Adaptive filter for fiber optic sensors

Abstract

The optical fiber sensor adaptive filter of the present invention includes a front-end digital filter and a back-end analog filter. The front-end digital filter adopts filtering using an FIR filter adaptive filtering coefficient, estimates the FIR length N through frequency deviation, obtains the correlation coefficient n, and sets N and corresponding weights according to the correlation coefficient and frequency accuracy, that is, the resolution requirement. The weight value is improved by an additional momentum method, which can avoid overfitting or underfitting and ensure filtering performance. The back-end analog filter adopts attenuation + variable impedance acquisition + transformer primary frequency selection and transformer secondary induction tuning, and finally filters through an adjustable harmonic filter and an LC frequency selection network, which can effectively filter noise interference and improve measurement accuracy.

Description

Technical Field

[0001] The invention belongs to the technical field of optical fiber sensing, and in particular relates to an adaptive filter for an optical fiber sensor. Background Art

[0002] Fiber optic sensors play an important role in multiple fields, including communications and industrial monitoring. In fiber optic sensing systems, modulation techniques are often used to convert the physical quantities to be measured (such as temperature, pressure, and displacement) into optical signals for transmission and processing. Modulation can be intensity modulation, phase modulation, or frequency modulation. Frequency modulation introduces a modulation frequency that is related to the physical quantity to be measured. By detecting the frequency changes of the modulated optical signal, information about the physical quantity to be measured can be inferred. Furthermore, when detecting certain physical quantities and subsequent signal processing, such as vibration measurement, only vibration signals within a specific frequency range can be accurately detected. Subsequent signal processing, such as filtering, amplification, and conversion, also needs to be performed within this specific frequency range.

[0003] During the optical fiber transmission process, it is inevitable that it will be interfered by various noises. Noise interference will cause signal frequency deviation and waveform distortion, affecting the accuracy and reliability of the signal. Therefore, it is necessary to provide an adaptive filter for the optical fiber sensor to effectively filter the noise interference. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an adaptive filter for optical fiber sensors, which filters the received front-end modulation frequency signal using the adaptive filter coefficient of the FIR filter, and processes the back-end signal using attenuation + variable impedance acquisition + transformer primary frequency selection and transformer secondary induction tuning, and finally filters the signal through an adjustable harmonic filter and an LC frequency selection network, thereby effectively filtering out noise interference.

[0005] The technical solution is to include a front-end digital filter and a back-end analog filter. The front-end digital filter receives the modulation signal of the optical fiber sensor, filters it using the adaptive filter coefficient of the FIR filter to remove noise, extract useful information, and convert it into an analog signal through a DA converter. The FIR filter includes an FIR controller, an adder, a multiplier, and an accumulator.

[0006] The back-end analog filter receives the analog signal, controls the intensity of the optical signal within a certain range through the attenuator, then obtains the optical signal through the tuner variable impedance, performs primary frequency selection on the transformer, performs inductive tuning on the secondary side of the transformer, and finally filters the optical signal through the tunable harmonic filter and the LC frequency selection network for output.

[0007] The steps of determining the coefficients of the adaptive filter coefficients are as follows:

[0008] Step 1: Obtain the desired frequency characteristic signal;

[0009] Step 2, obtaining an input signal;

[0010] Step 3, extracting the frequency characteristic component of the input signal;

[0011] Step 4, estimate the frequency deviation;

[0012] Step 5, estimate the FIR length N according to the frequency deviation;

[0013] Step 6: Estimate the corresponding weights of each convolutional layer;

[0014] In step 7, the input at each moment is multiplied by the weight coefficient by the multiplier, and finally the accumulator performs superposition to output the filtered signal.

[0015] Preferably, in order to avoid overfitting and underfitting, the FIR length N and the corresponding weight are determined by the frequency deviation and the frequency accuracy, and the weight is adjusted positively when the frequency deviation trend becomes smaller, and stopped when it becomes larger. The weight adjustment range is determined, and fine-tuning is performed within the weight adjustment range. The number of fine-tuning is set to 1 / 2 of the frequency accuracy.

[0016] Preferably, a momentum factor is introduced and an additional momentum method is used to improve the weight value correction process. Specifically, a part of the last or previous weight adjustment amount is superimposed on the weight adjustment amount calculated according to the current error as the actual weight adjustment amount this time.

[0017] Preferably, the attenuator includes an inductor L1 and a capacitor C1, one end of the inductor L1 and one end of the capacitor C1 are connected to the analog signal, the other end of the inductor L1 is respectively connected to the other end of the capacitor C1, the right end of the diode BD1, and one end of the capacitor C3, the cathode of the diode BD1 is connected to the detection signal amplitude, and the left end of the diode BD1 is connected to one end of the grounded capacitor C2;

[0018] The tuner includes a potentiometer RW1, an inductor L, and a capacitor C4. The potentiometer RW1, the inductor L, and the capacitor C4 constitute a variable impedance network. The inductor L, the capacitor C4, and the primary coil of the transformer T1 constitute a frequency selection network. High frequency is added to the primary coil of the transformer T1 through the capacitor C4, and low frequency is added to the primary coil of the transformer T1 through the inductor L. The secondary coil of the transformer T1 and the varactor diode BD2 are connected in parallel to form an inductive tuning network. The upper end of the potentiometer RW1, the upper end of the inductor L, and one end of the capacitor C4 are connected to the other end of the inductor L1. The other end of the capacitor C4 is connected to one end of the primary coil of the transformer T1. The other end of the primary coil of the transformer T1, the lower end of the potentiometer RW1, and the lower end of the inductor L are connected to ground. One end of the secondary coil of the transformer T1 is connected to the upper end of the varactor diode BD2, and the other end of the secondary coil of the transformer T1 is connected to the lower end of the varactor diode BD2.

[0019] The adjustable harmonic filter receives the tuner output signal and realizes the interference noise frequency component in the notch filtering tuning signal, including a resistor R4 and an electrolytic capacitor E2, one end of the resistor R4 and the upper end of the electrolytic capacitor E2 are connected to one end of the secondary coil of the transformer T1, the negative electrode of the electrolytic capacitor E2 is respectively connected to the positive electrode of the electrolytic capacitor E3 and one end of the potentiometer RW2, the other end of the resistor R4 is connected to the positive electrode of the electrolytic capacitor E4, the negative electrode of the electrolytic capacitor E3 is respectively connected to the negative electrode of the electrolytic capacitor E4 and the other end of the potentiometer RW2, the adjustable end of the potentiometer RW2 is connected to the ground through the resistor R5, and the adjustable end of the potentiometer RW1 is also connected to the noise frequency signal, and the noise frequency signal is used to control the frequency of the notch filtering;

[0020] The LC frequency selection network is composed of an inductor LN and a variable capacitance diode BD3, and finally outputs the filtered and conditioned optical fiber sensor signal through a capacitor C5.

[0021] The beneficial effects of the present invention are as follows: by filtering the received front-end modulation frequency signal with an FIR filter adaptive filter coefficient, and processing the back-end signal with attenuation + variable impedance acquisition + transformer primary frequency selection, transformer secondary induction tuning, and finally filtering with an adjustable harmonic filter and an LC frequency selection network, noise interference can be effectively filtered;

[0022] Among them, the filtering of the FIR filter adaptive filter coefficient estimates the FIR length N through the frequency deviation to obtain the correlation coefficient n, and sets N and the corresponding weight according to the correlation coefficient and frequency accuracy, that is, the resolution requirement. The additional momentum method is used to improve the weight value, which can avoid overfitting or underfitting and ensure the performance of the filter. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flow chart of the steps for determining the coefficients of the adaptive filter coefficients of the present invention.

[0024] Figure 2 It is a circuit principle diagram of the present invention. DETAILED DESCRIPTION

[0025] The technical solutions of various embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0026] The following is in conjunction with the instructions Figures 1 to 2 , the specific implementation methods of the present invention are further described in detail.

[0027] The fiber optic sensor adaptive filter includes a front-end digital filter and a back-end analog filter. The front-end digital filter receives the modulation signal of the fiber optic sensor and adopts the FIR filter adaptive filter coefficient to remove noise and extract useful information. The signal is converted into an analog signal through a DA converter (also known as digital-to-analog conversion). The FIR filter includes an FIR controller, an adder, a multiplier, and an accumulator. The FIR controller is the control core, which receives the expected frequency characteristic signal and the input signal (that is, the signal with the modulation frequency output after the fiber optic sensor is measured), estimates the frequency deviation, estimates the FIR length N and the corresponding weight, and sends a control signal to the adder, multiplier, and accumulator. The multiplier completes the multiplication of the input signal and the weight. When the adder is symmetrical in N, it first performs addition and then multiplication. The accumulator completes the accumulation of the weighted signals output by multiple multipliers to obtain the filtered output signal.

[0028] The back-end analog filter receives the analog signal and controls the optical signal intensity within a certain range through an attenuator. The signal is then acquired through a tuner with variable impedance, and the primary side of the transformer performs frequency selection to select a specific frequency range. The secondary side of the transformer performs inductive tuning, and the signal is finally filtered through a tunable harmonic filter to remove the detected noise frequency signal. The signal is then filtered through an LC frequency selection network to output signals outside the specific frequency range.

[0029] The steps of determining the coefficients of the adaptive filter coefficients are as follows:

[0030] Step 1: Obtain the desired frequency characteristic signal;

[0031] Step 2, obtaining an input signal;

[0032] Step 3: Extract the frequency characteristic components of the input signal. For example, a frequency sweep analysis method can be used to sweep the frequency within a certain frequency range to analyze the signal amplitude and frequency components.

[0033] Step 4, estimating the frequency deviation, for example, can use a correlation method to estimate the frequency deviation between the two signals by calculating the correlation between them. When the frequencies of the desired frequency characteristic signal and the input signal are close, the correlation between them will be high, otherwise, the correlation will be low;

[0034] Step 5. Estimate the FIR length N based on the frequency deviation, set the correlation coefficient n to be greater than 0.7, and when the frequency accuracy, that is, the resolution requirement, is high (for example, fiber optic vibration sensors, generally speaking, the distance resolution of fiber optic vibration sensors can reach the micron level. High-performance fiber optic vibration sensors can achieve a distance resolution of less than 0.1 micron. Some simpler fiber optic vibration sensors have a distance resolution of only hundreds of microns or microns. 0.1 micron is set as a high requirement), set N to (1-n)*2 / 0.1. To avoid overfitting and underfitting, the FIR length N and the corresponding weight are determined by the frequency deviation and frequency accuracy. The weight is adjusted positively when the frequency deviation trend decreases, and stops when it increases. Determine the weight adjustment range, and make fine adjustments within the weight adjustment range. The number of fine adjustments is set to 1 / 2 of the frequency accuracy.

[0035] Step 6: Estimate the corresponding weights of each convolutional layer, and initially set them to be evenly distributed, that is, (1 / N)*100%;

[0036] In step 7, the input at each moment is multiplied by the weight coefficient by the multiplier, and finally the accumulator performs superposition to output the filtered signal.

[0037] On the basis of the above scheme, it also includes the introduction of momentum factor and the use of additional momentum method to improve the weight value correction process. The specific approach is: a part of the last or previous weight adjustment amount is superimposed on the weight adjustment amount calculated according to the current error as the actual weight adjustment amount this time.

[0038] On the basis of the above scheme, the attenuator controls the intensity of the optical signal within a certain range, that is, attenuates the analog signal after FIR digital filtering and DA conversion, that is, the modulated frequency signal, and is controlled by the detection signal amplitude. Here, the detection signal amplitude can be the amplitude of the output signal of the optical fiber sensor, obtained by the peak detection circuit, or obtained by the amplitude detection of the modulated frequency signal by the amplitude detector. The specific structure adopted is: the attenuator is composed of an inductor L1 in parallel with the capacitor C1, a diode BD1 in series with the capacitor C2, and then in parallel with the capacitor C3. The diode BD1 is a varactor diode. The voltage corresponding to the detection signal amplitude is applied to the cathode of the varactor diode, and the impedance is adjusted to achieve attenuation. It can also be 1SV172. The current corresponding to the detection signal amplitude is applied to the cathode. As the current increases, its impedance decreases, and the impedance is adjusted to achieve attenuation. It includes an inductor L1 and a capacitor C1. One end of the inductor L1 and one end of the capacitor C1 are connected to the analog signal. The other end of the inductor L1 is respectively connected to the other end of the capacitor C1, the right end of the diode BD1, and one end of the capacitor C3. The cathode of the diode BD1 is connected to the detection signal amplitude, and the left end of the diode BD1 is connected to one end of the grounded capacitor C2.

[0039] The tuner is then obtained by the variable impedance of the potentiometer RW1. The potentiometer RW1 is a voltage-controlled potentiometer, which is reversely adjusted by the amplitude negative feedback of the detection signal. The inductor L, capacitor C4, and primary coil of the transformer select the frequency and select a specific frequency range. The secondary induction tuning of the transformer is restored to the frequency range of the optical signal intensity. The induced frequency range can be achieved by adding a voltage to the negative electrode of the varactor diode BD2. The potentiometer RW1, inductor L, and capacitor C4 form a variable impedance network. The inductor L, capacitor C4, and primary coil of the transformer T1 form a frequency selection network. The high frequency is transmitted through the capacitor C4 is added to the primary coil of transformer T1, and the low frequency is added to the primary coil of transformer T1 through inductor L. The secondary coil of transformer T1 and varactor diode BD2 are connected in parallel to form an inductive tuning network. The upper end of potentiometer RW1, the upper end of inductor L, and one end of capacitor C4 are connected to the other end of inductor L1. The other end of capacitor C4 is connected to one end of the primary coil of transformer T1. The other end of the primary coil of transformer T1, the lower end of potentiometer RW1, and the lower end of inductor L are connected to ground. One end of the secondary coil of transformer T1 is connected to the upper end of varactor diode BD2, and the other end of the secondary coil of transformer T1 is connected to the lower end of varactor diode BD2.

[0040] The adjustable harmonic filter receives the tuner output signal, and realizes the interference noise frequency component f in the notch filtering tuning signal through resistors R4 and R5 and electrolytic capacitor E2, wherein the notch is adjusted to the interference noise frequency by adjusting the resistance value of the potentiometer RW2. For example, the frequency generated by the adjustable harmonic filter and the interference noise frequency can be discriminated. When the frequency is the same, the voltage is zero. When the frequency is different, the voltage difference is added to RW2 for adjustment. The filter includes resistor R4 and electrolytic capacitor E2, one end of the resistor R4 and the upper end of the electrolytic capacitor E2 are connected to one end of the secondary coil of the transformer T1, the negative pole of the electrolytic capacitor E2 is respectively connected to the positive pole of the electrolytic capacitor E3 and one end of the potentiometer RW2, the other end of the resistor R4 is connected to the positive pole of the electrolytic capacitor E4, the negative pole of the electrolytic capacitor E3 is respectively connected to the negative pole of the electrolytic capacitor E4 and the other end of the potentiometer RW2, the adjustable end of the potentiometer RW2 is connected to the ground through the resistor R5, and the adjustable end of the potentiometer RW1 is also connected to the noise frequency signal, and the noise frequency signal is used to control the frequency of the notch filter;

[0041] The LC frequency selection network is composed of an inductor LN and a variable capacitance diode BD3, and finally outputs the filtered and conditioned optical fiber sensor signal through a capacitor C5.

[0042] The reverse excitation circuit receives the detection signal amplitude, filters it in the loop, reversely charges it, and adds it to the gate of the MOS tube M1, changing the drain voltage of the MOS tube M1. The drain voltage is superimposed with -5V and added to the potentiometer RW1, changing the impedance of the potentiometer RW1. In this way, by filtering the received front-end modulation frequency signal with the adaptive filter coefficient of the FIR filter, and processing the back-end signal with attenuation + variable impedance acquisition + transformer primary frequency selection and transformer secondary induction tuning, and finally filtering it with the adjustable harmonic filter and LC frequency selection network, the noise interference can be effectively filtered. The circuit includes capacitor C6, resistor R6, and inductor L4. One end of the capacitor C6, one end of the resistor R6, and one end of the inductor L4 are connected. The amplitude of the detection signal is received. The other end of capacitor C6 is connected to ground, the other end of resistor R6 is connected to one end of grounded capacitor C7, the other end of inductor L4 is respectively connected to one end of inductor L3 and the negative electrode of electrolytic capacitor E4, the other end of inductor L3 is connected to the gate of MOS transistor M1, the drain of MOS transistor M1 is connected to one end of inductor L2, the other end of inductor L2 and one end of grounded capacitor C8 are connected to the -5V power supply, the source of MOS transistor M1 is respectively connected to the negative electrode of diode D1, one end of grounded resistor R7, and the normally open contact of switch JP1, the anode of diode D1 is connected to ground, the normally closed contact of switch JP1 is connected to the -5V power supply, and the common end of switch JP1 is connected to the adjustable end of potentiometer RW1. When the present invention is used specifically, the front-end digital filter receives the modulation signal of the optical fiber sensor, filters the signal with the adaptive filter coefficient of the FIR filter, removes noise, extracts useful information, and converts the signal into an analog signal through the DA converter. The steps of determining the coefficient of the adaptive filter coefficient are as follows: obtaining the desired frequency characteristic signal and the input signal; extracting the frequency characteristic component of the input signal; estimating the frequency deviation by the correlation method; estimating the FIR length N according to the frequency deviation, setting N to (1-n)*2 / 0.1. In order to avoid overfitting and underfitting, the FIR length N and the corresponding weight are determined by the frequency deviation and frequency accuracy, and adopting When the frequency deviation trend decreases, the weight is adjusted positively. When it increases, stop. The weight adjustment range is determined and fine-tuned within the weight adjustment range. The number of fine-tuning times is set to 1 / 2 of the frequency accuracy. The corresponding weights of each convolutional layer are estimated and initially set to an equal weight, that is, (1 / N) * 100%. In step 7, the input at each moment is multiplied by the weight coefficient by the multiplier, and finally the accumulator is superimposed to output the filtered signal. The additional momentum method is used to improve the weight value correction process. Specifically, a part of the weight adjustment amount of the last or previous times is superimposed on the weight adjustment amount calculated according to the current error as the actual weight adjustment amount this time.

[0043] The back-end analog filter receives the analog signal and controls the optical signal intensity within a certain range through an attenuator. The signal is then acquired through a tuner with variable impedance, and the primary side of the transformer performs frequency selection to select a specific frequency range. The secondary side of the transformer performs inductive tuning, and the signal is finally filtered through a tunable harmonic filter to remove the detected noise frequency signal. The signal is then filtered through an LC frequency selection network to output signals outside the specific frequency range.

[0044] In this way, the front-end FIR filter with adaptive filter coefficient is adopted for filtering, and the back-end adopts attenuation + variable impedance acquisition + transformer primary frequency selection and transformer secondary induction tuning, and finally the adjustable harmonic filter and LC frequency selection network are used for filtering, which can effectively filter the noise interference.

Claims

1. An adaptive filter for an optical fiber sensor, comprising a front-end digital filter and a back-end analog filter, characterized in that: The front-end digital filter receives the modulation signal of the optical fiber sensor, adopts the FIR filter adaptive filter coefficient to remove noise, extract useful information, and converts it into an analog signal through a DA converter, wherein the FIR filter internally includes an FIR controller, an adder, a multiplier, and an accumulator; The back-end analog filter receives the analog signal, controls the intensity of the optical signal within a certain range through the attenuator, then obtains the optical signal through the tuner variable impedance, performs primary frequency selection on the transformer, performs inductive tuning on the secondary side of the transformer, and finally filters the optical signal through the tunable harmonic filter and the LC frequency selection network for output. The attenuator includes an inductor L1 and a capacitor C1, one end of the inductor L1 and one end of the capacitor C1 are connected to the analog signal, the other end of the inductor L1 is connected to the other end of the capacitor C1, the right end of the diode BD1, and one end of the capacitor C3 respectively, the cathode of the diode BD1 is connected to the detection signal amplitude, and the left end of the diode BD1 is connected to one end of the grounded capacitor C2; The tuner includes a potentiometer RW1, an inductor L, and a capacitor C4. The potentiometer RW1, the inductor L, and the capacitor C4 constitute a variable impedance network. The inductor L, the capacitor C4, and the primary coil of the transformer T1 constitute a frequency selection network. High frequency is added to the primary coil of the transformer T1 through the capacitor C4, and low frequency is added to the primary coil of the transformer T1 through the inductor L. The secondary coil of the transformer T1 and the varactor diode BD2 are connected in parallel to form an inductive tuning network. The upper end of the potentiometer RW1, the upper end of the inductor L, and one end of the capacitor C4 are connected to the other end of the inductor L1. The other end of the capacitor C4 is connected to one end of the primary coil of the transformer T1. The other end of the primary coil of the transformer T1, the lower end of the potentiometer RW1, and the lower end of the inductor L are connected to ground. One end of the secondary coil of the transformer T1 is connected to the upper end of the varactor diode BD2, and the other end of the secondary coil of the transformer T1 is connected to the lower end of the varactor diode BD2. The adjustable harmonic filter receives the tuner output signal and realizes the interference noise frequency component in the notch filtering tuning signal, including a resistor R4 and an electrolytic capacitor E2, one end of the resistor R4 and the upper end of the electrolytic capacitor E2 are connected to one end of the secondary coil of the transformer T1, the negative electrode of the electrolytic capacitor E2 is respectively connected to the positive electrode of the electrolytic capacitor E3 and one end of the potentiometer RW2, the other end of the resistor R4 is connected to the positive electrode of the electrolytic capacitor E4, the negative electrode of the electrolytic capacitor E3 is respectively connected to the negative electrode of the electrolytic capacitor E4 and the other end of the potentiometer RW2, the adjustable end of the potentiometer RW2 is connected to the ground through the resistor R5, and the adjustable end of the potentiometer RW1 is also connected to the noise frequency signal, and the noise frequency signal is used to control the frequency of the notch filtering; The LC frequency selection network is composed of an inductor LN and a varactor diode BD3, and finally outputs the filtered and conditioned optical fiber sensor signal through a capacitor C5; The steps of determining the coefficients of the adaptive filter coefficients are as follows: Step 1: Obtain the desired frequency characteristic signal; Step 2, obtaining an input signal; Step 3, extracting the frequency characteristic component of the input signal; Step 4, estimate the frequency deviation; Step 5: Estimate the FIR length N based on the frequency deviation and set N to (1-n)*2 / 0.

1. To avoid overfitting and underfitting, the FIR length N and the corresponding weight are determined by the frequency deviation and frequency accuracy. The weight is adjusted positively when the frequency deviation trend decreases, and stops when it increases. The weight adjustment range is determined, and fine-tuning is performed within the weight adjustment range. The number of fine-tuning times is set to 1 / 2 of the frequency accuracy. Step 6: Estimate the corresponding weights of each convolutional layer, and initially set them to be evenly distributed, that is, (1 / N)*100%; In step 7, the input at each moment is multiplied by the weight coefficient by the multiplier, and finally the accumulator performs superposition to output the filtered signal.

2. The optical fiber sensor adaptive filter according to claim 1, wherein: It also includes the introduction of momentum factor and the use of additional momentum method to improve the weight value correction process. The specific approach is: superimpose a part of the last or previous weight adjustment amount on the weight adjustment amount calculated according to the current error as the actual weight adjustment amount this time.

3. The optical fiber sensor adaptive filter according to claim 1, wherein: The variable impedance network is realized by adjusting the resistance value of the potentiometer RW1, which is a voltage-controlled potentiometer and is reversely regulated by the negative feedback of the amplitude of the detection signal; The negative feedback reverse regulation of the amplitude of the detection signal adopts a reverse excitation circuit; The reverse excitation circuit includes a capacitor C6, a resistor R6, and an inductor L4. One end of the capacitor C6, one end of the resistor R6, and one end of the inductor L4 receive the detection signal amplitude. The other end of the capacitor C6 is connected to the ground. The other end of the resistor R6 is connected to one end of the grounded capacitor C7. The other end of the inductor L4 is respectively connected to one end of the inductor L3 and the negative electrode of the electrolytic capacitor E4. The other end of the inductor L3 is connected to the gate of the MOS transistor M1. The drain of the MOS transistor M1 is connected to one end of the inductor L2. The other end of the inductor L2 and one end of the grounded capacitor C8 are connected to the power supply -5V. The source of the MOS transistor M1 is respectively connected to the negative electrode of the diode D1, one end of the grounded resistor R7, and the normally open contact of the switch JP1. The positive electrode of the diode D1 is connected to the ground. The normally closed contact of the switch JP1 is connected to the power supply -5V. The common end of the switch JP1 is connected to the adjustable end of the potentiometer RW1.

Images