High-precision data acquisition method and system of MEMS sensor

Abstract

The invention relates to a high-precision data acquisition method and system for an MEMS sensor, and the method comprises the following steps: carrying out the multi-channel parallel sampling of an initial electric signal outputted by the MEMS sensor, and obtaining an original data flow matrix; performing adaptive denoising processing on the original data stream matrix to obtain a denoised data set; performing signal phase reconstruction on the noise reduction data set through a phase unwrapping technology to obtain a continuous phase feature sequence; performing dynamic drift compensation on the continuous phase feature sequence to obtain a calibration signal vector; performing transient characteristic extraction on the calibration signal vector to obtain a sensor response characteristic curve; and performing high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technology to obtain a high-precision data acquisition result, thereby solving the technical problem of how to effectively remove noise interference generated in a multi-channel parallel sampling process.

Description

Technical Field

[0001] The present invention relates to the technical field of MEMS sensors, and in particular to a high-precision data acquisition method and system for MEMS sensors. Background Art

[0002] In modern industry and scientific research, the demand for high-precision data acquisition is growing, particularly in the field of microelectromechanical systems (MEMS) sensors. MEMS sensors, due to their small size, low power consumption, and high cost-effectiveness, have found widespread application in a variety of fields, including consumer electronics, medical devices, environmental monitoring, and the automotive industry. However, with the continuous expansion of application scenarios and increasing technical requirements, traditional data acquisition methods have gradually exposed their limitations, especially in areas such as signal noise suppression, accurate phase information extraction, and long-term stability. These issues directly affect the data acquisition accuracy and reliability of MEMS sensors, limiting their potential in higher-end applications.

[0003] Faced with these challenges, researchers have begun exploring more advanced data processing techniques to improve the data acquisition quality of MEMS sensors. For example, effectively removing the noise interference generated by multi-channel parallel sampling has become a key research topic. Furthermore, because MEMS sensor output signals often exhibit complex transient characteristics and potential dynamic drift, accurately capturing these characteristics and effectively compensating for them is also a pressing issue. Existing solutions often struggle to simultaneously meet the requirements of efficient denoising, precise phase reconstruction, and stable calibration, resulting in the inability to fully demonstrate the performance of MEMS sensors in practical applications.

[0004] Therefore, it is particularly important to develop a high-precision data acquisition method that comprehensively considers multiple factors, including noise reduction, phase reconstruction, dynamic drift compensation, and transient characteristic extraction. This new method not only needs to overcome the problems existing in traditional technologies but also needs to adapt to changes in different operating environments, ensuring that MEMS sensors can maintain high-precision data acquisition capabilities in complex environments. By introducing advanced methods such as adaptive quantization coding and phase unwrapping technology, it is expected to significantly improve the performance of MEMS sensors, expand their application range, and provide strong support for further development in related fields. Summary of the Invention

[0005] The main purpose of the present invention is to provide a high-precision data acquisition method and system for MEMS sensors, which solves the technical problem of how to effectively remove noise interference generated during multi-channel parallel sampling.

[0006] To achieve the above object, the present invention provides a high-precision data acquisition method for a MEMS sensor, comprising the following steps: Perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain the original data stream matrix; Performing adaptive denoising processing on the original data stream matrix to obtain a denoised data set; Reconstructing the signal phase of the noise reduction data set by phase unwrapping technology to obtain a continuous phase feature sequence; Performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; Extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve; Based on the adaptive quantization coding technology, high-precision data resampling is performed on the sensor response characteristic curve to obtain a high-precision data acquisition result.

[0007] Furthermore, the initial electrical signal is a weak electrical signal, and the initial electrical signal output by the MEMS sensor is sampled in parallel through multiple channels to obtain an original data stream matrix, including: The weak electrical signal of the MEMS sensor is differentially amplified by a high-impedance preamplifier to obtain a gain-matched signal group; Performing multi-channel clock synchronization on the gain matching signal group based on a synchronous trigger circuit to obtain a phase-locked sampling point sequence; The phase-locked sampling point sequence is dynamically adjusted through a programmable gain unit to obtain an original data stream matrix; wherein the original data stream matrix includes a sensor multi-axis output signal, a reference voltage signal, and a temperature compensation signal.

[0008] Furthermore, the adaptive denoising process is performed on the original data stream matrix to obtain a denoised data set, including: Performing multi-scale transformation decomposition on the original data stream matrix using orthogonal wavelet basis functions to obtain a wavelet coefficient sequence, and performing frequency band energy calculation on the wavelet coefficient sequence to obtain a frequency band energy distribution feature; wherein the frequency band energy distribution feature includes an approximate coefficient group, a detail coefficient set, and an edge feature identifier; An optimal threshold is estimated for the frequency band energy distribution characteristics based on the Bayesian risk criterion to obtain an adaptive threshold matrix, and a noise distribution model is performed on the adaptive threshold matrix using an interval correlation analysis method to obtain a noise probability density function; The noise probability density function is subjected to coefficient screening and reconstruction using a preset soft threshold shrinkage function to obtain a first-order denoised signal, and edge-preserving enhancement processing is performed on the first-order denoised signal to obtain an edge-enhanced signal set; wherein the edge-enhanced signal set includes a main frequency eigenvector, a noise suppression rate parameter, and a signal mutation point identifier; Performing a cross-channel consistency check on the edge enhancement signal set based on an adaptive correction factor to obtain a correction signal group, and performing residual noise suppression on the correction signal group through singular value filtering to obtain a second-order denoised signal; The second-order denoised signal is subjected to spectrum shaping processing by a time-varying filter to obtain a denoised data set.

[0009] Furthermore, the signal phase reconstruction of the noise reduction data set is performed by the phase unwrapping technology to obtain a continuous phase feature sequence, including: Performing a complex domain conversion on the noise reduction data set by Hilbert transform to obtain an analytical signal sequence; wherein the analytical signal sequence includes a real component, an imaginary component and an instantaneous amplitude feature; The phase angle of the analytical signal sequence is extracted based on the phase gradient descent method to obtain a wrapped phase sequence, and the phase jump point detection of the wrapped phase sequence is performed using a two-dimensional phase consistency constraint technology to obtain a phase jump mark set; wherein the phase jump mark set includes a jump position index, a jump amplitude value, and a jump direction identifier; Performing jump accumulation calculation on the phase jump marker set using a quality-guided path integration technique to obtain a phase correction amount, and performing 2π phase compensation on the wrapped phase sequence based on the phase correction amount using a phase unwrapping technique to obtain an initial unwrapped phase map; wherein the initial unwrapped phase map includes continuous phase values, phase residuals, and a quality factor matrix; The initial unwrapped phase image is calibrated with multi-sensor data through a multi-channel phase fusion method to obtain a continuous phase feature sequence; wherein the continuous phase feature sequence includes absolute phase values, phase continuity data and sensor relative phase delay characteristics.

[0010] Furthermore, the performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector includes: Performing system dynamic modeling on the continuous phase characteristic sequence through state space equations to obtain a sensor state transfer matrix; Characterizing the noise characteristics of the sensor state transfer matrix based on measurement covariance analysis to obtain the noise covariance structure of the sensor system, and estimating the state quantity of the sensor system noise covariance structure through forward recursive prediction technology to obtain a priori state estimation value; Performing Kalman gain calculation on the prior state estimate based on the minimum mean square error criterion to obtain an adaptive weight coefficient, and performing state correction processing on the adaptive weight coefficient through a state update equation to obtain a posterior state estimate; Numerical stability enhancement is performed on the posterior state estimator by using a square root information filtering technique to obtain a robust state vector, and outlier detection is performed on the robust state vector based on a residual hypothesis test to obtain an outlier identification set; The abnormal point identification set is locally reconstructed based on a piecewise polynomial smoothing kernel to obtain a repaired phase sequence, and the repaired phase sequence is corrected for temperature drift through a temperature compensation mechanism to obtain a calibration signal vector; wherein, the calibration signal vector includes a temperature compensation coefficient, a zero bias correction value, and a sensitivity correction parameter.

[0011] Furthermore, the step of extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve includes: Performing adaptive multi-scale decomposition on the calibration signal vector by empirical mode decomposition to obtain an intrinsic mode function set; wherein the intrinsic mode function set includes a multi-order component sequence, a residual trend term, and a modal energy density distribution; Based on the instantaneous frequency calculation framework, the non-stationary characteristics of the intrinsic mode function set are analyzed to obtain a time-frequency energy distribution spectrum, and the edge of the time-frequency energy distribution spectrum is enhanced by the frequency boundary sharpening technology to obtain a fine time-frequency structure; wherein the fine time-frequency structure includes the instantaneous frequency trajectory, the energy density matrix and the demarcation point feature set; Resonant modes of the fine time-frequency structure are identified through a subspace frequency tracking mechanism to obtain a modal parameter spectrum, and modal separation measurement is performed on the modal parameter spectrum based on a spectral manifold distance measure to obtain a modal separation eigenvector; wherein the modal separation eigenvector includes a modal center frequency, a modal damping ratio, and a modal coherence index; Performing group delay correction on the modal separation eigenvector based on a nonlinear phase compensation technique to obtain a phase-corrected spectrum, and performing amplitude modulation characteristic separation on the phase-corrected spectrum using a transient envelope extractor to obtain a modulation feature group; wherein the modulation feature group includes an envelope contour curve, a modulation depth parameter, and a transient response time constant; The modulation feature group is subjected to time domain remapping based on a multi-resolution reconstruction technique to obtain a sensor response characteristic curve; wherein the sensor response characteristic curve is a dynamic sensitivity function, an amplitude-frequency characteristic curve, and a phase-frequency characteristic curve.

[0012] Furthermore, the sensor response characteristic curve is subjected to high-precision data resampling based on the adaptive quantization coding technology to obtain a high-precision data acquisition result, including: Performing information entropy analysis on the sensor response characteristic curve through a non-uniform signal sampling mechanism to obtain a key point density distribution map; wherein the key point density distribution map includes an information gradient vector, a mutation point clustering structure, and a transient region boundary marker; Dynamically adjusting the bit depth of the key point density distribution map based on a segmented quantization precision allocation strategy to obtain an optimized bit depth allocation scheme, and reducing redundancy of the optimized bit depth allocation scheme through an adaptive quantization coding technology to obtain a compression mapping table; Performing high-precision interpolation reconstruction on the compressed mapping table using a fractional-order interpolation kernel function to obtain a fine sampling point sequence, and performing quantization error dispersion processing on the fine sampling point sequence based on orthogonal transform domain quantization to obtain a quantization coefficient matrix; wherein the quantization coefficient matrix includes basis vector combination weights, a quantization step size parameter, and a perceptual masking threshold; Performing quantization noise shaping on the quantization coefficient matrix based on nonlinear quantization boundary optimization to obtain a noise-shaped data stream, and performing signal recovery processing on the noise-shaped data stream through an adaptive inverse quantization technique to obtain a reconstructed data sequence; The reconstructed data sequence is subjected to residual correction processing through an error feedback compensation network to obtain a high-precision data acquisition result.

[0013] The present invention also provides a high-precision data acquisition system for a MEMS sensor, comprising: The first sampling module is used to perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain an original data stream matrix; A denoising module, configured to perform adaptive denoising on the original data stream matrix to obtain a denoised data set; A reconstruction module, configured to reconstruct the signal phase of the noise reduction data set by using a phase unwrapping technique to obtain a continuous phase feature sequence; A compensation module, configured to perform dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; an extraction module, configured to extract transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve; The second sampling module is used to perform high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technology to obtain a high-precision data acquisition result.

[0014] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the above methods are implemented.

[0016] The present invention provides a high-precision data acquisition method for a MEMS sensor, comprising the following steps: performing multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain a raw data stream matrix; performing adaptive denoising on the raw data stream matrix to obtain a de-noised data set; reconstructing the signal phase of the de-noised data set using a phase unwrapping technique to obtain a continuous phase characteristic sequence; performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; extracting transient characteristics from the calibration signal vector to obtain a sensor response characteristic curve; and performing high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technique to obtain a high-precision data acquisition result. This method solves the technical problem of how to effectively remove noise interference generated during multi-channel parallel sampling, implements high-precision data resampling of the sensor response characteristic curve using the adaptive quantization coding technique, and can further improve data resolution and accuracy while maintaining data integrity. The method dynamically adjusts the quantization level according to the specific characteristics of the signal, thereby achieving efficient and accurate data compression and representation, providing strong support for subsequent data analysis and applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 1 is a schematic diagram of the steps of a high-precision data acquisition method for a MEMS sensor according to one embodiment of the present invention; Figure 2 This is a block diagram of a high-precision data acquisition system for a MEMS sensor according to an embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] like Figure 1 As shown, Figure 1 This is a schematic diagram of the steps of a high-precision data acquisition method for a MEMS sensor in one embodiment of the present invention; In one embodiment of the present invention, a high-precision data acquisition method for a MEMS sensor is provided, comprising the following steps: Step S1: Perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain an original data stream matrix.

[0021] Specifically, multi-channel parallel sampling of the initial electrical signals output by MEMS sensors to generate a raw data stream matrix is ​​a fundamental step in achieving high-precision data acquisition. It ensures a rich and accurate data source for subsequent processing. Specifically, this step first requires designing a multi-channel parallel sampling architecture at the hardware level. This architecture can simultaneously receive signals from multiple MEMS sensors, representing different physical quantities or different dimensions of the same physical quantity. For example, in an environmental monitoring application, MEMS sensors may be used to simultaneously measure multiple environmental parameters such as temperature, humidity, and air pressure. Each sensor corresponds to a specific physical quantity, and each sensor outputs a continuously varying electrical signal. Subsequently, a synchronized triggering mechanism enables sampling operations on all channels to be initiated at the same time, ensuring data consistency and accuracy even with slight time differences between channels. This synchronization is crucial for subsequent data analysis, as it allows direct comparison or joint analysis of data from different channels, revealing deeper insights. During this process, the discretized electrical signals collected by each channel are stored in a two-dimensional array, forming the so-called raw data stream matrix. The term "raw" here means that this data has not yet undergone any filtering or correction processing, retaining its most original information characteristics. For example, in the environmental monitoring example mentioned above, if the system collects 100 samples per second from each sensor, a raw data stream matrix containing 300 samples (assuming three sensors) will be generated within one second. This unprocessed data provides the necessary input for the subsequent adaptive denoising process, ensuring the consistency and efficiency of the entire data processing pipeline. In addition, the data obtained in this way supports further complex analysis steps such as phase unwrapping and dynamic drift compensation, ultimately achieving a comprehensive understanding and optimized utilization of the MEMS sensor output signal.

[0022] Step S2: performing adaptive denoising processing on the original data stream matrix to obtain a denoised data set.

[0023] Specifically, adaptive denoising is performed on the raw data stream matrix to produce a de-noised data set. This process aims to remove noise interference from the collected data using intelligent algorithms, thereby improving data purity and reliability. First, implementing adaptive denoising requires the use of advanced signal processing techniques, such as wavelet transforms, adaptive filtering, or machine learning-based methods, to identify and eliminate noise. These methods automatically adjust parameters based on the data's characteristics to achieve optimal denoising results. For example, in environmental monitoring applications, MEMS sensors may be affected by external electromagnetic interference or other environmental factors, resulting in the collected data containing unwanted noise. In this case, adaptive denoising can dynamically select the most appropriate filtering strategy by analyzing the data characteristics of each channel in the raw data stream matrix. In practice, the system first performs a preliminary analysis of the raw data stream matrix to identify possible noise patterns. Based on these patterns, a corresponding denoising algorithm is selected and applied to the data of each channel. This approach not only accounts for correlations between different channels but also effectively addresses the impact of non-stationary noise. For example, in a scenario where temperature, humidity, and air pressure are simultaneously monitored, sudden interference may cause abnormal fluctuations in the data at a given moment. Adaptive denoising can detect this anomaly and smooth it out, making the resulting denoised data set more accurate and reliable. Furthermore, to further enhance the denoising effect, the system can also introduce a feedback mechanism to monitor the quality of the denoised data in real time and adjust the algorithm parameters based on the actual results. This ensures that the output data has a high signal-to-noise ratio, even in the face of complex environmental changes. This provides high-quality input data for subsequent steps such as phase unwrapping and dynamic drift compensation, thereby ensuring the stability and accuracy of the entire high-precision data acquisition process. For example, in applications where subtle changes in the environment are monitored continuously over long periods of time, data that has undergone adaptive denoising can more accurately reflect the true trends in environmental parameter changes, supporting scientific research and decision-making.

[0024] Step S3: reconstructing the signal phase of the noise reduction data set by using a phase unwrapping technique to obtain a continuous phase feature sequence.

[0025] Specifically, phase unwrapping is used to reconstruct the signal phase of a denoised data set, yielding a continuous phase signature sequence. This process is a crucial step in ensuring high-precision data acquisition. After adaptive denoising, a relatively clean denoised data set is obtained. However, the phase information within this data may still be discontinuous or fuzzy, particularly when sampling multiple channels in parallel. To restore accurate and continuous phase signatures, phase unwrapping is required. This technique analyzes the data characteristics of each channel in the denoised data set to identify and correct phase jumps caused by noise or other factors. Specifically, phase unwrapping first performs a preliminary phase estimate for each channel's data and then detects discontinuities by comparing the phase differences between adjacent data points. Once a phase jump is detected, the system automatically adjusts the phase value based on the trend of the local phase change, ensuring that the phase information across the entire data set is continuous and smooth. For example, in an environmental monitoring application, consider a MEMS sensor used to simultaneously measure changes in temperature, humidity, and air pressure. These microscopic variations in these physical quantities can cause subtle phase shifts in the sensor's output signal. If these changes are not correctly captured and interpreted, they will directly affect subsequent data analysis results. Therefore, phase unwrapping technology can effectively remove these discontinuities and restore the true phase information. Furthermore, phase unwrapping technology can combine data from multiple channels for joint processing, further improving the accuracy of phase reconstruction. For example, in the environmental monitoring example mentioned above, there may be certain correlations between different sensors. Leveraging this correlation can help more accurately infer the true phase information of each channel. After phase unwrapping, the data forms a continuous phase signature sequence. This sequence not only preserves the key characteristics of the original signal but also eliminates phase distortion caused by noise or other interference factors. This allows subsequent dynamic drift compensation steps to be performed based on more accurate and reliable phase information, ensuring the stability and accuracy of the entire high-precision data acquisition process. For example, in studies of long-term monitoring of subtle environmental changes, continuous phase signature sequences can provide more precise time series data, supporting in-depth analysis and prediction of environmental parameter trends.

[0026] Step S4: Dynamically compensate for the continuous phase characteristic sequence to obtain a calibration signal vector.

[0027] Specifically, dynamic drift compensation is performed on a continuous phase signature sequence to obtain a calibration signal vector. This process aims to eliminate long-term drift in the sensor output signal, ensuring data stability and accuracy. After phase unwrapping, a continuous and accurate phase signature sequence is obtained. However, this data may still be affected by time-dependent dynamic drift, which is often caused by environmental changes, sensor aging, or other long-term factors. To address these issues, dynamic drift compensation technology is introduced into the data processing workflow. This technology first comprehensively analyzes the continuous phase signature sequence to identify potential drift patterns and trends. Specifically, the system develops mathematical models to describe and predict drift behavior. These models can be based on historical data or pre-defined drift patterns. For example, in an environmental monitoring application, suppose a MEMS sensor is used to monitor long-term changes in temperature, humidity, and air pressure. Over time, the sensor's output signal may drift due to changes in ambient temperature or its own aging effects. To correct for this drift, the system calculates the difference between the current phase signature sequence and the ideal drift-free state at each time point and adjusts the corresponding phase value accordingly. This approach not only considers global drift trends but also adapts to local variations, achieving precise dynamic compensation. Furthermore, dynamic drift compensation can combine data from multiple channels for joint processing to improve compensation accuracy and reliability. For example, in the environmental monitoring example mentioned above, there may be correlation between different sensors. Leveraging this correlation can help more accurately infer the true drift of each channel. Data processed through dynamic drift compensation forms a calibrated signal vector, which not only eliminates errors caused by long-term drift but also preserves the key characteristics of the original signal. For example, in studies involving long-term monitoring of subtle environmental changes, the calibrated signal vector can provide more stable and reliable time series data, supporting in-depth analysis and prediction of environmental parameter trends. Subsequent transient characteristic extraction steps can then be performed based on this more accurate and stable signal vector, ensuring the stability and accuracy of the entire high-precision data acquisition process. Ultimately, data processed in this manner can provide powerful support for scientific research and practical applications, such as accurate weather forecasting and industrial process control. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, dynamic drift compensation can ensure data consistency over many years, allowing researchers to more accurately analyze climate change trends.

[0028] Step S5: extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve.

[0029] Specifically, transient characteristic extraction is performed on the calibration signal vector to generate a sensor response curve. This process aims to identify and extract the sensor's transient response characteristics from the data after dynamic drift compensation, providing more accurate information for subsequent data analysis and applications. After dynamic drift compensation, a stable and accurate calibration signal vector is obtained. This data contains information about the sensor's response under different operating conditions. To further unlock the valuable information within this data, transient characteristic extraction technology is incorporated into the data processing workflow. This technology analyzes the rapidly changing portions of the calibration signal vector to capture the sensor's transient behavior in response to external stimuli. Specifically, the system first performs comprehensive time and frequency domain analysis on the calibration signal vector to identify its transient characteristics. For example, in an environmental monitoring application, assuming a MEMS sensor is used to monitor long-term changes in temperature, humidity, and air pressure, the sensor will generate transient responses when exposed to sudden environmental changes (such as sudden temperature drops or rises). These transient responses typically manifest as rapid fluctuations or spikes in the signal, reflecting the sensor's sensitivity and response speed to external changes. To extract these characteristics, the system applies advanced signal processing algorithms, such as the Short-Time Fourier Transform (STFT), wavelet transform, or other methods suitable for processing non-stationary signals. These methods decompose the signal into distinct time-frequency components, more clearly revealing the details of transient characteristics. Furthermore, the system combines data from multiple channels for joint analysis to improve the accuracy of transient characteristic extraction. For example, in the environmental monitoring example mentioned above, there may be certain correlations between different sensors. Leveraging this correlation can help more accurately infer the true transient response of each channel. After transient characteristic extraction, the data forms a sensor response curve, which not only demonstrates the sensor's response characteristics under different conditions but also provides important information about its dynamic behavior. For example, in studies of long-term monitoring of subtle environmental changes, the sensor response curve can help researchers better understand the sensor's behavioral patterns and applicable range. This allows subsequent adaptive quantization and encoding steps to be based on a more accurate and detailed sensor response curve, ensuring the stability and accuracy of the entire high-precision data acquisition process. Ultimately, data processed in this way can provide strong support for scientific research and practical applications, such as accurate weather forecasting and industrial process control. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, the sensor response characteristic curve can ensure the quality and consistency of the data, allowing researchers to more accurately analyze climate change trends and make scientific predictions.

[0030] Step S6: performing high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technology to obtain a high-precision data acquisition result.

[0031] Specifically, adaptive quantization coding technology is used to resample the sensor response characteristic curve to achieve high-precision data acquisition results. This process aims to improve the accuracy and efficiency of the final data by optimizing data representation and compression. After extracting transient characteristics, detailed sensor response characteristic curves are obtained, which contain detailed information about the sensor's response under different conditions. However, to further improve the resolution and accuracy of the data and ensure efficient data transmission and processing within limited storage space, adaptive quantization coding technology is required for high-precision data resampling. First, adaptive quantization coding technology dynamically adjusts the quantization level based on the specific characteristics of the sensor response characteristic curve, thereby achieving efficient and accurate data compression and representation. Specifically, the system analyzes key points and changing trends in the sensor response characteristic curve to determine which parts require higher quantization accuracy and which parts can tolerate lower accuracy. For example, in an environmental monitoring application scenario, assuming a MEMS sensor is used to monitor long-term changes in temperature, humidity, and air pressure, the sensor response characteristic curve may exhibit a relatively flat trend in some time periods and rapid fluctuations or spikes in other time periods. For smooth sections, the system can use lower quantization precision to reduce data volume; for rapidly changing sections, higher quantization precision is used to ensure detail is not lost. The system then resamples the sensor response characteristic curve to generate a new set of data points that not only retains the key features of the original curve but also achieves an optimal signal-to-noise ratio during the quantization process. This approach not only improves data compression but also ensures data integrity and accuracy. For example, in the environmental monitoring example mentioned above, data processed through adaptive quantization coding can significantly reduce storage space requirements while maintaining accuracy, enabling long-term continuous monitoring. Furthermore, adaptive quantization coding technology can combine data from multiple channels for joint processing, further improving data consistency and reliability. For example, when simultaneously monitoring multiple environmental parameters, leveraging correlations between channels can help more accurately select the appropriate quantization strategy, thereby improving overall data quality. Ultimately, through this high-precision data resampling method, the system can obtain high-quality data acquisition results with high resolution and accuracy, while also being efficiently transmitted and stored within limited resources. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, adaptive quantization coding technology ensures data quality and consistency, enabling researchers to more accurately analyze climate change trends and make scientific predictions. This optimizes the entire high-precision data acquisition process, from initial signal acquisition to final data processing, providing strong support for scientific research and practical applications.

[0032] In a specific embodiment, the initial electrical signal is a weak electrical signal, and the initial electrical signal output by the MEMS sensor is sampled in parallel through multiple channels to obtain an original data stream matrix, including: The weak electrical signal of the MEMS sensor is differentially amplified by a high-impedance preamplifier to obtain a gain-matched signal group; Performing multi-channel clock synchronization on the gain matching signal group based on a synchronous trigger circuit to obtain a phase-locked sampling point sequence; The phase-locked sampling point sequence is dynamically adjusted through a programmable gain unit to obtain an original data stream matrix; wherein the original data stream matrix includes a sensor multi-axis output signal, a reference voltage signal, and a temperature compensation signal.

[0033] Specifically, when processing weak electrical signals, the initial electrical signals output by MEMS sensors are often very weak and susceptible to noise interference. Therefore, a series of precise operations are required to ensure accurate and reliable data acquisition. First, the weak electrical signals from the MEMS sensors are differentially amplified using a high-impedance preamplifier to produce a gain-matched signal set. The high-impedance preamplifier amplifies the weak signals generated by the sensors to a level that can be effectively processed while maintaining signal integrity and low noise. For example, in environmental monitoring applications, MEMS sensors are used to measure changes in temperature, humidity, and air pressure. The weak electrical signals output by these sensors may be difficult to directly use due to external interference or inherent sensor limitations. Using a high-impedance preamplifier not only significantly enhances signal strength but also effectively suppresses noise, ensuring high-quality data in subsequent processing steps. Next, the gain-matched signal set is synchronized across multiple channels using a synchronous trigger circuit to produce a phase-locked sampling point sequence. The synchronous trigger circuit plays a crucial role in this process, precisely controlling the sampling time of each channel to ensure that data from all channels is acquired at the same time. This synchronization is particularly important for multi-channel parallel sampling, as it ensures data consistency across channels, providing a solid foundation for subsequent data analysis. For example, in the environmental monitoring example mentioned above, if the system simultaneously collects data from multiple sensors, each representing a different physical quantity (such as temperature, humidity, and air pressure), a synchronized trigger circuit ensures that these sensors are sampled at the same instant, avoiding data inconsistencies caused by timing differences. This ensures data consistency and accuracy even in the face of complex environmental changes. Subsequently, a programmable gain unit dynamically adjusts the phase-locked sampling point sequence to generate a raw data stream matrix. The programmable gain unit dynamically adjusts the gain of each channel to accommodate varying signal strengths and ranges. For example, in some cases, the sensor output signal may exceed its normal operating range. In this case, the gain can be reduced to prevent signal saturation. In other cases, if the signal is too weak, the gain can be increased to improve the signal-to-noise ratio. This flexibility enables the system to cope with a variety of complex operating conditions and ensures data acquisition quality. In environmental monitoring applications, assuming the system needs to simultaneously process multiple signals from different sensors (such as multi-axis output signals, reference voltage signals, and temperature compensation signals), the programmable gain unit can dynamically adjust based on the specific characteristics of each signal to ensure that all signals are accurately acquired and processed. Ultimately, the raw data stream matrix generated through the above steps contains not only the multi-axis sensor output signals, but also rich information such as the reference voltage signal and temperature compensation signal. This data provides the necessary input for subsequent steps such as adaptive denoising, phase unwrapping, and dynamic drift compensation.For example, when using MEMS sensors for long-term meteorological data collection in weather stations, the multi-axis output signals in the raw data stream matrix can help researchers fully understand the changing trends of environmental parameters. Reference voltage signals are used to calibrate sensor outputs, ensuring data accuracy, and temperature compensation signals eliminate the effects of temperature changes on sensor performance, further improving data reliability. The entire process, from amplification and synchronization of the initial electrical signal to dynamic range adjustment, forms a complete high-precision data acquisition pipeline, ensuring high-quality and consistent final data. Data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In summary, by performing multi-channel parallel sampling of the initial electrical signals output by MEMS sensors, combined with technical techniques such as high-impedance preamplifiers, synchronous trigger circuits, and programmable gain units, high-precision data acquisition can be achieved, laying a solid foundation for subsequent data processing and analysis.

[0034] In a specific embodiment, the adaptive denoising process is performed on the original data stream matrix to obtain a denoised data set, including: Performing multi-scale transformation decomposition on the original data stream matrix using orthogonal wavelet basis functions to obtain a wavelet coefficient sequence, and performing frequency band energy calculation on the wavelet coefficient sequence to obtain a frequency band energy distribution feature; wherein the frequency band energy distribution feature includes an approximate coefficient group, a detail coefficient set, and an edge feature identifier; An optimal threshold is estimated for the frequency band energy distribution characteristics based on the Bayesian risk criterion to obtain an adaptive threshold matrix, and a noise distribution model is performed on the adaptive threshold matrix using an interval correlation analysis method to obtain a noise probability density function; The noise probability density function is subjected to coefficient screening and reconstruction using a preset soft threshold shrinkage function to obtain a first-order denoised signal, and edge-preserving enhancement processing is performed on the first-order denoised signal to obtain an edge-enhanced signal set; wherein the edge-enhanced signal set includes a main frequency eigenvector, a noise suppression rate parameter, and a signal mutation point identifier; Performing a cross-channel consistency check on the edge enhancement signal set based on an adaptive correction factor to obtain a correction signal group, and performing residual noise suppression on the correction signal group through singular value filtering to obtain a second-order denoised signal; The second-order denoised signal is subjected to spectrum shaping processing by a time-varying filter to obtain a denoised data set.

[0035] Specifically, adaptive denoising of the original data stream matrix to obtain a de-noised data set requires a series of sophisticated signal processing techniques to remove noise interference and ensure data purity and reliability. Specifically, this process begins with multi-scale transform decomposition. The original data stream matrix is ​​decomposed using orthogonal wavelet basis functions to generate a sequence of wavelet coefficients. The frequency band energy distribution characteristics of these coefficients are then calculated. For example, in environmental monitoring applications, consider MEMS sensors used to monitor long-term changes in temperature, humidity, and air pressure. The data output by these sensors may contain significant noise, impacting data accuracy and usability. Using orthogonal wavelet basis functions, the original data stream matrix can be decomposed into sequences of wavelet coefficients at different scales, each corresponding to a different frequency component. This decomposition not only reveals the primary signal characteristics but also effectively separates the noise components. Next, frequency band energy calculations on the wavelet coefficient sequence yield frequency band energy distribution features, including approximate coefficient groups, detail coefficient sets, and edge feature markers. These features describe the signal energy distribution across different frequency bands, helping to identify and distinguish useful signals from noise. For example, in the environmental monitoring example mentioned above, the approximate coefficient set typically contains information about the main signal trends, while the detail coefficient set contains information about high-frequency noise and other subtle signal variations. Edge feature identification is used to mark signal abrupt changes or edges, which is crucial for subsequent noise suppression and signal enhancement. This allows the system to comprehensively understand the signal's energy distribution characteristics, providing a foundation for further denoising. Estimating the optimal threshold for the band energy distribution characteristics based on the Bayesian risk criterion is a key next step. The Bayesian risk criterion provides a statistical method for selecting the optimal decision strategy given prior knowledge. Here, it is used to estimate the adaptive threshold matrix, which defines which wavelet coefficients should be considered noise and removed. To more accurately model the noise distribution, the system also uses interval correlation analysis to model the adaptive threshold matrix, resulting in a noise probability density function. This method not only considers the overall noise distribution characteristics but also captures local noise variation patterns, improving the accuracy of denoising. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, the noise probability density function can help the system better understand the noise characteristics, enabling more targeted denoising measures. Subsequently, the noise probability density function is reconstructed by filtering the coefficients using a preset soft threshold shrinkage function, resulting in a first-order denoised signal. The soft threshold shrinkage function is a commonly used denoising algorithm that reduces the impact of noise by adjusting the size of wavelet coefficients while preserving the key signal features. Furthermore, the system performs edge-preserving enhancement on the first-order denoised signal to produce an edge-enhanced signal set.The edge-enhanced signal set includes not only the dominant frequency eigenvector (i.e., the primary frequency component of the signal), but also noise suppression parameters and signal breakpoint identifiers. This information is crucial for subsequent signal correction and noise suppression. For example, in environmental monitoring applications, the edge-enhanced signal set can help researchers more clearly identify important signal features, effectively suppress noise, and improve data analysis accuracy. Next, a cross-channel consistency check is performed on the edge-enhanced signal set using an adaptive correction factor to produce a corrected signal set. The adaptive correction factor dynamically adjusts the correction parameters based on inter-channel correlations, ensuring data consistency and accuracy across all channels. For example, when simultaneously monitoring multiple environmental parameters, leveraging inter-channel correlations can help more accurately select the appropriate correction strategy, thereby improving overall data quality. Furthermore, residual noise is suppressed on the corrected signal set using singular value filtering, resulting in a second-order denoised signal. Singular value filtering is an effective denoising technique that decomposes the signal matrix and removes small singular values ​​to remove noise, further improving signal purity. Finally, the second-order denoised signal is spectrally reshaped using a time-varying filter to produce the final denoised data set. Time-varying filters can dynamically adjust filtering parameters based on the temporal variation of the signal, ensuring optimal denoising results over different time periods. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, time-varying filters can automatically adjust filtering strategies based on changing environmental conditions. This ensures that the resulting denoised data set retains the signal's key characteristics while effectively suppressing noise, improving data quality and reliability. The entire adaptive denoising process, encompassing multi-scale transform decomposition, band energy calculation, optimal threshold estimation, noise distribution modeling, coefficient screening and reconstruction, edge-preserving enhancement, cross-channel consistency verification, residual noise suppression, and spectrum shaping, forms a complete high-precision data processing chain, ensuring high-quality and consistent final data. Data processed in this manner not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In summary, adaptive denoising of raw data stream matrices can significantly improve data purity and reliability, laying a solid foundation for subsequent data analysis and applications.

[0036] In a specific embodiment, the signal phase reconstruction of the noise reduction data set by the phase unwrapping technology to obtain a continuous phase feature sequence includes: Performing a complex domain conversion on the noise reduction data set by Hilbert transform to obtain an analytical signal sequence; wherein the analytical signal sequence includes a real component, an imaginary component and an instantaneous amplitude feature; The phase angle of the analytical signal sequence is extracted based on the phase gradient descent method to obtain a wrapped phase sequence, and the phase jump point detection of the wrapped phase sequence is performed using a two-dimensional phase consistency constraint technology to obtain a phase jump mark set; wherein the phase jump mark set includes a jump position index, a jump amplitude value, and a jump direction identifier; Performing jump accumulation calculation on the phase jump marker set using a quality-guided path integration technique to obtain a phase correction amount, and performing 2π phase compensation on the wrapped phase sequence based on the phase correction amount using a phase unwrapping technique to obtain an initial unwrapped phase map; wherein the initial unwrapped phase map includes continuous phase values, phase residuals, and a quality factor matrix; The initial unwrapped phase image is calibrated with multi-sensor data through a multi-channel phase fusion method to obtain a continuous phase feature sequence; wherein the continuous phase feature sequence includes absolute phase values, phase continuity data and sensor relative phase delay characteristics.

[0037] Specifically, when using phase unwrapping to reconstruct the signal phase of a denoised data set and obtain a continuous phase feature sequence, the denoised data set must first be converted to the complex domain using the Hilbert transform to generate an analytic signal sequence. This process not only preserves the key information of the original signal but also provides additional instantaneous amplitude and phase information, which is crucial for subsequent phase reconstruction. Specifically, in an environmental monitoring application, assume that MEMS sensors are used to monitor long-term changes in temperature, humidity, and air pressure. After adaptive denoising, the output data from these sensors has been converted into a relatively clean denoised data set. Using the Hilbert transform, the system can convert this data into an analytic signal sequence containing real and imaginary components and instantaneous amplitude features. For example, the real component typically represents the main component of the signal, while the imaginary component provides information about the signal's phase variation. The instantaneous amplitude feature helps identify sudden changes or abnormal fluctuations in the signal. Next, the phase angle of the analytic signal sequence is extracted using the phase gradient descent method to obtain a wrapped phase sequence. Phase gradient descent is a commonly used phase extraction method that gradually approximates the true phase value by minimizing the phase error. Based on this, the system applies two-dimensional phase consistency constraints to detect phase jump points in the wrapped phase sequence, generating a set of phase jump markers. This constraint technique not only effectively identifies discontinuities (i.e., phase jump points) in the wrapped phase sequence, but also calculates the specific location, amplitude, and direction of each jump point. For example, in the environmental monitoring example mentioned above, if the sensor output signal is affected by external interference or inherent noise, phase jumps may occur, resulting in discontinuous phase information. Using two-dimensional phase consistency constraints, the system accurately detects these jump points and labels their locations and characteristics, providing a foundation for further phase correction. Subsequently, the phase jump marker set is cumulatively calculated using quality-guided path integration (QPI) to obtain the phase correction value. QPI dynamically adjusts the integration path based on the quality factor of each jump point to ensure accurate phase correction. The system then applies phase unwrapping technology to perform a 2π phase compensation on the wrapped phase sequence based on the phase correction value, generating an initial unwrapped phase map. The core of phase unwrapping lies in eliminating integer multiples of 2π phase jumps in the wrapped phase sequence, restoring continuous and accurate phase information. For example, when using MEMS sensors for long-term meteorological data collection in a weather station, the initial unwrapped phase map contains not only continuous phase values ​​but also provides information such as phase residuals and quality factor matrices, which are important for subsequent signal analysis and processing. Furthermore, to improve the accuracy and reliability of phase information, the system uses a multi-channel phase fusion method to jointly calibrate the initial unwrapped phase map with multi-sensor data, ultimately obtaining a continuous phase feature sequence.By combining data from multiple sensors, multi-channel phase fusion methods can more comprehensively capture the signal's phase characteristics and eliminate potential bias or errors inherent in individual sensors. For example, when simultaneously monitoring multiple environmental parameters, correlations between sensors may exist. Leveraging these correlations can help more accurately infer the true phase information for each channel. Data processed through multi-channel phase fusion forms a continuous sequence of phase signatures, encompassing not only absolute phase values ​​but also phase continuity data and sensor-by-sensor relative phase delay characteristics. This information is crucial for in-depth analysis of environmental parameter trends and their interrelationships. The entire process, from the Hilbert transform to the phase gradient descent method, to two-dimensional phase consistency constraints and quality-guided path integration techniques, is closely linked, forming a complete phase unwrapping pipeline. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, phase unwrapping can significantly improve data accuracy and consistency, enabling researchers to more clearly identify important phase features and effectively mitigate noise and phase jumps. Furthermore, multi-channel phase fusion methods can integrate data from multiple sensors, further enhancing the quality and reliability of phase information. Ultimately, data processed in this way not only supports scientific research but also provides strong support for practical applications, such as accurate weather forecasting and industrial process control. In short, performing phase unwrapping on a denoised data set significantly improves data purity and reliability, laying a solid foundation for subsequent data analysis and applications. The entire process, from generating the analytical signal sequence, extracting phase angles, detecting phase transition points, calculating transition accumulation, to multi-channel phase fusion, forms a complete high-precision phase reconstruction chain, ensuring high quality and consistency of the final data. Data processed in this way not only supports scientific research but also provides strong support for practical applications, such as accurate weather forecasting and industrial process control. In short, performing phase unwrapping on a denoised data set significantly improves data purity and reliability, laying a solid foundation for subsequent data analysis and applications.

[0038] In a specific embodiment, the step of performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector includes: Performing system dynamic modeling on the continuous phase characteristic sequence through state space equations to obtain a sensor state transfer matrix; Characterizing the noise characteristics of the sensor state transfer matrix based on measurement covariance analysis to obtain the noise covariance structure of the sensor system, and estimating the state quantity of the sensor system noise covariance structure through forward recursive prediction technology to obtain a priori state estimation value; Performing Kalman gain calculation on the prior state estimate based on the minimum mean square error criterion to obtain an adaptive weight coefficient, and performing state correction processing on the adaptive weight coefficient through a state update equation to obtain a posterior state estimate; Numerical stability enhancement is performed on the posterior state estimator by using a square root information filtering technique to obtain a robust state vector, and outlier detection is performed on the robust state vector based on a residual hypothesis test to obtain an outlier identification set; The abnormal point identification set is locally reconstructed based on a piecewise polynomial smoothing kernel to obtain a repaired phase sequence, and the repaired phase sequence is corrected for temperature drift through a temperature compensation mechanism to obtain a calibration signal vector; wherein, the calibration signal vector includes a temperature compensation coefficient, a zero bias correction value, and a sensitivity correction parameter.

[0039] Specifically, in the process of dynamically drift-compensating a continuous phase signature sequence to obtain a calibration signal vector, the system first needs to dynamically model the continuous phase signature sequence using state-space equations to generate a sensor state transition matrix. This process aims to establish a mathematical model to describe the sensor's dynamic behavior, thus providing a foundation for subsequent state estimation and drift compensation. Specifically, in an environmental monitoring application, assume that MEMS sensors are used to monitor long-term changes in temperature, humidity, and air pressure. After phase unwrapping, the output data from these sensors produces a relatively clean and continuous phase signature sequence. Using state-space equations, the system can convert these phase signature sequences into state variables and establish a corresponding state transition matrix, which describes how the sensor state changes over time. Next, the noise characteristics of the sensor state transition matrix are characterized using measurement covariance analysis (CCA), which is a statistical method used to assess the noise level and distribution characteristics in measurement data. Based on this, the system then applies forward recursive prediction techniques to estimate the state variables of the sensor system's CCA structure and generate a priori state estimates. For example, in the environmental monitoring example mentioned above, sensors may be affected by environmental changes or their own aging, causing drift in their output signals. Through measurement covariance analysis and forward recursive prediction techniques, the system can accurately estimate the sensor's state change trend and predict future state values, providing an important basis for subsequent drift compensation. Subsequently, the Kalman gain is calculated based on the minimum mean square error criterion on the prior state estimate to obtain adaptive weight coefficients. The Kalman gain is an optimization algorithm used to find the optimal trade-off between predicted and measured values ​​to minimize estimation error. The system then applies the state update equation to the adaptive weight coefficients to generate a posterior state estimate. For example, when using MEMS sensors for long-term meteorological data collection in a weather station, the Kalman gain calculation helps the system more precisely adjust the sensor's state estimate to ensure it aligns with actual conditions. The state update equation further corrects these estimates, making the resulting posterior state estimate more accurate and reliable. To enhance numerical stability, the system applies square root information filtering to the posterior state estimate to generate a robust state vector. Square root information filtering is an efficient filtering algorithm that improves the numerical stability of calculations while maintaining accuracy. Next, outlier detection is performed on the robust state vector based on residual hypothesis testing to obtain an outlier identification set. Residual hypothesis testing is a statistical method used to identify outliers or outliers in data. For example, in the environmental monitoring example above, if certain data points output by the sensor significantly deviate from the normal range, the system can mark them as outliers through residual hypothesis testing, providing guidance for further repair and correction.Next, a piecewise polynomial smoothing kernel is used to locally reconstruct the outlier point identification set, resulting in a repaired phase sequence. The piecewise polynomial smoothing kernel is an interpolation method that fills in outliers or missing data by fitting local data points, restoring a continuous and smooth phase sequence. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, if data from certain time periods is interfered with or corrupted, the system can repair it using the piecewise polynomial smoothing kernel to ensure data integrity and consistency. Finally, the repaired phase sequence is corrected for temperature drift using a temperature compensation mechanism to obtain a calibration signal vector. This temperature compensation mechanism dynamically adjusts the sensor output signal based on ambient temperature changes to eliminate the effects of temperature drift. The calibration signal vector includes not only temperature compensation coefficients but also bias correction values ​​and sensitivity correction parameters, which are critical for ensuring data accuracy. Throughout the entire process, from state-space equations to measurement covariance analysis, Kalman gain calculation, square root information filtering techniques, and residual hypothesis testing, each step is closely linked, forming a complete dynamic drift compensation process. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, dynamic drift compensation technology can significantly improve data accuracy and consistency, enabling researchers to more clearly identify important phase features and effectively suppress noise and drift. Furthermore, through temperature compensation, the system can eliminate the impact of temperature changes on sensor performance, further enhancing data quality and reliability. Ultimately, data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In short, dynamic drift compensation of continuous phase feature sequences can significantly improve data purity and reliability, laying a solid foundation for subsequent data analysis and applications. The entire process, from system dynamic modeling, noise characterization, state quantity estimation, Kalman gain calculation, state correction processing, numerical stability enhancement, outlier detection, to temperature drift correction, forms a complete high-precision calibration chain, ensuring high-quality and consistent final data. Data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In short, by dynamically compensating the continuous phase characteristic sequence, the purity and reliability of the data can be significantly improved, laying a solid foundation for subsequent data analysis and application.

[0040] In a specific embodiment, the step of extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve includes: Performing adaptive multi-scale decomposition on the calibration signal vector by empirical mode decomposition to obtain an intrinsic mode function set; wherein the intrinsic mode function set includes a multi-order component sequence, a residual trend term, and a modal energy density distribution; Based on the instantaneous frequency calculation framework, the non-stationary characteristics of the intrinsic mode function set are analyzed to obtain a time-frequency energy distribution spectrum, and the edge of the time-frequency energy distribution spectrum is enhanced by the frequency boundary sharpening technology to obtain a fine time-frequency structure; wherein the fine time-frequency structure includes the instantaneous frequency trajectory, the energy density matrix and the demarcation point feature set; Resonant modes of the fine time-frequency structure are identified through a subspace frequency tracking mechanism to obtain a modal parameter spectrum, and modal separation measurement is performed on the modal parameter spectrum based on a spectral manifold distance measure to obtain a modal separation eigenvector; wherein the modal separation eigenvector includes a modal center frequency, a modal damping ratio, and a modal coherence index; Performing group delay correction on the modal separation eigenvector based on a nonlinear phase compensation technique to obtain a phase-corrected spectrum, and performing amplitude modulation characteristic separation on the phase-corrected spectrum using a transient envelope extractor to obtain a modulation feature group; wherein the modulation feature group includes an envelope contour curve, a modulation depth parameter, and a transient response time constant; The modulation feature group is subjected to time domain remapping based on a multi-resolution reconstruction technique to obtain a sensor response characteristic curve; wherein the sensor response characteristic curve is a dynamic sensitivity function, an amplitude-frequency characteristic curve, and a phase-frequency characteristic curve.

[0041] Specifically, when extracting the transient characteristics of the calibration signal vector and obtaining the sensor response curve, the calibration signal vector must first be adaptively decomposed at multiple scales using empirical mode decomposition (EMD) to generate a set of eigenmode functions (IMFs). This process aims to decompose complex signals into multiple intrinsic mode functions (IMFs) with different time scales, enabling better analysis of their transient characteristics. Specifically, in an environmental monitoring application, assume that MEMS sensors are used to monitor long-term changes in temperature, humidity, and air pressure. After dynamic drift compensation, the output data from these sensors has been used to generate a relatively pure and stable calibration signal vector. Using EMD, the system can decompose these signals into multiple IMF component sequences and isolate the residual trend term and modal energy density distribution. For example, the IMF component sequence contains the different frequency components of the signal, while the residual trend term reflects the overall trend of the signal. The modal energy density distribution describes the energy distribution of each IMF component, helping to identify important features in the signal. Next, the non-stationary characteristics of the eigenmode function set are analyzed based on the instantaneous frequency calculation framework to obtain a time-frequency energy distribution map. The instantaneous frequency calculation framework is an effective method for analyzing nonstationary signals, revealing the frequency variations of a signal at different time points. Based on this, the system applies frequency boundary sharpening technology to enhance the edges of the time-frequency energy distribution spectrum, generating a refined time-frequency structure. Frequency boundary sharpening enhances the clarity of frequency boundaries, allowing the time-frequency energy distribution spectrum to more accurately reflect the actual characteristics of the signal. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, the time-frequency energy distribution spectrum includes not only the instantaneous frequency trajectory but also information such as the energy density matrix and the demarcation point feature set. This data is crucial for in-depth analysis of the changing trends and interrelationships of environmental parameters. Subsequently, the refined time-frequency structure is used to identify resonant modes using the subspace frequency tracking mechanism, resulting in a modal parameter spectrum. This subspace frequency tracking mechanism is an efficient modal identification algorithm that accurately identifies the frequencies and damping ratios of individual resonant modes in complex signals. The modal parameter spectrum is then subjected to a modal separation metric based on the spectral manifold distance measure to generate a modal separation feature vector. The spectral manifold distance measure is a statistical method used to evaluate the similarities and differences between modes. It can help the system distinguish different modal components more accurately. For example, in the environmental monitoring example mentioned above, if the signal output by the sensor contains multiple resonant modes, the system can accurately identify the center frequency, damping ratio, and modal coherence index of each mode through the subspace frequency tracking mechanism, providing a basis for further phase correction and modulation characteristic separation. To further improve phase accuracy, the system will apply nonlinear phase compensation technology to correct the group delay of the modal separation eigenvector to generate a phase-corrected spectrum.Nonlinear phase compensation technology adjusts the signal's phase characteristics to eliminate phase distortion caused by nonlinear effects. Next, a transient envelope extractor (TEE) separates the amplitude modulation characteristics of the phase-corrected spectrum to generate a modulation signature set. The TEE is a tool for separating the amplitude modulation characteristics of a signal. It can extract information such as the envelope contour, modulation depth parameters, and transient response time constant from the phase-corrected spectrum. For example, when using MEMS sensors for long-term meteorological data collection in a weather station, the modulation signature set includes not only the envelope contour but also information such as the modulation depth parameters and transient response time constant. These data are crucial for in-depth analysis of the signal's dynamic characteristics. Finally, the modulation signature set is remapped in the time domain using multi-resolution reconstruction technology to generate a sensor response characteristic curve. Multi-resolution reconstruction combines signal features at different resolutions to generate a comprehensive sensor response characteristic curve. This curve includes not only the dynamic sensitivity function but also information such as the amplitude-frequency and phase-frequency characteristics. For example, in the environmental monitoring example mentioned above, the sensor response characteristic curve can help researchers gain a more comprehensive understanding of the sensor's dynamic behavior and its response to environmental changes. The dynamic sensitivity function describes the sensor's sensitivity to changes in the input signal, the amplitude-frequency characteristic curve shows the sensor's response amplitude at different frequencies, and the phase-frequency characteristic curve reveals the sensor's phase response characteristics. The entire process, from empirical mode decomposition, instantaneous frequency calculation framework, frequency boundary sharpening technology, subspace frequency tracking mechanism, spectral manifold distance measure, nonlinear phase compensation technology, transient envelope extractor, to multi-resolution reconstruction technology, is closely linked to form a complete transient characteristic extraction process. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, transient characteristic extraction technology can significantly improve data accuracy and consistency, allowing researchers to more clearly identify important signal features and effectively suppress noise and phase distortion. Furthermore, through multi-resolution reconstruction technology, the system can integrate data from different resolutions, further improving the quality and reliability of sensor response characteristics. Ultimately, data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In summary, extracting transient characteristics from the calibration signal vector significantly improves data purity and reliability, laying a solid foundation for subsequent data analysis and applications. The entire process, encompassing adaptive multiscale decomposition, nonstationary characteristic analysis, resonant mode identification, modal separation metrics, group delay correction, amplitude modulation characteristic separation, and time-domain remapping, forms a complete, high-precision transient characteristic extraction chain, ensuring high-quality and consistent final data. Data processed in this manner not only supports scientific research but also provides strong support for practical applications, such as accurate weather forecasting and industrial process control.In short, by extracting the transient characteristics of the calibration signal vector, the purity and reliability of the data can be significantly improved, laying a solid foundation for subsequent data analysis and application.

[0042] In a specific embodiment, the high-precision data resampling of the sensor response characteristic curve based on the adaptive quantization coding technology to obtain a high-precision data acquisition result includes: Performing information entropy analysis on the sensor response characteristic curve through a non-uniform signal sampling mechanism to obtain a key point density distribution map; wherein the key point density distribution map includes an information gradient vector, a mutation point clustering structure, and a transient region boundary marker; Dynamically adjusting the bit depth of the key point density distribution map based on a segmented quantization precision allocation strategy to obtain an optimized bit depth allocation scheme, and reducing redundancy of the optimized bit depth allocation scheme through an adaptive quantization coding technology to obtain a compression mapping table; Performing high-precision interpolation reconstruction on the compressed mapping table using a fractional-order interpolation kernel function to obtain a fine sampling point sequence, and performing quantization error dispersion processing on the fine sampling point sequence based on orthogonal transform domain quantization to obtain a quantization coefficient matrix; wherein the quantization coefficient matrix includes basis vector combination weights, a quantization step size parameter, and a perceptual masking threshold; Performing quantization noise shaping on the quantization coefficient matrix based on nonlinear quantization boundary optimization to obtain a noise-shaped data stream, and performing signal recovery processing on the noise-shaped data stream through an adaptive inverse quantization technique to obtain a reconstructed data sequence; The reconstructed data sequence is subjected to residual correction processing through an error feedback compensation network to obtain a high-precision data acquisition result.

[0043] Specifically, when resampling sensor response curves for high-precision data acquisition using adaptive quantization coding technology to obtain high-precision data, the sensor response curves must first be subjected to information entropy analysis using a non-uniform signal sampling mechanism to generate a keypoint density distribution map. This process aims to identify important characteristic points in the signal and assign different sampling densities based on their information content to improve data compression efficiency and accuracy. Specifically, in an environmental monitoring application, assume that MEMS sensors are used to monitor temperature, humidity, and air pressure over a long period of time. Transient characteristic extraction of these sensor output data already yields relatively detailed sensor response curves. Using the non-uniform signal sampling mechanism, the system can perform information entropy analysis on these curves, identify keypoints containing the most information, and generate a keypoint density distribution map. For example, the keypoint density distribution map includes not only the information gradient vector (which describes the signal's information density variation) but also information such as the clustering structure of sudden changes and the identification of transient region boundaries. This data is crucial for subsequent dynamic bit depth adjustment and redundancy reduction. Next, the keypoint density distribution map is dynamically adjusted based on a pre-defined segmented quantization precision allocation strategy to obtain an optimized bit depth allocation solution. The piecewise quantization precision allocation strategy dynamically adjusts quantization precision based on the importance and complexity of different signal components. Based on this, the system applies adaptive quantization coding technology to reduce redundancy in the optimized bit depth allocation scheme, generating a compression mapping table. Adaptive quantization coding significantly improves data compression and transmission efficiency by removing unnecessary redundant information. For example, when using MEMS sensors in weather stations for long-term meteorological data collection, the optimized bit depth allocation scheme can minimize data volume while preserving important information, making data transmission and storage more efficient. Subsequently, the compression mapping table is reconstructed using a fractional-order interpolation kernel function for high-precision interpolation, resulting in a sequence of fine sampling points. The fractional-order interpolation kernel function is an efficient interpolation method that can restore a high-resolution version of the original signal while preserving signal details. Quantization error dispersion is then performed on the fine sampling point sequence using orthogonal transform domain quantization to generate a quantization coefficient matrix. Orthogonal transform domain quantization effectively disperses quantization error and improves signal quality by converting the signal to a transform domain (such as the Fourier transform or wavelet transform) and performing quantization based on this domain. For example, in the environmental monitoring example above, the quantization coefficient matrix includes not only the basis vector combination weights (which describe the signal representation in the transform domain), but also information such as the quantization step size parameter and the perceptual masking threshold. This data is important for further noise shaping and signal recovery. To further improve signal quality, the system applies nonlinear quantization boundary optimization to the quantization coefficient matrix to shape the quantization noise, generating a noise-shaped data stream.Nonlinear quantization boundary optimization adjusts the shape and position of the quantization boundary, distributing the quantization noise within a frequency range insensitive to the human ear, thereby improving perceptual quality. Next, adaptive inverse quantization technology is used to perform signal recovery on the noise-shaped data stream, generating a reconstructed data sequence. Adaptive inverse quantization restores the original signal through an inverse operation, ensuring data accuracy and integrity. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, adaptive inverse quantization can effectively recover high-quality signals, enabling researchers to more clearly identify important environmental trends. Finally, an error feedback compensation network is used to correct residual errors in the reconstructed data sequence, resulting in high-precision data acquisition results. The error feedback compensation network is an effective method for correcting errors generated during the reconstruction process. It compares the differences between the reconstructed data and the original data and gradually adjusts the model parameters to ultimately minimize the error. For example, in the environmental monitoring example mentioned above, the error feedback compensation network can significantly improve the accuracy of the reconstructed data, resulting in high-precision data acquisition results that include not only the dynamic sensitivity function but also information such as the amplitude-frequency characteristic curve and the phase-frequency characteristic curve. The entire process, from the non-uniform signal sampling mechanism, information entropy analysis, piecewise quantization precision allocation strategy, adaptive quantization coding technology, fractional-order interpolation kernel function, orthogonal transform domain quantization, nonlinear quantization boundary optimization, adaptive inverse quantization technology, and error feedback compensation network, is closely linked to form a complete high-precision data resampling process. For example, when using MEMS sensors for long-term meteorological data collection in weather stations, a high-precision data resampling method based on adaptive quantization coding technology can significantly improve data accuracy and consistency, enabling researchers to more clearly identify important environmental trends and effectively suppress noise and quantization errors. Furthermore, through the error feedback compensation network, the system can further enhance the quality and reliability of reconstructed data. Ultimately, data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control. In summary, high-precision data resampling of sensor response curves can significantly improve data purity and reliability, laying a solid foundation for subsequent data analysis and applications. The entire process, from information entropy analysis, dynamic bit depth adjustment, redundancy reduction, high-precision interpolation and reconstruction, quantization error dispersion, quantization noise shaping, signal recovery, and residual correction, forms a complete high-precision data resampling chain, ensuring the high quality and consistency of the final data. Data processed in this way not only supports scientific research but also provides strong support for practical applications such as accurate weather forecasting and industrial process control.In short, by performing high-precision data resampling of the sensor response characteristic curve, the purity and reliability of the data can be significantly improved, laying a solid foundation for subsequent data analysis and application.

[0044] The above describes the high-precision data acquisition method of the MEMS sensor in the embodiment of the present invention. The following describes the high-precision data acquisition system of the MEMS sensor in the embodiment of the present invention. Figure 2 An embodiment of a high-precision data acquisition system for a MEMS sensor according to an embodiment of the present invention includes: The first sampling module 21 is used to perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain an original data stream matrix; A denoising module 22 is configured to perform adaptive denoising on the original data stream matrix to obtain a denoised data set; A reconstruction module 23 is configured to reconstruct the signal phase of the noise reduction data set by using a phase unwrapping technique to obtain a continuous phase feature sequence; A compensation module 24 is configured to perform dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; An extraction module 25 is used to extract transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve; The second sampling module 26 is configured to perform high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technology to obtain a high-precision data acquisition result.

[0045] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the above method embodiment, which will not be repeated here.

[0046] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided, wherein the internal structure of the computer device can be as follows: Figure 3 As shown. The computer device includes a processor, memory, display screen, input device, network interface and database connected via a system bus. The processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0047] Those skilled in the art will understand that Figure 3The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0048] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the above-described method when executed by a processor. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0049] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware using a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media provided herein and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM.

[0050] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, apparatus, article, or method comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, apparatus, article, or method. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, apparatus, article, or method comprising the element.

[0051] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A high-precision data acquisition method for a MEMS sensor, characterized in that: The following steps are involved: Perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain the original data stream matrix; Performing adaptive denoising processing on the original data stream matrix to obtain a denoised data set; Reconstructing the signal phase of the noise reduction data set by phase unwrapping technology to obtain a continuous phase feature sequence; Performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; Extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve; Based on the adaptive quantization coding technology, high-precision data resampling is performed on the sensor response characteristic curve to obtain a high-precision data acquisition result.

2. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The initial electrical signal is a weak electrical signal. The initial electrical signal output by the MEMS sensor is sampled in parallel through multiple channels to obtain an original data stream matrix, including: The weak electrical signal of the MEMS sensor is differentially amplified by a high-impedance preamplifier to obtain a gain-matched signal group; Performing multi-channel clock synchronization on the gain matching signal group based on a synchronous trigger circuit to obtain a phase-locked sampling point sequence; The phase-locked sampling point sequence is dynamically adjusted through a programmable gain unit to obtain an original data stream matrix; wherein the original data stream matrix includes a sensor multi-axis output signal, a reference voltage signal, and a temperature compensation signal.

3. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The adaptive denoising process is performed on the original data stream matrix to obtain a denoised data set, comprising: Performing multi-scale transformation decomposition on the original data stream matrix using orthogonal wavelet basis functions to obtain a wavelet coefficient sequence, and performing frequency band energy calculation on the wavelet coefficient sequence to obtain a frequency band energy distribution feature; wherein the frequency band energy distribution feature includes an approximate coefficient group, a detail coefficient set, and an edge feature identifier; An optimal threshold is estimated for the frequency band energy distribution characteristics based on the Bayesian risk criterion to obtain an adaptive threshold matrix, and a noise distribution model is performed on the adaptive threshold matrix using an interval correlation analysis method to obtain a noise probability density function; The noise probability density function is subjected to coefficient screening and reconstruction using a preset soft threshold shrinkage function to obtain a first-order denoised signal, and edge-preserving enhancement processing is performed on the first-order denoised signal to obtain an edge-enhanced signal set; wherein the edge-enhanced signal set includes a main frequency eigenvector, a noise suppression rate parameter, and a signal mutation point identifier; Performing a cross-channel consistency check on the edge enhancement signal set based on an adaptive correction factor to obtain a correction signal group, and performing residual noise suppression on the correction signal group through singular value filtering to obtain a second-order denoised signal; The second-order denoised signal is subjected to spectrum shaping processing by a time-varying filter to obtain a denoised data set.

4. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The signal phase reconstruction of the noise reduction data set is performed by the phase unwrapping technology to obtain a continuous phase feature sequence, including: Performing a complex domain conversion on the noise reduction data set by Hilbert transform to obtain an analytical signal sequence; wherein the analytical signal sequence includes a real component, an imaginary component and an instantaneous amplitude feature; The phase angle of the analytical signal sequence is extracted based on the phase gradient descent method to obtain a wrapped phase sequence, and the phase jump point detection of the wrapped phase sequence is performed using a two-dimensional phase consistency constraint technology to obtain a phase jump mark set; wherein the phase jump mark set includes a jump position index, a jump amplitude value, and a jump direction identifier; Performing jump accumulation calculation on the phase jump marker set using a quality-guided path integration technique to obtain a phase correction amount, and performing 2π phase compensation on the wrapped phase sequence based on the phase correction amount using a phase unwrapping technique to obtain an initial unwrapped phase map; wherein the initial unwrapped phase map includes continuous phase values, phase residuals, and a quality factor matrix; The initial unwrapped phase image is calibrated with multi-sensor data through a multi-channel phase fusion method to obtain a continuous phase feature sequence; wherein the continuous phase feature sequence includes absolute phase values, phase continuity data and sensor relative phase delay characteristics.

5. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The performing dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector includes: Performing system dynamic modeling on the continuous phase characteristic sequence through state space equations to obtain a sensor state transfer matrix; Characterizing the noise characteristics of the sensor state transfer matrix based on measurement covariance analysis to obtain the noise covariance structure of the sensor system, and estimating the state quantity of the sensor system noise covariance structure through forward recursive prediction technology to obtain a priori state estimation value; Performing Kalman gain calculation on the prior state estimate based on the minimum mean square error criterion to obtain an adaptive weight coefficient, and performing state correction processing on the adaptive weight coefficient through a state update equation to obtain a posterior state estimate; Numerical stability enhancement is performed on the posterior state estimator by using a square root information filtering technique to obtain a robust state vector, and outlier detection is performed on the robust state vector based on a residual hypothesis test to obtain an outlier identification set; The abnormal point identification set is locally reconstructed based on a piecewise polynomial smoothing kernel to obtain a repaired phase sequence, and the repaired phase sequence is corrected for temperature drift through a temperature compensation mechanism to obtain a calibration signal vector; wherein, the calibration signal vector includes a temperature compensation coefficient, a zero bias correction value, and a sensitivity correction parameter.

6. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The step of extracting transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve includes: Performing adaptive multi-scale decomposition on the calibration signal vector by empirical mode decomposition to obtain an intrinsic mode function set; wherein the intrinsic mode function set includes a multi-order component sequence, a residual trend term, and a modal energy density distribution; Based on the instantaneous frequency calculation framework, the non-stationary characteristics of the intrinsic mode function set are analyzed to obtain a time-frequency energy distribution spectrum, and the edge of the time-frequency energy distribution spectrum is enhanced by the frequency boundary sharpening technology to obtain a fine time-frequency structure; wherein the fine time-frequency structure includes the instantaneous frequency trajectory, the energy density matrix and the demarcation point feature set; Resonant modes of the fine time-frequency structure are identified through a subspace frequency tracking mechanism to obtain a modal parameter spectrum, and modal separation measurement is performed on the modal parameter spectrum based on a spectral manifold distance measure to obtain a modal separation eigenvector; wherein the modal separation eigenvector includes a modal center frequency, a modal damping ratio, and a modal coherence index; Performing group delay correction on the modal separation eigenvector based on a nonlinear phase compensation technique to obtain a phase-corrected spectrum, and performing amplitude modulation characteristic separation on the phase-corrected spectrum using a transient envelope extractor to obtain a modulation feature group; wherein the modulation feature group includes an envelope contour curve, a modulation depth parameter, and a transient response time constant; The modulation feature group is subjected to time domain remapping based on a multi-resolution reconstruction technique to obtain a sensor response characteristic curve; wherein the sensor response characteristic curve is a dynamic sensitivity function, an amplitude-frequency characteristic curve, and a phase-frequency characteristic curve.

7. The high-precision data acquisition method for MEMS sensors according to claim 1, characterized in that: The method of performing high-precision data resampling on the sensor response characteristic curve based on the adaptive quantization coding technology to obtain a high-precision data acquisition result includes: Performing information entropy analysis on the sensor response characteristic curve through a non-uniform signal sampling mechanism to obtain a key point density distribution map; wherein the key point density distribution map includes an information gradient vector, a mutation point clustering structure, and a transient region boundary marker; Dynamically adjusting the bit depth of the key point density distribution map based on a segmented quantization precision allocation strategy to obtain an optimized bit depth allocation scheme, and reducing redundancy of the optimized bit depth allocation scheme through an adaptive quantization coding technology to obtain a compression mapping table; Performing high-precision interpolation reconstruction on the compressed mapping table using a fractional-order interpolation kernel function to obtain a fine sampling point sequence, and performing quantization error dispersion processing on the fine sampling point sequence based on orthogonal transform domain quantization to obtain a quantization coefficient matrix; wherein the quantization coefficient matrix includes basis vector combination weights, a quantization step size parameter, and a perceptual masking threshold; Performing quantization noise shaping on the quantization coefficient matrix based on nonlinear quantization boundary optimization to obtain a noise-shaped data stream, and performing signal recovery processing on the noise-shaped data stream through an adaptive inverse quantization technique to obtain a reconstructed data sequence; The reconstructed data sequence is subjected to residual correction processing through an error feedback compensation network to obtain a high-precision data acquisition result.

8. A high-precision data acquisition system for MEMS sensors, characterized in that: include: The first sampling module is used to perform multi-channel parallel sampling on the initial electrical signal output by the MEMS sensor to obtain an original data stream matrix; A denoising module, configured to perform adaptive denoising on the original data stream matrix to obtain a denoised data set; A reconstruction module, configured to reconstruct the signal phase of the noise reduction data set by using a phase unwrapping technique to obtain a continuous phase feature sequence; A compensation module, configured to perform dynamic drift compensation on the continuous phase characteristic sequence to obtain a calibration signal vector; an extraction module, configured to extract transient characteristics of the calibration signal vector to obtain a sensor response characteristic curve; The second sampling module is used to perform high-precision data resampling on the sensor response characteristic curve based on an adaptive quantization coding technology to obtain a high-precision data acquisition result.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images