Interferometer measurement system signal pre-processing method based on zipper type filter

Abstract

The invention provides an interferometer measurement system signal pre-processing method based on a zipper type filter. The method comprises the following steps of: multiplying an input signal by signals generated by two digital frequency synthesizers respectively, obtaining two paths of frequency mixing signals, obtaining an estimated input signal frequency through a phase resolving module after the two paths of frequency mixing signals pass through a low-pass filter respectively, and adjusting a filter coefficient in real time according to the change of the input signal frequency, therefore, the signal can be filtered more accurately. Wherein the filter adopts a zipper type structure to ensure smooth transition of signals. According to the invention, the signal-to-noise ratio of the signal can be improved, the corresponding noise can be suppressed, the resource occupation can be reduced, the power consumption of the system is reduced, and the amplitude, the phase, the direct current value and the frequency of the signal cannot be suddenly changed in the frequency change range of the whole signal.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of photoelectric signal processing, and relates to a signal pre-processing method of an interferometer measurement system. BACKGROUND

[0002] With the continuous progress of ultra-precision instrument manufacturing, semiconductor processing and other industries, higher requirements for the measurement accuracy and measurement speed of the interferometer measurement system are put forward in the manufacturing process.

[0003] From the measurement principle of the interferometer displacement measurement system, it can be known that the displacement to be measured is related to the phase change of the signal, and therefore the accuracy of the phase calculation value is very important. Ideally, the obtained digital signal should be a smooth sinusoidal signal, but in actual situations, various noises are introduced. For example, internal noise of electronic devices, electromagnetic interference introduced by device pin coupling, thermal noise, shot noise and ADC pre-driving noise, etc. The introduction of these noises will cause distortion of the sinusoidal waveform and reduce the signal-to-noise ratio. With the further improvement of the measurement speed, the range of Doppler shift changes, and the range of frequency changes of the measurement signal also increases.

[0004] If a general band-pass filter is used to process the measurement signal, in order to make all possible useful signals pass through, the passband bandwidth will be very wide. At this time, in order to achieve a better filtering effect, the filter order is required to be very high, which will lead to a high resource occupation, and at the same time, the harmonic suppression effect in the frequency difference range is poor. These will increase the error of phase calculation.

[0005] The present application proposes an interferometer measurement system signal pre-processing method based on a zipper filter, the center frequency of the passband of the signal pre-processing module of which can follow the frequency change of the measurement signal, and in the case of a relatively narrow passband bandwidth, all possible useful signals can pass through, which can improve the signal-to-noise ratio while suppressing the corresponding noise, reduce the occupation of resources, reduce the power consumption of the system, and ensure that the amplitude, phase, direct current value and frequency of the signal do not change suddenly within the entire frequency change range of the signal. SUMMARY

[0006] In view of the defects of the wide passband bandwidth and insufficient signal-to-noise ratio in the prior art measurement signal pre-processing means, the main purpose of the present application is to propose an interferometer measurement system signal pre-processing means, which can improve the filtering effect while meeting the dynamic range of the input signal, and reduce the occupation of resources.

[0007] An interferometer measurement system signal pre-processing method based on a zipper filter, the method comprising the following steps:

[0008] It is composed of sine digital frequency synthesizer 1, cosine digital frequency synthesizer 2, multipliers 3 and 4, low-pass filters 5 and 6, phase resolving module 7, phase differentiator 8, coefficient adjustment module 9, time alignment module 10, convolution filter module 11;The method consists of two parts, the first part is the frequency estimation module of the input signal: the input signal is multiplied with the two fixed frequency signals generated by the sine digital frequency synthesizer 1 and the cosine digital frequency synthesizer 2 respectively, and the mixed frequency signals obtained are filtered through the low-pass filters 5 and 6 respectively, and the two orthogonal signals after filtering high frequency signals are phase-resolved through the phase-resolving module 7, and the phase-resolved phase results are obtained through the phase differentiator 8 to obtain the estimated input signal frequency information. The second part is the zipper filter module: the frequency information of the input signal is obtained through the coefficient adjustment module 9 to obtain the filter coefficient, and the unprocessed input signal is processed through the time alignment module 10 and the obtained filter coefficient through the convolution filter module 11 to obtain the improved input signal.

[0009] The expression of the output signal can be represented by (12)

[0010] (12)

[0011] Wherein represents the amplitude of the signal, is the modulation frequency, is the modulation initial phase, is the frequency uncertainty, is the Doppler shift, is the input signal amplitude, is the DC component. Digital frequency synthesizer (DDS) is a public free IP core inside FPGA, through which an arbitrary frequency sine function or cosine function can be generated. The sine and cosine expressions generated by the DDS IP core are shown in equations (13) and (14):

[0012] (13)

[0013] (14)

[0014] Wherein represents the signal amplitude, represents the signal frequency. Multiply the digital synthesized signal with the input signal to obtain the signal as shown in equations (15) and (16).

[0015]

[0016] (15)

[0017]

[0018] (16)

[0019] From the above formula, it can be seen that the mixed signal contains three types of frequencies, one is the sum frequency term, one is the difference frequency term and the frequency difference term. To obtain a signal containing only the Doppler shift term, the high frequency part needs to be filtered out, so a low-pass filter with a cutoff frequency of is designed. After filtering out the high frequency, the low frequency part formula is simplified as formula (17) and (18):

[0020] (17)

[0021] (18)

[0022] When assuming , it is further simplified as formula (19) and (20):

[0023] (19)

[0024] (20)

[0025] The orthogonal signal is passed through the CORDIC module for phase solution, and the obtained phase result is shown in formula (21):

[0026] (21)

[0027] The phase result is passed through the phase differentiator, and after differentiating the time, the estimated frequency is shown in expression (22):

[0028] (22)

[0029] At this time, the estimated frequency and the actual frequency value only differ by a fixed coefficient , and the frequency instability is also a fixed value, plus the frequency difference. Therefore, the frequency of the measurement signal can be accurately estimated.

[0030] From the design diagram of the zipper filter, after the frequency of the current signal is estimated, the filter coefficient adjustment module adjusts the filter coefficient according to the estimated frequency to meet the bandwidth of the current signal, so as to achieve a better filtering effect. The core key of coefficient adjustment is to ensure that the signal in the whole frequency change range of the signal does not have sudden changes in amplitude, phase, direct current value and frequency. Therefore, the zipper filter is designed to ensure the smooth transition of the signal.

[0031] The zipper structure diagram of the filter in the application is shown in the drawing Figure 2 . Among them and representing the low and high cut-off frequencies of the first stage filter, and representing the low and high cut-off frequencies of the second stage filter, and representing the low and high cut-off frequencies of the third stage filter, and representing the low and high cut-off frequencies of the nth stage filter. Each stage filter represents the filter coefficients of the current stage. Meanwhile, the high and low cut-off frequencies of the entire filter cross each other, showing a zipper shape. When the input signal frequency is between the high cut-off frequency of the first stage filter and the low cut-off frequency of the second stage filter, i.e. the filter coefficients are selected as follows: two new frequency values are set, as shown in equation (23).

[0032] (23)

[0033] where is at the intersection of the cut-off frequencies of the two stage filters, the first stage filter coefficients are selected when , and the second stage filter coefficients are selected when . Thus, when the coefficients are switched, the frequency bands of the two stage filters both satisfy the current frequency, ensuring smooth transition of the signal. Meanwhile, the signal amplitude decay at the cut-off frequency is effectively avoided.

[0034] In the present application, the filter type is a band-pass filter, which produces a large attenuation to high frequency signals above and below, while producing no or very small attenuation to signals with frequencies between and . Meanwhile, the design method is based on window functions, and the filter type is an FIR filter, so the windowing process is beneficial to prevent spectrum leakage. When the filter order is N, the filter tap coefficients are:

[0035] (24)

[0036] After obtaining the filter coefficients, the coefficients are of floating-point type, but signal processing inside the FPGA requires converting the floating-point type to fixed-point number operation. Therefore, the filter coefficients need to be quantized, so the filter coefficients are multiplied by , where N represents the bit width of data quantization. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 ​​It is a kind of signal pre-processing method of interferometer measurement system based on zip filter.

[0038] Figure 2 It is a kind of signal pre-processing method of interferometer measurement system based on zip filter.

[0039] Figure 3 It is an embodiment of the signal pre-processing method of interferometer measurement system based on zip filter. DETAILED DESCRIPTION

[0040] The application will be further described in combination with specific embodiments.

[0041] Figure 3 An embodiment of the signal pre-processing method of interferometer measurement system based on zip filter is given. It is composed of sine digital frequency synthesizer 12, cosine digital frequency synthesizer 13, multipliers 14 and 15, low-pass filters 16 and 17, phase resolving module 18, phase differentiator 19, coefficient adjusting module 20, time alignment module 21, convolution filter module 22, band-pass filter 23, time alignment 24, phase resolving 25; the method is composed of two parts, the first part is the frequency estimation module of input signal: the input signal produces two fixed frequency signals with sine digital frequency synthesizer 12 and cosine digital frequency synthesizer 13 respectively, multiplies through multipliers 14 and 15, gets mixed frequency signals, the two orthogonal signals after filtering out high frequency signals are phase-resolved through phase resolving module 18, and the phase-resolved phase results get the estimated input signal frequency information through phase differentiator 19. The second part is the zip filter module: the frequency information of input signal gets filter coefficients through coefficient adjusting module 20, the unprocessed input signal gets improved input signal through convolution filter module 22 after time alignment module 21 and the obtained filter coefficients. In the measurement system of interferometer, the input signal at this time is often the measurement signal. Another reference signal is suppressed through band-pass filter 23 to improve the signal quality, and then enters time alignment module 24, and gets the resolved signal containing displacement information at phase resolving module 25 after the reference signal and the measurement signal are synchronized in sampling time.

Claims

1. A signal pre-processing method for an interferometer measurement system based on a zipper filter, characterized by: The method comprises a sine digital frequency synthesizer 1, a cosine digital frequency synthesizer 2, multipliers 3 and 4, low-pass filters 5 and 6, a phase calculation module 7, a phase differentiator 8, a coefficient adjustment module 9, a time alignment module 10, and a convolution filter module 11. The method comprises two parts: the first part is a frequency estimation module for the input signal: the input signal is mixed with two fixed-frequency signals generated by the sine digital frequency synthesizer 1 and the cosine digital frequency synthesizer 2, respectively, by multipliers 3 and 4. The resulting mixed signals are passed through low-pass filters 5 and 6, respectively. After filtering out the high-frequency signals, the two orthogonal signals are phase-calculated by a phase calculation module 7. The phase results are then passed through a phase differentiator 8 to obtain the estimated input signal frequency information. The second part is a zipper filter module: the frequency information of the input signal is passed through a coefficient adjustment module 9 to obtain filter coefficients. The unprocessed input signal is passed through a time alignment module 10 and the obtained filter coefficients are then passed through a convolution filter module 11 to obtain an improved input signal.

2. The frequency estimation module for real-time measurement of input signal frequency according to claim 1, wherein: The input signal expression can be expressed as (1): (1) in represents the amplitude of the signal, is the modulation frequency, is the initial phase of modulation, is the frequency uncertainty, is the Doppler shift, is the input signal amplitude, is the DC component. The expressions of the sine and cosine functions directly generated by the digital frequency synthesizer are shown in formulas (2) and (3) (2) (3) in represents the signal amplitude, Indicates the signal frequency. Multipliers 3 and 4 multiply and mix the signal generated by the digital frequency synthesizer with the input signal, and the resulting signals are shown in formulas (4) and (5). (4) (5) The mixed signal is passed through the cutoff frequency The signals after low-pass filters 5 and 6 are shown in formulas (6) and (7): (6) (7) CORDIC module 7 performs phase calculation on the quadrature signal, and the phase result is: (8) After the phase differentiator 8 differentiates the phase result with respect to time, the obtained frequency is shown in expression (9): (9) By subtracting the frequency uncertainty from the signal frequency at this time, the frequency of the input signal can be accurately estimated.

3. The zipper filter module for adjusting filter coefficients according to claim 1, wherein: The zipper structure block diagram of the filter is shown in Figure 2 of the accompanying drawings. and Represents the low cutoff frequency and high cutoff frequency of the first stage filter, and Represents the low cutoff frequency and high cutoff frequency of the second-stage filter, and Represents the low cutoff frequency and high cutoff frequency of the third-order filter, and Represents the low cutoff frequency and high cutoff frequency of the nth filter. Each filter level represents the filter coefficient of the current level. At the same time, the high and low cutoff frequencies of the entire filter intersect with each other, showing a zipper shape. When the input signal frequency is at the high cutoff frequency of the first filter, and the low cutoff frequency of the second stage filter Between, that is When , the filter coefficients are selected as follows: Set two new frequency values, as shown in formula (10): (10) In the formula At the intersection of the cutoff frequencies of the two-stage filters, when When , the first-stage filter coefficient is selected; when When , select the second-stage filter coefficients. The filter type in this method is a bandpass filter, which is designed using a window function-based method and the filter type is an FIR filter.

Images