noise mitigation module calculates the root mean square value of triaxial acceleration, and only when the fluctuation index is greater than or equal to the second sampling adjustment threshold and the loop impedance change rate of the state response seal module exceeds a preset impedance threshold of 15%, it is determined to be a valid intrusion event and the highest sampling frequency is maintained.

[0011] Preferably, the container-wide risk control terminal also includes a power degradation control module; the power degradation control module monitors the power supply voltage, and when it detects that the power supply voltage is lower than the set voltage threshold of 3.3V, it reduces the upper limit of the highest sampling frequency according to the ratio of the difference between the current power supply voltage and the set cutoff voltage of 3.0V to the difference between 3.3V and 3.0V; when the power degradation takes effect, the memory of the container-wide risk control terminal automatically adjusts the storage partition ratio and prioritizes the retention of high-risk status logs.

[0012] Preferably, the vehicle-mounted / ship-mounted sensing edge node includes an edge computing module; the edge computing module calculates the composite vector magnitude of the triaxial acceleration. When the deviation between the composite vector magnitude and the standard gravitational acceleration is less than a preset deviation threshold of 0.1g and the rate of change of environmental parameters is less than a preset rate threshold of 0.05% / s, the vehicle-mounted / ship-mounted sensing edge node only sends a single-byte status heartbeat packet to the container full-domain risk control terminal; after receiving the status heartbeat packet, the container full-domain risk control terminal supplements the steady-state interpolation data locally and controls the radio frequency front-end to enter a sleep state.

[0013] Preferably, the container-wide risk control terminal also includes an offline compression module. During the period of wireless communication disconnection with the cloud monitoring and control platform, the offline compression module uses a piecewise linear approximation algorithm to fit the time series data, saving only the slope, intercept, and start timestamp of the fitted line. After wireless communication is restored, the container-wide risk control terminal sends the fitted parameters to the cloud monitoring and control platform in batches. The cloud monitoring and control platform reconstructs the environmental state curve and completes the residual verification. The cloud monitoring and control platform includes a global risk situation assessment module and a dynamic decision threshold distribution module, which are used to summarize the monitoring data of multiple containers and correct the risk strategy parameters in the container-wide risk control terminal.

[0014] A multi-source sensing-based intelligent risk control method for container safety across the entire domain, used to operate a multi-source sensing-based intelligent risk control system for container safety across the entire domain, includes the following steps: Step S1: Receive sampled data from vehicle / shipborne sensing edge nodes through the spatiotemporal alignment and data fusion module, and use a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor; Step S2: The adaptive sampling rate control module calculates the fluctuation index based on the Euclidean distance change rate of the state tensor, and adjusts the sampling frequency of the vehicle / shipborne sensing edge node in a stepwise manner based on the fluctuation index; when the power supply voltage is lower than the set threshold, the adaptive sampling rate control module shrinks the upper limit of the highest sampling frequency based on the ratio of the power supply voltage to the set cutoff voltage. Step S3: Based on the state tensor matching state transition conditions, the local risk state machine control module drives the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than the preset threshold. Step S4: When the communication packet loss rate exceeds the preset threshold, the channel offset module controls the vehicle / shipborne sensing edge node to switch to buffer mode, and after recovery, the linear predictive coding algorithm is used to complete the interpolation recovery.

[0015] Compared with existing technologies, the present invention provides a multi-source sensing intelligent risk control system and method for container safety across the entire domain, which has the following advantages: 1. In the intelligent risk control of container safety across the entire domain through multi-source sensing, the sampling data of different sensing modules are corrected by the spatiotemporal alignment and data fusion module. This can reduce the time deviation caused by hardware clock drift and different sampling frequencies, and make the triaxial acceleration, ambient temperature and humidity and seal circuit impedance state accurately correspond under the same time reference. This can improve the temporal consistency of multi-source data, reduce misjudgments caused by data misalignment when the container shakes or its state changes rapidly, and provide more accurate and reliable data for risk state identification.

[0016] 2. The adaptive sampling rate control module can adjust the sampling frequency according to changes in the container's status. When the system is in a stable transportation state, the sampling frequency is reduced to decrease power consumption caused by data acquisition and wireless transmission. In the event of emergencies such as collisions or lock picking, the sampling frequency is increased in a timely manner to avoid missing short-lived risk signals. At the same time, by combining the loop impedance changes of the status response seal module to judge abnormal situations, false triggering caused by ordinary mechanical vibration can be reduced. This ensures timely capture of sudden risks while also taking into account the terminal's endurance.

[0017] 3. The local risk state machine control module can continue to make risk judgments based on the real-time status of the container when wireless communication is interrupted, and directly execute local protection operations such as log locking, audible and visual alarms and emergency communication in the event of physical intrusion or serious accident. Therefore, the container full-domain risk control terminal does not have to rely entirely on the cloud monitoring and control platform to issue control commands, which can avoid monitoring and control failures caused by communication disconnection, and enable the system to maintain continuous risk monitoring and local control capabilities in transportation environments with limited wireless signals. Attached Figure Description

[0018] Figure 1 This is a flowchart of the container safety risk control method of the present invention; Figure 2 This is a diagram of the local risk state machine control structure of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0020] A multi-source sensing, all-domain intelligent risk control system for container safety includes: The vehicle / shipborne sensing edge node includes a three-axis attitude sensing module, a status response sealing module, and an environmental temperature and humidity module, and is connected to the container full-domain risk control terminal. The container-based full-domain risk control terminal includes: The spatiotemporal alignment and data fusion module is used to receive sampling data from vehicle / shipborne sensing edge nodes, and uses a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor. The adaptive sampling rate control module is used to calculate the fluctuation index based on the Euclidean distance change rate of the state tensor, and adjust the sampling frequency of the vehicle / shipborne sensing edge node in a stepwise manner based on the fluctuation index; when the power supply voltage is lower than the set threshold, the adaptive sampling rate control module shrinks the upper limit of the highest sampling frequency based on the ratio of the power supply voltage to the set cutoff voltage. The local risk state machine control module is used to match state transition conditions based on state tensors and drive the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than a preset threshold. The channel offset module is used to control the vehicle / shipborne sensing edge nodes to switch to buffer mode when the communication packet loss rate exceeds a preset threshold, and to complete the interpolation recovery using a linear predictive coding algorithm after recovery.

[0021] Preferably, the spatiotemporal alignment and data fusion module establishes a unified reference time axis and maps the asynchronous sampling data of vehicle / shipborne sensing edge nodes onto the reference time axis based on the third-order Lagrange polynomial interpolation formula, generating a state tensor containing triaxial acceleration, temperature, humidity, and the state response seal module loop impedance state; wherein, when the jitter variance of the sampling timestamp sequence exceeds the preset jitter threshold, the spatiotemporal alignment and data fusion module switches from the third-order Lagrange polynomial interpolation downgrade to the first-order linear extrapolation mode.

[0022] Preferably, the adaptive sampling rate control module calculates the 2-norm as the fluctuation index based on the Hadamard product of the difference value of the continuous state tensor and the weight vector; the adaptive sampling rate control module calculates the sampling frequency based on the fluctuation index using the following formula. : ,in, The updated sampling frequency, The pre-set base sampling frequency, The preset maximum sampling frequency, This represents the volatility index at the current moment. The first sampling adjustment threshold is preset. The second sampling adjustment threshold is preset; in the volatility index Greater than or equal to the second sampling adjustment threshold At that time, the sampling frequency Set to the highest sampling frequency In the volatility index Less than or equal to the first sampling adjustment threshold At that time, the sampling frequency Set to the preset minimum sampling frequency .

[0023] Preferably, the local risk state machine control module includes a normal monitoring state, an environmental early warning state, a physical intrusion state, and a severe accident state; the state transition logic combines the fluctuation index, the environmental temperature and humidity limit, and the state response seal module loop impedance state to establish a composite adjudication matrix; the state transition condition requires a continuously preset number of discrete time sequence numbers to satisfy the composite adjudication matrix, and when the state recovers to a low risk level, a hysteresis threshold is superimposed.

[0024] Preferably, when the state transitions to the physical intrusion state or the severe accident state, the local risk state machine control module directly drives the container full-domain risk control terminal to lock the local data log, activate the high-frequency audible and visual alarm module, and control the short-range / long-range wireless communication module to switch to the emergency sudden communication mode to send alarm messages with location information.

[0025] Preferably, the adaptive sampling rate control module further includes a vibration and noise mitigation module; the vibration and noise mitigation module calculates the root mean square value of triaxial acceleration, and only when the fluctuation index is greater than or equal to the second sampling adjustment threshold and the loop impedance change rate of the state response seal module exceeds a preset impedance threshold of 15%, it is determined to be a valid intrusion event and the highest sampling frequency is maintained.

[0026] Preferably, the container-wide risk control terminal also includes a power degradation control module; the power degradation control module monitors the power supply voltage, and when it detects that the power supply voltage is lower than the set voltage threshold of 3.3V, it reduces the upper limit of the highest sampling frequency according to the ratio of the difference between the current power supply voltage and the set cutoff voltage of 3.0V to the difference between 3.3V and 3.0V; when the power degradation takes effect, the memory of the container-wide risk control terminal automatically adjusts the storage partition ratio and prioritizes the retention of high-risk status logs.

[0027] Preferably, the vehicle-mounted / ship-mounted sensing edge node includes an edge computing module; the edge computing module calculates the composite vector magnitude of the triaxial acceleration. When the deviation between the composite vector magnitude and the standard gravitational acceleration is less than a preset deviation threshold of 0.1g and the rate of change of environmental parameters is less than a preset rate threshold of 0.05% / s, the vehicle-mounted / ship-mounted sensing edge node only sends a single-byte status heartbeat packet to the container full-domain risk control terminal; after receiving the status heartbeat packet, the container full-domain risk control terminal supplements the steady-state interpolation data locally and controls the radio frequency front-end to enter a sleep state.

[0028] Preferably, the container-wide risk control terminal also includes an offline compression module. During the period of wireless communication disconnection with the cloud monitoring and control platform, the offline compression module uses a piecewise linear approximation algorithm to fit the time series data, saving only the slope, intercept, and start timestamp of the fitted line. After wireless communication is restored, the container-wide risk control terminal sends the fitted parameters to the cloud monitoring and control platform in batches. The cloud monitoring and control platform reconstructs the environmental state curve and completes the residual verification. The cloud monitoring and control platform includes a global risk situation assessment module and a dynamic decision threshold distribution module, which are used to summarize the monitoring data of multiple containers and correct the risk strategy parameters in the container-wide risk control terminal.

[0029] A multi-source sensing-based intelligent risk control method for container safety across the entire domain, used to operate a multi-source sensing-based intelligent risk control system for container safety across the entire domain, includes the following steps: Step S1: Receive sampled data from vehicle / shipborne sensing edge nodes through the spatiotemporal alignment and data fusion module, and use a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor; Step S2: The adaptive sampling rate control module calculates the fluctuation index based on the Euclidean distance change rate of the state tensor, and adjusts the sampling frequency of the vehicle / shipborne sensing edge node in a stepwise manner based on the fluctuation index; when the power supply voltage is lower than the set threshold, the adaptive sampling rate control module shrinks the upper limit of the highest sampling frequency based on the ratio of the power supply voltage to the set cutoff voltage. Step S3: Based on the state tensor matching state transition conditions, the local risk state machine control module drives the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than the preset threshold. Step S4: When the communication packet loss rate exceeds the preset threshold, the channel offset module controls the vehicle / shipborne sensing edge node to switch to buffer mode, and after recovery, the linear predictive coding algorithm is used to complete the interpolation recovery.

[0030] Example 1: When a container enters the stacking area of ​​a shipping hold after being hoisted at the port, the three-axis attitude perception module of the vehicle / shipborne sensing edge node samples the timestamp sequence. Sampling timestamp sequence of ambient temperature and humidity module Phase deviation caused by crystal oscillator drift is addressed by establishing a unified reference time axis through the spatiotemporal alignment and data fusion module in the container-wide risk control terminal. ,in, As the reference time, The default reference time interval is 100ms. For any sensor at the reference time, the discrete time index is... Physical quantity mapping The third-order Lagrange polynomial interpolation formula is used: ,in, This is the current interpolation term number. The product term number is... For the nearest reference time The four original sampling timestamps, To remove The original sampling timestamps that are used in the calculation of the current interpolation term, in addition to the original sampling timestamps. Original sampling time The corresponding raw sensor observations, through the above interpolation, map the physical motion, environmental temperature and humidity, and loop impedance state of the state response sealing module under different inherent sampling periods to a unified time series, reducing the phase deviation caused by asynchronous sampling of multi-source data, and generating... The state tensor at time step: ,in, , and They are respectively Triaxial acceleration at time t, for Temperature at any moment for Humidity at any time for The state response of the sealing module circuit impedance at any given time.

[0031] The adaptive sampling rate control module receives the state tensor corresponding to adjacent discrete time indices. and Since the time interval between adjacent state tensors is Using a discrete time index as the measurement period, the weighted difference 2-norm of adjacent state tensors is taken as the Euclidean distance change rate of the state tensor, and the fluctuation index is calculated according to the following formula. : ,in, For a pre-defined weight vector, to These correspond to the six physical components in the state tensor. This represents the Hadamard product, with the subscript 2 indicating the 2-norm, used in calculating the volatility index. Previously, the differences in physical quantities of each channel were first processed to be dimensionless: the differences in triaxial acceleration were divided by the standard gravitational acceleration, the differences in temperature and humidity were divided by the full-scale values ​​of the corresponding sensors, the differences in loop impedance of the state response sealing module were divided by the initial reference impedance value and converted into a percentage rate of change. The dimensionless components were multiplied by the corresponding weights in the weight vector, and then the 2-norm was calculated to avoid interference from different physical dimensions and numerical magnitudes on the calculation of the fluctuation index.

[0032] When a collision during loading and unloading machinery causes a sudden change in triaxial acceleration, the state tensor... As it changes, the volatility index As the sampling rate increases, the adaptive sampling rate control module adjusts the first sampling threshold. Second sampling adjustment threshold Divide the system into low-volatility, intermediate-range, and high-volatility zones, and switch the sampling frequency adjustment rule according to the range in which the volatility index falls. Within the intermediate-range, update the sampling frequency according to the following formula. : ,in, The updated sampling frequency, The preset base sampling frequency is 1Hz. The preset maximum sampling frequency is 100Hz. This represents the volatility index at the current moment. The first sampling adjustment threshold is 0.05. The second sampling adjustment threshold is 0.80, when At that time, Set as That is, 100Hz; when At that time, Set to the preset minimum sampling frequency That is, 0.1Hz; when At that time, according to the above formula in and Update between This results in a stepped sampling frequency control based on a three-fluctuation-range switching adjustment rule, which maintains low-frequency sampling during stable transportation and increases the sampling frequency when a collision or impact occurs, in order to collect pulse signals with shorter durations.

[0033] When containers are located at the bottom of the ship's hold, the metal shielding may cause a break in the wireless communication channel; during this period, if a tampering occurs, the local risk state machine control module continuously receives state tensors. The vibration noise mitigation module in the adaptive sampling rate control module takes the triaxial acceleration samples within the current sliding window as input, averages the sum of squares of the triaxial accelerations at each sampling time, and then takes the square root to obtain the root mean square value of the triaxial acceleration. This value is used to characterize the mechanical vibration intensity within the window. If only mechanical vibration causes the fluctuation index to be affected... If the impedance of the status response seal module increases, but the rate of change of the loop impedance does not exceed the preset impedance threshold of 15%, it is not considered a physical intrusion; when When the loop impedance change rate of the status response seal module exceeds 15%, it is determined to be a valid physical intrusion event, and the sampling frequency is maintained at 100Hz.

[0034] The local risk state machine control module is based on the state tensor Match the state transition conditions to bring the system from normal monitoring state. Switch to physical intrusion state When the wireless communication channel is disconnected, the local risk state machine control module directly drives the local protection module to perform control actions: lock the local data log in the container full-domain risk control terminal, activate the high-frequency audible and visual alarm module, and control the short-range / long-range wireless communication module to switch to emergency communication mode and send alarm messages with location information.

[0035] The local protection module includes a hardware log locking circuit, a high-frequency audible and visual alarm buzzer, an LED array, and a physical anti-tamper electrically controlled locking mechanism driven by an independent relay. The local risk state machine control module outputs high and low level trigger signals via the bus to control the hardware log locking circuit to pull high the write protection pin of the memory, switching the storage area containing the local data log to read-only mode; simultaneously, it controls the relay to close, activating the high-frequency audible and visual alarm buzzer and the LED array, and sends a self-locking level to the physical anti-tamper electrically controlled locking mechanism, keeping it locked. The module performs time-series alignment of asynchronous multi-source sampled data through a spatiotemporal alignment and data fusion module, and combines this with a volatility index-based approach. With adaptive sampling rate control and local risk state machine control, the container global risk control terminal can still continuously perform data acquisition, risk assessment and local protection actions when the physical state changes drastically and long-distance communication is limited, thus realizing local data acquisition, risk assessment and protection control.

[0036] Example 2: In this example, the container full-domain risk control terminal, vehicle / shipborne sensing edge nodes and cloud monitoring and control platform are deployed on the container test platform. The shielded compartment simulates the metal shielding of the sea hull and the high packet loss channel conditions. The test signal source is superimposed with Gaussian white noise with a signal-to-noise ratio of 20dB and 50Hz power frequency interference harmonics. The test platform is equipped with a loading and unloading mechanical collision application device and an electronic seal destruction simulation device.

[0037] The test samples include the present invention sample group, comparative sample group 1, comparative sample group 2, comparative sample group 3, comparative sample group 4, comparative sample group 5, and comparative sample group 6. The present invention sample group is equipped with a spatiotemporal alignment and data fusion module, an adaptive sampling rate control module including a vibration noise offset module, and a local risk state machine control module. Comparative sample group 1 uses a fixed 100Hz sampling frequency, comparative sample group 2 uses a fixed 1Hz sampling frequency, comparative sample group 3 uses an asynchronous acquisition and fusion method without third-order Lagrange polynomial interpolation spatiotemporal alignment, comparative sample group 4 uses a single-parameter adaptive sampling control method without loop impedance overlap verification of the state response seal module, and comparative sample group 5 adjusts the second sampling threshold. Set to 0.30, and adjust the second sampling threshold for comparison group six. Set to 0.95.

[0038] In each sample group, the basic sampling frequency Set to 1Hz, highest sampling frequency Set to 100Hz, minimum sampling frequency Set to 0.1Hz, first sampling adjustment threshold Set to 0.05, second sampling adjustment threshold Set to 0.80, the adaptive sampling rate control module adjusts according to the fluctuation index. The sampling frequency control rule for the adjustment range is as follows: reduce the sampling frequency when the change in physical motion is small, and increase the sampling frequency when the change in physical motion is large, so as to reduce power consumption during stable transportation and collect short-duration pulse signals during collisions and lock picking.

[0039] Under the combined conditions of mechanical collision during loading and unloading and lock picking intrusion, each sample group simultaneously collected triaxial acceleration, ambient temperature and humidity, and the state response of the seal module circuit impedance. Time, reference time The time is 10.0s, and the normalized vector of triaxial acceleration is... ;exist Time, reference time The time is 10.1s. The collision of the loading and unloading machinery causes the normalized vector of the triaxial acceleration to change abruptly. ;exist Time, reference time At 10.2s, the electronic seal destruction simulation device applied a pry lock intrusion, and the state response seal module circuit impedance change rate reached 18.5%.

[0040] The present invention sample group in The original input phase deviation at time t is 32ms. After the spatiotemporal alignment and data fusion module interpolates and maps the asynchronous sampled data, the fluctuation index... The value is 0.032, which is lower than the first sampling adjustment threshold of 0.05. Therefore, the adaptive sampling rate control module reduces the sampling frequency to 0.1Hz. At a certain time, the original input phase deviation increases to 35ms. The spatiotemporal alignment and data fusion module uses third-order Lagrange polynomial interpolation to correct the sampling timestamp jitter, resulting in a fluctuation index. The value is 0.741; this value falls between the first sampling adjustment threshold of 0.05 and the second sampling adjustment threshold of 0.80. The adaptive sampling rate control module updates the sampling frequency according to the linear proportional relationship corresponding to the intermediate adjustment range. Adjust to 92.21Hz and complete the sampling frequency increase within a 100ms response window.

[0041] exist At that time, the original input phase deviation of the sample group of this invention was 31ms, and the fluctuation index was... The sampling rate increases to 1.250, exceeding the second sampling adjustment threshold of 0.80. The adaptive sampling rate control module locks the sampling frequency to 100Hz. At this point, the impedance change rate of the state response seal module loop is 18.5%, exceeding the preset impedance threshold of 15%. Based on this, the vibration and noise mitigation module determines it as a valid physical intrusion event. The local risk state machine control module then determines the event based on the state tensor. Match the state transition conditions to bring the system from normal monitoring state. Switch to physical intrusion state When wireless communication is interrupted and the cloud monitoring and control platform does not issue control commands, the local data log is directly locked and a high-frequency audible and visual alarm is triggered.

[0042] During the aforementioned stable, impact, and intrusion tests, the physical intrusion response latency recorded by the sample group of this invention was 4.8 ms, the pulse risk capture rate was 99.2%, the average terminal power consumption was 18.2 mA, the pseudo-high-frequency false trigger rate was 1.2%, and the corresponding fluctuation index was... The values ​​are 0.032, 0.741, and 1.250, respectively, representing the dynamic sampling frequencies. The sampling frequencies are 0.1Hz, 92.21Hz, and 100Hz, respectively, and are adjusted according to the degree of change of the state tensor and the state response seal module loop impedance.

[0043] The original input phase deviation of each comparison sample group at the moment of impact was 35ms. Comparison sample group one used a fixed sampling frequency of 100Hz and did not calculate the fluctuation index. Its physical intrusion response latency is 12.5ms, the pulse risk capture rate is 99.1%, the average terminal power consumption is 85.2mA, the false high-frequency false trigger rate is 3.5%, and the battery life is shortened by 78.6%. This sample group maintains the highest sampling frequency and the pulse risk capture rate is close to that of the sample group of the present invention, but the average terminal power consumption is higher than that of the sample group of the present invention.

[0044] Comparison group two uses a fixed sampling frequency of 1Hz, and the fluctuation index is also not calculated. Under collision impact input, its physical intrusion response delay is 120.0ms, pulse risk capture rate is 68.1%, terminal average power consumption is 12.4mA, pseudo high frequency false trigger rate is 0.8%. Fixed low frequency sampling reduces terminal average power consumption, but when the collision pulse occurs between adjacent sampling times, the sampling data is insufficient to fully reflect the impact process, and the response delay increases accordingly.

[0045] Compared to sample group three, which retains a 35ms asynchronous sampling phase deviation, the acceleration data and the state response seal module loop impedance state were not mapped to a unified reference time axis using third-order Lagrange polynomial interpolation. Due to the phase misalignment, its fluctuation index... The updated sampling frequency is 0.412. The frequency was 48.15 Hz, the physical intrusion response delay was 45.2 ms, the pulse risk capture rate was 81.5%, the average power consumption of the terminal was 24.6 mA, the pseudo-high frequency false trigger rate was 14.8%, the phase misalignment caused the acceleration change and the loop impedance change to not correspond in the same discrete time sequence, and the fluctuation index was lower than 0.741 obtained by the sample group of this invention under the same impact input.

[0046] Comparison sample group four uses a single-parameter adaptive sampling control method, without utilizing the change rate of the state response seal module loop impedance to verify the acceleration change. Under an impact input with a 35ms phase deviation, its fluctuation index... The dynamic sampling frequency is 0.741. The frequency is 92.21Hz, the physical intrusion response delay is 4.8ms, the pulse risk capture rate is 99.2%, the average terminal power consumption is 31.5mA, the pseudo-high frequency false trigger rate is 28.4%, and the acceleration change caused by mechanical vibration can trigger high frequency sampling alone. Therefore, its response delay and pulse risk capture rate are the same as those of the sample group of this invention, but the number of times high frequency sampling is triggered by mechanical vibration increases.

[0047] The second sampling threshold was adjusted in the fifth comparison sample group. Set to 0.30, under an impact input with a phase deviation of 35ms, the fluctuation index... The value is 0.741, exceeding the second sampling adjustment threshold set for this sample group, and the sampling frequency rises to 100Hz; its physical intrusion response delay is 4.8ms, the pulse risk capture rate is 99.2%, the average terminal power consumption is 64.5mA, and the false high-frequency false trigger rate is 32.1%. During normal transportation, the fluctuation index generated by slight vibration is also easy to exceed 0.30, causing the sampling frequency to frequently remain at 100Hz.

[0048] The second sampling threshold was adjusted in the comparison sample group six. Set to 0.95, under an impact input with a phase deviation of 35ms, the fluctuation index... The value is 0.741, which does not reach the second sampling adjustment threshold set for this sample group; the dynamic sampling frequency... It is maintained at 1.0Hz; its physical intrusion response delay is 95.0ms, the pulse risk capture rate is 72.3%, the average power consumption of the terminal is 14.1mA, and the false high frequency false trigger rate is 1.0%. Since the collision impact did not trigger the highest sampling frequency, some pulse data during the collision was not collected, and the physical intrusion response delay increased accordingly.

[0049] First sampling adjustment threshold 0.05, second sampling adjustment threshold With a configuration of 0.80, the sample group of this invention reduces the sampling frequency to 0.1Hz in stable operating conditions, adjusts the sampling frequency to 92.21Hz in collision conditions, and increases the sampling frequency to 100Hz in lock picking and intrusion conditions; the corresponding pulse risk capture rate is 99.2%, the average power consumption of the terminal is 18.2mA, and the pseudo-high frequency false trigger rate is 1.2%.

[0050] Example 3: Before being installed on the container, the current container-wide risk control terminal first completes the sampling rate adjustment threshold calibration, placing the container in a static and stable parking state. The three-axis attitude sensing module, environmental temperature and humidity module, and state response seal module all continuously collect raw data for 10 seconds at a basic sampling frequency of 1Hz. The spatiotemporal alignment and data fusion module uses third-order Lagrange polynomial interpolation to map the asynchronous sampling data of each module to a unified reference time axis, generating state tensors corresponding to 100 discrete time indices. ,in, , and They are respectively Triaxial acceleration at time t, for Temperature at any moment for Humidity at any time for The state response module's loop impedance state at any given time, and the adaptive sampling rate control module calculates the state tensor difference between adjacent discrete time points. And combine the difference with the weight vector By performing the Hadamard product and then calculating the 2-norm of the resulting vector, the volatility index can be obtained. Find the volatility indices during the static steady-state parking period. The variance of the sample is calculated, and this variance is multiplied by a safety factor of 3 to determine the first sampling adjustment threshold. The value was set to 0.05. Subsequently, the container was placed on a standard drop test bench, and a mechanical impact acceleration of 5g was applied. The fluctuation index corresponding to the impact moment was calculated. The second sampling adjustment threshold is determined by using 80% of the volatility index as the saturation threshold. The value is 0.80, thus completing the calibration of the sampling rate adjustment threshold of the adaptive sampling rate control module.

[0051] The power degradation control module in the container-wide risk control terminal monitors the power supply voltage in real time. During long-distance transportation, when the battery voltage drops to At that time, the power degradation control module adjusts according to the current supply voltage. With the set cutoff voltage The difference is 3.3V and The ratio of the differences is used to proportionally shrink the upper limit of the highest sampling frequency. The updated upper limit of the highest sampling frequency is: ,in, The preset maximum sampling frequency is 100Hz. The current supply voltage. The set cutoff voltage is 3.0V. At 3.15V, the calculated values ​​are as follows: The sampling frequency is set to 50Hz, and the power degradation control module accordingly limits the sampling frequency output by the adaptive sampling rate control module to below 50Hz.

[0052] When power degradation control takes effect, the non-volatile memory of the container-wide risk control terminal synchronously adjusts the storage partition ratio, switching from normal monitoring state to normal operation. The corresponding storage partition ratio was reduced from 80% to 30%, while the storage partition ratio for high-risk state logs was increased from 20% to 70%. Priority was given to saving local risk state machine control modules switched to the physical intrusion state. and serious accident status The state tensor generated at time When the power supply voltage recovers to above 3.35V, the power degradation control module removes the shrinkage limit of the maximum sampling frequency and restores the storage partition ratio to the state before adjustment.

[0053] When containers are stacked in the ship's hold, the metal casing causes electromagnetic wave multipath attenuation. When more than 5 consecutive packets are lost between the vehicle / shipborne sensing edge node and the container-wide risk control terminal, the channel offset module controls the vehicle / shipborne sensing edge node to switch to a blind sampling buffer mode in its local non-volatile memory. The sampled data during the communication interruption, along with the corresponding data packet sequence number, is written to the local memory. After communication is restored, the container-wide risk control terminal determines the missing discrete-time sequence number based on the received data packet sequence number and uses a second-order linear predictive coding algorithm to interpolate and recover the state tensor at the missing time. ,in, for Reconstructing the state tensor at time step and This is the state tensor that has been corrected in the previous two adjacent discrete time steps. and For the second-order prediction coefficients, the autocorrelation function is calculated for the triaxial acceleration, temperature, humidity, and state response seal module loop impedance states in the state tensor. This is achieved by taking scalar data sequences from 20 discrete time points in the history of each channel, resulting in two second-order prediction coefficients for that channel. These two prediction coefficients are then multiplied by the data from the previous two discrete time points of that channel and summed. Finally, the reconstruction results from each channel are concatenated according to the component order of the state tensor to form a complete reconstructed state tensor. After inserting continuously missing state tensors using the above method, the volatility index... The residual variance is less than 0.005.

[0054] The local risk state machine control module is based on the volatility index. A composite adjudication matrix is ​​established based on the ambient temperature and humidity limits and the loop impedance state of the state response seal module. The rows of this matrix correspond sequentially to the normal monitoring state. Environmental early warning status Physical intrusion state and serious accident status The columns correspond sequentially to the fluctuation index range determination, the temperature and humidity boundary comparison results, and the discrete logic state of the loop impedance.

[0055] When volatility index Not higher than the first sampling adjustment threshold When both temperature and humidity are within their respective limits and the loop impedance is normal, the composite decision matrix is ​​in normal monitoring state. The corresponding judgment result is true when the temperature exceeds 50℃ or the humidity exceeds 90%, while the circuit impedance is normal, triggering an environmental warning state. The corresponding judgment result is true when the volatility index is true. Not lower than the second sampling adjustment threshold Furthermore, when the rate of change of the loop impedance of the status response seal module exceeds 15%, a physical intrusion state is detected. The corresponding judgment result is true when the temperature or humidity reaches a severe accident state. When the corresponding preset ambient temperature and humidity limits are met, a severe accident state is observed. The corresponding judgment result is true.

[0056] The conditions for any state transition must be continuous. A state transition can only be triggered if each discrete time sequence number satisfies the corresponding matrix judgment result. Taking physical intrusion judgment as an example, when the fluctuation index... Furthermore, if the rate of change of the loop impedance of the state response seal module exceeds 15%, and this condition is met for three consecutive discrete time intervals, the local risk state machine control module will switch the system from the normal monitoring state. Switch to physical intrusion state .

[0057] When the physical intrusion event is resolved and the system recovers to a low-risk level, the local risk state machine control module employs a hysteresis threshold. A recovery determination is performed, in which... The value is 0.15, when the volatility index satisfies... and continuously When the discrete time sequence number is less than 0.65, the system is in a physical intrusion state. Restore to normal monitoring status This avoids frequent state switching caused by mechanical vibration and slight fluctuations in circuit impedance.

[0058] Example 4: This example combines Figures 1 to 2 This document describes a multi-source sensing intelligent risk control system and method for container safety across the entire domain. Figure 1As shown, in step S1, the spatiotemporal alignment and data fusion module receives the sampling data of the vehicle / shipborne sensing edge nodes, and uses a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor. In step S2, the adaptive sampling rate control module calculates the fluctuation index based on the Euclidean distance change rate of the state tensor to adjust the sampling frequency in a stepwise manner, and shrinks the upper limit of the highest sampling frequency based on the voltage ratio when the power supply is low. Then, in step S3, the local risk state machine control module matches the state transition conditions based on the state tensor, and drives the local protection module to trigger control actions when communication is interrupted and the fluctuation index exceeds the limit. Finally, in step S4, the channel offset module controls the edge nodes to switch to buffer mode when the communication packet loss rate exceeds the limit, and uses a linear predictive coding algorithm to complete the interpolation recovery after recovery.

[0059] like Figure 2 As shown, the system takes the fluctuation index, ambient temperature and humidity limits, and the state response seal module loop impedance status as inputs. Combined with the state transition conditions satisfied by a continuously preset number of discrete time sequences, the system inputs a composite decision matrix. When the condition is true, the system switches to normal monitoring state, environmental warning state, physical intrusion state, or severe accident state. When the system is in physical intrusion state or severe accident state and communication is interrupted, the system triggers the local protection module to lock the local data log, activate the high-frequency audible and visual alarm module, and switch to emergency communication mode to send an alarm message with location information. At the same time, each state can be superimposed with a hysteresis threshold to recover to a low-risk level and return to normal monitoring state.

[0060] Example 5: After the container is sealed, the three-axis attitude sensing module, environmental temperature and humidity module, and state response seal module in the vehicle / shipborne sensing edge node initiate the on-site environmental baseline initialization calibration. The container's full-domain risk control terminal controls each module to continuously collect data for 50 discrete time numbers at a basic sampling frequency of 1Hz. The spatiotemporal alignment and data fusion module extracts the three-axis acceleration, temperature, humidity, and state response seal module loop impedance status within this 5-second period, calculates the arithmetic mean of each physical quantity, and forms the initial baseline bias vector. .

[0061] After completing baseline initialization, the spatiotemporal alignment and data fusion module starts from the dynamic state tensor. Subtracting the initial baseline bias vector The corresponding gravity component and initial loop impedance base value are used, and the loop impedance change rate of the state response seal module is converted into a percentage based on the initial loop impedance. After the above processing, the influence of different container body structure stiffness and sensor installation tilt angle on the initial component of triaxial acceleration can be reduced, so that the subsequent state tensor corresponds to the current loading state of the container.

[0062] During multimodal transport, containers alternately experience wet salt spray from sea transport and high-frequency vibrations from road transport. The spatiotemporal alignment and data fusion module maintains a dynamic sliding time window containing 100 discrete time indices. It periodically calculates the statistical variance of the triaxial acceleration and state response seal module loop impedance state within the window. When the mean variance within the sliding time window continuously deviates from the previous baseline by more than 10%, the adaptive sampling rate control module adjusts the weight vector. The weighting ratios of the triaxial acceleration components, environmental parameter components, and loop impedance components affect the fluctuation index. Adjustments are made as the transportation environment changes slowly.

[0063] weight vector It contains six components, corresponding to triaxial acceleration, temperature, humidity, and the state response seal module loop impedance state, respectively. During stable transportation, the initial weights of the first three triaxial acceleration components are all set to 0.25, the initial weights of the temperature and humidity components are all set to 0.05, and the initial weight of the loop impedance component is set to 0.15. The sum of the weights of each component remains at 1. When the variance of the rate of change of environmental parameters increases within the sliding time window, the adaptive sampling rate control module reduces the weights of the temperature and humidity components to 0.01 respectively, and distributes the released weights to the triaxial acceleration and loop impedance components according to the original proportion, reducing the impact of slowly changing environmental parameters on the fluctuation index. The impact of the calculation results.

[0064] Before the container-wide risk control terminal sends data to the cloud-based monitoring and control platform, the data anonymization module extracts the geographical location and equipment identification code from the local data log, generates an anonymous hash identifier using a one-way hash function, and then processes the state tensor. The encrypted hash identifier is packaged and sent to the cloud monitoring and control platform. The cloud monitoring and control platform performs monitoring and control based on the state tensor and distinguishes the corresponding container equipment by the anonymized hash identifier.

[0065] Example 6: The edge computing module in the current vehicle / shipborne sensing edge node calculates the triaxial acceleration composite vector magnitude in real time during the smooth section of container road transport: ,in, for The magnitude of the combined triaxial acceleration vector at time t. , and They are respectively The triaxial acceleration at time t is calculated based on the temperature and humidity differences between adjacent discrete time points. The temperature and humidity differences are divided by the full-scale value of the corresponding sensor, and then by the sampling time interval. The larger of the two absolute values ​​is taken as the current environmental parameter change rate.

[0066] when Compared with standard gravitational acceleration When the deviation is less than 0.1g and the rate of change of environmental parameters is less than 0.05% / s, the vehicle / shipborne sensing edge node suspends the transmission of full-sample data and only periodically sends single-byte status heartbeat packets through the low-power local bus. After receiving the single-byte status heartbeat packet, the container full-domain risk control terminal uses first-order linear interpolation to supplement the state tensor at discrete moments during the stationary period. It also controls the RF front end to enter sleep mode.

[0067] In this embodiment, the radio frequency front-end includes a radio frequency transmitting module in the vehicle / shipborne sensing edge node and a radio frequency receiving module in the container full-domain risk control terminal. In addition to the wireless radio frequency channel, the two are also connected in parallel by a low-power hardware bus trigger line. When the radio frequency front-end is in sleep mode, the low-power timer in the container full-domain risk control terminal briefly opens the receiving window every 100ms. When the vehicle / shipborne sensing edge node detects a change in attitude, the edge computing module sends a high-level pulse wake-up signal through the hardware bus pin, pulls up the hardware interrupt pin of the container full-domain risk control terminal, restores the radio frequency receiving module to normal receiving state, and wakes up the radio frequency transmitting module in the vehicle / shipborne sensing edge node.

[0068] When loading and unloading collisions cause Compared with standard gravitational acceleration When the deviation is greater than or equal to 0.1g, the edge computing module resumes full sampling data transmission. Thus, the RF front end remains dormant during stable transportation and resumes data transmission and reception during sudden changes in attitude.

[0069] Once containers enter the ocean shipping phase, long-range wireless communication links may be interrupted. The offline compression module in the container-wide risk control terminal processes continuously collected time-series temperature data. Perform piecewise linear approximation fitting and set the maximum allowable fitting error. Set to 0.2℃, for any observation value within a continuous time interval. The fitting results satisfy: ,in, Discrete time index The corresponding temperature data, The slope of the fitted line, The intercept of the fitted line, This represents the maximum permissible fitting error.

[0070] The offline compression module starts from the beginning timestamp of the current segment. Start by adding temperature observations sequentially and updating the data for that segment. and If the maximum allowable fitting error is still met after adding a new temperature observation, the observation is incorporated into the current segment; if the fitting error exceeds the limit after adding a new temperature observation, the new observation is incorporated into the current segment. When the current segment ends at the previous discrete sequence number, the next segment is established starting from the timestamp corresponding to that observation.

[0071] After each segment is fitted, the offline compression module only saves the segment in local memory. , and start timestamp The system removes redundant temperature observations within segments, arranges adjacent segments sequentially according to their start timestamps, and uses the start timestamp of the next segment as the end boundary of the previous segment. The last segment ends at the time when wireless communication is restored. With this method, the data storage space compression ratio during the shipping chain disruption reaches 8:1 to 12:1.

[0072] Once the container docks and the wireless communication link is re-established, the container-wide risk control terminal will control each segment... , and start timestamp Batch data is sent to the cloud monitoring and control platform, which sorts the segments according to their start timestamps and executes them within the corresponding time intervals. and The offline data points for each discrete time period are restored, and the segments are connected sequentially to reconstruct the environmental state curve for the entire time period.

[0073] At the common boundary moment between two adjacent segments, the cloud-based monitoring and control platform calculates the temperature reconstruction value at that boundary moment based on the fitting parameters of the previous and subsequent segments, respectively. The difference between the two temperature reconstruction values ​​is taken as the boundary residual, and the root mean square value is calculated for all boundary residuals. When the root mean square residual is greater than the preset residual threshold, the platform determines the boundary residual. At that time, the global risk situation assessment module generates a risk warning that the goods may deteriorate during the offline period.

[0074] The global risk situation assessment module aggregates the reconstructed environmental state curves, root mean square residuals, and risk warning results of multiple containers within the same time window, forming risk status data for multiple containers. The dynamic decision threshold distribution module corrects the risk strategy parameters in the container global risk control terminal based on the aggregated results and distributes the corrected risk strategy parameters to the corresponding container global risk control terminal. After receiving the risk strategy parameters, the container global risk control terminal writes them into its local parameter storage area and calls them in subsequent risk status determination.

[0075] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A multi-source sensing, all-domain intelligent risk control system for container safety, characterized in that, include: The vehicle / shipborne sensing edge node includes a three-axis attitude sensing module, a status response sealing module, and an environmental temperature and humidity module, and is connected to the container full-domain risk control terminal. The container-based full-domain risk control terminal includes: The spatiotemporal alignment and data fusion module is used to receive sampling data from vehicle / shipborne sensing edge nodes, and uses a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor. An adaptive sampling rate control module is used to calculate the fluctuation index based on the rate of change of the Euclidean distance of the state tensor, and to adjust the sampling frequency of the vehicle / shipborne sensing edge nodes in a stepwise manner based on the fluctuation index. When the supply voltage is lower than the set threshold, the adaptive sampling rate control module reduces the upper limit of the highest sampling frequency based on the ratio of the supply voltage to the set cutoff voltage. The local risk state machine control module is used to match state transition conditions based on state tensors and drive the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than a preset threshold. The channel offset module is used to control the vehicle / shipborne sensing edge nodes to switch to buffer mode when the communication packet loss rate exceeds a preset threshold, and to complete the interpolation recovery using a linear predictive coding algorithm after recovery.

2. The multi-source sensing container full-domain safety intelligent risk control system according to claim 1, characterized in that, The spatiotemporal alignment and data fusion module establishes a unified reference time axis. Based on the third-order Lagrange polynomial interpolation formula, it maps the asynchronous sampling data of vehicle / shipborne sensing edge nodes onto the reference time axis, generating a state tensor that includes triaxial acceleration, temperature, humidity, and the state response seal module loop impedance state. When the jitter variance of the sampling timestamp sequence exceeds the preset jitter threshold, the spatiotemporal alignment and data fusion module switches from third-order Lagrange polynomial interpolation to first-order linear extrapolation mode.

3. The multi-source sensing container full-domain safety intelligent risk control system according to claim 1, characterized in that, The adaptive sampling rate control module calculates the 2-norm as the fluctuation index based on the difference value of the continuous state tensor and the Hadamard product of the weight vector; The adaptive sampling rate control module calculates the sampling frequency based on the fluctuation index using the following formula. : ,in, The updated sampling frequency, The pre-set base sampling frequency, The preset maximum sampling frequency, This represents the volatility index at the current moment. The first sampling adjustment threshold is preset. The second sampling adjustment threshold is preset; in the volatility index Greater than or equal to the second sampling adjustment threshold At that time, the sampling frequency Set to the highest sampling frequency In the volatility index Less than or equal to the first sampling adjustment threshold At that time, the sampling frequency Set to the preset minimum sampling frequency .

4. The multi-source sensing container full-domain safety intelligent risk control system according to claim 1, characterized in that, The local risk state machine control module includes normal monitoring state, environmental early warning state, physical intrusion state, and severe accident state. The state transition logic combines the fluctuation index, environmental temperature and humidity limits, and the loop impedance of the state response seal module to establish a composite adjudication matrix. The state transition condition requires a number of continuously preset discrete time sequences to satisfy the composite adjudication matrix, and a hysteresis threshold is superimposed when the state recovers to a low risk level.

5. The multi-source sensing container full-domain safety intelligent risk control system according to claim 4, characterized in that, When the state transitions to the physical intrusion state or the severe accident state, the local risk state machine control module directly drives the container global risk control terminal to lock the local data log, activate the high-frequency audible and visual alarm module, and control the short-range / long-range wireless communication module to switch to the emergency sudden communication mode to send alarm messages with location information.

6. The multi-source sensing container full-domain safety intelligent risk control system according to claim 3, characterized in that, The adaptive sampling rate control module also includes a vibration and noise offset module. The vibration and noise offset module calculates the root mean square value of triaxial acceleration and determines it as a valid intrusion event and maintains the highest sampling frequency only when the fluctuation index is greater than or equal to the second sampling adjustment threshold and the loop impedance change rate of the state response seal module exceeds a preset impedance threshold of 15%.

7. The multi-source sensing container full-domain safety intelligent risk control system according to claim 1, characterized in that, The container-wide risk control terminal also includes a power degradation control module. The power degradation control module monitors the power supply voltage. When the power supply voltage is detected to be lower than the set voltage threshold of 3.3V, it reduces the upper limit of the highest sampling frequency based on the ratio of the difference between the current power supply voltage and the set cutoff voltage of 3.0V to the difference between 3.3V and 3.0V. When the power degradation takes effect, the memory of the container-wide risk control terminal automatically adjusts the storage partition ratio and prioritizes the retention of high-risk status logs.

8. The multi-source sensing intelligent risk control system for container safety across the entire domain, as described in claim 1, is characterized in that... The vehicle-mounted / ship-mounted sensing edge node includes an edge computing module. The edge computing module calculates the composite vector magnitude of the triaxial acceleration. When the deviation between the composite vector magnitude and the standard gravitational acceleration is less than a preset deviation threshold of 0.1g and the rate of change of environmental parameters is less than a preset rate threshold of 0.05% / s, the vehicle-mounted / ship-mounted sensing edge node only sends a single-byte status heartbeat packet to the container-wide risk control terminal. After receiving the status heartbeat packet, the container-wide risk control terminal supplements the steady-state interpolation data locally and controls the radio frequency front-end to enter a sleep state.

9. The multi-source sensing container full-domain safety intelligent risk control system according to claim 1, characterized in that, The container-wide risk control terminal also includes an offline compression module. During the period of wireless communication loss with the cloud monitoring and control platform, the offline compression module uses a piecewise linear approximation algorithm to fit the time series data, saving only the slope, intercept, and start timestamp of the fitted line. After wireless communication is restored, the container-wide risk control terminal sends the fitted parameters to the cloud monitoring and control platform in batches. The cloud monitoring and control platform reconstructs the environmental state curve and completes the residual verification. The cloud monitoring and control platform includes a global risk situation assessment module and a dynamic decision threshold distribution module, which are used to summarize the monitoring data of multiple containers and correct the risk strategy parameters in the container-wide risk control terminal.

10. A multi-source sensing intelligent risk control method for container safety across the entire domain, used to operate the multi-source sensing intelligent risk control system for container safety across the entire domain as described in claim 1, characterized in that, Includes the following steps: Step S1: Receive sampled data from vehicle / shipborne sensing edge nodes through the spatiotemporal alignment and data fusion module, and use a sliding window interpolation algorithm to eliminate hardware clock drift and generate a state tensor; Step S2: The adaptive sampling rate control module calculates the fluctuation index based on the Euclidean distance change rate of the state tensor, and adjusts the sampling frequency of the vehicle / shipborne sensing edge node in a stepwise manner based on the fluctuation index; when the power supply voltage is lower than the set threshold, the adaptive sampling rate control module shrinks the upper limit of the highest sampling frequency based on the ratio of the power supply voltage to the set cutoff voltage. Step S3: Based on the state tensor matching state transition conditions, the local risk state machine control module drives the local protection module to trigger control actions when communication is interrupted and the volatility index is greater than the preset threshold. Step S4: When the communication packet loss rate exceeds the preset threshold, the channel offset module controls the vehicle / shipborne sensing edge node to switch to buffer mode, and after recovery, the linear predictive coding algorithm is used to complete the interpolation recovery.

Images