Multimodal transport one-box system intelligent credible collaboration method based on artificial intelligence technology

Abstract

The invention relates to the technical field of multimodal transport, in particular to a multimodal transport one-box system intelligent credible collaboration method based on an artificial intelligence technology, comprising the following steps: S1, obtaining a standardized JSON data packet from transport data according to an agreed rule, performing triple encryption to obtain an encrypted data packet, and broadcasting the encrypted data packet to a block chain; s2, the verification node carries out consensus verification on the Internet of Things equipment data and the electronic waybill key field, and if the verification is passed, the verification is written into the block chain; s3, the digital twin system fuses the multi-source data and extracts calculation data, and artificial intelligence calculates an optimal loading and unloading sequence and a transportation path according to the calculation data; and S4, uploading the longitude and latitude data to the block chain node through the Beidou positioning terminal and the 5G / LoRa gateway along the way, and automatically identifying the abnormity in the transportation stage based on the geofence technology. According to the method, data rules are unified, efficient connection of cross-transportation modes is achieved, manual input errors or violation operation is prevented, the risk of goods damage is reduced through dynamic route planning, the transfer efficiency is improved, and logistics time and cost consumption are reduced.

Description

Technical Field

[0001] The present invention relates to the field of multimodal transport technology, and in particular to an intelligent and trustworthy collaborative method, system, device and storage medium for multimodal transport one-box system based on artificial intelligence technology. Background Art

[0002] Intermodal transport is a comprehensive transportation strategy that integrates multiple modes of transportation. By combining road, rail, air and sea transportation, it provides efficient and flexible door-to-door services, optimizes transportation efficiency, reduces transit time, and reduces costs, while promoting green logistics and reducing energy consumption and environmental pollution.

[0003] In the existing multimodal container transportation process, the cargo information of shippers, logistics companies and different transport entities is not shared and connected. The goods need to be transferred multiple times between different modes of transportation (such as sea, rail, and road). The multiple loading and unloading of goods increases the risk of damage to the goods, labor costs and time costs. Since the goods are not uniformly transported in containers, they often need to be repackaged or reinforced when switching between different modes of transportation, making it difficult to achieve standardization and scale of transportation, thus affecting transportation efficiency and service quality. This not only increases the complexity of operations, but also increases costs. In addition, some logistics hub facilities are not well-equipped, and dedicated multimodal The lack of intermodal transport terminals has led to inefficient container loading and unloading, frequent container overturning, and lagging cross-regional channel construction. In particular, the connection between inland areas and coastal ports is insufficient, which limits the coverage of the "one-box system"; in addition, the level of intelligence of logistics hubs is not high, resource scheduling relies too much on manual experience, and the allocation of cranes and trucks is highly subjective. When responding to sudden weather and equipment failures, the dynamic adjustment capability is insufficient, making it difficult to quickly optimize the scheduling plan. The fault tolerance rate is low, and there is a lack of an overall risk prediction mechanism, which leads to poor connection between transportation links and rising costs, restricting the efficient operation of the "multimodal transport one-box system" throughout the entire process. Summary of the Invention

[0004] The purpose of the present invention is to address the shortcomings of the prior art and provide an intelligent and reliable collaborative method for multimodal transport in one container based on artificial intelligence technology, comprising the following steps: S1: The transport data is converted into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcasted to the blockchain; S2: IoT device data and key fields of the electronic waybill, including bill of lading number, cargo category, and consignee and consignor, are collected in real time. Validation nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. S3: The digital twin system integrates multi-source data, including physical equipment data, business system data, and external environment data. It extracts and calculates data based on this multi-source data. Artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on this data, generates operation instructions, and sends them to on-site automation equipment. S4: Through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way, longitude and latitude data are uploaded to the blockchain node every minute, and anomalies in the transportation stage are automatically identified based on geo-fencing technology. When the physical device data exceeds the smart contract threshold, an alarm is pushed to the shipper and insurance company.

[0005] Preferably, in step S1, the transport data is converted into a standardized JSON format according to agreed rules to obtain a standardized JSON data packet, further comprising: Inputting the transport data including structured data and unstructured data through a data conversion gateway, wherein the structured data includes formats such as EDI messages, XML transport documents, and CSV, and the unstructured data includes scanned copies of paper documents and key fields extracted by the ORC engine and the NLP model; The transport data is output as a standardized JSON data packet according to the agreed rules, wherein the agreed rules include defining a container structured data model and spatiotemporal data rules. The container structured data model includes a unique identifier, physical parameters including size, load and box type, and safety inspection status including the temperature control range of the refrigerated container, the dangerous goods UN number and inspection status. The spatiotemporal data rule is that the timestamp is forced to use the ISO 8601 extended format for device-side encryption, and the location data integrates the WGS84 coordinate system and the UN / LOCODE port code.

[0006] Preferably, the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcasted to the blockchain, further comprising: S11: Using device-side AES-256 to symmetrically encrypt the original data of the standardized JSON data packet to obtain symmetrically encrypted data; S12: Dynamically generate a session key through the TLS protocol, and encrypt the symmetrically encrypted data according to the session key to obtain asymmetric encryption; S13: The asymmetric key is re-encrypted using the cloud-based AES-256 key. The re-encrypted data is hashed using SHA256 to generate an encrypted data packet and broadcast to the blockchain.

[0007] Preferably, in step S2, the verification node in the blockchain network performs consensus verification on the IoT device data and the key fields of the electronic waybill according to the encrypted data packet, further comprising: The verification nodes in the blockchain network conduct consensus verification on the weight data obtained by the smart weighing scale and the declared weight in the electronic waybill, the container number and ISO check code extracted by the container number recognition camera, and the temperature and humidity and cold chain protocol. The verification nodes include nodes of the current port, the next carrier, and the customs regulator. If more than half of the verification nodes confirm that the data is authentic and valid, the IoT device data and the key fields of the electronic waybill will be written into a new block to form an unchangeable electronic file. If any verification node detects that a field is missing or the value range is out of bounds, the write will be refused.

[0008] Preferably, in step S3, the digital twin system integrates multi-source data including physical equipment data, business system data and external environment data, extracts calculation data based on the multi-source data, and the artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on the calculation data, further comprising: S31: The digital twin system integrates multi-source data including equipment sensor data, container location and barcode information identified by AI vision, TOS terminal loading and unloading task lists and transportation schedules, and real-time weather and traffic conditions, and extracts calculation data including equipment load rate, container urgency, and route congestion coefficient based on the multi-source data; S32: Multi-agent deep deterministic policy gradient MADDPG assigns container lifting priorities to the rail crane based on the calculated data and plans a dynamic path for the AGV unmanned vehicle.

[0009] Preferably, in step S32, the multi-agent deep deterministic policy gradient MADDPG assigns container lifting priorities to the rail crane according to the calculated data and plans a dynamic path for the AGV unmanned vehicle further includes: S321: Discretize the digital twin station into a two-dimensional grid map to obtain a digital twin station topology map. The grid attributes include static obstacles and dynamic restricted areas. Static obstacles include buildings and fixed equipment areas, and dynamic restricted areas include the operating radius of the rail crane and the temporary container storage area. S322 uses artificial intelligence technology to generate an initial shortest path based on the digital twin terminal topology map, and uses a spatiotemporal grid method to predict whether there will be path conflicts in several future time intervals. If there is a path conflict, the AGV with the high-urgency container is assigned path priority, triggering other AGVs to make local detours to achieve dynamic real-time adjustment; Based on the digital twin station topology map, artificial intelligence technology is used to generate the initial shortest path, including: The open list stores the starting point S (f=0), and the closed list is used to store the explored nodes. It is initially empty. The cost function g(n) is defined as the actual moving cost from the starting point to the current node n. The adjacent grid distance is 1, and the diagonal grid distance is , define the cost function h(n) as the estimated cost of the Manhattan distance from the current node n to the end point G, and define the cost function f(n) as f(n)=g(n)+h(n); The loop selects the node with the smallest cost function f(n) from the open list as the current node, checks the adjacent grids in the eight directions of the upper, lower, left, right and diagonal directions of the current node, and filters out invalid nodes. If the node is not in the open list, it is added to the open list and the parent node is recorded. If it is already in the open list, the parent node and the cost are updated when the new cost function g(n) is smaller. If the end point G is added to the closed list, the parent node is traced back to generate a path, or the open list is empty and no path is returned, so the loop is stopped. When the path crosses a dynamic restricted area, the equipment operation plan for the area in the future several time intervals in the digital twin system is queried. If there is a conflict in the equipment operation plan, the dynamic restricted area is converted into a static obstacle and the path is replanned.

[0010] Preferably, the fault prediction and processing of the device sensor by the edge computing gateway further includes: Preprocessing the vibration data of the crane vibration sensor, temperature sensor, and truck load sensor through the edge computing gateway to obtain preprocessed data, and inputting the preprocessed data into the random forest model; The random forest model extracts the low-frequency energy, mid-frequency energy, and high-frequency energy of the vibration data, the sliding window mean and change rate of the temperature data, and the standardized ratio of the current load to the rated load. Each decision tree independently determines its category, and the voting results of all trees are counted to calculate the failure probability. If the failure probability is greater than or equal to 0.9, an emergency shutdown is triggered. If the failure probability is less than 0.9 and greater than or equal to 0.7, a high-risk alert is triggered; if it is less than 0.7, it is a low risk.

[0011] Based on the same concept, the present invention also provides an artificial intelligence-based multimodal transport one-box intelligent and reliable collaborative system, including: The data collaboration module converts the transport data into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, triple-encrypts the standardized JSON data packet to obtain an encrypted data packet, and broadcasts it to the blockchain; The verification module collects IoT device data and key fields of the electronic waybill, including the bill of lading number, cargo category, and consignee and consignor, in real time. The verification nodes in the blockchain network perform consensus verification on the IoT device data and the key fields of the electronic waybill. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. In the intelligent hub dynamic scheduling module, the digital twin system integrates multi-source data, including physical equipment data, business system data, and external environment data. It extracts and calculates data based on this multi-source data. Artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on this data, generates operation instructions, and sends them to on-site automation equipment. The tracking module uploads longitude and latitude data to the blockchain node every minute through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way. It automatically identifies anomalies in the transportation stage based on geo-fencing technology and pushes alarms to the shipper and insurance company when the physical device data exceeds the smart contract threshold.

[0012] Based on the same concept, the present invention also provides a computer device including a memory and one or more processors, wherein the memory stores computer code, and when the computer code is executed by the one or more processors, the one or more processors execute the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology as described in any one of the embodiments.

[0013] Based on the same concept, the present invention also provides a computer-readable storage medium, which stores computer code. When the computer code is executed, the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology as described in any of the embodiments is executed.

[0014] Compared with the prior art, the present invention has the following beneficial effects: (1) The transportation data of the present invention is converted into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, and the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcast to the blockchain, thereby breaking down data barriers, unifying operating rules, and achieving efficient connection across transportation modes.

[0015] (2) The present invention collects IoT device data and key fields of the electronic waybill including bill of lading number, cargo category, consignee and consignor in real time. The verification nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill. If the verification is passed, the data is written into the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. The verification nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill, ensuring strong consistency between business flow and logistics, and preventing manual entry errors or illegal operations.

[0016] (3) The present invention integrates multi-source data including physical equipment data, business system data and external environment data through the digital twin system, extracts calculation data based on the multi-source data, and artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on the calculation data, generates operation instructions and sends them to on-site automation equipment, thereby realizing dynamic planning of railway branch lines and short-haul road routes to expand the coverage of the "one-box-to-the-end" service, ensuring that the goods are not overturned throughout the entire process, thereby reducing the risk of cargo damage and improving transportation efficiency, improving the overall efficiency of logistics operations, and reducing time and cost consumption in the logistics process.

[0017] (4) The present invention uses the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way to upload longitude and latitude data to the blockchain node every minute, and compares the planned path with the actual path. If it exceeds the set rules, the transport path re-planning mechanism is triggered to improve the overall collaborative efficiency of multimodal transport. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Various other advantages and benefits will become apparent to those skilled in the art by reading the following detailed description of the preferred embodiment.The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the invention.

[0019] Figure 1 This is a flow chart of the multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology of the present invention; Figure 2 This is a verification flow chart of the intelligent and trustworthy collaborative method for multimodal transport with one container based on artificial intelligence technology of the present invention; Figure 3 This is a schematic diagram of the architecture of the multimodal transport one-box intelligent and trusted collaborative system based on artificial intelligence technology of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Obviously, the embodiments described are part of the embodiments of this application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without making creative work are within the scope of protection of this application.

[0021] Those skilled in the art will understand that, unless otherwise specified, the singular forms "a," "an," and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0022] The technical terms involved in the present invention are defined as follows: Blockchain is a decentralized, distributed database that uses cryptographic algorithms to record data in chronological blocks, maintained by multiple nodes in the network. Its core characteristics include decentralization, immutability, transparency, strong security, high availability, and the promotion of cross-sector collaboration. These features enable blockchain to maintain data consistency in a trustless environment, reduce transaction costs, enhance system security, and enable peer-to-peer value transfer.

[0023] Trusted system technology is a comprehensive set of technologies designed to ensure that computer and network systems operate securely and reliably in the face of various threats. It encompasses hardware and software security, encryption, security protocols, auditing, vulnerability management, intrusion detection, and other aspects, aiming to provide comprehensive security. This technology emphasizes the system's self-protection and resilience, responding to evolving threats through continuous monitoring and policy adjustments. This builds trust between users and the system, ensuring the confidentiality, integrity, availability, and non-repudiation of information.

[0024] IoT technology monitoring equipment: It is the process of remotely monitoring, controlling and managing physical equipment through technologies such as sensors, network communications and data processing to collect status data, analyze performance and predict maintenance needs.

[0025] Artificial intelligence decision-making: It uses core technology to simulate the human decision-making process, and through steps such as data collection, model building and training, feature selection and verification testing, it can analyze complex data and make fast and accurate decisions, thereby improving decision-making efficiency and quality.

[0026] First embodiment There is a batch of auto parts (high value, fragile items) shipped to a certain country; Transport route: container ship → port → railway train → China-Europe express train; Emergency: The ship is delayed by 4 hours and the transfer must be completed within 2 hours, otherwise the only railway train of the day will be missed.

[0027] See also Figure 1 As shown, the trusted system for multimodal transport document data exchange based on blockchain technology provided in this embodiment includes: Shipping companies, railway departments, road transport companies, etc. will be added to the jointly maintained blockchain network to establish a transportation data sharing chain, with each participant serving as a node on the chain.

[0028] S1: The transportation data is converted into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, and the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcast to the blockchain.

[0029] Preferably, in step S1, the transport data is converted into a standardized JSON format according to agreed rules to obtain a standardized JSON data packet, further comprising: Transport data, including structured and unstructured data, is input through a data conversion gateway. Structured data includes formats such as EDI messages, XML transport documents, and CSV. Unstructured data includes scanned copies of paper documents and key fields extracted by the ORC engine and NLP model. Specifically, in this embodiment, invalid fields in the transport data (such as duplicate records and garbled characters) are removed, missing values ​​are filled (transport time is supplemented by the mean of historical data), and data types are unified for data cleaning. Fields extracted by ORC / NLP are compared and verified with EDI messages at the JSON level, and original documents such as scanned bills of lading are hashed and uploaded to the blockchain to prevent tampering. Transport data is output as a standardized JSON data packet according to agreed rules. These agreed rules include defining a container structured data model and spatiotemporal data rules. The container structured data model includes the ISO 6346 unique identifier, physical parameters including dimensions, load capacity, and container type, and safety inspection status including the temperature control range of the refrigerated container, the dangerous goods UN number, and inspection status. The spatiotemporal data rules mandate that timestamps use the ISO 8601 extended format (accurate to milliseconds + time zone) for device-side encryption. Location data integrates the WGS84 coordinate system with the UN / LOCODE port code. Specifically, in this embodiment, the UN number and temperature control range are written to the blockchain in real time. Abnormal conditions trigger smart contract alarms, and port / rail / road carriers automatically identify transit nodes using UN / LOCODE. ISO 8601 timestamp encryption ensures precise synchronization of operation timing across time zones, enabling AGVs to read container electronic tags using RFID tags and implement hierarchical decryption permission control based on the permission mapping table in Table 1.

[0030] Standardized JSON data packet example: { "metadata": { / / Metadata layer: describes data source and processing information "data_source": { / / Data source classification "structured": [ / / Structured data source list { "type": "EDI", / / Data type: EDI Electronic Data Interchange "version": "EDIFACT-D96A", / / EDI standard version "raw_data_hash": "sha256(...)" / / SHA-256 hash value of the raw data (used to ensure data integrity and blockchain verification) }, { "type": "XML", / / Data type: XML Extensible Markup Language "schema": "CIM / SMDG", / / XML schema definition used "raw_data_hash": "sha256(...)" / / hash value of raw XML } ], "unstructured": [ / / List of unstructured data sources { "type": "scanned_document", / / Data type: scanned document "format": "PDF", / / file format "ocr_engine": "Google Vision API", / / OCR engine used "nlp_model": "BERT-Logistics", / / NLP model used "extracted_fields": ["..."], / / extracted key field list "raw_file_hash": "sha256(...)" / / SHA-256 hash of the original file } ], "gateway_id": "GATEWAY-EU-001", / / Unique identifier of the data conversion gateway "processing_timestamp": "2025-01-01T15:30:45.123Z" / / Gateway processing time (ISO 8601 format, encrypted on the device) }}, "container_model": { / / Container core data model "identifier": { / / container identification information "container_number": "MSKU 601234", / / Container physical number (ISO 6346 standard) "bic_code": "MSKU 001234" / / BIC International Container Bureau registration code }, "physical_parameters": { / / Physical parameters, used to determine the physical characteristics of the transport selected "size": "40 feet", / / container size "type": "High Cube", / / container type "tare_weight_kg": 3800, / / Container weight (kg) "max_payload_kg": 28400, / / Maximum payload (kg) "volume_cbm": 67.7 / / Internal volume (m 3 ) }, "safety_status": { / / Safety inspection status "cargo_type": "Dangerous Goods", / / Cargo type "un_number": "UN3480", / / United Nations dangerous goods number "inspection": { / / Inspection certificate information "certificate_id": "SAFE-20240624-001", / / Inspection certificate ID "expiry_date": "2025-12-31", / / Certificate validity period "is_valid": true / / Current validity status }, "reefer_settings": { / / Reefer-specific settings "temperature_range": [-25, -18], / / Temperature control range [minimum, maximum] (℃) "humidity_percent": 65, / / Humidity setting (%) "power_status": "ON" / / Power status } } }, "spatiotemporal_data": { / / spatiotemporal data layer "timestamp": "2025-06-24T14:25:30.456Z", / / Event timestamp (ISO 8601 format, device-side encryption) "location": { / / Fusion of WGS84 coordinates and UN / LOCODE location information "gps": { / / GPS coordinates (WGS84 coordinate system) "latitude": 51.92442, / / latitude "longitude": 4.46773, / / longitude "coordinate_system": "WGS84" / / coordinate system identifier }, "unlocode": "NLRTM", / / UN / LOCODE port code (Port of Rotterdam) "facility": "ECT Delta Terminal" / / Specific facility name }, "movement_status": "RAIL_LOADED" / / Standardized transport status enumeration value (railway loaded) }, "compliance_checks": { / / Compliance check results "iso_8601_verified": true, / / Is the timestamp format compliant? "unlocode_mapped": true / / Whether the coordinates are consistent with UN / LOCODE }, "blockchain_link": { / / Blockchain related information "previous_event_hash": "0x89a2...c7d1", / / The blockchain hash value of the previous event, forming an unalterable event chain "data_package_hash": "sha256(...)", / / SHA-256 hash value of the current data package, core data fingerprint "smart_contract_address": "0x742d...b9e3" / / Associated smart contract address, triggering automated business logic } } Table 1: Permission mapping table

[0031] Preferably, the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcasted to the blockchain, further comprising: S11: Using device-side AES-256 to symmetrically encrypt the original data of the standardized JSON data packet to obtain symmetrically encrypted data. Specifically, in this embodiment, a 256-bit key is generated using a random number generator, and the original data such as temperature and humidity are symmetrically encrypted using AES-256 to obtain symmetrically encrypted data; S12: Dynamically generate a session key through the TLS1.3 protocol, and encrypt the symmetrically encrypted data using the session key to obtain asymmetrical encryption. Specifically, in this embodiment, the blockchain node generates a public key and a private key. The public key is made public, and the private key is securely stored by the node. The node public key is used to encrypt the symmetrically encrypted data to prevent network eavesdropping and tampering; S13: The asymmetric key is re-encrypted using the cloud-based AES-256 key. The re-encrypted data is hashed with SHA256 to generate an encrypted data packet and broadcast to the blockchain. Specifically, in this embodiment, the security and integrity of static data are guaranteed. Finally, the encrypted data and hash value are stored in the blockchain to achieve tamper-proof and privacy protection.

[0032] S2: Real-time collection of IoT device data and key fields of the electronic waybill, including bill of lading number, cargo category, consignee and consignor, etc. The verification nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. Specifically, in this embodiment, the IoT device actively reports data to the edge computing gateway at a preset frequency.

[0033] See also Figure 2 As shown, in step S2, the verification node in the blockchain network performs consensus verification on the IoT device data and the key fields of the electronic waybill according to the encrypted data packet, further comprising: The verification nodes in the blockchain network perform consensus verification on the weight data obtained through the smart weighing scale and the declared weight in the electronic waybill, the container number and ISO check code (OCR reads ISO6346 code) extracted by the box number recognition camera, temperature and humidity, and the cold chain protocol. Among them, the verification nodes include nodes including the current port, the next carrier, and the customs regulator. Specifically, in this embodiment, the verification nodes also include 2 other transportation enterprise nodes.

[0034] The specific consensus verification process is as follows: Currently, the port calculates the discrepancy between the weight data obtained by the intelligent weighbridge and the weight declared in the electronic waybill. If the discrepancy exceeds 10%, a red alert is issued and the data is rejected. This automatically triggers the retrieval of the weighbridge calibration record, requiring manual review of the container seal video and re-verification of the container number validity. If the discrepancy does not exceed 10%, the container is considered valid. The next stage is for the carrier to check whether the real-time temperature and humidity data are within the cold chain temperature control range preset in the electronic waybill. If they are outside the range, a graded warning is issued based on the duration of the violation. If it lasts no more than 30 minutes, a yellow warning is issued, the backup refrigeration unit is activated, and the shipper is notified to choose to continue transportation or change the container. If it lasts more than 30 minutes, a red warning is issued and the data is rejected. At the same time, the carrier compares the historical temperature and humidity data to confirm that there are no sudden increases or decreases (for example, fluctuations of no more than 2°C in the past hour). The customs supervisor will check whether the weight declared in the electronic waybill is consistent with the customs declaration. If not, the data will be rejected and the shipper will be required to provide a written explanation. If the goods are dangerous goods, the customs supervisor will check whether the container number safety inspection status contains the correct UN number. If not or the number is incorrect, the data will be rejected. It will also be confirmed whether the binding relationship between the container number and the container number field in the electronic waybill is correct. If not or the binding is incorrect, it will be regarded as invalid data.

[0035] If more than half of the verification nodes confirm that the data is authentic and valid, the IoT device data and the key fields of the electronic waybill will be written into a new block to form an unchangeable electronic file. If any verification node detects that a field is missing or the value range is out of bounds, it will refuse to write and generate a verification report. Specifically, in this embodiment, data solidification is achieved through the block structure, and the block header records basic blockchain information such as the version number, timestamp, hash of the previous and next blocks, and the data fingerprint tree root. The transaction body stores key verification results to achieve an evidence storage effect that the data cannot be tampered with and the history can be traced.

[0036] Through this full-process data mutual verification mechanism, the operating standards of different transportation sections are automatically aligned, fundamentally avoiding repeated loading and unloading or box replacement due to information misalignment.

[0037] S3: The digital twin system integrates multi-source data including physical equipment data, business system data and external environment data, extracts calculation data based on the multi-source data, and artificial intelligence calculates the optimal loading and unloading sequence and transportation route based on the calculation data, generates operation instructions and sends them to on-site automation equipment.

[0038] Preferably, in step S3, the digital twin system integrates multi-source data including physical equipment data, business system data and external environment data, extracts calculation data based on the multi-source data, and the artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on the calculation data, further comprising: S31: The digital twin system integrates multi-source data including equipment sensor data, container location and barcode information recognized by AI vision, TOS terminal loading and unloading task list and transportation schedule, real-time weather and traffic conditions, and extracts calculation data including equipment load rate, container urgency and path congestion coefficient based on multi-source data. Specifically, in this embodiment, the equipment sensors are used to obtain sensor data including the motor current of the rail crane, rail crane motor temperature (82°C), AGV battery power (38%), position encoder, AGV unmanned vehicle battery status, speed, etc., and AI vision is used to recognize the container location (ship bay C7-32), barcode (MSKU 778901), etc. Business system data includes loading and unloading task lists from the TOS terminal operating system, rail / road transport timetables (departure time remaining 110 minutes), and external environmental conditions such as gusts of level 7 (affecting lifting safety) and congestion in area D of the yard. After obtaining multi-source data, sensor noise is removed and missing values ​​are repaired for data cleaning. Multi-source data is aligned with spatiotemporal labels. For example, the visually recognized container location is matched with the AGV unmanned vehicle coordinates to complete data fusion. Extracting and calculating data from multiple sources includes: Device load rate = current task duration / device maximum processing capacity Container urgency = the reciprocal of the time until the transport vehicle departs Path congestion coefficient = real-time AGV traffic / road section capacity.

[0039] S32: Multi-agent Deep Deterministic Policy Gradient (MADDPG) assigns container lifting priorities to rail cranes based on calculated data and plans dynamic paths for AGVs.

[0040] A multi-agent deep deterministic policy gradient (MADDPG) is constructed using the rail-mounted crane as the first agent and the automated guided vehicle (AGV) as the second agent. Specifically, in this embodiment, the rail-mounted crane makes the following decisions: hoisting priority for container MSKU 778901: first (i.e., the first agent); operating parameters: maximum swing angle: ≤5° (seismic requirement); hoisting speed: 0.8m / s (normally 1.2m / s); target coordinates: [X:120, Y:350] (temporary storage area); AGV path planning: Path A (shortest path): distance 350m, but requires passing through a high wind speed area → high safety risk; Path B (detour charging): distance 480m, battery replenishment possible → increased time consumption; Path C (tunnel passage): distance 420m, windproof but requires permission → optimal solution; Define the state space: including equipment load, container priority, path congestion, energy consumption, etc. Define the action space: assign container lifting sequence, AGV travel path speed and steering instructions to the rail crane; Define the reward function: including rail crane reward items and AGV reward items. Among them, the positive reward of the rail crane reward item is the number of containers lifted per unit time × the urgency weight, and the negative reward of the rail crane reward item is the proportion of equipment idle time × the penalty coefficient. The positive reward of the AGV reward item is the efficiency value of reaching the destination by the shortest path, and the negative reward of the AGV reward item is the number of path conflicts × the urgency loss weight.

[0041] Rail crane priority calculation: Urgency = 0.9 (high value) × e^(-0.1×110) × 1.3 (fragile) = 0.87; AGV path parameters: congestion coefficient = 0.7 (static obstacles) + 0.2 (dynamic AGV) = 0.9; safety factor = 0.6 (30% speed reduction due to gusts).

[0042] MADDPG technology enables millisecond-level dynamic decision-making in complex environments, intelligent resolution of equipment collaboration and resource conflicts, and automatic binding of physical operations and blockchain contracts for seamless transition of multimodal transport status throughout the entire process.

[0043] Preferably, in step S32, the multi-agent deep deterministic policy gradient MADDPG assigns container lifting priorities to the rail crane according to the calculated data and plans a dynamic path for the AGV unmanned vehicle further includes: S321: Discretize the digital twin station into a two-dimensional grid map to obtain a digital twin station topology map. The grid attributes include static obstacles and dynamic restricted areas. Static obstacles include buildings and fixed equipment areas, while dynamic restricted areas include the operating radius of the rail crane and the temporary container storage area. Specifically, in this embodiment, the station ground is divided into 1m*1m grids for spatial discretization. Time discretization is completed by generating 30 time slices in the future at intervals of 1s. If two AGVs occupy the same grid within the same time slice, they are marked as a conflict. S22: Based on the digital twin terminal topology, artificial intelligence technology is used to generate an initial shortest path. A space-time grid method is used to predict whether there will be path conflicts within several future time intervals. If a path conflict exists, path priority is assigned to the AGV with the most urgent container, triggering other AGVs to make partial detours to achieve dynamic real-time adjustments. Specifically, in this embodiment, the final decision is: [Terminal] → [Underground Tunnel] → [Charging Station] → [Railway Yard], with an estimated time of 8 minutes and 45 seconds (including 45 seconds of charging). Based on the digital twin station topology map, artificial intelligence technology is used to generate the initial shortest path, including: The open list stores the starting point S (f=0), and the closed list is used to store the explored nodes. It is initially empty. The cost function g(n) is defined as the actual moving cost from the starting point to the current node n. The adjacent grid distance is 1, and the diagonal grid distance is , define the cost function h(n) as the estimated cost of the Manhattan distance from the current node n to the end point G, and define the cost function f(n) as f(n)=g(n)+h(n); The loop selects the node with the smallest cost function f(n) from the open list as the current node, checks the adjacent grids in the eight directions of the upper, lower, left, right and diagonal directions of the current node, and filters out invalid nodes. If the node is not in the open list, it is added to the open list and the parent node is recorded. If it is already in the open list, the parent node and the cost are updated when the new cost function g(n) is smaller. If the end point G is added to the closed list, the parent node is backtracked to generate a path, or the open list is empty and no path is returned, the loop is stopped; when the path crosses a dynamic restricted area, the equipment operation plan for the area in the future several time intervals in the digital twin system is queried. If there is a conflict in the equipment operation plan, the dynamic restricted area is converted into a static obstacle and the path is replanned. Specifically, in this embodiment, the overlapping areas of the space-time cubes of different AGVs are detected and marked as conflict points. When a conflict is predicted, right-of-way is reallocated based on the container's urgency (defined as the inverse of the time until the transport vehicle departs). The original path is retained for the AGV carrying the high-urgency container, and other AGVs are forced to perform a local detour. That is, with the conflict point as the center, the Dijkstra algorithm is used to search for an alternative sub-path within a radius of 5 meters. The sub-path satisfies the requirement that the distance deviation is less than or equal to 15% of the initial path length and there are no new spatiotemporal conflicts. The adjusted path instructions are then input into the action space of the MADDPG model for collaborative optimization.

[0044] When the rail crane and AGV need to work together, the waiting time at the container handover point is incorporated into the state space. Specifically, in this embodiment, the AGV verifies the digital identity of the container (matching MSKU 778901 on the blockchain) before departure. The hash value of the rail crane's lifting operation parameters is recorded: HASH (speed 0.8m / s, angle 5°). When the AGV arrives at the railway yard, it automatically scans the container barcode, updates the blockchain status to "port section completed, waiting for railway loading", and triggers the smart contract to release the freight for the ocean section. If the estimated arrival time of the AGV is later than the idle time window of the rail crane, the MADDPG model is triggered to adjust synchronously, reducing the hoisting priority of the rail crane and adding a path speed instruction for the associated AGV.

[0045] Build a BI dashboard based on on-chain tracking data, integrating multi-dimensional data such as location, temperature and humidity, and handover records. Visualize the entire container process through a map and generate a transportation efficiency report. Combined with historical data to train a prediction model, it provides decision support for optimized route planning and resource scheduling, improving the overall collaborative efficiency of multimodal transport.

[0046] Intelligent decision-making uses AI algorithms to analyze data from across the entire delivery chain and dynamically optimize scheduling strategies. Combining real-time weather, road conditions, and equipment status data, AI automatically adjusts loading and unloading priorities or diversion routes. It also predicts cargo flow and empty container demand, proactively planning return container allocations to reduce empty trips. Ultimately, this allows for "one-box delivery" without unpacking or changing containers, reducing manual intervention costs and the risk of cargo damage.

[0047] Preferably, the fault prediction and processing of the device sensor by the edge computing gateway further includes: The vibration data of the crane vibration sensor, temperature sensor, and truck load sensor are preprocessed through the edge computing gateway to obtain preprocessed data, and the preprocessed data is input into the random forest model. Specifically, in this embodiment, an operation monitoring module is installed on the crane, truck and other equipment, and the vibration sensor is installed on the bearing and gear box at the key monitoring point of the crane to collect high-frequency vibration signals. The temperature sensor is used to monitor the motor winding and bearing temperature. The stress sensor is used for the load-bearing status and metal fatigue of the hook. The load sensor is installed on the bottom plate of the cargo box at the truck monitoring point to detect the pressure distribution. The GPS / IMU is installed to monitor the real-time position, acceleration, and turning angle of the crane. Sensor data is transmitted to the edge gateway via 5G. The edge computing layer uses a moving average method to average 10 consecutive data points, smoothing random fluctuations for pre-processing and noise reduction (removing interference). Fixed thresholds such as vibration thresholds and temperature are then set and directly deleted or replaced with the average of the previous and next data points to filter outliers (eliminating erroneous data). Only two core indicators are retained from the data obtained by the vibration sensor: the mean vibration intensity, which reflects the overall vibration level of the equipment, and the maximum vibration value, which captures sudden abnormal impacts. After local pre-processing, the data is uploaded to the cloud analysis platform. Through multiple types of sensors and edge computing technology, real-time tracking and intelligent management of equipment operating status are achieved. The random forest model extracts the low-frequency energy, mid-frequency energy, and high-frequency energy of the vibration data, the sliding window mean and change rate of the temperature data, and the normalized current load to rated load ratio. Each decision tree independently determines its category, counts the voting results of all trees, and calculates the failure probability. If the failure probability is greater than or equal to 0.9, an emergency shutdown is triggered. If the failure probability is less than 0.9 and greater than or equal to 0.7, a high-risk reminder is triggered and maintenance is required within 6 hours; if it is less than 0.7, it is low risk and is included in the regular maintenance plan. Specifically, in this embodiment, the platform uses a machine learning model By integrating and analyzing historical and real-time data, it can provide early warning of crane bearing failures 6-12 hours in advance, automatically generate maintenance work orders, and send them to operations personnel. Truck loading plans are optimized based on load and route data to reduce empty driving rates and the risk of overloading. When the predicted probability of bearing failure is >0.7, the impact of bearing fracture on the overall structure is simulated in the twin. If the probability of failure is >0.9 and spare parts are available in stock, an emergency replacement work order is issued and the nearest maintenance team is dispatched. If the probability of failure is >0.7 and spare parts are not available in stock, a reduced load is recommended, and the spare parts procurement process is automatically triggered. If the container tilt sensor exceeds 15° for 5 seconds, the AGV is triggered to make an emergency stop. If the temperature of the refrigerated container suddenly rises by 2°C / min, the AGV automatically reroutes to the maintenance area.

[0048] By connecting to the artificial intelligence dispatching system, integrating real-time data such as crane load, truck location, and yard capacity, using reinforcement learning algorithms to dynamically adjust resource allocation, and using risk prediction models trained with historical data and real-time sensor information, it can provide early warning of equipment failures and cargo anomalies, and push emergency instructions to the operating terminal in real time, reducing the inefficiency and unreasonable equipment allocation that may occur when manually operating yard equipment, ensuring the smooth transit of container transportation, and significantly improving the connection efficiency during container transportation.

[0049] S4: Through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way, longitude and latitude data are uploaded to the blockchain node every minute, and anomalies in the transportation stage are automatically identified based on geo-fencing technology. When the physical device data exceeds the smart contract threshold, an alarm is pushed to the shipper and insurance company.

[0050] Smart contract automatic verification: AGV plans the path to generate a digital fingerprint HASH (path), the actual driving path points are encrypted and uploaded to the chain (every 10 seconds to store evidence), and the smart contract compares the planned / actual path deviation The "transportation route is a contract" is realized to avoid the risk of human tampering with the route. The real-time effect is shown in Table 2 below: Table 2: Implementation Effects

[0051] By deeply binding AGV path planning with blockchain container digital identity, cross-transportation timetables, and environmental risk data, we can achieve trusted collaboration between physical transportation flow and information control flow, supporting the full-process responsibility traceability of the "one-box system".

[0052] The container has a built-in Beidou positioning terminal, which, combined with 5G base stations and LORA IoT gateways along the way, uploads location data to the blockchain node every minute. It automatically identifies the transportation stage such as sea transportation, railway stations, and road transportation through geographic fence technology. The location data is updated to the chain after consensus verification. The shipper and the carrier can query the container location in real time through the system and compare the planned path with the actual situation. If the sea transportation stays for more than 72 hours along the polygon within 200 nautical miles of the port, a detention alarm will be sent. If the railway station platform coordinates are within a circle with a radius of 500 meters, if the detention is greater than the planned time + 4 hours, a detention warning will be sent. If the road transportation is in the buffer zone of the planned path ± If the cargo type is frozen seafood, the system checks whether the temperature and humidity are within the smart contract range. If the smart contract threshold is exceeded, an alert is sent to the cargo owner and insurance company.

[0053] This embodiment builds a decentralized data sharing platform based on blockchain technology, encrypts and records information such as container box specifications, loading and unloading records, temperature and humidity status, etc., to solve the data mutual trust and tampering risks across transport modes; relies on artificial intelligence algorithms to integrate railway, port, and highway capacity data and equipment status in real time, such as crane load, yard capacity, vehicle location and other information, dynamically optimizes resource scheduling plans, replaces traditional manual experience-based decision-making, and reduces equipment idle rates; uses the Internet of Things and intelligent container tracking technology to achieve full-process transparent monitoring, combines artificial intelligence risk prediction models to provide early warnings of abnormal events such as equipment failures and route congestion, and simultaneously generates emergency detour or diversion plans; based on the multimodal transport data center, unifies document rules and operating standards, connects the information systems of shippers, logistics companies and different transport entities, eliminates conflicts in rules such as box adaptation and document circulation, ensures seamless connection of the "one box to the end" process, reduces box unloading and unpacking operations, comprehensively improves logistics efficiency and reduces cargo damage rates.

[0054] Second embodiment See also Figure 3 As shown, based on the same concept, this embodiment provides a multimodal transport one-box intelligent trustworthy collaborative system based on artificial intelligence technology, including: The basic software and hardware include domestic trusted computing servers, Kunpeng servers, domestic trusted computing operating systems such as Kylin, application middleware, TongWeb, TongRDS, TongHttpServer, database, DAMO database, Beidou positioning terminal, and artificial intelligence technology.

[0055] Based on secure and trusted hardware and trusted middleware, it consists of a full-link data collaboration module, intelligent hub and automation connection, IoT equipment monitoring and full-process container tracking, and connects with data information from external organizations and related units.

[0056] The data collaboration module converts the transport data into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, triple-encrypts the standardized JSON data packet to obtain an encrypted data packet and broadcasts it to the blockchain; The verification module collects IoT device data and key fields of the electronic waybill, including bill of lading number, cargo category, and consignee and consignor, in real time. The verification nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. In the intelligent hub dynamic scheduling module, the digital twin system integrates multi-source data, including physical equipment data, business system data, and external environment data. It extracts and calculates data based on this multi-source data. Artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on this data, generates operation instructions, and sends them to on-site automation equipment. The tracking module uploads longitude and latitude data to the blockchain node every minute through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way. It automatically identifies anomalies in the transportation stage based on geo-fencing technology and pushes alarms to the shipper and insurance company when the physical device data exceeds the smart contract threshold.

[0057] Third embodiment In this embodiment, a computer device is provided, including a memory and one or more processors. Computer code is stored in the memory. When the computer code is executed by the one or more processors, the one or more processors execute the steps of the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology in the first embodiment.

[0058] In some embodiments of the present application, a computer-readable storage medium is also provided. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology as described in any one of the first embodiments.

[0059] It can be understood that, for the aforementioned multimodal transport one-box intelligent trusted collaboration method based on artificial intelligence technology, if it is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer server, or a network device, etc.) to execute all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0060] Computer-readable storage media may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0061] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An intelligent and trustworthy collaborative method for multimodal transport with one container based on artificial intelligence technology, characterized by: The following steps are involved: S1: The transport data is converted into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, the standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcasted to the blockchain; S2: IoT device data and key fields of the electronic waybill, including bill of lading number, cargo category, and consignee and consignor, are collected in real time. Verification nodes in the blockchain network perform consensus verification on the IoT device data and key fields of the electronic waybill based on the encrypted data packet. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. S3: The digital twin system integrates multi-source data, including physical equipment data, business system data, and external environment data. It extracts and calculates data based on this multi-source data. Artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on this data, generates operation instructions, and sends them to on-site automation equipment. S4: Through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way, longitude and latitude data are uploaded to the blockchain node every minute, and anomalies in the transportation stage are automatically identified based on geo-fencing technology. When the physical device data exceeds the smart contract threshold, an alarm is pushed to the shipper and insurance company.

2. The multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology according to claim 1 is characterized by: In step S1, the transport data is converted into a standardized JSON format according to agreed rules to obtain a standardized JSON data packet, further comprising: Inputting the transport data including structured data and unstructured data through a data conversion gateway, wherein the structured data includes formats such as EDI messages, XML transport documents, and CSV, and the unstructured data includes scanned copies of paper documents and key fields extracted by the ORC engine and the NLP model; The transport data is output as a standardized JSON data packet according to the agreed rules, wherein the agreed rules include defining a container structured data model and spatiotemporal data rules. The container structured data model includes a unique identifier, physical parameters including size, load and box type, and safety inspection status including the temperature control range of the refrigerated container, the dangerous goods UN number and inspection status. The spatiotemporal data rule is that the timestamp is forced to use the ISO 8601 extended format for device-side encryption, and the location data integrates the WGS84 coordinate system and the UN / LOCODE port code.

3. The multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology according to claim 1 is characterized by: The standardized JSON data packet is triple-encrypted to obtain an encrypted data packet and broadcasted to the blockchain, further comprising: S11: Using device-side AES-256 to symmetrically encrypt the original data of the standardized JSON data packet to obtain symmetrically encrypted data; S12: Dynamically generate a session key through the TLS protocol, and encrypt the symmetrically encrypted data according to the session key to obtain asymmetric encryption; S13: The asymmetric key is re-encrypted using the cloud-based AES-256 key. The re-encrypted data is hashed using SHA256 to generate an encrypted data packet and broadcast to the blockchain.

4. The multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology according to claim 1 is characterized by: In step S2, the verification node in the blockchain network performs consensus verification on the IoT device data and the key fields of the electronic waybill according to the encrypted data packet, further comprising: The verification nodes in the blockchain network conduct consensus verification on the weight data obtained by the smart weighing scale and the declared weight in the electronic waybill, the container number and ISO check code extracted by the container number recognition camera, and the temperature and humidity and cold chain protocols. The verification nodes include nodes of the current port, the next carrier, and the customs regulator. If more than half of the verification nodes confirm that the data is authentic and valid, the IoT device data and the key fields of the electronic waybill will be written into a new block to form an unchangeable electronic file. If any verification node detects that a field is missing or the value range is out of bounds, the write will be refused.

5. The intelligent and trustworthy collaborative method for multimodal transport with one container based on artificial intelligence technology according to claim 1 is characterized in that: In step S3, the digital twin system integrates multi-source data including physical equipment data, business system data, and external environment data, extracts calculation data based on the multi-source data, and the artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on the calculation data, further including: S31: The digital twin system integrates multi-source data including device sensor data, container location and barcode information identified by AI visual recognition, TOS terminal loading and unloading task lists and transportation schedules, and real-time weather and traffic conditions. Based on the multi-source data, the system extracts calculation data including device load rate, container urgency, and path congestion coefficient. The system then predicts and processes device sensor faults through an edge computing gateway. S32: Multi-agent deep deterministic policy gradient MADDPG assigns container lifting priorities to the rail crane based on the calculated data and plans a dynamic path for the AGV unmanned vehicle.

6. The multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology according to claim 5 is characterized by: In step S32, the multi-agent deep deterministic policy gradient MADDPG assigns container lifting priorities to the rail crane according to the calculated data and plans a dynamic path for the AGV unmanned vehicle further including: S321: Discretize the digital twin station into a two-dimensional grid map to obtain a digital twin station topology map. The grid attributes include static obstacles and dynamic restricted areas. Static obstacles include buildings and fixed equipment areas, and dynamic restricted areas include the operating radius of the rail crane and the temporary container storage area. S322: Based on the digital twin terminal topology map, artificial intelligence technology is used to generate an initial shortest path, and a space-time grid method is used to predict whether there will be path conflicts in several future time intervals. If there is a path conflict, the AGV with the high-urgency container is assigned path priority, triggering other AGVs to make local detours to achieve dynamic real-time adjustment; Based on the digital twin station topology map, artificial intelligence technology is used to generate the initial shortest path, including: The open list stores the starting point S (f=0), and the closed list is used to store the explored nodes. It is initially empty. The cost function g(n) is defined as the actual moving cost from the starting point to the current node n. The adjacent grid distance is 1, and the diagonal grid distance is , define the cost function h(n) as the estimated cost of the Manhattan distance from the current node n to the end point G, and define the cost function f(n) as f(n)=g(n)+h(n); The loop selects the node with the smallest cost function f(n) from the open list as the current node, checks the adjacent grids in the eight directions of the current node, and filters out invalid nodes. If the node is not in the open list, it is added to the open list and the parent node is recorded. If the node is already in the open list, the parent node and cost are updated when the new cost function g(n) is smaller. If the end point G is added to the closed list, the parent node is backtracked to generate a path, or the loop stops when the open list is empty and no path is returned. When the path crosses a dynamic restricted area, the digital twin system queries the equipment operation plan for the area within several future time intervals. If there is a conflict in the equipment operation plan, the dynamic restricted area is converted into a static obstacle and the path is replanned.

7. The multimodal transport one-box intelligent and trustworthy collaborative method based on artificial intelligence technology according to claim 5 is characterized by: Fault prediction and processing of device sensors through edge computing gateways further includes: Preprocessing the crane vibration sensor, temperature sensor, truck load sensor data, and vibration data through the edge computing gateway to obtain preprocessed data, and inputting the preprocessed data into the random forest model; The random forest model extracts the low-frequency, mid-frequency, and high-frequency energy of the vibration data, the sliding window mean and rate of change of the temperature data, and the standardized ratio of the current load to the rated load. Each decision tree independently determines its category, and the voting results of all trees are counted to calculate the failure probability. If the failure probability is greater than or equal to 0.9, an emergency shutdown is triggered. If the failure probability is less than 0.9 and greater than or equal to 0.7, a high-risk alert is triggered; if it is less than 0.7, it is a low risk.

8. The multimodal transport one-box intelligent and trustworthy collaborative system based on artificial intelligence technology is characterized by: include: The data collaboration module converts the transport data into a standardized JSON format according to the agreed rules to obtain a standardized JSON data packet, triple-encrypts the standardized JSON data packet to obtain an encrypted data packet, and broadcasts it to the blockchain; The verification module collects IoT device data and key fields of the electronic waybill, including the bill of lading number, cargo category, and consignee and consignor, in real time. The verification nodes in the blockchain network perform consensus verification on the IoT device data and the key fields of the electronic waybill. If the verification passes, the data is written to the blockchain. If there is a data conflict, an early warning is issued and the data is corrected. In the intelligent hub dynamic scheduling module, the digital twin system integrates multi-source data, including physical equipment data, business system data, and external environment data. It extracts and calculates data based on this multi-source data. Artificial intelligence calculates the optimal loading and unloading sequence and transportation path based on this data, generates operation instructions, and sends them to on-site automation equipment. The tracking module uploads longitude and latitude data to the blockchain node every minute through the Beidou positioning terminal built into the container and the 5G / LoRa gateway along the way. It automatically identifies anomalies in the transportation stage based on geo-fencing technology and pushes alarms to the shipper and insurance company when the physical device data exceeds the smart contract threshold.

9. A computer device comprising a memory and one or more processors, wherein the memory stores computer code, and when the computer code is executed by the one or more processors, the one or more processors execute the steps of the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer code, wherein when the computer code is executed, the steps of the multimodal transport one-box intelligent and trusted collaboration method based on artificial intelligence technology as described in any one of claims 1 to 7 are executed.

Images