Method and system for monitoring the operational safety of heavy goods logistics

Abstract

The application discloses a kind of operation safety monitoring method and system of bulky logistics, it is related to logistics monitoring technical field;The method comprises: based on the multiple heterogeneous sensors arranged on bulky logistics assets, real-time multi-source state data is collected, and unified real-time state vector is generated after data fusion;The state vector is used to update the digital twin model of asset in real time;And the data digest of state vector is recorded in blockchain to ensure that data is not tamperable and traceable;Further, based on the digital twin model, dual-mode safety monitoring is performed, including real-time anomaly monitoring in copy mode and active risk prediction in simulation mode;Finally, a safety alert is generated according to the monitoring or prediction result.The application solves the problems of incomplete state perception, delayed risk warning and low data reliability in the prior art.

Description

Technical Field

[0001] This invention relates to the field of logistics monitoring technology, and in particular to a method and system for monitoring the operational safety of large-item logistics. Background Technology

[0002] Large-item logistics, especially the transportation of overweight, overlength, overwidth, or irregularly shaped goods, is an indispensable and crucial link in modern industry and infrastructure construction. However, due to the high center of gravity, high value, and high difficulty in loading, unloading, and transportation of such goods, their safety during transit faces severe challenges. External factors such as bumps, sharp turns, speed changes, complex road conditions, and inclement weather during transportation can all lead to cargo displacement, vehicle instability, and even major safety accidents such as overturning, causing huge economic losses and risks of personal injury.

[0003] To ensure the safe operation of large-item logistics, existing technologies typically employ multiple monitoring methods. Among these, the Global Positioning System (GPS) is widely used to track vehicle location and trajectory, but this only provides location information and cannot reflect the dynamic changes in the cargo itself and its carrying condition in real time. Some solutions add independent sensors, such as tilt sensors or temperature sensors, to monitor certain single physical parameters of the cargo. However, the data collected by these sensors is often fragmented and isolated, lacking effective multi-source data integration and correlation analysis capabilities, making it difficult to form a comprehensive and accurate assessment of the overall operational attitude and stability of the cargo. Furthermore, most existing monitoring systems remain at the level of passive response and post-event traceability. Safety alarms are usually triggered only after dangerous parameters (such as tilt angle) have exceeded preset static thresholds, often missing the optimal intervention window and failing to effectively prevent accidents. These systems generally lack the ability to proactively predict risks by combining information about the route ahead (such as curve curvature and slope) and real-time environmental factors (such as crosswind intensity). Meanwhile, in complex logistics chains involving multiple parties, the authenticity, integrity, and immutability of monitoring data are difficult to guarantee effectively. Once a safety incident occurs, the determination of responsibility and the tracing of the incident often become difficult due to data disputes.

[0004] Therefore, existing large-item logistics safety monitoring technologies suffer from problems such as incomplete status perception, delayed risk warning, lack of proactive prediction capabilities, and low data reliability. There is an urgent need for an innovative technical solution that can achieve comprehensive real-time status monitoring, proactive risk prediction, and ensure data reliability throughout the entire process. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for monitoring the operational safety of large-item logistics. To achieve the above objective, the embodiments of this invention employ the following technical solutions: In a first aspect, embodiments of the present invention propose a method for monitoring the operational safety of large-item logistics, comprising: Based on real-time multi-source status data collected by multiple heterogeneous sensors deployed on large-item logistics assets, a unified real-time status vector is generated through multi-sensor data fusion algorithm. The digital twin model of large-item logistics assets is updated using real-time state vectors. The digital twin model is a virtualized high-fidelity mapping of large-item logistics assets. A cryptographic hash operation is performed on the real-time state vector to generate a data digest, and the data digest, timestamp, and asset identifier are recorded on the distributed ledger of the blockchain to ensure the immutability and traceability of the real-time state vector; Dual-mode safety monitoring is performed based on a digital twin model. The steps of dual-mode safety monitoring include real-time anomaly monitoring in replication mode and proactive risk prediction in simulation mode. Real-time anomaly monitoring in replication mode includes comparing the real-time state of the digital twin model with a preset safety rule baseline to identify abnormal deviations from the current operating state. Proactive risk prediction in simulation mode includes performing forward-looking simulations on the digital twin model by combining preset route data and external environment data to predict potential instability risks of future path nodes. Security alerts are generated based on the results of real-time anomaly monitoring or proactive risk prediction.

[0006] Preferably, the plurality of heterogeneous sensors include: a microelectromechanical system accelerometer for measuring lateral, longitudinal and vertical triaxial acceleration, a gyroscope for measuring pitch, roll and yaw triaxial angular velocity, and a global positioning system module for acquiring geographic location and velocity information.

[0007] Preferably, the multi-sensor data fusion algorithm employs a state vector fusion method based on a Kalman filter, and the unified real-time state vector is calculated using the following formula: ;in, For discrete time steps, For the posterior estimate of the fusion state at the current moment, These are measurement vectors from multiple heterogeneous sensors. Here is the Kalman gain matrix. It is the identity matrix. For the observation matrix, This is a priori estimate of the state at the previous time step.

[0008] Preferably, the step of recording the data digest on the blockchain is executed by a smart contract deployed on a permissioned blockchain network. The smart contract automatically verifies the format of the data digest and writes it as a transaction into a new block, forming an immutable event log containing a series of state records linked in chronological order.

[0009] Preferably, the preset security rule baseline is defined through a node-based graphical rule editor. The editor is configured to receive the rule definition entered by the user by dragging and dropping functional nodes that represent trigger conditions, logical judgments, and response actions, and to establish connections, and to convert the rule definition into an executable security rule.

[0010] Preferably, the real-time anomaly monitoring step further includes: when the deviation between the real-time state of the digital twin model and the safety rule baseline exceeds a preset threshold, calculating a dynamic risk score, wherein the dynamic risk score is proportional to the magnitude and duration of the deviation.

[0011] Preferably, in proactive risk prediction, the route data includes the radius of curvature of the route. and road slope External environmental data includes real-time wind speed. And wind direction.

[0012] Preferably, the potential instability risk of future path nodes is predicted by calculating the Dynamic Stability Index (DSI). The calculation model for the Dynamic Stability Index is as follows: ,in For total mass, For the height of the asset center of gravity, At the current speed, The lateral acceleration is used; when the DSI is lower than the preset critical stability threshold, it is determined that there is a risk of instability.

[0013] Preferably, the safety alarm includes at least two levels: a secondary warning corresponding to the real-time anomaly monitoring, and a primary alarm corresponding to the proactive risk prediction and triggered when the DSI is below the critical stability threshold; the primary alarm triggers tactile feedback to the driver terminal and sends a suggested deceleration command to the vehicle control unit.

[0014] Secondly, embodiments of the present invention propose an operational safety monitoring system for large-item logistics, comprising: The data acquisition module is configured to collect real-time multi-source status data based on multiple heterogeneous sensors deployed on large-item logistics assets. The data fusion module is configured to process real-time multi-source state data through a multi-sensor data fusion algorithm to generate a unified real-time state vector. The digital twin module is configured to update the digital twin model of large-item logistics assets using real-time state vectors. The digital twin model is a virtualized high-fidelity mapping of the large-item logistics assets. The blockchain recording module is configured to perform cryptographic hash operations on the real-time state vector to generate a data digest, and record the data digest, timestamp, and asset identifier on the distributed ledger of the blockchain; The dual-mode monitoring module is configured to perform dual-mode security monitoring based on a digital twin model. Specifically, the dual-mode monitoring module is configured as follows: Real-time anomaly monitoring is performed in replication mode, including comparing the real-time state of the digital twin model with a preset safety rule baseline to identify abnormal deviations from the current operating state; and Active risk prediction is performed in simulation mode, including combining pre-set route data and external environment data to perform forward simulation of the digital twin model in order to predict the potential instability risk of future path nodes. The alarm generation module is configured to generate security alarms based on the results of real-time anomaly monitoring or proactive risk prediction by the dual-mode monitoring module.

[0015] Beneficial effects: This invention combines multi-source heterogeneous sensor data fusion with a digital twin model to achieve a comprehensive, high-fidelity dynamic mapping of the operational status of large-item logistics assets, overcoming the shortcomings of existing technologies that rely on single, fragmented data leading to incomplete status perception. It can compare the digital twin model with a safety baseline in real time to instantly detect current anomalies, and can also combine route and environmental data for simulation to proactively predict potential instability risks along future paths. This transforms safety monitoring from a reactive, post-event response to a proactive, pre-event warning, improving the timeliness and foresight of warnings. Furthermore, by recording the encrypted hash value of the status data on a blockchain, the immutability and full traceability of the monitoring data are ensured, providing a reliable data trust foundation for accident accountability and multi-party collaboration, solving the problems of data susceptibility to tampering and low reliability in existing technologies. In summary, this invention can reduce the probability of major safety accidents, protect asset and personnel safety, and improve the overall safety management level of large-item logistics. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart of a method for monitoring the operational safety of large-item logistics proposed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a large-item logistics operation safety monitoring system proposed in an embodiment of the present invention.

[0017] Figure 3This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0018] To more clearly illustrate the technical solutions in the embodiments of the invention or the prior art, the invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the invention, but does not constitute a limitation on the invention.

[0019] Firstly, referring to Figure 1 This invention provides a method for monitoring the operational safety of large-item logistics, which can be executed on a computer device. The computer device can be a server, a personal computer (PC), or an embedded large-item logistics operational safety monitoring system with sufficient computing power. The computer device typically includes, but is not limited to, a processor, memory, communication interface, and input / output devices. The memory stores computer program instructions, and the processor executes these instructions to implement the large-item logistics operational safety monitoring method of this invention.

[0020] Reference Figure 1 This invention provides a method for monitoring the operational safety of heavy-duty logistics. In the fields of heavy industry and infrastructure construction, heavy-duty logistics typically involves cross-regional transportation of ultra-long, ultra-wide, ultra-high, and ultra-heavy industrial core equipment. Typical transported objects include, but are not limited to: wind turbine blades exceeding 100 meters in length, substation converter transformers weighing hundreds of tons, core cutterheads and drive components of large tunnel boring machines, high-precision aerospace assembly components, and large chemical fractionation reaction towers with high centers of gravity. Such assets possess extremely high economic value, and their physical form and dynamic characteristics are complex. In long-distance road transport, inland waterway roll-on / roll-off shipping, and multimodal transport scenarios, heavy-duty logistics assets are susceptible to the coupled effects of complex external environments and internal dynamic factors.

[0021] For example, continuous unevenness on non-graded highways can easily cause resonance in the load-bearing system; crosswinds in mountainous and canyon areas can easily cause a surge in overturning moments for cargo with a high center of gravity; small-radius curves on highway ramps can easily cause centripetal imbalance in hydraulic multi-axle heavy trailers; in addition, non-standard operation by drivers also poses potential risks. The combination of these factors can easily induce serious safety accidents such as relative displacement caused by cargo lashing failure, global instability caused by shift of the load-bearing platform's center of gravity, and even overall overturning.

[0022] Combination Figure 1The flowchart shown illustrates the operational safety monitoring method for large-item logistics provided in this embodiment of the invention. It relies on a distributed computing architecture comprised of a high-precision sensor network deployed at the large-item logistics asset end, an edge computing gateway located within the vehicle's main control box, and a cloud data center deployed in a remote center. The method encompasses the following implementation steps: S101 is based on real-time multi-source status data collected by multiple heterogeneous sensors deployed on large-item logistics assets. It is processed by a multi-sensor data fusion algorithm to generate a unified real-time status vector.

[0023] This step aims to address the data silo problem in the transportation of large items by constructing a perception network that tracks the physical form and multi-degree-of-freedom kinematic state of large logistics assets (including vehicles and cargo). It involves distributed acquisition of raw sensor signals across multiple physical dimensions and low-latency fusion processing at the edge to extract high-precision unified state information representing the dynamic characteristics of the assets, providing reliable data support for subsequent twin-driven simulations. Specifically, it includes the following sub-steps: S1011. Deploy a heterogeneous sensor network and perform continuous data acquisition of multidimensional physical quantities.

[0024] For large-item logistics assets, it is necessary to deploy sensors targeting key stress and motion characteristics. Specifically, this includes the tractor unit of special transport vehicles, multi-axle trailer frames using hydraulic splicing technology, dedicated pallet systems, and key stress nodes of the cargo itself (such as transformer lifting lugs and wind turbine blade root flanges). At these nodes, heterogeneous sensor networks are deployed through rigid physical assembly and electrical connections. To characterize the six-degree-of-freedom motion state and micro-stress of the large-item logistics system in three-dimensional space, Preferably, the plurality of heterogeneous sensors include: a micro-electro-mechanical systems (MEMS) accelerometer for measuring lateral, longitudinal, and vertical triaxial acceleration; a gyroscope for measuring pitch, roll, and yaw triaxial angular velocities; and a Global Positioning System (GPS) module for acquiring geographic location and velocity information.

[0025] Specifically, MEMS accelerometers utilize the mechanical deformation of their internal microelectromechanical structures to sense external inertial forces, acquiring real-time data on instantaneous impact responses during transport, lateral centripetal acceleration when vehicles curve, longitudinal acceleration and deceleration impacts, and vertical high-frequency vibration acceleration induced by road surface excitation at high sampling frequencies (e.g., 100Hz to 500Hz). This provides fundamental data for assessing the effectiveness of lashing. Gyroscopes, based on the Coriolis force principle, accurately collect the instantaneous rotational motion trend of assets in three-dimensional space. Roll velocity is a core indicator for assessing the risk of asset rollover and instability, while yaw velocity reflects the tracking ability and tail-wagging risk of ultra-long trailer trains. Furthermore, the Global Positioning System (GPS) module refers to a multi-constellation compatible global navigation satellite system, which, combined with real-time dynamic carrier phase differential technology, provides high-precision absolute geographic latitude and longitude coordinates and altitude parameters for macroscopic trajectory tracking in wide-area environments.

[0026] S1012, Heterogeneous data preprocessing and spatiotemporal alignment at the edge.

[0027] Because the aforementioned sensors are physically distributed across different nodes of the large asset (e.g., the GPS antenna is located on top of the tractor, and the inertial measurement units are distributed at the rear of the trailer and at the equivalent plane of the cargo's center of gravity), and because each sensor has differences in internal clock crystal oscillator accuracy, sampling frequency, measurement range, system background noise characteristics, and underlying data communication protocol specifications (such as RS-485 serial bus, industrial Ethernet, etc.), this multi-source heterogeneity makes it difficult to directly apply the raw asynchronous data stream to high-precision safety decision-making.

[0028] Therefore, each sensor acquisition terminal transmits raw data frames with hardware timestamps to the edge computing gateway deployed on the transportation platform via the Controller Area Network (CAN) bus. The edge computing gateway first uses a precise time synchronization protocol to achieve time synchronization and millisecond-level alignment of the sensor nodes. Second, the gateway's processing core performs three-dimensional coordinate system unification processing based on the rotation matrix and translation vector calibrated by the sensors, uniformly transforming the heterogeneous sensor data into a reference coordinate system with the overall centroid of the large component system as the origin. Finally, it applies a sliding window mid-range filter or adaptive low-pass filter algorithm to perform basic signal denoising processing, eliminating pulse-type outlier noise caused by electromagnetic interference or road surface bounce, thereby outputting a time-synchronized and spatially aligned real-time multi-source state data sequence.

[0029] S1013, Deep fusion and state extraction of multi-sensor data based on Kalman filter.

[0030] After acquiring preprocessed real-time multi-source state data, a multi-sensor data fusion algorithm is invoked in the edge computing gateway microprocessor. In real-world operating environments, MEMS accelerometers and gyroscopes are susceptible to temperature drift and low-frequency white noise integral effects during long-term continuous operation, resulting in cumulative attitude calculation errors (integral drift). Simultaneously, GPS signals are prone to multipath effects or signal attenuation in areas with obstructed views, such as tunnels, tall buildings, or canyons, leading to abrupt changes or temporary failures in positioning coordinates. A single sensor source is insufficient to provide robust state estimation.

[0031] To overcome the aforementioned technical bottlenecks, preferably, the multi-sensor data fusion algorithm employs a state vector fusion method based on a Kalman filter, and the unified real-time state vector is calculated using the following formula:

[0032] In the formula: The discrete time step represents the current sampling and calculation period; It is a posterior estimate of the fused state at the current moment, which includes the optimal state information after fusion; These are measurement vectors from multiple heterogeneous sensors; This is the Kalman gain matrix, used to dynamically adjust the weights between the theoretically predicted state and the actual measured state; It is the identity matrix; The observation matrix represents the mapping relationship from the state space to the observation space; This is a priori estimate of the state at the previous time step.

[0033] During data fusion, the Kalman filter uses a dynamic prior model to perform optimal estimation with measurement data containing Gaussian noise. This is particularly effective when the current measurement noise covariance of the sensor is small (i.e., the measurement data has high reliability). Increasing the value of the matrix elements assigns a value to the observation. Greater weighting; conversely, when the measurement noise covariance increases, As element values ​​decrease, they rely more on prior state estimates. This mechanism effectively suppresses random Gaussian errors from multi-source sensors, outputting a smooth, continuous, and high-confidence unified real-time state vector, providing data support for subsequent digital twin model-driven operations.

[0034] S102, update the digital twin model of the large-item logistics asset using the real-time state vector, wherein the digital twin model is a virtualized high-fidelity mapping of the large-item logistics asset.

[0035] Digital twin technology, serving as a medium for interaction between physical entities and virtual digital spaces, is the infrastructure for intelligent management and control of large-item logistics assets. This step aims to deploy and maintain a virtual mirror model synchronized with the physical status of the large-item logistics assets within a cloud data center. This model possesses accurate geometric appearance and physical and mechanical properties, providing a visual and computational support environment for subsequent safety monitoring, logical rule analysis, and stress prediction.

[0036] S1021. Construct a static physical and geometric twin base model for large-item logistics assets.

[0037] Before the transportation mission begins, system engineers initialize and build a static virtual twin of the asset on a cloud server. This process generates an accurate 3D geometric mesh topology model by importing computer-aided design or 3D digital manufacturing drawings of large cargo.

[0038] The core characteristic of a digital twin model lies in its high physical fidelity. Rigorous physical properties and material dynamics parameters must be assigned to the geometric model. The backend configuration module requires input of the following parameters: the mass distribution of each component of the cargo, the three-dimensional absolute coordinates of the overall composite center of mass, the spatial multi-directional rotational inertia matrix about each principal axis of inertia, the nonlinear lateral stiffness and vertical radial stiffness of the multi-axle vehicle tires, the spring stiffness coefficient and damping attenuation coefficient of the hydraulic suspension system, and the tensile modulus and breaking strength limit of the restraint straps. The precise input of these parameters forms the underlying logical foundation for the digital twin model's simulation calculations.

[0039] S1022. Dynamic driving and optimization of digital twin models based on real-time state vectors and prediction results.

[0040] After the static base model is configured, during asset transportation, the vehicle-mounted edge computing gateway continuously sends the unified real-time state vector generated in step S101 to the cloud server, relying on mobile communication technology or broadband satellite communication network.

[0041] The cloud-based digital twin module extracts the position coordinates, spatial attitude angles, and kinematic derivatives from the vectors and inputs them as boundary conditions into the virtual physics engine. The physics engine drives the 3D model in virtual space in real time to generate translation, rotation, and deformation mappings consistent with the real world. For example, when the real-time state vector indicates that the vehicle is turning right in a mountainous area and the lateral centripetal acceleration is increasing, the digital twin rendering engine simultaneously presents the virtual vehicle's attitude of tilting to the left of the curve. At the same time, the solver calculates and updates the internal and external force states in real time based on the input stiffness and damping parameters, such as the change in the vertical ground load of the outer tire.

[0042] Furthermore, the digital twin model and the system prediction module form a closed-loop feedback mechanism. The digital model continuously receives prediction results generated in subsequent steps and makes real-time dynamic adjustments and optimizations based on these results. For example, if the prediction module indicates that a bumpy road section will occur in the future, the digital twin model can dynamically adjust the response parameter boundaries of the virtual hydraulic suspension in advance in the virtual environment. By evaluating the impact of parameter adjustments on overall stability through pre-simulation, it can achieve dynamic optimization of management strategies based on data and prediction.

[0043] S103, perform a cryptographic hash operation on the real-time state vector to generate a data digest, and record the data digest, timestamp, and asset identifier on the distributed ledger of the blockchain to ensure the immutability and traceability of the real-time state vector.

[0044] Large-item logistics involves multiple stakeholders, including manufacturers, logistics companies, insurance underwriters, transportation regulators, and third-party service providers. In past dispute resolution processes, determining liability for accidents has often been contentious due to information asymmetry, conflicting accounts of core data, or unilateral tampering. This step integrates cryptographic mechanisms with blockchain technology to reduce reliance on centralized storage architectures, build a decentralized, multi-party consensus-based distributed trust mechanism, and ensure the tamper-proof nature and credibility of core operational history data.

[0045] S1031. Perform cryptographic hash operations to extract the data digest.

[0046] Given the massive amount of real-time state vector data output from the edge during long-distance transportation, directly uploading all the original plaintext data to the blockchain would cause network congestion, increase consensus latency, and increase storage costs.

[0047] Therefore, an architectural strategy of "on-chain storage of certificates and off-chain storage of plaintext" is adopted. At the vehicle-mounted edge computing gateway or on-chain proxy encryption node, a cryptographic hash function (such as a 256-bit secure hash algorithm) is used to perform one-way hashing on the real-time state vector structure within a specific time window (such as a fixed period or a key event trigger point).

[0048] This hash operation maps an input dataset of arbitrary length to a fixed-length (256-bit) output string, known as a "data digest." The data digest is cryptographically collision-resistant and unidirectionally irreversible. Any tiny change in the original real-time state vector will cause an avalanche-like change in the generated data digest. Furthermore, it is impossible to reverse-engineer the original driving state data from the intercepted data digest, thus protecting both commercial privacy and data security.

[0049] S1032. Utilize blockchain smart contract mechanisms to achieve distributed consensus and evidence storage across the entire network.

[0050] Preferably, the step of recording the data digest on the blockchain is executed by a smart contract deployed on a permissioned blockchain network. The smart contract automatically verifies the format of the data digest and writes it as a transaction into a new block, forming an immutable event log containing a series of state records linked in chronological order.

[0051] Specifically, the permissioned blockchain architecture relies on digital certificate authorities to authenticate the identities of access nodes, allowing only authorized stakeholder nodes (such as insurance institutions, traffic management departments, and logistics company nodes) to participate in endorsement, sorting, and distributed ledger maintenance. Smart contracts are automated business logic code deployed in a blockchain virtual machine container.

[0052] Once the edge node generates a data digest, the gateway program assembles it together with a high-precision, trusted timestamp and a digital asset identifier (such as a vehicle identification number, waybill number, and encrypted digital signature) to uniquely identify the large logistics asset, and sends it to the permissioned blockchain network as a data transaction request.

[0053] Upon receiving the request, a smart contract node in the blockchain network automatically executes verification logic, including verifying the legality of the data digest format, the identity and permissions of the initiating node, and the authenticity of the digital signature. After successful endorsement verification, the transaction is submitted to the sorting node; the sorting node packages the transaction into a new block and broadcasts it to all ledger nodes in the network. Once all nodes reach a consensus based on the consensus algorithm, the block is appended to the distributed ledger database.

[0054] Because the distributed ledger maintains a complete copy on each authorized node, and each new block contains a hash pointer to the predecessor block, any unilateral act of tampering with historical state data requires simultaneous tampering with all subsequent blocks and synchronous implementation on more than half of the nodes in the network, which is difficult to achieve under the existing network architecture.

[0055] Therefore, in the event of a transportation safety incident, the joint investigation team can extract the original plaintext status data stored offline, either locally or in the cloud, recalculate the hash value using the same hash algorithm, and rigorously compare it with the data digest published on the blockchain. If the comparison results match, it can be cryptographically verified that the status data has not been tampered with since its generation and upload to the blockchain, thereby ensuring the immutability and traceability of the monitoring data and simplifying the trust establishment and compensation process in dispute resolution.

[0056] S104, Perform dual-mode security monitoring based on digital twin model. The steps of performing dual-mode security monitoring include real-time anomaly monitoring in replication mode and proactive risk prediction in simulation mode.

[0057] This step constitutes the analysis and decision-making center of the entire large-item logistics safety monitoring and management system. Unlike traditional monitoring systems that rely solely on single-dimensional sensors and set fixed alarm thresholds, this embodiment of the invention relies on a cloud-based, dynamically updated digital twin platform to construct a "replication-simulation" dual-modal parallel safety monitoring architecture. This aims to achieve a paradigm shift in large-item transportation safety monitoring from reactive post-event response to proactive pre-event prediction.

[0058] S1041, Implement a real-time anomaly monitoring mechanism in replication mode.

[0059] The replication mode focuses on frequently evaluating the operational status parameters of the digital twin model at the current time point to determine whether it is within the designed boundary threshold range. Preferably, the preset security rule baseline is defined through a node-based graphical rule editor. The editor is configured to receive rule definitions input by the user by dragging and dropping functional nodes representing trigger conditions, logical judgments, and response actions, and establish connections, and convert the rule definitions into executable security rules.

[0060] Due to the varying application scenarios in large-item logistics, safety constraint standards differ for different goods. The graphical editor provides a visual interactive interface, allowing engineers to drag and drop functional nodes such as "tilt angle" and "binding stress," and flexibly configure thresholds and logical connections to point to corresponding response action nodes. The intelligent compilation engine transforms these graphical rule definitions into underlying parsed safety monitoring rule scripts, thereby tailoring a rigorous safety rule baseline for individual transportation tasks.

[0061] When the digital twin model is in operation with the physical vehicle, the cloud monitoring engine reads the latest physical state vector in a high-frequency polling manner and inputs it into the safety rule engine for verification. Preferably, the real-time anomaly monitoring step further includes: when the deviation between the real-time state of the digital twin model and the safety rule baseline exceeds a preset threshold, calculating a dynamic risk score, wherein the dynamic risk score is proportional to the magnitude and duration of the deviation.

[0062] By introducing a dynamic risk scoring model, the drawback of frequent false alarms caused by a single fixed threshold is overcome. The calculus engine employs a time definite integral algorithm to numerically integrate and quantify the area of ​​the portion of the "real-time deviation - duration" two-dimensional curve that exceeds the safety baseline threshold. This continuous integration mechanism can accurately assess the cumulative safety threat level at the current moment, effectively filter out false alarms caused by transient impulse noise, and ensure the effective identification and risk weighting of potential hazards with longer durations (such as the prolonged maintenance of minor overspeed conditions).

[0063] S1042. Implement proactive risk forecasting mechanism and dynamic strategy optimization in simulation mode.

[0064] The simulation mode expands the defense boundaries of the monitoring system, giving it the ability to predict risks. Preferably, route data includes the radius of curvature and road slope, while external environmental data includes real-time wind speed and direction. The potential instability risk of future path nodes is predicted by calculating a dynamic stability index; when the dynamic stability index falls below a preset critical stability threshold, an instability risk is identified.

[0065] Because fully loaded heavy vehicles have enormous inertia and limited visibility, conventional response measures are often delayed in the face of sudden emergencies due to the long braking distance. Therefore, it is necessary to conduct dynamic risk simulations before the vehicle reaches the danger zone.

[0066] In the simulation calculations for proactive risk prediction, route data is obtained by calling cloud-based high-precision map services to pre-analyze the geometric and topological features of the road segment ahead; external meteorological environmental data is captured in real time through a meteorological interface. Given that large cargo has a large windward area, crosswind-induced aerodynamic overturning moment is one of the key physical factors leading to instability.

[0067] The dynamic stability index is calculated using a multivariable nonlinear function, taking into account the following core physical parameters: the total mass of the trailer and cargo; the equivalent comprehensive absolute center of mass height of the large-item logistics asset combination; the expected initial velocity of the vehicle entering the future predicted node during simulation; the expected lateral centrifugal acceleration calculated based on the geometric radius of curvature of the curve ahead and the expected velocity; characteristic parameters such as the ambient air density in the predicted node area; the road gradient of the road network ahead (including longitudinal slope and cross-sectional superelevation); and the absolute real-time ambient wind speed at the future predicted node.

[0068] To enhance the time series analysis capabilities and accuracy of proactive risk prediction, a Long Short-Term Memory (LSTM) neural network algorithm was deeply integrated to construct a time series prediction model before performing dynamic simulation. The driving behavior and state evolution in heavy-duty logistics transportation exhibit temporal correlation characteristics (such as periodic braking behavior on long downhill slopes in mountainous areas). The LSTM neural network, through its gated cell structure, effectively extracts and memorizes long-term temporal information.

[0069] In the implementation phase, massive amounts of multi-source heterogeneous data collected in the early stages (covering historical speed fluctuations, acceleration sequences, steering sequences, and environmental road condition labels) are input into the LSTM model for feature extraction and training. The trained LSTM model can output time series prediction results within the future prediction time window (such as expected trajectory offset and speed trend). This prediction result is used as a dynamic prior parameter input into the dynamic stability index model for numerical solution. When the simulation results show that the vertical ground support force of the inner wheel approaches zero, causing the calculated dynamic stability index to be lower than the preset critical stability safety threshold, it is determined that there is a risk of physical instability and overturning at the future path node.

[0070] Based on the prediction results, a management closed loop is further formed, and the relevant physical parameter boundaries or algorithm strategies in the cloud digital twin model are dynamically adjusted in real time. For example, the safety tolerance margin of the virtual suspension model is adjusted in advance, thereby optimizing the lead time of the safety management strategy.

[0071] S105 generates a security alert based on the results of real-time anomaly monitoring or proactive risk prediction.

[0072] After completing multi-dimensional real-time monitoring and in-depth predictive decision-making, the digital risk assessment results are transformed into actionable warnings and emergency intervention commands in the physical world. Preferably, the safety alarm includes at least two levels: a level two warning corresponding to real-time anomaly monitoring, which notifies the driver via visual prompts and voice broadcasts through the in-vehicle human-machine interface terminal; and a level one warning corresponding to proactive risk prediction, triggered when the dynamic stability index falls below the critical stability threshold, which triggers mandatory tactile feedback and sends a suggested deceleration command to the vehicle control unit via the vehicle bus to automatically implement torque reduction and active braking.

[0073] In practice, a tiered response control strategy is matched according to the urgency of the risk: For Level 2 alerts, a Level 2 alert is triggered when the replication pattern detects initial signs of deviation in the physical state of the asset, or when the dynamic risk score is in the low to medium safety range (e.g., slight fluctuations in the tension of unilateral fasteners but not below the safety lower limit). This alert signal is sent to the intelligent in-vehicle human-machine interface (HMI) terminal in the tractor's cab, providing visual cues (such as a flashing yellow warning icon) accompanied by clear voice prompts. This level of alert aims to guide the driver to autonomously and smoothly reduce speed, pull over, and conduct a physical check when road conditions are manageable, without triggering an overreaction.

[0074] The Level 1 alert serves as the highest level of safety intervention. When simulation mode and LSTM prediction determine that the vehicle faces a very high probability of serious loss of control risk when passing through the road ahead in a short period of time, conventional HMI visual or voice prompts are insufficient to address driver fatigue or slow reaction. In this case, the Level 1 alert will trigger a multi-sensory, high-intensity physical wake-up mechanism for the driver (such as triggering the seat belt pretensioner motor to tighten rapidly and the steering wheel eccentric motor to vibrate at high frequency) to forcibly relieve potential fatigue.

[0075] More importantly, the Level 1 alarm, while triggering haptic feedback, will simultaneously send a forced deceleration command with the highest execution weight to the Vehicle Control Unit (VCU) via a high-priority underlying vehicle control bus. Upon receiving this command, the chassis system or advanced driver assistance system with drive-by-wire capabilities will, under the support of the anti-lock braking algorithm and while maintaining trailer stability, collaboratively execute active torque reduction via the engine's electronic throttle, and coordinate with the hydraulic retarder and pneumatic service brakes to implement high-intensity, smooth, active braking. This underlying physical intervention effectively controls the actual speed of the heavy-duty vehicle group, ensuring it safely returns to the absolute safe speed threshold calculated based on the dynamic stability index before reaching the high-risk prediction point. Through these underlying physical execution methods, the evolution chain of a safety incident can be effectively blocked in the physical world, thereby preventing major logistics safety accidents.

[0076] Secondly, please refer to Figure 2 This invention also proposes an operational safety monitoring system for large-item logistics, adapted to the aforementioned operational safety monitoring method for large-item logistics. This system can comprehensively, in real-time, and reliably monitor the operational status of large-item logistics assets in transit and provide forward-looking risk warnings, thus solving problems such as incomplete status perception, delayed risk warnings, and low data reliability in the background technology. The system is described in detail below with reference to specific embodiments. The system includes a data acquisition module, a data fusion module, a digital twin module, a blockchain recording module, a dual-mode monitoring module, and an alarm generation module. These modules work collaboratively to form an integrated intelligent safety monitoring framework.

[0077] The data acquisition module is configured to collect real-time, multi-source status data from multiple heterogeneous sensors deployed on large-item logistics assets. Specifically, the hardware foundation of this module involves physically deploying a heterogeneous sensor network at key locations on the large-item logistics assets (such as the chassis, containers, or cargo itself of special transport vehicles). This network preferably integrates a high-precision MEMS accelerometer to capture dynamic acceleration changes along the asset's lateral, longitudinal, and vertical axes; a three-axis gyroscope to measure the asset's pitch, roll, and yaw angular velocities in real time for accurate attitude changes; and a GPS module to acquire high-precision geographic location, altitude, and real-time speed information. These sensors are connected to an onboard data gateway via wired or low-power wireless protocols (such as Zigbee or Bluetooth). This gateway is responsible for initial signal conditioning and timestamp synchronization of the raw sensor signals, ensuring the temporal consistency of the multi-source data and providing high-quality raw data input for subsequent data fusion.

[0078] The data fusion module is configured to process the real-time multi-source state data using a multi-sensor data fusion algorithm to generate a unified real-time state vector. This module is typically implemented in software within an onboard computing unit or edge computing device. Its core is running an optimized multi-sensor data fusion algorithm, such as the Kalman filter-based state estimation algorithm mentioned in the previous embodiment. This module receives a timestamped, multi-dimensional raw data stream from the data acquisition module and uses a pre-defined asset kinematic model to recursively predict and correct these heterogeneous data, which contain noise and drift. Through this process, the module effectively integrates acceleration, angular velocity, position, and velocity information, filters out random errors, and outputs a single, smooth, and highly accurate real-time state vector. This vector comprehensively describes the precise operating attitude and dynamic characteristics of the asset at any given time, serving as an authoritative data source for driving digital twin models and conducting safety analysis.

[0079] A digital twin module is configured to update the digital twin model of the large-item logistics asset using the real-time state vector. This module is a complex software system running on a server or cloud platform, responsible for maintaining a virtual copy that is synchronized with the physical asset in real time. Before the task begins, the module constructs a high-fidelity digital twin model based on the asset's geometric model, physical parameters (such as mass, center of mass height, and moment of inertia), and dynamic characteristics. During operation, the module continuously receives a unified real-time state vector from the data fusion module, which drives corresponding changes in the virtual model's position, orientation, and motion state in digital space, thereby achieving a precise mapping between the physical and digital worlds. Preferably, the module can be built according to the Asset Administration Shell (AAS) standard, managing the asset's static technical specifications and dynamic operational data through standardized sub-models, ensuring model interoperability and scalability.

[0080] The blockchain recording module is configured to perform cryptographic hash operations on the real-time state vector to generate a data digest, and record the data digest, timestamp, and asset identifier on the blockchain's distributed ledger. This module aims to provide immutable evidence for critical state data. It acts as an interface connecting the monitoring system to a permissioned blockchain network (such as Hyperledger Fabric). Whenever the data fusion module generates a new state vector, this module calls a secure hash function (such as SHA-256) to perform the operation, generating a unique, fixed-length data digest. Subsequently, the module executes a pre-built smart contract to package this data digest, the current timestamp, and the asset's unique identifier into a transaction, submitting it to the blockchain network for consensus and recording. By recording the data's "digital fingerprint" on the blockchain, the integrity and non-repudiation of the monitoring history are ensured, providing strong and reliable data support for potential incident tracing and liability determination.

[0081] The dual-mode monitoring module, configured to perform dual-mode security monitoring based on the digital twin model, is the core of the entire system's decision-making. This module is further configured to operate in two modes: Firstly, real-time anomaly monitoring is performed in replication mode. In this mode, the module performs high-frequency comparisons between real-time status parameters (such as real-time roll angle and lateral acceleration) provided by the digital twin module and a preset safety rule baseline. This baseline is defined by logistics safety experts through a graphical rule editor and includes various safety thresholds and logical conditions. Once a deviation from the safety baseline is detected in the real-time status, the module will determine it as an anomaly and can further calculate a dynamic risk score to comprehensively assess the severity and duration of the deviation.

[0082] Secondly, proactive risk prediction is performed in simulation mode. In this mode, the module combines pre-set high-precision route data (such as the radius of curvature and slope of the road ahead) and external environmental data (such as real-time wind speed) to perform forward-looking dynamic simulations on the digital twin model. Through simulation, the module can predict the dynamic stability of the asset when it is about to pass through future path nodes (such as sharp bends or steep slopes) and calculate key indicators such as DSI. When the predicted indicators are lower than the safety threshold, the module determines that there is a potential risk of instability.

[0083] The alarm generation module is configured to generate safety alarms based on the results of real-time anomaly monitoring or proactive risk prediction from the dual-mode monitoring module. This module is the system's execution end, responsible for transforming monitoring results into effective intervention measures. It triggers different levels of alarms based on the risk level output by the dual-mode monitoring module. For example, for minor deviations detected by real-time anomaly monitoring, a level two warning is generated, providing text or visual prompts to the driver via the in-vehicle screen. For potential instability risks or serious real-time anomalies detected by proactive risk prediction, a level one alarm is generated, providing the highest level of warning through rapid sound, strong tactile feedback on the driver's seat or steering wheel, etc. It can even proactively link with the vehicle's driver assistance system to send suggested deceleration commands, thereby achieving a closed-loop safety management system from passive response to proactive prevention.

[0084] The third aspect of this application provides a computer device for monitoring the operational safety of large-item logistics. Please refer to [link to relevant documentation]. Figure 3 The device includes a memory and a processor connected in series and in communication. The memory stores a computer program, and the processor reads the computer program and executes a method for monitoring the operational safety of large-item logistics as described in the first aspect of the embodiment. Specifically, the memory may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or last-in-first-out (FILO) memory, etc.; the processor may be, but is not limited to, microprocessors of the STM32F105 series, ARM (Advanced RISC Machines), x86 architecture processors, or processors with integrated NPUs (neural-network processing units). The working process, working details, and technical effects of the device provided in the third aspect of this embodiment can be found in the first aspect of the embodiment, and will not be repeated here.

[0085] This fourth aspect of the embodiment provides a computer-readable storage medium storing instructions containing the operational safety monitoring method for large-item logistics according to the first aspect of the embodiment. Specifically, the computer-readable storage medium stores instructions that, when executed on a computer, perform the operational safety monitoring method for large-item logistics as described in the first aspect. The computer-readable storage medium refers to a data storage medium, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or Memory Sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device systems. The working process, details, and technical effects of the computer-readable storage medium provided in this fourth aspect of the embodiment can be found in the first aspect of the embodiment, and will not be repeated here.

[0086] The fifth aspect of this embodiment provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform a method for monitoring the operational safety of large-item logistics as described in the first aspect of this embodiment. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0087] The embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the prior art, can be embodied in the form of software products. These computer software products can be stored in computer-readable storage media, such as ROM / RAM, magnetic disks, optical disks, etc., and include several instructions to cause a computer or programmable device to execute the methods of the various embodiments or certain parts of the foregoing embodiments.

[0089] Finally, it should be noted that although embodiments of the present invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for monitoring the operational safety of large-item logistics, characterized in that, include: Based on real-time multi-source status data collected by multiple heterogeneous sensors deployed on large-item logistics assets, a unified real-time status vector is generated through multi-sensor data fusion algorithm. The digital twin model of the large-item logistics asset is updated using the real-time state vector. The digital twin model is a virtualized high-fidelity mapping of the large-item logistics asset. A cryptographic hash operation is performed on the real-time state vector to generate a data digest, and the data digest, timestamp, and asset identifier are recorded on the distributed ledger of the blockchain to ensure the immutability and traceability of the real-time state vector; Dual-mode safety monitoring is performed based on the digital twin model. The steps of performing dual-mode safety monitoring include real-time anomaly monitoring in replication mode and proactive risk prediction in simulation mode. The real-time anomaly monitoring in replication mode includes comparing the real-time state of the digital twin model with a preset safety rule baseline to identify abnormal deviations from the current operating state. The proactive risk prediction in simulation mode includes performing forward-looking simulations on the digital twin model by combining preset route data and external environment data to predict potential instability risks of future path nodes. A security alert is generated based on the results of the real-time anomaly monitoring or the proactive risk prediction.

2. The method according to claim 1, characterized in that, The plurality of heterogeneous sensors include: a microelectromechanical system accelerometer for measuring lateral, longitudinal and vertical triaxial acceleration; a gyroscope for measuring pitch, roll and yaw triaxial angular velocity; and a global positioning system module for acquiring geographic location and velocity information.

3. The method according to claim 1, characterized in that, The multi-sensor data fusion algorithm employs a state vector fusion method based on a Kalman filter, and the unified real-time state vector is calculated using the following formula: ;in, For discrete time steps, For the posterior estimate of the fusion state at the current moment, These are measurement vectors from the multiple heterogeneous sensors. Here is the Kalman gain matrix. It is the identity matrix. For the observation matrix, This is a priori estimate of the state at the previous time step.

4. The method according to claim 1, characterized in that, The step of recording data summaries on the blockchain is executed by smart contracts deployed on a permissioned blockchain network. These smart contracts automatically verify the format of the data summaries and write them as a transaction into a new block, forming an immutable event log containing a series of state records linked in chronological order.

5. The method according to claim 1, characterized in that, The preset security rule baseline is defined through a node-based graphical rule editor, which is configured to receive rule definitions input by the user by dragging and dropping functional nodes representing trigger conditions, logical judgments and response actions, and establish connections, and convert the rule definitions into executable security rules.

6. The method according to claim 1, characterized in that, The real-time anomaly monitoring step further includes: when the deviation between the real-time state of the digital twin model and the safety rule baseline exceeds a preset threshold, calculating a dynamic risk score, wherein the dynamic risk score is proportional to the magnitude and duration of the deviation.

7. The method according to claim 1, characterized in that, In the proactive risk prediction, the route data includes the radius of curvature of the route. and road slope The external environmental data includes real-time wind speed. And wind direction.

8. The method according to claim 7, characterized in that, The prediction of potential instability risk of future path nodes is achieved by calculating the Dynamic Stability Index (DSI), and the calculation model of the DSI is as follows: ,in For total mass, For the height of the asset center of gravity, At the current speed, The lateral acceleration is used; when the DSI is lower than the preset critical stability threshold, it is determined that there is a risk of instability.

9. The method according to claim 1, characterized in that, The safety alarm includes at least two levels: a level two warning corresponding to the real-time anomaly monitoring, and a level one alarm corresponding to the proactive risk prediction and triggered when the DSI is below the critical stability threshold; the level one alarm triggers tactile feedback to the driver terminal and sends a suggested deceleration command to the vehicle control unit.

10. A safety monitoring system for the operation of large-item logistics, characterized in that, include: The data acquisition module is configured to collect real-time multi-source status data based on multiple heterogeneous sensors deployed on large-item logistics assets. The data fusion module is configured to process the real-time multi-source state data through a multi-sensor data fusion algorithm to generate a unified real-time state vector; The digital twin module is configured to update the digital twin model of the large-item logistics asset using the real-time state vector, wherein the digital twin model is a virtualized high-fidelity mapping of the large-item logistics asset. The blockchain recording module is configured to perform a cryptographic hash operation on the real-time state vector to generate a data digest, and record the data digest, timestamp, and asset identifier on the distributed ledger of the blockchain; A dual-mode monitoring module is configured to perform dual-mode security monitoring based on the digital twin model, wherein the dual-mode monitoring module is configured as follows: Real-time anomaly monitoring is performed in replication mode, including comparing the real-time state of the digital twin model with a preset security rule baseline to identify abnormal deviations from the current operating state; and Active risk prediction is performed in simulation mode, including combining preset route data and external environment data to perform forward simulation on the digital twin model in order to predict the potential instability risk of future path nodes. The alarm generation module is configured to generate a security alarm based on the results of real-time anomaly monitoring or proactive risk prediction by the dual-mode monitoring module.

Images