Method for detecting metal redundancy in thin slurry premixing process based on acoustic emission signal analysis

Abstract

The invention relates to a method for detecting metal redundancy in a thin slurry premixing process based on acoustic emission signal analysis, which comprises the following steps of: extracting an acoustic emission signal generated by collision and friction of the metal redundancy in the thin slurry premixing process as a standard acoustic emission signal template; the method comprises the following steps: windowing and segmenting acoustic emission signals generated in an online acquisition thin slurry premixing process, and performing cross-correlation analysis and template matching with a standard acoustic emission signal template; and comparing template matching result energy with background noise energy, and discovering and extracting an acoustic emission event contained in an acoustic emission signal based on wavelet packet noise reduction and energy analysis so as to realize online detection of the metal redundancy in the thin slurry premixing process. According to the method, the acoustic emission event can be rapidly, efficiently and accurately extracted from the acoustic emission signal containing noise interference, real-time and online detection of the thin slurry metal redundancy is achieved, the safety of the thin slurry premixing process is improved, the thin slurry preparation efficiency is improved, and the purposes of safety and efficiency improvement are achieved.

Description

Technical Field

[0001] This invention belongs to the field of production process status monitoring and metal foreign matter detection, specifically relating to a method for detecting metal foreign matter in a slurry premixing process based on acoustic emission signal analysis. Background Technology

[0002] The slurry premixing process is a key step in the solid propellant manufacturing process. It requires premixing, stirring, and homogenizing various components according to a specific formula ratio, mainly accomplished by a premixing device. During the slurry premixing process, there is a risk of foreign metal contaminants such as bolts, nuts, ball bearings, needle rollers, and iron pins entering the slurry. Since slurry is a high-energy material, the contamination of foreign metal contaminants can lead to product quality problems and even cause safety accidents such as combustion and explosion.

[0003] Slurry is opaque and contains metallic media (with a relatively high specific gravity). Metal powder agglomeration occurs in the early stage of premixing, and the premixing device has a warm water jacket in the pot wall, making it difficult to detect metal foreign matter in the premixing device. At present, active detection methods such as ultrasound, laser, and radar cannot be applied to the detection of metal foreign matter in high-energy slurry, and there is no effective means to detect metal foreign matter in slurry.

[0004] During the slurry premixing process, excess metal particles inside the premixing device collide and rub against the impeller and the pot wall, generating acoustic emission events. Acoustic emission technology collects and analyzes the acoustic emission signals generated during the slurry premixing process, extracts the acoustic emission events, and determines whether excess metal particles are present.

[0005] However, there are no known cases of applying acoustic emission technology to the detection of metal foreign matter in slurry. The key issue is that the noise signals generated during the slurry premixing process are characterized by high acquisition frequency and large signal processing requirements. Conventional noise signal processing methods, including time-domain analysis and spectral analysis, cannot meet the detection requirements of acoustic emission signals for metal foreign matter in slurry. Acoustic emission signal analysis needs to address issues such as the accuracy of acoustic emission signal extraction in high-noise environments, the effectiveness of algorithms, and the rapid response of the detection process. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention aims to provide a method for detecting metallic foreign matter that can rapidly, efficiently, and accurately extract acoustic emission events from acoustic emission signals containing noise interference, thereby enabling real-time, online detection of metallic foreign matter in the slurry premixing process.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] 1) Attach a high-frequency, high-sensitivity acoustic emission sensor to the outside of the pot wall of the slurry premixing device to collect acoustic emission signals generated during the slurry premixing process in real time and extract acoustic emission events caused by excess metal in real time.

[0009] 2) During the extraction of slurry premixing, the acoustic emission signals generated by the collision and friction of metal foreign matter are used as standard acoustic emission signal templates. By windowing and segmenting the online acquired acoustic emission signals, cross-correlation analysis and template matching are performed with the standard acoustic emission signal templates. The energy of the template matching results is compared with the energy of the background noise. Based on wavelet packet denoising and energy analysis, acoustic emission events contained in the noise signals are discovered and extracted, and metal foreign matter in the slurry is detected online.

[0010] The acoustic emission sensor used in step 1) is a Japanese Fuji 1045S, with a response frequency of 50Hz to 1300kHz.

[0011] Step 2) includes the following steps:

[0012] 2.1) Under experimental conditions, during the slurry premixing process, acoustic emission signals generated by the collision and friction of excess metal were extracted. The time series of the acoustic emission signals is x(n), n=1,2,...N;

[0013] 2.2) Search for a signal segment with exponential decay characteristics from the acoustic emission signal x(n), extract the signal segment and use it as the acoustic emission signal template Tem(k), 1≤k≤T, where T is the length of the template;

[0014] 2.3) Under experimental conditions, during the slurry premixing process, acoustic emission signals without metal residues were extracted. The time series of the acoustic emission signals was l0(i), i = 1, 2, ... T, where T is the length of the template.

[0015] 2.4) The time series y(m) of the acoustic emission signal acquired online is truncated using a rectangular window function ω(j), where y(m) is a segment of the acoustic emission signal acquired online. The window function is expressed as:

[0016]

[0017] The window function ω(j) performs M truncation operations on the acoustic emission signal time series y(m) with a sliding step size of 99T. y(m) is then divided into truncated signal sequences Y, where Y = [y1, y2, ..., y]. m ], 1≤m≤M;

[0018] 2.5), for a certain segment y in the truncated signal sequence Y m Perform cross-correlation operation with the acoustic emission signal template Tem(k):

[0019]

[0020] In the formula, d represents the vector subscript, q represents the offset of the acoustic emission signal template relative to the starting point of the truncated signal, 1≤q≤99T. The above cross-correlation operation is applied sequentially to each segment of the truncated signal sequence Y, and the template matching results of all segments are concatenated into R=[r1,r2,...,r m ], 1≤m≤M;

[0021] 2.6), Calculate the energy of the background noise U(0):

[0022]

[0023] In the formula, l0(i) is the acoustic emission signal without any metal foreign matter;

[0024] 2.7) Determine whether an acoustic emission event has occurred based on the energy of the template matching result. The specific steps are as follows:

[0025] 2.7.1) A rectangular window function of length T is used to slide and truncate the template matching result R. The sliding step size is an integer between 1 and 0.2T, which is taken as 1 here. Then, the energy of each segment of the matching result is calculated separately.

[0026]

[0027] 2.7.2), if U(p)≥αU(0), α≥5, and 5 is taken here, then it is considered that an acoustic emission event has occurred at this time, and p is the offset of the acoustic emission event from the start time of signal acquisition;

[0028] 2.7.3) When 3U(0)≤U(p)≤5U(0), perform wavelet packet denoising on the signal segment, and repeat step 2.6.1) on the denoised signal. If an acoustic emission event occurs, record p.

[0029] 2.8) Determine the time of acoustic emission event t = p / f based on the on-site sampling frequency. s f s At the sampling frequency, the detection of acoustic emission events proved the presence of metallic impurities within the slurry.

[0030] This invention is based on an improved template matching and noise reduction noise signal analysis algorithm. It collects and analyzes noise signals in the premixing process online, extracts acoustic emission events generated by collisions and friction of metal foreign objects, and detects metal foreign objects in the slurry premixing process online. It can quickly, efficiently, and accurately extract acoustic emission events from acoustic emission signals containing noise interference, realize real-time online detection of metal foreign objects in the slurry premixing process, remove metal foreign objects in a timely manner, ensure the safety of the slurry preparation process, improve the safety of slurry premixing operation, and improve the slurry preparation efficiency, thereby achieving the goal of safety and efficiency improvement.

[0031] The beneficial effects of this invention are as follows:

[0032] This invention detects excess metals by extracting acoustic emission events generated during the premixing of a slurry, enabling in-process detection of excess metals within the slurry; it provides non-destructive testing for excess metals in the slurry. The invention employs an acoustic emission sensor, which does not actively emit energy, thus meeting the safety requirements for energetic material mixing processes.

[0033] This invention extracts acoustic emission events using a time-domain template matching method, balancing the real-time performance and accuracy of the algorithm. By applying appropriate windowing, the template matching algorithm can reduce noise levels while maintaining low time complexity, extracting acoustic emission events from signals with low noise interference. For signals with high noise interference that are difficult to extract, wavelet packet denoising is used for secondary analysis, which can better distinguish high-frequency details of the signal.

[0034] This invention solves the problem that acoustic emission events at the signal boundary are easily missed after signal truncation by using a sliding step size smaller than the length of the rectangular window, thereby preventing the detection of unwanted metal objects. Attached Figure Description

[0035] Figure 1 This is a flowchart of a method for detecting metal foreign matter in a slurry premixing process based on acoustic emission signal analysis.

[0036] Figure 2 This is a simplified diagram of the acoustic emission metal foreign matter detection experimental device of Embodiment 1 of the present invention.

[0037] Figure 3 This is a time-domain waveform diagram of the acoustic emission signal in Embodiment 1 of the present invention.

[0038] Figure 4 This is a time-domain waveform diagram of the acoustic emission template in Embodiment 1 of the present invention.

[0039] Figure 5 This is the acoustic emission template spectrum diagram of Embodiment 1 of the present invention.

[0040] Figure 6 It is the acoustic emission signal after template matching in Embodiment 1 of the present invention.

[0041] Figure 7 This is Embodiment 1 of the present invention. Figure 4 The acoustic emission signal after wavelet packet denoising. Detailed Implementation

[0042] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0043] like Figure 1As shown, a method for detecting metal foreign matter in a slurry premixing device based on acoustic emission signal analysis is proposed. The method extracts acoustic emission signals generated by the collision and friction of metal foreign matter during the slurry premixing process as a standard acoustic emission signal template. A high-frequency, high-sensitivity acoustic emission sensor is attached to the outer wall of the slurry premixing device to collect acoustic emission signals online, including acoustic emission events caused by the movement of metal foreign matter and slurry premixing noise. The collected acoustic emission signals are windowed and segmented, and cross-correlation analysis and template matching are performed with the standard acoustic emission signal template. The energy of the template matching result is compared with the energy of the background noise. Based on wavelet packet denoising and energy analysis, the acoustic emission events contained within the signals are extracted to achieve metal foreign matter detection.

[0044] Reference Figure 2 This implementation case is based on a slurry premixing experimental platform for testing. The platform is based on... Figure 1 Signal processing flow setup: The premixing device consists of a stainless steel tank, a motor, and stainless steel blades. In this embodiment, both the stainless steel tank and blades are made of 304 stainless steel. Spherical metal impurities are selected and are also made of 304 stainless steel. An acoustic emission sensor is attached to the pot wall to detect metal impurities during the slurry premixing process.

[0045] The metal foreign matter detection platform for the slurry premixing process based on acoustic emission signal analysis involved in this implementation case can be used to test the slurry metal foreign matter detection method based on acoustic emission signal analysis. The specific test steps are as follows:

[0046] Step 1: Under experimental conditions, during the slurry premixing process, the acoustic emission signals generated by the collision and friction of excess metal were extracted. The time series of the acoustic emission signals is x(n), n=1,2,...N; (Refer to...) Figure 5 At 50000Hz, the frequency components of the acoustic emission signal are almost zero. According to the sampling theorem, the sampling frequency f is set in this example. s =100000Hz, time series length N = 1982000;

[0047] Step 2: Search for a signal segment with exponential decay characteristics from the acoustic emission signal x(n), extract the signal segment and use it as the acoustic emission signal template Tem(k), 1≤k≤T, where T is the length of the template; in this example, T=2000, Tem(k) contains 2000 sampling points and the duration is 0.02s.

[0048] Step 3: Under experimental conditions, during the slurry premixing process, the acoustic emission signal without metal residue is extracted. The time series of the acoustic emission signal is l0(i), i = 1, 2, ... T, where T is the length of the template. This signal is extracted to evaluate the background noise energy.

[0049] Step 4: Attach a high-frequency, high-sensitivity acoustic emission sensor (using a Japanese Fuji 1045S sensor with a response frequency of 50Hz–1300kHz) to the outer wall of the slurry premixing device to collect acoustic emission signals online from real-time acoustic emission events caused by the movement of excess metal during the slurry premixing process and from slurry premixing noise. Use a rectangular window function ω(j) to slide and truncate the time series y(m) of the online acoustic emission signal. The window function is expressed as:

[0050]

[0051] The window function ω(j) performs M truncation operations on the acoustic emission signal time series y(m) with a sliding step size of 99T. y(m) is then divided into truncated signal sequences Y, where Y = [y1, y2, ..., y]. m ], 1≤m≤M; the sliding step size is T shorter than the window function to avoid missing acoustic emission events at the boundary;

[0052] Step 5, truncate a segment y in the signal sequence Y. m A cross-correlation operation is performed with the acoustic emission event template signal Tem(k) to achieve template matching. The cross-correlation operation is as follows:

[0053]

[0054] In the formula, d represents the vector index with the same length as the template length T, q represents the offset of the template signal relative to the starting point of the truncated signal, 1≤q≤99T. The above cross-correlation operation is applied sequentially to each segment of the truncated signal sequence Y, and the template matching results of all segments are concatenated into R=[r1,r2,...,r m ], 1≤m≤M;

[0055] Step 6, calculate the energy of the background noise U(0):

[0056]

[0057] In the formula, l0(i) is the acoustic emission signal without any metal foreign matter;

[0058] Step 7: Determine whether an acoustic emission event has occurred based on the energy of the template matching result.

[0059] 7.1) Use a rectangular window function of length T to perform sliding truncation on the template matching result R, with a sliding step size of an integer between 1 and 0.2T, which is set to 1 here. Then calculate the energy of each truncated matching result fragment separately:

[0060]

[0061] When the sliding step size is 1, the interval between two calculations is It can achieve accurate detection;

[0062] 7.2) If U(p)≥αU(0) and α≥5, where 5 is taken, then it is considered that an acoustic emission event has occurred at this time, and p is the offset of the acoustic emission event from the start of signal acquisition.

[0063] 7.3) When 3U(0)≤U(p)≤5U(0), perform wavelet packet denoising on the signal segment, and repeat step 6.1) on the denoised signal to determine whether an acoustic emission event has occurred based on the signal energy, and record p;

[0064] Step 8: Determine the time of the acoustic emission event t = p / f based on the on-site sampling frequency. s f s The sampling frequency and the timing of the acoustic emission event indicate the presence of metallic impurities within the slurry.

[0065] Reference Figure 3 When a metal foreign object is placed in the acoustic emission signal, the acoustic emission event caused by the metal foreign object is observed in the time-domain waveform of the obtained signal. However, the duration is short and it is manifested as a pulse signal. At the same time, the signal is subject to strong noise interference with a large noise amplitude, and part of the signal is submerged in the noise.

[0066] Reference Figure 4 Signals with prominent acoustic emission waveforms were selected in the experiment. The acoustic emission component caused by metallic foreign matter in the signal was extracted as an acoustic emission template. The template is represented as a one-sided exponentially decaying waveform with a duration of 0.02 s in the waveform diagram. s At a sampling frequency of 100,000 Hz, the signal contains 2,000 sampling points. The template signal initially has a high amplitude, which then gradually decays to the noise level.

[0067] Reference Figure 5 The spectrum was obtained by performing a fast Fourier transform on the acoustic emission template. The peak frequency was 10751.1 Hz, which is consistent with the definition of the acoustic emission signal frequency range of 10 kHz to 100 kHz. The frequency components near 50 kHz are basically close to 0. Considering the sampling theorem and algorithm complexity, the sampling frequency of 100 kHz meets the sampling requirements.

[0068] Reference Figure 6 The acoustic emission template was used to perform template matching on the signals in the experiment, and three peak signals were detected. The windowed energy was calculated. The results showed that the acoustic emission event was extracted and the presence of metal foreign matter in the slurry was detected.

[0069] Reference Figure 7 ,right Figure 2In a certain experiment, wavelet packet denoising was performed on a signal with strong noise interference. The reconstructed signal waveform showed three peak signals. Energy comparison revealed three acoustic emission events, indicating the detection of excess metal within the slurry premixing device. For acoustic emission events where strong noise interference obscured the time-domain waveform, wavelet packet denoising achieved good extraction. Figure 6 A false peak appears at the end. Figure 7 This effectively eliminates spurious peaks, and wavelet packet denoising provides accurate results for areas where peak values ​​are unclear.

Claims

1. A method for detecting metal foreign matter in a slurry premixing process based on acoustic emission signal analysis, characterized in that: Includes the following steps: 1) Attach a high-frequency, high-sensitivity acoustic emission sensor to the outside of the pot wall of the slurry premixing device to collect acoustic emission signals generated during the slurry premixing process in real time and extract acoustic emission events caused by metal residues in real time. 2) During the extraction of slurry premixing, the acoustic emission signals generated by the collision and friction of metal foreign matter are used as standard acoustic emission signal templates. By windowing and segmenting the online acquired acoustic emission signals, cross-correlation analysis and template matching are performed with the standard acoustic emission signal templates. The energy of the template matching results is compared with the energy of the background noise. Based on wavelet packet denoising and energy analysis, acoustic emission events contained in the noise signals are discovered and extracted, and metal foreign matter in the slurry is detected online.

2. The method for detecting metal foreign matter in the slurry premixing process based on acoustic emission signal analysis according to claim 1, characterized in that, The acoustic emission sensor used in step 1) is a Japanese Fuji 1045S with a response frequency of 50Hz to 1300kHz.

3. The method for detecting metal foreign matter in the slurry premixing process based on acoustic emission signal analysis according to claim 1, characterized in that, Step 2) includes the following steps: 2.1) Under experimental conditions, during the slurry premixing process, acoustic emission signals generated by the collision and friction of excess metal were extracted. The time series of the acoustic emission signals is x(n), n=1,2,...N; 2.2) Search for a signal segment with exponential decay characteristics from the acoustic emission signal x(n), extract the signal segment and use it as the acoustic emission signal template Tem(k), 1≤k≤T, where T is the length of the template; 2.3) Under experimental conditions, during the slurry premixing process, acoustic emission signals without metal residues were extracted. The time series of the acoustic emission signals was l0(i), i = 1, 2, ... T, where T is the length of the template. 2.4) The time series y(m) of the acoustic emission signal acquired online is truncated using a rectangular window function ω(j), where y(m) is a segment of the acoustic emission signal acquired online. The window function is expressed as: The window function ω(j) performs M truncation operations on the acoustic emission signal time series y(m) with a sliding step size of 99T. y(m) is then divided into truncated signal sequences Y, where Y = [y1, y2, ..., y]. m ], 1≤m≤M; 2.5), for a certain segment y in the truncated signal sequence Y m Perform cross-correlation operation with the acoustic emission signal template Tem(k): In the formula, d represents the vector subscript, q represents the offset of the acoustic emission signal template relative to the starting point of the truncated signal, 1≤q≤99T. The above cross-correlation operation is applied sequentially to each segment of the truncated signal sequence Y, and the template matching results of all segments are concatenated into R=[r1,r2,...,r m ], 1≤m≤M; 2.6), Calculate the energy of the background noise U(0): In the formula, l0(i) is the acoustic emission signal without any metal foreign matter; 2.7) Determine whether an acoustic emission event has occurred based on the energy of the template matching result. The specific steps are as follows: 2.7.1) A rectangular window function of length T is used to slide and truncate the template matching result R. The sliding step size is an integer between 1 and 0.2T, which is taken as 1 here. Then, the energy of each segment of the matching result is calculated separately. 2.7.2), if U(p)≥αU(0), α≥5, and 5 is taken here, then it is considered that an acoustic emission event has occurred at this time, and p is the offset of the acoustic emission event from the start time of signal acquisition; 2.7.3) When 3U(0)≤U(p)≤5U(0), wavelet packet denoising is performed on the signal segment, and step 2.7.1) is repeated on the denoised signal to determine whether an acoustic emission event has occurred based on the signal energy. If an acoustic emission event has occurred, record p. 2.8) Determine the time of acoustic emission event t = p / f based on the on-site sampling frequency. s f s At the sampling frequency, the detection of acoustic emission events proved the presence of metallic impurities within the slurry.

Images