Piezoresistive multichannel pressure sensor array decoupling system and method

Abstract

The invention discloses a piezoresistive multichannel pressure sensor array decoupling system and method, and relates to the technical field of sensor circuits and signal processing. Firstly, a reference impedance matrix is established based on calibration data, and hardware-level preliminary decoupling is performed through a hybrid compensation network; then, impedance signals are collected in real time, a compensation matrix is dynamically updated through a self-regularization sparse algorithm, adjustable elements are driven, and complete decoupling control from static compensation to dynamic self-adaptive adjustment is achieved; finally, the equivalent impedance matrix approaches to a diagonal matrix, and independent, real-time and stable output of multichannel signals is realized. Besides, the system can automatically trigger re-calibration by monitoring the condition number of the equivalent impedance matrix when the condition number exceeds a threshold value, can automatically detect coupling increase or parameter drift during long-term operation of the system, and can automatically recover matrix precision through a calibration process to ensure long-term stability and reliability of the system.

Description

Technical Field

[0001] This application relates to the field of sensor circuits and signal processing technology, and in particular to a decoupling system and method for a piezoresistive multichannel pressure sensor array. Background Technology

[0002] In multi-channel piezoresistive sensor arrays, each sensing element changes its resistance and outputs a corresponding voltage signal when subjected to external force. However, due to factors such as wire capacitance, resistive coupling, substrate leakage current, and differences in input impedance between array nodes, significant electrical coupling effects occur between different channels, causing adjacent nodes to respond when a single point is excited. This cross-talk phenomenon is particularly pronounced in high-density flexible arrays, causing the output signal of the sensor matrix to no longer have a one-to-one correspondence with the force applied to the element, resulting in measurement errors, image distortion, and positioning ambiguity.

[0003] In practical applications, resistive coupling can cause severe signal crosstalk, with channel signals interfering with each other and failing to accurately reflect the actual pressure values ​​of each node. This leads to a decrease in system sensitivity, and the output voltage becomes nonlinear due to the influence of neighboring units, resulting in measurement errors.

[0004] Traditional decoupling methods often employ pure hardware compensation or software algorithms. However, pure hardware compensation structures are complex and have limited ability to respond to dynamic coupling, while software algorithm decoupling, although flexible, is susceptible to noise and matrix ill-conditioning. Summary of the Invention

[0005] Therefore, it is necessary to provide a piezoresistive multichannel pressure sensor array decoupling system and method to address the above-mentioned technical problems.

[0006] The following technical solution is adopted in this specification: This specification provides a decoupling system and method for a piezoresistive multichannel pressure sensor array, including: Based on a piezoresistive multichannel pressure sensor array, a multichannel coupling impedance matrix is ​​established; wherein, the off-diagonal elements of the multichannel coupling impedance matrix represent the interconnection coupling between channels, and the diagonal elements represent the piezoresistive unit impedance of the current channel; The voltage signals of each piezoresistive unit of the multi-channel coupled impedance matrix are acquired in real time and digitized to obtain a time-varying impedance matrix; and the voltage signals of the multi-channel coupled impedance matrix in the zero-input state are acquired in calibration mode to form a reference impedance matrix. Based on the reference impedance matrix, a preliminary physical decoupling of the coupling impedance between channels is performed at the hardware level through a hybrid compensation network to obtain the static compensation matrix. Based on the time-varying impedance matrix, a self-regularized sparse algorithm is used to dynamically update the value of the static compensation matrix with the goal of minimizing the norm of the off-diagonal elements of the static compensation matrix, and convert it into an analog voltage signal; and based on the analog voltage signal, the adjustable element in the hybrid compensation network is driven to perform impedance adjustment. The output is a decoupled voltage signal that corresponds one-to-one with the pressure of each piezoresistive unit.

[0007] Furthermore, the construction of the reference impedance matrix includes: Under reference pressure conditions, a known excitation is injected by periodically scanning the voltage signals at each node of the multi-channel coupling impedance matrix. and measure Construct the reference impedance matrix : ; in, Indicates a known stimulus; This represents the voltage matrix actually acquired.

[0008] Furthermore, the preliminary physical decoupling of the coupling impedance between channels through a hybrid compensation network, resulting in a static compensation matrix, includes: The off-diagonal terms of the reference impedance matrix are calculated, and the initial physical static compensation matrix is ​​set according to the ideal cancellation requirements. Map the initial physical static compensation matrix to realizable circuit elements; The mapped physical compensation branches are connected to the array according to the topology. The inter-channel compensation branch is obtained by connecting compensation elements in parallel between nodes i and j; the ground compensation branch is obtained by connecting a capacitor in parallel with the reference ground at node i or by connecting an active current source. After the physical connection is completed, a static compensation matrix is ​​obtained; The initial physical static compensation matrix is ​​expressed as: ; The static compensation matrix is ​​expressed as: ; in, Represents the off-diagonal terms of the reference impedance matrix; Represents the initial physical static compensation matrix; Represents the reference impedance matrix; Represents the static compensation matrix; The off-diagonal residual energy of the static compensation matrix is ​​expressed as: ; like If the current static compensation is within an acceptable threshold, it is retained; otherwise, the constraint optimization is performed by solving a minimization problem with physical implementation constraints. The minimization problem with physical implementation constraints is expressed as: ; in, express ; The off-diagonal part of the matrix is ​​represented; the set S represents the space of realizable elements; This represents the overall equivalent compensation impedance matrix of the module.

[0009] Furthermore, the dynamic updating of the static compensation matrix includes: According to the preset sampling period, the rows and columns of the multi-channel coupling impedance matrix are scanned to collect real-time voltage and current. For each pair of elements (i,j) in the matrix, cross-correlation / autocorrelation is calculated using a sliding time window of length W to estimate the impedance element. The calculation formula is as follows: ; The symbol * denotes complex conjugation; This represents the element in the i-th row and j-th column of the equivalent impedance matrix estimated at time t; ) represents the complex conjugate of the current signal; () represents the current signal applied to the j-th excitation channel at the k-th sampling time; (k) represents the voltage signal acquired at the i-th measurement channel at the k-th sampling time; W represents the length of the sliding time window used for impedance estimation; Construct the current estimation matrix And calculate the equivalent matrix residual based on the current estimated matrix and the static compensation matrix; Based on the equivalent matrix residue, calculate the residual vector; The equivalent matrix residual is expressed as: ; The residual vector is calculated using the following formula: ; Based on the self-regularized sparse algorithm, the static compensation matrix is ​​updated element by element; The core optimization objective function of the self-regularized sparse algorithm is expressed as: ; For elements Perform gradient descent and discretize to obtain the update formula: ; The gradient term can be expanded from the sampled data as follows: ; like ( If the threshold is set, the element update is frozen to reduce meaningless noise disturbances; for items that are zero for a long time, the corresponding compensation branch can be turned off in hardware to save power.

[0010] Furthermore, the hybrid compensation network adopts a hybrid topology, including: Inter-channel interconnection compensation branch: set between adjacent channel nodes to cancel lateral coupling, including resistors, capacitors and inductors; Ground compensation branch: Located between each channel and the reference ground, it is used to cancel parasitic coupling and common-mode interference of the ground wire, including capacitive elements and controllable current sources.

[0011] This specification provides a piezoresistive multichannel pressure sensor array decoupling system, including: Piezoresistive sensor array module: used to establish a multi-channel coupling impedance matrix based on a piezoresistive multi-channel pressure sensor array; wherein, the off-diagonal elements of the multi-channel coupling impedance matrix represent the interconnection coupling between channels, and the diagonal elements represent the piezoresistive unit impedance of the current channel; Signal acquisition module: used to acquire and digitize the voltage signals of each piezoresistive unit of the multi-channel coupled impedance matrix in real time to obtain a time-varying impedance matrix; and to acquire the voltage signals of the multi-channel coupled impedance matrix in the zero-input state in calibration mode to form a reference impedance matrix. Compensation network module: Used to perform preliminary physical decoupling of the coupling impedance between channels at the hardware level based on the reference impedance matrix and through a hybrid compensation network, so as to obtain a static compensation matrix; Matrix calculation and control module: Based on the time-varying impedance matrix, it dynamically updates the value of the static compensation matrix using a self-regulating sparse algorithm, aiming to minimize the norm of the off-diagonal elements of the static compensation matrix, and converts it into an analog voltage signal; and based on the analog voltage signal, it drives the adjustable elements in the hybrid compensation network to perform impedance adjustment. Output module: Used to output decoupled voltage signals that correspond one-to-one with the pressure of each piezoresistive unit.

[0012] The above-mentioned technical solutions adopted in this specification can achieve the following beneficial effects: This specification provides a decoupling system and method for a piezoresistive multi-channel pressure sensor array. By modeling the multi-channel sensing system as a linear impedance matrix network and introducing a hybrid topology compensation network at the circuit level, adaptive dynamic adjustment is performed at the algorithm level, achieving real-time independent response and high-precision decoupling output for each channel in the array. Through this decoupling method, the multi-channel sensing system can achieve self-diagnosis and self-recovery during long-term operation, ensuring stable diagonalization and compensation accuracy of the impedance matrix.

[0013] Furthermore, the dynamic decoupling system for sensor arrays described in this invention can achieve real-time suppression of signal interference caused by electrical mutual coupling in multi-node piezoresistive sensor arrays, and is more suitable for flexible pressure sensor arrays, electronic skin, tactile detection panels and other multi-point array sensing and measurement applications. Attached Figure Description

[0014] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0015] Figure 1 This document provides a schematic flowchart of a decoupling method for a piezoresistive multichannel pressure sensor array. Figure 2 This is a schematic diagram of the piezoresistive sensor array structure provided in this specification; Figure 3 This is a schematic diagram of the hybrid compensation network topology circuit provided in this specification; Figure 4 This specification provides a schematic diagram of the overall structure of a piezoresistive multichannel pressure sensor array decoupling system. Figure 5 This is a schematic diagram illustrating the principle of matrix diagonalization provided in this manual; Figure 6 This is a schematic diagram of the SRS-A algorithm provided in this manual; Figure 7 This is a schematic diagram of the dynamic compensation closed-loop control provided in this manual; Figure 8 This is a schematic diagram of the matrix condition number monitoring and recalibration logic provided in this manual. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this specification, all other implementations obtained by those skilled in the art without creative effort are based on this, and there is an urgent need for a novel decoupling method that works collaboratively at the hardware and algorithm levels, capable of achieving real-time independence and high-precision output of the sensing channel under low-frequency dynamic conditions.

[0017] To address the aforementioned issues, this invention proposes a joint decoupling scheme combining physical compensation and dynamic diagonalization. This scheme uses the impedance matrix as the system model, achieving partial physical decoupling by establishing a hybrid compensation network at the circuit level. Then, it utilizes a self-regulating sparse algorithm (SRS-A) for dynamic numerical adjustment, ultimately resulting in an equivalent impedance matrix. By approximating a diagonal matrix, independent, real-time, and stable output of multi-channel signals is achieved. Furthermore, this invention proposes a matrix condition number monitoring and automatic recalibration mechanism, which can automatically detect increased coupling or parameter drift during long-term system operation and self-recover matrix accuracy through the calibration process, ensuring long-term system stability and reliability.

[0018] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0019] Figure 1 This is a schematic diagram of the overall structure of a piezoresistive multi-channel pressure sensor array decoupling system as described in this specification, specifically including: The system comprises a piezoresistive sensor array module, a signal acquisition module, a compensation network module, a matrix calculation and control module, and an output module. Its core idea is to model the multi-channel sensing system as a linear impedance matrix network. By introducing a hybrid topology compensation network at the circuit level and performing adaptive dynamic adjustments at the algorithm level, it achieves real-time independent response and high-precision decoupled output for each channel in the array.

[0020] Piezoresistive sensor array module: Composed of an N×M matrix structure of piezoresistive units, the resistance of each unit changes with external force; the array units are interconnected via conductive lines, forming a coupling impedance matrix Z. The diagonal elements of the coupling impedance matrix Z... The piezoresistive element impedance corresponding to the channel of interest in the current measurement, drive, or decoupling calculation; off-diagonal elements. This indicates parasitic or interconnect coupling between channels. The signal characteristics of this module are low-frequency, quasi-static variations, with a typical operating frequency range of 0.1Hz to 1kHz.

[0021] Signal acquisition and calibration module: Employing a matrix scanning method, this module consists of a multiplexer array, differential amplifier, reference voltage source, and analog-to-digital converter (ADC). It periodically scans the voltage signals at each node of the coupling impedance matrix Z, amplifies and filters out high-frequency noise, and digitizes the sampled signals before transmitting them to the control module. This completes the calibration of the reference impedance matrix under zero-input conditions in calibration mode. The sampling frequency is typically set to 5 to 10 times the operating frequency to ensure the time-domain resolution of the decoupled calculation.

[0022] Hybrid compensation network module: Located between the sensor array and the acquisition circuit, this module provides physical layer compensation for inter-channel coupling impedance. The compensation network employs a hybrid topology, including inter-channel interconnect compensation branches and ground compensation branches. Inter-channel interconnect compensation branches are located between adjacent channel nodes to cancel lateral coupling and include resistors, capacitors, inductors, or combinations thereof, with impedances denoted as... The design principle is The ground compensation branch is located between each channel and the reference ground to cancel parasitic coupling and common-mode interference. It includes a capacitor component and a controllable active current component. The active current component uses an operational amplifier to form a negative impedance converter (NIC), which generates a reverse current to cancel parasitic effects. The overall equivalent compensation impedance matrix of this module is Z. N Its parameters are automatically updated by the control module based on real-time calculation results.

[0023] The matrix calculation and control module, mainly composed of a microcontroller (MCU) or field-programmable logic device (FPGA), includes an arithmetic unit, a storage unit, a digital-to-analog converter (DAC), and a communication interface. It is the core calculation and adjustment unit of the sensor array decoupling system. This module's functions include impedance matrix estimation, error function calculation, adaptive algorithm execution, and the output and control of compensation signals.

[0024] This module estimates the impedance matrix of the current array based on the collected voltage and current data. : ; in, (t) represents the element in the i-th row and j-th column of the impedance matrix estimated at the current time t, which is used to characterize the equivalent impedance generated by the j-th excitation channel to the i-th measurement channel. When i=j, it corresponds to the equivalent impedance of this channel. When i≠j, it corresponds to the coupling impedance between channels. (t) represents the voltage response signal acquired at the i-th measurement channel or node at time t; (t) represents the current signal applied to the j-th excitation channel at time t; This represents the inner product operation, used to describe the correlation between two sets of signals in the time domain.

[0025] According to the impedance matrix The time-varying matrix is ​​obtained. And calculate the residual coupling error: ; The system employs an improved adaptive algorithm (SRS-A), using the residual coupling error as the optimization objective for the compensation matrix Z. N (t) is iteratively updated in real time. And the updated... It is converted into an analog voltage signal and output to the variable element in the compensation network via a DAC to achieve real-time impedance adjustment.

[0026] Output module: Includes a multi-channel buffer amplifier and a data interface. This module receives the decoupled voltage signals from each channel, performs isolation amplification and low-pass filtering, and then outputs them to an external data acquisition system. The output signals correspond one-to-one with the pressure changes of each sensing unit, enabling completely independent responses between channels.

[0027] This system aims at matrix diagonalization. At the hardware level, a hybrid compensation network is used to physically cancel out the main coupling terms. At the software level, an adaptive sparse regularization algorithm (SRS-A) is used to estimate and finely compensate for the remaining coupling in real time, thereby achieving real-time and stable decoupling of the multi-channel piezoresistive array.

[0028] The piezoresistive multi-channel pressure sensor array decoupling system operates in five steps: initialization, static compensation, real-time sampling and estimation, parameter updating and output, and condition detection and recalibration, as detailed below: 1) System initialization: used for calibration and decoupling from static physical properties. Label the symbol: Let N be the number of channels; Let be the sampled port voltage vector; For port drive current vector (or equivalent excitation); This is the original (time-varying) impedance matrix of the array; To compensate for the equivalent impedance matrix generated by the network; , which is the system's equivalent impedance; It is denoted as the off-diagonal part of the region matrix; It is denoted as the diagonal part of the region matrix.

[0029] Static physical decoupling: Under conditions of no external force or reference pressure (system no-load or calibration state), inject known excitation port by port in a row and column scan. and measure Construct the reference impedance matrix and save it. As an initial model: If a single column / row excitation is used, the j-th column is filled by energizing the j-th channel separately and measuring the responses of the other channels.

[0030] Calculate the off-diagonal terms of the reference impedance matrix Set the initial physical static compensation matrix according to the ideal cancellation requirements. for: .

[0031] 2) Static compensation Map the initial physical static compensation matrix to realizable circuit elements: use L / C for purely imaginary terms (inductive, capacitive); use active current sources / NICs (negative impedance transformers) for equivalent compensation for real parts (dissipative) or terms that exceed the realizable range of passive elements; if an extreme value is encountered that cannot be realized, then perform a constrained least squares approximation on that term.

[0032] 3) Real-time sampling and estimation Connect the mapped physical compensation branches to the array according to the topology; Inter-channel compensation branch: A compensation element is connected in parallel between nodes i and j (parallel admittance effect) ); Ground compensation branch: A capacitor or an active current source is connected in parallel with the reference ground at node i (to suppress common-mode and ground loop coupling).

[0033] After physical access is completed, the equivalent matrix under static compensation is obtained. : .

[0034] 4) Parameter update and output Calculate the off-diagonal residual energy after static compensation : like Within an acceptable threshold (determined by design accuracy), retain the current static compensation; otherwise, perform constraint optimization corrections by solving a minimization problem with physical realization constraints. Here, the set S represents the space of realizable components (e.g., allowing only pure virtual jX, or limiting the range of capacitor / inductor values, active implementation of amplitude limiting, etc.); the optimal approximation of this problem can be obtained using constrained quadratic programming or least squares solvers. And configure the hardware accordingly.

[0035] During system operation, sampling is performed according to the set sampling period. Perform the following operations: Sampling: V(t) and I(t) are obtained by row and column scanning or parallel ADC synchronous acquisition.

[0036] Sliding window statistical estimation (for noise reduction): For each pair of (i,j), calculate the cross-correlation / autocorrelation using a sliding time window of length W to estimate the impedance element. If only voltage is measured, the equivalent voltage can be calculated using a known excitation reference or differential measurement. The symbol * represents complex conjugation, and at low frequencies it represents real-valued operations.

[0037] Construct the current estimation matrix And calculate the equivalent matrix residual (including static compensation): Calculate the residual vector or scalar error: .

[0038] Applying the SRS-A algorithm to the equivalent impedance matrix The element-wise update; the objective function of the SRS-A algorithm is expressed as: For elements Gradient descent and discretization yield the update formula (per cycle): The gradient term can be expanded from the sampled data as (approximate form): , Alternatively, a more direct, error-driven LMS approximation update can be used: in, The residual of the i-th channel is expressed as: ;parameter , , Determined by system settings (suggested range:) ).

[0039] Sparse threshold control: If ( If the threshold is set, the element update is frozen to reduce meaningless noise disturbances; for items that are zero for a long time, the corresponding compensation branch can be turned off in hardware to save power.

[0040] 5) Condition detection and recalibration The mapping principle of the resulting impedance decoupling network is to map the complex compensation value. Mapping to a realizable element at the target frequency f0. If (Pure Emptiness): If Using series inductors ;like Use parallel / series capacitors ;like Parallel / series resistors or active circuits (NICs) can be used to achieve equivalent negative impedance or inject compensation current. For non-single frequency scenarios, narrowband or multi-segment compensation networks should be designed first, or active injection should be used to obtain phase control.

[0041] This invention provides multiple hardware access methods. The inter-channel compensation branch directly connects the compensation element between nodes i and j on the PCB (parallel connection changes admittance). If it is an active scheme, an adjustable current injection point is connected in parallel along this path, and the output of the op-amp / current source controlled by the DAC is connected to the two nodes to produce an equivalent current injection point. The ground compensation branch connects a capacitor in parallel between node i and ground or connects an active current source to adjust the diagonal terms and suppress common-mode components. The digital / analog interface control module outputs voltage to the drive circuit (voltage-controlled variable elements or drive operational amplifiers to generate current) via a DAC (resolution ≥ 12 bits). For rapid updates, it is recommended to pre-configure several combinations of switching elements (switching type) in hardware to achieve rapid switching over a wide range, and then use active fine-tuning for fine adjustment.

[0042] Calculate the equivalent matrix condition number for each update cycle: like (e.g., 50-100), or residuals If there is no downward trend or an upward trend within a certain number of periods, the system triggers a full matrix recalibration: online updates are paused, and the calibration process is re-executed to obtain a new matrix. And reset .

[0043] If a single channel anomaly is detected (such as a sensor open circuit or short circuit), the channel is locked, its row and column data are removed, and block diagonalization is continued on other channels to maintain system robustness.

[0044] Special explanations are needed for physical compensation selection strategies in specific scenarios: Coupling is primarily based on proximity (typical piezoresistive array): employs local interconnect compensation + ground-to-ground branches, satisfying only for i and j. The physical implementation of the terms, SRS-A is responsible for the software suppression of distant sub-terms.

[0045] Coupling is capacitively dominant (high frequency or long wire): use series / parallel inductors L for anti-phase compensation or parallel capacitors to adjust the diagonal term; for those with high phase requirements, active phase injection should be used first.

[0046] Coupling with real part (dissipation) or negative resistance effect: Equivalent negative resistance compensation is achieved by injecting reverse current using an active NIC or controllable current source; the software layer controls the amplitude and phase of these active terms to ensure stability.

[0047] Broadband / multi-frequency scenarios: Multi-segment passive compensation (multi-order π / T network) combined with active small-amplitude phase adjustment is adopted; SRS-A performs independent updates in each sub-band or uses weighted error integration.

[0048] The complete mathematical process for selecting the physical compensation strategy is as follows (per cycle): A. Read in .

[0049] B. Estimate : .

[0050] C. Calculate the residual matrix ,and .

[0051] D. For each Perform an SRS-A update: .

[0052] E. For the updated Perform a realizability projection (mapped to the allowed set of devices S): if an element is outside the device range, correct it to the nearest realizable value using constrained least squares.

[0053] F. The DAC outputs the corresponding control value, and after the hardware adjustment is completed, the next cycle begins.

[0054] G. Periodic calculation The decision on whether to recalibrate will be made according to the rules.

[0055] Furthermore, the technical solutions provided in this application will be described in detail below with reference to the embodiments.

[0056] like Figure 1 As shown, the overall system structure of the present invention includes: a piezoresistive sensor array module 1, a signal acquisition and calibration module 2, a hybrid compensation network module 3, a matrix calculation and control module 4, and an output and feedback module 5.

[0057] The piezoresistive sensor array module 1 detects external pressure and outputs a corresponding voltage signal. The signal acquisition and calibration module 2 includes a multiplexed analog-to-digital converter (ADC) and a preamplifier differential amplifier, used for acquiring and initially calibrating the array signal. The hybrid compensation network module 3 works in conjunction with the matrix calculation and control module 4; the former performs physical compensation and active injection, while the latter performs dynamic numerical compensation by executing the self-regulating sparse algorithm (SRS-A). The system output includes both analog voltage output and digital bus output, enabling data interaction with external control systems or host computers.

[0058] like Figure 2 As shown, this embodiment employs a 4×4 piezoresistive sensor array. The array consists of 16 piezoresistive units R 11 To R 44 The system is composed of rows and columns arranged in a matrix. Each unit node is connected between the corresponding row drive line and column sampling line. The rows are driven sequentially by a microcontroller unit (MCU) according to timing, and the columns are sampled by an ADC module.

[0059] The coupling problem that is common in piezoresistive arrays is that there are cross-coupling paths formed by parasitic capacitance and parasitic resistance between adjacent cells. These paths are particularly significant when the array area increases or the wire spacing decreases.

[0060] Therefore, in the subsequent compensation design, this invention eliminates signal aliasing and measurement errors caused by this coupling through matrix diagonalization and dynamic compensation algorithms.

[0061] like Figure 3 As shown, the hybrid compensation network module adopts a hybrid topology structure of inter-channel interconnection + ground-to-ground branch + active injection.

[0062] Each signal channel (channel 1, channel 2, channel 3) corresponds to a set of output nodes in the array, and adjacent channels are interconnected via interconnection branches R. 12 C 12 R 23 C 23 These are connected to cancel out the coupling impedance between channels. Each channel has a ground branch below it, including capacitors C1, C2, and C3, and controllable current sources I1, I2, and I3. The capacitor branches are used to compensate for low-frequency parasitic effects, and the active current sources are driven by the DAC output to achieve dynamically adjustable active current injection.

[0063] This hybrid structure possesses both the stability of passive compensation and the flexibility of active control, enabling real-time compensation for nonlinear changes in matrix elements.

[0064] like Figure 4 As shown, the system's workflow includes the following steps: Signal acquisition: The multi-channel voltage signals of the sensor array are synchronously sampled by the ADC acquisition module; Impedance matrix estimation: The control module calculates the current equivalent impedance matrix Z based on the sampled data. Residual calculation: Calculate the error residual based on the difference between the predicted output and the actual sampled signal; SRS-A algorithm update: Perform self-regular sparse update and dynamically adjust the compensation matrix Zn(t); DAC output control: Converts the updated result into an analog signal and outputs it to the compensation network via a DAC; Compensation Execution: The hybrid compensation network performs physical and active compensation to achieve matrix diagonalization. Feedback monitoring: The system detects the output signal in real time and updates the error estimate to achieve closed-loop dynamic compensation.

[0065] like Figure 5 As shown, the core mathematical idea of ​​this invention is: to design a compensation matrix. Make the equivalent impedance matrix Approaching a diagonal matrix. In the original matrix Z, the off-diagonal terms represent the inter-channel coupling impedance. Compensation matrix. It is formed by the physical compensation branch and the algorithmic dynamic compensation, and cancels out the off-diagonal components through superposition. When When the diagonal elements of the non-diagonal matrix are close to zero, the channels in the system are electrically decoupled, achieving signal independence and non-interference. This diagonalization method is universal and can be applied to array matrices of any size.

[0066] like Figure 6 As shown, the SRS-A algorithm is the core algorithm for dynamic compensation in this invention. Its process includes initialization, impedance estimation, error calculation, sparse constraint update, regular smoothing, output control, and convergence detection. The algorithm adopts a stepwise update approach, and its core recursive equation is:

[0067] Where μ is the step size factor, λ1 is the sparsity constraint coefficient, and λ2 is the regularization smoothing coefficient. This algorithm rapidly approaches zero non-critical coupling terms through sparsity constraints and suppresses noise interference through regularization, achieving stable convergence. Compared to traditional LMS or RLS algorithms, SRS-A exhibits stronger sparsity preservation and adaptability to dynamic changes.

[0068] like Figure 7 As shown, this invention constructs a complete closed-loop dynamic control system. The signal from the piezoresistive array is acquired by the ADC module and then input to the MCU control unit. The MCU internally executes the SRS-A algorithm to calculate compensation parameters based on the real-time sampled data. (t), and the calculation result is converted into an analog control signal through the DAC output module.

[0069] These signals drive the controllable current sources I1~I in the hybrid compensation network. n The system adjusts the voltage and current distribution of the physical branches to dynamically correct the coupling error of the array nodes. The compensated output voltage is simultaneously fed back to the MCU to form a real-time update closed loop. The system also has digital control signals (dashed arrows) for timing triggering and synchronization control, and the host computer can be debugged and parameters set through the communication interface. This closed-loop structure enables the system to maintain a decoupled state even when array characteristics change (such as temperature drift or aging).

[0070] like Figure 8 As shown, to prevent matrix degradation due to environmental changes or component drift after long-term system operation, this invention introduces a condition number monitoring and automatic recalibration mechanism.

[0071] The system calculates the residual energy E(t) and the matrix condition number in each iteration cycle. Its definition is: when Below the set threshold When, the system continues to execute the main algorithm loop; when When the threshold is exceeded, it indicates matrix instability, and the system automatically triggers the recalibration process. The recalibration process includes: executing the recalibration command and measuring the reference matrix. Update the initial value of the compensation matrix. Then it re-enters the SRS-A algorithm loop.

[0072] Through this mechanism, the system can achieve self-diagnosis and self-recovery during long-term operation, ensuring the stable diagonalization and compensation accuracy of the impedance matrix.

[0073] The dynamic decoupling system for sensor arrays described in this invention can suppress signal interference caused by electrical mutual coupling in multi-node piezoresistive sensor arrays in real time. It is applicable to flexible pressure sensor arrays, electronic skin, tactile detection panels and other multi-point array sensing and measurement applications.

[0074] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A decoupling method for a piezoresistive multi-channel pressure sensor array, characterized in that, include: Based on a piezoresistive multichannel pressure sensor array, a multichannel coupling impedance matrix is ​​established; wherein, the off-diagonal elements of the multichannel coupling impedance matrix represent the interconnection coupling between channels, and the diagonal elements represent the piezoresistive unit impedance of the current channel; The voltage signals of each piezoresistive unit of the multi-channel coupled impedance matrix are acquired in real time and digitized to obtain a time-varying impedance matrix; and the voltage signals of the multi-channel coupled impedance matrix in the zero-input state are acquired in calibration mode to form a reference impedance matrix. Based on the reference impedance matrix, a preliminary physical decoupling of the coupling impedance between channels is performed at the hardware level through a hybrid compensation network to obtain the static compensation matrix. Based on the time-varying impedance matrix, a self-regularized sparse algorithm is used to dynamically update the value of the static compensation matrix with the goal of minimizing the norm of the off-diagonal elements of the static compensation matrix, and convert it into an analog voltage signal; and based on the analog voltage signal, the adjustable element in the hybrid compensation network is driven to perform impedance adjustment. The output is a decoupled voltage signal that corresponds one-to-one with the pressure of each piezoresistive unit.

2. The decoupling method for a piezoresistive multi-channel pressure sensor array as described in claim 1, characterized in that, The construction of the reference impedance matrix includes: Under reference pressure conditions, a known excitation is injected by periodically scanning the voltage signals at each node of the multi-channel coupling impedance matrix. and measure Construct the reference impedance matrix : ； in, Indicates a known stimulus; This represents the voltage matrix actually acquired.

3. The decoupling method for a piezoresistive multi-channel pressure sensor array as described in claim 1, characterized in that, The preliminary physical decoupling of the coupling impedance between channels through a hybrid compensation network, resulting in a static compensation matrix, includes: The off-diagonal terms of the reference impedance matrix are calculated, and the initial physical static compensation matrix is ​​set according to the ideal cancellation requirements. Map the initial physical static compensation matrix to realizable circuit elements; The mapped physical compensation branches are connected to the array according to the topology. The inter-channel compensation branch is obtained by connecting compensation elements in parallel between nodes i and j; the ground compensation branch is obtained by connecting a capacitor in parallel with the reference ground at node i or by connecting an active current source. After the physical connection is completed, a static compensation matrix is ​​obtained; The initial physical static compensation matrix is ​​expressed as: ； The static compensation matrix is ​​expressed as: ； in, Represents the off-diagonal terms of the reference impedance matrix; Represents the initial physical static compensation matrix; Represents the reference impedance matrix; Represents the static compensation matrix; The off-diagonal residual energy of the static compensation matrix is ​​expressed as: ； like If the current static compensation is within an acceptable threshold, it is retained; otherwise, the constraint optimization is performed by solving a minimization problem with physical implementation constraints. The minimization problem with physical implementation constraints is expressed as: ； in, express ; The off-diagonal part of the matrix is ​​represented; the set S represents the space of realizable elements; This represents the equivalent compensation impedance matrix of the entire module.

4. The decoupling method for a piezoresistive multi-channel pressure sensor array as described in claim 1, characterized in that, The value of the dynamically updated static compensation matrix includes: According to the preset sampling period, the rows and columns of the multi-channel coupling impedance matrix are scanned to collect real-time voltage and current. For each pair of elements (i,j) in the matrix, cross-correlation / autocorrelation is calculated using a sliding time window of length W to estimate the impedance element. The calculation formula is as follows: ； The symbol * denotes complex conjugation; This represents the element in the i-th row and j-th column of the equivalent impedance matrix estimated at time t; ) represents the complex conjugate of the current signal; () represents the current signal applied to the j-th excitation channel at the k-th sampling time; (k) represents the voltage signal acquired at the i-th measurement channel at the k-th sampling time; W represents the length of the sliding time window used for impedance estimation; Construct the current estimation matrix And calculate the equivalent matrix residual based on the current estimated matrix and the static compensation matrix; Based on the equivalent matrix residue, calculate the residual vector; The equivalent matrix residual is expressed as: ； The residual vector is calculated using the following formula: ； Based on the self-regularized sparse algorithm, the static compensation matrix is ​​updated element by element; The core optimization objective function of the self-regularized sparse algorithm is expressed as: ； For elements Perform gradient descent and discretize to obtain the update formula: ； The gradient term can be expanded from the sampled data as follows: ； like ( If the threshold is set, the element update is frozen to reduce meaningless noise disturbances; for items that are zero for a long time, the corresponding compensation branch can be turned off in hardware to save power.

5. The decoupling method for a piezoresistive multi-channel pressure sensor array as described in claim 1, characterized in that, The hybrid compensation network adopts a hybrid topology, including: Inter-channel interconnection compensation branch: set between adjacent channel nodes to cancel lateral coupling, including resistors, capacitors and inductors; Ground compensation branch: Located between each channel and the reference ground, it is used to cancel parasitic coupling and common-mode interference of the ground wire, including capacitive elements and controllable current sources.

6. A piezoresistive multi-channel pressure sensor array decoupling system, characterized in that, include: Piezoresistive sensor array module: used to establish a multi-channel coupling impedance matrix based on a piezoresistive multi-channel pressure sensor array; wherein, the off-diagonal elements of the multi-channel coupling impedance matrix represent the interconnection coupling between channels, and the diagonal elements represent the piezoresistive unit impedance of the current channel; Signal acquisition module: used to acquire and digitize the voltage signals of each piezoresistive unit of the multi-channel coupled impedance matrix in real time to obtain a time-varying impedance matrix; and to acquire the voltage signals of the multi-channel coupled impedance matrix in the zero-input state in calibration mode to form a reference impedance matrix. Compensation network module: Used to perform preliminary physical decoupling of the coupling impedance between channels at the hardware level based on the reference impedance matrix and through a hybrid compensation network, so as to obtain a static compensation matrix; Matrix calculation and control module: Based on the time-varying impedance matrix, it dynamically updates the value of the static compensation matrix using a self-regulatory sparse algorithm, aiming to minimize the norm of the off-diagonal elements of the static compensation matrix, and converts it into an analog voltage signal; and based on the analog voltage signal, it drives the adjustable elements in the hybrid compensation network to perform impedance adjustment. Output module: Used to output decoupled voltage signals that correspond one-to-one with the pressure of each piezoresistive unit.

Images