Level calibration and error control method and device for electronic truck scale, and medium

Abstract

The embodiment of the invention discloses a horizontal calibration and error control method and device for an electronic truck scale, and a medium, belongs to the technical field of electronic truck scales, and solves the problem that error calibration is difficult to carry out on the electronic truck scale timely and efficiently in the prior art. Comprising the following steps: acquiring pressure distribution data corresponding to a splicing interface of adjacent weighing platforms in the electronic truck scale, and acquiring three-dimensional inclination angle data corresponding to each section of weighing platform; based on a weighing signal sequence sent by a weighing sensor, a dynamic weighing track of the vehicle on the multi-section weighing platform is reconstructed; carrying out fusion analysis on the pressure distribution data and the three-dimensional inclination angle data along the dynamic weighing track to determine the distribution characteristics of the horizontal deviation on the spatial position of the weighing platform; and analyzing the overall horizontal deviation vector into a local deviation component associated with the corresponding weighing platform segment based on the distribution characteristics, correcting the original weighing value based on the local deviation component, and outputting a compensated final weight result.

Description

Technical Field

[0001] This application relates to the field of electronic truck scale technology, and in particular to a method, equipment and medium for level calibration and error control of electronic truck scales. Background Technology

[0002] Electronic truck scales, as large-scale weighing and measurement devices, are widely used in ports, logistics, and industrial and mining enterprises. For applications requiring temporary or frequent relocation of weighing stations, mobile, multi-section electronic truck scales are favored for their flexibility.

[0003] However, this movable, multi-section splicing structure also presents inherent technical challenges. Due to factors such as uneven ground, uneven settlement caused by long-term use, or vehicle impacts, level deviations can easily occur between the sections of the weighing platform and within the platform itself. These level deviations can severely affect the force state of the load cells, causing the weighing center to shift, thus introducing significant weighing errors and impacting the fairness and accuracy of the weighing process.

[0004] In existing technologies, the leveling of electronic truck scales mainly relies on manual measurement and mechanical leveling using a level at regular intervals. This method is highly dependent on the operator's experience, is inefficient, and often requires stopping the weighing operation before testing. In emergency situations, it is difficult to calibrate and control the error of the electronic truck scale in a timely and efficient manner. Summary of the Invention

[0005] This application provides a method, device, and medium for the level calibration and error control of electronic truck scales, which addresses the following technical problems: In the prior art, the level calibration of electronic truck scales heavily relies on the operator's experience, resulting in low efficiency. Furthermore, this method often requires stopping the weighing task before testing, making it difficult to calibrate and control the error of the electronic truck scale in a timely and efficient manner under emergency conditions.

[0006] The embodiments of this application adopt the following technical solutions: This application provides a method for level calibration and error control of an electronic truck scale. The method includes: acquiring pressure distribution data at the interface between adjacent weighing platforms in the electronic truck scale, and acquiring three-dimensional tilt angle data for each weighing platform section, during weighing operations; reconstructing the dynamic weighing trajectory of the vehicle on multiple weighing platforms based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale; spatiotemporally labeling the pressure distribution data and three-dimensional tilt angle data based on the dynamic weighing trajectory; fusing and analyzing the spatiotemporally labeled pressure distribution data and three-dimensional tilt angle data along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial location of the weighing platform; and based on the distribution characteristics, resolving the overall horizontal deviation vector into local deviation components associated with the corresponding weighing platform sections, correcting the original weighing value based on the local deviation components, and outputting the compensated final weight result.

[0007] In one implementation of this application, the dynamic weighing trajectory of a vehicle on a multi-section weighing platform is reconstructed based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale. Specifically, this includes: using an adaptive threshold detection algorithm to determine peak events generated by wheel axle passage in the weighing signal sequence, and determining the sensor number and timestamp corresponding to each peak event; determining reference peak events with an interval time difference less than a preset time threshold, merging and decoupling them into axle load events based on the waveform correlation of the sensor signals corresponding to the reference peak events, and correcting the timestamps of the axle load events; constructing a time-series chain based on the chronological order of the corrected timestamps, and determining the instantaneous speed of the vehicle based on the spatiotemporal information between the weighing sensors corresponding to consecutive events in the time-series chain, so as to verify the rationality of the time-series chain through the instantaneous speed; and reconstructing the dynamic weighing trajectory based on the spatial topological mapping relationship between the weighing sensors and the weighing platform segments, and the verified time-series chain.

[0008] In one implementation of this application, the dynamic weighing trajectory is reconstructed based on the spatial topological mapping relationship between the weighing sensor and the weighing platform segment, and the verified time-series chain. Specifically, this includes: mapping axle load events to two-dimensional trajectory points with timestamps based on the spatial topological mapping relationship and the verified time-series chain, and constructing a graph optimization model with the trajectory points as vertices; adding constraint edges between the vertices of the graph optimization model; wherein the constraint edges include at least one of uniform motion constraint based on time interval, minimum turning radius constraint, and fixed vehicle wheelbase constraint; globally optimizing the two-dimensional coordinates of the trajectory point vertices by solving the graph optimization model to obtain optimized trajectory points that meet the edge constraints; and performing curve fitting on the optimized trajectory points to generate a continuous two-dimensional planar trajectory function as the reconstructed dynamic weighing trajectory.

[0009] In one implementation of this application, spatiotemporal labeling of pressure distribution data and three-dimensional tilt angle data is performed based on a dynamic weighing trajectory. Specifically, this includes: establishing a spatiotemporal coordinate system with vehicle travel time as the line and the platform segment space as the surface, based on the dynamic weighing trajectory; determining the platform segments corresponding to the vehicle at different times based on the pressure distribution data collected at each moment in the spatiotemporal coordinate system, and determining the splicing interface spatial identifier associated with the pressure distribution data based on the platform segments; labeling the spatial identifier of the platform segment to which the three-dimensional tilt angle data belongs based on the installation position of each platform segment, and adding a vehicle load time phase identifier to the three-dimensional tilt angle data based on the vehicle's speed and acceleration state in the dynamic weighing trajectory; aligning the labeled pressure distribution data and three-dimensional tilt angle data according to the time series to generate a multi-source dataset with spatiotemporal labels.

[0010] In one implementation of this application, pressure distribution data and three-dimensional tilt angle data labeled with spatiotemporal tags are fused and analyzed along a dynamic weighing trajectory to determine the distribution characteristics of horizontal deviation at the spatial location of the weighing platform. Specifically, this includes: extracting observation data sequences composed of pressure distribution characteristics and three-dimensional tilt angle readings sequentially from a multi-source dataset along discrete spatiotemporal points of the dynamic weighing trajectory; generating corresponding reference data sequences based on vehicle axle load information corresponding to each discrete spatiotemporal point; obtaining multi-source residual sequences based on the residuals corresponding to the observation data sequences and reference data sequences at each discrete spatiotemporal point; determining feature waveform vectors based on the spatial evolution patterns corresponding to the multi-source residual sequences; constructing a spatial topology graph with weighing platform segments, splicing interfaces, and sensors as nodes and physical connection relationships as edges; loading the feature waveform vectors as initial features of the corresponding nodes into the spatial topology graph; inputting the spatial topology graph with the loaded initial features into a graph neural network; and outputting the distribution characteristics of horizontal deviation corresponding to each weighing platform segment through the graph neural network.

[0011] In one implementation of this application, based on distribution characteristics, the overall horizontal deviation vector is parsed into local deviation components associated with the corresponding weighing platform segments. Specifically, this includes: concatenating the distribution characteristics of the horizontal deviations corresponding to each weighing platform segment to generate an overall horizontal deviation vector; using the deviation severity index corresponding to the overall horizontal deviation vector as the source target, determining the contribution of each node to the generation of the overall horizontal deviation vector by using the gradient values ​​of the initial feature waveform vectors of each node corresponding to the source target in the spatial topology map; weighting the overall horizontal deviation vector based on the contribution to obtain the local deviation components associated with each weighing platform segment; and determining the local compensation coefficients used to correct the weighing sensor readings of the corresponding weighing platform segment based on each local deviation component.

[0012] In one implementation of this application, the original measured weight value is corrected based on the local deviation component, and a compensated final weight result is output. Specifically, this includes: at each sampling moment during the weighing process, determining the instantaneous force vector direction acting on the load-bearing points of each weighing sensor according to the dynamic weighing trajectory and the preset vehicle load distribution model; determining the equivalent correction value of the original measured weight value under the preset virtual horizontal reference plane and the same force direction based on the original reading, local deviation component, and instantaneous force vector direction of each weighing sensor at each sampling moment; and matching the equivalent correction values ​​of all weighing sensors at all sampling moments according to the time sequence to obtain the compensated final weight result.

[0013] In one implementation of this application, the equivalent correction values ​​of all weighing sensors at all sampling times are matched according to the time series to obtain the compensated final weight result. Specifically, this includes: generating multiple sets of candidate vehicle parameters based on the original measured weight values ​​and dynamic weighing trajectories corresponding to the weighing sensors; instantiating multiple digital twins carrying different candidate parameters based on local deviation components to generate a parallel simulation cluster; driving the parallel simulation cluster to perform synchronous physical simulation based on the candidate vehicle parameters it carries, generating a virtual multidimensional simulation time series corresponding to the time series of the equivalent correction values ​​of each weighing sensor; using the actual equivalent correction value time series of each weighing sensor as the observation benchmark, performing multidimensional time series similarity matching with the virtual multidimensional simulation time series generated by the digital twins; selecting digital twins with a matching degree higher than a preset threshold to form a trusted consensus set; and performing confidence-weighted fusion based on the vehicle weight parameters and simulation matching confidence levels corresponding to the trusted consensus set, using the fusion result as the compensated final weight result.

[0014] This application provides a horizontal calibration and error control device for an electronic truck scale, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to: acquire pressure distribution data at the interface between adjacent weighing platforms in the electronic truck scale, and acquire three-dimensional tilt angle data corresponding to each weighing platform section, in a weighing operation state; reconstruct the dynamic weighing trajectory of the vehicle on multiple weighing platforms based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale; perform spatiotemporal labeling on the pressure distribution data and three-dimensional tilt angle data based on the dynamic weighing trajectory; perform fusion analysis on the spatiotemporally labeled pressure distribution data and three-dimensional tilt angle data along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial position of the weighing platform; and, based on the distribution characteristics, resolve the overall horizontal deviation vector into local deviation components associated with the corresponding weighing platform sections, correct the original weighing value based on the local deviation components, and output the compensated final weight result.

[0015] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: acquire pressure distribution data at the interface between adjacent weighing platforms in an electronic truck scale, and acquire three-dimensional tilt angle data corresponding to each weighing platform section, during a weighing operation; reconstruct the dynamic weighing trajectory of the vehicle on multiple weighing platforms based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale; perform spatiotemporal labeling on the pressure distribution data and three-dimensional tilt angle data based on the dynamic weighing trajectory; fuse and analyze the spatiotemporally labeled pressure distribution data and three-dimensional tilt angle data along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial location of the weighing platform; and, based on the distribution characteristics, resolve the overall horizontal deviation vector into local deviation components associated with the corresponding weighing platform sections, correct the original weighing value based on the local deviation components, and output the compensated final weight result.

[0016] The at least one technical solution adopted in this application embodiment can achieve the following beneficial effects: By reconstructing the dynamic weighing trajectory, this application embodiment establishes a unified spatiotemporal coordinate benchmark, accurately quantifies the dynamic movement process of the vehicle, and enables all subsequent analyses to be correlated with the vehicle's position and attitude in real time, solving the dependence of traditional methods on the completely stationary state of the vehicle. Spatiotemporal labeling assigns clear temporal and spatial attributes to each data point, transforming it from disordered streaming data into structured information. Furthermore, this application embodiment determines the distribution characteristics through fusion analysis, identifying the dynamic law of horizontal deviation changing with load position, thereby accurately locating the error source. In addition, this application embodiment decomposes and compensates the overall error by analyzing components and allocating it to specific responsible segments for precise compensation, improving the reliability of compensation and ultimately ensuring the metrological fairness of the output results. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart of a method for level calibration and error control of an electronic truck scale provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of an electronic truck scale level calibration and error control device provided in an embodiment of this application.

[0018] Figure label: 200: Level calibration and error control equipment for electronic truck scales; 201: Processor; 202: Memory. Detailed Implementation

[0019] This application provides a method, device, and medium for the horizontal calibration and error control of an electronic truck scale.

[0020] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0021] The technical solutions proposed in the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Figure 1 This is a flowchart illustrating a method for level calibration and error control of an electronic truck scale, as provided in an embodiment of this application. Figure 1 As shown, the method for level calibration and error control of electronic truck scales includes the following steps: S101. In the weighing operation state, acquire the pressure distribution data corresponding to the splicing interface of adjacent weighing platforms in the electronic truck scale, and acquire the three-dimensional tilt angle data corresponding to each weighing platform.

[0023] In one implementation of this application, a pressure sensor is installed at a stress point below the splicing interface of adjacent weighing platforms in the electronic truck scale, such as near the sensor support. When the weighing platform experiences slight misalignment or concentrated stress due to horizontal deviation, the pressure sensor can output the pressure values ​​at each measuring point in real time, thereby synthesizing pressure distribution data reflecting the stress state of the interface. Furthermore, digital three-dimensional tilt sensors are fixedly installed on the main structure of each independent weighing platform, such as at the center or both ends of the platform. Each tilt sensor operates independently, measuring the tilt angle of its respective weighing platform segment relative to the horizontal reference plane in real time, and outputting three-dimensional tilt angle data including forward / backward tilt and left / right tilt.

[0024] S102. Based on the weighing signal sequence sent by the weighing sensor in the electronic truck scale, reconstruct the dynamic weighing trajectory of the vehicle on the multi-section weighing platform.

[0025] In one implementation of this application, an adaptive threshold detection algorithm is used to identify peak events generated by wheel axle passage in the weighing signal sequence, and to determine the sensor number and timestamp corresponding to each peak event. Reference peak events with interval time differences less than a preset time threshold are identified. Based on the waveform correlation of the sensor signals corresponding to the reference peak events, these events are merged and decoupled into axle load events, and the timestamps of the axle load events are corrected. A time-series chain is constructed based on the chronological order of the corrected timestamps, and the instantaneous speed of the vehicle is determined based on the spatiotemporal information between the weighing sensors corresponding to consecutive events in the time-series chain. The instantaneous speed is used to verify the rationality of the time-series chain. Based on the spatial topological mapping relationship between the weighing sensors and the weighing platform segments, and the verified time-series chain, the dynamic weighing trajectory is reconstructed.

[0026] Specifically, this embodiment reads the signal data streams of all weighing sensors in real time at a fixed sampling frequency. To eliminate signal baseline drift and transient noise interference, this embodiment also employs an adaptive threshold algorithm. Specifically, a floating threshold based on the recent signal mean and variance is dynamically calculated. When the signal strength of a sensor exceeds this floating threshold and its rate of change simultaneously exceeds a threshold, it is determined to be a valid peak event. At this time, the timestamp of the event is recorded, and the sensor number that generated the signal is marked. All identified peak events are scanned, and events with an occurrence time interval less than a preset time interval are grouped together as a reference peak event group. The original sensor signal waveform segments corresponding to each event in this group are extracted, and the cross-correlation of these waveform segments is calculated. If the cross-correlation coefficient exceeds a preset threshold, it is determined that these peaks are caused by the same axle passing over adjacent sensors, and they are then merged and decoupled into a single axle load event. Subsequently, a weighted average timestamp is calculated using the timestamp and amplitude of each peak in the group as weights, serving as a more accurate occurrence time of this axle load event.

[0027] Furthermore, all axle load events are sorted according to their corrected timestamps to form a time-series chain. The instantaneous speed of the vehicle in this segment is calculated by dividing the known physical distance between the sensors corresponding to two consecutive events in the vehicle's direction of travel by their time difference. If the calculated instantaneous speed exhibits a sudden, physically inconsistent change, such as a sudden increase from 5 km / h to 30 km / h, it is determined that there may be incorrect peak matching or omissions in the time-series chain. Events in this abnormal segment are then re-associated or marked as suspicious points, thereby ensuring the physical rationality of the time-series chain. Finally, based on the spatial topological mapping relationship between the weighing sensors and the weighing platform segments, and the verified time-series chain, the dynamic weighing trajectory is reconstructed.

[0028] In one implementation of this application, the dynamic weighing trajectory is reconstructed based on the spatial topological mapping relationship between the weighing sensor and the weighing platform segment, and the verified time-series chain. Specifically, this includes mapping axle load events as two-dimensional trajectory points with timestamps based on the spatial topological mapping relationship and the verified time-series chain, and constructing a graph optimization model with the trajectory points as vertices. Constraint edges are added between the vertices of the graph optimization model; wherein the constraint edges include at least one of uniform motion constraints based on time intervals, minimum turning radius constraints, and fixed vehicle wheelbase constraints. By solving the graph optimization model, the two-dimensional coordinates of the trajectory point vertices are globally optimized to obtain optimized trajectory points that meet the edge constraints. Curve fitting is performed on the optimized trajectory points to generate a continuous two-dimensional planar trajectory function, which serves as the reconstructed dynamic weighing trajectory.

[0029] Specifically, this embodiment of the application pre-stores a two-dimensional plan view of the weighing platform. By calling the pre-stored two-dimensional plan view of the weighing platform, the installation coordinates of each weighing sensor are determined, thereby determining the spatial topological mapping relationship. Each axle load event in the verified time-series chain is mapped to the corresponding two-dimensional coordinates according to the sensor number that triggered it, thereby generating a series of discrete two-dimensional trajectory points with timestamps. Subsequently, the problem is represented by a graph optimization model, and each discrete trajectory point is defined as a vertex to be optimized in the model, and the parameter to be optimized for the vertex is the two-dimensional coordinates of that point.

[0030] Furthermore, in order to smooth and correct the trajectory using the inherent motion characteristics of the vehicle, specific constraint edges are added between the vertices of the graph model in this embodiment. First, uniform motion constraint edges are added between adjacent trajectory point vertices, which expect the displacement between adjacent points to be proportional to their time difference. Second, fixed wheelbase constraint edges are added between trajectory points belonging to different axles of the same vehicle. In addition, minimum turning radius constraint edges can be added for trajectory points that are not traveling in a straight line to prevent unreasonable sharp turns in the optimized trajectory. These constraint edges collectively encode the vehicle's kinematic knowledge into the graph optimization model.

[0031] Furthermore, the constructed graph optimization model is input into a nonlinear least squares optimization solver. The solver aims to adjust the coordinates of all vertices so that the adjusted coordinates are as close as possible to the initial observation position while satisfying the vehicle kinematic rules defined on all constraint edges to a greater extent. Through iterative calculation, the global optimal solution is determined, and a set of optimized trajectory point coordinates after smoothing and physical correction is output. The optimized discrete trajectory points are used as new control points, and algorithms such as cubic spline curve fitting are employed to generate a two-dimensional planar trajectory function. This two-dimensional planar trajectory function can accurately describe the motion of reference points on the vehicle, such as the center of gravity, on the weighing platform plane over time. It should be noted that this function not only directly serves as the reconstructed dynamic weighing trajectory, but also, by differentiating this function, the instantaneous velocity and acceleration vectors of the vehicle at any given time can be obtained, providing continuous dynamic information for subsequent force direction estimation and load analysis.

[0032] S103. Based on the dynamic weighing trajectory, spatiotemporal labels are applied to the pressure distribution data and three-dimensional tilt angle data.

[0033] In one implementation of this application, a spatiotemporal coordinate system is established based on the dynamic weighing trajectory, with vehicle travel time as the line and the platform segment space as the surface. Within this coordinate system, based on the pressure distribution data collected at each moment, the corresponding platform segments for the vehicle at different times are determined, and spatial identifiers for the splicing interface associated with the pressure distribution data are determined based on these platform segments. According to the installation position of each platform segment, the spatial identifiers of the respective platform segments are labeled on the three-dimensional tilt data, and vehicle load time phase identifiers are added to the three-dimensional tilt data based on the vehicle's velocity and acceleration state in the dynamic weighing trajectory. The labeled pressure distribution data and three-dimensional tilt data are aligned according to time series to generate a multi-source dataset with spatiotemporal labels.

[0034] Specifically, a unified spatiotemporal coordinate system is established based on the dynamic weighing trajectory. The horizontal axis represents the time line, and the vertical axis represents discrete platform segment spaces. For example, the entire platform is divided into spatial units such as Section 1, the interface between Sections 1 and 2, and Section 2. The pressure distribution data collected at each moment is processed. Based on the current moment, the system queries the dynamic weighing trajectory to determine which section (or two sections) of the platform the vehicle's center of gravity is currently on. Based on the vehicle chassis geometry and trajectory, the main load-bearing interface is inferred. For example, when the vehicle's center of gravity is at the front of Section 2, the interface between Sections 1 and 2 is the main load-bearing surface. The pressure data at that moment is labeled with its associated interface spatial identifier, such as P1-2, representing the interface between Sections 1 and 2, thus clarifying which specific location the pressure data reflects.

[0035] Furthermore, for the three-dimensional tilt data, firstly, based on the physical installation location of each tilt sensor, a fixed spatial identifier for its corresponding weighing platform segment is assigned, such as S2, indicating that this data comes from the second weighing platform segment. Next, the dynamic weighing trajectory is analyzed, and the real-time speed and acceleration of the vehicle are calculated. Based on these dynamic parameters, a vehicle load time phase identifier is added to the tilt data. For example, when the absolute value of acceleration is greater than a threshold, it is marked as an acceleration / deceleration phase; when the speed is stable and the vehicle covers the weighing platform, it is marked as a completely stationary phase; when the vehicle is partially on the weighing platform, it is marked as an off-center load phase. All pressure and tilt data with labeled spatial and time phase identifiers are given the same timestamp. Subsequently, all data streams are aligned according to the time series. Finally, the aligned data is organized into a multi-source dataset with spatiotemporal labels, where each row of data includes a timestamp, spatial location identifier, load phase identifier, and corresponding sensor reading.

[0036] S104. The pressure distribution data and three-dimensional tilt angle data labeled with spatiotemporal tags are fused and analyzed along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial position of the weighing platform.

[0037] In one implementation of this application, along discrete spatiotemporal points of the dynamic weighing trajectory, an observation data sequence consisting of pressure distribution features and three-dimensional tilt angle readings is sequentially extracted from a multi-source dataset. Furthermore, a corresponding reference data sequence is generated based on the vehicle axle load information corresponding to each discrete spatiotemporal point. Based on the residuals between the observation data sequence and the reference data sequence at each discrete spatiotemporal point, a multi-source residual sequence is obtained. Based on the spatial evolution pattern corresponding to the multi-source residual sequence, a feature waveform vector is determined. A spatial topology graph is constructed using weighing platform segments, splicing interfaces, and sensors as nodes and physical connection relationships as edges. The feature waveform vector is loaded as the initial feature of the corresponding node into the spatial topology graph. The spatial topology graph with the initial feature is input into a graph neural network, which outputs the distribution features of the horizontal deviation corresponding to each weighing platform segment.

[0038] Specifically, along the reconstructed dynamic weighing trajectory, discrete spatiotemporal points are selected at fixed time intervals. At each spatiotemporal point, two sets of key data are extracted from a multi-source dataset with spatiotemporal labels: first, the observation data sequence, including the pressure distribution characteristics of the main load-bearing splicing interface and the three-dimensional inclination angle readings of the corresponding weighing platform segment at that moment; second, the reference data sequence, which is calculated using a pre-set theoretical model calibrated under ideal horizontal conditions. This theoretical model calculates the ideal values ​​of the above observation data under ideal conditions without horizontal deviation, based on the vehicle axle load and vehicle position at the current moment. Next, the difference between the observation data sequence and the reference data sequence at each spatiotemporal point, i.e., the residual, is calculated. For example, the difference between the actual resultant pressure and the theoretical resultant pressure, and the difference between the actual inclination angle and the theoretical inclination angle are calculated. The residuals of all spatiotemporal points along the entire trajectory are arranged in chronological order to obtain a multi-source residual sequence. Subsequently, feature extraction is performed on the multi-source residual sequence, and it is converted into a feature waveform vector using signal processing techniques.

[0039] Furthermore, a spatial topology map is constructed based on the mechanical structure of the weighing platform. In the map, each platform segment, each splicing interface, and each weighing sensor is treated as an independent node, and the edges between nodes are established based on their actual physical connections. The feature waveform vector extracted for each physical entity in the previous step is used as the initial feature of that node and loaded onto the corresponding node in the spatial topology map. At this point, the topology map not only describes the physical structure of the weighing platform, but each node also carries dynamic anomaly information exhibited during weighing. The spatial topology map loaded with node features is then input into a pre-trained graph neural network. The output layer of the neural network generates a distribution feature representing the horizontal deviation of each platform segment node, such as a two-dimensional vector, quantifying the deviation angle of the segment in the forward tilt and lateral tilt directions, respectively.

[0040] S105. Based on the distribution characteristics, the overall horizontal deviation vector is analyzed into local deviation components associated with the corresponding weighing platform segments. Based on the local deviation components, the original weighing value is corrected, and the compensated final weight result is output.

[0041] In one implementation of this application, the distribution characteristics of the horizontal deviations corresponding to each weighing platform segment are concatenated to generate an overall horizontal deviation vector. Using the deviation severity index corresponding to the overall horizontal deviation vector as the source target, the contribution of each node to the generation of the overall horizontal deviation vector is determined by the gradient value of the initial feature waveform vector of each node in the spatial topology graph corresponding to the source target. Based on the contribution, the overall horizontal deviation vector is weighted to obtain the local deviation components associated with each weighing platform segment. Based on each local deviation component, local compensation coefficients for correcting the weighing sensor readings of the corresponding weighing platform segment are determined.

[0042] Specifically, the horizontal deviation distribution features output by the graph neural network, belonging to each segment node of the weighing platform (e.g., each node outputs a two-dimensional vector containing pitch and roll angles), are concatenated in segment order to form a comprehensive horizontal deviation vector describing the overall horizontal state of the entire weighing platform system. Then, a quantification target requiring tracing is set, such as calculating the magnitude of this comprehensive vector, i.e., the combined magnitude of deviations across all segments, as a deviation severity index. The magnitude of this index reflects the severity of the current overall levelness of the weighing platform system. Utilizing the differentiability of the graph neural network, a backpropagation calculation is performed. That is, the gradient of the deviation severity index set in the previous step with respect to the initial feature waveform vectors of each node in the graph neural network input layer is calculated. The magnitude of this gradient quantifies the impact of a small change in the feature waveform vector of a certain node on the final overall deviation severity index. This calculation is performed on all nodes to obtain the contribution of each node.

[0043] The calculated contributions of each node are normalized to a sum of 1, resulting in a set of weights. Then, the overall horizontal deviation vector is decomposed, and the deviation amount in the overall vector is allocated to each weighing platform segment based on the contribution weights, forming a local deviation component associated with each segment. Specifically, the segment containing nodes with higher contribution values ​​will be allocated a larger proportion of the deviation amount. This application embodiment pre-stores a compensation coefficient lookup table, which defines the correspondence between different local deviation components and the correction amount of the weighing sensor readings. By querying this table, a set of local compensation coefficients used to correct the readings of all weighing sensors on that segment can be determined.

[0044] In one implementation of this application, at each sampling moment during the weighing process, the instantaneous force vector direction acting on the load-bearing points of each weighing sensor is determined based on the dynamic weighing trajectory and a preset vehicle load distribution model. Based on the original readings, local deviation components, and instantaneous force vector directions of each weighing sensor at each sampling moment, the equivalent correction value of the original measured weight value under a preset virtual horizontal reference plane and the same force direction is determined. The equivalent correction values ​​of all weighing sensors at all sampling moments are matched and processed according to a time series to obtain the compensated final weight result.

[0045] Specifically, at each sampling moment, the vehicle's current position and attitude are first determined based on the dynamic weighing trajectory. Then, combined with a pre-set vehicle load distribution model that describes how the vehicle chassis structure distributes the total weight to each axle and how the weight of each axle is transmitted through the contact surface between the tires and the weighing platform, the instantaneous force vector direction acting on each load-bearing point of the weighing sensor is determined. This instantaneous force vector direction is not always vertically downward; it changes with the vehicle's position and attitude on the inclined weighing platform. For example, on a slope, the direction of the force will deviate towards the slope normal direction.

[0046] Furthermore, for the raw reading of each weighing sensor at each moment, the following correction calculation is performed: First, based on the local deviation component of the weighing platform segment to which the sensor belongs, i.e., the tilt angle, and the instantaneous force vector direction calculated in the previous step, the measurement error introduced by the non-horizontal mounting plane of the sensor and the non-perpendicular force direction is calculated using the mechanical formulas of three-dimensional spatial coordinate transformation and force decomposition. Then, this error is calculated in reverse to determine the reading that the sensor should output if it were installed on an absolutely horizontal virtual reference surface and under the same instantaneous force direction. This reading is the equivalent correction value.

[0047] Furthermore, the equivalent correction values ​​generated by all sensors at all sampling times are aligned and converged along a unified time axis. The compensated final weight result is then obtained using a state estimation method.

[0048] In one implementation of this application, the equivalent correction values ​​of all weighing sensors at all sampling times are matched according to a time series to obtain the compensated final weight result. Specifically, based on the original measured weight values ​​and dynamic weighing trajectories corresponding to the weighing sensors, multiple sets of candidate vehicle parameters are generated. Based on local deviation components, multiple digital twins carrying different candidate parameters are instantiated to generate a parallel simulation cluster. The parallel simulation cluster is driven to perform synchronous physical simulation based on the candidate vehicle parameters it carries, generating a virtual multidimensional simulation time series corresponding to the time series of the equivalent correction values ​​of each weighing sensor. The actual equivalent correction value time series of each weighing sensor is used as the observation benchmark and matched with the virtual multidimensional simulation time series generated by the digital twins for multidimensional time series similarity. Digital twins with a matching degree higher than a preset threshold are selected to form a trusted consensus set. Based on the vehicle weight parameters and simulation matching confidence levels corresponding to the trusted consensus set, a confidence-weighted fusion is performed, and the fusion result is used as the compensated final weight result.

[0049] Specifically, based on the vehicle's wheelbase, number of axles, and other contour information obtained during this weighing, and combined with common vehicle model data from the historical database, a large set of candidate vehicle parameters varying within a reasonable range is generated. Each set of parameters mainly includes an assumed total vehicle weight, an assumed center of gravity position on the vehicle, and possible load distributions. These parameter sets cover multiple possibilities for the vehicle weight under the current observation. The determined local deviation components, i.e., the actual tilt state of each weighing platform segment, are obtained and used as the fixed structural basis for the digital twin. Then, for each set of candidate vehicle parameters, a corresponding digital twin is quickly instantiated. All twins constitute a parallel simulation cluster, driving the entire cluster to run physical simulation synchronously. The simulation process reproduces the real timeline of this weighing, simulating the entire process of the vehicle entering, stopping, and exiting, and records the reading time series that each virtual weighing sensor should output under the assumed parameters of each twin, i.e., the virtual multidimensional simulation time series.

[0050] Furthermore, the actual equivalent correction value time series of each weighing sensor is used as the real observation benchmark. For each digital twin, the multidimensional dynamic time warping distance between its generated virtual simulation time series and the real observation series is calculated. This algorithm calculates the distance value by comparing the similarity of the curve shape of the entire dynamic process, and the calculated distance value is converted into a similarity score. This score comprehensively evaluates the overall degree of consistency between the simulation result of the twin and the actual observation data of all sensors. This application embodiment also sets a similarity score threshold, and filters out all digital twins with scores higher than the threshold to form a credible consensus set. The similarity score of each credible twin is normalized and used as its simulation matching confidence weight. The vehicle weight parameters carried by all twins in the credible consensus set are weighted and averaged according to their confidence weights to achieve the final compensated weight result.

[0051] Figure 2 This is a schematic diagram of the structure of an electronic truck scale leveling and error control device provided in an embodiment of this application. Figure 2As shown, the level calibration and error control device 200 for an electronic truck scale includes: at least one processor 201; and a memory 202 communicatively connected to the at least one processor 201. The memory 202 stores instructions executable by the at least one processor 201, which, when executed, enable the at least one processor 201 to: acquire pressure distribution data at the interface between adjacent weighing platforms in the electronic truck scale, and acquire three-dimensional tilt angle data corresponding to each weighing platform section, during weighing operations; reconstruct the dynamic weighing trajectory of the vehicle on multiple weighing platforms based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale; perform spatiotemporal labeling on the pressure distribution data and three-dimensional tilt angle data based on the dynamic weighing trajectory; perform fusion analysis on the spatiotemporally labeled pressure distribution data and three-dimensional tilt angle data along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial position of the weighing platform; and, based on the distribution characteristics, resolve the overall horizontal deviation vector into local deviation components associated with the corresponding weighing platform sections, correct the original weighing value based on the local deviation components, and output the compensated final weight result.

[0052] This application provides a non-volatile computer storage medium storing computer-executable instructions. These instructions are configured to: acquire pressure distribution data at the interface between adjacent weighing platforms in an electronic truck scale, and acquire three-dimensional tilt angle data corresponding to each weighing platform section, during a weighing operation; reconstruct the dynamic weighing trajectory of the vehicle on multiple weighing platforms based on the weighing signal sequence sent by the weighing sensors in the electronic truck scale; perform spatiotemporal labeling on the pressure distribution data and three-dimensional tilt angle data based on the dynamic weighing trajectory; fuse and analyze the spatiotemporally labeled pressure distribution data and three-dimensional tilt angle data along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the spatial location of the weighing platform; and, based on the distribution characteristics, resolve the overall horizontal deviation vector into local deviation components associated with the corresponding weighing platform sections, correct the original weighing value based on the local deviation components, and output the compensated final weight result.

[0053] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0054] The above descriptions are merely embodiments of this application and are not intended to limit the scope of this application. For those skilled in the art, various modifications and variations can be made to the embodiments of this application. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions in the embodiments of this application.

Claims

1. A method for horizontal calibration and error control of an electronic truck scale, characterized in that, The method comprises: In the weighing operation state, the pressure distribution data corresponding to the adjacent scale platform splicing interface in the electronic truck scale and the three-dimensional inclination data corresponding to each section of the scale platform are acquired; Based on the weighing signal sequence sent by the weighing sensor in the electronic truck scale, the dynamic weighing trajectory of the vehicle on the multi-section scale platform is reconstructed; Based on the dynamic weighing trajectory, the pressure distribution data and the three-dimensional inclination data are labeled in space-time; The pressure distribution data and the three-dimensional inclination data labeled in space-time are fused and analyzed along the dynamic weighing trajectory to determine the distribution characteristics of the horizontal deviation in the space position of the scale platform; Based on the distribution characteristics, the overall horizontal deviation vector is analyzed into local deviation components associated with the corresponding scale platform segments, and the original weight measurement value is corrected based on the local deviation components to output the final compensated weight result.

2. The method for horizontal calibration and error control of an electronic truck scale according to claim 1, characterized in that, The reconstruction of the dynamic weighing trajectory of the vehicle on the multi-section scale platform based on the weighing signal sequence sent by the weighing sensor in the electronic truck scale specifically comprises: The peak value events generated by the wheel axle pressure in the weighing signal sequence are determined by an adaptive threshold detection algorithm, and the sensor number and timestamp corresponding to each peak value event are determined; The reference peak value events with an interval time difference less than a preset time threshold are determined, the reference peak value events are merged and decoupled into axle load events according to the waveform correlation of the sensor signals corresponding to the reference peak value events, and the timestamps of the axle load events are corrected; The time sequence chain is constructed according to the order of the corrected timestamps, and the instantaneous speed of the vehicle is determined according to the space-time information between the weighing sensors corresponding to the continuous events in the time sequence chain, so that the time sequence chain is reasonably checked through the instantaneous speed; The dynamic weighing trajectory is reconstructed according to the spatial topological mapping relationship between the weighing sensors and the scale platform segments and the checked time sequence chain.

3. The method for horizontal calibration and error control of an electronic truck scale according to claim 2, characterized in that, The reconstruction of the dynamic weighing trajectory according to the spatial topological mapping relationship between the weighing sensors and the scale platform segments and the checked time sequence chain specifically comprises: According to the spatial topological mapping relationship and the checked time sequence chain, the axle load events are mapped into two-dimensional trajectory points with timestamps, and a graph optimization model is constructed with the trajectory points as vertices; Constraint edges are added between the vertices of the graph optimization model; wherein the constraint edges at least include one of uniform motion constraint based on time interval, minimum turning radius constraint and vehicle wheelbase fixed constraint; The two-dimensional coordinates of the trajectory point vertices are globally optimized by solving the graph optimization model to obtain the optimized trajectory points conforming to the edge constraints; The optimized trajectory points are curve fitted to generate a continuous two-dimensional plane trajectory function as the reconstructed dynamic weighing trajectory.

4. The method for horizontal calibration and error control of an electronic truck scale according to claim 1, wherein, The labeling of the pressure distribution data and the three-dimensional inclination data in space-time based on the dynamic weighing trajectory specifically comprises: A space-time coordinate system with vehicle travel time as the line and scale platform segment space as the plane is established based on the dynamic weighing trajectory; In the space-time coordinate system, based on the pressure distribution data collected at each time, the scale segments corresponding to the vehicle at different times are determined to determine the splicing interface space identifier associated with the pressure distribution data based on the scale segments; According to the installation position of each scale section, the space identifier of the scale section to which the three-dimensional inclination data belongs is labeled, and according to the speed and acceleration state of the vehicle in the dynamic weighing track, the three-dimensional inclination data is additionally labeled with the vehicle load time phase identifier; Align the pressure distribution data and the three-dimensional inclination data labeled with the space-time label in time sequence to generate a multi-source data set with a space-time label.

5. The method for horizontal calibration and error control of an electronic truck scale according to claim 4, characterized in that, The pressure distribution data and the three-dimensional inclination data labeled with the space-time label are fused and analyzed along the dynamic weighing track to determine the distribution characteristics of the horizontal deviation in the scale space position, specifically including: Along the discrete space-time points of the dynamic weighing track, the observation data sequence composed of pressure distribution characteristics and three-dimensional inclination readings is extracted in the multi-source data set in turn, and the corresponding reference data sequence is generated according to the vehicle axle load information corresponding to each discrete space-time point; Based on the observation data sequence and the reference data sequence, the multi-source residual sequence is obtained based on the residual corresponding to each discrete space-time point, and the feature waveform vector is determined based on the space evolution mode corresponding to the multi-source residual sequence; Taking the scale section, the splicing interface and the sensor as the node and the physical connection relationship as the edge, a space topology graph is constructed, and the feature waveform vector is loaded as the initial feature of the corresponding node to the space topology graph; Input the space topology graph loaded with the initial feature into the graph neural network, and output the distribution characteristics of the horizontal deviation corresponding to each scale section through the graph neural network.

6. The method for horizontal calibration and error control of an electronic truck scale according to claim 5, wherein, The distribution characteristics of the horizontal deviation corresponding to each scale section are spliced to generate an overall horizontal deviation vector; The deviation severity index corresponding to the overall horizontal deviation vector is taken as the traceability target, and the gradient value of the initial feature waveform vector of each node in the space topology graph through the traceability target is determined to determine the contribution degree of each node to the generation of the overall horizontal deviation vector; Based on the contribution degree, the overall horizontal deviation vector is weighted to obtain the local deviation component associated with each scale section; According to each local deviation component, a local compensation coefficient for correcting the weighing sensor reading of the corresponding scale section is determined. The original measured weight value is corrected based on the local deviation component to output the compensated final weight result, specifically including:

7. The method for horizontal calibration and error control of an electronic truck scale according to claim 1, wherein, At each sampling time during the weighing process, the instantaneous force vector direction acting on each weighing sensor bearing point is determined according to the dynamic weighing track and the preset vehicle load distribution model; ​ Determine the original measurement weight value in the preset virtual horizontal reference plane and the equivalent correction value in the same force direction based on the original reading of each weighing sensor corresponding to each sampling time, the local deviation component, and the instantaneous force vector direction; Match the equivalent correction values of all weighing sensors at all sampling times in time sequence to obtain the compensated final weight result.

8. The method for horizontal calibration and error control of an electronic truck scale according to claim 7, characterized in that, The matching processing of the equivalent correction values of all weighing sensors at all sampling times in time sequence to obtain the compensated final weight result specifically includes: Based on the original measurement weight value corresponding to the weighing sensor and the dynamic weighing trajectory, generate multiple groups of candidate vehicle parameters, instantiate multiple digital twins carrying different candidate parameters based on the local deviation component, generate a parallel simulation cluster; Drive the parallel simulation cluster to perform synchronous physical simulation based on the candidate vehicle parameters carried by each, generate a virtual multi-dimensional simulation time sequence corresponding to the equivalent correction value time sequence of each weighing sensor; Match the real equivalent correction value time sequence of each weighing sensor as an observation reference with the virtual multi-dimensional simulation time sequence generated by the digital twin for multi-dimensional time sequence similarity matching; Screen out digital twins with matching degree higher than a preset threshold to constitute a trusted consensus set; Based on the vehicle weight parameters and simulation matching confidence corresponding to the trusted consensus set, perform confidence weighted fusion, and take the fusion result as the compensated final weight result.

9. A level calibration and error control apparatus for an electronic truck scale, comprising: The device includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to perform the method of any one of claims 1-8.

10. A non-transitory computer storage medium storing computer-executable instructions that, when executed, cause a computer to perform: The computer executable instructions can perform the method of any one of claims 1-8. The computer executable instructions can perform the method of any one of claims 1-8.

Images