An ETV non-inductive weighing method and system based on multi-level checking

Abstract

The application relates to the technical field of intelligent warehousing, and particularly discloses an ETV non-inductive weighing method and system based on multilevel checking, which comprises the following steps: deploying a sensor network in an ETV channel, collecting container weight and position data, constructing a transportation digital track and visually displaying the track, fusing and analyzing weight mutation, track deviation and time deviation indexes, generating a hierarchical early warning through an adaptive threshold model, generating a TV vehicle scheduling scheme and an optimal path through a multi-objective optimization algorithm according to the early warning level and the site condition, establishing a full-process data chain, ensuring data continuity through an automatic repair mechanism, providing multi-dimensional traceability query, and regularly evaluating system performance and self-diagnosing based on a historical data machine learning optimization model and strategy. The application solves the problems of data faults in the checking and transfer link and the coordination blind area between levels in the prior art, realizes real-time detection, efficient processing and data traceability of the whole process of the multilevel freight station ETV non-inductive weighing, and improves the transfer efficiency and reliability of the container.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing technology, and in particular to an ETV-based contactless re-weighing method and system based on multi-level verification. Background Technology

[0002] Multi-level air cargo terminals, as a core component of air cargo hubs, handle the entire process of palletizing, transshipment, and reweighing of containers, and their efficiency directly impacts flight loading timeliness. Currently, cargo terminals generally adopt a model of palletizing at each floor, transshipment via ETV, and manual reweighing on the first floor. This model only includes a single reweighing step, with no intermediate verification steps. As a result, 70% of weight anomalies are only discovered after the container has been transshipped to the first floor. Returning abnormal containers to the original floor for processing is time-consuming, costly, and inefficient.

[0003] Although the freight station has established a freight production system and an ETV system, the data of the two are not linked. There is a lack of real-time monitoring of the containers during inter-floor transfer, resulting in a data gap between the verification and transfer process. Anomalies caused by cargo shifting or path deviation during ETV transfer cannot be detected in real time and can only be discovered when the containers reach the first-floor reweighing point. This further increases the frequency of rework and wastes transport capacity. In addition, the ETV scheduling lacks a dynamic path planning mechanism, and the fixed turnaround routes easily cause congestion in the floor passages. The one-way transfer time is long, the operation error rate is high, and it is difficult to adapt to the high-efficiency operation requirements of multi-story freight stations.

[0004] Therefore, there is an urgent need for a multi-level verification-based ETV seamless re-verification method and system to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-level verification-based ETV seamless re-weighing method, comprising the following steps:

[0006] Deploy a sensor network in the ETV lane to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through dynamic dashboards.

[0007] Based on sensor data fusion analysis of indicators such as weight mutation, trajectory deviation and timeliness deviation, an adaptive threshold model is used to generate graded early warnings, which are then transmitted to the central system in real time.

[0008] TV vehicle dispatching plan is dynamically generated based on the warning level and site conditions. The optimal route is planned through a multi-objective optimization algorithm, and navigation instructions are sent to the vehicle terminal.

[0009] Establish a full-process data chain system to ensure data integrity from anomaly detection to processing completion, ensure data chain continuity through automatic repair mechanisms, and provide traceability and query functions;

[0010] The system optimizes detection models and scheduling strategies based on historical data through machine learning, regularly assesses system performance, and self-diagnoses its operational status.

[0011] Furthermore, the steps of deploying a sensor network in the ETV lane to continuously collect container weight and location data, construct a digital transportation trajectory, detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard include:

[0012] Weight monitoring points are set up at fixed intervals along the ETV lane, and each monitoring point is equipped with a set of weighing sensors of specific accuracy. Position beacons are added at the turns of the lane and the connection between floors.

[0013] Establish a sensor data acquisition network and transmit real-time data to a central server via a wireless network;

[0014] Based on the collected real-time data, a digital trajectory for container transportation is constructed, including three dimensions: weight-time curve, location-time trajectory, and speed-time curve. A transportation status assessment model is also established to calculate the stability index of the transportation process in real time.

[0015] The design incorporates a multi-level monitoring dashboard. The first level displays the overall transportation status, the second level displays detailed data for individual containers, and the third level displays abnormal warning information.

[0016] Furthermore, the step of analyzing weight mutations, trajectory deviations, and timeliness deviations based on sensor data fusion, generating tiered early warnings using an adaptive threshold model, and transmitting them to the central system in real time includes:

[0017] A weighted fusion method is used to calculate the comprehensive anomaly index, where the weight coefficients are obtained by training based on historical data and an adaptive adjustment mechanism is set to establish a multi-level early warning system with different early warning levels.

[0018] Based on machine learning algorithms, the thresholds of each indicator are dynamically adjusted according to factors such as historical abnormal data, environmental factors, and operator skill level.

[0019] An efficient data compression protocol is adopted, and an early warning confirmation mechanism is designed. If the recipient fails to confirm receipt of the early warning within a preset time, the early warning level will be automatically upgraded and transmitted to the central system in real time.

[0020] Furthermore, the steps of dynamically generating a TV vehicle dispatch plan based on the warning level and site conditions, planning the optimal route through a multi-objective optimization algorithm, and issuing navigation commands to the vehicle terminal include:

[0021] Establish a mathematical model that includes multiple optimization objectives, including minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs;

[0022] Based on real-time site monitoring data, the optimal processing route is dynamically generated to avoid congested areas and peak hours.

[0023] Automatically generate a scheduling scheme containing complete processing instructions, the scheduling scheme including path navigation information, processing time limit requirements, and exception handling guidelines;

[0024] It tracks the execution of scheduling instructions in real time and automatically issues warnings for execution deviations.

[0025] Furthermore, the steps of establishing a full-process data chain system to achieve data integrity from anomaly detection to processing completion, ensuring data chain continuity through an automatic repair mechanism, and providing traceability and query functions include:

[0026] Establish data quality inspection rules to automatically mark and repair abnormal data;

[0027] When a data transmission interruption or loss is detected, the data reconstruction program is automatically started, and the complete data chain is restored through three steps: data retransmission, data interpolation processing, and data verification.

[0028] The intelligent query interface allows for multi-dimensional queries based on container number, time range, anomaly type, and other criteria.

[0029] Based on historical anomaly data, an optimization suggestion report is automatically generated, which includes sensor deployment optimization, threshold adjustment suggestions, and scheduling strategy improvements.

[0030] Furthermore, the steps of optimizing the detection model and scheduling strategy based on historical data through machine learning, and periodically evaluating system performance and self-diagnosing its operating status include:

[0031] Based on historical operation data and anomaly handling records, machine learning algorithms are used to optimize the anomaly detection model and scheduling strategy.

[0032] Mobile monitoring devices enable real-time reception of early warning information, viewing of processing instructions, and feedback of processing results, ensuring timely handling of anomalies.

[0033] Construct a key performance indicator system, which includes indicators such as anomaly detection timeliness, processing completion rate, and data integrity, and generate system performance evaluation reports regularly;

[0034] Through automatic diagnostic algorithms, the status of sensors, network connection quality, and system operation are checked regularly to detect potential problems and provide early warnings.

[0035] Furthermore, the present invention also discloses an ETV contactless re-weighing system based on multi-level verification, comprising:

[0036] The data acquisition module is used to deploy a sensor network in the ETV lane to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard.

[0037] The generation module is used to analyze weight mutation, trajectory deviation and timeliness deviation indicators based on sensor data fusion, generate graded early warnings using an adaptive threshold model, and transmit them to the central system in real time.

[0038] The scheduling module is used to dynamically generate TV vehicle scheduling plans based on the warning level and site conditions, plan the optimal route through a multi-objective optimization algorithm, and send navigation instructions to the vehicle terminal.

[0039] The module is used to establish a full-process data chain system, realize data integrity after anomaly detection and processing, ensure data chain continuity through an automatic repair mechanism, and provide traceability and query functions;

[0040] The diagnostic module is used to optimize detection models and scheduling strategies based on historical data through machine learning, periodically evaluate system performance, and self-diagnose the operating status.

[0041] Furthermore, the scheduling module includes:

[0042] The optimization unit is used to establish a mathematical model that includes multiple optimization objectives, such as minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs.

[0043] The first generation unit is used to dynamically generate the optimal processing path based on real-time site monitoring data, avoiding congested areas and peak hours;

[0044] The second generation unit is used to automatically generate a scheduling scheme containing complete processing instructions. The scheduling scheme includes path navigation information, processing time limit requirements, and exception handling guidelines.

[0045] The tracking unit is used to track the execution of scheduling instructions in real time and automatically issue warnings for execution deviations.

[0046] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described ETV seamless re-weighing method based on multi-level verification.

[0047] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described ETV seamless re-weighting method based on multi-level verification.

[0048] The beneficial effects of this application are as follows:

[0049] This invention achieves real-time data acquisition during container transportation by deploying a sensor network in ETV channels, generates tiered early warnings through multi-dimensional data fusion analysis, dynamically generates TV vehicle scheduling schemes through multi-objective optimization algorithms, constructs a full-process data chain to ensure continuous data traceability, and finally optimizes models and strategies based on historical data using machine learning. This solves the problems of data gaps and inter-level coordination blind spots in the verification-transfer process in existing technologies, enabling real-time anomaly detection, efficient processing, and data traceability throughout the entire ETV seamless reweighing process at multi-level freight stations, thereby improving container transfer efficiency and reweighing reliability. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of a method flow proposed in an embodiment of this application.

[0051] Figure 2 This is a schematic diagram of the system structure proposed in an embodiment of the present invention.

[0052] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0053] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0054] like Figure 1 As shown, this application provides a multi-level verification-based ETV seamless re-weighing method, including the following steps:

[0055] S1, deploy a sensor network in the ETV channel to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard.

[0056] S2, based on sensor data fusion analysis of weight mutation, trajectory deviation and timeliness deviation indicators, uses an adaptive threshold model to generate graded early warnings and transmits them to the central system in real time;

[0057] S3 dynamically generates TV vehicle dispatching plans based on warning levels and site conditions, plans the optimal route through multi-objective optimization algorithms, and sends navigation instructions to the vehicle terminal.

[0058] S4, establish a full-process data chain system to achieve data integrity from anomaly detection to processing completion, ensure data chain continuity through automatic repair mechanism, and provide traceability and query functions;

[0059] S5 uses machine learning based on historical data to optimize detection models and scheduling strategies, regularly evaluates system performance, and self-diagnoses its operating status.

[0060] As described in steps S1-S5 above, this invention achieves real-time data collection during container transportation by deploying a sensor network in the ETV channel, generates hierarchical early warnings by combining multi-dimensional data fusion analysis, dynamically generates TV vehicle scheduling schemes through multi-objective optimization algorithms, constructs a full-process data chain to ensure continuous and traceable data, and finally relies on historical data to machine learning optimization models and strategies. This solves the problems of data gaps and inter-level coordination blind spots in the verification-transfer link in the existing technology, realizes real-time anomaly detection, efficient processing and data traceability of the entire process of seamless ETV reweighing in multi-level freight stations, and improves container transfer efficiency and reweighing reliability.

[0061] Current technology only sets up fixed verification points within each floor of the freight station. There is no real-time monitoring of the containers during inter-floor transport. If discrepancies occur due to cargo shifting causing weight abnormalities, deviations from the preset route by the transport vehicle (TV), or delays in transport time, these issues can only be detected upon arrival at the target verification point. This necessitates the TV vehicle to return along the original route, increasing one-way travel time. Furthermore, anomaly tracing relies on manual verification of paper documents against system data, further extending the average tracing time and creating a significant data gap between verification and transport. Therefore, it is necessary to establish a data collection link during transport to achieve real-time anomaly detection and collaborative processing, avoiding efficiency losses caused by coordination blind spots.

[0062] The core shortcomings of existing technologies lie in their reliance on fixed-point data collection, lack of data support during transshipment, delayed anomaly detection, absence of dynamic scheduling mechanisms, fixed turnaround routes for transit vehicles (TVs), susceptibility to congestion leading to extended processing times, discontinuous data chains, and reliance on manual traceability, resulting in low efficiency. To address these issues, this invention utilizes a full-channel sensor network to achieve real-time data collection during transshipment, filling data gaps. It achieves precise, tiered early warning through multi-dimensional data fusion and an adaptive threshold model, avoiding false alarms and missed alarms. A multi-objective optimization algorithm dynamically plans TV vehicle routes, improving scheduling efficiency. A full-process data chain and automatic repair mechanism ensure data continuity, reducing traceability costs. Machine learning continuously optimizes the model to adapt to changes in freight station conditions, forming a targeted closed-loop solution.

[0063] In one embodiment, the steps of deploying a sensor network in the ETV lane to continuously collect container weight and location data, construct a digital transportation trajectory, detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard include:

[0064] S11. Weight monitoring points are set up at fixed intervals along the ETV channel. Each monitoring point is equipped with a weighing sensor group of specific accuracy. Position beacons are added at channel bends and floor connections to achieve precise positioning of the container.

[0065] S12 uses an industrial IoT protocol to establish a sensor data acquisition network, transmits real-time data to a central server via a wireless network, and designs a data compression algorithm to reduce the amount of data transmission and ensure that data is not lost when the network is congested.

[0066] S13, based on the collected real-time data, construct the container transportation digital trajectory, including three dimensions: weight time curve, position time trajectory, and speed time curve, and establish a transportation status assessment model to calculate the transportation process stability index in real time.

[0067] The S14 features a multi-level monitoring dashboard. The first level displays the overall transportation status, the second level displays detailed data for individual containers, and the third level displays abnormal warning information. It supports multi-dimensional data filtering and custom alarm rule settings.

[0068] As described in steps S11-S14 above, by deploying a sensor network throughout the ETV channel to continuously collect weight and location data, relying on the Industrial Internet of Things to build a stable data transmission link, constructing a three-dimensional transportation digital trajectory based on real-time data and establishing a stability evaluation model, and finally realizing the visualization of transportation status through multi-level monitoring dashboards, thereby filling the data collection blind spots in the inter-floor transshipment process of the container and achieving full-process data coverage and real-time monitoring of the verification-transshipment link.

[0069] The transshipment routes for containers in multi-level freight stations include complex sections such as straight passages, turns, and floor connections. Current technology only installs fixed monitoring equipment at floor entrance checkpoints and first-floor reweighing points, lacking any data collection methods during transit. When container cargo shifts due to road bumps, causing changes in weight distribution or the transshipment vehicle deviating from its preset route to avoid obstacles, the anomaly is only detected upon reaching the first-floor reweighing point. This necessitates the transshipment vehicle turning back, increasing one-way travel time and leading to a significant daily rework volume due to such delayed detection. Therefore, it is necessary to deploy sensing equipment throughout the transshipment route to collect weight and location data in real time, enabling immediate anomaly detection and avoiding unnecessary backtracking.

[0070] The system employs a fixed-interval weight monitoring system along the ETV lanes, with each monitoring point equipped with a specific precision load cell array. Position beacons are added at lane bends and floor connections to achieve precise container positioning and create a comprehensive sensing network, resolving data blind spots. Based on the actual length of the ETV lanes at the freight station, 30 weight monitoring points are deployed at fixed 5-meter intervals. Each monitoring point uses a load cell array of four strain gauge load cells with an accuracy of ±0.2kg, consistent with the accuracy of the sensors at the floor calibration points to ensure data calibration consistency. Two UWB position beacons are added at lane bends and floor connections to achieve real-time container positioning within a ±0.5-meter range using triangulation principles. This method covers the entire path of the container from the floor verification point to the first-floor reweighing point. For example, container X123 starts from entrance E2 on the third floor, passing through the straight passage on the third floor, the turning section from the third floor to the second floor, the straight passage on the second floor, the connection point from the second floor to the first floor, and the straight passage on the first floor to the F3 exit platform on the first floor. Its weight can be collected at each 5-meter monitoring point, and the beacons at the turning points and connection points can accurately locate its coordinates, avoiding monitoring omissions due to the complexity of the path. Compared with traditional fixed-point monitoring, the spatial coverage of anomaly detection can be expanded from 2 points to the entire passage, ensuring that any anomalies occurring at any point during the transport of the container can be detected.

[0071] A sensor data acquisition network is established using an Industrial Internet of Things (IIoT) protocol. Real-time data is transmitted to a central server via a wireless network. A data compression algorithm is designed to reduce data transmission volume, ensuring no data loss during network congestion and guaranteeing the real-time performance and integrity of data transmission, thus solving the problem of data loss during peak hours. A star-shaped data acquisition network is constructed using the MQTT IIoT protocol, with each sensor node acting as a client and the central server as a broker. Data is transmitted at a frequency of "weight data once per second and location data once per 0.5 seconds". A differential compression algorithm is designed to transmit only the initial value and the deviation value (300kg, +0.1kg, -0.1kg) for three consecutive weight data (e.g., 300kg, 300.1kg, 299.9kg), reducing the data transmission volume by 60%. At the same time, a 16GB local cache is built into the sensor node to automatically store data when the network is congested and retransmit it according to the timestamp after the connection is restored. The cargo terminal experiences peak container transshipment from 10:00 to 12:00 daily. During this time, the concurrent data volume from 30 weight monitoring points and 10 location beacons reaches 50Mbps. Traditional transmission methods are prone to data loss due to insufficient network bandwidth. However, this step, through protocol optimization and compression algorithms, can control the data transmission volume during peak hours to within 20Mbps. Combined with a local caching mechanism, this ensures that the data loss rate is reduced.

[0072] Based on the collected real-time data, a digital trajectory for container transportation is constructed, including three dimensions: weight-time curve, position-time trajectory, and speed-time curve. A transportation status assessment model is also established to calculate the stability index of the transportation process in real time. This allows for the digital reconstruction of the transportation process and the quantitative assessment of anomalies. Technically, with time as the horizontal axis, continuously collected weight data is plotted as a weight-time curve. For example, the curve for container X123 from 10:10 to 10:15 shows the weight decreasing from 300kg to 298kg. The position coordinates (x, y, z) are plotted as a position-time trajectory (e.g., the movement trajectory from (10, 20, 6) to (15, 20, 6)). Speed ​​is calculated based on the time difference between adjacent positions, generating a speed-time curve. The transportation status assessment model sets three stability indicators: weight stability (three consecutive weight deviations ≤ 1.5%), trajectory stability (deviation from the preset route ≤ 1 meter), and speed stability (deviation from the preset speed of 0.6m / s ≤ 0.2m / s). Periods that do not meet these indicators are calculated and marked in real time. Its physical significance lies in the ability to intuitively trace the entire process of an anomaly through a three-dimensional digital trajectory. For example, when container X123 passed through the second-floor turning section at 10:12:30, the weight-time curve showed a weight decrease of 1.8 kg within 1 second (exceeding the 1.5% threshold), while the position-time trajectory showed a deviation of 1.1 meters from the preset route (exceeding the 1-meter threshold). Combining these two factors, it can be determined that the center of gravity shift was caused by cargo displacement, rather than sensor error. This model can improve the accuracy of anomaly identification and avoid misjudgments caused by fluctuations in a single data point.

[0073] The system features a multi-tiered monitoring dashboard. The first tier displays the overall transportation status, the second tier shows detailed data for individual containers, and the third tier displays anomaly warnings. It supports multi-dimensional data filtering and custom alarm rule settings. This design aims to achieve precise information distribution and improve the collaborative efficiency of various roles. Technically, the first-tier dashboard uses a heatmap to display the number of TV trucks in each channel and the percentage of abnormal containers. The second-tier dashboard displays the real-time weight, current location, estimated arrival time, and a 3D digital trajectory thumbnail for each individual container. The third-tier dashboard lists anomaly information according to warning levels (Level 1, Level 2, Level 3) (e.g., "X123, 50 meters in the second-floor channel, weight fluctuation 1.8%"). It supports multi-dimensional filtering by floor, TV truck number, and anomaly type, allowing dispatchers to customize alarm rules, such as setting the weight fluctuation threshold for fragile items to 1%. Dispatchers can quickly locate areas of concentrated anomalies through the first-tier dashboard, TV truck drivers can view the real-time status of their transported containers through the second-tier dashboard, and maintenance personnel can prioritize handling Level 3 warnings through the third-tier dashboard, solving the problem of information overload in traditional single-interface systems. The time required for each role to obtain the necessary information is shortened, and the efficiency of anomaly response is improved.

[0074] In one embodiment, the step of analyzing weight mutations, trajectory deviations, and timeliness deviations based on sensor data fusion, generating tiered early warnings using an adaptive threshold model, and transmitting them to the central system in real time includes:

[0075] S21 uses a weighted fusion method to calculate the comprehensive anomaly index, where the weight coefficients are obtained by training based on historical data and an adaptive adjustment mechanism is set to establish a multi-level early warning system with different levels of early warning.

[0076] S22, based on machine learning algorithms, dynamically adjusts the thresholds of various indicators according to factors such as historical abnormal data, environmental factors, and operator skill level, and sets up a threshold smooth transition mechanism to avoid false alarms caused by sudden threshold changes;

[0077] S23 employs a high-efficiency data compression protocol to ensure that early warning information is transmitted to relevant systems in a short time. It is designed with an early warning confirmation mechanism. If the recipient does not confirm receipt of the early warning within a preset time, the early warning level will be automatically upgraded and transmitted to the central system in real time.

[0078] As described in steps S21-S23 above, by extracting multi-dimensional indicators from the weight, position, and time data collected by the sensors, a weighted fusion and adaptive threshold model is used to generate graded early warnings. An efficient transmission and confirmation mechanism ensures that the early warning information is delivered to the central system in real time, thereby achieving accurate identification, graded response, and efficient transmission of anomalies during container transport. This solves the problems of false alarms, missed alarms, and delayed early warning response caused by traditional single-indicator judgment.

[0079] When containers are transported between floors, anomalies can have multiple sources. Cargo shifting can cause sudden weight changes, such as a 2kg loss within one second; sudden turns by the transport vehicle (TV) can cause trajectory deviations, such as a 1.2-meter deviation from the preset route; and congestion in the passageway can cause transport delays. A single-dimensional anomaly, such as a 1.2% weight fluctuation, may be a sensor error rather than a genuine anomaly. It is necessary to combine this with other data, such as no trajectory deviation and no timeout, for a comprehensive judgment. Furthermore, the operating conditions of the freight station change with the environment and personnel, and fixed thresholds are prone to false alarms or missed alarms under complex conditions. Therefore, it is necessary to construct a multi-dimensional indicator fusion judgment system and dynamically adjust thresholds according to operating conditions to achieve accurate anomaly identification and tiered response.

[0080] The weight dimension includes indicators of weight mutation degree and fluctuation frequency; the trajectory dimension includes indicators of path deviation degree and speed anomaly degree; and the time dimension includes indicators of transportation timeout and abnormal dwell time. This allows for the construction of a multi-dimensional anomaly feature system, addressing the limitations of relying on a single indicator. Weight mutation degree is defined as the ratio of weight change per unit time (1 second) to the initial weight (e.g., container X123 initially weighs 300 kg, decreasing by 3 kg in 1 second, a mutation degree of 1%). Fluctuation frequency is defined as "the number of times the weight change exceeds the threshold (1%) within 1 minute" (e.g., 3 times). Path deviation degree is defined as "the straight-line distance between the current position and the preset route" (e.g., 1.2 meters). Speed ​​anomaly degree is defined as "the absolute value of the difference between the actual speed and the preset speed (0.6 m / s)" (e.g., 0.5 m / s). Transportation timeout is defined as "the difference between the actual transshipment time and the standard time (10 minutes)" (e.g., +5 minutes). Abnormal dwell time is defined as "the unplanned dwell time within the passageway" (e.g., 1 minute). Multi-dimensional indicators can comprehensively characterize abnormal features. For example, container X123 exhibited a weight change of 1.2% (exceeding the 1% threshold), fluctuation frequency of 2 times (not exceeding the 3 times threshold), path deviation of 0.3 meters (not exceeding the 1-meter threshold), and speed anomaly of 0.1 m / s (not exceeding the 0.2 m / s threshold), with no timeout or stoppage. This comprehensive assessment classifies it as a minor anomaly, avoiding misjudgment of a severe anomaly by a single weight indicator. Technically, multi-dimensional indicators can reduce the bias of anomaly identification by 60%, laying the foundation for subsequent accurate early warnings.

[0081] A weighted fusion method is used to calculate the comprehensive anomaly index. The weight coefficients are trained based on historical data and an adaptive adjustment mechanism is set up to establish a multi-level early warning system. Different early warning levels are set, and intelligent fusion can achieve quantitative classification of anomalies, solving the problem of insufficient adaptability caused by fixed index weights. The weighted fusion formula is "Comprehensive Anomaly Value = Weight Index × 0.4 + Trajectory Index × 0.3 + Time Index × 0.3". The weight coefficients are trained using historical data from 4,500 containers (including 300 real anomalies) from June to August 2024.

[0082] The adaptive adjustment mechanism is set so that the weight of the trajectory indicator is increased to 0.4 in rainy weather (weight and time are reduced to 0.35 and 0.25 respectively), and the weight indicator weight is increased to 0.45 when a new driver is operating the vehicle (trajectory and time are reduced to 0.25 and 0.3 respectively). The multi-level early warning system has three threshold levels: Level 1 warning (comprehensive value 0.3-0.5), Level 2 warning (0.5-0.8), and Level 3 warning (>0.8). For example, when container X123 is transporting goods in rainy weather, the weight indicator is 0.3, the trajectory indicator is 0.6, and the time indicator is 0.2. The comprehensive value calculated according to the rainy weather weight is 0.3×0.35+0.6×0.4+0.2×0.25=0.0105+0.24+0.05=0.395, triggering a Level 1 warning. The dynamic adjustment of weights can adapt to the abnormal characteristics under different working conditions, and the graded early warning allows the response resources to accurately match the severity of the abnormality. Specifically, Level 1 warnings only notify the driver, while Level 3 warnings are handled by designated personnel. This method can improve the accuracy of early warning classification and avoid resource waste and insufficient response.

[0083] Based on machine learning algorithms, this system dynamically adjusts the thresholds of various indicators according to historical anomaly data, environmental factors, and operator skill levels. A threshold smoothing transition mechanism is implemented to avoid false alarms caused by sudden threshold changes. Dynamic threshold optimization improves anomaly identification accuracy and addresses the adaptability limitations of fixed thresholds. A random forest algorithm is used to construct the threshold adjustment model. Input parameters include historical anomaly data from the past 30 days, such as the average amplitude of weight mutations, environmental factors like rainfall and temperature, and operator skill levels, such as driver anomaly handling scores over the past three months. The output is the dynamic threshold for each indicator. The initial threshold for weight mutation severity is set at 1.5%. When rainfall exceeds 50mm for three consecutive days, the model automatically adjusts it to 1.7%, with each adjustment not exceeding 0.2%. The smoothing transition mechanism is implemented through a weighted calculation of new threshold = old threshold × 0.7 + adjustment value × 0.3, avoiding sudden daily threshold changes. For example, when container X123 is transporting goods in rainy weather, a sudden weight change of 1.6% should trigger an alarm according to the original threshold of 1.5%. However, the dynamic threshold is adjusted to 1.7%, and after a smooth transition, it becomes 1.56%. The 1.6% change does not exceed the threshold, thus avoiding false alarms. The threshold dynamically changes with operating conditions to match the actual occurrence of anomalies (normal weight fluctuations caused by bumps in rainy weather do not require an alarm), and the smooth transition prevents continuous false alarms caused by sudden threshold changes. This mechanism can reduce the false alarm rate in rainy weather and improve system adaptability.

[0084] A high-efficiency data compression protocol is adopted to ensure that early warning information is transmitted to relevant systems within a short time. An early warning confirmation mechanism is designed so that if the recipient does not confirm receipt of the early warning within a preset time, the warning level is automatically upgraded and transmitted to the central system in real time. This ensures the real-time nature of the early warning information and the effectiveness of the response, solving the problems of transmission delay and response lag. The CoAP protocol is used to compress early warning information (including container number, anomaly type, and comprehensive value), and the data packet size is controlled within 128 bytes, ensuring delivery to the vehicle terminal and dispatch console within 1 second. The early warning confirmation mechanism has a 30-second confirmation time limit. The recipient needs to click the "confirm" button on the terminal to provide feedback. If no confirmation is received, the warning is automatically upgraded (Level 1 to Level 2, Level 2 to Level 3), and the upgrade information is synchronously transmitted to the central system log. For example, if a Level 1 early warning for container X123 is sent to the TV-05 vehicle terminal and the driver does not confirm within 30 seconds, the system automatically upgrades the warning to Level 2 and synchronizes it to the central system, allowing the dispatcher to intervene and urge handling. The high-efficiency protocol is adapted to the cargo station's wireless network environment (delay ≤ 1 second during peak hours), the confirmation mechanism ensures that early warnings are not ignored, and the upgrade mechanism strengthens the response to emergency anomalies. The reduced transmission delay of early warning information also reduced the non-response rate, ensuring timely handling of anomalies.

[0085] In one embodiment, the steps of dynamically generating a TV vehicle dispatching plan based on the warning level and site conditions, planning the optimal route through a multi-objective optimization algorithm, and issuing navigation instructions to the vehicle terminal include:

[0086] S31. Establish a mathematical model that includes multiple optimization objectives, including minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs. Use optimization algorithms to solve for the optimal scheduling scheme and control the algorithm solution time.

[0087] S32 dynamically generates the optimal processing route based on real-time site monitoring data, avoiding congested areas and peak hours. The route planning takes into account multiple factors such as the current location of the TV vehicle, the traffic status of the passage, and the urgency of the task.

[0088] S33, automatically generate a scheduling scheme containing complete processing instructions, the scheduling scheme including path navigation information, processing time limit requirements, and exception handling guidelines;

[0089] S34 tracks the execution of scheduling instructions in real time and automatically issues warnings for execution deviations.

[0090] As described in steps S31-S34 above, a multi-objective optimization mathematical model is established to solve the TV vehicle scheduling scheme. The optimal path is dynamically planned by combining real-time site data. A scheduling scheme containing complete instructions is generated and the execution status is tracked. This achieves efficient scheduling and accurate navigation of TV vehicles during anomaly handling, solving the problems of long processing time, channel congestion, and ambiguous instructions caused by traditional fixed-path scheduling.

[0091] This involves establishing a mathematical model with multiple optimization objectives, including minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs. An optimization algorithm is used to solve for the optimal scheduling scheme, and the algorithm's solution time is controlled. Multi-objective collaborative optimization can overcome the limitations of a single objective, ensuring the global optimality of the scheduling scheme. It is important to note that the objective function of the mathematical model is defined as: min (α × processing time + β × channel occupancy rate + γ × operating cost), where α = 0.5, β = 0.3, and γ = 0.2. The weights were determined through training on 120 anomaly handling data points from June to August 2024. Processing time is the travel time of the TV vehicle from its current location to the anomaly handling point and then to the target recovery point. Channel occupancy impact is the real-time congestion rate of the channels along the route. Operating cost is the travel distance × cost per unit mileage. The NSGA-II algorithm is used for solving the problem, with a population size of 50 and 30 iterations, keeping the solution time within 1 second to meet real-time scheduling requirements. The multi-objective model can find a balance between time, occupancy, and cost, avoiding overall efficiency loss due to single-objective optimization. Solving within 1 second ensures the solution can adapt to real-time changes in the site. This model can improve the overall efficiency of anomaly handling.

[0092] Based on real-time site monitoring data, an optimal processing path is dynamically generated to avoid congested areas and peak hours. The path planning considers multiple factors, including the current location of the TV vehicles, the traffic flow status of the passageways, and the urgency of the task. Real-time data-driven path adjustments address the issue of fixed paths failing to adapt to dynamic site changes. Real-time site monitoring data includes the number of TV vehicles in each passageway (e.g., 3 vehicles in the main passageway and 1 vehicle in the backup passageway on the second floor), average driving speeds (e.g., 0.3 m / s in the main passageway and 0.6 m / s in the backup passageway), and peak hours (10:00-12:00). The path planning algorithm transforms the passageway traffic status into a cost function, which is:

[0093]

[0094] in, Represents the total cost of the path. Indicates the path length (unit: meters, such as 150 meters in X123). This represents the current number of TV vehicles on the route (e.g., 1 vehicle in the backup lane). The maximum number of TV vehicles that the route can accommodate (e.g., 3 vehicles in the backup lane), of which, the congested section ( The cost coefficient for congestion-free sections is 1.5 (0% congestion rate) and 1.0 (0% congestion rate), ensuring that routes prioritize routes with low congestion rates. Task urgency (level 3 warning weight 1.2, level 2 0.9, level 1 0.7) is used as the heuristic function weight. For example, when TV-05 processes a level 3 warning for X123, the current location is 15 meters from the second-floor passage. The original fixed route requires passing through the main second-floor passage (congested). The dynamically planned route is: 15 meters from the second-floor passage - left turn into the backup passage (0.6 m / s) - straight for 50 meters to the processing point - right turn into the first-floor secondary passage - arrive at the F3 re-processing point. The total distance is 20 meters shorter than the fixed route, and the time is reduced by 3 minutes. Real-time data ensures that the route avoids sudden congestion, and multiple factors are considered to make the route both efficient and adaptable to task priority. Dynamic route planning can reduce the deviation between actual travel time and planned time, reducing congestion delays compared to fixed routes.

[0095] The system automatically generates a scheduling plan containing complete processing instructions. This plan includes path navigation information, processing time limits, and anomaly handling guidelines. Standardized instructions improve the standardization and efficiency of anomaly handling, resolving operational delays caused by ambiguous traditional instructions. Path navigation information uses segmented guidance, such as "10:05-10:07: Proceed straight for 50 meters along the backup route, maintaining a speed of 0.6 m / s," achieving meter-level navigation in conjunction with location beacons. Time limits are set according to the warning level (Level 3 warning ≤ 8 minutes, Level 2 ≤ 12 minutes, Level 1 ≤ 15 minutes), and are linked to the TV vehicle's real-time location to update the remaining time. The anomaly handling guidelines pre-set operating steps for different anomaly types, such as for sudden weight changes: "1. After stopping, check the tightness of the cargo securing straps; 2. Re-weigh the cargo using a handheld scale to confirm the weight; 3. Re-secure the cargo and click 'Continue Transportation'." For example, in X123's three-level early warning dispatch scheme, the navigation information clearly states "depart from the current location at 10:05, arrive at the second-floor C2 processing point at 10:07," with a time limit requirement of "complete processing and depart before 10:13." The guidelines detail the cargo inspection steps. Standardized instructions eliminate driver operational uncertainties, such as not knowing how to re-weigh items, while time constraints ensure processing rhythm and avoid delays. Complete instructions can shorten the driver's abnormal handling operation time and improve operational standardization.

[0096] The system tracks the execution of dispatch instructions in real time, automatically issuing warnings for deviations. Closed-loop monitoring ensures the effective implementation of the dispatch plan and resolves deviations during instruction execution. The TV vehicle's coordinates are collected in real time via location beacons (once per second). The central system compares the actual position with the planned path and calculates the deviation distance. For example, if the deviation between the actual position of TV-05 and the planned path is 0.8 meters, a deviation threshold (0.5 meters) is set. When the threshold is exceeded, a warning message is immediately sent to the onboard terminal, such as "0.8 meters deviation from the planned path, please adjust to the backup lane." Simultaneously, the system synchronizes the information to the monitoring dashboard. For instance, when TV-05 is handling X123 and deviates 0.6 meters from the path to avoid an obstacle, the system issues a warning within 1 second, and the driver adjusts back to the correct path within 20 seconds. Real-time tracking can promptly correct execution deviations, avoiding path extensions caused by deviations, such as entering dead ends requiring a turnback or lane conflicts. This mechanism reduces the execution deviation rate, ensuring the dispatch plan is completed as scheduled.

[0097] In one embodiment, the steps of establishing a full-process data chain system to achieve data integrity upon anomaly detection and processing completion, ensuring data chain continuity through an automatic repair mechanism, and providing traceability and query functions include:

[0098] S41, Establish data quality inspection rules to automatically mark and repair abnormal data, and ensure the integrity of the data chain;

[0099] S42, when a data transmission interruption or loss is detected, the data reconstruction program is automatically started, and the complete data chain is restored through three steps: data retransmission, data interpolation processing, and data verification.

[0100] S43, through the intelligent query interface, can be used to query by multiple dimensions such as container number, time range, and anomaly type. The query results are displayed in a visual form, including dynamic display of transportation trajectory, display of weight change curve, and display of processing record timeline.

[0101] S44. Based on historical anomaly data, an optimization suggestion report is automatically generated. The optimization suggestion report includes sensor deployment optimization, threshold adjustment suggestions, scheduling strategy improvement, etc.

[0102] As described in steps S41-S44 above, the continuity of the entire data chain is ensured by establishing a data quality inspection and automatic repair mechanism, data traceability is achieved by relying on a multi-dimensional intelligent query interface, and optimization suggestion reports are generated based on historical data. A complete data management system of data collection, quality control, repair traceability and optimization iteration is constructed to solve the problems of data dispersion, inability to recover lost data and difficulty in traceability in the traditional verification-transfer links, and to achieve complete recording and efficient utilization of data throughout the container transportation process.

[0103] Establish data quality inspection rules to automatically mark and repair abnormal data, ensuring the integrity of the data chain. This allows for source control of data quality and addresses the problem of abnormal data interfering with analysis. The data quality inspection rules include: weight data must be within the range of 0-500kg; location data deviation must be ≤0.5 meters (exceeding this range is marked as "location abnormal"); timestamps must be continuous (intervals exceeding 1 second are marked as "time chain break"). Repair methods are designed for different anomaly types: weight exceeding limits is replaced with the arithmetic mean of the preceding and following 5 seconds of data (e.g., X123's 300kg-999kg-298kg is repaired to 300kg-299kg-298kg); location abnormalities are supplemented using linear interpolation (e.g., if the location jumps from (10,20) to (15,20), missing points are filled with (12.5,20)); and time chain breaks are marked as "repaired" by filling in missing values. The data includes weight and location data collected by the sensor network and the warning timestamps generated above. Automatic tagging can quickly identify unreliable data (such as outliers caused by sensor malfunctions) and perform targeted repairs to ensure the data chain remains continuous at anomaly points (e.g., the weight curve of X123 has no abrupt breakpoints), providing a reliable foundation for subsequent traceability and analysis. This rule can improve data accuracy and reduce erroneous analysis caused by data anomalies.

[0104] When a data transmission interruption or loss is detected, the data reconstruction program is automatically initiated. Through three steps—data retransmission, data interpolation, and data verification—the complete data chain is restored, resolving chain breaks caused by sudden data loss and ensuring continuous data flow throughout the entire process. Data transmission interruptions are identified by the central system via heartbeat detection. The first step, data retransmission, involves the system sending a retransmission command to the faulty sensor, requesting the return of data from the lost period (e.g., 10:10:00-10:10:10). The second step, data interpolation, involves using cubic spline interpolation to generate missing data when retransmission fails (e.g., if the weight data for X123 drops from 298kg to 297kg during this period, interpolation generates continuous data for the eight intermediate time points). The third step, data verification, compares the reconstructed data with the preceding and following valid data; if the error is ≤0.5%, it is considered valid (e.g., the error between the interpolated data of 297.5kg and the preceding and following actual values ​​of 298kg and 297kg is 0.17%, meeting the requirement). For example, if X123 loses 5 seconds of data due to a network interruption at 10:10:05, and the system fails to retransmit the data after restarting, continuous data is generated through interpolation and verified, thus restoring the data link to full integrity. The three-step procedure progresses step-by-step: first, it attempts to acquire the original data; if that fails, it generates reliable replacement data using an algorithm; and finally, it verifies the data to ensure quality, preventing permanent data loss due to a single point of failure. Technically, this procedure can improve the data recovery rate from 60% to 95%, and maintain a data link continuity rate above 99.9%.

[0105] The intelligent query interface allows for multi-dimensional queries by container number, time range, and anomaly type. Query results are displayed visually, including dynamic displays of transport trajectories, weight change curves, and processing record timelines. This improves data traceability efficiency and overcomes the limitations of traditional single-query and text-based displays. The intelligent query interface supports combined condition searches: container number (e.g., X123), time range (e.g., 2024-10-01 10:00-10:30), and anomaly type (e.g., weight mutation). The visualization uses WebGL technology: the dynamic transport trajectory display shows the container's movement path within the channel (highlighted in red for abnormal sections), the weight change curve is presented as a line graph (highlighted in yellow for warning thresholds), and the processing record timeline sequentially displays nodes such as "10:05 Warning - 10:07 Scheduling - 10:10 Processing Completed." For example, when querying the weight change record of X123 between 10:00 and 10:30, the interface can dynamically play its trajectory from the second floor to the first floor, highlighting it in red at 10:12, and simultaneously displaying the curve showing the weight decreasing from 300kg to 297kg at that time. The timeline also indicates the corresponding secondary warning and handling process. Multi-dimensional queries can accurately locate the required data (such as querying only the weight anomaly of X123), and visualization transforms abstract data into intuitive graphics (such as trajectory animations and curves), greatly reducing the difficulty of understanding. This interface can shorten the time for a single query and improve the efficiency of anomaly tracing.

[0106] Based on historical anomaly data, an optimization suggestion report is automatically generated. This report includes suggestions for sensor deployment optimization, threshold adjustment, and scheduling strategy improvement. Data-driven continuous system optimization overcomes the lag inherent in traditional methods relying on manual experience-based iteration. Historical anomaly data includes early warning records, processing results, and statistical queries for this step, such as 15 weight abrupt changes occurring at the second-floor corridor bend in the past 30 days. The report generation uses an association rule algorithm to analyze the correlation between anomalies and the environment and equipment. Sensor deployment optimization suggestions are based on high-frequency anomaly points; for example, if the anomaly rate at the second-floor bend is 30%, it is recommended to add two more sensors. Threshold adjustment suggestions are based on false alarm analysis; for example, if the false alarm rate for fragile item weight abrupt changes is 8%, it is recommended to adjust the threshold from 1.5% to 1.2%. Scheduling strategy improvements are based on processing time statistics; for example, if the average processing time for a level 3 early warning exceeds 10 minutes, it is recommended to add a backup TV cart. For example, the October optimization report stated, "The sensor data loss rate at 50 meters in the second-floor corridor is 7%, it is recommended to replace it with an industrial-grade sensor," and "The processing time for level 3 early warnings via the backup corridor has been reduced by 4 minutes, it is recommended to increase the priority of using the backup corridor." The report uncovers potential problems from the data and proposes targeted improvement measures, providing a data-driven basis for system optimization rather than relying on subjective judgment. Technically, the report can reduce the monthly anomaly rate, improve processing efficiency, and drive the system to continuously adapt to the operational needs of the freight station.

[0107] In one embodiment, the steps of optimizing the detection model and scheduling strategy based on historical data through machine learning, periodically evaluating system performance, and self-diagnosing operational status include:

[0108] S51, based on historical operation data and anomaly handling records, uses machine learning algorithms to optimize the anomaly detection model and scheduling strategy;

[0109] S52, through mobile monitoring devices, supports real-time reception of early warning information, viewing of processing instructions, and feedback of processing results, ensuring timely handling of anomalies;

[0110] S53, Construct a key performance indicator system, which includes indicators such as anomaly detection timeliness, processing completion rate, and data integrity, and generate system performance evaluation reports regularly;

[0111] S54 uses an automatic diagnostic algorithm to periodically check sensor status, network connection quality, and system operating status, and provides early warnings of potential problems to ensure stable system operation.

[0112] As described in steps S51-S54 above, the anomaly detection model and scheduling strategy are optimized based on historical data using machine learning algorithms. Real-time response to anomaly handling is achieved by relying on mobile monitoring equipment. A key performance indicator system is constructed to evaluate system performance. Potential problems are detected in advance through automatic diagnostic algorithms, forming a closed-loop system of model optimization, real-time response, performance evaluation, and fault early warning. This ensures that the multi-level verification ETV contactless weighing system continuously adapts to operational needs as operating conditions change, improves anomaly handling efficiency and system stability, and solves the problems of traditional reliance on manual experience iteration, response lag, and sudden faults.

[0113] Based on historical operational data and anomaly handling records, machine learning algorithms are used to optimize the anomaly detection model and scheduling strategy. This data-driven approach enables dynamic adaptation of the system's core functions, addressing the insufficient adaptability of fixed models and strategies. Historical operational data includes sensor-collected data and early warning data; anomaly handling records include the scheduler and processing time. The optimization algorithm uses a random forest to optimize the anomaly detection model, taking six indicators, including the degree of weight mutation, as input and outputting the early warning level. Five-fold cross-validation is used to adjust the tree depth to 15 layers, improving classification accuracy. Deep reinforcement learning is employed to optimize the scheduling strategy, using processing time and channel occupancy as reward functions, iterating 10,000 times to train the agent and generate a scheduling scheme. For example, before optimization, the anomaly detection model had a 15% misclassification rate for minor weight mutations and slight trajectory deviations. After training with historical data, the model adjusts the feature weights for these samples, decreasing the weight of weight mutation from 0.4 to 0.35 and increasing the weight of trajectory deviation from 0.3 to 0.35, reducing the misclassification rate to 5%. After optimizing the scheduling strategy, the average processing time for level three early warnings is shortened. Machine learning algorithms can extract patterns from historical data regarding changes in operating conditions (such as extended periods of traffic congestion) and automatically adjust model parameters and strategy logic, enabling the system to adapt to new operating conditions without manual intervention. Technically, this step improves the accuracy of anomaly detection and enhances the overall efficiency of scheduling strategies.

[0114] Mobile monitoring devices enable real-time reception of early warning information, viewing of processing instructions, and feedback of processing results, ensuring timely handling of anomalies. This breaks down the spatial limitations of fixed terminals, enabling full-scenario response to anomalies and resolving response delays. The mobile monitoring devices utilize industrial-grade tablets (supporting IP65 protection) and communicate with the central system via a dedicated 4G network, with data transmission latency ≤1 second. Functional modules include: early warning reception (a pop-up displaying a level-three early warning, "X123 experiences a 2% weight change at 50 meters on the second floor"), instruction viewing, and result feedback. For example, when maintenance personnel are patrolling the first floor, if the mobile device receives a level-three early warning from X123, they can immediately view the navigation to "re-weigh at 50 meters on the second floor" in the instruction, process the issue, and upload a photo of the re-weighing. The entire process reduces response time. Mobile devices free operators from the constraints of the monitoring room, allowing them to participate in anomaly handling in real-time from any location, such as the corridor or maintenance room, eliminating response delays caused by spatial barriers. This device can shorten the average response time for anomaly handling and reduce delay rates.

[0115] A key performance indicator (KPI) system is constructed, including indicators such as anomaly detection timeliness, processing completion rate, and data integrity. System performance evaluation reports are generated regularly to quantify the system's operational status, providing a clear basis for optimization and addressing the lack of quantitative standards in traditional evaluations. Technically, anomaly detection timeliness is defined as "the time difference between anomaly occurrence and system warning" (target ≤ 1 second); processing completion rate is "the number of anomalies processed on time / the total number of anomalies" (target ≥ 95%); and data integrity is "the percentage of complete data chains" (target ≥ 99%). Indicator data comes from sensor timestamps, warning times, processing records, and data chain status. Evaluation reports are generated weekly, using radar charts to display the deviation between actual and target values ​​for each indicator. For example, if the actual anomaly detection timeliness is 1.2 seconds, the deviation is +0.2 seconds, and areas for improvement that did not meet the target are indicated, such as "optimizing the sensor sampling frequency to 0.5 seconds / time." For instance, the report for the first week of October shows a processing completion rate of 92% (below 95%) because the Level 3 warning processing timed out 3 times, suggesting "adding one backup TV vehicle." Quantitative indicators make system performance measurable and comparable. Reports, through deviation analysis, pinpoint weaknesses (such as insufficient resources causing substandard completion rates), providing precise directions for subsequent optimization. This system enhances the interpretability of system performance and improves the effectiveness of targeted improvement measures.

[0116] Through automated diagnostic algorithms, sensor status, network connection quality, and system operating status are regularly monitored. Potential problems are identified and warned in advance to ensure stable system operation. This allows for early identification of potential faults, preventing sudden downtime and addressing the limitations of manual inspections. Sensor status monitoring compares current accuracy with initial accuracy, issuing warnings when the deviation exceeds 50%. Network connection quality monitoring tracks the number of data transmission failures within one hour (threshold 5), issuing warnings when these thresholds are exceeded. System operating status monitoring tracks CPU usage (threshold 80%) and memory usage (threshold 70%), issuing warnings when these thresholds are exceeded. The automated diagnostic algorithm executes daily at 2 AM, pushing warning information to mobile monitoring devices (e.g., "Weighing sensor error at 30 meters on the second floor is 0.4 kg, calibration recommended"). For example, if the algorithm detects 6 data transmission failures at the location beacon at the second-floor corner (exceeding the 5-failure threshold), an immediate warning is sent. Maintenance personnel can then calibrate the beacon that day, preventing positioning failure the following day (potentially causing 2 hours of downtime). By continuously monitoring key parameters of equipment and the system, the automated diagnostic algorithm identifies degradation trends before faults occur, triggering maintenance early and shifting from reactive emergency repairs to proactive prevention. This algorithm can reduce the average downtime due to sudden system failures, and also reduce equipment maintenance costs.

[0117] In one embodiment, the present invention also discloses an ETV contactless re-weighing system based on multi-level verification, comprising:

[0118] Data acquisition module 1 is used to deploy a sensor network in the ETV lane to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard.

[0119] The generation module 2 is used to analyze weight mutation, trajectory deviation and timeliness deviation indicators based on sensor data fusion, generate graded early warnings using an adaptive threshold model, and transmit them to the central system in real time.

[0120] The scheduling module 3 is used to dynamically generate TV vehicle scheduling plans based on the warning level and site conditions, plan the optimal route through a multi-objective optimization algorithm, and send navigation instructions to the vehicle terminal.

[0121] Module 4 is established to create a full-process data chain system, enabling data integrity through anomaly detection and processing, ensuring data chain continuity through an automatic repair mechanism, and providing traceability and query functions.

[0122] Diagnostic module 5 is used to optimize detection models and scheduling strategies based on historical data through machine learning, periodically evaluate system performance, and self-diagnose the operating status.

[0123] In one embodiment, the scheduling module includes:

[0124] The optimization unit is used to establish a mathematical model that includes multiple optimization objectives, such as minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs.

[0125] The first generation unit is used to dynamically generate the optimal processing path based on real-time site monitoring data, avoiding congested areas and peak hours;

[0126] The second generation unit is used to automatically generate a scheduling scheme containing complete processing instructions. The scheduling scheme includes path navigation information, processing time limit requirements, and exception handling guidelines.

[0127] The tracking unit is used to track the execution of scheduling instructions in real time and automatically issue warnings for execution deviations.

[0128] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described ETV seamless re-weighing method based on multi-level verification.

[0129] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described ETV seamless re-weighting method based on multi-level verification.

[0130] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0131] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0132] The above description is merely a preferred embodiment of the present invention and does not limit the scope of this application. Any equivalent results or equivalent process transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of protection of this application.

Claims

1. A method for seamless ETV re-weighing based on multi-level verification, characterized in that, Includes the following steps: Deploy a sensor network in the ETV lane to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through dynamic dashboards. Based on sensor data fusion analysis of indicators such as weight mutation, trajectory deviation and timeliness deviation, an adaptive threshold model is used to generate graded early warnings, which are then transmitted to the central system in real time. TV vehicle dispatching plan is dynamically generated based on the warning level and site conditions. The optimal route is planned through a multi-objective optimization algorithm, and navigation instructions are sent to the vehicle terminal. Establish a full-process data chain system to ensure data integrity from anomaly detection to processing completion, ensure data chain continuity through automatic repair mechanisms, and provide traceability and query functions; The system optimizes detection models and scheduling strategies based on historical data through machine learning, regularly assesses system performance, and self-diagnoses its operational status.

2. The ETV seamless re-weighing method based on multi-level verification according to claim 1, characterized in that, The steps of deploying a sensor network in the ETV lane to continuously collect container weight and location data, construct a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard include: Weight monitoring points are set up at fixed intervals along the ETV lane, and each monitoring point is equipped with a set of weighing sensors of specific accuracy. Position beacons are added at the turns of the lane and the connection between floors. Establish a sensor data acquisition network and transmit real-time data to a central server via a wireless network; Based on the collected real-time data, a digital trajectory for container transportation is constructed, including three dimensions: weight-time curve, location-time trajectory, and speed-time curve. A transportation status assessment model is also established to calculate the stability index of the transportation process in real time. The design incorporates a multi-level monitoring dashboard. The first level displays the overall transportation status, the second level displays detailed data for individual containers, and the third level displays abnormal warning information.

3. The ETV seamless re-weighing method based on multi-level verification according to claim 1, characterized in that, The steps of analyzing weight mutations, trajectory deviations, and timeliness deviations based on sensor data fusion, generating graded early warnings using an adaptive threshold model, and transmitting them to the central system in real time include: A weighted fusion method is used to calculate the comprehensive anomaly index, where the weight coefficients are obtained by training based on historical data and an adaptive adjustment mechanism is set up to establish a multi-level early warning system with different early warning levels. Based on machine learning algorithms, the thresholds of each indicator are dynamically adjusted according to historical anomaly data, environmental factors, and operator skill level. An efficient data compression protocol is adopted, and an early warning confirmation mechanism is designed. If the recipient fails to confirm receipt of the early warning within a preset time, the early warning level will be automatically upgraded and transmitted to the central system in real time.

4. The ETV seamless re-weighing method based on multi-level verification according to claim 1, characterized in that, The steps of dynamically generating a TV vehicle dispatch plan based on the warning level and site conditions, planning the optimal route through a multi-objective optimization algorithm, and issuing navigation instructions to the vehicle terminal include: Establish a mathematical model that includes multiple optimization objectives, including minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs; Based on real-time site monitoring data, the optimal processing route is dynamically generated to avoid congested areas and peak hours. Automatically generate a scheduling scheme containing complete processing instructions, the scheduling scheme including path navigation information, processing time limit requirements and exception handling guidelines; It tracks the execution of scheduling instructions in real time and automatically issues warnings for execution deviations.

5. The ETV seamless re-weighing method based on multi-level verification according to claim 1, characterized in that, The steps of establishing a full-process data chain system to achieve data integrity upon anomaly detection and processing, ensuring data chain continuity through an automatic repair mechanism, and providing traceability and query functions include: Establish data quality inspection rules to automatically mark and repair abnormal data; When a data transmission interruption or loss is detected, the data reconstruction program is automatically started, and the complete data chain is restored through three steps: data retransmission, data interpolation processing, and data verification. The intelligent query interface allows for multi-dimensional queries based on container number, time range, and anomaly type. Based on historical anomaly data, an optimization suggestion report is automatically generated, which includes sensor deployment optimization, threshold adjustment suggestions, and scheduling strategy improvements.

6. The ETV seamless re-weighing method based on multi-level verification according to claim 1, characterized in that, The steps of optimizing the detection model and scheduling strategy based on historical data through machine learning, and periodically evaluating system performance and self-diagnosing its operating status include: Based on historical operation data and anomaly handling records, machine learning algorithms are used to optimize the anomaly detection model and scheduling strategy. Mobile monitoring devices enable real-time reception of early warning information, viewing of processing instructions, and feedback of processing results, ensuring timely handling of anomalies. Construct a key performance indicator system, which includes anomaly detection timeliness, processing completion rate, and data integrity, and generate system performance evaluation reports regularly; Through automatic diagnostic algorithms, the status of sensors, network connection quality, and system operation are regularly checked to detect potential problems and provide early warnings.

7. An ETV contactless re-weighing system based on multi-level verification, characterized in that, include: The data acquisition module is used to deploy a sensor network in the ETV lane to continuously collect container weight and location data, build a digital transportation trajectory and detect abnormal fluctuations in real time, and visualize the transportation status through a dynamic dashboard. The generation module is used to analyze weight mutation, trajectory deviation and timeliness deviation indicators based on sensor data fusion, generate graded early warnings using an adaptive threshold model, and transmit them to the central system in real time. The scheduling module is used to dynamically generate TV vehicle scheduling plans based on the warning level and site conditions, plan the optimal route through a multi-objective optimization algorithm, and send navigation instructions to the vehicle terminal. The module is used to establish a full-process data chain system, realize data integrity after anomaly detection and processing, ensure data chain continuity through an automatic repair mechanism, and provide traceability and query functions; The diagnostic module is used to optimize detection models and scheduling strategies based on historical data through machine learning, periodically evaluate system performance, and self-diagnose the operating status.

8. The ETV contactless re-weighing system based on multi-level verification according to claim 7, characterized in that, The scheduling module includes: The optimization unit is used to establish a mathematical model that includes multiple optimization objectives, such as minimizing anomaly handling time, minimizing channel occupancy impact, and minimizing operating costs. The first generation unit is used to dynamically generate the optimal processing path based on real-time site monitoring data, avoiding congested areas and peak hours; The second generation unit is used to automatically generate a scheduling scheme containing complete processing instructions. The scheduling scheme includes path navigation information, processing time limit requirements, and exception handling guidelines. The tracking unit is used to track the execution of scheduling instructions in real time and automatically issue warnings for execution deviations.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images