Self-adaptive detection method for signal spectrum peak value

Abstract

The invention relates to the technical field of signal processing, in particular to a self-adaptive detection method for a signal spectrum peak value, which comprises the following steps: acquiring a time domain-amplitude input signal; converting the time domain-amplitude signal into a frequency-amplitude signal, performing histogram division, and calculating the occurrence frequency of the amplitude value in each interval; constructing a frequency sequence, searching a potential peak value position by using the frequency sequence, and searching a peak value candidate region through the potential peak value position; and screening the candidate peak points by using the difference value of the discrete frequencies of the candidate peak points and the amplitude values of the candidate peak points. The method solves the problems that an existing peak value detection method is weak in sensitivity to frequency resolution, and effective peak values are wrongly detected.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to an adaptive detection method for a signal spectrum peak value. Background Art

[0002] Peak detection technology is a technology used to measure the maximum value (positive peak) or minimum value (negative peak) in a signal. It has a wide range of application scenarios and covers multiple fields. It can be applied to equipment fault diagnosis, status monitoring and signal analysis in the fields of electricity, vibration, acoustics, etc.

[0003] The peak detection algorithm in the existing MATLAB signal processing toolbox is highly dependent on manually set thresholds, has difficulty in handling noise, and is less sensitive to frequency resolution, which leads to problems such as incorrect detection of valid peaks. Summary of the invention

[0004] In view of the shortcomings of the existing methods, the present invention solves the problem that the existing peak detection method has weak sensitivity to frequency resolution and has the problem of erroneous detection of effective peaks.

[0005] The technical solution adopted by the present invention is: an adaptive detection method for signal spectrum peak value comprises the following steps:

[0006] Step 1: Collect time domain-amplitude input signal;

[0007] As a preferred embodiment of the present invention, a vibration sensor is used to collect input signals.

[0008] Step 2: Convert the time domain-amplitude signal into a frequency-amplitude signal, perform histogram division, and calculate the frequency of the amplitude value in each interval;

[0009] As a preferred implementation of the present invention, the FFT algorithm is used to convert the time-domain-amplitude signal into a frequency-amplitude signal.

[0010] As a preferred implementation of the present invention, the histCounts function is used to perform frequency calculation.

[0011] Step 3: construct a frequency sequence, use the frequency sequence to find the potential peak position, and then use the potential peak position to find the peak candidate area;

[0012] As a preferred embodiment of the present invention, step three specifically includes:

[0013] Step 31: Search the frequency sequence for all frequency index numbers whose frequency values ​​do not exceed the first threshold;

[0014] Step 32: When the frequency values ​​after a certain frequency index number do not exceed the first threshold, the region formed by all frequency index numbers after the corresponding frequency index number is a peak candidate region;

[0015] Step 33: Taking the minimum amplitude value of all frequency indexes in the peak candidate area as the second threshold, and taking the peak points corresponding to the frequency indexes exceeding the second threshold as candidate peak points.

[0016] Step 4: Screen the candidate peak points using the difference of the discrete frequencies of the candidate peak points and the amplitude values ​​of the candidate peak points;

[0017] As a preferred embodiment of the present invention, step 4 specifically includes:

[0018] Step 41, performing difference calculation on adjacent values ​​in the discrete frequency sequence of the candidate peak point to obtain a discrete frequency difference sequence;

[0019] As a preferred implementation of the present invention, the function for difference calculation is adjacentDiffs.

[0020] Step 42, obtaining the element numbers of a plurality of pairs of discrete frequency sequences corresponding to when the value of the discrete frequency difference sequence is equal to the third threshold;

[0021] Step 43: retain the maximum amplitude value corresponding to the element number of the discrete frequency sequence in the pair.

[0022] As a preferred embodiment of the present invention, an adaptive detection system for a signal spectrum peak value includes: a memory for storing instructions executable by a processor; and a processor for executing the instructions to implement an adaptive detection method for a signal spectrum peak value.

[0023] As a preferred embodiment of the present invention, a computer readable medium stores a computer program code, and when the computer program code is executed by a processor, an adaptive detection method for a signal spectrum peak is implemented.

[0024] Beneficial effects of the present invention:

[0025] 1. The present invention uses a frequency sequence and a first threshold to search for a peak candidate area, and then uses an adaptive second threshold to screen the candidate peak points in the peak candidate area to reduce manual intervention; and then improves the accuracy of peak detection by eliminating small adjacent value noise, which is suitable for spectrum analysis scenarios of different signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flow chart of an adaptive detection method for a signal spectrum peak value of the present invention;

[0027] Figure 2is the amplitude and frequency histogram of the present invention;

[0028] Figure 3 This is a schematic diagram of the present invention in which the first threshold is used to screen potential peak positions and the second threshold is used to screen peak candidate regions;

[0029] Figure 4 It is a partial enlargement of the schematic diagram of the peak candidate area screening using the first and second thresholds of the present invention;

[0030] Figure 5 is a schematic diagram of candidate peak points of the present invention;

[0031] Figure 6 is a schematic diagram of a peak point of the third threshold value of the present invention;

[0032] Figure 7 It is a local enlargement of the peak point schematic diagram of the third threshold value of the present invention. DETAILED DESCRIPTION

[0033] The present invention is further described below in conjunction with the accompanying drawings and embodiments. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore it only shows the components related to the present invention.

[0034] like Figure 1 As shown, a method for adaptively detecting a signal spectrum peak comprises the following steps:

[0035] Step 1: Collect time domain-amplitude input signal;

[0036] The input signal includes: a time domain-amplitude signal with a fixed frequency generated when a structure vibrates, such as a bridge structure, a building structure, etc.; and a vibration sensor is used to collect the input signal.

[0037] Step 2: Obtain the corresponding frequency-amplitude signal by using the FFT algorithm on the time domain-amplitude signal, divide the frequency-amplitude into histograms, and calculate the frequency of the amplitude value in each interval;

[0038] FFT is the fast Fourier transform algorithm; the amplitude of the input signal is divided into histograms, and the number of division intervals num_bins is defined, which is usually set to 200 to ensure that the signal is divided in detail; the frequency counts and interval boundary edges of each interval are calculated through the histCounts function.

[0039] like Figure 2 As shown, a histogram of the frequency corresponding to the amplitude value is obtained according to different amplitude intervals;

[0040] Step 3: construct a frequency sequence, use the frequency sequence to find the potential peak position, and then use the potential peak position to find the peak candidate area;

[0041] Step 31: Search the frequency sequence for all frequency index numbers whose frequency values ​​do not exceed the first threshold;

[0042] For example, the frequency series counts i =[1,2,5,7,1,2,2,3], where i is the frequency index number, find the potential peak area whose frequency is less than the first threshold;

[0043] Step 32: When the frequency values ​​after a certain frequency index number do not exceed the first threshold, the region formed by all frequency index numbers after the corresponding frequency index number is a peak candidate region;

[0044] Determine whether each element in the potential peak area has a frequency greater than the first threshold 2. It should be noted that the last element may be greater than 2, but since it must have a peak, it is included in the peak candidate area by default. The frequencies of index numbers 1 and 2 are count1=1 and count2=2, respectively, while count3 and count4 are 5 and 7, respectively, which are greater than 2, and are not valid positions. Count5=1, count6=2, and count7=2 are all less than 2, which meet the conditions. Count8 is in the candidate area by default. Therefore, the area composed of indexes 5 to 8 is the peak candidate area.

[0045] Step 33: taking the minimum amplitude value of all frequency indexes in the peak candidate area as the second threshold, and taking the peak point corresponding to the frequency index exceeding the second threshold as the candidate peak point;

[0046] Select the minimum amplitude value of each index number in the peak candidate area as the second threshold; that is, use the find function to find all positions that are not less than the second threshold, extract the frequency-amplitude corresponding to these positions, and preliminarily screen out a set of peak positions x_peaks and a corresponding set of amplitude values ​​y_peaks; Figure 3 In the figure, the green line is the potential peak area with the first threshold ≤ 2. According to the second threshold of the red dotted line, the peak candidate area is found, that is, the right side of the red dotted line is the peak candidate area; Figure 5 are all candidate peak points in the peak candidate area; the first threshold can be customized and set, and in this embodiment, the first threshold is 2.

[0047] In view of the fact that there may be multiple adjacent peak points among the candidate peak points, resulting in inaccurate peak extraction, the adjacent points in the same peak area are screened by calculating the difference between adjacent points, and the point with the largest peak value is retained;

[0048] Step 4: Screen the candidate peak points using the difference of the discrete frequencies of the candidate peak points and the amplitude values ​​of the candidate peak points;

[0049] Step 41: taking the minimum amplitude value of all frequency indexes in the peak candidate area as the second threshold, and taking the peak points corresponding to the frequency indexes exceeding the second threshold as candidate peak points;

[0050] Step 42, obtaining the element numbers of a plurality of pairs of discrete frequency sequences corresponding to when the value of the discrete frequency difference sequence is equal to the third threshold;

[0051] Step 43, retaining the maximum amplitude value corresponding to the element number of the discrete frequency sequence in the pair;

[0052] For example, assuming that the frequency resolution is 1, that is, the third threshold delta_x=1, delta_x is a custom parameter; according to the second threshold, all amplitude value sequences y_peaks and their corresponding discrete frequency sequences x_peaks are found as follows:

[0053] x_peaks=[1,2,6,8,9,20,30,40], y_peaks=[0.5,0.15,0.12,0.11,0.13,0.1,0.1,0.1];

[0054] Through the function adjacentDiffs = x_peaks j+1 -x_peaks j Calculate the difference between adjacent peak points, j is the element number in x_peaks; get the discrete frequency difference sequence Dis = [1,4,2,1,11,10,10]; among them, adjacentDiffs = delta_x calculated for the two pairs of positions 1 and 4, so the discrete frequency horizontal distance between x_peaks1 and x_peaks2, x_peaks4 and x_peaks5 is 1, which meets the adjacent condition; compare the corresponding amplitude values ​​y_peaks j , retain the maximum values ​​y_peaks1 and y_peaks5, that is, 0.5 is the maximum value compared with 0.15, and 0.13 is the maximum value compared with 0.11; and delete y_peaks2 and y_peaks4, thereby reducing the influence of false detection and noise on the detection results.

[0055] Due to the influence of frequency domain resolution, multiple adjacent peak points may be detected in the same peak candidate area; calculate the difference adjacentDiffs between adjacent peak points, and filter out adjacent points whose difference is equal to delta_x; by comparing the amplitude values ​​of adjacent points, delete the smaller points and retain the largest point as the final peak result.

[0056] like Figure 6 As shown, the frequency x_peaks and the corresponding amplitude value y_peaks of the peak point after filtering, Figure 7 It corresponds to Figure 6 A partial enlarged view of .

[0057] The present invention can adaptively determine the second threshold value to reduce manual intervention; at the same time, improve the accuracy of peak detection by eliminating small adjacent value noise, which is suitable for spectrum analysis scenarios of different signals; and use the third threshold value to screen adjacent peak points, retaining the largest point as the final peak result.

[0058] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An adaptive detection method for a signal spectrum peak, characterized in that: The following steps are involved: Step 1: Collect time domain-amplitude input signal; Step 2: Convert the time domain-amplitude signal into a frequency-amplitude signal, perform histogram division, and calculate the frequency of the amplitude value in each interval; Step 3: construct a frequency sequence, use the frequency sequence to find the potential peak position, and then use the potential peak position to find the peak candidate area; Step 4: Screen the candidate peak points using the difference between the discrete frequencies of the candidate peak points and the amplitude values ​​of the candidate peak points.

2. The method for adaptively detecting a signal spectrum peak according to claim 1, characterized in that: Step three specifically includes: Step 31: Search the frequency sequence for all frequency index numbers whose frequency values ​​do not exceed the first threshold; Step 32: When the frequency values ​​after a certain frequency index number do not exceed the first threshold, the region formed by all frequency index numbers after the corresponding frequency index number is a peak candidate region; Step 33: Taking the minimum amplitude value of all frequency indexes in the peak candidate area as the second threshold, and taking the peak points corresponding to the frequency indexes exceeding the second threshold as candidate peak points.

3. The adaptive detection method for signal spectrum peak according to claim 1, characterized in that: Step 4 specifically includes: Step 41, performing difference calculation on adjacent values ​​in the discrete frequency sequence of the candidate peak point to obtain a discrete frequency difference sequence; Step 42, obtaining the element numbers of a plurality of pairs of discrete frequency sequences corresponding to when the value of the discrete frequency difference sequence is equal to the third threshold; Step 43: retain the maximum amplitude value corresponding to the element number of the discrete frequency sequence in the pair.

4. The method for adaptively detecting a signal spectrum peak according to claim 3, characterized in that: The function for difference calculation is adjacentDiffs.

5. The method for adaptively detecting a signal spectrum peak according to claim 1, characterized in that: The input signal is collected using a vibration sensor.

6. The method for adaptively detecting a peak value of a signal spectrum according to claim 1, characterized in that: The FFT algorithm is used to convert the time-domain-amplitude signal into a frequency-amplitude signal.

7. The method for adaptively detecting a signal spectrum peak according to claim 1, characterized in that: The histCounts function is used for frequency calculation.

8. An adaptive detection system for signal spectrum peaks, characterized in that include: a memory for storing instructions executable by a processor; A processor, configured to execute instructions to implement the adaptive detection method for a signal spectrum peak as described in any one of claims 1 to 7.

9. A computer readable medium storing a computer program code, characterized in that: When the computer program code is executed by a processor, the computer program code implements the adaptive detection method for signal spectrum peak according to any one of claims 1 to 7.

Images