System and method for testing vibration rectification error of accelerometer

Abstract

The invention relates to the technical field of data processing, in particular to an accelerometer vibration rectification error test system and method, and the system comprises a self-adaptive signal collection module, a bidirectional cooperative transmission module, a composite modal statistics module and a cross validation error modeling module. The adaptive signal acquisition module generates a feature mark, and the feature mark drives the bidirectional cooperative transmission module to adjust the depth of a buffer area and injects a timestamp verification identifier; a composite modal statistics module selects a wavelet packet time-frequency fusion algorithm based on a timestamp missing proportion, and aligns a physical cycle segmentation window to extract a time-frequency entropy mean statistical magnitude; a cross validation error modeling module fuses statistics and vibration parameters to generate a quantitative model, and a confidence evaluation factor distribution diagram triggers double-path feedback. Invisible window optimization improves data quality in a key period, and frequency domain weight redistribution optimizes a feature extraction strategy. The problems of insufficient dynamic data real-time processing efficiency and transient feature extraction misalignment caused by data acquisition and analysis function separation in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a vibration rectification error testing system and method for an accelerometer. Background Art

[0002] Accelerometer vibration rectification error is a distortion phenomenon in accelerometer measurements, essentially caused by the nonlinear physical properties of the sensing element. When the device is exposed to a symmetrical alternating acceleration environment, the average acceleration output should ideally be zero. However, the nonlinear characteristics of the sensing element cause the response amplitudes to positive and negative acceleration to be unequal, directly resulting in an asymmetric bias in the output waveform. Subsequently, the time averaging calculation applied to this asymmetric waveform inevitably produces a non-zero DC offset component. This DC offset component is equivalent to an additional constant bias error, which is superimposed on the actual measured acceleration value, ultimately reducing the acceleration measurement accuracy.

[0003] Existing accelerometer vibration rectification error testing has the following technical pain points: the existing test system uses discrete hardware and independent data processing units, and there is a lack of deep integration between the data acquisition module and the back-end analysis software. Testers need to manually switch between different devices to complete signal acquisition, filtering, and data recording, and then rely on third-party tools for statistical analysis, resulting in a fragmented operational process. Especially when observing dynamic data changes in a vibration environment, because key statistical quantities cannot be extracted in real time, the original data must be repeatedly exported and the extreme values ​​and means within a specific time period must be manually calculated. For example, when analyzing the accelerometer response drift during vibration, testers need to intercept vibration waveform data from different time periods and perform extreme value search and mean calculation section by section, which significantly extends the analysis cycle. In addition, because hardware sampling control and software processing are not coordinated and optimized, key feature points of high-frequency vibration signals are omitted, causing the rectification error calculation to deviate from the actual working conditions. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides an accelerometer vibration rectification error testing system and method. The present invention solves the technical problems of low efficiency of real-time segmented statistics of dynamic data changes and insufficient accuracy of key transient response feature extraction during the accelerometer vibration rectification error test due to the functional separation of the test system data acquisition hardware and the back-end analysis software and the lack of deep collaborative optimization.

[0005] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows: In a first aspect, the present invention provides an accelerometer vibration rectification error testing system, comprising: an adaptive signal acquisition module, which acquires the current signal output by the accelerometer and performs analog-to-digital conversion to output the original digital signal, extracts the frequency domain energy distribution parameters of the original digital signal, generates a signature based on the frequency domain energy distribution parameters, appends the signature to the digital signal sequence, and outputs the digital signal sequence with the signature; The bidirectional collaborative transmission module receives the digital signal sequence output by the adaptive signal acquisition module, analyzes the characteristic mark in the digital signal sequence, dynamically calculates the buffer depth adjustment amount according to the frequency domain energy distribution parameter value carried in the characteristic mark, sets the ring buffer depth to the buffer depth adjustment amount, synchronously detects the state transition edge of the characteristic mark, appends a timestamp check mark to the corresponding data frame header when the transition edge occurs, and outputs the data stream integrated with the timestamp check mark to the composite modal statistics module; The composite modal statistics module receives the data stream output by the bidirectional collaborative transmission module, parses the timestamp check mark in the data stream, detects the missing ratio of the timestamp check mark, and when the missing ratio exceeds a preset threshold, selects the wavelet packet time-frequency fusion algorithm from the preset algorithm library, aligns the signal physical period segmentation statistical window, and generates statistical data based on the signal analysis results. The statistical data includes at least the mean time-frequency entropy and is marked with the corresponding algorithm selection mark; The cross-validation error modeling module receives the statistical data output by the composite modal statistics module, extracts the algorithm selection identifier in the statistical data, and adopts the range dynamic weighted calculation when the algorithm selection identifier indicates the use of the wavelet packet time-frequency fusion algorithm; when the algorithm selection identifier indicates the use of the preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, it adopts the fuzzy cognitive mapping calculation, and fuses the statistical data to output the vibration rectification error quantification model and the confidence assessment factor distribution map.

[0006] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the adaptive signal acquisition module includes a multi-core analog-to-digital conversion array; The multi-core analog-to-digital conversion array inputs the current signal output by the accelerometer, performs analog-to-digital conversion and outputs the original digital signal; Extract the first 10ms time window of the original digital signal and calculate the energy proportion of the signal in the first 10ms time window in the 10-500Hz frequency band; When the energy ratio exceeds 20%, the sampling rate increase instruction is output to the multi-core analog-to-digital conversion array, a feature tag is generated based on the energy ratio, and the feature tag is transmitted to the bidirectional collaborative transmission module.

[0007] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the bidirectional collaborative transmission module is configured as follows: receiving a digital signal sequence with characteristic marks output by the adaptive signal acquisition module, and parsing the characteristic marks in the digital signal sequence; Calculate the buffer depth adjustment amount based on the energy ratio value in the feature mark and the preset scaling factor, where the buffer depth adjustment amount is the product of the initial depth value superimposed on the energy ratio value and the preset scaling factor; Set the ring buffer depth to the buffer depth adjustment amount; Detect the state transition edge of the characteristic mark. When the state transition edge occurs, add a timestamp check mark to the data frame header, and output the data stream integrated with the timestamp check mark to the composite modal statistics module.

[0008] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the composite modal statistics module is configured as follows: Receive the data stream output by the bidirectional cooperative transmission module, and parse the timestamp verification identification sequence from the data stream; Count the number of timestamp verification sequence losses per second; When the number of losses exceeds 5%, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library; Align the received data stream with the signal physical period segmentation statistical window; Perform wavelet packet decomposition to the fifth layer and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all sub-bands, and statistical data including the mean time-frequency entropy is generated; the wavelet packet time-frequency fusion algorithm selection identifier is marked, and the statistical data marked with the wavelet packet time-frequency fusion algorithm selection identifier is output to the cross-validation error modeling module.

[0009] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the composite modal statistics module is further configured to: Create a statistical data object including a time-frequency entropy mean field based on the time-frequency entropy mean, the signal summary hash stored in the composite modal statistics module, and the statistical window boundary coordinates; An algorithm selection identifier is marked on the statistical data object, and the statistical data object marked with the algorithm selection identifier is output to the cross-validation error modeling module.

[0010] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the cross-validation error modeling module is configured as follows: Receive the statistical data object identified by the annotation algorithm selected by the composite modal statistics module as a statistical data set; Extract algorithmic selection identifiers from statistical data sets; When the algorithm selection flag indicates the use of the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the range dynamic weighting algorithm to perform calculations; When the algorithm selection flag indicates to use a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the fuzzy cognitive mapping algorithm to perform calculations; Receive external input vibration frequency and amplitude parameters; The calculation results of the range dynamic weighting algorithm or the fuzzy cognitive mapping algorithm are integrated with the vibration frequency and amplitude parameters to generate a vibration rectification error quantization model; The confidence assessment factor is calculated based on the vibration rectification error quantization model, and a confidence assessment factor distribution graph is output.

[0011] Furthermore, in the accelerometer vibration rectification error testing system of the present invention, the composite modal statistics module is further configured as follows: Receive the user's time interval selection instruction, parse the user's time interval selection instruction, and obtain the coordinates of the time interval start and end points; intercepting a data segment between a starting point and an ending point from the data stream output by the bidirectional cooperative transmission module; For the data at the starting point of the time interval of the intercepted data segment, the frequency domain energy distribution parameter extraction method of the adaptive signal acquisition module is used to calculate the frequency domain main frequency component; The physical extension length is calculated based on the main frequency component in the frequency domain, where the physical extension length is equal to an integer N multiplied by 1 divided by the main frequency component, where N is an integer; Set the optimization statistics window boundary. The starting point of the optimization statistics window boundary is the starting point of the time interval, and the end point is the starting point of the time interval plus the physical extension length. Within the optimization statistics window boundary, reparse the timestamp check mark sequence and detect the missing timestamp check mark ratio; When the missing ratio exceeds the preset threshold, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library to align and optimize the physical period segmentation of the data stream within the statistical window, perform wavelet packet decomposition to the fifth layer, and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all subbands, and statistical data including the mean time-frequency entropy is generated, and the algorithm selection identifier is marked.

[0012] Furthermore, the accelerometer vibration rectification error testing system of the present invention further includes: After the cross-validation error modeling module generates a confidence assessment factor distribution graph, it calculates the average confidence assessment factor of the distribution graph and outputs the average confidence assessment factor feedback signal to the composite modal statistics module and the adaptive signal acquisition module; When the feedback signal value received by the composite modal statistics module is lower than 0.9, the invisible window optimization process is executed: Receive user selection instructions and parse time interval coordinates, intercept data segments and calculate the main frequency components, set the optimization window and reprocess the data, and output the regenerated statistical data; When the feedback signal value received by the adaptive signal acquisition module is lower than 0.9, the confidence assessment factor distribution diagram is analyzed to identify the frequency interval with a confidence level less than 0.8, and the weight distribution ratio of the frequency interval in the frequency domain energy distribution parameter is reduced.

[0013] Furthermore, the accelerometer vibration rectification error testing system of the present invention further includes: The input of the bidirectional cooperative transmission module is the digital signal sequence with feature marks output by the adaptive signal acquisition module; The input of the composite modal statistics module is the data stream with timestamp verification mark output by the bidirectional collaborative transmission module; The input of the cross-validation error modeling module is the statistical data selected by the labeling algorithm output by the composite modal statistics module.

[0014] Second, see Figure 1 The present invention provides an accelerometer vibration rectification error testing method, which is applied to the accelerometer vibration rectification error testing system, comprising: Step 1: Acquire the current signal output by the accelerometer and perform analog-to-digital conversion to output the original digital signal, extract the frequency domain energy distribution parameters of the original digital signal, generate a feature marker based on the frequency domain energy distribution parameters, attach the feature marker to the digital signal sequence, and output the digital signal sequence with the feature marker; Step 2: Receive the digital signal sequence output from step 1, parse the characteristic mark in the digital signal sequence, dynamically calculate the buffer depth adjustment amount based on the frequency domain energy distribution parameter value carried in the characteristic mark, set the ring buffer depth to the buffer depth adjustment amount, synchronously detect the state transition edge of the characteristic mark, and append a timestamp check mark to the corresponding data frame header when the transition edge occurs, and output the data stream integrated with the timestamp check mark to step 3; Step 3: Receive the data stream output by step 2, parse the timestamp check mark in the data stream, detect the missing ratio of the timestamp check mark, and when the missing ratio exceeds a preset threshold, select a wavelet packet time-frequency fusion algorithm from a preset algorithm library, align the signal physical period segmentation statistical window, generate statistical data based on the signal analysis results, the statistical data at least includes the time-frequency entropy mean, and mark the corresponding algorithm selection mark; Step 4: Receive the statistical data outputted in step 3, extract the algorithm selection identifier in the statistical data, and when the algorithm selection identifier indicates the use of the wavelet packet time-frequency fusion algorithm, use the range dynamic weighted calculation; when the algorithm selection identifier indicates the use of a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, use the fuzzy cognitive mapping calculation, and fuse the statistical data to output the vibration rectification error quantization model and the confidence assessment factor distribution diagram.

[0015] Beneficial effects of the present invention: The present invention drives the bidirectional collaborative transmission module to dynamically adjust the buffer depth and inject a timestamp verification mark through feature marking, thereby solving the problem of data acquisition and analysis being separated, breaking through the limitation of fixed sampling mode, and avoiding the omission of high-frequency transient features; the composite modal statistics module selects the wavelet packet time-frequency fusion algorithm based on the timestamp missing ratio, aligns the physical period segmentation window to extract the time-frequency entropy mean statistic, and improves the dynamic data segmentation statistical efficiency and transient feature extraction accuracy; the cross-validation error modeling module fuses statistics and vibration parameters to generate a vibration rectification error quantization model, and the confidence assessment factor distribution diagram triggers dual-path feedback, including invisible window optimization to improve data quality in key time periods and frequency domain weight redistribution to optimize feature extraction strategies, forming a closed-loop evolution mechanism of hardware acquisition and software analysis, maintaining feature transmission integrity through standardized data stream encapsulation, eliminating information attenuation of multi-tool transfer, and ultimately achieving improved real-time processing efficiency of dynamic data and enhanced accuracy of key transient response feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0017] Figure 1 The present invention provides a flow chart of a method for testing vibration rectification error of an accelerometer. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the embodiment described is only a module embodiment of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The technical solutions provided by each embodiment of the present invention are described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described in detail below.

[0019] In a first aspect, the present invention provides an accelerometer vibration rectification error testing system, comprising: an adaptive signal acquisition module, which acquires the current signal output by the accelerometer and performs analog-to-digital conversion to output the original digital signal, extracts the frequency domain energy distribution parameters of the original digital signal, generates a signature based on the frequency domain energy distribution parameters, appends the signature to the digital signal sequence, and outputs the digital signal sequence with the signature; The bidirectional collaborative transmission module receives the digital signal sequence output by the adaptive signal acquisition module, analyzes the characteristic mark in the digital signal sequence, dynamically calculates the buffer depth adjustment amount according to the frequency domain energy distribution parameter value carried in the characteristic mark, sets the ring buffer depth to the buffer depth adjustment amount, synchronously detects the state transition edge of the characteristic mark, appends a timestamp check mark to the corresponding data frame header when the transition edge occurs, and outputs the data stream integrated with the timestamp check mark to the composite modal statistics module; The composite modal statistics module receives the data stream output by the bidirectional collaborative transmission module, parses the timestamp check mark in the data stream, detects the missing ratio of the timestamp check mark, and when the missing ratio exceeds a preset threshold, selects the wavelet packet time-frequency fusion algorithm from the preset algorithm library, aligns the signal physical period segmentation statistical window, and generates statistical data based on the signal analysis results. The statistical data includes at least the mean time-frequency entropy and is marked with the corresponding algorithm selection mark; The cross-validation error modeling module receives the statistical data output by the composite modal statistics module, extracts the algorithm selection identifier in the statistical data, and adopts the range dynamic weighted calculation when the algorithm selection identifier indicates the use of the wavelet packet time-frequency fusion algorithm; when the algorithm selection identifier indicates the use of the preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, it adopts the fuzzy cognitive mapping calculation, and fuses the statistical data to output the vibration rectification error quantification model and the confidence assessment factor distribution map.

[0020] Technical Description of the Adaptive Signal Acquisition Module: The adaptive signal acquisition module first acquires the current signal output by the accelerometer and performs an analog-to-digital conversion operation to convert the analog current signal into a raw digital signal. This conversion process uses a multi-core parallel processing architecture to improve signal acquisition efficiency. After the conversion is complete, the module performs frequency domain energy analysis on the raw digital signal and calculates the energy distribution parameters of the signal in a specific frequency band through a fast Fourier transform. The frequency domain energy distribution parameters reflect the core spectral characteristics of the signal. Based on the frequency domain energy distribution parameters, a digital feature marker is generated. The feature marker encoding carries a summary of the signal's spectral characteristics. The feature marker is appended to the end of the digital signal sequence, forming a digital signal sequence output with the feature marker. This step achieves the initial integration of the hardware acquisition layer and feature analysis.

[0021] Technical Description of the Bidirectional Collaborative Transmission Module: The bidirectional collaborative transmission module receives a digital signal sequence with a signature and parses the signature embedded in the digital signal sequence. The frequency domain energy distribution parameter value in the signature is used to calculate the buffer depth adjustment. The calculation process dynamically determines the ring buffer storage capacity based on the product of the energy percentage value and the preset scaling factor, added to the initial depth value. After setting the ring buffer depth, the module detects the signature state transition event. The state transition triggers the addition of a timestamp checksum, which is embedded in the corresponding data frame header. The timestamp checksum provides a microsecond timing reference. The data stream with the timestamp checksum is finally output to the downstream module. This process establishes a feature-driven transmission mechanism to ensure the integrity of high-frequency transient signals.

[0022] Technical Description of the Composite Modal Statistics Module: The Composite Modal Statistics Module receives a data stream with a timestamp checksum and parses the timestamp checksum sequence within the data stream. It counts the number of timestamp checksum losses per second, which reflects the reliability of data transmission. When the number of timestamp losses exceeds a set threshold, it calls a wavelet packet time-frequency fusion algorithm from a preset algorithm library. The algorithm is selected based on the quality of data transmission. The module aligns the signal's physical period to divide the statistical window, with the window boundaries matching the signal's natural period. It performs wavelet packet decomposition to a fixed number of layers, generating multi-level subband frequency domain components. It calculates the energy entropy of each subband and aggregates the output time-frequency entropy mean. The time-frequency entropy mean is included in the statistical data as a core statistic. The statistical data is labeled with the algorithm selected and then output. This step addresses the efficiency of segmented statistics for dynamic signals.

[0023] Technical Description of the Cross-Validation Error Modeling Module: The Cross-Validation Error Modeling Module receives statistical data labeled with an algorithm selection identifier. The algorithm selection identifier is extracted from the statistical data, and the algorithm selection identifier declares the upstream processing logic. When the algorithm selection identifier indicates the wavelet packet time-frequency fusion algorithm, a dynamic range weighting calculation is performed. When the algorithm selection identifier indicates another statistical algorithm, a fuzzy cognitive mapping calculation is performed. The module receives external vibration frequency and amplitude parameters. The algorithm calculation results are combined with the vibration parameters to generate a vibration rectification error quantization model. The quantization model represents the mapping relationship between the error and the vibration parameters. Based on the quantization model, a confidence assessment factor is calculated and a two-dimensional confidence assessment factor distribution map is generated. The distribution map indicates the spatial distribution of the model reliability.

[0024] Invisible window optimization technology description: The composite modal statistics module receives the user's box-selected time interval instruction and parses the instruction to obtain the coordinates of the time interval's starting and ending points. The corresponding time interval data segment is intercepted from the data stream. The main frequency component of the data frequency domain at the starting point is extracted. The component calculation method is consistent with the adaptive signal acquisition module. The physical extension length is calculated based on the main frequency component, and the extension length adapts to the signal period characteristics. The optimized statistical window boundary is set, and the boundary range covers an integer multiple of the signal period. Within the optimized window, the timestamp sequence is re-parsed, the missing ratio is detected, the statistical algorithm is selected, and wavelet packet decomposition is performed. The optimized time-frequency entropy mean is calculated to generate statistical data and annotate the algorithm identifier. This mechanism improves the accuracy of critical period analysis.

[0025] The dual-path feedback control technology describes the cross-validation error modeling module, which calculates the spatial mean of the confidence factor distribution map and outputs an average confidence factor feedback signal. This feedback signal is transmitted to the composite modal statistics module and the adaptive signal acquisition module. When the feedback signal value received by the composite modal statistics module falls below a threshold, the invisible window optimization process is triggered to regenerate statistical data. When the feedback signal value received by the adaptive signal acquisition module falls below a threshold, the confidence factor distribution map is analyzed to identify low-confidence frequency intervals. The weight allocated to these low-confidence frequency intervals in the frequency domain energy distribution parameters is reduced. This closed-loop system achieves the co-evolution of hardware sampling and software analysis.

[0026] System-Level Data Flow Technology Description: The adaptive signal acquisition module outputs a digital signal sequence with feature tags to the bidirectional collaborative transmission module. The bidirectional collaborative transmission module outputs a data stream with timestamp verification tags to the composite modal statistics module. The composite modal statistics module outputs statistical data, labeled with algorithm selection tags, to the cross-validation error modeling module. The data stream uses a three-level standardized encapsulation format: feature tags carry signal spectral characteristics, timestamp tags ensure timing synchronization, and algorithm tags declare processing methods. This structured data flow maintains feature integrity and eliminates information degradation when transferring data across modules.

[0027] The present invention performs signal acquisition and signature generation in step 1; implements feature-driven transmission control in step 2; completes dynamic signal statistics and algorithm selection in step 3; and performs error modeling and confidence assessment in step 4. The signature-marked digital signal sequence output from step 1 serves as input for step 2, the timestamp-bearing data stream output from step 2 serves as input for step 3, and the labeled statistical data output from step 3 serves as input for step 4. This four-stage process forms a "acquisition, transmission, statistics, modeling" technology chain, with each step maintaining a strictly closed-loop input and output loop.

[0028] Specifically, the accelerometer vibration rectification error test system of the present invention, the adaptive signal acquisition module includes a multi-core analog-to-digital conversion array; The multi-core analog-to-digital conversion array inputs the current signal output by the accelerometer, performs analog-to-digital conversion and outputs the original digital signal; Extract the first 10ms time window of the original digital signal and calculate the energy proportion of the signal in the first 10ms time window in the 10-500Hz frequency band; When the energy ratio exceeds 20%, the sampling rate increase instruction is output to the multi-core analog-to-digital conversion array, a feature tag is generated based on the energy ratio, and the feature tag is transmitted to the bidirectional collaborative transmission module.

[0029] The adaptive signal acquisition module processes the accelerometer's output current signal through a multi-core analog-to-digital converter array. This multi-core analog-to-digital converter array utilizes a parallel conversion channel architecture, with each core independently performing analog-to-digital conversion. Oversampling technology is used during the conversion process to suppress quantization noise and output a high-fidelity original digital signal. The multi-core array's collaborative operation improves signal acquisition throughput efficiency.

[0030] The module intercepts a fixed-duration segment at the beginning of the original digital signal as an analysis window. It then performs a frequency domain transform on this segment, using a fast Fourier transform algorithm to extract the signal's energy distribution parameters within the target frequency band. These energy distribution parameters quantitatively reflect the concentration of signal energy within the target frequency band and characterize the core spectral characteristics of the current vibration environment.

[0031] A dynamic sampling strategy is triggered based on the results of energy distribution parameter analysis. When the energy percentage of the target frequency band exceeds a set threshold, a sampling rate increase command is generated and sent to the multi-core analog-to-digital conversion array. Upon receiving the command, the array switches the sampling frequency configuration to enhance the capture of high-frequency signal components. This dynamic sampling mechanism avoids the problem of missing transient features in fixed sampling mode.

[0032] Frequency-domain energy distribution parameters are digitally encoded to generate signatures. The signature data structure includes energy percentage values ​​and frequency band identification information. The signature is appended to the end of the original digital signal sequence, forming a signature-labeled digital signal sequence. The signature is transmitted as a hardware-level feature summary to the bidirectional collaborative transmission module, providing decision-making support for downstream buffer optimization.

[0033] A multi-core analog-to-digital conversion array converts analog current signals into digital signals, providing the input basis for frequency domain analysis. Fixed-duration analysis window capture ensures temporal consistency in feature calculations, eliminating feature fluctuations caused by random capture.

[0034] Fast Fourier transform processing converts the time-domain signal into a frequency-domain energy distribution. Calculating the energy percentage of the target frequency band focuses on vibration-sensitive frequency bands while filtering out interference from non-critical noise bands. This step establishes a mapping between the signal's physical characteristics and digital features.

[0035] The energy percentage threshold comparison directly drives sampling rate switching. When high-frequency energy is concentrated, the sampling rate is increased to ensure transient feature integrity, while when low-frequency energy dominates, the standard sampling rate is maintained to conserve resources. This control mechanism enables adaptive optimization of acquisition strategies.

[0036] Energy distribution parameters are encoded to generate signatures, converting physical features into machine-processable digital identifiers. These signatures are then transmitted bound to the original signal, preserving the temporal and spatial correlation of the signature data. This step completes the transfer of features from the hardware acquisition layer to the data transmission layer.

[0037] Specifically, the accelerometer vibration rectification error test system of the present invention has a bidirectional collaborative transmission module configured as follows: receiving a digital signal sequence with characteristic marks output by the adaptive signal acquisition module, and parsing the characteristic marks in the digital signal sequence; Calculate the buffer depth adjustment amount based on the energy ratio value in the feature mark and the preset scaling factor, where the buffer depth adjustment amount is the product of the initial depth value superimposed on the energy ratio value and the preset scaling factor; Set the ring buffer depth to the buffer depth adjustment amount; Detect the state transition edge of the characteristic mark. When the state transition edge occurs, add a timestamp check mark to the data frame header, and output the data stream integrated with the timestamp check mark to the composite modal statistics module.

[0038] The bidirectional collaborative transmission module receives the signature-tagged digital signal sequence output by the adaptive signal acquisition module. The module parses the signature embedded in the digital signal sequence, which includes the frequency-domain energy distribution parameter. The parsing process extracts the energy fraction from the signature, which reflects the signal's energy concentration in the target frequency band.

[0039] Based on the extracted energy contribution value and a preset scaling factor, the module calculates the buffer depth adjustment. This calculation uses a product-addition mechanism, where the buffer depth adjustment is the product of the initial depth value, the energy contribution value, and the preset scaling factor. The calculation logic correlates signal feature strength with storage resource requirements, dynamically adapting to high-frequency transient signal processing needs.

[0040] The module applies the calculated buffer depth adjustment to the ring buffer depth configuration. The ring buffer uses an end-to-end storage structure, and the depth adjustment determines the number of data frames the buffer can accommodate. The configuration process updates the buffer storage space in real time, aligning the buffer depth with the intensity of signal characteristic changes. This buffer depth adjustment eliminates the risk of resource redundancy or overflow associated with fixed buffer configurations.

[0041] The module continuously monitors changes in the signature state and identifies sudden changes in the signature value using a transition edge detection circuit. When a signature state change occurs, the module embeds a timestamp checksum in the corresponding data frame header. The timestamp checksum includes precise clock information and marks the absolute time of the signal signature change. The timestamp is added to the data frame header to ensure traceability.

[0042] After adding the timestamp checksum, the module integrates the original digital signal sequence, signature, and timestamp checksum into a structured data stream. This structured data stream is output via a high-speed transmission channel to the composite modal statistics module. This output process maintains the signal timing integrity and signature correlation, establishing a timing alignment foundation for downstream statistical analysis.

[0043] The feature tag analysis provides the energy percentage numerical input, which is the core parameter of subsequent calculations. The analysis action establishes a direct connection between the upstream feature and the transmission control.

[0044] The energy percentage multiplied by the preset scaling factor determines the magnitude of the adjustment. This multiplication reflects the degree to which signal strength influences resource demand. The initial depth value is superimposed to achieve a gradual adjustment of resource allocation.

[0045] The ring buffer depth setting directly applies the previous calculation result. The end-to-end nature of the ring structure ensures data continuity. This depth adjustment mechanism overcomes the limitations of fixed buffer configurations.

[0046] State transition edge detection identifies the moment of a feature mutation. Timestamp identification is bound to an absolute time reference, solving multi-device timing synchronization issues. Adding location selection data frame headers enables identification for rapid positioning.

[0047] The original signal, signature, and timestamp are integrated. Structured encapsulation maintains the integrity of signature transmission. High-speed transmission channels minimize transmission latency.

[0048] Specifically, the accelerometer vibration rectification error test system of the present invention has a composite modal statistics module configured as follows: Receive the data stream output by the bidirectional cooperative transmission module, and parse the timestamp verification identification sequence from the data stream; Count the number of timestamp verification sequence losses per second; When the number of losses exceeds 5%, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library; Align the received data stream with the signal physical period segmentation statistical window; Perform wavelet packet decomposition to the fifth layer and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all sub-bands, and statistical data including the mean time-frequency entropy is generated; the wavelet packet time-frequency fusion algorithm selection identifier is marked, and the statistical data marked with the wavelet packet time-frequency fusion algorithm selection identifier is output to the cross-validation error modeling module.

[0049] The composite modal statistics module receives the data stream output by the bidirectional collaborative transmission module, which includes vibration signal data with timestamp checksums. The module first parses the timestamp checksum sequence from the data stream. The timestamp checksum contains the precise time information of signal acquisition. The parsing process identifies and extracts the timestamp checksum content and establishes a timestamp sequence database for subsequent integrity analysis.

[0050] The module performs integrity checks on the timestamp checksum sequence and counts the number of timestamp checksum losses per second. This number reflects the degree of timing synchronization signal loss during data transmission. This statistics uses a sliding time window counting mechanism to quantify data transmission reliability. This step generates data transmission quality assessment parameters.

[0051] Based on the loss statistics, the module dynamically selects a statistical algorithm. When the loss count exceeds a preset threshold, it calls the wavelet packet time-frequency fusion algorithm from a pre-set library of algorithms. This algorithm selection mechanism links data transmission quality with the analysis method. In scenarios with high loss rates, a highly robust algorithm is used to improve analysis stability. The pre-set library includes a variety of noise suppression algorithms.

[0052] The module performs physical period-segmented window alignment on the received data stream. This alignment process uses zero-crossing detection to identify the inherent period boundaries of the signal and adjusts the statistical window start and end positions to ensure that the window length matches an integer multiple of the signal's physical period. This synchronization eliminates period truncation errors and ensures the accuracy of subsequent feature calculations. After window alignment, data is processed within the complete period.

[0053] Wavelet packet decomposition is performed to a fixed number of decomposition levels, converting the time-domain signal into a multi-level sub-band frequency domain component. Wavelet packet decomposition employs a tree-structured frequency band partitioning, with each sub-band corresponding to a signal component within a specific frequency range. A fixed decomposition depth ensures consistent feature extraction, balancing computational complexity with resolution requirements.

[0054] The energy entropy characteristic of each subband component is calculated. Energy entropy represents the complexity of the subband signal energy distribution. Energy entropy calculation is based on the probability distribution of subband signal energy and quantifies the degree of uncertainty in the frequency domain component. The energy entropy calculation results of all subbands are aggregated to output the mean time-frequency entropy, which represents the global time-frequency characteristics. The mean time-frequency entropy integrates the joint characteristics of the signal in the time and frequency domains.

[0055] Generate a statistical data set including the mean time-frequency entropy, which integrates the mean time-frequency entropy and the original signal summary information. The statistical data is annotated with the wavelet packet time-frequency fusion algorithm selection identifier, which clearly records the algorithm type used. The annotated structured data is output to the downstream modeling module via a high-speed interface.

[0056] Specifically, in the accelerometer vibration rectification error testing system of the present invention, the composite modal statistics module is further configured as follows: Create a statistical data object including a time-frequency entropy mean field based on the time-frequency entropy mean, the signal summary hash stored in the composite modal statistics module, and the statistical window boundary coordinates; An algorithm selection identifier is marked on the statistical data object, and the statistical data object marked with the algorithm selection identifier is output to the cross-validation error modeling module.

[0057] The Composite Modal Statistics module creates a structured statistical data object based on the time-frequency entropy mean data, the signal summary hash value stored internally, and the statistical window boundary coordinates. The time-frequency entropy mean data is derived from the aggregated results of wavelet packet decomposition and characterizes the global time-frequency characteristics of the signal. The signal summary hash value is generated by applying a hash algorithm to the original digital signal and provides a basis for data integrity verification. The statistical window boundary coordinates record the start and end time points of the analysis window and specify the time range for feature calculations. These three elements are integrated to form the core data structure, including the time-frequency entropy mean field.

[0058] After creating a statistical data object, the module annotates the object's metadata with an algorithm selection identifier. This identifier specifies the specific statistical algorithm used in the data processing process and explicitly records the application status of the wavelet packet time-frequency fusion algorithm or other pre-defined algorithms. This annotation permanently binds the algorithm selection identifier to the statistical data object, providing downstream modules with a basis for tracing the processing method.

[0059] After labeling, the module outputs the statistical data object containing the selected identifiers by the labeling algorithm to the cross-validation error modeling module. This output process uses a structured data transmission protocol, with a header that includes a data type identifier and a checksum. The data structure maintains the association between the mean time-frequency entropy, signal summary hash value, window boundary coordinates, and the algorithm selection identifier, thus preventing information attenuation during data transfer.

[0060] Specifically, the accelerometer vibration rectification error testing system of the present invention has a cross-validation error modeling module configured as follows: Receive the statistical data object identified by the annotation algorithm selected by the composite modal statistics module as a statistical data set; Extract algorithmic selection identifiers from statistical data sets; When the algorithm selection flag indicates the use of the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the range dynamic weighting algorithm to perform calculations; When the algorithm selection flag indicates to use a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the fuzzy cognitive mapping algorithm to perform calculations; Receive external input vibration frequency and amplitude parameters; The calculation results of the range dynamic weighting algorithm or the fuzzy cognitive mapping algorithm are integrated with the vibration frequency and amplitude parameters to generate a vibration rectification error quantization model; The confidence assessment factor is calculated based on the vibration rectification error quantization model, and a confidence assessment factor distribution graph is output.

[0061] The Cross-Validation Error Modeling module receives the statistical data object output by the Composite Modal Statistics module, labeled with the algorithm selection identifier, as input for the statistical data set. The module first parses the algorithm selection identifier within the statistical data object. The algorithm selection identifier explicitly indicates the specific algorithm type used in the upstream statistical processing. The identifier parsing process distinguishes the application scenarios of the wavelet packet time-frequency fusion algorithm from other pre-defined statistical algorithms.

[0062] When the algorithm selection flag indicates the use of the wavelet packet time-frequency fusion algorithm, the module inputs the statistical data set into the range dynamic weighting algorithm to perform the calculation. The range dynamic weighting algorithm calculates the data range based on the mean time-frequency entropy characteristics and dynamically assigns weight coefficients to enhance the contribution of high-frequency transient characteristics. The calculation process focuses on analyzing the degree of dispersion of the signal's time-frequency characteristics.

[0063] When the algorithm selection flag indicates the use of a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, the module inputs the statistical data set into the fuzzy cognitive mapping algorithm to perform the calculation. The fuzzy cognitive mapping algorithm constructs a nonlinear relationship network between input features and output results, and handles data uncertainty through fuzzy rule reasoning. This calculation process is adaptable to low-quality data scenarios.

[0064] The module receives vibration frequency and amplitude parameters from external test equipment. The frequency parameter describes the fundamental frequency of the excitation signal, while the amplitude parameter quantifies the intensity of the vibration environment. The parameter input interface supports access to real-time environmental monitoring data, ensuring that the model matches the physical working conditions.

[0065] The fusion process couples the algorithm's calculation results with vibration frequency and amplitude parameters in multiple dimensions. This coupling operation maps heterogeneous data into a unified feature space through matrix transformation, establishing a correlation matrix between signal features and vibration parameters. The fusion output generates a vibration rectification error quantization model, which represents the mapping relationship between the accelerometer output error and vibration parameters in functional form.

[0066] Based on the vibration rectification error quantization model, the module calculates confidence assessment factors. These factors reflect the model's reliability level in different feature regions. The calculation process combines residual analysis and feature coverage density assessment. The module outputs a confidence assessment factor distribution map, which uses a heat map to mark the boundaries between high-confidence and low-confidence regions, providing a visual criterion for the reliability of the test results.

[0067] The labeling algorithm selects the identified statistical data object as its core input. The algorithm then resolves the identified object to determine the subsequent calculation path. The identification content directly determines the statistical algorithm selected. The range dynamic weighting algorithm enhances the contribution of high-frequency features, and the fuzzy cognitive mapping algorithm addresses data uncertainty. The algorithm branches match the upstream data processing method. External vibration parameters provide a physical working condition benchmark, frequency parameters define the excitation characteristics, and amplitude parameters quantify the environmental intensity. Parameter inputs synchronize the model with the actual vibration environment. The algorithm output is coupled with the vibration parameters through matrix transformation to construct a unified feature space. The fusion process eliminates dimensional differences and generates a quantitative mathematical model. Confidence factors are calculated based on model residuals and feature coverage, and spatial distribution heat maps identify areas of model reliability. Visual output supports test conclusion and decision-making.

[0068] Specifically, in the accelerometer vibration rectification error testing system of the present invention, the composite modal statistics module is further configured as follows: Receive the user's time interval selection instruction, parse the user's time interval selection instruction, and obtain the coordinates of the time interval start and end points; intercepting a data segment between a starting point and an ending point from the data stream output by the bidirectional cooperative transmission module; For the data at the starting point of the time interval of the intercepted data segment, the frequency domain energy distribution parameter extraction method of the adaptive signal acquisition module is used to calculate the frequency domain main frequency component; The physical extension length is calculated based on the main frequency component in the frequency domain, where the physical extension length is equal to an integer N multiplied by 1 divided by the main frequency component, where N is an integer; Set the optimization statistics window boundary. The starting point of the optimization statistics window boundary is the starting point of the time interval, and the end point is the starting point of the time interval plus the physical extension length. Within the optimization statistics window boundary, reparse the timestamp check mark sequence and detect the missing timestamp check mark ratio; When the missing ratio exceeds the preset threshold, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library to align and optimize the physical period segmentation of the data stream within the statistical window, perform wavelet packet decomposition to the fifth layer, and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all subbands, and statistical data including the mean time-frequency entropy is generated, and the algorithm selection identifier is marked.

[0069] The Composite Modal Statistics module receives user input via a graphical user interface (GUI) to select a time interval. The command parsing process maps the screen coordinates of the user's selection into start and end timestamps in the system's time coordinate system. This coordinate conversion is based on the calibration parameters of the data stream's time axis, accurately aligning the user's intent with the signal's time dimension. The coordinates of the time interval's start and end points provide a benchmark for locating key time periods.

[0070] This module extracts data segments between the start and end points of the data stream output by the bidirectional collaborative transmission module. The extraction operation locates the data stream location based on timestamp coordinates and extracts complete signal data blocks within the target time interval. The data segments include the original vibration signal and its associated timestamp checksum sequence, preserving the integrity of the original data structure. The extraction process inherits the data stream's timing characteristics and associated markers.

[0071] The dominant frequency component of the signal near the starting point of the time interval of the intercepted data segment is calculated in the frequency domain. This calculation method reuses the core algorithm process of the adaptive signal acquisition module: Fast Fourier Transform is used to analyze the spectrum of the signal near the starting point, and the frequency component with the highest energy concentration is identified as the dominant frequency component. The reuse of feature extraction methods ensures the consistency of frequency domain analysis across the entire system.

[0072] The physical extension length is calculated based on the dominant frequency component in the frequency domain. The physical extension length consists of integer multiples of the signal period length. The calculation formula reflects the signal's physical periodicity: the physical extension length is equal to the integer N multiplied by 1 divided by the dominant frequency component. The calculation process adapts to the signal's fundamental frequency characteristics, ensuring that the extension range matches the signal's physical oscillation patterns.

[0073] Set optimized statistical window boundaries. The starting point of the boundary remains unchanged from the start point of the user-selected time interval, and the end point extends to the starting point plus the physical extension length. Window boundary reconstruction eliminates incomplete cycles caused by manual truncation. The optimized window covers the original area of ​​interest and the physically extended area, enhancing the physical rationality of feature calculations.

[0074] Reparse the timestamp checksum sequence within the optimized statistical window. This parsing operation verifies the continuity and integrity of timestamps within the window and establishes a timestamp sequence database. Timestamp sequence parsing provides the input for subsequent integrity checks.

[0075] Detects the percentage of timestamp checksum missing identifiers, counting the percentage of missing identifiers relative to the total expected number of identifiers. This missing ratio is calculated using a sliding window counting mechanism to quantify the reliability of data transmission within the optimization window. This parameter reflects the data quality status during critical periods.

[0076] When the missing data ratio exceeds a preset threshold, a wavelet packet time-frequency fusion algorithm is selected from a pre-set library of algorithms. This algorithm selection mechanism prioritizes highly robust processing methods to accommodate potential noise interference during critical periods. The threshold comparison result directly determines the type of statistical algorithm.

[0077] Alignment optimizes the physical period segmentation of the data stream within the statistical window. The segmentation operation uses zero-crossing detection to identify the inherent period boundaries of the signal and adjusts the data segment boundaries within the window to match the complete signal period. Period alignment eliminates truncation errors and ensures the accuracy of subsequent feature calculations.

[0078] Wavelet packet decomposition is performed up to the fifth level, converting the time-domain signal within the optimization window into 32 subband frequency-domain components. Fixed-level decomposition balances feature resolution and computational efficiency, with subband division covering the full spectral range of the target frequency band.

[0079] Calculate the energy entropy characteristics of all decomposed subbands. Energy entropy quantifies the complexity of the energy distribution of each subband signal. The entropy value calculation is based on the subband energy probability distribution function and outputs a multidimensional energy entropy feature vector.

[0080] The mean time-frequency entropy is calculated based on the energy entropy eigenvectors of all subbands. The mean operation aggregates the subband features into a global statistic. The mean time-frequency entropy characterizes the joint time-frequency distribution of the signal within the optimization window.

[0081] Generate statistical data including the mean time-frequency entropy. The data structure integrates the mean time-frequency entropy, the signal summary hash value, and the optimized window boundary coordinates. The statistical data is annotated with the selection flag of the wavelet packet time-frequency fusion algorithm and the feature generation method is declared.

[0082] Specifically, the accelerometer vibration rectification error testing system of the present invention further includes: After the cross-validation error modeling module generates a confidence assessment factor distribution graph, it calculates the average confidence assessment factor of the distribution graph and outputs the average confidence assessment factor feedback signal to the composite modal statistics module and the adaptive signal acquisition module; When the feedback signal value received by the composite modal statistics module is lower than 0.9, the invisible window optimization process is executed: Receive user selection instructions and parse time interval coordinates, intercept data segments and calculate the main frequency components, set the optimization window and reprocess the data, and output the regenerated statistical data; When the feedback signal value received by the adaptive signal acquisition module is lower than 0.9, the confidence assessment factor distribution diagram is analyzed to identify the frequency interval with a confidence level less than 0.8, and the weight distribution ratio of the frequency interval in the frequency domain energy distribution parameter is reduced.

[0083] After the cross-validation error modeling module generates a confidence factor distribution map, it calculates the spatial mean of this distribution map to generate an average confidence factor. This average confidence factor is then transmitted as a feedback signal via the system's internal bus to the composite modal statistics module and the adaptive signal acquisition module. The feedback signal is numerically quantified to represent the global reliability of the vibration rectification error quantization model.

[0084] The composite modal statistics module monitors the value of the received feedback signal in real time. When the feedback signal value falls below a set threshold, the invisible window optimization process is triggered. This triggering mechanism automatically activates the optimization process based on the numerical comparison results. The invisible window optimization process consists of four consecutive operations: receiving user selection instructions and parsing time interval coordinates; intercepting the target data segment and calculating the dominant frequency component; setting a physically extended optimization window; and re-executing the complete statistical processing process. The process outputs regenerated statistical data to enhance the quality of features during critical time periods.

[0085] The adaptive signal acquisition module receives the average confidence factor feedback signal in parallel. The module analyzes the feedback signal values ​​and extracts the confidence factor distribution map data structure. The module analyzes the spatial data of the distribution map to identify frequency intervals with confidence levels below a specified standard. For these identified low-confidence frequency intervals, the weight allocated to these intervals in the frequency domain energy distribution parameters is reduced. This weight adjustment process dynamically optimizes the feature extraction strategy, focusing on signal features in high-confidence frequency bands.

[0086] Calculating the spatial mean of the confidence factor distribution graph provides a comprehensive assessment of model reliability. A feedback signal value below a threshold indicates insufficient system reliability, triggering an optimization mechanism. Bus transmission ensures synchronous signal distribution.

[0087] Low-confidence states activate the invisible window optimization process, user-selected commands guide analysis focus, physically expanded windows enhance feature plausibility, and data reprocessing improves input quality. Confidence assessment factor distribution graphs analyze and identify low-confidence frequency intervals, and weight allocation ratios are adjusted to reduce the impact of interfering frequency bands. Dynamic optimization of feature extraction strategies allows subsequent acquisition to focus on key features.

[0088] Optimization of the statistics module addresses data quality issues, while optimization of the acquisition module addresses feature generation accuracy issues. These two pathways collaborate to form a closed-loop enhancement system, jointly improving confidence in subsequent modeling.

[0089] Specifically, the accelerometer vibration rectification error testing system of the present invention further includes: The input of the bidirectional cooperative transmission module is the digital signal sequence with feature marks output by the adaptive signal acquisition module; The input of the composite modal statistics module is the data stream with timestamp verification mark output by the bidirectional collaborative transmission module; The input of the cross-validation error modeling module is the statistical data selected by the labeling algorithm output by the composite modal statistics module.

[0090] The adaptive signal acquisition module outputs a signature-tagged digital signal sequence to the bidirectional collaborative transmission module. This signature-tagged digital signal sequence consists of the original digital signal block and a signature block, which encodes frequency-domain energy distribution parameters. The output sequence uses a differential signal transmission protocol, and the protocol frame header includes a data type identifier and a length check code to ensure interference resistance. This output serves as the input for the bidirectional collaborative transmission module.

[0091] After receiving input, the bidirectional collaborative transmission module performs signature parsing and timestamp injection, outputting a data stream with a timestamp checksum to the composite modal statistics module. This data stream with a timestamp checksum is encapsulated using a timestamp tag bound to the signal frame. The timestamp tag, located in the data frame header, accurately records the system clock count at the time the signature tag transitions. The data stream is transmitted via a direct memory access channel, maintaining millisecond-level transmission latency. This output constitutes the input source for the composite modal statistics module.

[0092] The composite modal statistics module receives the data stream with timestamp checksums, performs timestamp integrity checks and statistical analysis, and outputs statistical data labeled with the algorithm selection identifier to the cross-validation error modeling module. The statistical data labeled with the algorithm selection identifier is structured as a triple: the core feature of the mean time-frequency entropy, the signal summary hash value, and the algorithm selection identifier. This triple is encapsulated using a protocol and transmitted over a high-speed serial bus. The encapsulation header includes a version identifier and a cyclic redundancy check code. This output serves as the input to the cross-validation error modeling module.

[0093] Second, see Figure 1 The accelerometer vibration rectification error testing method of the present invention is applied to the accelerometer vibration rectification error testing system, comprising: Step 1: Acquire the current signal output by the accelerometer and perform analog-to-digital conversion to output the original digital signal, extract the frequency domain energy distribution parameters of the original digital signal, generate a feature marker based on the frequency domain energy distribution parameters, attach the feature marker to the digital signal sequence, and output the digital signal sequence with the feature marker; Step 2: Receive the digital signal sequence output from step 1, parse the characteristic mark in the digital signal sequence, dynamically calculate the buffer depth adjustment amount based on the frequency domain energy distribution parameter value carried in the characteristic mark, set the ring buffer depth to the buffer depth adjustment amount, synchronously detect the state transition edge of the characteristic mark, and append a timestamp check mark to the corresponding data frame header when the transition edge occurs, and output the data stream integrated with the timestamp check mark to step 3; Step 3: Receive the data stream output by step 2, parse the timestamp check mark in the data stream, detect the missing ratio of the timestamp check mark, and when the missing ratio exceeds a preset threshold, select a wavelet packet time-frequency fusion algorithm from a preset algorithm library, align the signal physical period segmentation statistical window, generate statistical data based on the signal analysis results, the statistical data at least includes the time-frequency entropy mean, and mark the corresponding algorithm selection mark; Step 4: Receive the statistical data outputted in step 3, extract the algorithm selection identifier in the statistical data, and when the algorithm selection identifier indicates the use of the wavelet packet time-frequency fusion algorithm, use the range dynamic weighted calculation; when the algorithm selection identifier indicates the use of a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, use the fuzzy cognitive mapping calculation, and fuse the statistical data to output the vibration rectification error quantization model and the confidence assessment factor distribution diagram.

[0094] Step 1: Obtain the current signal output by the accelerometer through a multi-core analog-to-digital conversion array, perform parallel analog-to-digital conversion and output the original digital signal. A fixed-length analysis window is intercepted for the original digital signal, and the energy distribution parameters of the signal in the window in the target frequency band are calculated using fast Fourier transform. The energy distribution parameters characterize the core spectral characteristics of the signal. A digital feature marker is generated based on the energy distribution parameters, and the feature marker encoding contains the energy percentage value and frequency band identification information. The feature marker is appended to the end of the digital signal sequence to form a digital signal sequence output with a feature marker. This step realizes the conversion of physical signals to digital features, providing a feature-driven basis for downstream collaborative transmission.

[0095] Step 2: Receive a digital signal sequence with a signature and parse the signature content in the sequence. Calculate the buffer depth adjustment based on the energy percentage value in the signature and the preset scaling factor. The depth adjustment is obtained by superimposing the initial depth value on the product of the energy percentage value and the scaling factor. Set the ring buffer depth to the calculated adjustment to dynamically match the signal feature strength. Synchronously detect the signature state transition edge, and when the transition edge occurs, embed a timestamp check mark in the corresponding data frame header. The timestamp mark records the microsecond system clock count value of the signature mutation. Output the data stream with the integrated timestamp mark to the downstream statistics module to establish a transmission timing benchmark.

[0096] Step 3: Receive a data stream with a timestamp checksum and parse the timestamp sequence in the data stream. Count the number of timestamp loss per second, which reflects the integrity of data transmission. When the number of timestamp loss exceeds a set threshold, invoke the wavelet packet time-frequency fusion algorithm from the preset algorithm library. Perform a physical period segmentation window alignment operation on the data stream, matching the inherent period boundaries of the signal through zero-crossing detection. Perform wavelet packet decomposition to the fifth level to generate multi-level subband frequency domain components. Calculate the energy entropy characteristics of each subband and aggregate them to output a global statistic of the mean time-frequency entropy. Generate statistical data containing the mean time-frequency entropy and annotate the algorithm selection flag.

[0097] Step 4: Receive statistical data identified by the algorithm selection. When the algorithm identifier indicates the wavelet packet time-frequency fusion algorithm, the data is processed using dynamic range weighting calculation; when other preset algorithms are indicated, fuzzy cognitive mapping calculation is used. Receive external input vibration frequency and amplitude parameters. The vibration frequency defines the excitation fundamental frequency characteristics, and the amplitude quantifies the environmental intensity. The fusion algorithm calculation results and vibration parameters are combined to generate a vibration rectification error quantization model. The model establishes a functional mapping relationship between the error and the vibration parameters. Calculate the confidence assessment factor based on model residual analysis, output the confidence assessment factor distribution map, and use the distribution map to annotate the model reliability spatial distribution in the form of a heat map.

[0098] The adaptive signal acquisition module of the present invention collects the accelerometer current signal through a multi-core analog-to-digital conversion array, and extracts the frequency domain energy distribution parameters of the front time window signal in the target frequency band. A feature tag is generated based on the frequency domain energy distribution parameters, and the feature tag carries a summary of the signal spectrum characteristics. The feature tag is attached to the digital signal sequence output to realize the real-time transmission of the hardware acquisition layer features to the transmission layer. The feature tag drives the bidirectional collaborative transmission module to dynamically calculate the buffer depth adjustment amount, and the ring buffer depth changes dynamically with the value of the frequency domain energy distribution parameter. The feature tag state jump triggers the injection of the timestamp check mark, and the microsecond clock mark is embedded in the data frame header. The hardware feature tag directly controls the resource configuration and timing synchronization of the transmission layer, breaking through the limitations of the fixed sampling transmission mode.

[0099] The data stream with timestamp checksums output by the bidirectional collaborative transmission module is fed into the composite modal statistics module. This module parses the timestamp checksum sequence and calculates the percentage of timestamp loss per second. When the loss percentage exceeds a threshold, a wavelet packet time-frequency fusion algorithm is invoked from a pre-set algorithm library. The algorithm selects a correlation with data transmission quality: a high-robustness algorithm is automatically used in high-loss scenarios, and a high-precision algorithm is used in low-loss scenarios. The statistical window is segmented by aligning the signal's physical period, and period boundaries are matched using zero-crossing detection. Wavelet packet decomposition is performed up to the fifth layer to calculate subband energy entropy and output the mean time-frequency entropy statistic. The statistical data annotation algorithm selects a method for processing identifier declarations, forming a dynamic adaptation mechanism for analysis strategies and data quality.

[0100] The statistical data identified by the labeling algorithm is input into the cross-validation error modeling module. Based on the algorithm's identification, the module selects dynamic range weighting or fuzzy cognitive mapping calculations, integrating external vibration frequency and amplitude parameters to generate a vibration rectification error quantification model. After the confidence assessment factor distribution graph is output, the average confidence assessment factor is calculated as a feedback signal. When the value falls below the threshold, dual-path optimization is triggered: the composite modal statistics module performs invisible window optimization, reconstructing the physical period window based on the user-selected interval and regenerating statistical data; the adaptive signal acquisition module reduces the weight allocation ratio of low-confidence frequency bands and optimizes the feature extraction strategy. This feedback loop enables the coordinated evolution of hardware sampling and software analysis.

[0101] The system maintains feature integrity through standardized data streams: the adaptive signal acquisition module outputs a digital signal sequence with signature tags (raw signal block + signature tag block), the bidirectional collaborative transmission module outputs a data stream with timestamp verification identifiers (signal frames are bound to timestamp identifiers), and the composite modal statistics module outputs statistical data (time-frequency entropy mean + algorithm identifier) ​​labeled with algorithm selection identifiers. This encapsulation format eliminates information degradation when transferring data from multiple tools and ensures the cross-module linkage of signature tags, timing information, and processing methods.

[0102] The present invention constructs a three-level feature transmission chain of hardware feature tagging, transmission timing identification, and statistical algorithm identification. Through feature-driven transmission configuration, timing quality adaptation analysis, and dual-path feedback optimization, it realizes deep collaboration between collection and analysis, and ultimately achieves improved real-time segmented statistical efficiency of dynamic data and enhanced accuracy of transient feature extraction.

[0103] The specific embodiment of the present invention is based on the accelerometer vibration rectification error test scenario. It addresses the problem of low dynamic data statistical efficiency and inaccurate transient feature extraction caused by the functional separation of data acquisition hardware and back-end analysis software. The technical problem is solved through the following technical contents: The adaptive signal acquisition module acquires accelerometer current signals in real time via a multi-core analog-to-digital conversion array. The array uses parallel conversion channels to synchronously process multiple signals, outputting the raw digital signal and capturing the preceding time window data segment. A Fast Fourier Transform (FFT) is performed on the data segment to calculate the energy distribution parameters of the target frequency band. When the energy percentage exceeds a set threshold, a sampling rate increase command is triggered. Based on the energy distribution parameters, a signature is generated and appended to the digital signal sequence. This process enables real-time hardware-level feature extraction, addressing the issue of missing high-frequency transient features in fixed sampling modes. The signature carries a summary of the spectral characteristics to drive downstream collaborative optimization.

[0104] The bidirectional collaborative transmission module receives signature-tagged digital signal sequences and analyzes the energy percentage values ​​within the signatures. Based on these values ​​and a preset scaling factor, it dynamically calculates buffer depth adjustments and sets the ring buffer depth in real time. It also synchronously detects signature state transitions and embeds a microsecond-level timestamp checksum in the corresponding data frame header. This structured data stream is output to the composite modal statistics module, addressing timing misalignment caused by discrete devices. The transport layer uses signatures to drive resource allocation, preventing data overflow in high-vibration scenarios.

[0105] The composite modal statistics module parses the timestamp checksum identifier sequence in the data stream and calculates the percentage of identifier loss per second. When the loss percentage exceeds a threshold, the wavelet packet time-frequency fusion algorithm is invoked. Window alignment is performed on the data stream using physical period segmentation, and zero-crossing detection is used to match the signal's inherent period boundaries. Five-layer wavelet packet decomposition is performed to calculate subband energy entropy and output the mean time-frequency entropy statistic. The statistical data annotation algorithm selects the identifier declaration processing method. When the user selects a time period of interest, invisible window optimization is triggered: the dominant frequency component of the starting point is extracted, the physical window is expanded to an integer multiple of the period based on the formula N×1 / dominant frequency component, and the statistical process is re-executed within the window. This mechanism eliminates artificial truncation errors and improves the accuracy of impulse response analysis.

[0106] The cross-validation error modeling module selects either the dynamic range weighting or fuzzy cognitive mapping algorithm based on the algorithm identifier. It integrates the time-frequency entropy mean statistic with external vibration frequency and amplitude parameters to generate a vibration rectification error quantification model. After the confidence assessment factor distribution graph is output, the average confidence assessment factor is fed back to the statistics and acquisition module. If the confidence assessment factor falls below the threshold, the statistics module performs invisible window optimization to regenerate the data. The acquisition module analyzes the distribution graph to identify low-confidence frequency bands and reduces the weight allocation to these bands. This dual-path feedback loop forms a co-evolutionary closed loop between hardware sampling and software analysis.

[0107] The system transmits spectral characteristics through a sequence of signature-tagged digital signals. A data stream with timestamp verification ensures timing synchronization, and statistical data labeled with algorithm selection declares processing methods. Structured encapsulation eliminates information degradation associated with multi-tool transfers, enabling end-to-end optimization from current signal input to error model output.

[0108] The technical features of the present invention are explained as follows: In this method, the adaptive signal acquisition module performs a fast Fourier transform on the raw digital signal to calculate the signal's energy concentration within a specific frequency band. The target frequency band covers the core spectral range of the accelerometer's vibration response. The energy distribution parameters characterize the signal's energy distribution in the frequency domain. The frequency domain energy distribution parameters are digitally encoded to generate signatures, which serve as machine-readable summaries of the signal's spectral characteristics.

[0109] Dynamic buffer depth adjustment strategy: The bidirectional cooperative transmission module calculates the ring buffer depth adjustment based on the energy percentage value in the signature and a preset scaling factor. The buffer depth adjustment is the product of the initial depth value and the energy percentage value multiplied by the scaling factor. The ring buffer depth is set to this calculated adjustment, ensuring that the storage capacity matches the changes in signal signature strength in real time.

[0110] The wavelet packet time-frequency fusion algorithm, which is invoked by the composite modal statistics module when the percentage of missing timestamps exceeds a threshold, performs a five-layer wavelet packet decomposition, decomposing the time-domain signal into multiple subband frequency-domain components. The algorithm then calculates the energy entropy characteristics of each subband, quantifying the complexity of the subband signal's energy distribution. The energy entropy of all subbands is aggregated to produce the mean time-frequency entropy, which represents the global time-frequency characteristics of the signal.

[0111] The cross-validation error modeling module uses a dynamic range weighting algorithm. When the algorithm selection flag indicates the wavelet packet time-frequency fusion algorithm, it performs a dynamic range weighting calculation. The algorithm calculates the data range based on the mean time-frequency entropy feature and dynamically assigns weight coefficients to enhance the contribution of high-frequency transient features to the model. The weight assignment is positively correlated with the degree of feature dispersion.

[0112] The fuzzy cognitive mapping algorithm, when the algorithm selection flag indicates a non-wavelet packet time-frequency fusion algorithm, the modeling module performs a fuzzy cognitive mapping calculation. The algorithm constructs a nonlinear relationship network between input features and output results, and handles data uncertainty through fuzzy rule reasoning. The cognitive mapping network is suitable for modeling scenarios with low-quality data.

[0113] Invisible window optimization strategy: The composite modal statistics module receives a user-selected time interval command and extracts the dominant frequency component of the data at the start of the time interval. Based on this dominant frequency component, the physical extension length is calculated, which is an integer multiple of the signal period length. The optimized statistical window boundaries are set to cover the original interval and the physically extended area, and data analysis and feature calculations are re-performed within the window.

[0114] A dual-path feedback control strategy distributes the average confidence factor of the confidence factor distribution map as the feedback signal. When the composite modal statistics module receives feedback signals below a threshold, it triggers invisible window optimization to regenerate statistics. The adaptive signal acquisition module analyzes the distribution map to identify low-confidence frequency intervals and reduces the weight allocated to these intervals in the frequency domain energy distribution parameters. This dual-path collaboratively optimizes hardware sampling and software analysis accuracy.

Claims

1. An accelerometer vibration rectification error test system, characterized in that: include; an adaptive signal acquisition module, which acquires the current signal output by the accelerometer and performs analog-to-digital conversion to output the original digital signal, extracts the frequency domain energy distribution parameters of the original digital signal, generates a signature based on the frequency domain energy distribution parameters, appends the signature to the digital signal sequence, and outputs the digital signal sequence with the signature; The bidirectional collaborative transmission module receives the digital signal sequence output by the adaptive signal acquisition module, analyzes the characteristic mark in the digital signal sequence, dynamically calculates the buffer depth adjustment amount according to the frequency domain energy distribution parameter value carried in the characteristic mark, sets the ring buffer depth to the buffer depth adjustment amount, synchronously detects the state transition edge of the characteristic mark, appends a timestamp check mark to the corresponding data frame header when the transition edge occurs, and outputs the data stream integrated with the timestamp check mark to the composite modal statistics module; The composite modal statistics module receives the data stream output by the bidirectional collaborative transmission module, parses the timestamp check mark in the data stream, detects the missing ratio of the timestamp check mark, and when the missing ratio exceeds a preset threshold, selects the wavelet packet time-frequency fusion algorithm from the preset algorithm library, aligns the signal physical period segmentation statistical window, and generates statistical data based on the signal analysis results. The statistical data includes at least the mean time-frequency entropy and is marked with the corresponding algorithm selection mark; The cross-validation error modeling module receives the statistical data output by the composite modal statistics module, extracts the algorithm selection identifier in the statistical data, and adopts the range dynamic weighted calculation when the algorithm selection identifier indicates the use of the wavelet packet time-frequency fusion algorithm; when the algorithm selection identifier indicates the use of the preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, it adopts the fuzzy cognitive mapping calculation, and fuses the statistical data to output the vibration rectification error quantification model and the confidence assessment factor distribution map.

2. The accelerometer vibration rectification error testing system according to claim 1, characterized in that: The adaptive signal acquisition module includes a multi-core analog-to-digital conversion array; The multi-core analog-to-digital conversion array inputs the current signal output by the accelerometer, performs analog-to-digital conversion and outputs the original digital signal; Extract the first 10ms time window of the original digital signal and calculate the energy proportion of the signal in the first 10ms time window in the 10-500Hz frequency band; When the energy ratio exceeds 20%, the sampling rate increase instruction is output to the multi-core analog-to-digital conversion array, a feature tag is generated based on the energy ratio, and the feature tag is transmitted to the bidirectional collaborative transmission module.

3. The accelerometer vibration rectification error testing system according to claim 2, characterized in that: The bidirectional collaborative transmission module is configured as follows: receiving a digital signal sequence with characteristic marks output by the adaptive signal acquisition module, and parsing the characteristic marks in the digital signal sequence; Calculate the buffer depth adjustment amount based on the energy ratio value in the feature mark and the preset scaling factor, where the buffer depth adjustment amount is the product of the initial depth value superimposed on the energy ratio value and the preset scaling factor; Set the ring buffer depth to the buffer depth adjustment amount; Detect the state transition edge of the characteristic mark. When the state transition edge occurs, add a timestamp check mark to the data frame header, and output the data stream integrated with the timestamp check mark to the composite modal statistics module.

4. The accelerometer vibration rectification error testing system according to claim 3, characterized in that: The configuration of the composite modal statistics module is: Receive the data stream output by the bidirectional cooperative transmission module, and parse the timestamp verification identification sequence from the data stream; Count the number of timestamp verification sequence losses per second; When the number of losses exceeds 5%, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library; Align the received data stream with the signal physical period segmentation statistical window; Perform wavelet packet decomposition to the fifth layer and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all sub-bands, and statistical data including the mean time-frequency entropy is generated; the wavelet packet time-frequency fusion algorithm selection identifier is marked, and the statistical data marked with the wavelet packet time-frequency fusion algorithm selection identifier is output to the cross-validation error modeling module.

5. The accelerometer vibration rectification error testing system according to claim 4, characterized in that: The Composite Modal Statistics module is also configured to: Create a statistical data object including a time-frequency entropy mean field based on the time-frequency entropy mean, the signal summary hash stored in the composite modal statistics module, and the statistical window boundary coordinates; An algorithm selection identifier is marked on the statistical data object, and the statistical data object marked with the algorithm selection identifier is output to the cross-validation error modeling module.

6. The accelerometer vibration rectification error testing system according to claim 5, characterized in that: The cross-validation error modeling module is configured as: Receive the statistical data object identified by the annotation algorithm selected by the composite modal statistics module as a statistical data set; Extract algorithmic selection identifiers from statistical data sets; When the algorithm selection flag indicates the use of the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the range dynamic weighting algorithm to perform calculations; When the algorithm selection flag indicates to use a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, the statistical data set is input into the fuzzy cognitive mapping algorithm to perform calculations; Receive external input vibration frequency and amplitude parameters; The calculation results of the range dynamic weighting algorithm or the fuzzy cognitive mapping algorithm are integrated with the vibration frequency and amplitude parameters to generate a vibration rectification error quantization model; The confidence assessment factor is calculated based on the vibration rectification error quantization model, and a confidence assessment factor distribution graph is output.

7. The accelerometer vibration rectification error testing system according to claim 6, characterized in that: The Composite Modal Statistics module is also configured to: Receive the user's time interval selection instruction, parse the user's time interval selection instruction, and obtain the coordinates of the time interval start and end points; intercepting a data segment between a starting point and an ending point from the data stream output by the bidirectional cooperative transmission module; For the data at the starting point of the time interval of the intercepted data segment, the frequency domain energy distribution parameter extraction method of the adaptive signal acquisition module is used to calculate the frequency domain main frequency component; The physical extension length is calculated based on the main frequency component in the frequency domain, where the physical extension length is equal to an integer N multiplied by 1 divided by the main frequency component, where N is an integer; Set the optimization statistics window boundary. The starting point of the optimization statistics window boundary is the starting point of the time interval, and the end point is the starting point of the time interval plus the physical extension length. Within the optimization statistics window boundary, reparse the timestamp check mark sequence and detect the missing timestamp check mark ratio; When the missing ratio exceeds the preset threshold, the wavelet packet time-frequency fusion algorithm is selected from the preset algorithm library to align and optimize the physical period segmentation of the data stream within the statistical window, perform wavelet packet decomposition to the fifth layer, and calculate the energy entropy of all decomposed subbands; The mean time-frequency entropy is calculated based on the energy entropy of all subbands, and statistical data including the mean time-frequency entropy is generated, and the algorithm selection identifier is marked.

8. The accelerometer vibration rectification error testing system according to claim 7, characterized in that: Also includes: After the cross-validation error modeling module generates a confidence assessment factor distribution graph, it calculates the average confidence assessment factor of the distribution graph and outputs the average confidence assessment factor feedback signal to the composite modal statistics module and the adaptive signal acquisition module; When the feedback signal value received by the composite modal statistics module is lower than 0.9, the invisible window optimization process is executed: Receive user selection instructions and parse time interval coordinates, intercept data segments and calculate the main frequency components, set the optimization window and reprocess the data, and output the regenerated statistical data; When the feedback signal value received by the adaptive signal acquisition module is lower than 0.9, the confidence assessment factor distribution diagram is analyzed to identify the frequency interval with a confidence level less than 0.8, and the weight distribution ratio of the frequency interval in the frequency domain energy distribution parameter is reduced.

9. The accelerometer vibration rectification error testing system according to claim 8, characterized in that: Also includes: The input of the bidirectional cooperative transmission module is the digital signal sequence with feature marks output by the adaptive signal acquisition module; The input of the composite modal statistics module is the data stream with timestamp verification mark output by the bidirectional collaborative transmission module; The input of the cross-validation error modeling module is the statistical data selected by the labeling algorithm output by the composite modal statistics module.

10. A method for testing an accelerometer vibration rectification error, applied to the accelerometer vibration rectification error testing system according to any one of claims 1 to 9, characterized in that: include: Step 1: Acquire the current signal output by the accelerometer and perform analog-to-digital conversion to output the original digital signal, extract the frequency domain energy distribution parameters of the original digital signal, generate a feature marker based on the frequency domain energy distribution parameters, attach the feature marker to the digital signal sequence, and output the digital signal sequence with the feature marker; Step 2: parse the characteristic markers in the digital signal sequence, dynamically calculate the buffer depth adjustment amount based on the frequency domain energy distribution parameter value carried in the characteristic markers, set the ring buffer depth to the buffer depth adjustment amount, synchronously detect the state transition edge of the characteristic markers, and append a timestamp check mark to the corresponding data frame header when the transition edge occurs. Output the data stream with the integrated timestamp check mark to step 3; Step 3: parse the timestamp checksum in the data stream and detect the missing ratio of the timestamp checksum. When the missing ratio exceeds a preset threshold, select the wavelet packet time-frequency fusion algorithm from the preset algorithm library, align the signal physical period to divide the statistical window, and generate statistical data based on the signal analysis results. The statistical data includes at least the mean time-frequency entropy and is marked with the corresponding algorithm selection identifier. Step 4, extract the algorithm selection mark in the statistical data. When the algorithm selection mark indicates the use of the wavelet packet time-frequency fusion algorithm, the range dynamic weighted calculation is adopted. When the algorithm selection mark indicates the use of a preset statistical algorithm other than the wavelet packet time-frequency fusion algorithm, the fuzzy cognitive mapping calculation is adopted. The statistical data is fused to output the vibration rectification error quantization model and the confidence assessment factor distribution map.

Images