Surface acoustic wave sensor detection circuit based on LMS adaptive filtering

Abstract

The invention belongs to the technical field of surface acoustic wave sensor signal processing, and discloses a surface acoustic wave sensor detection circuit based on LMS adaptive filtering, which comprises an analog detection circuit and a digital detection circuit, the analog detection circuit comprises an SAW oscillation circuit and a frequency mixing circuit, the digital detection circuit adopts an adaptive filtering module to filter a signal output by the analog detection circuit, and the frequency demodulator receives the filtered signal and performs frequency demodulation on the signal; the SAW oscillation circuit comprises a reference channel and a measurement channel, a reference channel signal and a measurement channel signal pass through the frequency mixing circuit to obtain an offset frequency signal, and the frequency of the offset frequency signal is the frequency of a physical quantity to be measured. Through the LMS adaptive filtering module, coefficients can be dynamically adjusted to suppress time-varying noise, the defect of fixed parameters of a traditional analog filter is overcome, the signal-to-noise ratio is improved, and the LMS adaptive filter is suitable for industrial automation, aerospace and other detection scenes.

Description

Technical Field

[0001] This invention belongs to the field of surface acoustic wave (SAW) sensor signal processing technology, and relates to a surface acoustic wave sensor detection circuit based on LMS adaptive filtering. Background Technology

[0002] Surface acoustic wave (SAW) sensors, with their advantages of small size, high sensitivity, and strong resistance to electromagnetic interference, have shown broad application prospects in environmental monitoring (such as temperature, humidity, and gas concentration detection), industrial automation (such as pressure and flow sensing), and medical diagnostics (such as biomolecular recognition). The core principle of SAW sensors is that changes in external physical quantities cause a frequency shift in the SAW resonator, and the detection circuit senses this change by measuring this frequency shift. Therefore, the accuracy of the detection circuit's frequency signal processing and its anti-interference capability directly determine the sensor's final performance.

[0003] The majority of existing SAW sensor detection circuits use the mixing method, which involves building two oscillation circuits, one as a reference circuit and the other as the measurement branch signal. The frequency offset of the signal is obtained by mixing the two signals and then passing them through a bandpass filter. The mixing method mainly relies on the analog signal processing architecture. The traditional mixing method uses analog filters with fixed parameters (such as RC or LC filters) to suppress noise, but it faces two major problems: First, the parameters are fixed. The filter parameters (such as cutoff frequency and bandwidth) are fixed by passive components such as resistors and capacitors, which cannot adapt to the dynamic changes of the sensor output signal (such as frequency drift of environmental noise and resonant frequency shift caused by sensor aging). When the noise frequency is close to the signal frequency, the analog filter is difficult to balance passband flatness and stopband attenuation, resulting in a significant decrease in signal-to-noise ratio (SNR) (Shen Xiaoqing, Yu Qinghua, Qiu Bin, Wang Zhihu, Xia Wei. Design and implementation of analog flat bandpass filter [J]. Automation and Instrumentation, 2022(2):130-132137); Second, the device aging effect. Analog components (such as capacitors and inductors) are easily affected by environmental factors such as temperature and humidity, resulting in parameter drift, causing the filter characteristics to deviate from the design value and insufficient long-term stability. Under long-term working conditions, the detection accuracy deteriorates significantly (M. Sathiyanathan; K. Anandhakumar; S. Jaganathan; CS Subashkumar, "Remaining Useful Life Prediction of Small and Large Signal Analog CircuitsUsing Filtering Algorithms," in Smart Systems for Industrial Applications, 2022, pp.93-113).

[0004] Adaptive finite impulse response (FIR) filters based on the Least Mean Squares (LMS) algorithm automatically adjust filter coefficients to suppress specific frequency interference by minimizing the error signal energy. With the maturity of digital signal processing technology, adaptive filters can be implemented using Field Programmable Gate Arrays (FPGAs), offering advantages such as high accuracy, strong stability, and ease of integration. FPGA-based LMS adaptive filtering systems process signals in real time and can automatically adjust filter coefficients according to changes in the noise environment, effectively suppressing various types of noise interference and greatly improving signal detection accuracy (C. Safarian, T. Ogunfunmi, WJ Kozacky and BK Mohanty, "FPGA implementation of LMS-based FIR adaptive filter for real-time digital signal processing applications," 2015 IEEE International Conference on Digital Signal Processing (DSP), 2015, pp. 1251-1255). Furthermore, digital implementation of filtering avoids problems such as temperature drift and aging in analog circuits, enhancing system reliability.

[0005] However, existing LMS algorithms face challenges when applied to SAW sensor detection circuits. LMS alters the filter's transfer function through real-time adjustment of digital weighting coefficients; it is programmable, discrete, and nonlinear. SAW sensors, on the other hand, achieve their sensor function through the geometry of their interdigital electrodes (IDT) on a piezoelectric substrate, fixedly and continuously. Once fabricated, their characteristics (such as amplitude response, phase response, and group delay) are essentially immutable, making it impossible to directly adjust the device's physical properties using digital algorithms. Summary of the Invention

[0006] To address the above problems, this invention provides a high-precision, low-power, and real-time surface acoustic wave (SAW) sensor signal detection system and method. An adaptive filtering algorithm effectively solves the problems of existing SAW sensors being susceptible to signal interference and having low measurement accuracy in complex environments. By using a hybrid analog and digital circuit system, the invention overcomes the limitation of directly adjusting the physical characteristics of SAW sensors using the LMS filtering algorithm, thus meeting the high-precision measurement requirements of fields such as industrial automation, aerospace, and automotive electronics.

[0007] The technical solution of the present invention:

[0008] A surface acoustic wave (SAW) sensor detection circuit based on LMS adaptive filtering includes an analog detection circuit and a digital detection circuit. The analog detection circuit includes a SAW oscillation circuit and a mixer circuit. The digital detection circuit uses an adaptive filtering module to filter the signal output from the analog detection circuit. A frequency demodulator receives the filtered signal and performs frequency demodulation on the signal. The SAW oscillation circuit includes a reference channel and a measurement channel. The reference channel signal and the measurement channel signal are mixed by a mixer circuit to obtain a bias signal, the frequency of which is the frequency of the physical quantity to be measured.

[0009] Furthermore, the SAW oscillation circuit adopts a pierce oscillation circuit topology, including:

[0010] The four resistors are R1, R2, R3, and R4.

[0011] The seven capacitors are C1, C2, C3, C4, C5, C6, and C7.

[0012] Two inductors, L1 and L2;

[0013] DC power supply V1 and high-frequency transistor Q1;

[0014] Capacitor C1 and resistor R3 are connected in parallel to the base and emitter of high-frequency transistor Q1; resistor R4 is connected in series with the emitter of high-frequency transistor Q1 and grounded; resistor R1 is connected in parallel to the base and collector of high-frequency transistor Q1; the collector of high-frequency transistor Q1 is connected in series with resistor R2; to prevent high-frequency signal grounding, inductor L1 is connected in series with DC power supply V1 and grounded, and inductor L2 is connected in series with resistor R4 and grounded; to suppress the influence of current transients on the oscillation circuit, capacitors C6 and C7 are connected in series and then in parallel with DC power supply V1; capacitor C2 is connected in parallel with the collector and emitter of high-frequency transistor Q1; capacitor C3 is connected in parallel with the emitter of high-frequency transistor Q1 and grounded; capacitor C4 is connected in series with the SAW sensor and in parallel between the emitter and collector of high-frequency transistor Q1; the signal output terminal is connected to capacitors C2 and C4 respectively, and capacitor C5 is connected in series at the signal output terminal to reduce the influence of DC on the output signal.

[0015] The reference channel circuit and the measurement channel circuit both use the above-mentioned SAW oscillation circuit. The difference is that the SAW components in the measurement channel are in the environment under test.

[0016] Furthermore, the mixing circuit consists of a microstrip directional coupler and two mixing diodes. The signal output terminal of the measurement channel circuit is connected to the input terminal 1 of the microstrip directional coupler, and the signal output terminal of the reference channel circuit is connected to the isolation terminal 4 of the microstrip directional coupler. The two mixing diodes can be selected as HSMS-282 diodes with high frequency, low noise, and low reverse current. One mixing diode is forward-biased and connected to the through terminal 2 of the microstrip directional coupler, and the other mixing diode is reverse-biased and connected to the coupling terminal 3 of the microstrip directional coupler. The negative terminal of the mixing diode at the through terminal 2 is connected to the positive terminal of the mixing diode at the coupling terminal 3 to the output terminal port2, and the signal is transmitted to the adaptive filtering module.

[0017] The further adaptive filtering module includes a signal delay chain submodule, a forward filter submodule, an error calculation submodule, and a coefficient update submodule. The signal input to the mixer circuit is processed by a shift register chain in the signal delay chain submodule to form a finite impulse response window of the input signal, and the input signal is stored to form an input vector. Each register stage uses fixed-point numbers and passes the input vector generated by the signal delay chain submodule to the forward filter submodule. The forward filter submodule processes the input vector from the signal delay chain submodule using an FIR filter. A 32-tap FIR filter is used, and its mathematical model is as follows: ,in This represents the SAW sensor input signal delayed by i sampling periods. This represents the filter tap coefficients for the i-th sampling period at time n. This represents the filtered output signal; the forward filter submodule instantiates a parallel multiplier with the same number of filter taps, performs accumulation calculations using a five-level Wallace tree structure, and finally outputs the mathematical model calculation result to the error calculation submodule; the error calculation submodule selects the valid bits from the output signal of the forward filter submodule for output. Then with the desired signal Subtracting them yields the instantaneous error signal. The obtained error signal is transmitted to the coefficient update submodule for processing. The coefficient update submodule is responsible for updating and calculating the filter coefficients, and the mathematical model is as follows: ,in This represents the updated filter tap coefficients for the i-th sampling period. Let represent the filter tap coefficients for the i-th sampling period at time n, and u represent the convergence step size factor. This represents the instantaneous error signal at time n. This represents the SAW sensor input signal delayed by i sampling periods. The coefficient update submodule updates the filter tap coefficients for each instantaneous signal until the error signal converges, completing the adaptive processing of the filter tap coefficients, and then transmits the adaptively processed signal to the frequency demodulator.

[0018] The further frequency demodulator uses two registers to delay the input signal, compares the delayed signal with the current signal, and detects rising and falling edges. A counter is used to reset and start a new cycle count upon zero-crossing detection. The counter increments each clock cycle until the next zero-crossing point or the maximum value is reached. Whenever a zero-crossing is detected, the current counter value is stored in a buffer, and the buffer index is updated. If not enough samples have been collected, the count is directly accumulated. If the average number of cycles has been reached or exceeded, a sliding window approach is used, subtracting the oldest value and adding the latest value. Finally, the frequency value is obtained by dividing the system clock frequency by the average number of cycles.

[0019] This invention provides a surface acoustic wave (SAW) sensor detection circuit based on adaptive filtering. The working principle is as follows: when the SAW sensor in the measurement path senses a change in the surrounding environment, the output frequency of the oscillation circuit in the measurement path changes, while the frequency of the reference path remains unchanged. The two signals are mixed to obtain a frequency difference signal, which needs to undergo adaptive digital filtering to filter out noise. Finally, the filtered signal is demodulated by a frequency demodulator to obtain the frequency value of the signal under test.

[0020] The beneficial effects of this invention are:

[0021] This invention uses an LMS adaptive FIR filter to dynamically adjust coefficients to suppress time-varying noise, overcoming the fixed parameter defects of traditional analog filters and improving the signal-to-noise ratio.

[0022] Existing frequency demodulation techniques mostly employ fixed threshold detection, which is susceptible to noise interference, leading to period measurement errors. This invention uses bidirectional zero-crossing detection in the frequency demodulator module, simultaneously detecting the zero-crossing points of both the rising and falling edges, thus avoiding the misjudgment problems associated with single-threshold detection.

[0023] The core digital signal processing (adaptive filtering and frequency demodulation) is implemented based on the FPGA platform, which makes full use of the parallel processing capability of the FPGA to meet high real-time requirements. At the same time, it is easy to integrate the system and reduce hardware complexity. The digital implementation is naturally immune to temperature drift and aging of analog devices, and the system has good long-term stability. Parameters (such as filter order, step size u, reference frequency, etc.) can be configured by changing parameters, which can flexibly adapt to different types of SAW sensors and diverse detection scenarios. Attached Figure Description

[0024] Figure 1A schematic diagram of the detection circuit of a surface acoustic wave sensor based on adaptive filtering provided by the present invention;

[0025] Figure 2 This is a schematic diagram of the SAW oscillator circuit based on the pierce structure of this invention;

[0026] Figure 3 This is a schematic diagram of the mixer circuit structure of the present invention;

[0027] Figure 4 This is a block diagram of the LMS adaptive filtering module of the present invention;

[0028] Figure 5 This is a structural diagram of the LMS adaptive filter of the present invention;

[0029] Figure 6 Figure 1 shows the MATLAB simulation diagram of the LMS algorithm of this invention. Figure 2 shows the original sine signal, Figure 3 shows the randomly generated noise signal, Figure 4 shows the superimposed signal of the sine signal and the noise signal, and Figure 5 shows the signal after processing by the LMS algorithm.

[0030] Figure 7 Figure 1 shows the simulation results of the LMS adaptive filtering module of the present invention. Figure 2 shows the original sine signal, Figure 3 shows the signal after noise is added, Figure 4 shows the sine signal after adaptive filtering, and Figure 5 shows the error signal after adaptive filtering. Detailed Implementation

[0031] The present invention will be further described below with reference to the accompanying drawings:

[0032] This invention provides a surface acoustic wave sensor detection circuit based on LMS adaptive filtering, such as... Figure 1 As shown, the present invention includes a SAW measurement channel, a reference channel, and a mixing circuit connected in sequence. The mixing circuit transmits the mixed frequency signal to an adaptive filtering module for filtering. The filtered signal is then transmitted to a frequency demodulator unit for frequency measurement, ultimately achieving the detection function.

[0033] The oscillation circuit adopts a pierce oscillation circuit topology, as follows: Figure 2 As shown, in this invention, an equivalent circuit of a surface acoustic wave device is used instead of the actual device to facilitate subsequent circuit measurements. The change in the measurement path is determined by the change in dynamic inductance in the equivalent circuit. By adjusting the circuit parameters, the oscillation circuit can output both a reference signal and the device's test signal.

[0034] The mixer circuit is designed using a microstrip directional coupler and an HSMS-282c mixer diode, as shown below. Figure 3As shown, the signal output terminal of the measurement channel circuit is connected to the input terminal 1 of the microstrip directional coupler, and the signal output terminal of the reference channel circuit is connected to the isolation terminal 4 of the microstrip directional coupler. The signal is mixed and then transmitted to the adaptive filtering module.

[0035] The output signal model of the SAW sensor at time n is: , Represents useful signal components, Represents environmental noise and circuit noise. The actual observed mixed signal. Through adaptive filtering, from... Extract The estimated value.

[0036] LMS adaptive pre-filtering module, such as Figure 4 As shown, it includes a signal delay chain submodule, a forward filter submodule, an error calculation submodule, and a coefficient update submodule. The signal input to the mixer circuit passes through a 32-stage shift register chain in the signal delay chain submodule to form a finite impulse response window of the input signal, storing 32 input signals to form an input vector. The input vector formed by the signal delay chain is passed to the forward filter module.

[0037] The forward filter module instantiates 32 parallel multipliers, employing a five-stage Wallace tree-structured accumulation architecture. The first stage involves 16 parallel additions, each with 2 operands; the second stage involves 8 parallel additions, each with 4 operands; the third stage involves 4 parallel additions, each with 8 operands; the fourth stage involves 2 parallel additions, each with 16 operands; and the fifth stage outputs the filtered result. A further error calculation module selects the valid output from the filtered output. Then with the desired signal Subtracting them yields the instantaneous error signal. The obtained error signal is transmitted to the coefficient update submodule for processing. The coefficient update submodule is responsible for calculating the update of the filter coefficients, and the mathematical model is as follows: The coefficient update submodule performs coefficient update calculations for each instantaneous signal until the error signal converges, completing the adaptive processing of the signal and transmitting the adaptively processed signal to the frequency demodulation module. Simulations of the LMS algorithm in MATLAB are as follows: Figure 6 Figure (d) shows the signal after processing by the LMS algorithm. In the initial stage, the filter is converging from 0.01 ms to 0.012 ms, and the noise is significantly suppressed from 0.012 ms to 0.024 ms, and the sine wave is clearly visible.

[0038] The adaptive filtering module uses the Zynq-XC7Z020 development board as its processor. This board is equipped with a dual-core ARM Cortex-A9 processor, and its parallel computing capabilities significantly accelerate the LMS adaptive filtering algorithm. It can also be integrated onto the same FPGA for convenient data transmission and display. The aforementioned digital circuit modules are implemented using a hardware description language, and simulation and debugging results are obtained using Modelsim software. Figure 7 As shown in Figure (d), the error signal after adaptive filtering is rapidly converged within a hundred clock cycles. After convergence, the output signal fluctuates very little, and the module's mathematical operations are in line with theoretical expectations.

[0039] The signal processed by the adaptive filtering module is delayed by two registers. The delayed signal is compared with the current signal, and rising and falling edges are detected. A counter is reset and a new cycle count begins upon zero-crossing detection. The counter increments every clock cycle until the next zero-crossing point or the maximum value is reached. Whenever a zero-crossing is detected, the current counter value is stored in a buffer, and the buffer index is updated. Finally, the frequency value is obtained by dividing the system clock frequency by the average period.

Claims

1. A surface acoustic wave sensor detection circuit based on LMS adaptive filtering, characterized in that, The LMS adaptive filtering-based surface acoustic wave sensor detection circuit includes an analog detection circuit and a digital detection circuit. The analog detection circuit includes a SAW oscillation circuit and a mixer circuit. The digital detection circuit uses an adaptive filtering module to filter the signal output from the analog detection circuit. The frequency demodulator receives the filtered signal and demodulates the frequency of the signal. The SAW oscillation circuit includes a reference channel circuit and a measurement channel circuit. The reference channel signal and the measurement channel signal are mixed by a mixer circuit to obtain a bias signal. The frequency of the bias signal is the frequency of the physical quantity to be measured.

2. The surface acoustic wave sensor detection circuit based on LMS adaptive filtering according to claim 1, characterized in that, The SAW oscillator circuit adopts a pierce oscillator circuit topology, including: The four resistors are R1, R2, R3, and R4. The seven capacitors are C1, C2, C3, C4, C5, C6, and C7. Two inductors, L1 and L2; DC power supply V1 and high-frequency transistor Q1; Capacitor C1 and resistor R3 are connected in parallel to the base and emitter of high-frequency transistor Q1; resistor R4 is connected in series with the emitter of high-frequency transistor Q1 and grounded; resistor R1 is connected in parallel to the base and collector of high-frequency transistor Q1; the collector of high-frequency transistor Q1 is connected in series with resistor R2; to prevent high-frequency signal grounding, inductor L1 is connected in series with DC power supply V1 and grounded, and inductor L2 is connected in series with resistor R4 and grounded; to suppress the influence of current transients on the oscillation circuit, capacitors C6 and C7 are connected in series and then in parallel with DC power supply V1; capacitor C2 is connected in parallel with the collector and emitter of high-frequency transistor Q1; capacitor C3 is connected in parallel with the emitter of high-frequency transistor Q1 and grounded; capacitor C4 is connected in series with the SAW sensor and in parallel between the emitter and collector of high-frequency transistor Q1; the signal output terminal is connected to capacitors C2 and C4 respectively, and capacitor C5 is connected in series at the signal output terminal to reduce the influence of DC on the output signal.

3. The surface acoustic wave sensor detection circuit based on LMS adaptive filtering according to claim 2, characterized in that, Both the reference channel circuit and the measurement channel circuit use the aforementioned SAW oscillation circuit, the difference being that the SAW components in the measurement channel are in the environment under test.

4. The surface acoustic wave sensor detection circuit based on LMS adaptive filtering according to claim 2, characterized in that, The mixing circuit consists of a microstrip directional coupler and two mixing diodes. The signal output terminal of the measurement channel circuit is connected to the input terminal 1 of the microstrip directional coupler, and the signal output terminal of the reference channel circuit is connected to the isolation terminal 4 of the microstrip directional coupler. The two mixing diodes can be selected as HSMS-282 diodes with high frequency, low noise, and low reverse current. One mixing diode is connected to the through terminal 2 of the microstrip directional coupler in the forward direction, and the other mixing diode is connected to the coupling terminal 3 of the microstrip directional coupler in the reverse direction. The negative terminal of the mixing diode at the through terminal 2 is connected to the positive terminal of the mixing diode at the coupling terminal 3 to the output terminal port2, and the signal is transmitted to the adaptive filtering module.

5. The surface acoustic wave sensor detection circuit based on LMS adaptive filtering according to claim 4, characterized in that, The adaptive filtering module includes a signal delay chain submodule, a forward filter submodule, an error calculation submodule, and a coefficient update submodule. The signal input to the mixer circuit is processed by a shift register chain in the signal delay chain submodule to form a finite impulse response window of the input signal, and the input signal is stored to form an input vector. Each register stage uses fixed-point numbers and passes the input vector generated by the signal delay chain submodule to the forward filter submodule. The forward filter submodule processes the input vector from the signal delay chain submodule using an FIR filter. A 32-tap FIR filter is used, and its mathematical model is as follows: ,in This represents the SAW sensor input signal delayed by i sampling periods. This represents the filter tap coefficients for the i-th sampling period at time n. This represents the filtered output signal; the forward filter submodule instantiates a parallel multiplier with the same number of filter taps, performs accumulation calculations using a five-level Wallace tree structure, and finally outputs the mathematical model calculation result to the error calculation submodule; the error calculation submodule selects the valid bits from the output signal of the forward filter submodule for output. Then with the desired signal Subtracting them yields the instantaneous error signal. The obtained error signal is transmitted to the coefficient update submodule for processing. The coefficient update submodule is responsible for updating and calculating the filter coefficients. The mathematical model is as follows: ,in This represents the updated filter tap coefficients for the i-th sampling period. Let represent the filter tap coefficients for the i-th sampling period at time n, and u represent the convergence step size factor. This represents the instantaneous error signal at time n. This represents the SAW sensor input signal delayed by i sampling periods; The coefficient update submodule updates the filter tap coefficients for each instantaneous signal until the error signal converges, completing the adaptive processing of the filter tap coefficients, and then transmits the adaptively processed signal to the frequency demodulator.

6. The surface acoustic wave sensor detection circuit based on LMS adaptive filtering according to claim 5, characterized in that, The frequency demodulator uses two registers to delay the input signal, compares the delayed signal with the current signal, and detects rising and falling edges. A counter is used to reset and start a new cycle count upon zero-crossing detection. The counter increments each clock cycle until the next zero-crossing point or the maximum value is reached. Whenever a zero-crossing is detected, the current counter value is stored in a buffer, and the buffer index is updated. If not enough samples have been collected, the counter is directly incremented. If the average number of cycles has been reached or exceeded, a sliding window approach is used to subtract the oldest value and add the latest value. Finally, the frequency value is obtained by dividing the system clock frequency by the average number of cycles.

Images