A method for estimating crossing time based on the Akaike information criterion of waveform similarity.

Abstract

This invention relates to a transit time estimation method based on the Akaike Information Criterion of waveform similarity, comprising the following steps: extracting effective ultrasonic waveform data; designing an FIR bandpass filter to perform zero-phase delay filtering and noise reduction processing on the effective ultrasonic waveform data; estimating transit time based on the Akaike Information Criterion of waveform similarity; and achieving super-resolution transit time correction based on phase compensation.

Description

Technical Field

[0001] This invention belongs to the field of fluid measurement technology and relates to a transit time estimation method based on the Akaike information criterion of waveform similarity, which is used for transit time measurement in ultrasonic flow meters. Background Technology

[0002] Accurate and reliable ultrasonic time-of-flight (TOF) estimation algorithms are crucial for ultrasonic flow meters, especially small-diameter downhole ultrasonic flow meters. The measurement sensitivity of flow meters in small-diameter pipes is lower than that in large-diameter pipes, which places higher demands on the resolution of TOF estimation. Furthermore, ultrasonic measurements in downhole flow meters suffer from waveform distortion, spike noise, and harmonic interference, requiring highly robust TOF estimation methods for accurate flow measurement. Existing TOF estimation methods, including thresholding, window function detection, cross-correlation, and the Akaike Information Criterion (AIC), each have their limitations. For example, the accuracy of thresholding TOF estimation can be affected by significant spike noise and fluctuating signal levels, and it depends on selecting an appropriate threshold coefficient, making its robustness and versatility insufficient for practical industrial needs. Window function detection methods suffer from high computational costs, making them unsuitable for real-time online flow measurement. Cross-correlation methods rely on selecting a suitable reference signal for accuracy, and their accuracy is easily affected by waveform distortion. Compared to other methods, the AIC method offers advantages such as ease of implementation, no need for pre-setting parameters, and insensitivity to waveform distortion. However, the AIC method has poor noise immunity to random noise and requires a high signal-to-noise ratio for the measured signal.

[0003] This invention proposes an AIC TOF estimation algorithm based on waveform similarity. While retaining the advantages of the traditional AIC TOF estimation algorithm, such as not requiring pre-set parameters and being unaffected by waveform distortion, it utilizes the similarity between two consecutive waveforms as prior knowledge to further improve the TOF estimation's ability to resist spike noise and random noise. Compared with the neighborhood TOF estimation algorithm based on the AIC framework developed in recent years (X. Qu, "Novel automatic first-arrival picking method for ultrasonic sound-speed tomography," Japanese Journal of Applied Physics, vol. 54, no. 7S1, pp. 07HF10, 2015.) and the spectral entropy TOF estimation algorithm (B. Bühling, "Improving onsetpicking in ultrasonic testing by using a spectral entropy criterion," The Journal of the Acoustical Society of America, vol. 155, no. 1, pp. 544-554, 2024.), the waveform similarity-based AIC TOF estimation algorithm proposed in this invention simply changes the variance by calculating the covariance of the AIC function. It does not require a reference waveform or additional processing steps, has lower computational complexity, and is more robust to spike noise. Summary of the Invention

[0004] This invention aims to improve the accuracy, noise immunity, and resolution of transit time estimation algorithms in ultrasonic flow meters, thereby achieving accurate flow measurement. To achieve this goal, this invention proposes a Time-of-Flight (TOF) estimation algorithm based on the similarity-based Akaike Information Criterion (SAIC). This method retains the advantages of the AIC TOF framework, such as not requiring pre-set parameters and being unaffected by waveform distortion. It further enhances accuracy and noise immunity by utilizing the similarity in continuously measured waveforms as prior information, and improves measurement resolution by employing a phase compensation algorithm to overcome sampling frequency limitations. The technical solution of this invention is as follows:

[0005] A method for estimating transit time based on the Akaike information criterion of waveform similarity includes the following steps:

[0006] Step 1: Extract effective ultrasound waveform data

[0007] Based on the actual sound path, flow range, and sound velocity of the liquid in the ultrasonic flow meter, the signal arrival time is estimated; sampling points are taken before and after the estimated signal arrival time to form effective ultrasonic waveform data.

[0008] Step 2: Design an FIR bandpass filter to perform zero-phase-delay filtering and noise reduction on the effective ultrasonic waveform data to obtain a zero-phase-delay filtered signal;

[0009] Step 3: Estimating transit time based on the Akaike information criterion of waveform similarity;

[0010] The signal cross-correlation method is used to calculate the Akaike information SAIC(k) based on waveform similarity for the zero-phase-delay filtered signals acquired by the same ultrasonic sensor in two consecutive acquisitions.

[0011] The calculated Akaike information based on waveform similarity is exponentially sharpened, and the corresponding weight w(k) is calculated:

[0012]

[0013] SAIC min This represents the minimum SAIC value based on Akaike information with similar waveforms;

[0014] The minimum value is located using weights, which is the ultrasound signal transit time t in this case. SAIC :

[0015]

[0016] Step 4: Achieve super-resolution transit time correction based on phase compensation.

[0017] In step one, based on the number of transmitted pulses N p Flowmeter ultrasonic signal frequency f c and the ultrasonic measurement sampling frequency f s Extracting data before and after the estimated signal arrival time. Each sampling point constitutes an effective ultrasonic waveform data.

[0018] Step two is as follows: based on the ultrasonic signal frequency f c With ultrasonic measurement sampling frequency f s An FIR bandpass filter was designed to process the effective ultrasonic waveform data sequentially through filtering, time-domain flipping, inverse filtering, and secondary flipping to obtain a filtered signal with zero phase delay.

[0019] In step three, the formula for calculating the Akaike information SAIC(k) based on waveform similarity for the zero-phase-delay filtered signals acquired by the same ultrasonic sensor in two consecutive acquisitions using the signal cross-correlation method is as follows:

[0020] SAIC(k)=klog(cov(s1(1,k),s2(1,k)))+Nlog(cov(s1(k+1,N),s2(k+1,N)))

[0021] Where k is the signal sampling point, N is the total signal length, and cov(·) represents the cross-correlation calculation of the signals.

[0022] Step four is as follows: using t SAIC Corresponding sampling point k SAIC Starting from this point, several sampling points are extracted, and the initial phase θ(k) of this signal segment is calculated using Hilbert transform. SAIC ):

[0023]

[0024] Where S(K) is the Hilbert transform of the original signal s(k), and K fc The center frequency f c The corresponding Hilbert transform points, imag(·) and real(·) are calculated by taking the real and imaginary parts respectively;

[0025] After obtaining the initial phase, the final transit time t is calculated by correcting the waveform similarity estimation result using the Akaike Information Criterion.

[0026]

[0027] The beneficial effects of this invention are as follows:

[0028] 1. The advantages of transit time estimation under the Akaike Information Criterion framework, which does not require pre-setting parameters and is not affected by waveform distortion, are retained.

[0029] 2. Design the Akaike information equation based on the similarity of continuously measured waveforms to improve the accuracy and noise resistance of transit time measurement;

[0030] 3. Design a super-resolution transit time estimation algorithm based on phase compensation to improve the transit time measurement resolution. Attached Figure Description

[0031] The following figures illustrate selected embodiments of the present invention and are exemplary rather than exhaustive or limiting, wherein:

[0032] Figure 1 The ultrasonic flowmeter path diagram corresponding to the measurement method of the present invention includes Z-type and π-type ultrasonic flowmeters;

[0033] Figure 2 Flowchart for zero-phase-delay filtering and noise reduction of ultrasonic signals;

[0034] Figure 3A flowchart for estimating transit time based on the Akaike Information Criterion of waveform similarity;

[0035] Figure 4 The flowchart shows the super-resolution transit time correction process based on phase compensation.

[0036] Figure 5 and Figure 6 The diagram compares the performance of different transit time estimation methods, including the traditional Akaike Information Criterion Transit Time Estimation (AIC), the Weighted Akaike Information Criterion Transit Time Estimation (WAIC), the Spectral Entropy TOF Estimation Algorithm (SEC), and the method proposed in this invention (SAIC). Detailed Implementation

[0037] The basic scheme of the present invention includes the following steps:

[0038] Step 1: Extract effective ultrasound waveform data

[0039] Based on the actual sound path, flow meter range, and liquid sound velocity within the flow meter, the signal arrival time is estimated; based on the number of transmitted pulses N... p Flowmeter ultrasonic signal frequency f c and the ultrasonic measurement sampling frequency f s Extracting data before and after the estimated signal arrival time. Each sampling point constitutes the effective ultrasound waveform data;

[0040] Step 2: Zero-phase delay filtering and noise reduction of the ultrasonic signal;

[0041] Based on the frequency f of the ultrasonic signal c With ultrasonic measurement sampling frequency f s A FIR (Finite Impulse Response) bandpass filter H is designed to sequentially filter, time-domain flip, inverse filter, and double flip the input effective ultrasonic waveform data, and finally output a filtered signal with zero phase delay.

[0042] Step 3: Estimating transit time based on the Akaike information criterion of waveform similarity;

[0043] In ultrasonic flow meters, the repetition period for ultrasonic signal transmission and reception is typically 1 ms. During this period, the change in flow velocity between two consecutive ultrasonic measurements is negligible, while the similarity of the measured waveforms can be used to improve the accuracy of transit time estimation and noise immunity. Using the filtered signals s1 and s2 with zero phase delay acquired by the same ultrasonic sensor in two consecutive measurements as inputs, the corresponding Akaike information SAIC(k) based on waveform similarity is calculated:

[0044] SAIC(k)=klog(cov(s1(1,k),s2(1,k)))+Nlog(cov(s1(k+1,N),s2(k+1,N)))

[0045] Where k is the signal sampling point, N is the total signal length, and cov(·) represents the signal cross-correlation calculation;

[0046] The calculated Akaike Information SAIC based on waveform similarity is exponentially sharpened, and the corresponding weight w(k) is calculated:

[0047]

[0048] SAIC min This represents the minimum SAIC value based on Akaike information with similar waveforms;

[0049] The minimum value is located using weights, which is the ultrasound signal transit time t in this case. SAIC :

[0050]

[0051] Step 4: Achieve super-resolution transit time correction below the sampling interval using the phase compensation method.

[0052] Limited by the sampling frequency, the Akaike information criterion's transit time suffers from insufficient temporal resolution. This can be addressed through phase compensation to achieve super-resolution transit time correction below the sampling interval. (The last part, "t," appears to be an unrelated fragment and is omitted from the translation.) SAIC Corresponding sampling point k SAIC Starting from this point, extract the following sections. The signal is sampled at several points, and the initial phase θ(k) of the signal is calculated using Hilbert transform. SAIC ):

[0053]

[0054] Where S(K) is the Hilbert transform of the original signal s(k), and K fc The center frequency f c The corresponding Hilbert transform points, imag(·) and real(·), are calculated by taking the real and imaginary parts, respectively.

[0055] After obtaining the initial phase, the waveform similarity Akaike information criterion estimation results can be corrected, and the final transit time t can be calculated:

[0056]

[0057] The embodiments of the present invention will now be described in detail with reference to the specification and accompanying drawings.

[0058] Figure 1The acoustic path diagram of the ultrasonic flowmeter corresponding to the measurement method of the present invention includes the traditional Z-type ( Figure 1 a) and π-type flow meter ( Figure 1 (b) During the measurement process, the ultrasonic signal generates a square wave pulse signal with 20 cycles and a center frequency of 1MHz at a repetition frequency of 1kHz. At the same time as the excitation, another receiving sensor begins to collect the ultrasonic signal and records the waveform, which is then uploaded to the host computer.

[0059] Figure 2 The flowchart for zero-phase delay filtering and denoising of ultrasonic signals is presented here, using the Z-transform as the tool. First, based on the center frequency and sampling frequency of the ultrasonic measurement signal, a bandpass filter is designed, whose transfer function's Z-transform is H(Z). This bandpass filter is then applied to the measurement signal x(t). Defining X(z) as the Z-transform of x(t), the filtered result is X(z)H(Z). The time-domain flipped signal yields X(z)H(Z). -1 )H(Z -1 After bandpass filtering the inverted signal again, we can obtain X(z). -1 )H(Z -1 The signal is then flipped in the time domain again, resulting in X(z)H(Z). -1 If the gain of the bandpass filter itself, |H(z)|, can be ensured to be 1, then the phase changes generated by H(Z) in the filtering result can cancel each other out, thus achieving zero-phase-delay ultrasonic signal filtering and denoising.

[0060] Figure 3 A flowchart for estimating transit time based on the Akaike Information Criterion of waveform similarity;

[0061] In ultrasonic flow meters, the repetition period for ultrasonic signal transmission and reception is typically 1 ms. During this period, the change in flow velocity between two consecutive ultrasonic measurements is negligible, and the similarity of the measured waveforms can be used to improve the accuracy of transit time estimation and noise immunity. Using the same ultrasonic sensor's received signals from two consecutive measurements as input, the corresponding Akaike Information Equation based on waveform similarity is calculated:

[0062] SAIC(k)=klog(cov(s1(1,k),s2(1,k)))+Nlog(cov(s1(k+1,N),s2(k+1,N)))

[0063] Where k is the signal sampling point, N is the total signal length, and cov(·) represents the cross-correlation calculation of the signals.

[0064] The calculated waveform similarity Akaike information equation is exponentially sharpened, and the corresponding weights of the equation are calculated:

[0065]

[0066] SAICmin This represents the minimum value of the SAIC equation.

[0067] The minimum value of the equation is located using the equation weights, which in this case is the ultrasound signal transit time.

[0068]

[0069] Figure 4 Flowchart of super-resolution transit time correction based on phase compensation

[0070] Crossing time t using the Akaike Information Criterion SAIC Corresponding sampling point k SAIC Starting from this point, extract the following sections. The signal is sampled at several points, and the initial phase θ(t) of the signal is calculated using Hilbert transform. SAIC ):

[0071]

[0072] Where S(K) is the Hilbert transform of the original signal s(k), and K fc The center frequency f c The corresponding Hilbert transform points, imag(·) and real(·), are calculated by taking the real and imaginary parts, respectively.

[0073] After obtaining the initial phase, the waveform similarity Akaike information criterion estimation results can be corrected, and the final transit time can be calculated:

[0074]

[0075] Figure 5 The figure shows a comparison of the effects of different transit time estimation methods. The results in the figure show that when there is spike noise and additive random noise in the ultrasonic measurement signal, the transit time detection results of the AIC, WAIC and SEC methods are shifted backward and deviate from the actual arrival time, while the SAIC method of the present invention can accurately measure the transit time.

[0076] Because the SEC method is too sensitive to noise, and the WAIC and AIC methods are not significantly different, therefore... Figure 6 In long-term testing, only the AIC method and the SAIC method of this invention were compared. As can be seen from the figure, the SAIC method of this invention has higher accuracy and robustness.

Claims

1. A method for estimating transit time based on the Akaike information criterion of waveform similarity, comprising the following steps: Step 1: Extract effective ultrasound waveform data Based on the actual sound path, flow range, and sound velocity of the liquid in the ultrasonic flow meter, the signal arrival time is estimated; sampling points are taken before and after the estimated signal arrival time to form effective ultrasonic waveform data. Step 2: Design an FIR bandpass filter to perform zero-phase-delay filtering and noise reduction on the effective ultrasonic waveform data to obtain a zero-phase-delay filtered signal; Step 3: Estimating transit time based on the Akaike information criterion of waveform similarity; The Akaike information SAIC(k) based on waveform similarity is calculated using the signal cross-correlation method for two consecutive zero-phase-delay filtered signals acquired by the same ultrasonic sensor: SAIC(k)=klog(cov(s1(1,k),s2(1,k)))+Nlog(cov(s1(k+1,N),s2(k+1,N))) in, cov(·) represents the cross-correlation calculation of signals; The calculated Akaike information based on waveform similarity is exponentially sharpened, and the corresponding weight w(k) is calculated: Among them, SAIC min This represents the minimum value of Akaike Information SAIC based on waveform similarity, where N is the total signal length and k is the signal sampling point; Let the ultrasonic measurement sampling frequency be f. s The minimum value is located using weights, which is the ultrasound signal transit time t. SAIC : Step 4: Implement super-resolution transit time correction based on phase compensation. The method is as follows: using t SAIC Corresponding sampling point k SAIC Starting from this point, several sampling points are extracted, and the initial phase θ(k) of this signal segment is calculated using Hilbert transform. SAIC ): Where S(K) is the Hilbert transform of the original signal s(k), and K fc The center frequency f of the ultrasound is represented by c The corresponding Hilbert transform points, imag(·) and real(·) are calculated by taking the real and imaginary parts respectively; After obtaining the initial phase, the final transit time t is calculated by correcting the waveform similarity estimation result using the Akaike Information Criterion.

2. The method for estimating the crossing time of the Akaike information criterion according to claim 1, characterized in that, In step one, based on the number of transmitted pulses N p Flowmeter ultrasonic center frequency f c and the ultrasonic measurement sampling frequency f s Extracting data before and after the estimated signal arrival time. Each sampling point constitutes an effective ultrasonic waveform data.

3. The method for estimating the crossing time of the Akaike information criterion according to claim 1, characterized in that, Step two is as follows: based on the ultrasound center frequency f c With ultrasonic measurement sampling frequency f s An FIR bandpass filter was designed to process the effective ultrasonic waveform data sequentially through filtering, time-domain flipping, inverse filtering, and secondary flipping to obtain a filtered signal with zero phase delay.

Images