Intelligent load prediction and adjustment method in dynamic weighing system

Abstract

The invention relates to the technical field of dynamic weighing, and discloses an intelligent load prediction and adjustment method in a dynamic weighing system, which comprises the following steps of: separating a weighing signal into a space domain static load characteristic (pressure point distribution sparsity and gravity center geometric vector) and a time domain dynamic inertial interference component (acceleration coupling noise) through space-time double-domain decoupling; the frequency spectrum aliasing problem of a traditional filtering algorithm is eliminated from a signal source; the future interference trend is predicted based on the real-time motion state of the carrier, a feed-forward compensation signal is generated through a pre-training dynamic basis function library, and dynamic errors are actively counteracted; in combination with sensor array space consistency detection and lightweight online learning, decoupling model parameters are dynamically corrected. According to the method, space geometric analysis and kinematics feedforward prediction are combined, a traditional signal processing framework is strided out, error source separation and real-time compensation are achieved, and the dynamic weighing precision of transient working conditions such as sudden stop and speed change is remarkably improved.

Description

Technical Field

[0001] The invention relates to an intelligent load prediction and adjustment method in a dynamic weighing system, belonging to the technical field of dynamic weighing. Background Art

[0002] Existing dynamic weighing systems usually rely on filtering algorithms to remove noise interference, but traditional filtering algorithms are prone to spectrum aliasing when processing high-frequency noise and low-frequency signals, resulting in distortion of static load signals. This makes it impossible for weighing accuracy to meet high-demand application scenarios under transient conditions such as emergency stops and speed changes. Especially in the fields of e-commerce warehousing, automated logistics, and smart manufacturing, weighing errors pose significant challenges to work efficiency and cost control.

[0003] In addition, most of the existing dynamic weighing methods are based on the processing of a single sensor signal, and fail to effectively consider the spatial distribution characteristics of the sensor array and the motion state of the vehicle, resulting in insufficient stability and reliability of the system in complex dynamic environments. Even with the support of high-precision sensors, traditional technologies still find it difficult to achieve rapid adaptation to different vehicle types and dynamic environments, and the update process is complex and costly. Therefore, the existing dynamic weighing technology still has technical bottlenecks in handling the rapidly changing dynamic load of the vehicle, improving weighing accuracy, and adapting to diverse application needs. Summary of the invention

[0004] The present invention provides an intelligent load prediction and adjustment method in a dynamic weighing system, the main purpose of which is to solve the problem of insufficient weighing accuracy caused by inertial interference and dynamic error in the dynamic weighing system.

[0005] To achieve the above object, the present invention provides an intelligent load prediction and adjustment method in a dynamic weighing system, comprising the following steps: S1, signal acquisition: obtaining the real-time output signal of the weighing sensor array and synchronously receiving the motion state data of the vehicle, the motion state data including acceleration, speed and posture information; S2, time-space dual domain decoupling: Spatial domain analysis: based on the spatial distribution characteristics of the load cell array, extract the spatial domain characteristics of the static load, the spatial domain characteristics include the sparse distribution of pressure points and the geometric vector of the center of gravity position; Time domain analysis: According to the real-time motion state data of the vehicle, a dynamic inertial interference model is constructed to generate a time domain noise component; the dynamic inertial interference model is expressed as: , in, is the noise component in the time domain, For real-time acceleration data, For real-time speed data, is the rate of change of acceleration, , , The inertial disturbance coefficient is fitted by the calibration data of the standard test vehicle; S3, feedforward compensation: matching the type of the vehicle through the pre-trained dynamic basis function library, predicting the interference evolution trend at future moments based on the dynamic inertial interference model, generating a feedforward compensation signal, and reversely superimposing the feedforward compensation signal to the real-time output signal to obtain a corrected static load value; S4, closed-loop self-healing calibration: When it is detected that the corrected static load value deviates from the preset threshold value, the residual error distribution pattern is traced based on the spatial consistency characteristics of the sensor array, and the weight parameters of the spatiotemporal dual-domain decoupling are adjusted by the gradient descent method to complete dynamic error suppression.

[0006] As a preferred embodiment, in step S2: spatial domain analysis uses a sparse coding algorithm to separate the static load component in the weighing sensor array signal, wherein the sparse coding basis function is generated by typical load distribution experimental data; time domain analysis constructs a second-order inertial interference equation through the acceleration and velocity data of the vehicle to calculate the dynamic noise component in real time.

[0007] As a preferred embodiment, the pre-training method of the dynamic basis function library includes: extracting the common motion-load coupling characteristics of three types of vehicles, namely forklifts, AGVs and conveyor belts, through transfer learning to generate basic feature vectors; when adding a new vehicle type, fine-tuning the basic feature vector through online learning based on the mapping relationship between its 10-20 groups of real-time motion states and weighing data.

[0008] As a preferred implementation, in the closed-loop self-healing calibration: the residual error distribution pattern is detected through the spatial consistency of the sensor array to locate the incompletely decoupled inertial interference component; a lightweight online learning algorithm is used to adjust the weight parameters of the spatiotemporal dual-domain decoupling, and the weight parameters include the basis function weights of the spatial domain sparse coding and the coefficients of the time domain inertial interference model.

[0009] As a preferred embodiment, the matching method of the dynamic basis function library includes: calling pre-trained forklift motion basis functions, AGV motion basis functions or conveyor belt motion basis functions according to the vehicle type; when the vehicle type is not pre-stored, matching the closest basis function group based on the spectral characteristics of the real-time motion state data.

[0010] As a preferred embodiment, in the time-space dual-domain decoupling: the spatial domain feature extraction has a higher priority than the time domain interference modeling, and the decoupling operations of the two are completed synchronously within a 10ms period; the geometric vector of the spatial domain feature is calculated through the topological relationship of the sensor array, and the topological relationship includes the spacing between adjacent sensors and the pressure distribution gradient.

[0011] As a preferred embodiment, the method for generating the feedforward compensation signal includes: predicting the amplitude of the inertial interference component within the next 5-10ms based on the acceleration change rate of the vehicle; scaling the predicted value through the migration adaptation coefficient of the dynamic basis function library to generate a compensation signal matching the current vehicle.

[0012] As a preferred implementation, the triggering condition for the closed-loop self-healing calibration is: the fluctuation amplitude of the corrected static load value exceeds 1.5 times of the preset threshold within three consecutive sampling periods; the learning rate of the gradient descent method is dynamically adjusted according to the distribution pattern of the residual error.

[0013] As a preferred embodiment, it also includes the update of the dynamic basis function library, wherein the update mechanism of the dynamic basis function library includes: every time 100 sets of mapping relationships between vehicle motion states and weighing data are added, the global optimization of the basis function library is triggered once; during the optimization process, the common characteristics of the historical basis function groups are retained, and only the feature vectors that conflict with the newly added data are updated.

[0014] As a preferred embodiment, the weight parameter initialization method of the time-space dual-domain decoupling is: based on the calibration data of the standard test vehicle, the initial weights are fitted by the least squares method; the standard test vehicle includes the benchmark motion load curves of three types of equipment: forklifts, AGVs and conveyor belts.

[0015] Compared with the problems described in the background technology, the beneficial effects of the present invention are: combining spatial geometric analysis with kinematic feedforward prediction, decoupling the dynamic weighing signal into static load characteristics in the spatial domain (such as sparse distribution of pressure points, center of gravity geometric vector) and dynamic inertial interference components in the time domain (such as acceleration coupling noise), realizing error separation at the signal source, avoiding static load information distortion caused by time-frequency aliasing in traditional filtering algorithms, and realizing collaborative decoupling of spatial geometric characteristics and kinematic predictions, fundamentally eliminating signal coupling effects, and improving weighing stability under dynamic conditions; taking the vehicle motion state as a feedforward variable, and predicting future interference trends based on the real-time motion state of the vehicle, which is more suitable for transient scenarios such as emergency stops and speed changes, ensuring weighing accuracy in the form of prediction, and realizing zero-cost rapid adaptation of new vehicles (such as 10 groups of sample fine-tuning) through cross-device common feature migration and lightweight online learning, and combining the spatial consistency detection of the sensor array, automatically tracing the residual error distribution mode and dynamically correcting the decoupling model, so as to achieve low-cost use effect, which is more conducive to meeting the needs of e-commerce warehousing scenarios for high-density and high-dynamic weighing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the intelligent load prediction and regulation method of the present invention; Figure 2 It is a flow chart of the intelligent load prediction and adjustment method in the dynamic weighing system of the present invention; Figure 3 It is a diagram of the error detection and correction process in the dynamic weighing system of the present invention.

[0017] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] The embodiment of the present application provides an intelligent load prediction and adjustment method in a dynamic weighing system, which includes the following steps: S1, signal acquisition: obtaining the real-time output signal of the weighing sensor array and synchronously receiving the motion state data of the vehicle, wherein the motion state data includes acceleration, speed and posture information;

[0020] S2, time-space dual-domain decoupling: Spatial domain analysis: Based on the spatial distribution characteristics of the weighing sensor array, the spatial domain characteristics of the static load are extracted, and the spatial domain characteristics include the sparse distribution of pressure points and the geometric vector of the center of gravity position; Time domain analysis: According to the real-time motion state data of the vehicle, a dynamic inertial interference model is constructed to generate a time domain noise component; the dynamic inertial interference model is expressed as: , in, is the noise component in the time domain, It is real-time acceleration data, which comes from the acceleration sensor in the weighing sensor array. During the data acquisition process, the real-time acceleration signal is filtered to remove static interference to ensure that only dynamic acceleration changes are reflected. Real-time speed data is obtained by the speed sensor in real time and properly filtered to reduce static errors. The accuracy of speed data directly affects the accuracy of the interference prediction model. Therefore, the calibration and maintenance of the speed sensor is very important during system operation. The acceleration change rate is obtained through the differential equation of the vehicle's motion state. It is specifically calculated as the time derivative of the acceleration signal, reflecting the speed change trend of the vehicle. The acceleration change rate provides instant feedback on the dynamic characteristics of the vehicle and is the key to building a dynamic inertial interference model. , , The inertial disturbance coefficients are fitted by the calibration data of the standard test vehicle; these coefficients are fitted by the experimental data of the standard test vehicle. Through regression analysis, the least squares method is used to fit the acceleration, velocity, and acceleration change rate to ensure that these coefficients can accurately reflect the inertial characteristics of the vehicle under different dynamic loads. This process avoids overfitting through multiple cross-validations to ensure that the model can adapt to different vehicle types.

[0021] S3, feedforward compensation: match the type of the vehicle through the pre-trained dynamic basis function library, predict the interference evolution trend at future times based on the dynamic inertial interference model, generate a feedforward compensation signal, and reversely superimpose the feedforward compensation signal on the real-time output signal to obtain a corrected static load value; S4, closed-loop self-healing calibration: when it is detected that the deviation between the corrected static load value and the preset threshold exceeds the limit, trace the residual error distribution pattern based on the spatial consistency characteristics of the sensor array, and adjust the weight parameters of the spatiotemporal dual-domain decoupling through the gradient descent method to complete dynamic error suppression.

[0022] As a preferred embodiment, in step S2: spatial domain analysis uses a sparse coding algorithm to separate the static load component in the weighing sensor array signal, wherein the sparse coding basis function is generated by typical load distribution experimental data; time domain analysis constructs a second-order inertial interference equation through the acceleration and velocity data of the vehicle to calculate the dynamic noise component in real time.

[0023] As a preferred embodiment, the pre-training method of the dynamic basis function library includes: extracting the common motion-load coupling characteristics of three types of vehicles, namely forklifts, AGVs and conveyor belts, through transfer learning to generate basic feature vectors; when adding a new vehicle type, fine-tuning the basic feature vector through online learning based on the mapping relationship between its 10-20 groups of real-time motion states and weighing data.

[0024] As a preferred implementation, in the closed-loop self-healing calibration: the residual error distribution pattern is detected through the spatial consistency of the sensor array to locate the incompletely decoupled inertial interference component; a lightweight online learning algorithm is used to adjust the weight parameters of the spatiotemporal dual-domain decoupling, and the weight parameters include the basis function weights of the spatial domain sparse coding and the coefficients of the time domain inertial interference model.

[0025] As a preferred embodiment, the matching method of the dynamic basis function library includes: calling pre-trained forklift motion basis functions, AGV motion basis functions or conveyor belt motion basis functions according to the vehicle type; when the vehicle type is not pre-stored, matching the closest basis function group based on the spectral characteristics of the real-time motion state data.

[0026] As a preferred embodiment, in the time-space dual-domain decoupling: the spatial domain feature extraction has a higher priority than the time domain interference modeling, and the decoupling operations of the two are completed synchronously within a 10ms period; the geometric vector of the spatial domain feature is calculated through the topological relationship of the sensor array, and the topological relationship includes the spacing between adjacent sensors and the pressure distribution gradient.

[0027] As a preferred embodiment, the method for generating the feedforward compensation signal includes: predicting the amplitude of the inertial interference component within the next 5-10ms based on the acceleration change rate of the vehicle; scaling the predicted value through the migration adaptation coefficient of the dynamic basis function library to generate a compensation signal matching the current vehicle.

[0028] As a preferred implementation, the triggering condition for the closed-loop self-healing calibration is: the fluctuation amplitude of the corrected static load value exceeds 1.5 times of the preset threshold within three consecutive sampling periods; the learning rate of the gradient descent method is dynamically adjusted according to the distribution pattern of the residual error.

[0029] As a preferred embodiment, it also includes the update of the dynamic basis function library, wherein the update mechanism of the dynamic basis function library includes: every time 100 sets of mapping relationships between vehicle motion states and weighing data are added, the global optimization of the basis function library is triggered once; during the optimization process, the common characteristics of the historical basis function groups are retained, and only the feature vectors that conflict with the newly added data are updated.

[0030] As a preferred embodiment, the weight parameter initialization method of the time-space dual-domain decoupling is: based on the calibration data of the standard test vehicle, the initial weights are fitted by the least squares method; the standard test vehicle includes the benchmark motion load curves of three types of equipment: forklifts, AGVs and conveyor belts.

[0031] See also Figure 2 and Figure 3 , Figure 2 The flowchart of the intelligent load prediction and adjustment method in the dynamic weighing system of the present invention. First, the sensor signal and motion state data are input. Next, the spatial domain decoupling is performed, and the spatial domain analysis includes sparse coding to separate the static load component, the basis function from the experimental data and the calculation of the topological relationship (such as spacing / gradient). Then, the time domain decoupling is performed, and the time domain analysis includes constructing the third-order inertia equation and calculating the dynamic noise component in real time. The whole decoupling process is completed synchronously within a 10ms period, and finally the static load characteristics and the time domain noise components are output. Figure 3This is a diagram of the error detection and correction process in the dynamic weighing system of the present invention. First, the system detects the deviation between the static load value and the adjustment value. According to the deviation value, the system enters the error detection module for detection. Next, the system performs a spatial consistency test to determine whether the response of the sensor array is consistent. Based on this, the system updates the relevant system parameters and restores stability. As the error detection is further advanced, the system adjusts the model parameters through a progressive reduction method to reduce the impact of the error, ultimately ensuring the stability and accuracy of the system. In addition, the system gradually corrects and optimizes the weighing accuracy by adjusting the spatiotemporal decoupling weights.

[0032] Embodiment 1: During the implementation process, firstly, multiple weighing sensor arrays are installed on the vehicle, and the real-time motion state data of the vehicle is collected synchronously. The motion state data includes acceleration, velocity and posture information. Through these data, the system can fully reflect all factors affecting weighing generated by the vehicle during the dynamic working process. While collecting signals, the system ensures the data synchronization of all sensors to avoid measurement errors caused by inconsistent data timing. In the weighing sensor array, the output signal of each sensor is transmitted to the signal processing module of the system together with the motion state data. Then, based on the spatial distribution characteristics of the sensor array, the system uses a sparse coding algorithm to separate the static load component from the dynamic noise component in the weighing signal. The extraction of spatial domain features mainly characterizes the spatial distribution of the load through the geometric vector of the center of gravity position and the sparsity of the pressure point distribution. Specifically, the geometric vector of the center of gravity is calculated by the distance between adjacent sensors and the pressure distribution gradient, and the spatial domain features are analyzed before the time domain features to ensure that their priority is higher than the modeling of dynamic inertial interference. Subsequently, using the acceleration, velocity and posture information of the vehicle, the system constructs a dynamic inertial interference model in real time. The model calculates and extracts the noise component in the time domain through the vehicle's motion state data. After the output signal of each sensor is decoupled in both time and space domains, it can effectively separate the static load from the dynamic inertial interference, thereby avoiding the error caused by spectrum aliasing in traditional filtering algorithms.

[0033] The system generates corresponding compensation signals through a pre-trained dynamic basis function library, such as according to the vehicle type and motion state. The generation of the compensation signal is based on the motion state of the vehicle, combined with the previous dynamic inertial interference model to predict the interference evolution trend at future moments. The feedforward compensation signal actively corrects the dynamic load error by reverse superposition, thereby effectively improving the real-time accuracy of the weighing system. When the vehicle type changes or is added, the system will fine-tune the dynamic adaptation coefficient in the basis function library through online learning. Specifically, the system will update the basis function library through transfer learning methods based on the mapping relationship between 10-20 sets of real-time motion data and weighing data of the newly added vehicle to ensure that the new vehicle can be quickly adapted at zero cost and maintain weighing accuracy.

[0034] When the system detects that the deviation between the corrected static load value and the preset threshold exceeds the limit, it enters the closed-loop self-healing calibration mechanism. At this time, based on the spatial consistency characteristics of the sensor array, the system locates the incompletely decoupled inertial interference component by tracing the residual error distribution pattern. Through the gradient descent method, the system dynamically adjusts the weight parameters of the time-space dual-domain decoupling to achieve dynamic error suppression. Through the spatial consistency detection algorithm of the sensor array, the system can promptly detect errors caused by hardware failures or external interference and effectively correct them. The learning rate of the gradient descent method is dynamically adjusted according to the distribution pattern of the residual error to ensure the accuracy and stability of the error calibration. The system is divided into four major modules: signal acquisition module, time-space decoupling module, compensation signal generation module, and closed-loop calibration module. Each module has independent functions and responsibilities, and the data flow and control logic between modules are closely connected. The specific process is as follows: signal acquisition module: real-time acquisition of weighing sensor signals, and synchronous collection of vehicle motion state data; spatiotemporal decoupling module: perform spatial and temporal domain signal analysis to separate static loads from dynamic interference; compensation signal generation module: based on dynamic basis function library and kinematic model, generate feedforward compensation signal; closed-loop calibration module: when the error exceeds the limit, automatically trigger closed-loop calibration, adjust decoupling model parameters by gradient descent method, and correct dynamic error. This embodiment combines spatial geometry analysis with kinematic feedforward prediction, breaking through the bottleneck of signal decoupling in traditional dynamic weighing systems. In practical applications, the system can effectively improve the accuracy of dynamic weighing, especially in transient conditions such as emergency stop and speed change. The modular design of the system makes the technical solution easy to implement and maintain, and can automatically adapt according to different vehicle types without additional hardware support, and has good market application prospects. Through precise signal decoupling and real-time dynamic compensation, the present invention can provide high-precision and high-stability dynamic weighing solutions in complex weighing environments, especially in high-density and high-dynamic e-commerce warehousing environments.

[0035] Embodiment 2: In the signal acquisition stage of this embodiment, first ensure that the real-time motion state data (including acceleration, velocity and posture information) of the weighing sensor array and the vehicle can be collected synchronously. These data are collected and transmitted to the system by the weighing sensor and the inertial measurement unit (IMU) at the same time. In the specific implementation, a timing synchronization mechanism is adopted to ensure the timing consistency of the sensor data and avoid weighing errors caused by signal delay or data timing disorder. The optimized time-space dual-domain decoupling step is divided into two sub-steps: spatial domain feature extraction and time domain interference modeling. Spatial domain feature extraction: On the original basis, the specific calculation method of the spatial domain feature is further clarified. The spatial domain characteristics of the static load include the sparse distribution of pressure points and the geometric vector of the center of gravity. During implementation, the geometric vector of the static load is accurately calculated by analyzing the positional relationship of the weighing sensor array (sensor spacing and pressure distribution gradient of adjacent sensors). This part no longer relies on the traditional filtering algorithm, but adopts a geometric analysis method based on the sensor array topology. The algorithm analyzes the local load distribution of each sensor and then calculates the center of gravity position of the entire array. Time domain interference modeling: Further improve the time domain model and clarify the generation process of dynamic inertial interference signals. In this process, the acceleration, velocity and attitude data of the vehicle work together to construct the second-order inertial interference equation. By calculating the noise component of the acceleration signal in real time, the weight of the time domain noise component is dynamically adjusted. The source and function of each variable have been clarified to ensure that the model can adapt to the inertial interference of different vehicles under different working conditions.

[0036] The specific method of generating the feedforward compensation signal is as follows: through the pre-trained dynamic basis function library, according to the real-time motion state data of the vehicle (acceleration, velocity change rate, etc.), the change trend of the inertial interference signal in the next 5-10 milliseconds is accurately predicted, and then based on the predicted interference trend, the weighing signal is corrected by reverse superposition. For each type of vehicle, the basis functions in the dynamic basis function library have been pre-trained through transfer learning, and online fine-tuning is performed according to the newly added vehicle data to ensure that the compensation signal can be dynamically adapted according to the specific characteristics of the vehicle, and the inertial interference amplitude in the next few milliseconds is predicted based on the real-time motion state data of the vehicle. Through the pre-trained dynamic basis function library, the motion-load coupling characteristics of each vehicle type are extracted, and the coefficients in the basis function library are fine-tuned through transfer learning. Based on the predicted results of the acceleration change rate, the system reversely superimposes the compensation signal to correct the weighing signal. This process accurately predicts future dynamic errors and avoids weighing deviations caused by inertial interference.

[0037] The optimized closed-loop self-healing calibration mechanism specifically includes the following steps, such as error detection and tracing: when the deviation between the corrected static load value and the preset threshold exceeds the predetermined range, the system locates the incompletely decoupled inertial interference component by analyzing the spatial consistency characteristics of the sensor array. The rapid identification of dynamic errors is ensured by tracing the residual error distribution pattern; dynamic adjustment by gradient descent method: based on the detected residual error distribution pattern, the weight parameters of the time-space dual-domain decoupling are adjusted using the gradient descent method. The key parameters include the basis function weights of the spatial domain sparse coding and the coefficients of the time domain inertial interference model. The learning rate and adjustment step size of the gradient descent method will be dynamically optimized according to the error distribution pattern to ensure the accuracy and stability of the adjustment process, and the spatial domain feature extraction adopts a sparse coding algorithm to analyze the sparsity of the pressure point distribution and the center of gravity geometric vector through the topological relationship of the sensor array. The time domain interference modeling uses a dynamic inertial interference model to model acceleration, velocity and its rate of change, calculate the dynamic noise component in real time, and use these noise components to predict future dynamic interference trends. Through this method, the static load and the dynamic interference signal can be effectively decoupled, which belongs to an extended implementation method known to ordinary technicians in this field.

[0038] The dynamic basis function library is updated in real time during the implementation process. Whenever 100 sets of mapping relationships between vehicle motion states and weighing data are added, the global optimization process of the basis function library is triggered. During the optimization process, the common characteristics of the historical basis function groups are retained, and only the feature vectors that conflict with the newly added data are adjusted. This optimization process can ensure zero-cost rapid adaptation of new vehicle types without increasing the hardware cost of the system. During the implementation of the entire system, the following specific optimization paths are adopted to ensure the accuracy and efficiency of the algorithm, that is, the sparse coding algorithm is used to extract static load components in the spatial domain analysis to ensure that the signal is decoupled without being affected by the spectrum aliasing problem in the traditional filtering algorithm. The time domain interference model can accurately predict and suppress dynamic interference signals in transient conditions by modeling the second-order equations of acceleration and velocity data.

[0039] When the system is running, all real-time data (acceleration, velocity, acceleration rate) are transmitted to the system from each sensor through a synchronization mechanism. Through pre-set models and algorithms, these data are analyzed in real time and necessary conversions are performed to ensure the time consistency of the data. The data of each sensor will undergo a decoupling and correction process in the system to ensure that each variable can directly affect the improvement of weighing accuracy. For example, the extraction of spatial domain features: through the relative position relationship of the sensor array (such as sensor spacing and pressure distribution gradient), the system extracts the spatial domain features of the static load. The extraction of these features takes precedence over the construction of the dynamic interference model to ensure that the static load can be accurately processed in time and space decoupling; the establishment of a dynamic inertial interference model: by combining the acceleration, velocity and posture information of the vehicle, a dynamic inertial interference model is constructed. This model identifies and predicts the trend of dynamic errors through time series data analysis, dynamically adjusts the weight coefficient, and ensures that the interference signal of each sensor data can be effectively eliminated, which belongs to an extended implementation method known to ordinary technicians in this field.

[0040] Embodiment 3: This embodiment provides a further optimized intelligent load prediction and adjustment method in a dynamic weighing system. In this embodiment, first ensure that the weighing sensor array and the motion state data (acceleration, speed, posture) of the vehicle are collected synchronously. The signal acquisition adopts a timing synchronization mechanism to ensure the timing consistency of the sensor data to avoid errors caused by data timing disorder. The output signal of each sensor will be transmitted to the signal processing module in real time, and the module will further process the signal to ensure the validity of each signal source.

[0041] Time-space dual-domain decoupling optimization: The optimization of the time-space dual-domain decoupling step includes two parts: space-domain analysis and time-domain interference modeling; Space-domain analysis: This embodiment uses a sparse coding algorithm to analyze the spatial characteristics of the static load through the topological relationship of the sensor array. Specifically, the sensor spacing and the pressure distribution gradient are used to calculate the geometric vector of the center of gravity position, and the spatial domain features are extracted in combination with the sparsity of the pressure point distribution. In this way, the spatial domain feature extraction takes precedence over the time-domain interference modeling, thereby ensuring that the static load signal can be accurately extracted and avoiding distortion caused by spectrum aliasing. Time-domain analysis: In the process of time-domain decoupling, the real-time motion state data (acceleration, velocity and attitude information) of the vehicle is combined to construct a dynamic inertial interference model. The model is based on the time series of acceleration and velocity, and the dynamic noise component is calculated through the second-order inertial interference equation. This step ensures that the noise component can be accurately modeled and interference suppression is performed by dynamically adjusting the weight coefficient in real time. Feedforward compensation and optimization of dynamic basis function library: Active correction of dynamic errors is achieved through feedforward compensation. In this embodiment, the dynamic basis function library uses a pre-training method to generate basic feature vectors to adapt to the load and motion state coupling characteristics of different vehicle types. The real-time motion state data (acceleration, speed, posture change) of each vehicle will be input into the dynamic basis function library, and the system will fine-tune the basis function through the transfer learning mechanism to adapt to the dynamic behavior of the new vehicle. Feedforward compensation signal generation: The compensation signal generated by the dynamic basis function library predicts the amplitude of the inertial interference in the next 5-10 milliseconds based on the acceleration change rate of the vehicle, and corrects the weighing signal by reverse superposition. This process plays a feedforward role in the system, effectively avoiding the weighing error caused by inertial interference and ensuring the stability of the system under transient conditions such as emergency stop and speed change. Closed-loop self-healing calibration and error correction: When the deviation between the corrected static load value and the preset threshold exceeds the limit, the system automatically enters the closed-loop self-healing calibration mode.

[0042] Spatial consistency detection: Through the spatial consistency characteristics of the sensor array, the system analyzes the distribution pattern of the residual error and locates the inertial interference component that is not completely decoupled. According to the results of the spatial consistency detection, the system will dynamically adjust the weight parameters of the spatiotemporal dual-domain decoupling to optimize the error suppression process. At this time, a lightweight online learning algorithm is used to adjust the weight parameters to ensure that the dynamic error of the system can be effectively suppressed. Dynamic adjustment of the gradient descent method: In order to improve the accuracy of the closed-loop calibration, the gradient descent method is used to dynamically adjust the weight parameters. By analyzing the distribution pattern of the residual error, the learning rate of the gradient descent method will be adjusted in real time according to the characteristics of the error to ensure the accuracy and stability of the error correction process. Update mechanism of the dynamic basis function library: In order to improve the adaptability and stability of the system, this embodiment introduces an update mechanism for the dynamic basis function library. When 100 sets of vehicle motion state and weighing data mapping relationships are added, the system triggers the global optimization of the basis function library. During the optimization process, the common characteristics of the historical basis function group are retained, and only the feature vectors that conflict with the newly added data are updated to ensure the rapid adaptation of the new vehicle type without increasing the hardware burden of the system.

[0043] Embodiment 4: This embodiment further explains the formula used in the dynamic inertial interference model: , in: : The time domain noise component is the final output of the interference signal; : The real-time acceleration data of the vehicle comes from the acceleration sensor of the weighing sensor array; : The real-time speed data of the vehicle is obtained through the speed sensor and has been filtered to remove static interference; : The acceleration change rate is obtained through the differential equation of the vehicle's motion state, reflecting the dynamic trend of the vehicle's speed change; the coefficient , , It is the inertial interference coefficient obtained by fitting the calibration data of the standard test vehicle to ensure that the model can accurately reflect the dynamic behavior of different vehicles. The value and calculation method of each coefficient are determined by regression analysis of experimental data to ensure that it is suitable for practical application. It is obtained by fitting the calibration data of standard test vehicles, including three typical vehicles: forklifts, AGVs (automatic guided vehicles) and conveyor belts. The standard for selecting these vehicles is based on their wide application in e-commerce warehousing and automated logistics. The calibration process includes measuring the acceleration, speed and posture changes of each vehicle under static and dynamic conditions. Data acquisition uses a high-precision inertial measurement unit (IMU) and a weighing sensor array to ensure that each collected data has high accuracy and low error. In the fitting process, for example, the least squares method can be used for regression analysis to match the collected motion state data (such as acceleration, speed and posture changes) with the weighing data to calculate the optimal inertial interference coefficient. Multiple cross-validations are used in the fitting process to avoid overfitting and ensure that the model can adapt to different dynamic load conditions. Finally, the inertial interference coefficient obtained reflects the inertial characteristics of the vehicle in a dynamic environment through the calibration data of the standard vehicle, and can effectively eliminate dynamic interference caused by acceleration changes, speed fluctuations, etc. in practical applications.

[0044] In practical applications, the inertial interference coefficient It will be used to correct the interference error in the dynamic weighing system, especially in transient conditions such as emergency stop and speed change of the vehicle. By applying these coefficients to the dynamic inertial interference model, the system can perform feedforward compensation based on real-time data collection, effectively improve the weighing accuracy and stabilize the output results of the system. The specific implementation method is as follows: Step 1: Signal acquisition. The signal acquisition module synchronously collects the dynamic load data of the vehicle through a high-precision weighing sensor array, and cooperates with the inertial measurement unit (IMU) to obtain acceleration, speed and attitude information. All data are synchronized through a timing synchronization mechanism to ensure timing consistency, thereby eliminating errors caused by data timing disorder.

[0045] Step 2: Time-space dual-domain decoupling. The spatial domain feature extraction in this step calculates the topological relationship of the sensor array, determines the local pressure distribution of each sensor, and further extracts the spatial domain features of the static load (such as the sparse distribution of pressure points and the center of gravity geometric vector). These spatial domain features are analyzed before the time domain interference modeling to ensure the accurate extraction of static load data and avoid distortion caused by spectrum aliasing. The time domain interference modeling analyzes the acceleration and velocity data by constructing a second-order inertial interference equation to calculate the dynamic noise component. The key to this part is to calculate the trend of the interference signal through real-time acceleration data and velocity change rate, and then provide input for the feedforward compensation module.

[0046] Step 3: Feedforward compensation. During the feedforward compensation process, future interference trends are predicted based on the previous dynamic inertial interference model. Through the pre-trained dynamic basis function library, the system can generate targeted compensation signals for different vehicle types (such as forklifts, AGVs, conveyor belts, etc.). This compensation signal is reversely superimposed on the real-time output signal to correct the static load value, thereby eliminating the impact of inertial interference. The generation of compensation signals relies on prediction models, which are based on the kinematic characteristics of the vehicle and historical load data, and are continuously fine-tuned and optimized through transfer learning algorithms to adapt to the dynamic behavior of the new vehicle.

[0047] Step 4: Closed-loop self-healing calibration. When the deviation of the corrected static load value is detected to be out of limit, the system will trigger the closed-loop self-healing calibration mechanism. This mechanism is based on the spatial consistency characteristics of the sensor array and uses the residual error distribution pattern to locate the incompletely decoupled inertial interference components. The gradient descent method is used to adjust the weight parameters of the spatiotemporal dual-domain decoupling to achieve dynamic error suppression. This calibration step can not only be adjusted in real time according to the spatial consistency detection of the sensor array, but also through the parameters of the online learning optimization process to ensure that the error in long-term operation remains within an acceptable range. These are all extended implementation methods known to ordinary technicians in this field.

[0048] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An intelligent load prediction and adjustment method in a dynamic weighing system, characterized in that: The following steps are involved: S1, signal acquisition: obtaining the real-time output signal of the weighing sensor array and synchronously receiving the motion state data of the vehicle, the motion state data including acceleration, speed and posture information; S2, time-space dual domain decoupling: Spatial domain analysis: based on the spatial distribution characteristics of the load cell array, extract the spatial domain characteristics of the static load, the spatial domain characteristics include the sparse distribution of pressure points and the geometric vector of the center of gravity position; Time domain analysis: According to the real-time motion state data of the vehicle, a dynamic inertial interference model is constructed to generate a time domain noise component; the dynamic inertial interference model is expressed as: ， in, is the noise component in the time domain, For real-time acceleration data, For real-time speed data, is the rate of change of acceleration, , , The inertial disturbance coefficient is fitted by the calibration data of the standard test vehicle; S3, feedforward compensation: matching the type of the vehicle through the pre-trained dynamic basis function library, predicting the interference evolution trend at future moments based on the dynamic inertial interference model, generating a feedforward compensation signal, and reversely superimposing the feedforward compensation signal to the real-time output signal to obtain a corrected static load value; S4, closed-loop self-healing calibration: When it is detected that the corrected static load value deviates from the preset threshold value, the residual error distribution pattern is traced based on the spatial consistency characteristics of the sensor array, and the weight parameters of the spatiotemporal dual-domain decoupling are adjusted by the gradient descent method to complete dynamic error suppression.

2. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 1, characterized in that: In step S2: spatial domain analysis uses a sparse coding algorithm to separate the static load component in the weighing sensor array signal, wherein the sparse coding basis function is generated by typical load distribution experimental data; time domain analysis constructs a second-order inertial interference equation through the acceleration and velocity data of the vehicle to calculate the dynamic noise component in real time.

3. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 2, characterized in that: The pre-training method of the dynamic basis function library includes: extracting the common motion-load coupling characteristics of three types of vehicles, namely forklifts, AGVs and conveyor belts, through transfer learning to generate basic feature vectors; when adding a new vehicle type, fine-tuning the basic feature vector through online learning based on the mapping relationship between its 10-20 groups of real-time motion states and weighing data.

4. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 3, characterized in that: In the closed-loop self-healing calibration: the residual error distribution pattern is detected by the spatial consistency of the sensor array to locate the incompletely decoupled inertial interference component; a lightweight online learning algorithm is used to adjust the weight parameters of the spatiotemporal dual-domain decoupling, and the weight parameters include the basis function weights of the spatial domain sparse coding and the coefficients of the time domain inertial interference model.

5. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 1, characterized in that: The matching method of the dynamic basis function library includes: calling pre-trained forklift motion basis functions, AGV motion basis functions or conveyor belt motion basis functions according to the vehicle type; when the vehicle type is not pre-stored, matching the closest basis function group based on the spectrum characteristics of the real-time motion state data.

6. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 1, characterized in that: In the time-space dual-domain decoupling: the spatial domain feature extraction has a higher priority than the time domain interference modeling, and the decoupling operations of the two are completed synchronously within a 10ms period; the geometric vector of the spatial domain feature is calculated through the topological relationship of the sensor array, and the topological relationship includes the spacing between adjacent sensors and the pressure distribution gradient.

7. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 6, characterized in that: The method for generating the feedforward compensation signal includes: predicting the amplitude of the inertial interference component within the next 5-10ms according to the acceleration change rate of the vehicle; scaling the predicted value through the migration adaptation coefficient of the dynamic basis function library to generate a compensation signal matching the current vehicle.

8. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 7, characterized in that: The triggering condition of the closed-loop self-healing calibration is: the fluctuation amplitude of the corrected static load value exceeds 1.5 times of the preset threshold within three consecutive sampling periods; the learning rate of the gradient descent method is dynamically adjusted according to the distribution pattern of the residual error.

9. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 1, characterized in that: It also includes the update of the dynamic basis function library, where the update mechanism of the dynamic basis function library includes: every time 100 sets of mapping relationships between vehicle motion states and weighing data are added, the global optimization of the basis function library is triggered once; during the optimization process, the common characteristics of the historical basis function groups are retained, and only the feature vectors that conflict with the newly added data are updated.

10. The intelligent load prediction and adjustment method in a dynamic weighing system according to claim 1, characterized in that: The weight parameter initialization method of the space-time dual-domain decoupling is: based on the calibration data of the standard test vehicle, the initial weight is fitted by the least squares method; the standard test vehicle includes the benchmark motion load curves of three types of equipment: forklifts, AGVs and conveyor belts.

Images