A feature point recognition method based on ultrasonic echo signal contour

Abstract

The present invention discloses a signal processing method for ultrasonic echoes emitted by an ultrasonic transducer, which relates to the field of flow velocity measurement by time difference method and gas ultrasonic flowmeter; what needs to be solved is the filtering processing of low signal-to-noise ratio ultrasonic echo signals and the identification of echo signal feature points, with high precision and high measurement range; the signal processing scheme includes a signal preprocessing scheme and an ultrasonic echo feature point identification method, reads the echo digital signal collected in the experiment, uses an FIR filter to preprocess the original signal, and then calculates the time difference by using a feature point identification method based on the echo signal profile. This patent can realize synchronous signal processing of signals collected under multiple channels and obtain accurate sound time data.

Description

Technical Field

[0001] The present invention relates to the field of signal processing methods for ultrasonic transducer echo signals, and in particular to a feature point recognition method based on ultrasonic echo signal profile. Background Art

[0002] The echo signal of the ultrasonic transducer is processed based on the time difference method. It is widely used in the fields of petroleum and natural gas industry, power machinery, energy metering, etc. It is used to measure the flow velocity or flow statistics of the fluid. It can be used in many industrial scenarios, such as measuring the flow rate of gas in a pipeline or realizing non-contact measurement of the flow field in the impeller blade.

[0003] The principle of time difference method, such as Figure 2 As shown in the figure, there is gas flowing in the pipeline. The ultrasonic wave emitted by the upstream transducer T1 is reflected in the flow field and received by the downstream transducer T2. This direction is defined as downstream. Similarly, the ultrasonic wave emitted by T2 is received by T1. This propagation direction is defined as upstream. The downstream / reverse propagation time of the ultrasonic wave in the fluid is:

[0004]

[0005] Where: v is the average gas flow velocity (unit: m / s), θ is the angle between the emission direction of the ultrasonic transducer and the direction of the airflow in the tube (unit: rad); L is the sound path (unit: m), and c is the propagation speed of the ultrasonic wave in the medium (unit: m / s).

[0006] The time difference between upstream and downstream is:

[0007]

[0008] Since the ultrasonic propagation velocity is much greater than the average gas flow velocity in the measurement area, that is, c 2 >>v 2 , the gas flow rate formula is simplified to:

[0009]

[0010] One of the current mainstream signal processing methods is to approximate the actual echo signal by establishing an echo signal mathematical model, and to obtain the echo propagation time by searching the model parameters through an algorithm. However, it is difficult for the current echo mathematical model to match the echo signals of all transducers. Another mainstream signal processing method is the feature point recognition method based on the echo signal energy. This method has a large range of measurable flow velocities and has a certain anti-interference ability. However, the amount of calculation of this method is relatively large, and the response speed is slow during online measurement. Summary of the invention

[0011] The present invention is made to solve the above problems, and its purpose is to provide a feature point recognition method based on ultrasonic echo signal profile with wide measurement range and high measurement accuracy. The present invention adopts the following technical solutions:

[0012] The present invention provides a feature point recognition method based on the contour of an ultrasonic echo signal, characterized in that it comprises the following steps: step S1, reading a collected ultrasonic signal file and performing signal segmentation processing on the ultrasonic file; step S2, selecting a preprocessed signal segment, that is, setting a starting point and a length of the preprocessed signal segment; step S3, inputting the preprocessed signal segment into an FIR bandpass filter for preprocessing operation, thereby improving the signal-to-noise ratio of the original echo signal; step S4, converting the original echo signal into an echo signal approximating a spindle by removing noise interference in a non-central wave frequency segment, and performing normalization processing; step S5, Step S5, obtaining the peak of the echo signal approximating the spindle, fitting the peak through a multinomial function to obtain a curve function, and calculating the maximum value of the curve change rate by taking the derivative of the curve function; Step S6, obtaining the trough of the echo signal approximating the spindle, fitting the trough through a multinomial function to obtain a curve function, and calculating the minimum value of the curve change rate; Step S7, calculating the average value based on the maximum value and the minimum value of the change rate and tracing the average value back to the echo signal, obtaining the characteristic point of the echo signal through periodic calculation, and saving the data at the characteristic point to the computer hard disk.

[0013] The feature point recognition method based on the ultrasonic echo signal profile provided by the present invention may also have such a technical feature, wherein the FIR bandpass filter in step S3 includes: the high and low cut-off frequencies are respectively set to the center frequency of the echo signal ±10kHz; the window function is set to the Hanning window; the filter order is calculated by the following formula: Where fs is the sampling frequency of the signal in kHz, n is the number of points in the transition bandwidth, and δ is the corresponding window function coefficient.

[0014] The feature point recognition method based on the ultrasonic echo signal profile provided by the present invention may also have such a technical feature, wherein the normalization processing in step S4 includes the following steps: step S4-1, selecting the signal segment to be processed in the echo signal, and setting the starting point and length of the signal segment; step S4-2, obtaining the amplitude of each sampling point within the length of the signal segment, and dividing the amplitude of each sampling point of the signal segment length by the maximum value of the absolute value of the amplitude of the sampling point in the signal segment, thereby performing normalization processing.

[0015] The feature point recognition method based on the ultrasonic echo signal profile provided by the present invention may also have such technical features, wherein step S5 also includes the following sub-steps: step S5-1, obtaining the maximum peak point of the echo signal of the approximate spindle, and taking five peaks forward based on the maximum peak point; step S5-2, fitting the six peaks through a multinomial function to obtain a curve S(x), and performing a derivative operation on the curve to calculate the maximum value of the curve S'(x), thereby obtaining the position definition point a where the rate of change of the curve S(x) is the largest.

[0016] The feature point recognition method based on the ultrasonic echo signal profile provided by the present invention may also have such a technical feature, wherein step S6 also includes the following sub-steps: step S6-1, obtaining the minimum trough point of the echo signal approximating the spindle, and taking five troughs forward based on the minimum trough point; step S6-2, fitting the six troughs through a polynomial function to obtain a curve, and performing a derivative operation on the curve to obtain a position definition point b where the curve change rate is the smallest.

[0017] The feature point recognition method based on the ultrasonic echo signal contour provided by the present invention may also have such technical features, wherein step S7 also includes the following sub-steps: step S7-1, placing the position definition point a at the maximum curve change rate and the position definition point b at the minimum curve change rate on the time axis to take the average value, and tracing back the average value to the echo signal as c; step S7-2, defining the period where point c of the echo signal is located as the starting period, and based on the starting period, taking the starting points of four consecutive periods thereafter, calculating the average value of the four period starting points on the time axis, thereby obtaining the calculated feature points of the echo signal.

[0018] Function and Effect of the Invention

[0019] According to the feature point recognition method based on the ultrasonic echo signal profile of the present invention, due to the use of the time difference method and the improved method based on the echo profile algorithm, by finding a cycle with the largest amplitude increase in the rising segment of the echo, and determining the feature points through this cycle, it mainly includes a signal preprocessing scheme and an ultrasonic echo feature point recognition method, reading the echo digital signal collected in the experiment, using the FIR filter to preprocess the original signal, and then calculating the time difference by using the feature point recognition method based on the echo signal profile. The method of the present invention has a wide range of flow velocity measurement and low calculation time, and can be widely applicable in both offline and online measurements. The present invention can quickly calculate the time difference between ultrasonic forward and reverse flow at different flow rates; the present invention can also realize synchronous signal processing of signals collected under multiple channels and obtain accurate sound time data. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a schematic flow chart of a feature point recognition method based on an ultrasonic echo signal profile in an embodiment of the present invention;

[0021] Figure 2 Schematic diagram of the principle of the time difference method according to an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of comparison of echo signals under different flow velocities according to an embodiment of the present invention;

[0023] Figure 4 is a schematic diagram of an online measurement hardware block diagram of an embodiment of the present invention;

[0024] Figure 5 Schematic diagram of filtering effect of an embodiment of the present invention;

[0025] Figure 6 It is a schematic diagram for explaining the signal processing method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the feature point recognition method based on the ultrasonic echo signal profile of the present invention is specifically described below in combination with embodiments and drawings.

[0027] <Example>

[0028] The signal processing method of this embodiment is a method summarized by verifying and comparing the commonalities of various echoes under different flow velocities. Figure 3 As shown in the figure, the echo signals of flow velocity from 0-40m / s were collected experimentally. The collected digital echo signals were filtered and it was observed that the overall shape of the echo signals under different flow velocities after filtering was similar to a sine wave with amplitude modulated, and its peak value gradually increased and then gradually decreased to zero. And in the rising stage of the echo, the change trend showed a high degree of consistency. The zero-crossing point where the signal amplitude changes from negative to positive is defined as the starting point of the signal cycle. A method is designed to determine the position of the characteristic point of the echo signal on the time axis by finding the period with the largest change in the amplitude of the echo signal in the rising stage of the echo signal, and then searching for the zero-crossing point backwards.

[0029] Figure 1 It is a flow chart of a feature point recognition method based on ultrasonic echo signal profile in an embodiment of the present invention.

[0030] like Figure 1 As shown, the feature point recognition method based on the ultrasonic echo signal profile includes the following steps:

[0031] Step S1, reading the collected ultrasonic signal file and performing signal segmentation processing on the ultrasonic signal file;

[0032] Step S2, selecting a preprocessed signal segment, i.e. setting a starting point and a length of the preprocessed signal segment;

[0033] In this embodiment, the starting point of the preprocessed signal segment is set to the 2000th point, and the length is set to 800 sampling points.

[0034] Step S3, inputting the preprocessed signal segment into a FIR bandpass filter for preprocessing operation, thereby improving the signal-to-noise ratio of the original echo signal;

[0035] Step S4, converting the original echo signal into an echo signal approximating a spindle by removing noise interference in the non-central wave frequency band, and performing normalization processing;

[0036] Figure 5 Schematic diagram of filtering effect of an embodiment of the present invention;

[0037] In this embodiment, the FIR bandpass filter includes parameters such as high and low cutoff frequencies, a window function, and a filter order Nc; the high and low cutoff frequencies of the FIR bandpass filter are respectively set to the center frequency ±10kHz of the echo signal, the window function is selected as a Hanning window, and the number of taps (i.e., the filter order) is calculated by the following formula:

[0038]

[0039] Where fs is the sampling frequency of the signal, the unit is kHz, n is the number of points in the transition bandwidth, δ is the corresponding window function coefficient, in this example, 50 points in the transition bandwidth are taken, and the coefficient corresponding to the Hanning window is 6.2; specifically, the signal-to-noise ratio of the echo signal is first enhanced by filtering, and the original echo signal is converted into an echo signal approximating the spindle, such as Figure 5 As shown; then select the starting point and length of the signal segment to be processed, divide the amplitude of each sampling point within the length of the signal by the maximum absolute value of the amplitude of the sampling point in the signal, and perform normalization processing for subsequent signal processing.

[0040] Figure 6 is a schematic diagram illustrating a signal processing method according to an embodiment of the present invention;

[0041] Step S5, obtaining the peak of the echo signal approximating the spindle, fitting the peak through a multi-function to obtain a curve function, and taking the derivative of the curve function to calculate the maximum value of the curve change rate;

[0042] In this embodiment, step S5 also includes the following sub-steps: step S5-1, obtaining the maximum peak point of the echo signal of the approximate spindle, and taking five peaks forward based on the maximum peak point; step S5-2, fitting the six peaks through a multinomial function to obtain a curve S(x), and performing a derivative operation on the curve to calculate the maximum value of the curve S'(x), thereby obtaining a position definition point a at which the rate of change of the curve S(x) is the largest;

[0043] Specifically, find the maximum peak point of this echo signal, and take 5 peaks forward to ensure that enough contour information of the echo rising segment can be obtained. Fit the 6 peak positions through a polynomial function to obtain the curve S(x)=3.67051-0.0309642x+ 0.0000836237Ex 2 -0.00000006.57956x 3 , and calculate the first derivative of the fitted curve function S(x) to obtain the first derivative function of the curve S'(x) = -0.0309642 + 0.000167247x - 0.000000197387x 2 , and find the maximum value of S'(x), that is, the position where the fitting curve function S(x) has the largest rate of change is defined as point a(423, -0.404), such as Figure 6 shown.

[0044] Step S6, obtaining the trough of the echo signal approximating the spindle, fitting the trough by a multi-functional method to obtain a curve function, and calculating the minimum value of the curve change rate;

[0045] In this embodiment, step S6 also includes the following sub-steps: step S6-1, obtaining the minimum trough point of the echo signal of the approximate spindle, and taking five troughs forward based on the minimum trough point; step S6-2, fitting the six troughs through a polynomial function to obtain a curve, and performing a derivative operation on the curve to obtain a position definition point b where the curve change rate is the smallest.

[0046] Specifically, find the minimum trough point of the echo signal, and take 5 troughs forward, fit these 6 trough positions through a cubic polynomial, and find the point where the curve has the minimum rate of change b (416, -0.518), as shown in Figure 6 Take the average value of a and b on the time axis, and then trace it back to the echo signal and record it as c(419,-0.462), as shown in Figure 6 shown.

[0047] Step S7, based on the maximum value and the minimum value of the rate of change, calculate the average value and trace the average value back to the echo signal, obtain the characteristic point of the echo signal through periodic calculation, and save the data at the characteristic point to the computer hard disk.

[0048] In this embodiment, since the overall shape of the echo signal after filtering is similar to an amplitude modulated sine wave, its peak value gradually increases and then gradually decreases to zero. The zero-crossing point where the signal amplitude changes from negative to positive is defined as the starting point of the signal cycle. Specifically, in order to improve the stability of the feature point, the period where point c is located is defined as the starting period, and the starting points of four consecutive periods are taken after the starting period. The average value (504.83, 0) of these four starting points on the time axis is determined as the feature point of the echo signal, and this point is recorded as the feature point of the ultrasonic echo calculation, such as Figure 6 As shown, the acquired characteristic point sound time data is saved in the computer hard disk.

[0049] Example Function and Effect

[0050] According to the feature point recognition method based on the ultrasonic echo signal profile of the present invention, due to the use of the time difference method and the improved method based on the echo profile algorithm, by finding a cycle with the largest amplitude increase in the rising segment of the echo, and determining the feature points through this cycle, it mainly includes a signal preprocessing scheme and an ultrasonic echo feature point recognition method, reading the echo digital signal collected in the experiment, using the FIR filter to preprocess the original signal, and then calculating the time difference by using the feature point recognition method based on the echo signal profile. The method of the present invention has a wide range of flow velocity measurement and low calculation time, and can be widely applicable in both offline and online measurements. The present invention can quickly calculate the time difference between ultrasonic forward and reverse flow at different flow rates; the present invention can also realize synchronous signal processing of signals collected under multiple channels and obtain accurate sound time data.

[0051] The above embodiments are only used to illustrate specific implementation modes of the present invention, and the present invention is not limited to the description scope of the above embodiments.

Claims

1. A signal processing method for ultrasonic echoes emitted by an ultrasonic transducer, for analyzing the contour of an ultrasonic echo with good resonance characteristics, identifying feature points, and calculating the time difference of the forward and reverse flow of the ultrasonic echo signal, characterized in that: The following steps are involved: Step S1, reading the collected ultrasonic signal file and performing signal segmentation processing on the ultrasonic signal file; Step S2, selecting a preprocessed signal segment, i.e. setting a starting point and a length of the preprocessed signal segment; Step S3, inputting the preprocessed signal segment into a FIR bandpass filter for preprocessing operation, thereby improving the signal-to-noise ratio of the original echo signal; Step S4, converting the original echo signal into a spindle-shaped echo signal by removing noise interference in the non-central wave frequency band, and performing normalization processing; Step S5, obtaining the peak of the echo signal, fitting the peak through a multi-function to obtain a curve function, and taking the derivative of the curve function to calculate the maximum value of the curve change rate; Step S6, obtaining the trough of the echo signal, fitting the trough by a multi-function, obtaining a curve function, and calculating the minimum value of the curve change rate; Step S7, based on the maximum value of the rate of change and the minimum value of the rate of change, calculate the average value and trace the average value back to the echo signal, obtain the characteristic point of the echo signal through periodic calculation, and save the data at the characteristic point to the computer hard disk, Wherein, step S5 includes the following sub-steps: Step S5-1, obtaining the maximum peak point of the echo signal, and taking five peaks forward based on the maximum peak point; Step S5-2, fitting the six peaks through a multinomial function to obtain a curve S(x), and performing a derivative operation on the curve to obtain a curve S ′ (x), thereby obtaining the position definition point a where the rate of change of the curve S(x) is the largest. Step S6 includes the following sub-steps: Step S6-1, obtaining the minimum trough point of the echo signal, and taking five troughs forward based on the minimum trough point; Step S6-2, fitting the six troughs through a multinomial function to obtain a curve, and performing a derivative operation on the curve to obtain a position definition point b where the curve has the minimum rate of change. Step S7 includes the following sub-steps: Step S7-1, placing the position definition point a where the curve change rate is the largest and the position definition point b where the curve change rate is the smallest on the time axis to obtain an average value, and tracing the average value back to the echo signal as c; Step S7-2, defines the period where the echo signal point c is located as the starting period, and based on the starting period, takes the starting points of four consecutive periods, calculates the average value of the starting points of the four periods on the time axis, thereby obtaining the calculated characteristic points of the echo signal.

2. The signal processing method for ultrasonic echo emitted by an ultrasonic transducer according to claim 1, characterized in that: in, The FIR bandpass filter in step S3 includes: The high and low cut-off frequencies are respectively set to the center frequency of the echo signal ±10kHz; The window function is set to Hanning window; The filter order Nc is calculated by the following formula: Where fs is the sampling frequency of the signal in kHz, n is the number of points in the transition bandwidth, and δ is the corresponding window function coefficient.

3. The signal processing method for ultrasonic echo emitted by an ultrasonic transducer according to claim 1, Features: Wherein, step S4 also includes the following sub-steps: Step S4-1, selecting a signal segment to be processed in the echo signal, and setting a starting point and length of the signal segment; Step S4-2, obtaining the amplitude of each sampling point within the signal segment length, and performing normalization processing by dividing the amplitude of each sampling point within the signal segment length by the maximum absolute value of the amplitude of the sampling point within the signal segment.

Images