Unmanned shipping system based on intelligent weighing

Abstract

The invention relates to the technical field of intelligent logistics and unmanned shipping, in particular to an unmanned shipping system based on intelligent weighing, which comprises a multi-modal data acquisition module, a distributed weighing and dynamic optimization module, a loading execution module, a detection and verification module and an electronic shipping order generation module. The system calculates the loading sequence and path of goods through a dynamic loading optimization model, and monitors the gravity center change and load distribution of the vehicle by using an inertial sensor and a piezoelectric sensor to realize real-time adjustment and optimization; through the combination of an electronic shipping order and a digital twin entity, the system supports whole-process state tracking and abnormal early warning, and the transportation safety and the management efficiency are ensured; according to the invention, the intelligence and efficiency of logistics operation are obviously improved, and comprehensive technical support is provided for modern logistics and factory dispatching.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent logistics and unmanned shipping, and in particular to an unmanned shipping system based on intelligent weighing. Background Art

[0002] The unmanned shipping system is of great significance to the cargo shipping link of modern logistics and factories. The shipping process is a key link in the flow of goods, and its efficiency and accuracy directly affect the operating efficiency and logistics costs of the enterprise. In the traditional shipping method, manual operation is prone to errors, and the efficiency of vehicles entering and leaving the factory is inefficient, leading to congestion, chaotic management and other problems. In the prior art (Chinese invention patent, publication number: CN113108884A, name: An AI unattended weighing and shipping method) it is mainly achieved through the following technical means: using RFID cards and remote card readers to record and identify vehicle information; using a floor scale in a fixed weighing area to weigh the initial weight and the weight after loading of the vehicle; and storing and verifying the transportation data through a cloud database. However, this technology has the following defects: The existing solution only determines the vehicle loading status by weight difference and cannot monitor the position of the vehicle's center of gravity, which may cause vehicle instability during loading and pose a safety hazard; unexpected situations during loading (such as incorrect placement of cargo or weight deviation) cannot dynamically adjust the loading order or path in real time, and the system adaptability is low; during the synchronization of cargo information recorded by RFID cards with the cloud database, there is a lack of real-time monitoring and feedback mechanism, and it is unable to respond to dynamic changes in the logistics process; the loading process does not combine cargo size with the compartment space model, which may lead to space waste or improper loading. Summary of the invention

[0003] In view of the many problems existing in the above-mentioned prior art, the present invention provides an unmanned shipping system based on intelligent weighing. The present invention is based on a technical framework of multimodal data collection, dynamic optimization, real-time monitoring and full-process tracking, and utilizes distributed weighing and dynamic optimization models, and optimizes the cargo loading sequence and path through the combination of reinforcement learning and cellular automation rules. At the same time, dynamic center of gravity monitoring and deviation verification ensure loading safety, and digital twin entity synchronization technology realizes full-process status tracking and management. The present invention significantly improves logistics efficiency, load balancing and transportation safety.

[0004] An unmanned shipping system based on intelligent weighing, comprising: A multimodal data processing module is used to collect multimodal data including vehicle feature data, driver feature data and cargo feature data; perform quantum encoding on the multimodal data, map the multimodal data to a high-dimensional space using a kernel function method, extract core feature associations between data through high-dimensional association analysis, generate associated feature data, and generate digital twin entity data in combination with task information; Distributed weighing and optimization module, which is used to calculate the weight distribution model through vehicle contact pressure data, and generate a dynamic loading optimization model using reinforcement learning optimization algorithm in combination with vehicle specification information and cargo characteristic data; The loading execution module includes a loading robot, which is used to generate loading task data according to the dynamic loading optimization model data and perform cargo loading operations; monitor the changes in the center of gravity of the vehicle through inertial sensors, capture the changes in the load during cargo loading through piezoelectric sensors, and adjust the loading sequence and path in real time in combination with the dynamic interactive prediction algorithm to optimize the loading process; the dynamic center of gravity distribution data generated during the loading process and the cargo placement status data together constitute the loading status data; The detection and verification module is used to collect the vehicle's detection data after loading is completed, generate verification results based on the comparison of the detection data with the loading status data, and generate an electronic shipping order when the verification passes.

[0005] Preferably, the spectral scanning device in the multimodal data processing module transmits and receives spectral data reflected by the vehicle surface through a multi-band light source, and generates vehicle spectral feature data in combination with a preset spectral feature matching algorithm; the facial recognition camera classifies and extracts the driver's facial features through a convolutional neural network algorithm to generate the driver's facial feature data; the three-dimensional laser scanner obtains the three-dimensional size data of the cargo surface through laser beam ranging technology to generate cargo size feature data.

[0006] Preferably, the chemical sensor in the multimodal data processing module detects the molecular composition of the cargo material by gas phase molecular analysis, combines the material characteristic model matching algorithm stored in the database, generates material characteristic description data of the cargo, and inputs the data into the multimodal data processing module as part of the cargo characteristic data.

[0007] Preferably, the pressure sensor array in the distributed weighing and optimization module is installed in a grid arrangement in the vehicle's driving path, and the vehicle's weight distribution data is generated by real-time collection of pressure data at the vehicle tire contact points and combined with the vehicle's axle weight distribution model.

[0008] Preferably, the distributed weighing and optimization module utilizes the deep Q network structure in the reinforcement learning optimization algorithm, inputs vehicle weight distribution data, three-dimensional dimensional feature data of the cargo and vehicle specification information, and outputs a dynamic loading optimization model of the vehicle, wherein the training objectives of the reinforcement learning optimization algorithm include minimization of loading balance, center of gravity stability and loading time efficiency.

[0009] Preferably, the dynamic loading optimization model performs dynamic evolution simulation through cellular automaton rules, the cellular state includes the current cargo position, the vehicle center of gravity position and the cargo priority, the cellular state change rules include the impact of the newly added cargo position on the center of gravity of the adjacent cells, and the simulation results include loading task data and cargo priority ranking.

[0010] Preferably, the loading robot in the loading execution module has a six-axis robotic arm structure, combined with a path planning algorithm, to perform precise grabbing and placement operations of goods according to loading task data; the loading robot monitors the changes in the center of gravity of the vehicle in real time through an inertial sensor, and optimizes the loading path in combination with a dynamic interactive prediction algorithm, wherein the dynamic interactive prediction algorithm predicts the impact of loading order adjustments based on a recurrent neural network.

[0011] Preferably, the dynamic interactive prediction algorithm combines cargo size feature data and vehicle loading space model data, recalculates loading sequence and path optimization parameters after each cargo placement operation, and generates real-time adjusted loading task data.

[0012] Preferably, the dynamic detection device in the detection and verification module collects the actual load distribution data of the vehicle through a pressure sensor, and generates the actual center of gravity data and vibration mode data of the vehicle in combination with an inertial sensor and a vibration sensor. The comparison algorithm calculates the deviation range according to the vehicle design load model and loading status data, and generates a verification result based on the deviation range.

[0013] Preferably, the electronic shipping order includes vehicle information, cargo information, shipping time, verification results and deviation range evaluation indicators, and the electronic shipping order is synchronized to the transportation management system through the digital twin entity for subsequent status tracking of logistics operations.

[0014] Compared with the prior art, the advantages and beneficial effects of the present invention are: The present invention realizes load balancing optimization through dynamic center of gravity monitoring technology: using inertial sensors and dynamic center of gravity calculation methods, the center of gravity changes of the vehicle are monitored in real time to ensure the stability and safety of the vehicle during loading.

[0015] The present invention realizes real-time adjustment of the loading process through a dynamic interactive prediction algorithm: by combining cargo size feature data with loading space model data, the loading sequence and path parameters are dynamically optimized, thus significantly improving loading efficiency and adaptability.

[0016] The present invention realizes the status tracking of the whole logistics process through the digital twin entity: the electronic shipping order is synchronized with the digital twin entity to realize the dynamic monitoring of the status and abnormal warning during the transportation process.

[0017] The present invention achieves maximum utilization of the compartment space through space matching optimization technology: based on the precise matching of cargo characteristics and loading space, waste of loading space is avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a structural block diagram of the system of the present invention; Figure 2 Schematic diagram of the workflow of the multimodal data processing module in the present invention; Figure 3 A schematic diagram of a dynamic optimization workflow for loading an execution module in the present invention; Figure 4 It is a schematic diagram of the flow of the detection and verification module in the present invention. DETAILED DESCRIPTION

[0019] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0020] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "include", "comprising", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0021] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0022] like Figure 1 As shown, an unmanned shipping system based on intelligent weighing includes: like Figure 2 As shown, a multimodal data processing module is used to collect multimodal data including vehicle feature data, driver feature data and cargo feature data; perform quantum encoding on the multimodal data, map the multimodal data to a high-dimensional space using a kernel function method, extract core feature associations between data through high-dimensional association analysis, generate associated feature data, and generate digital twin entity data in combination with task information; The collection of multimodal data includes three types of data: vehicle feature data, driver feature data, and cargo feature data, which correspond to different collection equipment and technical paths: The vehicle characteristic data is transmitted to the vehicle surface by a spectral scanning device using a multi-band light source and receiving reflected spectral data. The spectral data is classified and processed using a spectral characteristic matching algorithm (such as an algorithm based on correlation analysis) to generate the vehicle's spectral characteristic data.

[0023] The driver’s facial features are collected by a facial recognition camera and classified by a convolutional neural network (CNN) model to extract the driver’s facial feature data. At the same time, the driver’s voice signal is collected by a voice sensor to extract its frequency characteristics as supplementary data.

[0024] The cargo characteristic data uses a 3D laser scanner to capture the 3D external dimensions of the cargo and generate dimensional characteristic data; the chemical sensor detects the volatile gas components on the surface of the cargo and generates chemical characteristic description data.

[0025] The collected multimodal data is quantum encoded, and the kernel function method is used to map the nonlinear data to a high-dimensional feature space to capture the implicit correlation in the data. Kernel function mapping method: Through the Gaussian kernel function, the input data Transformed into a mapping vector in feature space , and its calculation formula is:

[0026] in, is the input data point; represents the Euclidean distance; is the bandwidth parameter of the kernel function, which is used to control the amplitude attenuation of the Gaussian function. Through this mapping, the nonlinear characteristics of multimodal data can be transformed into a linear distribution in a high-dimensional space, which is convenient for subsequent analysis.

[0027] In high-dimensional space, the singular value decomposition (SVD) method is used to extract the core correlation features between multimodal data. The core steps of this method include: Constructing a multimodal data matrix , where each column represents a feature data. Perform singular value decomposition: ,in: and are the left singular matrix and the right singular matrix respectively; is a diagonal matrix, and the values ​​on the diagonal are singular values. Select the largest front singular values ​​and their corresponding singular vectors to generate associated feature data, which are used to represent the core association between data.

[0028] After extracting the associated feature data, it is combined with the mission information (including the transport mission number, cargo list, planned loading information, etc.) to generate digital twin entity data. The digital twin entity is a dynamically updated virtual model that reflects the status of vehicles, cargo, and missions in real time, providing global data support for subsequent modules.

[0029] Implementation example, example application: A logistics company uses this system for unmanned shipment. When a truck enters the factory area: the spectral scanning device captures the external spectral feature data of the vehicle (such as the type of paint and damage). The facial recognition camera confirms the identity of the driver, and the voice sensor verifies the driver's voice characteristics. The dimensions (length, width and height) of the cargo are measured by a three-dimensional laser scanner to be 2.5m×1.8m×1.2m, and the chemical sensor detects that the surface composition of the cargo is polypropylene. The collected multimodal data is quantum encoded, and the feature vector is generated using Gaussian kernel function mapping. High-dimensional correlation analysis: The correlation features between multimodal data are extracted by singular value decomposition. The correlation results show that the vehicle load capacity matches the cargo size by 0.95 (range 0-1, 1 is a perfect match). The weight of the driver's operating habits on the vibration of the cargo is 0.7. The above results are integrated into a digital twin entity, including the correlation information of the vehicle, driver and cargo, and synchronized to the optimization module in real time.

[0030] Preferably, the spectral scanning device in the multimodal data processing module transmits and receives spectral data reflected by the vehicle surface through a multi-band light source, and generates vehicle spectral feature data in combination with a preset spectral feature matching algorithm; the facial recognition camera classifies and extracts the driver's facial features through a convolutional neural network algorithm to generate the driver's facial feature data; the three-dimensional laser scanner obtains the three-dimensional size data of the cargo surface through laser beam ranging technology to generate cargo size feature data.

[0031] The multimodal data processing module is used in the present invention to realize the accurate collection and digital transformation of vehicle, driver and cargo characteristics to support the subsequent data processing and optimization process in the intelligent weighing system. The module consists of a spectral scanning device, a facial recognition camera and a three-dimensional laser scanner, which are respectively used for the characteristic data collection of vehicles, drivers and cargoes.

[0032] The working principle of the spectral scanning device is to emit a beam of light to the vehicle surface through a multi-band light source, and obtain the reflected spectral data through a receiving device. The reflected spectral data can reflect the optical properties of the vehicle surface material, such as coating color, texture and damage. In order to improve accuracy, the device combines the spectral feature matching algorithm to analyze and classify the collected spectral data.

[0033] Spectral feature matching algorithm, based on matching score calculation:

[0034] in, Indicates the matching score, which is used to judge the spectral similarity; Indicates the actual collected spectral reflectance; represents the reference reflectance of the preset spectral model; Indicates the number of spectral channels. Finally, the device outputs vehicle spectral feature data to provide input for subsequent load distribution analysis and vehicle status monitoring.

[0035] The facial recognition camera uses the convolutional neural network (CNN) algorithm to classify and extract the driver's facial features. The specific implementation process includes capturing the driver's facial image, extracting multi-level local features (such as texture information of the eyes, nose, mouth, etc.) through the convolution layer, and outputting feature vectors through the fully connected layer to generate the driver's facial feature data.

[0036] Using the grayscale value of the facial image as input, the feature map is calculated through layer-by-layer convolution operations, and finally the driver identity feature vector is generated in the classification layer. For example, the feature extraction output vector for the driver is:

[0037] in, The feature vector representing the driver; Represents the numerical values ​​of different feature dimensions. The output data is used to confirm the driver's identity and record driving behavior characteristics.

[0038] The 3D laser scanner captures the 3D dimension data of the cargo surface through laser beam ranging technology. The device measures the time difference from laser emission to return by rotating the laser transmitter and receiver, and calculates the spatial distance by combining the principle of triangulation. The laser ranging formula is:

[0039] in, Indicates the measured distance; represents the speed of light; Represents the time difference between laser emission and reception. Using the continuously rotating ranging results, the three-dimensional point cloud data of the goods can be reconstructed and the dimensional feature data of the goods can be generated to accurately describe the length, width, height and surface shape of the goods.

[0040] Accurately analyze the optical properties of the vehicle surface to detect the vehicle's coating status and surface damage, and avoid vehicle identification errors during weighing and shipping. Extract the driver's facial features through a convolutional neural network to ensure the accuracy and safety of the driver's identity. Generate three-dimensional dimensional feature data of the cargo through laser ranging technology to provide necessary input for loading planning and load balancing.

[0041] Embodiment: In a certain unmanned shipping system, a truck enters the factory area for multimodal data collection. The spectral scanning device emits a multi-band light source, receives the reflection spectrum of the vehicle surface, and determines after analysis that the coating material of the vehicle surface is metallic paint, and the surface damage area accounts for 2% of the total coating area. The facial recognition camera captures the driver's facial feature image, extracts the feature vector through the CNN algorithm, and compares it with the driver database in the system to confirm that the driver is "Zhang San". The three-dimensional laser scanner obtains the three-dimensional point cloud data of the goods, and after analysis, the length, width and height of the goods are generated as 2.5m, 1.8m and 1.2m respectively. There is no depression on the surface of the goods, which is a regular cuboid. Finally, the above data are summarized into the multimodal data processing module to generate multimodal feature data of the vehicle, driver and goods as the data input of the digital twin entity for use by the subsequent optimization module of the system.

[0042] Preferably, the chemical sensor in the multimodal data processing module detects the molecular composition of the cargo material by gas phase molecular analysis, combines the material characteristic model matching algorithm stored in the database, generates material characteristic description data of the cargo, and inputs the data into the multimodal data processing module as part of the cargo characteristic data.

[0043] The chemical sensor detects volatile gas molecules released from the surface of the cargo through gas phase molecular analysis. Specifically, the sensor absorbs volatile gas molecules into the detection chamber, converts the molecules into measurable ion signals through an ionization device, and then analyzes the mass-to-charge ratio (m / z) of the ions through a mass analyzer to generate characteristic data of the molecular composition.

[0044] Mass-to-charge ratio analysis: The detection formula is as follows:

[0045] in, represents the mass-to-charge ratio; represents the mass of the ion; Indicates ionic charge. This method allows rapid and accurate separation and identification of the main molecular components of cargo materials, such as polypropylene (PP) or polyethylene (PE) in plastics.

[0046] The molecular composition characteristic data obtained by the test needs to be further matched with the material characteristic model in the system database to generate cargo material characteristic description data. The algorithm calculates the similarity of the molecular characteristic vectors to screen out the most suitable material type. The characteristic vector matching formula is:

[0047] in, Represents the similarity score; Denote the detection feature vector and the database feature vector, respectively. Quantity; represents the dimension of the feature vector; Represents the cosine value of the angle between feature vectors. When the matching threshold set by the system is exceeded, the system marks the material property description data as a matching material type and inputs the matching result into the multimodal data processing module as part of the cargo characteristic data.

[0048] The detection results of chemical sensors are integrated with data from other modules (such as 3D laser scanners) to form complete cargo feature data. For example, combining the material properties of the cargo with the 3D size data can provide more accurate input for subsequent loading optimization models.

[0049] Example: In a logistics transportation mission, a batch of goods needs to be weighed and loaded through an unmanned shipping system. The goods contain a variety of materials with complex components and similar appearance. To ensure the accuracy of loading planning, the system enables the chemical sensor module. The chemical sensor collects volatile gas molecules from the surface of the goods and generates ion signals through an ionization device. The main ion characteristics detected by the mass analyzer are as follows: Mass-to-charge ratio Corresponding to isobutylene; Mass-to-charge ratio Corresponding to the polypropylene molecular fragment.

[0050] Based on the comprehensive analysis of molecular signals, the system confirmed that the main material of the cargo was polypropylene (PP).

[0051] The detection results are compared with the database model through the feature vector matching algorithm to obtain the similarity score , which is higher than the matching threshold of 0.9 set by the system. The system confirms that the material of the cargo is polypropylene (PP) and generates the characteristic description data of "polypropylene material". The generated material characteristic description data is input into the multimodal data processing module together with the cargo size characteristics (length 2.5m, width 1.8m, height 1.2m) measured by the 3D laser scanner, providing complete cargo characteristic data for the loading optimization module.

[0052] Distributed weighing and optimization module, which is used to calculate the weight distribution model through vehicle contact pressure data, and generate a dynamic loading optimization model using reinforcement learning optimization algorithm in combination with vehicle specification information and cargo characteristic data; When the vehicle enters the weighing area, the pressure sensor array on the ground collects the pressure data of the vehicle tire contact points in real time. The pressure sensors are arranged in a grid, and each node corresponds to a sampling point, which can capture the detailed characteristics of the vehicle weight distribution. Weight distribution model calculation: Based on the physical model of the vehicle axle weight distribution, the collected pressure data is converted into weight distribution:

[0053] in, Indicates the gross vehicle weight; Indicates The pressure value of each sensor node; Indicates The force bearing area corresponding to each node; Represents the total number of sensor nodes. Finally, the weight distribution model describes the load distribution of each axle of the vehicle and provides basic data for load balance analysis.

[0054] The cargo characteristic data (such as size and material properties) obtained by the multimodal data processing module is integrated with the vehicle specification information (such as maximum load, wheelbase and compartment size). The system simulates the cargo size and weight data, analyzes the impact of different cargo placement schemes on weight distribution in the compartment, and generates a preliminary load distribution strategy.

[0055] With the support of the above basic data, the system uses reinforcement learning algorithm to optimize loading strategy and generate dynamic loading optimization model. The main logic of reinforcement learning is to find the optimal loading sequence and position through repeated trials (iterations) to maximize the load balance and loading efficiency of the vehicle.

[0056] After the optimization is completed, the system generates a dynamic loading optimization model, which includes key information such as the order of cargo placement, specific location coordinates, and expected weight distribution. The model serves as the input of the loading execution module to guide the loading robot to complete subsequent operations.

[0057] Preferably, the pressure sensor array in the distributed weighing and optimization module is installed in a grid arrangement in the vehicle's driving path, and the vehicle's weight distribution data is generated by real-time collection of pressure data at the vehicle tire contact points and combined with the vehicle's axle weight distribution model.

[0058] The pressure sensor array is installed in a grid layout in the vehicle's driving path to ensure that the sensors can cover all tire contact areas of the vehicle. Each sensor node collects the distribution of the vehicle load on the sensor plane by measuring the pressure value of the vehicle tire at the contact point. The sensor nodes are arranged at a fixed spacing, and each node represents the pressure value of a grid unit. When the vehicle passes through the sensor array, each node records real-time pressure data. ,in These data are uploaded to the processing module in real time through the data acquisition unit.

[0059] Combining the collected pressure data with the vehicle's geometric characteristics (such as tire position and axle spacing), the system uses the vehicle axle weight distribution model to calculate the weight distribution of the vehicle on each axle. The contact pressure value of each tire has a linear relationship with the total load on the axle it is on, and the calculation formula is as follows:

[0060] in, Indicates Total weight of the axles; Indicates The pressure value of each sensor node; Representation and Node The corresponding effective contact area; Indicates the number of sensor nodes belonging to this axle. By aggregating the weight data of all axles, the system generates an overall weight distribution model of the vehicle.

[0061] The weight distribution data includes the load data of each axle and the center of gravity position of the vehicle. The system generates the following steps: axle load data, calculates the load weight of each axle, and determines whether the actual load of each axle exceeds the vehicle design specification. Center of gravity position calculation, combines the load data and the axle spacing to calculate the center of gravity position of the vehicle, and the output result is used to judge the balance of load distribution.

[0062] The sensor array can collect the vehicle's contact pressure data in real time, ensuring that the system can quickly reflect the vehicle's weight distribution. Through grid layout and comprehensive analysis of multi-node pressure data, the weight distribution model can accurately describe the load on each axle and the position of the vehicle's center of gravity. Through dynamic monitoring of the vehicle's weight distribution, overload or uneven load can be identified in a timely manner to ensure the safety of the vehicle in subsequent transportation. The weight distribution data provides important input for the system's loading optimization module, helping to generate efficient loading solutions.

[0063] In an implementation example, a logistics company weighs and performs load analysis on a truck through a distributed weighing and optimization module in its unmanned shipping system.

[0064] When the vehicle passes through the pressure sensor array, the sensor collects the pressure value of each tire contact point. Assume that the pressure data of the front wheel is and , the pressure data of the rear wheel is and The sensor array automatically transmits the pressure value to the data processing module.

[0065] The system combines the effective contact area of ​​the sensor nodes (assumed to be ) Calculate the shaft load:

[0066]

[0067] Calculate the total weight of the vehicle as .

[0068] The center of gravity of the vehicle is calculated based on the wheelbase and load data to confirm whether the load distribution of the vehicle is uniform.

[0069] Generate weight distribution data, including: front axle load of 500kg, accounting for 32.89% of the total load; rear axle load of 1020kg, accounting for 67.11% of the total load; center of gravity position is located at the rear 65% of the vehicle wheelbase.

[0070] Preferably, the distributed weighing and optimization module utilizes the deep Q network structure in the reinforcement learning optimization algorithm, inputs vehicle weight distribution data, three-dimensional dimensional feature data of the cargo and vehicle specification information, and outputs a dynamic loading optimization model of the vehicle, wherein the training objectives of the reinforcement learning optimization algorithm include minimization of loading balance, center of gravity stability and loading time efficiency.

[0071] Reinforcement learning is an adaptive optimization method based on trial and error mechanism. Deep Q network improves the optimization ability in high-dimensional state space by combining traditional Q learning with deep neural network. In the present invention: The state space includes the current weight distribution of the vehicle, the state of the cargo to be loaded (position, size, weight, etc.), and the distribution of available space in the compartment. Each loading operation includes selecting the specific location and order of cargo placement. Based on the three goals of loading balance, center of gravity stability, and loading time efficiency, the reward function is specifically defined as follows:

[0072] in, Indicates in status Take action The value of the reward received; Indicates the load balance score, which measures the balance of the load on the front and rear axles of the vehicle; It represents the center of gravity stability score, which measures the stability of the vehicle's load center of gravity shift; Indicates the loading time consumption, used to penalize inefficient operations; , , It is expressed as the weight coefficient of each target and adjusted dynamically according to actual needs.

[0073] The deep Q network processes input data through a convolutional neural network (CNN) structure: vehicle weight distribution data, as a two-dimensional matrix input, represents the current load status of each area in the car. The three-dimensional dimensional feature data of the cargo is provided by the multimodal data processing module and converted into a standardized dimensional feature vector after input into the CNN. Vehicle specification information, including the maximum load of the car, size restrictions, etc., is added to the optimization process as a constraint condition of the network.

[0074] In each round of decision making, the deep Q network calculates the Q values ​​of all possible actions based on the current state and selects the action with the highest Q value as the optimal decision. The dynamic loading optimization model contains the following core contents: the placement of each piece of cargo ( coordinates); loading sequence (dynamically adjusted based on cargo characteristics and spatial layout); expected load distribution and center of gravity position.

[0075] Using deep learning technology, the system can quickly generate efficient loading solutions in complex, multi-dimensional state spaces. By optimizing the loading position, the load deviation between the front and rear axles of the vehicle is significantly reduced to ensure transportation safety. The order and position of cargo placement are dynamically adjusted to prevent the vehicle from rolling over or losing control due to center of gravity shift. Based on the optimization of the reward function for loading time efficiency, the system can effectively reduce loading time and improve the overall efficiency of logistics.

[0076] Example: A logistics company needs to load a batch of goods into a truck. The goods have the following characteristics: Cargo A: weight 1,500kg, size 2m×1.5m×1.2m; Cargo B: weight 800kg, size 1.8m×1.2m×1.0m; Cargo C: weight 1,200kg, dimensions 1.6m×1.4m×1.1m.

[0077] Input data: vehicle weight distribution data, the front axle load is 3,000kg, the rear axle load is 6,500kg, and there is a serious imbalance problem; cargo size and weight characteristics, generated by the multimodal data processing module; vehicle specification information, the cabin size is 10m×2.5m×2.5m, and the maximum load is 10,000kg.

[0078] Optimization process: In the initial state, the vehicle's weight distribution is seriously biased towards the rear axle, and the center of gravity is too far back; in the first step of optimization, cargo A is placed at the front of the car, the front axle load is increased to 5,500kg, and the rear axle load is reduced to 5,000kg; in the second step of optimization, cargo B is placed in the middle left of the car to optimize the load balance between the front and rear axles and avoid the shift of the cargo's center of gravity; in the third step of optimization, cargo C is placed in the middle of the rear of the car to fill the space gap and improve the utilization rate of the car space.

[0079] Output results: The loading solutions generated by the dynamic loading optimization model include: Cargo A placement position: (1m, 1m, 0m); Cargo B placement position: (5m, 0.5m, 0m); Cargo C placement position: (8m, 1.25m, 0m).

[0080] After optimized load distribution, the front axle load is 5,500kg, the rear axle load is 4,500kg, and the center of gravity is located in the middle of the car, which significantly improves load balance and stability.

[0081] Preferably, the dynamic loading optimization model performs dynamic evolution simulation through cellular automaton rules, the cellular state includes the current cargo position, the vehicle center of gravity position and the cargo priority, the cellular state change rules include the impact of the newly added cargo position on the center of gravity of the adjacent cells, and the simulation results include loading task data and cargo priority ranking.

[0082] The dynamic loading optimization model uses cellular automaton (CA) rules to perform dynamic evolution simulation to optimize the cargo loading order and location. A cellular automaton is a discrete dynamic system consisting of a set of rule-driven cells whose state evolves over time. Based on the local interaction characteristics of cellular automata, the present invention simulates the impact of new cargo on the vehicle's center of gravity, spatial distribution, and priority sorting during cargo loading.

[0083] Cell state and spatial modeling include: Cell state definition, each cell represents a fixed position in the compartment space, and the cell state includes the following information: current cargo position, indicating whether the cell is occupied and the number of the occupied cargo; vehicle center of gravity position, the coordinates of the vehicle center of gravity calculated according to the position and weight of the cargo in the compartment; cargo priority, defining the priority of cargo in the loading task, which is used to guide the loading order. Spatial modeling, divide the compartment into several three-dimensional cellular units ( Coordinate system), each cell unit represents a fixed small space in the car. The state change of each cell depends on its own characteristics and the interaction with adjacent cells.

[0084] The cell state change rules are based on the impact of the new cargo on the center of gravity, space occupancy and priority ranking of adjacent cells during the loading process. The specific rules include: Center of gravity offset rule: the location of the newly added cargo will affect the overall center of gravity distribution of the vehicle. The system updates the contribution of each cell to the overall center of gravity offset in real time. The calculation formula is:

[0085] in, Indicates the center of gravity offset; Indicates The weight of the additional cargo; Indicates The distance from the newly added cargo to the current center of gravity of the vehicle; Indicates the current gross weight of the vehicle.

[0086] Neighboring influence rules: New cargo will occupy space, and its status update will affect the availability of adjacent cells. The system marks the available space by detecting the occupancy of adjacent cells.

[0087] Priority sorting rules, the priority of goods is determined by their volume, weight and the importance of the transportation task, and the priority sorting is dynamically adjusted as the cell state changes.

[0088] During the simulation, the system starts from the initial cellular state and gradually updates the three-dimensional cellular state of the carriage according to the cellular rules. The placement of new cargo will trigger the iteration of the cellular rules, and finally obtain the global loading plan. The output results include: loading task data, the loading position of each cargo ( Cargo priority sorting and optimized loading order are used to guide the loading robot to perform tasks.

[0089] The local rules and global evolution characteristics of cellular automata can respond to the state changes of newly added goods in real time and dynamically adjust the loading strategy. Through the center of gravity offset rule, the system can optimize the center of gravity distribution of the vehicle to ensure loading balance and transportation stability. The neighboring influence rule maximizes the use of compartment space and avoids space waste caused by irregular goods. The priority sorting rule ensures that important goods are loaded first, while reducing loading time and improving transportation efficiency.

[0090] Example: In a logistics task, the system needs to load three pieces of cargo into a carriage. The characteristics of the cargo are as follows: Cargo A: weight 1,000kg, size 1.5m×1.0m×0.8m, priority 3; Cargo B: weight 800kg, size 1.2m×1.0m×0.6m, priority 1; Cargo C: weight 600kg, dimensions 1.0m×0.8m×0.5m, priority 2.

[0091] Initial state: The car is divided into 10×5×3 cells (each cell is 1m³). The current center of gravity is located at the bottom center of the car ( ). All cell states are empty.

[0092] Simulation process: In the first step, cargo B is loaded first, and its volume occupies the cell The center of gravity offset is calculated as (Shift forward). The cell state is updated, and the available space of adjacent cells is reduced. Step 2: Cargo C is loaded in the middle of the carriage. Its placement is . The center of gravity offset is adjusted to (Return to the central axis). Step 3: Cargo A is loaded at the rear of the carriage. Its position is , ensuring that the remaining space is maximized. The center of gravity is ultimately balanced in the middle of the car ( ).

[0093] Output results: loading task data, placement coordinates and loading order of cargo B, C, and A. Cargo priority ranking, B>C>A (optimized and generated based on cargo characteristics and importance).

[0094] The loading execution module includes a loading robot, which is used to generate loading task data according to the dynamic loading optimization model data and perform cargo loading operations; monitor the changes in the center of gravity of the vehicle through inertial sensors, capture the changes in the load during cargo loading through piezoelectric sensors, and adjust the loading sequence and path in real time in combination with the dynamic interactive prediction algorithm to optimize the loading process; the dynamic center of gravity distribution data generated during the loading process and the cargo placement status data together constitute the loading status data; The loading robot generates loading task data through dynamic loading optimization model data, including the order of placing goods, position coordinates and loading path. The loading robot has a multi-axis mechanical arm structure and a flexible clamping device, which can accurately grasp and place goods. The optimal transportation trajectory of the goods from the initial position to the final placement position is determined by the path planning algorithm.

[0095] The inertial sensor measures the acceleration change of the vehicle and calculates the center of gravity of the vehicle in real time. The monitoring data of the inertial sensor is combined with the vehicle geometry model to update the dynamic center of gravity distribution data of the vehicle to determine the load balance.

[0096] The piezoelectric sensor is installed at the contact point between the robot arm and the cargo to capture the load changes during cargo loading, especially to detect whether the weight of the cargo is consistent with the expected data and whether the cargo is placed correctly. The output of the piezoelectric sensor is directly used as the real-time adjustment input of the dynamic interaction prediction algorithm.

[0097] The dynamic interactive prediction algorithm is based on a recurrent neural network (RNN). It uses multi-step time series data to predict the impact of each step in the loading process and adjust the loading strategy in real time. The input data includes dynamic center of gravity distribution data, cargo placement status data, and loading path data. The output results include the current optimal placement of the cargo, the next loading path, and sequence adjustment suggestions.

[0098] During the entire loading process, the system generates dynamic center of gravity distribution data and cargo placement status data through inertial sensors and piezoelectric sensors respectively. Dynamic center of gravity distribution data reflects the current load balance and center of gravity position of the vehicle. Cargo placement status data includes the specific placement position, angle and fixation of each piece of cargo. These two types of data together constitute the loading status data and are updated to the system in real time to provide support for subsequent detection and verification modules.

[0099] Preferably, Figure 3 As shown, the loading robot in the loading execution module has a six-axis robotic arm structure, which is combined with a path planning algorithm to perform precise grabbing and placement operations of goods according to loading task data; the loading robot monitors the center of gravity changes of the vehicle in real time through inertial sensors, and optimizes the loading path in combination with a dynamic interactive prediction algorithm, wherein the dynamic interactive prediction algorithm predicts the impact of loading sequence adjustment based on a recurrent neural network.

[0100] The loading robot in the loading execution module adopts a six-axis mechanical arm structure, combined with a path planning algorithm and a dynamic interactive prediction algorithm, to achieve accurate grabbing, placement and path optimization of goods in the unmanned shipping system. The core is to monitor the changes in the center of gravity of the vehicle in real time through inertial sensors, and use recurrent neural networks (RNN) to predict the impact of loading path adjustments on the center of gravity distribution and task efficiency, thereby achieving dynamic optimization. The loading robot is equipped with a six-degree-of-freedom mechanical arm, which enables it to operate flexibly in space and adapt to the complex changes of different cargo sizes and loading spaces.

[0101] Based on the cargo position and target position generated by the loading task data, the movement trajectory of the robot arm is calculated through path optimization to ensure efficient handling of the cargo. Optimization formula:

[0102] in, Indicates the total handling time; Indicates the path Distance to each transport point; Indicates the penalty value for angle changes on the path; Represents the angle weight factor. The robot dynamically adjusts the path during execution to avoid obstacles and ensure the optimal angle change to improve loading efficiency.

[0103] The inertial sensor measures the acceleration and tilt changes of the vehicle during loading, and the real-time position of the vehicle's center of gravity is calculated in combination with the vehicle's geometric model. The center of gravity calculation formula is:

[0104] in, Indicates the center of gravity of the vehicle; Indicates The weight of the cargo; Indicates The location of the cargo; Indicates the current gross weight of the vehicle. The system dynamically evaluates load balance through center of gravity change data to ensure that the center of gravity of the vehicle is always maintained within the designed safety range during loading.

[0105] The dynamic interactive prediction algorithm is based on a recurrent neural network (RNN) and uses time series prediction to analyze the impact of loading order adjustment on the loading path and vehicle center of gravity. The input includes real-time dynamic center of gravity data, cargo placement status data, and path planning results. The output is the optimized loading path and adjusted loading order. Core prediction formula:

[0106] in, Indicates the loading path after optimization in the next step; Indicates the current path prediction result; Represents the input data of the current time step, including the center of gravity change data of the inertial sensor, cargo status data, etc. The dynamic interactive prediction algorithm is characterized by being able to respond to changes in external input in real time, optimize the overall path planning when the cargo loading sequence is adjusted, and ensure vehicle load balance and improved loading efficiency.

[0107] Dynamic center of gravity distribution data reflects the center of gravity distribution of the vehicle under the current loading progress and is used to evaluate load balance. Cargo placement status data records the placement position, angle and fixed status of each piece of cargo. The two types of data together constitute the loading status data, which is updated to the system in real time to provide accurate input for subsequent processes.

[0108] The combination of the six-axis robotic arm structure and the path planning algorithm significantly improves cargo loading efficiency while reducing energy consumption during handling. The inertial sensor monitors the changes in the vehicle's center of gravity in real time, and dynamically adjusts the loading order through a dynamic interactive prediction algorithm to ensure that the vehicle maintains center of gravity stability during loading. The recurrent neural network predicts the impact of loading path adjustments, effectively avoiding confusion in cargo placement order or path conflicts during loading. Loading status data is generated in real time to provide accurate vehicle status information for subsequent detection and verification modules.

[0109] In this embodiment, an unmanned delivery system needs to load the following goods into a vehicle: Cargo A: weight 1,000kg, dimensions 1.8m×1.2m×1.0m; Cargo B: weight 800kg, size 1.5m×1.0m×0.8m; Cargo C: weight 1,200kg, dimensions 2.0m×1.5m×1.2m.

[0110] The task data generated by the dynamic loading optimization model includes: Cargo A is placed at the front of the carriage ( ); Cargo B is placed in the middle of the carriage ( ); The cargo C is placed at the rear of the carriage ( ).

[0111] The loading robot moves the goods to the designated location one by one according to the trajectory optimized by the path planning algorithm: the transportation path of cargo A is planned as a straight path to reduce the turning time; cargo B avoids the trajectory of loaded cargo A midway to optimize the transportation angle to save the path length; cargo C avoids obstacles at the rear of the car through dynamic path adjustment. During the transportation process, the inertial sensor detects that the center of gravity of the vehicle gradually shifts from the front axle to the middle, and the system dynamically adjusts the placement of cargoes B and C to ensure the stability of the center of gravity.

[0112] Dynamic center of gravity distribution data: the front axle load of the vehicle accounts for 40% of the total weight, the rear axle load accounts for 60%, and the center of gravity is located on the central axis of the vehicle; cargo placement status data includes the coordinates, placement angle and fixing status of each piece of cargo.

[0113] Preferably, the dynamic interactive prediction algorithm combines cargo size feature data and vehicle loading space model data, recalculates loading sequence and path optimization parameters after each cargo placement operation, and generates real-time adjusted loading task data.

[0114] The dynamic interactive prediction algorithm is an optimization method based on the cargo size feature data and vehicle loading space model data in the intelligent weighing unmanned shipping system. It aims to recalculate the loading sequence and path optimization parameters after each cargo placement operation and generate real-time adjusted loading task data. Through this dynamic adjustment mechanism, the system can adapt to changes in the actual loading process (such as space utilization, center of gravity offset) and optimize the overall loading plan.

[0115] The cargo size feature data includes the length, width, height and spatial shape information of the cargo, which comes from the multimodal data processing module. The vehicle loading space model data uses 3D modeling technology to divide the available loading space of the vehicle into regular grid units, and mark the available status of each unit (available, occupied, unavailable). The dynamic interactive prediction algorithm matches the cargo size feature data with the available units in the current loading space model and selects the optimal placement area. Matching calculation formula:

[0116] in, represents the matching score; Indicates The volume of available space units; Indicates the volume of the current cargo; Indicates the total number of available space units.

[0117] The dynamic interactive prediction algorithm is based on a recurrent neural network (RNN) and combines the size characteristics of the goods and the real-time loading status data generated after the placement operation to recalculate the loading sequence and path optimization parameters.

[0118] After each cargo placement, the algorithm updates the loading space model in real time, marking the current cargo occupancy status and the availability of adjacent areas. Through the time series prediction function, the impact of the current operation on the subsequent cargo loading path and sequence is analyzed. Optimization formula:

[0119] in, Indicates the loading task data after optimization in the next step; Indicates the current task status; Represents the current input data (including cargo size characteristics and loading space model data).

[0120] Loading order optimization dynamically adjusts the loading order according to the priority of the goods, the current placement and the remaining available space, giving priority to placing larger goods to maximize space utilization. Path optimization parameters recalculate the optimal parameters of the loading path to avoid unnecessary repeated movements and complex paths.

[0121] After each cargo placement operation, the system can dynamically adjust the subsequent loading sequence and path planning to ensure the optimization of the overall loading process. By updating the loading space model and center of gravity position in real time, the vehicle load is evenly distributed and the transportation risk is reduced. The precise matching of cargo size and loading space significantly improves the utilization efficiency of the compartment space and avoids the reduction of loading capacity due to space waste. The dynamic adjustment of path optimization parameters reduces invalid actions during loading and improves the overall operation efficiency.

[0122] Example: In a logistics loading task, the system needs to load the following goods into the carriage: Cargo A: weight 1,200kg, dimensions 1.8m×1.2m×1.0m; Cargo B: weight 800kg, size 1.5m×1.0m×0.8m; Cargo C: weight 1,000kg, dimensions 2.0m×1.5m×1.2m.

[0123] Initial loading task data generation, initial calculation results of dynamic loading optimization model: Cargo A is placed at the front of the carriage first, at ( ); Cargo B is placed in the middle left of the carriage at ( ); Cargo C is placed at the rear of the carriage at ( ).

[0124] Dynamic adjustment after cargo A is placed: After cargo A is placed, the system updates the loading space model, marking the area occupied by cargo A and the available status of adjacent units; the dynamic interactive prediction algorithm analyzes the current center of gravity offset and detects that the front axle load increases to 40%, the rear axle load decreases to 60%, and the center of gravity position is slightly forward. The system adjusts the loading order and adjusts the placement of cargo B to the back of the middle of the car to balance the center of gravity distribution.

[0125] Path optimization after placing cargo B: After placing cargo B, the path optimization parameters are recalculated, and the repeated moving parts contained in the original path are removed; the total path length is reduced by 20%.

[0126] Generate real-time adjusted loading task data: The final loading order is adjusted to: Cargo A>Cargo C>Cargo B. The loading path optimization results show that the total path length is reduced and the handling time is shortened by 15%.

[0127] like Figure 4 As shown, the detection and verification module is used to collect the vehicle's detection data after loading is completed, generate a verification result based on the comparison of the detection data with the loading status data, and generate an electronic shipping order when the verification passes.

[0128] Preferably, the dynamic detection device in the detection and verification module collects the actual load distribution data of the vehicle through a pressure sensor, and generates the actual center of gravity data and vibration mode data of the vehicle in combination with an inertial sensor and a vibration sensor. The comparison algorithm calculates the deviation range according to the vehicle design load model and loading status data, and generates a verification result based on the deviation range.

[0129] The dynamic detection device consists of a pressure sensor, an inertial sensor and a vibration sensor. Each sensor works together to collect real-time data of the vehicle. The pressure sensor collects pressure data at the contact point of the vehicle tire to calculate the actual load distribution data of the vehicle. Load distribution calculation formula:

[0130] in, Indicates the current gross weight of the vehicle; Indicates The pressure value collected by a pressure sensor; Indicates The effective contact area of ​​each pressure node; Indicates the total number of pressure sensor nodes.

[0131] The inertial sensor monitors the acceleration changes of the vehicle in real time and calculates the actual center of gravity data in combination with the vehicle geometry model. The vibration sensor collects the vibration frequency and amplitude of the vehicle during driving and generates the vehicle's vibration mode data to determine the fixity and stability of the cargo.

[0132] The deviation comparison algorithm calculates the deviation range between the actual state of the vehicle and the ideal state based on the vehicle design load model and loading status data. Input data: actual load distribution data, actual center of gravity data and vibration mode data; vehicle design load model, including the vehicle's maximum load limit, front and rear axle load balancing parameters and cargo fixing requirements; loading status data, including cargo placement, load distribution and the vehicle's current center of gravity position. Deviation calculation formula:

[0133] in, Indicates the center of gravity deviation; Indicates the current center of gravity position of the vehicle; Indicates the ideal center of gravity position in the design load model. Deviation range generation, set the deviation threshold according to the vehicle design requirements (such as the center of gravity deviation does not exceed 10%), when the calculation result exceeds the threshold, generate the verification result.

[0134] Based on the calculation results of the deviation comparison algorithm, a verification report is generated, including the load distribution deviation results, whether it meets the balance requirements of the vehicle design load model. The center of gravity deviation results, whether the current center of gravity meets the deviation limit of the design center of gravity. The vibration mode deviation results: whether the vibration is within the safe range.

[0135] The actual load, center of gravity and vibration status of the vehicle can be fully monitored before transportation to ensure that the load status of each vehicle meets safety requirements. Deviation comparison is performed in combination with the loading status data, and adjustment suggestions are generated in a timely manner when problems are found to optimize the loading results. The fixity of the goods is judged through vibration mode data to avoid safety accidents caused by loose goods or center of gravity shift. The verification results generated by the deviation range provide high-precision basic data support for the subsequent generation of electronic shipping orders.

[0136] In the implementation example, a logistics company detects a loaded truck in the intelligent weighing and unmanned shipping system. The cargo includes: Cargo A: weight 1,200kg, location (1m, 1m, 0m); Cargo B: weight 800kg, location (5m, 1m, 0m); Cargo C: weight 1,000kg, location (9m, 1.5m, 0m).

[0137] Dynamic detection data collection: The pressure sensor collects the vehicle's front axle load of 5,000kg, the rear axle load of 6,000kg, and the total weight of 11,000kg; the inertial sensor calculates the current center of gravity position to be in the middle of the car ( ); The vibration sensor detected a vibration frequency of 8 Hz and an amplitude of 0.02 m.

[0138] Deviation comparison calculation: Load distribution deviation, according to the vehicle design load model, the front and rear axle load ratio should be 45:55, the actual deviation is 5%, and the verification is passed. Center of gravity deviation, the ideal center of gravity position is the middle of the car ( ), the actual deviation is , the verification passed. Vibration mode deviation, the detected vibration frequency and amplitude are within the vehicle's allowable range (frequency <10Hz, amplitude <0.03m), the verification passed.

[0139] The verification results are generated, including: total weight: 11,000kg, in line with the vehicle design load model; front and rear axle load ratio: 45:55, deviation 5%, verification passed; center of gravity position: , deviation 8.3%, verification passed; vibration mode: safe, verification passed. The results show that the vehicle load is balanced, the center of gravity is stable, and the vibration is normal. An electronic shipping order is generated and the vehicle is allowed to be shipped.

[0140] Preferably, the electronic shipping order includes vehicle information, cargo information, shipping time, verification results and deviation range evaluation indicators, and the electronic shipping order is synchronized to the transportation management system through the digital twin entity for subsequent status tracking of logistics operations.

[0141] The data of the electronic shipping order comes from the output of the detection and verification module and the loading execution module, including: Vehicle information: basic data such as vehicle registration number, vehicle type and maximum load; Cargo information: cargo number, specifications (length, width, height), weight and material characteristics; Shipping time: the shipping confirmation time after the vehicle passes the inspection module verification; Verification results: whether the vehicle's loading status meets the design requirements, including indicators such as load distribution balance, center of gravity offset range, and vibration mode; Deviation range evaluation index: various deviation results calculated by the comparison algorithm, such as the front and rear axle load ratio deviation, center of gravity position deviation, and whether the vibration amplitude meets the safety range.

[0142] Data integration and document generation: The system compiles the above data into a standardized electronic document format (such as XML or JSON) through the data integration module to ensure the integrity and parsability of the document content.

[0143] The digital twin entity is a dynamic model of the vehicle and cargo in a virtual environment, reflecting its physical state and operation process in real time. The content of the electronic shipping order is synchronized to the digital twin entity through the edge computing node. The digital twin entity will update the relevant status, including: the current status of the vehicle (position, load distribution); cargo loading status (list of loaded cargo and its specific placement); verification results and deviation indicators.

[0144] The synchronized digital twin entity will be connected to the transportation management system through a network interface to achieve real-time tracking and sharing of logistics status.

[0145] Through electronic shipping orders and digital twin entities, logistics managers can monitor the transportation status of vehicles and goods in real time, tracking content including: the current location of the vehicle and the transportation progress; whether there are any abnormalities in the loading status (such as loose goods, shifted center of gravity); and whether the transportation plan needs to be dynamically adjusted (such as reallocating goods or changing the transportation route).

[0146] When the digital twin entity detects that the deviation exceeds the safe range (such as excessive center of gravity deviation), the system will automatically send an alarm message to the transportation management system and recommend adjustment plans.

[0147] The electronic shipping order integrates all-round information of vehicles, goods, verification results and deviation indicators to ensure the transparency of logistics task information. The synchronization function of the digital twin entity ensures real-time update of dynamic data during transportation and provides full traceability of logistics operations support. The deviation range evaluation indicator provides a safety guarantee for logistics tasks and can warn of potential risks in advance. By connecting with the transportation management system, managers can obtain the status of vehicles and goods at any time to improve logistics management efficiency.

[0148] In the example, a logistics company uses an intelligent weighing unmanned shipping system to transport a batch of goods for customers. The electronic shipping order generation and synchronization process of the shipping task is as follows: Electronic delivery order data integration: Vehicle information: Vehicle number is "L001", vehicle type is "12T flatbed truck", and maximum load is 12,000kg; Cargo information: Cargo A (weight 1,000kg, dimensions 1.5m×1.2m×1.0m, material characteristics: "polyethylene"), Cargo B (weight 800kg, dimensions 1.2m×1.0m×0.8m, material characteristics: "metal"); Shipping time: 10:30 on January 15, 2024; Verification results: Load distribution balance: the load ratio of the front and rear axles is 45:55, with a deviation of 5%, which meets the design requirements; Center of gravity deviation range: deviation 8%, within the safety range; Vibration mode: frequency is 7Hz, amplitude is 0.015m, which is within the safety range; Deviation range evaluation indicators: All deviation indicators are within the set threshold range and the verification passes.

[0149] Electronic delivery order generation and synchronization: The system integrates the above data to generate an electronic delivery order and synchronizes it to the digital twin entity. The digital twin entity is updated as follows: Current load: Total weight 1,800kg, front axle load 800kg, rear axle load 1,000kg; Cargo status: Cargo A and Cargo B have been loaded and placed at the front and middle of the carriage respectively; Verification result: All indicators passed.

[0150] During the transportation process, logistics managers monitor the vehicle status in real time through the transportation management system. The digital twin entity shows that the vehicle has reached the entrance of the highway and is currently in normal condition. When the vibration amplitude of the carriage is detected to increase to 0.03m (exceeding the safe range) during transportation, the system automatically alarms and recommends slowing down to prevent the goods from loosening.

[0151] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0152] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. An unmanned shipping system based on intelligent weighing, characterized in that: include: A multimodal data processing module, used for collecting multimodal data including vehicle characteristic data, driver characteristic data and cargo characteristic data; Quantum encoding is performed on the multimodal data, the multimodal data is mapped to a high-dimensional space using a kernel function method, and the core feature associations between the data are extracted through high-dimensional association analysis to generate associated feature data, and digital twin entity data is generated in combination with task information; Distributed weighing and optimization module, which is used to calculate the weight distribution model through vehicle contact pressure data, and generate a dynamic loading optimization model using reinforcement learning optimization algorithm in combination with vehicle specification information and cargo characteristic data; The loading execution module includes a loading robot, which is used to generate loading task data according to the dynamic loading optimization model data and perform cargo loading operations; monitor the changes in the center of gravity of the vehicle through inertial sensors, capture the changes in the load during cargo loading through piezoelectric sensors, and adjust the loading sequence and path in real time in combination with the dynamic interactive prediction algorithm to optimize the loading process; the dynamic center of gravity distribution data generated during the loading process and the cargo placement status data together constitute the loading status data; The detection and verification module is used to collect the vehicle's detection data after loading is completed, generate verification results based on the comparison of the detection data with the loading status data, and generate an electronic shipping order when the verification passes.

2. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The spectrum scanning device in the multimodal data processing module generates vehicle spectrum feature data by transmitting and receiving spectrum data reflected from the vehicle surface through a multi-band light source and combining with a preset spectrum feature matching algorithm; The facial recognition camera classifies and extracts the driver's facial features through a convolutional neural network algorithm to generate driver facial feature data; the three-dimensional laser scanner obtains three-dimensional size data of the cargo surface through laser beam ranging technology to generate cargo size feature data.

3. The unmanned shipping system based on intelligent weighing according to claim 2 is characterized in that: The chemical sensor in the multimodal data processing module detects the molecular composition of the cargo material by gas phase molecular analysis, combines the material characteristic model matching algorithm stored in the database, generates material characteristic description data of the cargo, and inputs the data into the multimodal data processing module as part of the cargo characteristic data.

4. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The pressure sensor array in the distributed weighing and optimization module is installed in a grid-like arrangement in the vehicle's driving path, and the vehicle's weight distribution data is generated by collecting pressure data of the vehicle's tire contact points in real time and combining it with the vehicle's axle weight distribution model.

5. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The distributed weighing and optimization module uses the deep Q network structure in the reinforcement learning optimization algorithm, inputs vehicle weight distribution data, three-dimensional size feature data of the cargo and vehicle specification information, and outputs a dynamic loading optimization model for the vehicle, wherein the training objectives of the reinforcement learning optimization algorithm include minimization of loading balance, center of gravity stability and loading time efficiency.

6. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The dynamic loading optimization model is dynamically evolved through cellular automation rules. The cellular state includes the current cargo position, the vehicle center of gravity position and the cargo priority. The cellular state change rules include the impact of the newly added cargo position on the center of gravity of the adjacent cells. The simulation results include loading task data and cargo priority ranking.

7. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The loading robot in the loading execution module has a six-axis robotic arm structure, combined with a path planning algorithm, to perform precise grabbing and placement operations of goods according to loading task data; the loading robot monitors the changes in the center of gravity of the vehicle in real time through an inertial sensor, and optimizes the loading path in combination with a dynamic interactive prediction algorithm, wherein the dynamic interactive prediction algorithm predicts the impact of loading order adjustments based on a recurrent neural network.

8. The unmanned shipping system based on intelligent weighing according to claim 7 is characterized in that: The dynamic interactive prediction algorithm combines cargo size feature data and vehicle loading space model data to recalculate loading sequence and path optimization parameters after each cargo placement operation and generate real-time adjusted loading task data.

9. The unmanned shipping system based on intelligent weighing according to claim 1 is characterized in that: The dynamic detection device in the detection and verification module collects the actual load distribution data of the vehicle through a pressure sensor, and generates the actual center of gravity data and vibration mode data of the vehicle in combination with an inertial sensor and a vibration sensor. The comparison algorithm calculates the deviation range according to the vehicle design load model and loading status data, and generates a verification result based on the deviation range.

10. The unmanned shipping system based on intelligent weighing according to claim 1, characterized in that: The electronic shipping order includes vehicle information, cargo information, shipping time, verification results and deviation range evaluation indicators. The electronic shipping order is synchronized to the transportation management system through the digital twin entity for subsequent status tracking of logistics operations.

Images