Whole vehicle type automatic weighing system based on dynamic transportation process
By collecting deformation angle data in real time and automatically identifying it through the dynamic benchmark management module, combined with load calculation and edge computing, the accuracy drift and data silo problems of vehicle load detection are solved, achieving high-precision, maintenance-free dynamic monitoring.
Patent Information
- Application Number
- CN202610986554.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-03
- Publication Date
- 2026-08-04
AI Technical Summary
Existing vehicle load detection technologies suffer from several drawbacks: measurement accuracy is easily affected by changes in the environment and vehicle condition; the calibration process is complex; data silos are common; and it is difficult to achieve high precision, maintenance-free operation, adaptive measurement, and dynamic monitoring.
The system uses a data acquisition unit to collect deformation angle data in real time, and combines it with a dynamic reference management module to automatically identify the unloaded static state. The load weight is calculated in real time by the load calculation module, and edge computing is enabled for local storage when communication is interrupted. A multi-axis load model is constructed and attitude compensation is performed to generate work records.
It achieves high-precision, maintenance-free, adaptive measurement for vehicle load detection, solves the long-term accuracy drift problem, ensures measurement accuracy and data continuity, and supports dynamic management throughout the entire process.
Smart Images

Figure CN122505385A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of vehicle load detection technology, specifically relating to a whole-vehicle automatic weighing system based on dynamic transportation processes. Background Technology
[0002] Currently, vehicle load detection mainly relies on fixed-point static weighing using weighbridges. This method requires vehicles to drive into a dedicated area, stop, and repeatedly enter and exit the weighbridge, which has limitations such as low efficiency, inability to obtain real-time load data, and difficulty in meeting the dynamic monitoring needs of the entire logistics and sanitation operation process.
[0003] To overcome the shortcomings of static weighing, vehicle-mounted weighing technology has emerged. Existing solutions typically involve installing displacement or strain gauges in the vehicle's suspension system (such as leaf springs) to indirectly calculate the load by measuring deformation. However, the following technical bottlenecks still exist in practical applications: Firstly, measurement accuracy is easily affected by changes in the environment and the vehicle's own condition. Sensors are susceptible to temperature drift and zero drift; leaf springs experience fatigue and creep after long-term use; and factors such as changes in tire pressure and driving on slopes or bumpy roads can all cause measurement parameters to drift. Existing systems mostly undergo a single static calibration before leaving the factory or during installation, with a fixed no-load reference value. They lack an effective self-correction mechanism and cannot adapt to parameter changes caused by sensor drift, suspension fatigue, etc. After long-term use, accuracy significantly decreases, requiring frequent returns to the factory or manual recalibration, resulting in high maintenance costs.
[0004] Secondly, the calibration process is complex and dependent on specific conditions. Traditional calibration requires vehicles to enter the weighbridge multiple times under various operating conditions, such as empty and fully loaded, and then manually record data and fit curves. This operation is cumbersome and difficult to promote on a large scale in a fleet. In particular, for special vehicles such as sanitation vehicles and engineering vehicles, which operate in complex environments and frequently switch between empty and fully loaded states, existing technologies are unable to achieve seamless integration of source measurement and dynamic monitoring.
[0005] Third, data silos are prevalent. Most vehicle-mounted weighing systems only provide local display or simple recording, lacking deep integration with business management systems, and thus failing to achieve centralized monitoring, automatic analysis, and coordinated scheduling of vehicle load data.
[0006] Therefore, how to achieve high-precision, maintenance-free, and adaptive weighing of vehicle-mounted weighing systems and integrate them into the entire dynamic management process is a technical problem that urgently needs to be solved in this field. Summary of the Invention
[0007] To address the aforementioned problems in the existing technology, this invention provides a vehicle-mounted automatic weighing system based on a dynamic transportation process. The objective of this invention can be achieved through the following technical solution: A vehicle-mounted automated weighing system based on a dynamic transportation process includes: The data acquisition unit is used to collect deformation angle data installed at various suspension system measuring points in real time during vehicle operation, and generate timestamps and vehicle identifiers. The data processing and communication unit is used to preprocess and cache the collected data locally, obtain the vehicle location and operating conditions, upload the data to the backend server, and enable local storage to resume transmission when communication is interrupted. The back-end service and computing unit includes: The dynamic reference management module is used to automatically identify the time period when the vehicle is in an unloaded, stationary, and horizontal state according to the operating conditions, and store the deformation angle data collected during that time period as the dynamic unloaded reference value; when the deviation between the newly identified unloaded reference value and the system's initial reference value exceeds the threshold, the reference is automatically triggered to update the reference value and replace the original reference value with the current dynamic unloaded reference value. The load calculation module is used to calculate the vehicle load in real time based on real-time deformation angle data and dynamic no-load reference value, combined with the load calculation model. The work record generation module is used to associate the load weight with timestamps, locations, and vehicle identifiers to generate work records.
[0008] Specifically, identifying a vehicle as being in an unloaded, stationary, and horizontal state includes determining it to be in an unloaded, stationary, and horizontal state when the following conditions are met simultaneously: The longitudinal tilt angle and lateral tilt angle detected by the vehicle attitude sensor are both less than the preset horizontal threshold. The fluctuation range of the collected deformation angle data within the preset time window is less than the preset fluctuation threshold.
[0009] Specifically, the dynamic reference management module is further configured with a preset first time threshold and a second time threshold, wherein the second time threshold is greater than the first time threshold; the deformation angle data collected in the current time period is confirmed as a valid unloaded reference sampling point only after the above conditions are met continuously and the duration reaches the second time threshold; the arithmetic mean or median value of the valid unloaded reference sampling points for a preset number of consecutive times is used as the final dynamic unloaded reference value stored in the database.
[0010] Specifically, the determination method for the deviation exceeding the threshold includes two trigger modes: Mode 1: The absolute deviation between the dynamic no-load reference value and the initial reference value identified in a single instance exceeds the first preset deviation threshold, and this state occurs three times consecutively. Mode 2: The deviation between the sliding average of all dynamic no-load reference values identified within 24 consecutive hours and the initial reference value exceeds a second preset deviation threshold, wherein the second preset deviation threshold is less than the first preset deviation threshold; When any of the patterns is satisfied, the baseline update process is triggered.
[0011] Specifically, the benchmark update process includes: Write the current dynamic no-load baseline value to the baseline history table and mark the update timestamp; Calculate the difference between the new benchmark value and the previous version benchmark value. If the difference exceeds the third preset deviation threshold, wherein the third preset deviation threshold is greater than the first preset deviation threshold, a benchmark change warning message is generated to prompt manual inspection of the sensor and suspension system status. The real-time reference pointer used in the load calculation module is set to the new reference value, and all subsequent load calculations are based on the new reference value.
[0012] Specifically, the background service and calculation unit also includes a benchmark rollback module, which is used to continuously monitor the load calculation error when a vehicle passes through an external weighbridge during a preset observation period after the benchmark is updated; if the weighbridge comparison error exceeds a preset accuracy threshold for a preset number of consecutive times, the benchmark will be automatically rolled back to the previous version and the benchmark update function will be locked until manual confirmation.
[0013] Specifically, the load calculation model is constructed as follows: Collect multi-axis deformation angle data and corresponding standard weight values of vehicles passing over weighbridges under different load conditions to construct a training dataset; Using the deformation angle of each axis as the input feature and the standard weight value as the output label, the load sub-model of each axis is trained by using multivariate nonlinear regression, support vector regression, or lightweight neural network algorithm. The outputs of each axle load sub-model are summed to obtain the overall vehicle load calculation model, and each axle sub-model is assigned a dynamically adjustable confidence weight.
[0014] Specifically, the load calculation module also includes an attitude compensation submodule, which is used to perform projection transformation or coordinate system rotation on the deformation angle data based on the real-time attitude data of the vehicle, convert the measured value in the non-horizontal state to the equivalent value in the horizontal state, and then input the compensated data into the load calculation model.
[0015] Specifically, the data processing and communication unit also includes an edge computing submodule, which is deployed on the vehicle terminal side and has a lightweight version of the load calculation model built in. When the communication network is interrupted for a long time or the backend server is unavailable, the edge computing submodule calculates the vehicle load weight in real time based on the locally cached dynamic empty load benchmark value and stores the calculation results locally. After the communication network is restored, the local calculation results and the original collected data are uploaded to the backend server for consistency verification and data supplementation.
[0016] Specifically, the work record generation module also generates electronic fence matching information: the backend server has preset geographic coordinate ranges of electronic fences for multiple work points; when it receives the load data and location information uploaded by the vehicle, it automatically determines whether the location falls within any electronic fence; if it does, the corresponding work point name, fence number and falling time are appended to the work record, and the work is automatically collected into the report of the corresponding point; if it does not fall within any electronic fence, it is marked as a non-fixed-point work and classified separately.
[0017] Specifically, the data acquisition unit is equipped with at least two angle sensors at each suspension system measuring point to form a redundant backup. When the main sensor data is abnormal or communication is interrupted, it automatically switches to the redundant sensor for data acquisition and generates sensor fault alarm information.
[0018] Specifically, the backend server is also used to: automatically obtain the standard weight value of the corresponding time node from the external business system when a vehicle is detected passing through the external weighbridge, and bind the standard weight value with the deformation angle data collected at the same time as a training sample for the load calculation model.
[0019] The beneficial effects of this invention are as follows: Dynamic no-load reference adaptive update solves the problem of long-term accuracy drift: Existing vehicle-mounted weighing systems mostly use a one-time static calibration before leaving the factory, with a fixed no-load reference value. After long-term use, factors such as sensor temperature drift, zero drift, leaf spring fatigue, and tire pressure changes cause measurement parameters to drift, resulting in a significant decrease in accuracy. This invention innovatively introduces a dynamic reference management module, which automatically identifies periods of no-load static horizontal state based on multi-dimensional working condition information such as vehicle speed, lifting mechanism angle, and vehicle posture, and stores the deformation angle data collected during these periods as dynamic no-load reference values. When the deviation between the newly identified reference value and the initial reference value exceeds a threshold, the system automatically triggers a reference update, replacing the original reference value with the current reference value for subsequent load calculations. This mechanism is equivalent to the system automatically performing "zero calibration" in every no-load state, promptly eliminating accumulated errors, fundamentally ensuring long-term measurement accuracy, and requiring no manual intervention throughout the process, significantly reducing maintenance costs.
[0020] Dual-mode triggering and benchmark rollback ensure update reliability and system safety: To address different causes of benchmark drift, this invention employs two triggering modes: Mode 1 targets sudden, large jumps (such as sensor damage or broken leaf springs), triggering after three consecutive instances of exceeding the tolerance; Mode 2 targets gradual, slow drifts (such as component aging or wear), triggering with a 24-hour moving average deviation, and ensuring the second threshold is lower than the first threshold, thus sensitively capturing slow changes. After the update, an observation period is initiated, verifying the update effect through comparison with weighbridges. If multiple consecutive errors exceed the tolerance, the benchmark rollback module automatically reverts to the previous version and locks the update function, forming a complete fault-tolerance mechanism to prevent erroneous benchmarks from affecting measurement accuracy and ensuring system operational safety.
[0021] Multi-axis load model and attitude compensation improve measurement accuracy in complex working conditions: This invention constructs a load sub-model with the deformation angle of each axle as input and the standard weight value as output. The outputs of each axle are summed and dynamically assigned confidence weights. Compared to a single measurement point scheme, this more comprehensively reflects the load distribution of the entire vehicle, improving measurement accuracy. Simultaneously, the attitude compensation sub-module projects the deformation angle based on real-time vehicle tilt data, converting non-horizontal state measurements into equivalent horizontal values. This effectively eliminates measurement errors caused by slopes and inclined road surfaces, significantly improving measurement accuracy under complex road conditions.
[0022] Edge computing operates in a degraded manner to ensure data continuity in network blind spots: The vehicle terminal has a built-in edge computing submodule. When the communication network is interrupted for a long time or the background is unavailable, it can calculate the vehicle's load weight in real time based on the locally cached dynamic empty load benchmark value and store it locally. After the network is restored, it automatically uploads the data and performs consistency verification and data replenishment. This design solves the problem of data loss in remote road sections and signal blind spots, ensuring that the system can still maintain basic metering functions under extreme conditions, and that the data is complete and traceable. Attached Figure Description
[0023] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.
[0024] Figure 1 This is a schematic diagram of the overall system architecture of a vehicle-mounted automatic weighing system based on a dynamic transportation process according to the present invention. Figure 2 This is a flowchart of the dynamic benchmark update process of the present invention; Figure 3 This is a schematic diagram of the electronic fence matching and operation record collection of the present invention. Detailed Implementation
[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.
[0026] Please see Figures 1-3 A vehicle-mounted automatic weighing system based on a dynamic transportation process includes: The data acquisition unit is used to collect deformation angle data installed at various suspension system measuring points in real time during vehicle operation, and generate timestamps and vehicle identifiers. The data processing and communication unit is used to preprocess and cache the collected data locally, obtain the vehicle location and operating conditions, upload the data to the backend server, and enable local storage to resume transmission when communication is interrupted. The back-end service and computing unit includes: The dynamic reference management module is used to automatically identify the time period when the vehicle is in an unloaded, stationary, and horizontal state according to the operating conditions, and store the deformation angle data collected during that time period as the dynamic unloaded reference value; when the deviation between the newly identified unloaded reference value and the system's initial reference value exceeds the threshold, the reference is automatically triggered to update the reference value and replace the original reference value with the current dynamic unloaded reference value. The load calculation module is used to calculate the vehicle load in real time based on real-time deformation angle data and dynamic no-load reference value, combined with the load calculation model. The work record generation module is used to associate the load weight with timestamps, locations, and vehicle identifiers to generate work records.
[0027] Specifically, identifying a vehicle as being in an unloaded, stationary, and horizontal state includes determining it to be in an unloaded, stationary, and horizontal state when the following conditions are met simultaneously: The longitudinal tilt angle and lateral tilt angle detected by the vehicle attitude sensor are both less than the preset horizontal threshold. The fluctuation range of the collected deformation angle data within the preset time window is less than the preset fluctuation threshold.
[0028] Specifically, the dynamic reference management module is further configured with a preset first time threshold and a second time threshold, wherein the second time threshold is greater than the first time threshold; the deformation angle data collected in the current time period is confirmed as a valid unloaded reference sampling point only after the above conditions are met continuously and the duration reaches the second time threshold; the arithmetic mean or median value of the valid unloaded reference sampling points for a preset number of consecutive times is used as the final dynamic unloaded reference value stored in the database.
[0029] Specifically, the determination method for the deviation exceeding the threshold includes two trigger modes: Mode 1: The absolute deviation between the dynamic no-load reference value and the initial reference value identified in a single instance exceeds the first preset deviation threshold, and this state occurs three times consecutively. Mode 2: The deviation between the sliding average of all dynamic no-load reference values identified within 24 consecutive hours and the initial reference value exceeds a second preset deviation threshold, wherein the second preset deviation threshold is less than the first preset deviation threshold; When any of the patterns is satisfied, the baseline update process is triggered.
[0030] Specifically, the benchmark update process includes: Write the current dynamic no-load baseline value to the baseline history table and mark the update timestamp; Calculate the difference between the new benchmark value and the previous version benchmark value. If the difference exceeds the third preset deviation threshold, wherein the third preset deviation threshold is greater than the first preset deviation threshold, a benchmark change warning message is generated to prompt manual inspection of the sensor and suspension system status. The real-time reference pointer used in the load calculation module is set to the new reference value, and all subsequent load calculations are based on the new reference value.
[0031] Specifically, the background service and calculation unit also includes a benchmark rollback module, which is used to continuously monitor the load calculation error when a vehicle passes through an external weighbridge during a preset observation period after the benchmark is updated; if the weighbridge comparison error exceeds a preset accuracy threshold for a preset number of consecutive times, the benchmark will be automatically rolled back to the previous version and the benchmark update function will be locked until manual confirmation.
[0032] Specifically, the load calculation model is constructed as follows: Collect multi-axis deformation angle data and corresponding standard weight values of vehicles passing over weighbridges under different load conditions to construct a training dataset; Using the deformation angle of each axis as the input feature and the standard weight value as the output label, the load sub-model of each axis is trained by using multivariate nonlinear regression, support vector regression, or lightweight neural network algorithm. The outputs of each axle load sub-model are summed to obtain the overall vehicle load calculation model, and each axle sub-model is assigned a dynamically adjustable confidence weight.
[0033] Specifically, the load calculation module also includes an attitude compensation submodule, which is used to perform projection transformation or coordinate system rotation on the deformation angle data based on the real-time attitude data of the vehicle, convert the measured value in the non-horizontal state to the equivalent value in the horizontal state, and then input the compensated data into the load calculation model.
[0034] Specifically, the data processing and communication unit also includes an edge computing submodule, which is deployed on the vehicle terminal side and has a lightweight version of the load calculation model built in. When the communication network is interrupted for a long time or the backend server is unavailable, the edge computing submodule calculates the vehicle load weight in real time based on the locally cached dynamic empty load benchmark value and stores the calculation results locally. After the communication network is restored, the local calculation results and the original collected data are uploaded to the backend server for consistency verification and data supplementation.
[0035] Specifically, the work record generation module also generates electronic fence matching information: the backend server has preset geographic coordinate ranges of electronic fences for multiple work points; when it receives the load data and location information uploaded by the vehicle, it automatically determines whether the location falls within any electronic fence; if it does, the corresponding work point name, fence number and falling time are appended to the work record, and the work is automatically collected into the report of the corresponding point; if it does not fall within any electronic fence, it is marked as a non-fixed-point work and classified separately.
[0036] Specifically, the data acquisition unit is equipped with at least two angle sensors at each suspension system measuring point to form a redundant backup. When the main sensor data is abnormal or communication is interrupted, it automatically switches to the redundant sensor for data acquisition and generates sensor fault alarm information.
[0037] Specifically, the backend server is also used to: automatically obtain the standard weight value of the corresponding time node from the external business system when a vehicle is detected passing through the external weighbridge, and bind the standard weight value with the deformation angle data collected at the same time as a training sample for the load calculation model.
[0038] Example This embodiment uses a garbage collection truck of a municipal sanitation group as an example to provide a detailed description of the vehicle-mounted automatic weighing system based on dynamic transportation process described in this invention.
[0039] 1. System Composition and Installation This system consists of three parts: a data acquisition unit, a data processing and communication unit, and a back-end service and computing unit.
[0040] Data Acquisition Unit: Depending on the vehicle model, 4 to 6 tilt sensors are installed between the leaf springs of each axle and the frame beam (e.g., 4 for 4x2 models and 6 for 6x4 models). The sensors collect suspension deformation angle data in real time, simultaneously generating timestamps and identifying information such as the vehicle's VIN and license plate number. To improve reliability, each measuring point can be configured with redundant sensors (i.e., two sensors are set up at each installation location as backups). When the primary sensor data is abnormal or communication is interrupted, the system automatically switches to the redundant sensor at the same measuring point and generates a sensor fault alarm message, which is pushed to the management platform.
[0041] Data processing and communication unit: Installed in the driver's cab, this unit integrates a signal acquisition module, a Beidou positioning module, a 4G / 5G communication module, a local encrypted storage chip, and an edge computing submodule. This unit filters and preprocesses multi-channel angle signals to remove noise, acquires real-time vehicle position and attitude (longitudinal tilt angle, lateral tilt angle) and other operating condition information, and uploads data packets to the backend server. When the vehicle enters a communication dead zone (such as an underground parking lot or mountainous area), the data is automatically stored locally and transmitted sequentially once the network is restored.
[0042] Backend service and computing unit: Deployed on a cloud server, it includes a dynamic baseline management module, a load calculation module (including an attitude compensation sub-module), a job record generation module, a baseline rollback module, and an electronic fence matching engine.
[0043] 2. Identification of Unloaded Static Horizontal State and Establishment of Dynamic Benchmark The system automatically identifies when a vehicle is unloaded, stationary, and level by the following methods: The vehicle attitude sensor detects that both the longitudinal tilt angle and the lateral tilt angle are less than a preset horizontal threshold (set to ≤3° in this embodiment); Meanwhile, if the fluctuation amplitude of the collected deformation angle data within a preset time window (e.g., 2 consecutive seconds) is less than the preset fluctuation threshold (e.g., the maximum deviation is less than 0.5°), it indicates that the vehicle is stationary and has no load change.
[0044] Once the above conditions are met, the system initially determines the system to be in an unloaded, stationary state. To prevent brief, unstable interference, the system also sets a first time threshold (3 seconds) and a second time threshold (10 seconds, greater than the first time threshold). Only when the above conditions are met consecutively and the duration reaches the second time threshold (10 seconds) is the deformation angle data collected in the current time period confirmed as a valid unloaded reference sampling point. After collecting 5 consecutive valid sampling points, the arithmetic mean is taken as the final dynamic unloaded reference value stored in the database, thereby filtering out interference such as road surface vibrations and sensor noise.
[0045] 3. Dynamic no-load baseline update and dual-mode triggering The system uses the factory static calibration value as the initial reference value, and the dynamic reference management module continuously monitors the reference deviation and automatically updates using two trigger modes: Mode 1 (Sudden Change Trigger): If the absolute deviation between the dynamic no-load reference value and the initial reference value exceeds the first preset deviation threshold (±3%) and occurs three times consecutively, it is determined that the sensor is abnormal or the suspension components have changed significantly, and an update is triggered immediately.
[0046] Mode 2 (Gradual Change Trigger): If the deviation between the sliding average of all dynamic no-load reference values over 24 consecutive hours and the initial reference value exceeds a second preset deviation threshold (±1.5%), it is determined to be a slow drift of the component (such as leaf spring fatigue or sensor aging), and an automatic update is triggered. The second threshold is lower than the first threshold to ensure timely response to gradual drift.
[0047] The baseline update process is as follows: Write the new baseline value to the baseline history table and mark the update timestamp; Calculate the difference between the new benchmark value and the previous version benchmark value. If the difference exceeds the third preset deviation threshold (±5%, which is greater than the first threshold), generate a "benchmark change warning" message and push it to the administrator's mobile phone through the background to prompt manual inspection of the sensor and suspension system. The real-time reference pointer used in the load calculation module is set to the new reference value, and all subsequent load calculations are based on the new reference value.
[0048] 4. Baseline rollback protection mechanism After the baseline update, a preset observation period (24 hours) begins. During this period, the system continuously monitors the load calculation error when vehicles pass over the weighbridge: the background automatically obtains the standard weight value of the weighbridge from the sanitation logistics system and compares it with the real-time calculated value of the system. If the weighbridge comparison error exceeds the preset accuracy threshold (±2%) for 5 consecutive times, the baseline update is deemed abnormal. The baseline rollback module automatically restores the baseline to the previous version and locks the automatic update function until manual confirmation and unlocking by the management personnel. For example, if after an update, the weighbridge comparison errors for three consecutive times are 2.3%, 2.5%, and 2.2%, all exceeding the threshold, the system executes a rollback and issues an alarm. After investigation, it was found that the sensor installation was loose, causing the data abnormality. After manual tightening, recalibration was performed.
[0049] 5. Load calculation model construction and attitude compensation Static calibration before leaving the factory: After the angle sensors are installed, the vehicle needs to undergo weight calibration before leaving the factory. The vehicle is placed on a horizontal platform, and a special calibration device is used to apply simulated loads of different pressures to each axle of the vehicle (for example, applying pressure values corresponding to no load, 25% load, 50% load, 75% load, and full load in sequence) to simulate the actual state of the vehicle under different load conditions. The calibration device outputs pressure values in real time, and at the same time, the angle sensors collect deformation angle data at each measuring point. The system matches and records different pressure values (i.e., simulated load weight) with the corresponding angle values. Through multiple sets of pressure-angle correspondences, the initial calibration of the entire vehicle's equipment is completed. Only after the calibration is qualified can the vehicle leave the factory.
[0050] Model Construction: Vehicles enter the weighbridge under various operating conditions at the test track, including empty, 25% load, 50% load, 75% load, and fully loaded, stopping for 30 seconds each time to collect data on the deformation angle of each axle and the corresponding standard weight value. The backend uses the deformation angle of each axle as input features and the standard weight value as the output label, employing a multivariate nonlinear regression or lightweight neural network algorithm to train a sub-model for the load of each axle. The outputs of each axle are then summed to obtain the overall vehicle load calculation model. Confidence weights are dynamically assigned based on the error performance of each axle sensor during training.
[0051] Attitude compensation: When a vehicle stops on a slope (such as a 5° inclined road surface), the attitude compensation submodule performs coordinate system rotation projection on the deformation angle data based on the real-time detected longitudinal tilt angle (e.g., 5.2°), converts the non-horizontal measurement value to the horizontal equivalent value, and then inputs it into the model for calculation, thereby eliminating the measurement error caused by the road surface tilt.
[0052] 6. Edge computing and offline metering The vehicle-mounted data processing and communication unit incorporates an edge computing submodule and a lightweight load calculation model (with model parameters reduced to 30% of the original model). When the vehicle enters a communication dead zone (such as mountainous areas or underground parking garages) and the network interruption exceeds a set time (e.g., 5 minutes), the edge computing submodule automatically switches to offline mode: based on the latest dynamic empty load baseline value cached locally, it independently calculates the vehicle's load weight in real time and encrypts and stores the calculation results in the local storage chip. After the vehicle returns to a network coverage area, the system automatically uploads the local calculation results and the original collected data to the backend. The backend performs timestamp alignment and hash value consistency verification. After confirming that there are no errors, the data is added to the database, ensuring that data from operations in remote road sections is not lost and is traceable.
[0053] 7. Electronic fence matching and operation record generation The backend server has pre-set geographic coordinates (e.g., a circular fence with a radius of 50 meters) for each work site (such as a residential area, transfer station, or processing plant). During operation, vehicles upload load data and BeiDou location information in real time, and the backend automatically determines whether the vehicle's location falls within any of the electronic fences. If the object falls into a fence, the name of the work site, the fence number, and the time of the fall will be automatically added to the work record, and the work will be collected into the daily / weekly / monthly report of the corresponding site (including the number of work, total net weight, average load, etc.). If it does not fall within any fence, it is marked as "non-fixed-point operation" and classified and counted separately.
[0054] Meanwhile, the backend server automatically obtains the standard weight value of each vehicle passing the weighbridge from the business system and binds it with the deformation angle data collected at that moment, using it as a continuous training sample for the load calculation model to achieve online incremental optimization of the model.
[0055] 8. System Workflow Summary After the vehicles are put into operation, the system continuously collects sensor data and uploads it to the cloud. In the initial stage, the backend establishes a dynamic empty load baseline value during idle periods and trains the load model using weighbridge data. During daily operation, the dynamic baseline management module automatically monitors baseline drift, triggers updates (abrupt or gradual changes) when necessary, and verifies the data after an observation period, rolling back in case of anomalies. When the vehicle is operating on a slope, attitude compensation automatically takes effect; when the network is interrupted, the edge computing submodule takes over the metering; all operation records are matched and collected with the electronic fence. Administrators can view vehicle load, baseline change history, abnormal alarms, and various statistical reports in real time through a web page or mobile client, realizing dynamic monitoring and accurate accounting of the entire waste collection and transportation process.
[0056] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.
Claims
1. A vehicle-mounted automatic weighing system based on a dynamic transportation process, characterized in that, include: The data acquisition unit is used to collect deformation angle data installed at various suspension system measuring points in real time during vehicle operation, and generate timestamps and vehicle identifiers. The data processing and communication unit is used to preprocess and cache the collected data locally, obtain the vehicle location and operating conditions, upload the data to the backend server, and enable local storage to resume transmission when communication is interrupted. The back-end service and computing unit includes: The dynamic reference management module is used to automatically identify the time period when the vehicle is in an unloaded, stationary, and horizontal state according to the operating conditions, and store the deformation angle data collected during that time period as the dynamic unloaded reference value; when the deviation between the newly identified unloaded reference value and the system's initial reference value exceeds the threshold, the reference is automatically triggered to update the reference value and replace the original reference value with the current dynamic unloaded reference value. The load calculation module is used to calculate the vehicle load in real time based on real-time deformation angle data and dynamic no-load reference value, combined with the load calculation model. The work record generation module is used to associate the load weight with timestamps, locations, and vehicle identifiers to generate work records.
2. The system according to claim 1, characterized in that, The identification of a vehicle being in an unloaded, stationary, and horizontal state specifically includes determining that a vehicle is in an unloaded, stationary, and horizontal state when the following conditions are met simultaneously: The longitudinal tilt angle and lateral tilt angle detected by the vehicle attitude sensor are both less than the preset horizontal threshold. The fluctuation range of the collected deformation angle data within the preset time window is less than the preset fluctuation threshold.
3. The system according to claim 2, characterized in that, The dynamic reference management module is also set with a preset first time threshold and a second time threshold, the second time threshold being greater than the first time threshold; only after the above conditions are met continuously and the duration reaches the second time threshold will the deformation angle data collected in the current time period be confirmed as a valid unloaded reference sampling point. The arithmetic mean or median of the valid no-load reference sampling points for a consecutive preset number of times is used as the final dynamic no-load reference value stored in the database.
4. The system according to claim 1, characterized in that, The determination method for deviations exceeding the threshold includes two trigger modes: Mode 1: The absolute deviation between the dynamic no-load reference value and the initial reference value identified in a single instance exceeds the first preset deviation threshold, and this state occurs three times consecutively. Mode 2: The deviation between the sliding average of all dynamic no-load reference values identified within 24 consecutive hours and the initial reference value exceeds a second preset deviation threshold, wherein the second preset deviation threshold is less than the first preset deviation threshold; When any of the patterns is satisfied, the baseline update process is triggered.
5. The system according to claim 4, characterized in that, The benchmark update process includes: Write the current dynamic no-load baseline value to the baseline history table and mark the update timestamp; Calculate the difference between the new benchmark value and the previous version benchmark value. If the difference exceeds the third preset deviation threshold, wherein the third preset deviation threshold is greater than the first preset deviation threshold, a benchmark change warning message is generated to prompt manual inspection of the sensor and suspension system status. The real-time reference pointer used in the load calculation module is set to the new reference value, and all subsequent load calculations are based on the new reference value.
6. The system according to claim 5, characterized in that, The background service and calculation unit also includes a benchmark rollback module, which is used to continuously monitor the load calculation error when a vehicle passes through an external weighbridge during a preset observation period after the benchmark is updated; if the weighbridge comparison error exceeds the preset accuracy threshold for a preset number of consecutive times, the benchmark will be automatically rolled back to the previous version and the benchmark update function will be locked until manual confirmation.
7. The system according to claim 1, characterized in that, The load calculation model is constructed as follows: Collect multi-axis deformation angle data and corresponding standard weight values of vehicles passing over weighbridges under different load conditions to construct a training dataset; Using the deformation angle of each axis as the input feature and the standard weight value as the output label, the load sub-model of each axis is trained by using multivariate nonlinear regression, support vector regression, or lightweight neural network algorithm. The outputs of each axle load sub-model are summed to obtain the overall vehicle load calculation model, and each axle sub-model is assigned a dynamically adjustable confidence weight.
8. The system according to claim 7, characterized in that, The load calculation module also includes an attitude compensation submodule, which is used to perform projection transformation or coordinate system rotation on the deformation angle data based on the real-time attitude data of the vehicle, convert the measured value in the non-horizontal state to the equivalent value in the horizontal state, and then input the compensated data into the load calculation model.
9. The system according to claim 1, characterized in that, The data processing and communication unit also includes an edge computing submodule, which is deployed on the vehicle terminal side and has a lightweight version of the load calculation model built in. When the communication network is interrupted for a long time or the backend server is unavailable, the edge computing submodule calculates the vehicle load weight in real time based on the locally cached dynamic empty load benchmark value and stores the calculation results locally. After the communication network is restored, the local calculation results and the original collected data are uploaded to the backend server for consistency verification and data supplementation.
10. The system according to claim 1, characterized in that, The work record generation module also generates electronic fence matching information: the backend server has preset geographic coordinate ranges of electronic fences for multiple work points; when it receives the load data and location information uploaded by the vehicle, it automatically determines whether the location falls within any electronic fence; if it does, the corresponding work point name, fence number and time of falling are appended to the work record, and the work is automatically collected into the report of the corresponding point; if it does not fall within any electronic fence, it is marked as a non-fixed-point work and classified separately.
11. The system according to claim 1, characterized in that, The data acquisition unit is equipped with at least two angle sensors at each suspension system measuring point to form a redundant backup. When the main sensor data is abnormal or communication is interrupted, it automatically switches to the redundant sensor for data acquisition and generates sensor fault alarm information.
12. The system according to claim 1, characterized in that, The backend server is also used to: automatically obtain the standard weight value of the corresponding time node from the external business system when a vehicle is detected passing through the external weighbridge, and bind the standard weight value with the deformation angle data collected at the same time as a training sample for the load calculation model.
