A dynamic weighing error self-correction system based on multi-modal edge collaboration

Abstract

The present invention relates to the field of dynamic weighing technology, and discloses a dynamic weighing error self-correction system based on multi-modal edge collaboration, including: a multi-modal data acquisition module, a spatio-temporal alignment module, a dynamic error propagation modeling module, and a collaborative calibration and suppression module. The system realizes the precise positioning of the error source and the whole-process tracking of the propagation path through the real-time construction of the vibration energy transfer topology map and the dynamic determination of the path priority. Combining with the piezoelectric ceramic reverse vibration wave neutralization technology, it directly blocks the error diffusion and solves the industry problems of fuzzy error tracing and compensation lag in traditional solutions; through the collaborative design of vibration wave speed correction and dynamic interpolation of temperature signals, it realizes the spatio-temporal consistency compensation of cross-modal data and avoids compensation logic conflicts in the cold chain logistics temperature mutation scenario; based on the intelligent sleep of edge nodes and the self-sensing mechanism of hidden faults, it ensures the long-term stable operation of the system in the scenarios of continuous weighing on highways and high industrial loads.

Description

Technical Field

[0001] The present invention relates to a dynamic weighing error self - calibration system based on multi - modal edge collaboration, and belongs to the technical field of dynamic weighing. Background Art

[0002] Dynamic weighing is widely used in scenarios such as highway toll - by - weight, cold - chain logistics cargo monitoring, and industrial production line quality control. Traditional dynamic weighing systems rely on strain - type sensors to collect weight signals and use algorithms such as temperature compensation and vibration filtering to eliminate environmental interference. However, under complex working conditions, the coupling effect of multiple physical fields causes the weighing error to exhibit dynamic propagation characteristics, and the existing technologies have the following limitations: 1. Existing systems usually perform independent calibration or global mean compensation on individual sensor nodes. However, vibration energy is transmitted between multiple nodes through the mechanical structure, resulting in the error still spreading along a fixed path after calibration. For example, a local temperature mutation caused by the opening and closing of the cold - storage door in cold - chain logistics will be transmitted to adjacent weighing units through the metal bracket, triggering a chain - like reading jump. 2. To improve accuracy, the industry generally adopts multi - sensor (vibration, temperature, weighing) data fusion technology. However, the difference in the acquisition timing of high - frequency vibration signals (1000 Hz) and low - frequency temperature signals (10 Hz) leads to a mismatch between the update of the temperature compensation coefficient and the vibration phase. In the weighing scenario of a chemical reaction kettle, this asynchronous compensation will cause periodic oscillations in the readings, requiring frequent manual intervention. 3. The mainstream solution adopts a threshold - triggered calibration strategy, but does not consider the dynamic change of the error propagation path. For example, in the highway dynamic weighing system during peak traffic flow, fixed - period calibration causes 90% of the edge nodes to consume computing power during inactive periods, while the calibration of key nodes lags behind, resulting in the interruption of the continuity of weighing data.

[0003] In addition, traditional methods assume that the mechanical connection state of the weighing platform is constant. However, during long - term use, bolt loosening and micro - cracks in welding will change the vibration transmission path. In the dynamic weighing scenario of a port crane, such hidden faults will cause sudden weighing inaccuracy, and the fault tracing takes up to several hours. To improve the above problems, the industry has tried to improve accuracy by increasing sensor density, optimizing filtering algorithms, etc. However, sensor redundancy exacerbates data conflicts, and complex algorithms increase the edge - computing burden. Summary of the Invention

[0004] The present invention provides a dynamic weighing error self - calibration system based on multi - modal edge collaboration, and its main purpose is to solve the problems of uncontrollable error propagation path, multi - modal data compensation conflict, and inefficient calibration resource allocation in dynamic weighing.

[0005] To achieve the above object, a dynamic weighing error self - calibration system based on multi - modal edge collaboration provided by the present invention includes a multi - modal data acquisition module: configured to collect vibration signals of a weighing platform through a vibration sensor array at a first sampling frequency, collect temperature signals through a temperature sensor array at a second sampling frequency, and collect weighing signals through a weighing sensor array at a third sampling frequency;

[0006] A spatio - temporal alignment module: connected to the multi - modal data acquisition module and configured to: generate an interpolated temperature data stream by performing cubic spline interpolation on the temperature signal, where the interpolation node interval is dynamically adjusted according to the main frequency of the vibration signal, and the main frequency is determined by spectral analysis when the weighing platform runs without load; calculate the propagation time difference of the vibration signal from the source node to the target node based on the vibration wave velocity of the weighing platform material and the spatial position coordinates of the sensor nodes, and align the sampling moments of the vibration signal and the weighing signal with the time difference as the offset; output a spatio - temporally synchronized multi - modal data stream to the dynamic error propagation modeling module;

[0007] A dynamic error propagation modeling module: connected to the spatio - temporal alignment module and configured to calculate the vibration energy attenuation rate between nodes in real - time based on a predefined mechanical connection topology. The calculation formula for the vibration energy attenuation rate is: , where, is the integral value of the power spectrum of the vibration signal of the source node, calculated by performing Fourier transform on the vibration signal of the source node in the spatio - temporally synchronized multi - modal data stream; is the integral value of the power spectrum of the vibration signal of the target node, and the calculation method is the same as that of ; is the spatial distance between the source node and the target node; when the coherence coefficient of the vibration signal between nodes is detected to be lower than a preset threshold, update the vibration coupling weight in the mechanical connection topology; generate a vibration propagation path priority list sorted in ascending order of the energy attenuation rate according to the updated vibration coupling weight and the energy attenuation rate; output the vibration propagation path priority list to the collaborative calibration and suppression module;

[0008] Collaborative Calibration and Suppression Module: Connected to the dynamic error propagation modeling module, configured to query the vibration propagation path priority list when the signal deviation of a weighing sensor exceeds a preset percentage threshold of its range. If the sensor is the starting node of the current highest priority path, it is marked as an error source node; trigger the self-calibration program of the error source node, where the self-calibration program includes disconnecting the sensor power supply to reset the zero point and updating the sensitivity compensation coefficient according to the temperature data in the spatio-temporal synchronized multimodal data stream; send a suppression instruction to all downstream nodes of the error source node, and the duration of the suppression instruction is dynamically calculated based on the maximum energy decay time of the propagation path, and drive the piezoelectric ceramic sheet of the downstream node to output a reverse vibration wave with a phase opposite to that of the interference vibration wave to neutralize the vibration energy transmitted to this node.

[0009] As a preferred embodiment of the present invention, the first sampling frequency is high-frequency sampling, with a frequency range of 800 Hz to 1200 Hz, the third sampling frequency is medium-frequency sampling, with a frequency range of 80 Hz to 150 Hz, and the second sampling frequency is low-frequency sampling, with a frequency range of 5 Hz to 20 Hz.

[0010] As a preferred embodiment of the present invention, the update logic of the vibration coupling weight includes: the initial weight is 1.0, indicating a rigid connection; when the power value of the vibration signal rises by more than 25% in a preset frequency band and the coherence coefficient drops to 0.2 to 0.4, the weight is adjusted down to 0.4 to 0.6; when the coherence coefficient is lower than 0.2, the weight is reset to zero and the connection relationship between nodes is removed.

[0011] As a preferred embodiment of the present invention, the time alignment error of the spatio-temporal alignment module is less than 1 millisecond, and the alignment rule includes: using the rising edge of the weighing signal as a reference, applying a programmable delay to the vibration signal; the delay step accuracy is 0.05 millisecond to 0.2 millisecond.

[0012] As a preferred embodiment of the present invention, the zero point reset operation of the self-calibration program includes: disconnecting the weighing sensor power supply for 3 milliseconds to 10 milliseconds; performing linear interpolation compensation on the weighing signal during the pause period after calibration, and the length of the interpolation window does not exceed 5 sampling periods.

[0013] As a preferred embodiment of the present invention, the amplitude error control range of the reverse vibration wave is 3% to 8%, the phase control accuracy is ±5°, and the duration is dynamically adjusted according to the maximum energy decay time of the propagation path, and the adjustment coefficient is 1.1 to 1.3 times.

[0014] As a preferred embodiment of the present invention, the method for generating a driving signal for a piezoelectric ceramic sheet includes: extracting the time-domain waveform of the currently propagating vibration wave and performing a Hilbert transform to obtain the instantaneous phase; generating a reverse driving voltage signal after increasing the phase by 180°, and pre-compensating the signal amplitude through the calibrated frequency-amplitude curve of the piezoelectric ceramic sheet, with the compensated amplitude error being less than 5%.

[0015] As a preferred embodiment of the present invention, the self-calibration program triggering conditions include: the weighing platform is in an unloaded state or the static load fluctuation is less than 5% of the rated value; when the dynamic load changes exceed the threshold, calibration is immediately terminated and historical valid parameters are restored.

[0016] As a preferred embodiment of the present invention, the dynamic update rule for the main frequency of the vibration signal is: when the load change of the weighing platform exceeds 30% of the rated value, the no-load spectrum analysis is re-executed; the adjustment range of the interpolation node interval after the main frequency update does not exceed 5% of the historical value.

[0017] Compared with the problems described in the background art, the beneficial effects of the present invention are as follows: For scenarios such as error accumulation during continuous weighing on highways or instantaneous weighing distortion caused by sudden temperature changes in cold chain logistics, through the real-time construction of the vibration energy transfer topology map and the dynamic determination of path priorities, the precise positioning of the error source and the full tracking of the propagation path are realized. Combining the piezoelectric ceramic reverse vibration wave neutralization technology, the error diffusion is directly blocked at the mechanical structure level, solving the problems of fuzzy error tracing and compensation lag in traditional dynamic weighing. The temperature drift compensation is deeply bound to the vibration phase change, realizing the millisecond-level self-consistency of the compensation logic in scenarios with sudden temperature changes such as cold chain logistics, avoiding the vicious cycle of compensation conflicts and precision oscillations caused by data asynchrony in traditional multi-sensor systems; Based on the edge node cooperation mechanism with propagation path priorities, the system can autonomously identify key paths and dynamically allocate calibration resources, and during long-term use, through the detection of abnormal resonance peaks in the vibration spectrum and the dynamic update of connection weights, when hidden faults such as bolt loosening and structural micro-damage occur, the system automatically reconstructs the propagation path model and triggers an early warning mechanism, enabling the error correction system to always be synchronized with the true state of the physical device, greatly improving the long-term operation reliability in industrial scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is the system block diagram of the present invention;

[0019] Figure 2 is the schematic diagram of signal and data flow of the present invention;

[0020] Figure 3 is the core architecture diagram of the sensor collaborative processing system of the present invention.

[0021] The realization of the object, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0022] 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.

[0023] The embodiment of the present application provides a dynamic weighing error self-correction system based on multi-modal edge collaboration. It includes a multi-modal data acquisition module: configured to collect vibration signals of the weighing platform through a vibration sensor array at a first sampling frequency, collect temperature signals through a temperature sensor array at a second sampling frequency, and collect weighing signals through a weighing sensor array at a third sampling frequency, where the first sampling frequency is higher than the third sampling frequency, and the third sampling frequency is higher than the second sampling frequency; the spatial position coordinates of the vibration sensor array, temperature sensor array, and weighing sensor array are pre-stored in the system memory;

[0024] A spatio-temporal alignment module: connected to the multi-modal data acquisition module, configured to: perform cubic spline interpolation on the temperature signal to generate an interpolated temperature data stream, and the interpolation node interval is dynamically adjusted according to the main frequency of the vibration signal, and the main frequency is determined by spectral analysis when the weighing platform runs without load; based on the vibration wave velocity of the weighing platform material and the spatial position coordinates of the sensor nodes, calculate the propagation time difference of the vibration signal from the source node to the target node, and use the time difference as the offset to align the sampling moments of the vibration signal and the weighing signal; output the spatio-temporally synchronized multi-modal data stream to the dynamic error propagation modeling module;

[0025] A dynamic error propagation modeling module: connected to the spatio-temporal alignment module, configured to calculate the vibration energy attenuation rate between nodes in real time based on a predefined mechanical connection topology, and the calculation formula of the vibration energy attenuation rate is: , where is the integral value of the power spectrum of the vibration signal of the source node, which is calculated by performing Fourier transform on the vibration signal of the source node in the spatio-temporally synchronized multi-modal data stream; is the integral value of the power spectrum of the vibration signal of the target node, and the calculation method is the same as ; is the spatial distance between the source node and the target node, in meters; when the coherence coefficient of the vibration signal between nodes is detected to be lower than the preset threshold, update the vibration coupling weight in the mechanical connection topology; according to the updated vibration coupling weight and energy attenuation rate, generate a vibration propagation path priority list sorted in ascending order of energy attenuation rate; output the vibration propagation path priority list to the collaborative calibration and suppression module;

[0026] Collaborative Calibration and Suppression Module: Connected to the dynamic error propagation modeling module, configured to query the vibration propagation path priority list when the signal deviation of a certain load cell exceeds the preset percentage threshold of its range. If the sensor is the starting node of the current highest-priority path, it is marked as an error source node; trigger the self-calibration program of the error source node, and the self-calibration program includes disconnecting the sensor power supply to reset the zero point and updating the sensitivity compensation coefficient according to the temperature data in the spatio-temporal synchronized multimodal data stream; send a suppression instruction to all downstream nodes of the error source node, and the duration of the suppression instruction is dynamically calculated according to the maximum energy decay time of the propagation path, and drive the piezoelectric ceramic sheet of the downstream node to output a reverse vibration wave with a phase opposite to that of the interference vibration wave to neutralize the vibration energy transmitted to this node.

[0027] As a preferred embodiment of the present invention, the first sampling frequency is high-frequency sampling, and its frequency range is 800 Hz to 1200 Hz. The third sampling frequency is medium-frequency sampling, and its frequency range is 80 Hz to 150 Hz. The second sampling frequency is low-frequency sampling, and its frequency range is 5 Hz to 20 Hz.

[0028] As a preferred embodiment of the present invention, the update logic of the vibration coupling weight includes: the initial weight is 1.0, indicating a rigid connection; when the power value of the vibration signal rises by more than 25% in the preset frequency band and the coherence coefficient drops to 0.2 to 0.4, the weight is adjusted down to 0.4 to 0.6; when the coherence coefficient is lower than 0.2, the weight is reset to zero and the connection relationship between nodes is removed.

[0029] As a preferred embodiment of the present invention, the time alignment error of the spatio-temporal alignment module is less than 1 millisecond, and the alignment rule includes: taking the rising edge of the weighing signal as a reference, applying a programmable delay to the vibration signal; the delay step accuracy is 0.05 millisecond to 0.2 millisecond, and it is implemented by FPGA hardware.

[0030] As a preferred embodiment of the present invention, the zero point reset operation of the self-calibration program includes: disconnecting the power supply of the load cell for 3 milliseconds to 10 milliseconds; after calibration, performing linear interpolation compensation on the weighing signal during the pause period, and the length of the interpolation window does not exceed 5 sampling periods.

[0031] As a preferred embodiment of the present invention, the amplitude error control range of the reverse vibration wave is 3% to 8%, the phase control accuracy is ±5°, and the duration is dynamically adjusted according to the maximum energy decay time of the propagation path, and the adjustment coefficient is 1.1 to 1.3 times.

[0032] As a preferred embodiment of the present invention, the method for generating a driving signal of a piezoelectric ceramic sheet includes: extracting the time-domain waveform of the currently propagating vibration wave and performing a Hilbert transform to obtain the instantaneous phase; generating a reverse driving voltage signal after increasing the phase by 180°, and pre-compensating the signal amplitude through the calibrated frequency-amplitude curve of the piezoelectric ceramic sheet, and the amplitude error after compensation is less than 5%.

[0033] As a preferred embodiment of the present invention, the triggering conditions for the self-calibration program include: the weighing platform is in an unloaded state or the static load fluctuation is less than 5% of the rated value; when the dynamic load change exceeds the threshold, the calibration is immediately terminated and the historical valid parameters are restored.

[0034] As a preferred embodiment of the present invention, the dynamic update rule of the main frequency of the vibration signal is: when the load change of the weighing platform exceeds 30% of the rated value, the no-load spectrum analysis is re-executed; the adjustment range of the interpolation node interval after the main frequency update does not exceed 5% of the historical value. The persistent storage rule of the mechanical connection topology of the system is: the current topology map is stored in the non-volatile memory of the edge node every 20 seconds to 60 seconds; the stored data includes node coordinates, directed edge weights, and timestamps, and the data format is binary encoding or JSON array.

[0035] See Figure 1 — Figure 3 , Figure 1 is the system block diagram of the dynamic weighing error self-correction system based on multi-modal edge collaboration of the present invention. The figure shows the main modules of the system, including: a multi-modal data acquisition module, a spatio-temporal alignment module, a dynamic error propagation modeling module, and a collaborative calibration and suppression module. The multi-modal data acquisition module respectively acquires vibration signals, temperature signals, and weighing signals through a vibration sensor array, a temperature sensor array, and a weighing sensor array. The spatio-temporal alignment module is connected to the multi-modal data acquisition module, responsible for interpolating the temperature signal to generate a temperature data stream, and dynamically adjusting the interval of the interpolation nodes according to the main frequency of the vibration signal to ensure the spatio-temporal synchronization of various signals. The dynamic error propagation modeling module is based on the mechanical connection topology, calculates the vibration energy attenuation rate between nodes in real time, updates the vibration coupling weight, and generates a vibration propagation path priority list. The collaborative calibration and suppression module receives the vibration propagation path priority list, triggers the self-calibration program when detecting an error source node, and suppresses the interference signal through the reverse vibration wave to complete the error correction. Figure 2This is a schematic diagram of signal and data flow for the dynamic weighing error self - calibration system based on multi - modal edge collaboration of the present invention. The figure shows the flow process of signals and data among various modules. First, signals from the vibration sensor array, temperature sensor array, and weighing sensor array are collected by the multi - modal data acquisition module and transmitted to the spatio - temporal alignment module according to their respective sampling frequencies. The spatio - temporal alignment module performs interpolation processing on the temperature signal and dynamically adjusts according to the main frequency of the vibration signal to ensure the spatio - temporal synchronization of the temperature signal, vibration signal, and weighing signal. The synchronized data stream is transmitted to the dynamic error propagation modeling module, which calculates the vibration energy attenuation rate, updates the vibration coupling weight, and generates a vibration propagation path priority list. Finally, the data stream is sent to the collaborative calibration and suppression module, which, according to the priority list, triggers the self - calibration process after detecting the error source node and suppresses the interference signal through the reverse vibration wave technology, thus realizing error correction and suppression. See Figure 3 This is the core architecture diagram of the sensor collaborative processing system of the present invention, which includes three key components: the error source node, the vibration sensor node, and the weighing sensor node. Among them, the error source node receives the vibration spectrum characteristics from the vibration sensor node through the real - time data interaction channel and synchronously sends the error compensation parameters to the weighing sensor node. The vibration sensor node is responsible for collecting the three - axis vibration waveforms of the mechanical structure. After receiving the compensation parameters, the weighing sensor node performs dynamic load calibration. The three form an error suppression topology network through a closed - loop feedback mechanism to accurately realize vibration source identification, error transfer modeling, and multi - modal data collaborative processing.

[0036] Example 1: In this example, the multi - modal data acquisition module collects vibration signals, temperature signals, and weighing signals through the vibration sensor array, temperature sensor array, and weighing sensor array respectively. The sampling frequency of each sensor array is dynamically adjusted according to the actual working conditions. That is, for the vibration sensor array: according to the change of system load, the sampling frequency is dynamically adjusted to ensure higher - frequency vibration data at higher loads and lower the sampling frequency at no - load or lighter loads. In this way, the system can reduce the unnecessary data processing burden while ensuring data accuracy. For the temperature sensor array: the sampling frequency is relatively low, usually 5Hz to 20Hz, but in an environment with sudden temperature changes, the temperature sensor will enhance its data acquisition accuracy to avoid weighing errors caused by temperature fluctuations. In this way, the system can more effectively process sensor data and can be adjusted according to the working conditions in practical applications, thus improving the stability and reliability of the system.

[0037] The dynamic error propagation modeling module determines the vibration propagation path by calculating the vibration energy attenuation rate between nodes. In this process, in the calculation formula of the attenuation rate is the integral value of the power spectrum of the vibration signal of the source node, which is obtained by Fourier transform and normalized based on a predetermined sensor sensitivity during all calculations. is the integral value of the power spectrum of the vibration signal of the target node, and the calculation method is the same as that of to ensure that the vibration differences between the source node and the target node can be compared during actual operation. is the physical distance between the source node and the target node, which is determined by the spatial coordinates of the sensor. The spatial distances between all nodes are pre-stored during system initialization to ensure the accuracy of subsequent calculations. This model ensures the degree of vibration coupling between each node and can update the connection weights between nodes according to the actual situation to further improve the accuracy of system error correction.

[0038] Example 2: In the dynamic error propagation modeling module of the system in this example, it involves the formula for the vibration energy attenuation rate: , where: is the integral value of the power spectrum of the vibration signal of the source node, which is derived from the Fourier transform of the vibration signal of the source node. The specific calculation method is obtained through the vibration signal of the spatio-temporal synchronous data stream; is the integral value of the power spectrum of the vibration signal of the target node, which is calculated by Fourier transform and is consistent with the calculation method of the source node; is the spatial distance between the source node and the target node, with the unit of meter, which can be calculated by the spatial difference of the sensor position coordinates.

[0039] In this example, the spatio-temporal alignment method of the vibration signal, temperature signal, and weighing signal is specifically described. Especially the accuracy of temperature signal interpolation. The cubic spline interpolation method is used to generate the interpolated temperature data stream, and its node interval is dynamically adjusted according to the main frequency of the vibration signal. The main frequency is obtained through the spectral analysis during the no-load operation of the weighing platform, and the temperature signal interpolation interval is adjusted in combination with the actual data of the vibration signal. During this process, it is ensured that the accuracy error of the temperature signal interpolation does not exceed 0.2 milliseconds, and the interpolated temperature data stream for each interpolation is re-calibrated to cope with external temperature changes. In the dynamic error propagation modeling, when constructing the dynamic error propagation path, a more refined weight update rule is used. For example, when the ratio of the power spectrum of the vibration signal to the power spectrum of the target node signal shows a significant change and the coherence coefficient decreases, the vibration coupling weight will be down-regulated. This update mechanism not only depends on the power difference between signals but also combines the vibration coupling state between nodes for real-time adjustment, ensuring the dynamic adaptability of the calibration path. This move avoids the problems of error accumulation and compensation lag caused by static vibration paths in traditional methods.

[0040] To ensure that the self-calibration process can be accurately executed under high load conditions, this embodiment further optimizes the zero reset and calibration strategies. For example, when it is found that the signal deviation of a certain load cell exceeds the preset threshold, the system will automatically enter the self-calibration mode. During this process, the system will first cut off the power supply of the error source node and perform a power-off for 3 milliseconds to 10 milliseconds to reset its zero point. In addition, during calibration, for the weighing signals during the pause period, the system adopts a linear interpolation algorithm to compensate for the weighing signal deviation during this period, and the length of the interpolation window does not exceed 5 sampling periods. In this way, the high precision and fast response of calibration are ensured. And the application path of the vibration wave reverse neutralization technology is to ensure that the driving signal of the piezoelectric ceramic sheet can generate an anti-phase reaction with the vibration wave. The specific method is to extract the time-domain waveform of the currently propagating vibration wave, perform a Hilbert transform to obtain the instantaneous phase, and generate a reverse driving voltage signal through phase reversal. The signal amplitude is compensated according to the calibrated frequency-amplitude curve to ensure that the amplitude error after compensation remains between 3% and 8%, and the phase error is controlled within ±5°. And in order to cope with potential hidden faults that may occur during long-term operation, this embodiment introduces an abnormal resonance peak detection technology based on the vibration spectrum. When an abnormality appears in the vibration signal, the system will automatically update the vibration coupling weight and recalculate the propagation path. This mechanism can effectively avoid changes in the vibration transmission path caused by the aging or micro-cracks of mechanical components, thereby improving the long-term stability and reliability of the system.

[0041] Embodiment 3: The core of this embodiment is to process the error correction problem of the weighing platform through multi-modal data acquisition and spatio-temporal alignment. For each sensor (vibration, temperature, and weighing), different sampling frequencies are used for data acquisition, and different signals are aligned through the spatio-temporal alignment module to achieve precise fusion of multi-modal data. Specifically, the multi-modal data acquisition module: Each sensor (vibration sensor, temperature sensor, weighing sensor) performs data acquisition through an independent sampling frequency. The vibration sensor samples at a high frequency (for example, 800 Hz to 1200 Hz), the temperature sensor samples at a low frequency (for example, 5 Hz to 20 Hz), and the weighing sensor samples at a medium frequency (for example, 80 Hz to 150 Hz) for data acquisition. Such a sampling frequency setting helps to fully capture different signals and provides accuracy guarantee for subsequent data processing; The spatio-temporal alignment module: The task of this module is to align the temperature signal with the vibration signal and the weighing signal. Specifically, the temperature signal generates a temperature data stream through the cubic spline interpolation method and dynamically adjusts the interpolation node interval. The adjustment of the interpolation node is based on the main frequency of the vibration signal. To ensure the time alignment of all signals, the main frequency of the vibration signal can be determined according to the spectral analysis when the weighing platform is unloaded, and the temperature data interpolation interval is adjusted based on this frequency. These are all extended implementation methods known to those of ordinary skill in the art.

[0042] In the process of error propagation modeling, this embodiment optimizes the calculation of the attenuation of vibration energy between nodes. For example, the calculation of vibration energy attenuation: For each pair of sensor nodes and , we use the following formula to calculate the vibration energy attenuation rate: , where is the power spectrum integral value of the vibration signal of the source node, is the power spectrum integral value of the vibration signal of the target node, is the spatial distance between the source node and the target node. The power spectra of the source node and the target node are calculated through Fourier transform. The power spectrum integral value reflects the intensity of the signal. is the physical distance between nodes, which can be calculated through the spatial coordinates of the sensors. The purpose of this formula is to quantify the attenuation degree of the vibration signal during propagation. In practical applications, during the calculation of the power spectrum integral values and , based on the spatial position coordinates of each node during system initialization, these coordinate information can be stored in advance, and the attenuation rate between each pair of nodes can be calculated in combination with the spatial distance. The coupling weight between nodes is dynamically adjusted through the updated vibration attenuation rate. Its weight update mechanism: At each moment, if the coherence coefficient of the vibration signal between nodes is lower than the preset threshold, the vibration coupling weight may be dynamically adjusted. Initially, the weight between nodes is set to 1.0, indicating a rigid connection; when the change amplitude of the power spectrum of the vibration signal exceeds the set threshold (such as a difference of 25%), and the coherence coefficient drops to between 0.2 and 0.4, the weight between nodes will be adjusted down to 0.4 to 0.6; when the coherence coefficient is lower than 0.2, the connection relationship between nodes will be removed.

[0043] The self-calibration process is initiated when the system detects an error source node. In the original calibration logic, more refined control steps for the calibration process are added to ensure that the zero reset operation can still be completed quickly and accurately under high-load conditions. Zero reset and interpolation compensation: When the deviation of a certain sensor is detected to exceed the preset threshold of its range, the system will first perform self-calibration on the sensor. The self-calibration process will first reset the zero point by powering off and dynamically update the sensitivity compensation coefficient according to the temperature data in the multimodal data stream. For the data missing during the power-off of the sensor, the system will use the linear interpolation method for compensation, and the length of the interpolation window is limited between 5 sampling periods to ensure the continuity and accuracy of the data during calibration. And this embodiment further optimizes the usage method of the reverse vibration wave. When the error source node is identified, the system will send a suppression instruction to its downstream node and output a reverse vibration wave with a phase opposite to that of the interference vibration wave through the piezoelectric ceramic sheet to neutralize the vibration energy. The amplitude error control range of this process is set to 3% to 8%, the phase control accuracy is ±5°, and the duration is dynamically adjusted according to the maximum energy attenuation time of the propagation path, and the adjustment coefficient is set to 1.1 to 1.3 times. To further improve the long-term stability and anti-interference ability of the system, this embodiment improves the fault detection and automatic reconstruction mechanism of the system, such as latent fault detection and path reconstruction: By introducing the vibration spectrum abnormal resonance peak detection technology, the system can automatically detect latent faults (such as bolt loosening, structural micro-cracks, etc.) when the vibration signal appears abnormally and automatically update the vibration propagation path model. This mechanism effectively improves the stability of the system during long-term use, avoids the error accumulation and system failures caused by mechanical damage, and can send fault reports in an extended implementation manner, which all belong to the extended implementation manners known to those of ordinary skill in the art.

[0044] Embodiment 4: This embodiment specifically defines the calculation method of the power spectrum and its background. The integral value of the power spectrum and are obtained through Fourier transform. The specific steps are as follows: sample the vibration signals of each node, and the sampling frequency is dynamically adjusted according to the specific environment of the node to ensure the capture accuracy of the vibration signals under high load; use Fourier transform to convert the collected time-domain vibration signals into frequency-domain signals to obtain the power spectrum of the vibration signals; and are the integral values of the power spectra of the source node and the target node respectively, and the calculation formula is: , where is the power spectrum of the vibration signal, which represents the power distribution of the signal at different frequency components. In this system, the power spectral density is obtained after the vibration signal undergoes Fourier transform and is used as a parameter to describe the change of vibration intensity with frequency; and They are the minimum and maximum frequency values of the signal spectrum (e.g., minimum 50 Hz, maximum 200 Hz), and the parameters in the formula represents the physical distance between nodes. Usually, during system initialization, the sensors are spatially located, and the coordinate information of the nodes is pre-stored. The distance between nodes is dynamically calculated according to the spatial position coordinates of the sensors during the data stream transmission process.

[0045] The goal of the spatio-temporal alignment module is to ensure the accurate synchronization of data from different sensors, especially in the alignment of temperature, vibration, and weighing signals. In this embodiment, the interpolation accuracy control of the temperature signal is further optimized. Based on the no-load spectrum analysis of the weighing platform, the main frequency of the vibration signal is obtained, and the interpolation node interval of the temperature signal is dynamically adjusted according to the main frequency. Specifically, the main frequency of the vibration signal is recalculated whenever the load changes, and the updated main frequency value will dynamically adjust the time node interval of the temperature signal interpolation. The interpolation method of the temperature signal uses cubic spline interpolation. This interpolation method can effectively reduce the interpolation error and ensure the smoothness and accuracy of the temperature data stream, ensuring that the spatio-temporal alignment error between low-frequency and high-frequency signals is less than 1 millisecond. Through this dynamic adjustment mechanism, the interpolation of temperature data can effectively avoid data timing conflicts in temperature mutation scenarios and improve the overall stability and accuracy of the system. During the self-calibration process, when the signal deviation of a certain weighing sensor exceeds the preset percentage threshold, the system will automatically trigger the self-calibration program. And in this embodiment, the missing part of the data during calibration will be compensated using the linear interpolation method. For example, during the power-off period of the sensor, the system will perform interpolation compensation based on the previous and subsequent data points to ensure that the continuity of the data stream is not affected during a short power-off period of 3 milliseconds to 10 milliseconds. The length of the interpolation window does not exceed 5 to 15 sampling periods. In this way, the signal gaps caused by power-off during the calibration process can be effectively eliminated, ensuring the real-time performance of the system and the integrity of the data. In addition, after calibration, the system will dynamically update the temperature compensation coefficient according to the spatio-temporal synchronized data stream, thereby further improving the accuracy and stability of self-calibration. And in this embodiment, the generation and adjustment mechanism of the reverse vibration wave is specifically that the amplitude error control range of the reverse vibration wave is 3% to 8%, and the phase error is controlled within ±5°. This control accuracy ensures that the phase of the reverse vibration wave is opposite to that of the interference wave by real-time monitoring the vibration states of each node in the system, suppressing the propagation of interference signals. The duration is dynamically calculated according to the maximum energy attenuation time of the vibration propagation path and is appropriately adjusted through an adjustment coefficient (usually 1.1 to 1.3 times). This adjustment coefficient is optimized according to the different characteristics of the propagation path to improve the suppression effect of the reverse vibration wave.

[0046] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A dynamic weighing error self-correction system based on multi-modal edge collaboration, characterized in that It includes a multi-modal data acquisition module: configured to collect vibration signals of a weighing platform through a vibration sensor array at a first sampling frequency, collect temperature signals through a temperature sensor array at a second sampling frequency, and collect weighing signals through a weighing sensor array at a third sampling frequency; A spatio-temporal alignment module: connected to the multi-modal data acquisition module, configured to: generate an interpolated temperature data stream by performing cubic spline interpolation on the temperature signal, and dynamically adjust the interpolation node interval according to the main frequency of the vibration signal, where the main frequency is determined by spectral analysis when the weighing platform runs without load; calculate the propagation time difference of the vibration signal from the source node to the target node based on the vibration wave velocity of the weighing platform material and the spatial position coordinates of the sensor nodes, and align the sampling moments of the vibration signal and the weighing signal with the time difference as the offset; output a spatio-temporally synchronized multi-modal data stream to the dynamic error propagation modeling module; Dynamic error propagation modeling module: Connected to the spatio-temporal alignment module, configured to calculate the vibration energy attenuation rate between nodes in real time based on a predefined mechanical connection topology. The calculation formula for the vibration energy attenuation rate is: , where is the integral value of the power spectrum of the vibration signal of the source node, calculated by performing a Fourier transform on the vibration signal of the source node in the spatio-temporally synchronized multimodal data stream; is the integral value of the power spectrum of the vibration signal of the target node, and the calculation method is the same as that of ; is the spatial distance between the source node and the target node; when the coherence coefficient of the vibration signal between nodes is detected to be lower than a preset threshold, update the vibration coupling weight in the mechanical connection topology; generate a vibration propagation path priority list sorted in ascending order of the energy attenuation rate according to the updated vibration coupling weight and energy attenuation rate; output the vibration propagation path priority list to the collaborative calibration and suppression module; A collaborative calibration and suppression module: connected to the dynamic error propagation modeling module, configured to, when the signal deviation of a certain weighing sensor exceeds a preset percentage threshold of its range, query the vibration propagation path priority list, and if the sensor is the starting node of the current highest-priority path, mark it as an error source node; trigger the self-calibration program of the error source node, where the self-calibration program includes disconnecting the sensor power supply to reset the zero point and updating the sensitivity compensation coefficient according to the temperature data in the spatio-temporally synchronized multi-modal data stream; send a suppression instruction to all downstream nodes of the error source node, where the duration of the suppression instruction is dynamically calculated according to the maximum energy decay time of the propagation path, and drive the piezoelectric ceramic sheet of the downstream node to output a reverse vibration wave with a phase opposite to that of the interfering vibration wave to neutralize the vibration energy transmitted to this node.

2. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, characterized in that The first sampling frequency is high-frequency sampling, with a frequency range of 800 Hz to 1200 Hz, the third sampling frequency is medium-frequency sampling, with a frequency range of 80 Hz to 150 Hz, and the second sampling frequency is low-frequency sampling, with a frequency range of 5 Hz to 20 Hz.

3. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, characterized in that, The update logic of the vibration coupling weight includes: the initial weight is 1.0, indicating a rigid connection; when the power value of the vibration signal rises by more than 25% in a preset frequency band and the coherence coefficient drops to 0.2 to 0.4, the weight is reduced to 0.4 to 0.6; when the coherence coefficient is lower than 0.2, the weight is reset to zero and the connection relationship between nodes is removed.

4. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, wherein The time alignment error of the spatio-temporal alignment module is less than 1 millisecond, and the alignment rule includes: taking the rising edge of the weighing signal as a reference, applying a programmable delay to the vibration signal; the delay step accuracy is 0.05 millisecond to 0.2 millisecond.

5. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, wherein, The zero point reset operation of the self-calibration program includes: disconnecting the weighing sensor power supply for 3 milliseconds to 10 milliseconds; performing linear interpolation compensation on the weighing signal during the pause period after calibration, and the interpolation window length does not exceed 5 sampling periods.

6. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, characterized in that, The amplitude error control range of the reverse vibration wave is 3% to 8%, the phase control accuracy is ±5°, and the duration is dynamically adjusted according to the maximum energy decay time of the propagation path, and the adjustment coefficient is 1.1 to 1.3 times.

7. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, characterized in that, The method for generating a driving signal of a piezoelectric ceramic sheet includes: extracting the time-domain waveform of the currently propagating vibration wave and performing Hilbert transform to obtain the instantaneous phase; generating a reverse driving voltage signal after increasing the phase by 180°, and pre-compensating the signal amplitude through the calibrated frequency-amplitude curve of the piezoelectric ceramic sheet, and the amplitude error after compensation is less than 5%.

8. The dynamic weighing error self-correction system based on multi-modal edge collaboration according to claim 1, wherein The triggering conditions for the self-calibration program include: the weighing platform is in an unloaded state or the static load fluctuation is less than 5% of the rated value; when the dynamic load changes exceed the threshold, the calibration is immediately terminated and the historical valid parameters are restored.

9. The dynamic weighing error self - correction system based on multi - modal edge collaboration according to claim 1, characterized in that, The dynamic update rule for the main frequency of the vibration signal is: when the load change of the weighing platform exceeds 30% of the rated value, the no-load spectrum analysis is re-executed; the adjustment range of the interpolation node interval after the main frequency update does not exceed 5% of the historical value.

Images