A method for weighing a CAN-based shaft pin type load cell

Abstract

The application relates to the technical field of weighing sensor signal processing, and discloses a CAN-based shaft pin type weighing sensor weighing method.The method comprises the following steps: acquiring structural parameters and performing double-bridge redundant sensor positioning and symmetrical arrangement; collecting strain signals and converting the strain signals into CAN compatible data packets; performing differential coding and anti-interference transmission processing on the data packets; performing dynamic filtering and purification on the signals; compensating and verifying the signals by fusing temperature data; and finally calculating the load according to a calibration model and generating storage records.The application effectively improves the signal anti-interference capability and measurement precision, and realizes high reliability and high stability dynamic weighing.

Description

Technical Field

[0001] This invention relates to the field of weighing sensor signal processing technology, and in particular to a weighing method for a CAN-based pin-type weighing sensor. Background Technology

[0002] In the field of weighing sensor signal processing technology, existing solutions related to CAN-based pin-type weighing sensors typically employ a single-bridge sensor combined with a fixed filtering algorithm. These solutions suffer from limitations such as susceptibility to electromagnetic interference in signal transmission, low efficiency in redundant data fusion, and poor adaptability to dynamic environments. Existing methods largely rely on traditional encoding methods and static threshold filtering, which are prone to signal distortion and baud rate inaccuracies in strong electromagnetic interference scenarios, making it difficult to meet the high-precision weighing requirements of dual-bridge redundant pin-type sensors. For the joint processing of dual-bridge redundant sensor signals and CAN bus communication, existing technologies generally suffer from high differential signal encoding redundancy, insufficient noise spectrum feature extraction, and lagging dynamic updates of filtering coefficients. This makes it difficult to form a continuous processing flow of signal acquisition—encoding—transmission—filtering—compensation in complex electromagnetic environments, resulting in a decrease in the signal-to-noise ratio and insufficient real-time performance of the weighing signal. Summary of the Invention

[0003] This invention provides a CAN-based method for weighing load cells with pin-type load cells, which solves the problem of how to achieve high-precision transmission and processing of weighing signals in environments with strong electromagnetic interference by using a dual-bridge redundant pin-type load cell to communicate with the CAN bus and by employing an improved Shannon entropy coding and dynamic filtering algorithm.

[0004] To address the aforementioned technical problems, this invention provides a CAN-based method for weighing load cells using a pin-type load cell, comprising:

[0005] The installation parameters of the load-bearing structure are obtained, and the positioning processing of the dual-bridge redundant sensors is performed. Based on the symmetrical arrangement of the redundant bridge circuits according to finite element analysis, the installation configuration parameters are formed. The installation parameters of the load-bearing structure include the axle pin size, the thickness of the load-bearing plate, the load distribution range, and the mechanical constraints of the installation environment. The positioning processing includes determining the spatial coordinates of the installation position of the dual-bridge redundant sensors, analyzing the symmetry of the redundant bridge circuits, and verifying the compatibility of the mechanical interfaces.

[0006] Acquire strain signals from dual-bridge redundant sensors, perform 24-bit high-precision analog-to-digital conversion and frame formatting, add CRC checksum, and generate CAN protocol compatible data packets;

[0007] Acquire CAN protocol compatible data packets, perform differential signal encoding processing, shielded twisted pair impedance matching and CRC verification, and generate anti-interference transmission signals;

[0008] The anti-interference transmission signal is acquired, and time-domain noise feature analysis, dynamic filter coefficient calculation based on wavelet packet energy analysis, and sliding window convolution processing are performed to generate a clean digital signal.

[0009] Acquire purified digital signals, perform temperature sensor data fusion based on Kalman filtering, multinomial fitting compensation and threshold comparison, and generate verification data;

[0010] Once the verification data is obtained, the load distribution model calibrated by finite element analysis is used for calculation, unit conversion, and precision rounding to generate and store records.

[0011] Furthermore, the installation parameters of the load-bearing structure are obtained, dual-bridge redundant sensor positioning processing is performed, and the redundant bridge circuits are symmetrically arranged based on finite element analysis to form installation configuration parameters, specifically including:

[0012] Strain signals of the load-bearing structure are acquired using dual-bridge redundant sensors;

[0013] The dual-bridge redundant sensor includes two bridge circuits, which are respectively arranged in the key stress areas of the load-bearing structure to achieve redundant signal acquisition;

[0014] The dual-bridge redundant sensor is tightly integrated with a pin-type weighing structure through a high-precision strain gauge.

[0015] Furthermore, strain signals from dual-bridge redundant sensors are acquired, and 24-bit high-precision analog-to-digital conversion, frame formatting, and CRC checksum addition are performed to generate CAN protocol-compatible data packets, specifically including:

[0016] Simultaneous acquisition of strain signals from both bridges;

[0017] The synchronous acquisition process is based on a hardware trigger signal, and the sampling timestamps of the two bridges are consistent;

[0018] It features a high-bandwidth analog front-end circuit, including a constant current source excitation and a differential input amplifier, to suppress common-mode interference.

[0019] Furthermore, acquiring strain signals from the dual-bridge redundant sensor also includes:

[0020] Real-time monitoring of the voltage fluctuation range of the sensor output and automatic adjustment of the sampling window;

[0021] The built-in diagnostic module detects abnormal signals such as disconnection, short circuit, or over-limit voltage and records the abnormal event log.

[0022] Furthermore, a 24-bit high-precision analog-to-digital conversion is performed, specifically including:

[0023] It employs a 24-bit high-resolution ADC module, which features low noise and high linearity.

[0024] The input range of the ADC is dynamically adjusted by sampling the maximum and minimum values ​​in real time.

[0025] Furthermore, the process includes performing 24-bit high-precision analog-to-digital conversion and frame formatting, as well as adding CRC checksums.

[0026] The data is encapsulated into a digital signal frame according to a preset data structure, which includes sampling timestamp, bridge number, sample value and status identification information.

[0027] Further, differential signal encoding processing is performed, specifically including:

[0028] CAN differential coding technology is used to convert single-ended digital signals into differential signals;

[0029] By logical flipping and encoding mapping, the corresponding differential level sequence is generated.

[0030] Furthermore, differential signal encoding processing also includes:

[0031] Insert the necessary compensation bits according to the bit filling rules defined in the CAN protocol.

[0032] Furthermore, CAN protocol-compatible data packets are acquired, and differential signal encoding processing is performed to obtain the expression for the differential encoded sequence, which includes:

[0033]

[0034] in, This is a coding efficiency metric. For the first CAN data packet One bit; This refers to the length of the data packet. Let be the statistical probability of the k-th bit; This is the bit error rate metric for the k-th bit. This refers to the sampling frequency parameter;

[0035] Further, a differential level sequence is generated:

[0036]

[0037] in, This represents the differential voltage amplitude. This refers to the packet identifier weight coefficient; This is the initial value of the characteristic impedance of the transmission line; The differential voltage amplitude of the k-th bit; It is the hyperbolic tangent function.

[0038] Furthermore, the expression for obtaining the anti-interference transmission signal, performing time-domain noise feature analysis, dynamic filter coefficient calculation, and sliding window convolution to generate the purified digital signal also includes:

[0039] Constructing frequency domain feature vectors:

[0040]

[0041] in, Wavelet packet energy characteristics; The wavelet decomposition level; Number the frequency sub-bands; This represents the total number of frequency sub-bands. For the first Layer wavelet coefficients of the sub-band; For the first The signal-to-noise ratio of the subband; For the first The center frequency of the sub-band;

[0042] Furthermore, Input to the adaptive filter coefficient calculation model:

[0043]

[0044] in, These are the filter weight coefficients; , 2 represents the filter index; This represents the total number of filters; It is a natural exponential function; This is the temperature compensation factor; For the first The design bandwidth of each filter; Frequency resolution;

[0045] Establish a time-varying filtering model:

[0046]

[0047] in, For time indexing; This is the time offset; Let be the amplitude of the input signal at time t; The attenuation coefficient of the window function; The width of the sliding window; Let be the purified digital signal at time t.

[0048] The key innovations of this invention include:

[0049] (1) After acquiring the CAN data packet, differential signal encoding processing is performed to generate a differential encoding sequence, and the baud rate parameter is extracted to perform shielded twisted pair impedance matching to form transmission configuration parameters.

[0050] (2) By performing CRC verification on the transmission configuration parameters, an anti-interference transmission signal is generated, and time-domain noise characteristics are analyzed during signal transmission to extract the main interference frequency for dynamic filtering coefficient calculation.

[0051] (3) In the dynamic filtering process, the sliding window convolution technique is applied to process the real-time filtering parameters, generate the purified digital signal, and use it for subsequent temperature compensation processing.

[0052] The following are its main beneficial effects:

[0053] (1) By using differential signal encoding processing and shielded twisted pair impedance matching, the anti-interference capability of signal transmission can be effectively improved, ensuring the stability of signal transmission in a strong electromagnetic interference environment.

[0054] (2) Through CRC verification and time-domain noise feature analysis, the main interference frequency can be accurately identified and extracted, the calculation of dynamic filtering coefficients can be optimized, and the accuracy and real-time performance of signal processing can be improved.

[0055] (3) By processing the real-time filtering parameters through sliding window convolution technology, digital signals can be effectively purified in complex electromagnetic environments, ensuring the accuracy and reliability of subsequent temperature compensation. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating a CAN-based weighing method using a pin-type load cell, as provided in an embodiment of this application. Detailed Implementation

[0057] Example 1: Refer to Figure 1 This is a flowchart illustrating a CAN-based pin-type load cell weighing method according to an embodiment of the present invention. The process may include at least steps S100-S600:

[0058] S110. Obtain the installation parameters of the load-bearing structure, perform dual-bridge redundant sensor positioning processing, and form installation configuration parameters based on the symmetrical arrangement of redundant bridge circuits according to finite element analysis.

[0059] S200: Acquires strain signals from dual-bridge redundant sensors, performs 24-bit high-precision analog-to-digital conversion and frame formatting, adds CRC checksum, and generates CAN protocol compatible data packets;

[0060] S300: Acquire CAN protocol compatible data packets, perform differential signal encoding processing, shielded twisted pair impedance matching and CRC verification, and generate anti-interference transmission signals;

[0061] S400: Acquire anti-interference transmission signals, perform time-domain noise feature analysis, calculate dynamic filter coefficients based on wavelet packet energy analysis and perform sliding window convolution processing to generate purified digital signals;

[0062] S500: Acquires purified digital signals, performs temperature sensor data fusion based on Kalman filtering, polynomial fitting compensation and threshold comparison, and generates verification data;

[0063] S600: Obtain the verification data, calculate the load distribution model based on the finite element analysis calibration, perform unit conversion and precision rounding, and generate a storage record.

[0064] Step S100 includes at least steps S110-S130:

[0065] S110. Obtain the installation parameters of the load-bearing structure, perform dual-bridge redundant sensor positioning processing, and obtain the installation configuration parameters.

[0066] Specifically, the steps are based on pre-set installation parameters of the load-bearing structure as input. These parameters include, but are not limited to, pin dimensions, load plate thickness, load distribution range, and mechanical constraints of the installation environment. First, the geometric shape and key dimensions of the load-bearing structure are obtained using 3D scanning technology or by importing a CAD model, forming a preliminary installation parameter dataset. Further, based on the physical characteristics and installation requirements of the dual-bridge redundant sensors, a positioning process is performed. This positioning process includes determining the spatial coordinates of the dual-bridge redundant sensor installation position, analyzing the symmetry of the redundant bridge circuits, and verifying the compatibility of the mechanical interfaces to ensure a reasonable sensor layout on the load-bearing structure. Specifically, the mounting surface of the dual-bridge redundant sensors is matched with the contact surface of the load-bearing structure, and the stress concentration at the mounting point is estimated using finite element analysis to select the optimal sensor positioning scheme. This process also includes evaluating the impact of mechanical vibration and temperature changes during installation on sensor performance, forming comprehensive installation parameters. Further, the results of the above positioning process are integrated to generate complete installation configuration parameters, covering the sensor installation angle, fixing method, and layout scheme of the redundant bridge circuits. These installation and configuration parameters serve as output fields for this step, which will be used in subsequent sensor layout calls for S120 to achieve orderly data transfer and functional integration between modules.

[0067] S120. Extract mechanical fixing points from the installation configuration parameters, arrange redundant bridge circuits symmetrically, and generate sensor layout.

[0068] Based on the installation configuration parameters output by S110, the mechanical fixing point information is further extracted, including the coordinates of the mounting holes, the size of the fixing bolts, and their relative positional relationships. Specifically, the mechanical fixing points are used as a reference to perform a symmetrical arrangement of redundant bridge circuits. This symmetrical arrangement of redundant bridge circuits ensures an equivalent spatial distribution by calculating the strain sensing direction and sensitivity of each bridge circuit of the dual-bridge redundant sensor. To this end, a geometric mapping algorithm is used to map the mechanical fixing points to the installation positions of the sensor bridge circuits, while simultaneously verifying the consistency of the installation angle and load direction of each bridge circuit. Furthermore, for the arrangement of redundant bridge circuits, a reasonable wiring scheme is designed in conjunction with the electrical connection requirements of the sensors to avoid signal line crosstalk and mechanical stress concentration. The arrangement process also includes planning electrical isolation and shielding measures between redundant bridge circuits to improve signal reliability. After completing the above arrangement, a complete sensor layout data structure is generated, which records in detail the spatial coordinates, fixing methods, and electrical connection information of each bridge circuit. This sensor layout is used as an output field of this step for S130 to optimize the fixing structure, achieving spatial and functional coordination of sensor installation.

[0069] S130. Perform stress distribution analysis on the sensor layout and generate a fixed structure;

[0070] After receiving the sensor layout data output by S120, stress distribution analysis is first performed on the mounting points of each bridge circuit in the sensor layout. Specifically, finite element analysis software is used to model the load-bearing structure and sensor mounting locations, inputting the spatial coordinates of the sensor layout and the mechanical fixing method to simulate the stress state under actual load. During the analysis, the uniformity of stress on the sensor bridge circuits and potential stress concentration areas are emphasized, and the stiffness and stability of the fixing structure are evaluated. Further, based on the stress distribution results, the fixing structure design scheme is adjusted, including the shape, size, and installation material selection of the fasteners, and the mechanical connection method is optimized to disperse stress concentration. This optimization process also considers thermal expansion, vibration effects, and long-term fatigue performance during installation to ensure the durability of the fixing structure and the stability of the sensor signal. After completing the optimization design, the final fixing structure is generated, and the manufacturing process parameters and assembly sequence of the fasteners are defined. The fixing structure serves as the output field of this step, used for the acquisition of load-bearing structure parameters and the use of sensor signal acquisition in the subsequent S200 module, achieving seamless integration from mechanical design to signal acquisition.

[0071] Step S200 includes at least steps S210-S230:

[0072] S210. Obtain the load-bearing structural parameters in the fixed structure, perform synchronous acquisition of dual-bridge strain signals, and obtain the original analog signal;

[0073] Specifically, the steps take the fixed structure output by S130 as input. First, strain signals of the load-bearing structure are acquired using a dual-bridge redundant sensor installed on the fixed structure. The dual-bridge redundant sensor includes two bridge circuits, respectively arranged in the key stress areas of the load-bearing structure to achieve redundant signal acquisition. The dual-bridge redundant sensor is tightly integrated with a pin-type weighing structure through high-precision strain gauges, enabling it to sense minute deformations caused by loads. Further, the dual-bridge strain signals are acquired synchronously. This synchronous acquisition process is based on a hardware trigger signal to ensure that the sampling timestamps of the two bridge circuits are consistent, avoiding signal distortion due to time deviations. Specifically, the synchronous acquisition is configured with a high-bandwidth analog front-end circuit, including a constant current source excitation and a differential input amplifier, to suppress common-mode interference and improve signal quality. During the acquisition process, the system monitors the voltage fluctuation range of the sensor output in real time and automatically adjusts the sampling window to prevent signal overload or saturation. For abnormal signals such as open circuits, short circuits, or over-limit voltages, the system detects them through a built-in diagnostic module and records the abnormal event log for subsequent maintenance and analysis. After the dual-bridge strain signal is synchronously acquired, an original analog signal containing the voltage outputs of the two bridges is formed. This signal has high timeliness and integrity, and serves as the output field "original analog signal" of this step. It is then used by the analog-to-digital conversion module of S220 for digital processing to achieve continuous data transmission and functional connection.

[0074] S220: Extract the voltage fluctuation range from the original analog signal, perform 24-bit high-precision analog-to-digital conversion, and generate a digital signal frame;

[0075] Based on the raw analog signal output from the S210, the voltage fluctuation range is first analyzed, and a 24-bit high-precision analog-to-digital conversion (ADC) is performed to determine the appropriate ADC input range and gain configuration. Specifically, the system uses a 24-bit high-resolution ADC module, which features low noise and high linearity, suitable for accurate sampling of weak signals. The voltage fluctuation range is calculated using the maximum and minimum values ​​sampled in real time, dynamically adjusting the ADC input range to avoid quantization errors and signal truncation. The ADC sampling process is clock-synchronized to ensure strict correspondence between the sampling points of the dual-bridge signals, maintaining signal synchronization and consistency. After sampling, the digital signal is filtered to remove high-frequency noise and interference components, and a digital filtering algorithm is used to improve the signal-to-noise ratio. Further, the dual-bridge digital signals are frame-formatted and encapsulated into digital signal frames according to a preset data structure, containing information such as sampling timestamps, bridge numbers, sampled values, and status identifiers. This digital signal frame is managed through internal buffering, supporting continuous output of real-time data streams and ensuring data integrity and timing continuity. The system marks and triggers an alarm mechanism for abnormal data during the conversion process, such as sample loss or overflow, and records detailed logs. Finally, the digital signal frame serves as the output field of this step, used by S230 for subsequent CRC checks and CAN protocol-compatible data packet generation, achieving reliable digital signal transmission and protocol adaptation.

[0076] S230: Add CRC checksum to the digital signal frame to generate a CAN protocol compatible data packet;

[0077] After receiving the digital signal frame output from S220, the system first performs Cyclic Redundancy Check (CRC) encoding on its content. Specifically, based on the data length and protocol specifications of the digital signal frame, the system calculates a checksum using a standard CRC algorithm and appends it to the end of the data frame to form a complete checksum field. This CRC checksum is used for error detection in subsequent data transmission, ensuring the integrity of the data packet in environments with strong electromagnetic interference. Further, the data frame with the CRC checksum is encapsulated according to the Controller Area Network (CAN) protocol format to generate a CAN protocol-compatible data packet. The encapsulation process includes setting the data packet's identifier (ID), data length code (DLC), data fields, and control fields to ensure the data packet conforms to the CAN bus communication standard. The data packet is encapsulated through a software protocol stack, supporting multiple transmission priorities and error handling mechanisms. After encapsulation, the system performs integrity verification on the data packet, verifying the consistency between the CRC code and the data content to ensure the encapsulation process is error-free. In abnormal situations, the system automatically retransmits the erroneous data packet and records the transmission status log. Finally, the CAN protocol-compatible data packet is used as the output field of this step for S310 to perform differential signal encoding processing, complete the subsequent CAN bus differential transmission, and realize a seamless connection from digital signal acquisition to bus transmission.

[0078] Step S300 includes at least steps S310-S330:

[0079] S310. Obtain CAN protocol compatible data packets, perform differential signal encoding processing, and obtain differential encoded sequences;

[0080] Specifically, the steps involve taking the CAN protocol-compatible data packet output by the S230 as input. First, the data packet is parsed to extract its data and control fields. The CAN data packet conforms to the Controller Area Network (CAN) protocol standard and includes an identifier (ID), a data length code (DLC), data fields, and a CRC checksum. During parsing, the data packet is read byte-by-byte using a software protocol stack to confirm its integrity and conformity to the specifications. Further, differential signal encoding is performed based on the binary data content of the data packet. Specifically, CAN differential encoding technology is used to convert single-ended digital signals into differential signals to enhance anti-interference capabilities. The CAN differential encoding technology generates a corresponding differential level sequence by logically flipping and encoding each bit in the original data stream, where logic "1" and logic "0" correspond to the voltage difference between CAN_H (high level line) and CAN_L (low level line), respectively. During encoding, the system inserts necessary compensation bits according to the bitstuffing rules defined by the CAN protocol to prevent prolonged unchanged levels from causing clock synchronization failure at the receiving end. This process is completed by a dedicated CAN controller chip in the embedded microcontroller, with built-in hardware supporting differential encoding and bit stuffing operations to ensure real-time performance and accuracy of the encoding. Furthermore, the encoding module monitors the transmission priority and error flags of the input data packets, re-encoding or marking abnormal data. After encoding, a differentially encoded sequence containing differential level sequences is generated. This sequence conforms to the CAN physical layer standard and enables high-reliability transmission on the differential bus. This differentially encoded sequence serves as the output field of this step, available for use by the S320's transmission configuration module to configure and match subsequent physical layer signal transmission, ensuring stable data transmission and interference resistance.

[0081] In another embodiment, based on the CAN protocol-compatible data packet output by S230, differential signal encoding processing is first performed. The CAN data packet contains an identifier, a data field, and a CRC checksum field, and the binary data stream is extracted through protocol parsing. An improved Shannon entropy encoding calculation is then performed on the data stream, defined by formula ①:

[0082]

[0083] in:

[0084] : Coding efficiency metric, used to quantify the anti-interference capability of differential coding;

[0085] For the first CAN data packet Each bit has a value range of [1, n].

[0086] The data packet length (in bits) is derived from the DLC field of the CAN data packet.

[0087] The statistical probability of the k-th bit is calculated from the historical transmission error rate of the data packet.

[0088] The bit error rate index for the k-th bit is calculated from the voltage margin of the anti-interference transmission signal;

[0089] The sampling frequency parameter is derived from the baud rate parameter;

[0090] Furthermore, the calculation result of formula ① is input into the geometric mapping algorithm to generate a differential level sequence:

[0091]

[0092] in:

[0093] Differential voltage amplitude;

[0094] The packet identifier weight coefficient is determined by the CAN protocol priority field;

[0095] The initial value of the characteristic impedance of the transmission line is derived from the shielded twisted pair parameter library;

[0096] The differential voltage amplitude of the k-th bit is used as the output element of the differential coded sequence;

[0097] : Hyperbolic tangent function.

[0098] The output field "Differential Coded Sequence" in this step is called by the S320's transmission configuration module, and its data structure includes time-aligned... Sequences and associations Encoding efficiency parameters.

[0099] S320. Extract the baud rate parameter from the differential coded sequence, perform shielded twisted pair impedance matching, and generate transmission configuration parameters.

[0100] Based on the differential encoding sequence output by the S310, the baud rate parameters implicit in the sequence are first extracted and analyzed. These baud rate parameters define the transmission rate of the CAN bus and typically include key timing parameters such as bit time, synchronization segment, propagation segment, and phase buffer segment. Through timing characteristic analysis of the differential encoding sequence, the system identifies the current transmission baud rate setting, serving as the basis for subsequent physical layer configuration. Further, based on the baud rate parameters, impedance matching design of the shielded twisted pair is performed. Specifically, considering the characteristic impedance of the shielded twisted pair (typically 120 ohms), and combining the frequency characteristics of the differential signal and the transmission distance, matching network parameters are calculated, including component values ​​such as terminating resistance, distributed capacitance, and inductance. The matching design employs transmission line theory and network analysis methods to ensure minimal signal reflection, reduced transmission loss, and intact signal waveform. Further, based on the intensity of environmental electromagnetic interference (EMI) and the shielding effect of the transmission environment, the shielding layer grounding method and grounding impedance are adjusted to optimize anti-interference capabilities. This impedance matching process is implemented through a hardware configuration interface, allowing dynamic adjustment of matching parameters to adapt to different transmission conditions and wiring lengths. After completing the impedance matching design, transmission configuration parameters are generated, including baud rate settings, termination matching resistor values, shielding grounding schemes, and transmission line parameters. These transmission configuration parameters serve as output fields for the S330's signal transmission module, used for driving and transmitting physical layer signals to ensure efficient and stable transmission of differential coded sequences on the bus.

[0101] S330. Perform CRC verification on the transmission configuration parameters and generate an anti-interference transmission signal;

[0102] Specifically, the steps take the transmission configuration parameters output by the S320 as input. First, a Cyclic Redundancy Check (CRC) verification is performed on this parameter set to confirm the integrity and correctness of the configuration data. The CRC verification uses a standard polynomial compatible with the CAN protocol, covering all key fields of the transmission configuration parameters, including baud rate, terminating resistor, shielding grounding scheme, and impedance matching component parameters. During verification, the system calculates the CRC code through an embedded verification module and compares it with the checksum in the configuration parameters. If a mismatch is found, an error handling mechanism is triggered, an exception log is recorded, and a reconfiguration request is initiated. After successful verification, an anti-interference transmission signal for physical layer transmission is generated based on the transmission configuration parameters. This signal is converted into differential voltage signals on the CAN_H and CAN_L lines by a differential driver, with the specific driving level and timing strictly adhering to the CAN physical layer standard. During the generation of the anti-interference transmission signal, the output impedance and level amplitude of the driver are adjusted in conjunction with the electrical characteristics of the shielded twisted pair to minimize signal reflection and crosstalk. The signal transmission module incorporates a temperature compensation circuit and an electromagnetic compatibility (EMC) filter to further enhance signal stability. The system monitors the voltage waveform and timing of the transmitted signal in real time, capturing abnormal waveforms via a built-in oscilloscope to support fault diagnosis and maintenance. The anti-interference transmission signal serves as the output field for this step, accessible to the S400's signal input module for subsequent dynamic filtering and noise suppression, completing a high-quality conversion and transmission loop from digital data to physical signals.

[0103] Step S400 includes at least steps S410-S430:

[0104] S410. Acquire the anti-interference transmission signal, perform time-domain noise characteristic analysis, and obtain the noise spectrum diagram;

[0105] Specifically, the steps involve taking the anti-interference transmission signal output from the S330 as input. First, the high-speed sampling module acquires time-domain data from this signal. The anti-interference transmission signal includes the differential voltage waveforms of the CAN_H and CAN_L lines. The sampling frequency is set at several MHz to capture high-frequency and transient interference components in the signal. During sampling, the system uses synchronous clock control to ensure time consistency and data continuity at sampling points, avoiding the impact of sampling jitter on subsequent analysis. The acquired time-domain signal data undergoes preprocessing, including DC component removal and baseline drift correction, to eliminate the influence of ambient temperature changes and slow changes in transmission line characteristics on the signal baseline. Further, a time-domain noise feature analysis algorithm is employed, specifically including autocorrelation function calculation and short-time energy analysis, to distinguish between random noise and periodic interference components in the signal. This analysis is implemented using a sliding window technique, with the window length and step size dynamically adjusted according to the actual noise spectrum characteristics to balance time and frequency resolution. Furthermore, wavelet transform is combined to perform multi-scale decomposition of the time-domain signal, extracting noise features from different frequency bands and enhancing the ability to identify non-stationary noise. During the analysis, the system marks abnormal sudden interference events and records the interference timestamp and amplitude through an event triggering mechanism for subsequent filter parameter adjustment. The noise feature analysis results are presented in the form of a spectrum, reflecting the energy distribution and interference intensity of the signal at different frequencies. This spectrum is generated by the digital signal processing unit and includes a three-dimensional data structure with frequency, amplitude, and time axes, supporting parameter extraction for subsequent dynamic filtering algorithms. The noise spectrum serves as the output field of this step, which is called by the algorithm loading module of the S420 to realize the calculation and generation of dynamic filtering parameters, completing the continuous processing link from physical signal acquisition to noise feature extraction.

[0106] S420. Extract the main interference frequency from the noise spectrum diagram, perform dynamic filter coefficient calculation, and generate real-time filter parameters.

[0107] Based on the noise spectrum output by the S410, the first step involves automatically identifying the energy peaks in the spectrum and using a peak detection algorithm to locate the main interference frequency. This algorithm combines threshold judgment and frequency neighborhood comparison to eliminate background noise and secondary frequency components, accurately selecting the frequency bands that have the greatest impact on signal quality. During the identification process, frequency drift and multipath interference effects are considered, and the estimated value of the main interference frequency is dynamically adjusted through time series analysis. Further, dynamic filter coefficient calculation is performed for the main interference frequency. The calculation method is based on adaptive filtering theory, combining a Kalman filter and a Least Mean Squares (LMS) algorithm to adjust the filter coefficients in real time to match the current noise environment. Specifically, a Finite Impulse Response (FIR) filter is used, and the coefficients are dynamically updated through an iterative optimization algorithm to ensure the stability and response speed of the filtering performance under time-varying interference. During the calculation, the system considers parameters such as the filter order, cutoff frequency, and stopband attenuation, balancing filtering accuracy and computational resource consumption through a multi-objective optimization method. Furthermore, based on the time-varying characteristics of the noise spectrum, the update frequency of the filter parameters is dynamically adjusted to achieve real-time adaptation of the filter parameters. To prevent the filter from over-responding to occasional interference, a robust control mechanism is introduced, combining historical filter coefficients and error feedback to suppress drastic fluctuations in the filter coefficients. Finally, a real-time filter parameter data packet containing the filter coefficient vector, filter structure parameters, and update timestamps is generated. This data packet is transmitted to the signal purification module via a high-speed bus, supporting continuous filtering operations. The real-time filter parameters serve as the output field of this step, available for use by the S430's signal purification module to achieve efficient filtering of interference signals, completing the closed-loop control from noise feature identification to filter parameter generation.

[0108] S430: Perform sliding window convolution processing on the real-time filtering parameters to generate a purified digital signal;

[0109] Specifically, the steps involve taking the real-time filtering parameters output by S420 as input and combining them with the digital representation of the anti-interference transmission signal output by S330 to perform sliding window convolution processing, specifically, convolutional filtering based on a sliding window. The sliding window convolution processing first segments the input signal according to a preset window length. The window length is dynamically adjusted based on the filter order and signal sampling rate to ensure the timeliness and stability of the filter response. Within each window, the filter coefficient vector from the real-time filtering parameters is used to perform convolution calculations on the signal. Specifically, the filter coefficients are multiplied point-by-point with the signal samples within the window and summed to obtain the filtered signal value. This convolution process is hardware-accelerated by a Digital Signal Processor (DSP), supporting high-speed real-time processing. To avoid boundary effects, the system uses symmetrical expansion or zero-padding techniques to process window edge data, ensuring the continuity and distortion-free nature of the filtered output. During the filtering process, the system monitors the signal amplitude and spectral characteristics of the filter output in real time and dynamically adjusts the filter coefficients using an error feedback mechanism to further optimize the filtering effect. For abnormal signal segments, such as sudden interference or signal loss, the system identifies them through the anomaly detection module and triggers rapid reconfiguration of filter parameters to maintain the continuity of filtering performance. After convolutional filtering, a continuous purified digital signal is generated, exhibiting significant noise suppression and signal fidelity. This purified digital signal is output as a digital data stream, including a timestamp, signal amplitude, and filtering status identifier, supporting subsequent calls from the S510's temperature compensation and data verification modules, achieving high-quality conversion from physical to digital signals. This purified digital signal serves as the output field of this step, used for S510 temperature compensation, completing the closed-loop data transmission of the dynamic filtering algorithm processing module.

[0110] In another embodiment, the main interference frequency is extracted based on the noise spectrum generated by S410. A frequency domain feature vector is constructed using an improved wavelet packet energy analysis method.

[0111]

[0112] in:

[0113] Wavelet packet energy characteristics are used to identify the main interference frequency band;

[0114] The wavelet decomposition level is [1, 6], and its value range is determined by the resolution of the noise spectrum.

[0115] The frequency sub-bands are numbered based on the sampling rate of the anti-interference transmission signal;

[0116] Total number of frequency subbands;

[0117] For the first Layer The wavelet coefficients of the subband are derived from the time-frequency matrix of the noise spectrum.

[0118] For the first The signal-to-noise ratio of a subband can be derived from the power spectral density ratio in the spectrum.

[0119] : No. The center frequency of the sub-band.

[0120] Furthermore, The input to the adaptive filter coefficient calculation model is defined in formula ③ as follows:

[0121]

[0122] in:

[0123] : Filter weight coefficients (the first weight coefficient of the dynamic filter) (each weighting coefficient)

[0124] , 2: Filter Index;

[0125] Total number of filters;

[0126] : Natural exponential function;

[0127] This is the temperature compensation factor;

[0128] For the first The design bandwidth of each filter;

[0129] The frequency resolution is derived from the number of FFT (Fast Fourier Transform) points in the noise spectrum.

[0130] The output field "Real-time Filtering Parameters" in this step is called by the signal purification module of S430. Its data entity includes the weight coefficient matrix generated according to formula ③ and the associated frequency band division rules.

[0131] Furthermore, a time-varying filtering model is established by performing a sliding window convolution on the real-time filtering parameters:

[0132]

[0133] in:

[0134] Time index;

[0135] Time offset;

[0136] The amplitude of the input signal at time t is derived from the sampling sequence of the anti-interference transmission signal;

[0137] The window function attenuation coefficient is derived from the baud rate in the transmission configuration parameters.

[0138] The width of the sliding window is equal to two CAN data packet transmission cycles.

[0139] The purified digital signal at time t;

[0140] Technical effects of this section: Improved Shannon entropy coding enhances the anti-interference capability of differential signals; dynamic noise suppression is achieved by combining wavelet packet energy analysis and adaptive filtering; and finally, a high-fidelity purified signal is generated through a time-varying convolution model.

[0141] Step 500 includes at least steps S510-S530:

[0142] S510: Acquire the purified digital signal, perform temperature sensor data fusion, and obtain the temperature compensation coefficient;

[0143] Specifically, the steps involve using the purified digital signal output from the S430 as input, while simultaneously receiving ambient temperature data collected by a temperature sensor installed near the load cell. First, the purified digital signal and the temperature sensor data are time-aligned, employing a hardware clock synchronization mechanism to ensure consistent timestamps for both types of data, avoiding compensation errors caused by sampling time differences. Further, a multi-sensor data fusion algorithm is used to perform weighted averaging of the multi-point measurement data from the temperature sensor, eliminating the influence of local temperature fluctuations and forming a stable ambient temperature input. The temperature sensor data fusion process employs a Kalman filter method, combining historical temperature data and current sampled values ​​to dynamically adjust the temperature estimation results, improving the accuracy and timeliness of temperature measurement. Further, based on the ambient temperature, a preset temperature compensation model is invoked. This model, based on the temperature response characteristics of the pivot-type load cell, uses polynomial fitting or lookup table interpolation to calculate the temperature compensation coefficient. During the calculation process, the system considers the temperature gradient, the thermal expansion coefficient of the sensor material, and the temperature sensitivity characteristics of the resistance strain gauge, forming comprehensive compensation parameters. The compensation coefficient is generated using a real-time calculation mechanism, which dynamically responds to changes in ambient temperature, ensuring the continuity and stability of the compensation. To prevent abnormal temperature data from affecting the compensation effect, the system sets a reasonable temperature range threshold. When the temperature exceeds this range, an alarm is triggered, and the most recent valid compensation coefficient is used as a replacement, ensuring the robustness of the compensation process. The temperature compensation coefficient is used as an output field of this step for the drift correction module of the S520 to realize a continuous processing link between temperature compensation and signal correction.

[0144] S520: Extract the zero-point drift from the temperature compensation coefficient, perform polynomial fitting compensation, and generate the corrected signal.

[0145] Based on the temperature compensation coefficient output by the S510, the first step involves parsing the zero-point drift information contained therein. This zero-point drift reflects the baseline offset characteristics of the sensor under different temperature conditions. Specifically, the system extracts the zero-point drift trend by comparing historical calibration data with the real-time temperature compensation coefficient, forming a zero-point drift curve. Further, a polynomial fitting algorithm is used to mathematically model the zero-point drift curve. Commonly used fitting orders are second- or third-order polynomials. During the fitting process, the least squares method is used to optimize the fitting parameters, ensuring the model's accuracy in describing the zero-point drift. This fitting model can be dynamically updated, adjusting the fitting curve in conjunction with the real-time temperature compensation coefficient to reflect the influence of the current ambient temperature on the zero point. Further, the polynomial fitting model is applied to the purified digital signal output by the S430 to perform zero-point drift compensation processing. Specifically, the zero-point drift value obtained from the fitting is subtracted point by point from the purified digital signal to obtain the corrected signal. This correction process is implemented in a digital signal processor (DSP), supporting real-time streaming computation to ensure the timeliness of signal correction. To avoid signal distortion due to overcompensation, the system sets upper and lower limits for the compensation amplitude. When these limits are exceeded, the compensation value is automatically restricted, and abnormal compensation events are recorded. During the calibration process, the system monitors the signal stability and noise level, and adjusts the polynomial fitting parameters using a feedback mechanism to improve the accuracy and adaptability of the compensation. Finally, a calibrated signal is generated, including zero-point drift compensation. This signal retains the dynamic characteristics of the original signal and eliminates baseline drift caused by temperature. This calibrated signal serves as the output field of this step, available for use by the S530's data verification module, effectively linking temperature compensation and signal calibration.

[0146] S530. Perform CRC check on the corrected signal and compare it with the threshold to generate verification data;

[0147] Specifically, the steps take the corrected signal output from the S520 as input and first perform a Cyclic Redundancy Check (CRC) verification on the signal. The CRC check uses a standard polynomial that matches the digital signal frame, covering all data bits of the signal to ensure data integrity during transmission and processing. The verification process is implemented by an embedded hardware module, which calculates the CRC code of the input signal in real time and compares it with the checksum attached to the signal. If the CRC check fails, the system triggers an error handling mechanism, including a signal resampling request, error log recording, and anomaly alarm, to prevent erroneous data from entering subsequent processing stages. Further, the system performs threshold comparison processing on the corrected signal. Specifically, multiple threshold ranges are set, including a normal range, a warning range, and an abnormal range, determined according to the sensor's calibration standards and safety specifications. Threshold judgments are performed point-by-point on the input signal to identify whether the signal exceeds the allowable range. For signals within the warning range, the system records a warning event and triggers a soft alarm, but data flow is still allowed; for signals exceeding the abnormal range, the system interrupts data transmission and triggers a hard alarm to prevent erroneous data from affecting system decisions. During the threshold comparison process, the system combines signal temporal continuity analysis to filter out occasional transient anomalies and reduce the false alarm rate. Furthermore, signals that pass CRC verification and meet the threshold requirements are marked as verified data and an integrity verification flag is attached. This flag is used by subsequent modules to quickly identify data validity, improving processing efficiency. The verified data is output in data packet form, containing the corrected signal value, timestamp, and integrity flag, supporting subsequent calls from the S610 weight calculation module, achieving closed-loop processing of temperature compensation, correction, and data verification. The verified data serves as the output field of this step, used for S610 weight calculation, completing the functional closed loop of the temperature compensation and data verification modules.

[0148] Step S600 includes at least steps S610-S630:

[0149] S610. Obtain the verification pass data, perform load distribution model calculation, and obtain the original weight value;

[0150] Specifically, the steps take the verification data output by S530 as input, which includes a high-integrity digital signal after temperature compensation, polynomial zero-point drift correction, and CRC check. First, the verification data is sorted and buffered according to timestamp order to ensure the temporal continuity and completeness of the input data for the load distribution model calculation. Further, based on a pre-established load distribution model that integrates the physical characteristics, structural mechanical parameters, and sensor layout information of the pin-type load cell, the original weight value is calculated. Specifically, the model employs a multi-point load transfer analysis method, mapping the corrected signals output by the dual-bridge redundant sensors to the load distribution state of the load-bearing structure. Combining the sensor sensitivity matrix and geometric position, the force contribution of each bridge circuit is calculated. The load distribution model calculation is pre-calibrated using Finite Element Analysis (FEA) software, including the stiffness coefficients of the sensor bridge circuits, nonlinear deformation response, and temperature influence correction parameters, achieving accurate response prediction under complex load conditions. During the calculation process, the system removes outliers from the input data, employs statistical methods to detect outliers in signal fluctuations, and eliminates abnormal readings caused by transient interference or sensor malfunctions. The load distribution model calculation uses an iterative numerical method combined with the Newton-Raphson algorithm to solve the nonlinear equations, ensuring computational stability and convergence speed. Furthermore, the system dynamically adjusts model parameters based on real-time environmental parameters, such as structural temperature and vibration frequency, improving the adaptability and accuracy of the calculation. Finally, the load distribution model output includes the raw weight value under the combined load, which is the basic weight data before unit conversion and precision adjustment. This raw weight value serves as the output field for this step, available for use by the S620's "Data Formatting" module, enabling continuous conversion from verified data to the raw weight value, supporting subsequent unit conversion and precision optimization.

[0151] S620: Extract significant digits from the original weight value, perform unit conversion and precision rounding, and generate standard weight data;

[0152] Specifically, the steps take the raw weight value output by S610 as input and first perform significant digit extraction on this raw value. This extraction is based on preset accuracy standards and sensor resolution, employing numerical truncation and rounding algorithms to remove invalid or noise-introduced low-order values, ensuring data stability and readability of subsequent displays. The system dynamically adjusts the length of significant digits according to the specific needs of the weighing application, supporting switching between multiple accuracy levels to meet the accuracy requirements of different industrial environments. Further, unit conversion is performed. This conversion is based on the International System of Units (SI) and user-defined units of measurement, supporting conversions between kilograms (kg), pounds (lb), tons (t), and other units. During the conversion process, the system calls the built-in unit conversion table and conversion factor library, and fine-tunes the unit conversion factor using a real-time temperature compensation coefficient to reduce measurement errors caused by environmental changes. The precision rounding operation uses either Banker's Rounding or rounding to zero; the specific rounding method is determined by system configuration parameters, ensuring statistical consistency and numerical fairness in the rounding process. Furthermore, the system performs anomaly detection on the standard weight data, identifying physically unreasonable weight jumps or exceeding limits, triggering a data verification mechanism to prevent erroneous data from entering the output stage. After generation, the standard weight data is appended with a timestamp and status identifier, recording the specific time of data generation and processing status. This data is managed through an internal buffer, supporting multi-threaded concurrent access and real-time updates. This standard weight data serves as the output field for this step, available for use by the S630's "Storage Output" module, enabling the conversion from raw weight to user-friendly data, completing the standardization of weighing information, and ensuring the accuracy and consistency of subsequent storage and display.

[0153] S630. Timestamp the standard weight data and generate a storage record;

[0154] Specifically, the steps take the standard weight data output by the S620 as input and first perform timestamp processing on this data. The timestamp is based on the system's internal high-precision real-time clock (RTC), recording the specific date and time of data generation. The time format conforms to the ISO 8601 standard and supports precise identification at the year, month, day, hour, minute, second, and millisecond levels. The timestamp data is periodically calibrated with an external time server through a hardware clock synchronization module to ensure the accuracy and consistency of the timestamp. Further, the timestamp and standard weight data are integrated into a data structure to form a complete storage record. This storage record includes the weight value, timestamp, data source identifier, processing status, and calibration identifier. The calibration identifier field records the corresponding temperature compensation and zero-point calibration information, derived from the temperature compensation and data verification process of the S500 module, specifically including the calibration cycle, calibration parameter version, and calibration result status. During the integration process, the system uses a structured data format, supporting binary and text format storage to meet different storage media and access requirements. Furthermore, the system performs integrity verification on the storage record, using a CRC checksum to verify the integrity of the record content and prevent data corruption during storage. For storage media, the system supports multi-level storage schemes, including local non-volatile memory (NVM), external flash memory, and cloud databases, automatically selecting the storage path based on the storage strategy. Storage operations are implemented through a transaction mechanism to ensure the atomicity and consistency of data writes, preventing data loss and write conflicts. In abnormal situations, the system records storage failure logs and triggers retry or alternative storage schemes. The storage record, as an output field of this step, is fed back to the "Installation and Configuration Parameters" module of S100, realizing closed-loop feedback of calibration data, supporting dynamic optimization of sensor installation and configuration, and forming a complete closed-loop operation of the weighing system. The storage record is also available for external display devices and data management systems to enable real-time display and historical query of weighing information.

Claims

1. A weighing method for a CAN-based pivot-type load cell, characterized in that, include: The installation parameters of the load-bearing structure are obtained, the positioning processing of the dual-bridge redundant sensors is carried out, and the redundant bridge circuits are symmetrically arranged based on finite element analysis to form the installation configuration parameters. The installation parameters of the load-bearing structure include the dimensions of the axle pins, the thickness of the load-bearing plate, the load distribution range, and the mechanical constraints of the installation environment; the positioning process includes determining the spatial coordinates of the installation positions of the dual-bridge redundant sensors, analyzing the symmetry of the redundant bridge circuits, and verifying the compatibility of the mechanical interfaces. Acquire strain signals from dual-bridge redundant sensors, perform 24-bit high-precision analog-to-digital conversion and frame formatting, add CRC checksum, and generate CAN protocol compatible data packets; Acquire CAN protocol compatible data packets, perform differential signal encoding processing, shielded twisted pair impedance matching and CRC verification, and generate anti-interference transmission signals; Specifically, it includes: The expression for the differentially encoded sequence obtained after acquiring CAN protocol-compatible data packets and performing differential signal encoding processing includes: ； in, This is a coding efficiency metric. For the first CAN data packet One bit; This refers to the length of the data packet. Let be the statistical probability of the k-th bit; This is the bit error rate metric for the k-th bit. This refers to the sampling frequency parameter; Generate differential level sequences: ； in, This represents the differential voltage amplitude. This refers to the packet identifier weight coefficient; This is the initial value of the characteristic impedance of the transmission line; It is the hyperbolic tangent function; The anti-interference transmission signal is acquired, and time-domain noise feature analysis, dynamic filter coefficient calculation based on wavelet packet energy analysis, and sliding window convolution processing are performed to generate a cleaned digital signal; the expression for generating the cleaned digital signal includes: Constructing frequency domain feature vectors: ； in, Wavelet packet energy characteristics; The wavelet decomposition level; Number the frequency sub-bands; This represents the total number of frequency sub-bands. For the first Layer wavelet coefficients of the sub-band; For the first The signal-to-noise ratio of the subband; For the first The center frequency of the sub-band; Will Input to the adaptive filter coefficient calculation model: ； in, These are the filter weight coefficients; , 2 represents the filter index; This represents the total number of filters; It is a natural exponential function; This is the temperature compensation factor; For the first The design bandwidth of each filter; Frequency resolution; Establish a time-varying filtering model: ； in, For time indexing; This is the time offset; Let be the amplitude of the input signal at time t; The attenuation coefficient of the window function; The width of the sliding window; The purified digital signal at time t; Acquire purified digital signals, perform temperature sensor data fusion based on Kalman filtering, multinomial fitting compensation and threshold comparison, and generate verification data; Once the verification data is obtained, the load distribution model calibrated by finite element analysis is used for calculation, unit conversion, and precision rounding to generate and store records.

2. The method according to claim 1, characterized in that, The installation parameters of the load-bearing structure are obtained, and the positioning process of the dual-bridge redundant sensors is performed. Based on the symmetrical arrangement of the redundant bridge circuits according to finite element analysis, the installation configuration parameters are formed, specifically including: Strain signals of the load-bearing structure are acquired using dual-bridge redundant sensors; The dual-bridge redundant sensor includes two bridge circuits, which are respectively arranged in the key stress areas of the load-bearing structure to achieve redundant signal acquisition; The dual-bridge redundant sensor is tightly integrated with a pin-type weighing structure through a high-precision strain gauge.

3. The method according to claim 1, characterized in that, The system acquires strain signals from dual-bridge redundant sensors, performs 24-bit high-precision analog-to-digital conversion and frame formatting, adds CRC checksums, and generates CAN protocol-compatible data packets, specifically including: Simultaneous acquisition of strain signals from both bridges; The synchronous acquisition process is based on a hardware trigger signal, and the sampling timestamps of the two bridges are consistent; It features a high-bandwidth analog front-end circuit, including a constant current source excitation and a differential input amplifier, to suppress common-mode interference.

4. The method according to claim 1 or 3, characterized in that, Acquiring strain signals from dual-bridge redundant sensors also includes: Real-time monitoring of the voltage fluctuation range of the sensor output and automatic adjustment of the sampling window; The built-in diagnostic module detects abnormal signals such as disconnection, short circuit, or over-limit voltage and records the abnormal event log.

5. The method according to claim 1, characterized in that, Perform 24-bit high-precision analog-to-digital conversion, specifically including: It employs a 24-bit high-resolution ADC module, which features low noise and high linearity. The input range of the ADC is dynamically adjusted by sampling the maximum and minimum values ​​in real time.

6. The method according to claim 1, characterized in that, Performs 24-bit high-precision analog-to-digital conversion and frame formatting, CRC checksum addition, and also includes: The data is encapsulated into a digital signal frame according to a preset data structure, which includes sampling timestamp, bridge number, sample value and status identification information.

7. The method according to claim 1, characterized in that, Differential signal encoding processing is performed, specifically including: CAN differential coding technology is used to convert single-ended digital signals into differential signals; By logical flipping and encoding mapping, the corresponding differential level sequence is generated.

Images