Microphone welding direction detection method and device, computer equipment and medium

Abstract

The invention relates to the technical field of microphone welding quality detection. The microphone welding direction detection method comprises the steps that acoustic feedback information of a microphone to be detected is obtained, the acoustic feedback information is processed, high-frequency interference and background noise are removed, filtered information is obtained, and the microphone welding direction is detected. Extracting frequency spectrum feature information in a preset frequency range from the filtered information, generating target frequency band information, performing phase calculation on the target frequency band information to obtain actual phase value information, inputting the actual phase value information into a welding direction judgment model, generating a feature difference value, and judging the welding direction based on the feature difference value. And determining the microphone with the wrong welding direction. The welding direction detection device has the effect of improving the reliability of welding direction detection.

Description

Technical Field

[0001] The present invention relates to the technical field of microphone welding quality detection, and in particular to a microphone welding direction detection method, device, computer equipment and medium. Background Art

[0002] At present, the detection of microphone welding direction mainly relies on manual inspection and CCD visual inspection technology. Manual inspection is usually completed by operators through naked eye observation, but due to the influence of human subjective factors, negligence or misjudgment is prone to occur, resulting in inaccurate detection results. In addition, manual inspection is inefficient and it is difficult to meet the requirements of modern production lines for efficient inspection.

[0003] Although the existing CCD visual inspection has certain automation capabilities, it also has shortcomings. First, the equipment cost of the CCD system is relatively high, especially the investment in high-precision color cameras and related hardware, which increases the construction cost of the overall inspection system; second, CCD visual inspection has high requirements for ambient light and working conditions, and it is difficult to adapt to the complex working environment of the production workshop. In addition, the inspection of the CCD system requires a dedicated station design, which increases the additional pick-and-place time, further reducing the overall efficiency of the production line.

[0004] The above-mentioned related technologies have the following defects: the detection method is costly and complicated, and it is difficult to achieve fast and accurate detection requirements in the production line. Especially in the mass production process, it is easy to have unstable detection due to environmental interference or the complexity of the operating process. Summary of the invention

[0005] In order to improve the reliability of welding direction detection, the present application provides a microphone welding direction detection method, device, computer equipment and medium.

[0006] The above-mentioned invention objective of the present application is achieved through the following technical solutions: A method for detecting a welding direction of a microphone, the method comprising: Obtaining acoustic feedback information of the microphone under test; Processing the acoustic feedback information to remove high-frequency interference and background noise to obtain filtered information; Extracting spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; By performing phase calculation on the target frequency band information, actual phase value information is obtained; The actual phase value information is input into a welding direction determination model to generate a characteristic difference value, and based on the characteristic difference value, a microphone with a wrong welding direction is determined.

[0007] By adopting the above technical solution, by obtaining the acoustic feedback information of the microphone to be tested and filtering it, high-frequency interference and background noise can be effectively removed to ensure that the extracted signal is pure and reliable; by extracting the spectrum feature information within the preset frequency range from the filtered information and generating the target frequency band information, the pertinence and effectiveness of the analysis can be guaranteed; by performing phase calculation on the target frequency band information, the actual phase value information is obtained, which can accurately reflect the phase characteristics of the microphone to be tested; the actual phase value information is input into the welding direction determination model, and the calculation and analysis of the characteristic difference value by the model can quickly determine whether the welding direction is correct. This technical solution can effectively avoid the problems of low efficiency and large errors caused by relying on manual or high-cost detection equipment in traditional methods, and at the same time improve the accuracy and automation of welding direction determination, which is suitable for the high-efficiency detection needs in mass production, thereby improving the production quality of microphones and reducing the failure rate.

[0008] In a preferred example, the present application may be further configured as follows: the actual phase value information is obtained by performing phase calculation on the target frequency band information, including: Dividing the target frequency band information into time domain signals of multiple fixed time windows, performing Fourier transform on the time domain signals of the multiple fixed time windows to obtain multiple frequency domain signals; Acquire corresponding complex forms based on multiple frequency domain signals, and calculate initial phase information according to the complex forms; Performing phase unwrapping processing on the initial phase information to generate phase unwrapped information; The phase-unwrapped information is subjected to time series weighting processing to generate actual phase value information.

[0009] By adopting the above technical solution, by dividing the target frequency band information into time domain signals of multiple fixed time windows and performing Fourier transform on them, the frequency domain feature information can be effectively extracted, the complex form is obtained based on the frequency domain signal and the initial phase information is calculated, thereby ensuring the integrity of the phase feature, and by performing phase unwrapping processing on the initial phase information, the discontinuity problem caused by the phase jump is solved, and smooth and accurate phase-unwrapped information is generated. The phase-unwrapped information is subjected to time series weighted processing, and the weight is calculated in combination with the energy distribution of the time window, thereby generating actual phase value information reflecting the overall characteristics.

[0010] In a preferred example, the present application may be further configured as follows: calculating the initial phase information according to the complex form includes: obtaining a real signal and an imaginary signal from the complex form; The initial phase information is calculated by the following formula: , wherein the θ represents the initial phase information, the arctan represents the inverse tangent function, and the Re(f) and the Im(f) respectively represent the real signal and the imaginary signal in the complex form.

[0011] By adopting the above technical solution, the real signal and imaginary signal in the complex form are extracted, which correspond to the cosine component and sine component of the frequency component respectively, thereby ensuring the integrity and accuracy of the signal. By combining the inverse tangent function to calculate the initial phase information, the phase angle value of each frequency point can be accurately restored. At the same time, the phase angle is corrected according to the quadrant characteristics of the complex number, thus avoiding the ambiguity problem in the phase calculation.

[0012] In a preferred example, the present application may be further configured as follows: performing phase unwrapping processing on the initial phase information to generate information after phase unwrapping includes: Performing a continuity check on the initial phase information to identify phase jump points between adjacent time points; The phase jump point is corrected based on the following formula: , where the θ 展开 (t) is the information after the phase is expanded, θ(t) is the phase jump point, and k is the adjustment coefficient at the jump point.

[0013] By adopting the above technical solution, by checking the continuity of the initial phase information, the phase jump points caused by periodicity between adjacent time points can be accurately identified, avoiding the error of phase data caused by jumps; based on the identification results of the phase jump points, the jump points are corrected through formulas, and the phase values ​​are reasonably adjusted to fall within the continuous range, thereby generating information after phase expansion. This technical solution overcomes the mutation problem caused by jumps in traditional phase calculations and ensures the time series continuity of phase information.

[0014] In a preferred example, the present application may be further configured as follows: performing time series weighted processing on the phase-unwrapped information to generate actual phase value information includes: The time weight is calculated according to the signal energy value of the phase-unwrapped information in each time window. The calculation formula of the time weight is: , wherein w(t) represents the weight of the t-th time window, E(t) represents the square of the signal energy of the t-th time window, and ∑ t E(t) represents the sum of the signal energies in all time windows; Based on the time weight, the information after the phase expansion is weighted, and the weighted calculation formula is: , where the θ加权 represents the actual phase value information, w(t) is the weight of the t-th time window, and θ 展开 (t) is the information after the phase is unfolded.

[0015] By adopting the above technical solution, by calculating the square of the signal energy of each time window and combining the normalization operation to generate the time weight, the significance of the high-energy window can be fully reflected, so that the time window with strong signal energy contributes more to the final phase value, and the influence of the low-energy window on the overall calculation result is reasonably weakened, thereby effectively improving the noise resistance and accuracy of the calculation result. At the same time, based on the time weight, the information after phase expansion is weighted, which can comprehensively reflect the overall characteristics of the signal in the target frequency band and avoid the result deviation caused by signal fluctuations or noise in traditional methods. This technical solution significantly improves the stability and reliability of the phase calculation results, reduces the possibility of being affected by random changes or abnormal points of the signal, and is particularly suitable for high-precision phase analysis in complex signal environments.

[0016] In a preferred example, the present application may be further configured as follows: the actual phase value information is input into the welding direction determination model to generate a characteristic difference value, including: The welding direction determination model extracts features from the actual phase value information to obtain actual phase features; The actual phase feature is compared point by point with the standard feature of the normal welding direction pre-stored in the welding direction determination model, and the feature difference value between the actual phase feature and the standard feature is calculated.

[0017] By adopting the above technical solution, the welding direction determination model can be used to extract features from the actual phase value information, thereby refining the actual phase features that can accurately reflect the characteristics of the welding direction. By comparing the extracted actual phase features with the standard features of the normal welding direction pre-stored in the model point by point, the feature difference value between the actual features and the standard features can be accurately calculated to quantify the degree of deviation in the welding direction. This technical solution can overcome the shortcomings of traditional methods that are difficult to capture complex signal features. Through the deep feature extraction of the model, the comprehensiveness and robustness of the features are guaranteed; the point-by-point comparison method not only improves the calculation accuracy, but also can locate specific deviation features and improve the reliability of the determination. Through this method, the abnormality of the welding direction can be quickly and accurately detected, providing a scientific basis for welding quality control, while avoiding the misjudgment problem caused by human judgment or simple feature comparison in traditional detection methods, and further improving the accuracy and efficiency of welding direction determination.

[0018] In a preferred example, the present application may be further configured as follows: the microphone that determines the welding direction error based on the characteristic difference value includes: Extracting a deviation threshold of a normal welding direction from the welding direction determination model, wherein the deviation threshold of the normal welding direction is used to define an allowable range of characteristic differences in the normal welding direction; The characteristic difference value is compared with the deviation threshold of the normal welding direction. If the characteristic difference value is less than or equal to the deviation threshold of the normal welding direction, the welding direction is determined to be correct; if the characteristic difference value is greater than the deviation threshold of the normal welding direction, the welding direction is determined to be incorrect.

[0019] By adopting the above technical solution, by extracting the deviation threshold of the normal welding direction from the welding direction determination model, the allowable range of the characteristic difference of the normal welding direction can be clarified, and a specific quantitative determination standard can be provided. The calculated characteristic difference value is compared with the extracted normal welding direction deviation threshold one by one. When the characteristic difference value is less than or equal to the deviation threshold of the normal welding direction, it can be determined that the welding direction meets the requirements of the normal welding direction; when the characteristic difference value is greater than the deviation threshold of the normal welding direction, it can be determined that there is a deviation in the welding direction, thereby identifying the microphone with the wrong welding direction.

[0020] The second object of the invention is achieved by the following technical solutions: A microphone welding direction detection device, the microphone welding direction detection device comprising: An acoustic signal acquisition module, used to obtain acoustic feedback information of the microphone to be tested; A signal preprocessing module, used to process the acoustic feedback information, remove high-frequency interference and background noise, and obtain filtered information; A spectrum feature extraction module, used to extract spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; A phase calculation and feature generation module, used to obtain actual phase value information by performing phase calculation on the target frequency band information; The welding direction determination module is used to input the actual phase value information into the welding direction determination model to generate a characteristic difference value, and determine the microphone with the wrong welding direction based on the characteristic difference value.

[0021] By adopting the above technical solution, by obtaining the acoustic feedback information of the microphone to be tested and filtering it, high-frequency interference and background noise can be effectively removed to ensure that the extracted signal is pure and reliable; by extracting the spectrum feature information within the preset frequency range from the filtered information and generating the target frequency band information, the pertinence and effectiveness of the analysis can be guaranteed; by performing phase calculation on the target frequency band information, the actual phase value information is obtained, which can accurately reflect the phase characteristics of the microphone to be tested; the actual phase value information is input into the welding direction determination model, and the calculation and analysis of the characteristic difference value by the model can quickly determine whether the welding direction is correct. This technical solution can effectively avoid the problems of low efficiency and large errors caused by relying on manual or high-cost detection equipment in traditional methods, and at the same time improve the accuracy and automation of welding direction determination, which is suitable for the high-efficiency detection needs in mass production, thereby improving the production quality of microphones and reducing the failure rate.

[0022] The third objective of the present application is achieved through the following technical solutions: A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned microphone welding direction detection method when executing the computer program.

[0023] The fourth objective of the present application is achieved through the following technical solutions: A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the microphone welding direction detection method are implemented.

[0024] In summary, the present application includes at least one of the following beneficial technical effects: 1. By obtaining the acoustic feedback information of the microphone to be tested and filtering it, high-frequency interference and background noise can be effectively removed to ensure that the extracted signal is pure and reliable; by extracting the spectral feature information within the preset frequency range from the filtered information and generating the target frequency band information, the pertinence and effectiveness of the analysis can be ensured; by performing phase calculation on the target frequency band information, the actual phase value information is obtained, which can accurately reflect the phase characteristics of the microphone to be tested; the actual phase value information is input into the welding direction determination model, and the calculation and analysis of the characteristic difference value by the model can quickly determine whether the welding direction is correct. This technical solution can effectively avoid the problems of low efficiency and large errors caused by reliance on manual or high-cost detection equipment in traditional methods, while improving the accuracy and automation of welding direction determination, and is suitable for efficient detection needs in mass production, thereby improving the production quality of microphones and reducing the failure rate; 2. By calculating the square of the signal energy of each time window and combining the normalization operation to generate the time weight, the significance of the high-energy window can be fully reflected, so that the time window with strong signal energy contributes more to the final phase value, and the influence of the low-energy window on the overall calculation result is reasonably weakened, thereby effectively improving the noise resistance and accuracy of the calculation result. At the same time, based on the time weight, the information after phase expansion is weighted, which can comprehensively reflect the overall characteristics of the signal in the target frequency band and avoid the result deviation caused by signal fluctuations or noise in traditional methods. This technical solution significantly improves the stability and reliability of the phase calculation results, reduces the possibility of being affected by random changes or abnormal points of the signal, and is particularly suitable for high-precision phase analysis in complex signal environments; 3. The welding direction determination model can be used to extract features from the actual phase value information, thereby refining the actual phase features that can accurately reflect the characteristics of the welding direction. By comparing the extracted actual phase features with the standard features of the normal welding direction pre-stored in the model point by point, the feature difference value between the actual features and the standard features can be accurately calculated to quantify the degree of deviation in the welding direction. This technical solution can overcome the shortcomings of traditional methods that are difficult to capture complex signal features. Through the deep feature extraction of the model, the comprehensiveness and robustness of the features are guaranteed; the point-by-point comparison method not only improves the calculation accuracy, but also can locate specific deviation features and improve the reliability of the determination. Through this method, the abnormality of the welding direction can be quickly and accurately detected, which provides a scientific basis for welding quality control. At the same time, it avoids the problem of misjudgment caused by human judgment or simple feature comparison in traditional detection methods, and further improves the accuracy and efficiency of welding direction determination. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of a method for detecting a microphone welding direction in an embodiment of the present application; Figure 2 It is a flowchart for implementing step S40 in a method for detecting a welding direction of a microphone in an embodiment of the present application; Figure 3 It is a flowchart for implementing step S402 in a method for detecting a welding direction of a microphone in an embodiment of the present application; Figure 4 It is a flowchart for implementing step S403 in a method for detecting a welding direction of a microphone in an embodiment of the present application; Figure 5 It is a flowchart for implementing step S404 in a method for detecting a welding direction of a microphone in an embodiment of the present application; Figure 6 is a flowchart for implementing step S50 in a method for detecting a microphone welding direction in an embodiment of the present application; Figure 7is another implementation flow chart of step S50 in the microphone welding direction detection method in one embodiment of the present application; Figure 8 This is a principle block diagram of a microphone welding direction detection device in one embodiment of the present application; Fig. 9 is a schematic diagram of a device in an embodiment of the present application; Fig.10 This is a phase contrast diagram of the microphone welding direction in one embodiment of the present application. DETAILED DESCRIPTION

[0026] The present application is further described in detail below in conjunction with the accompanying drawings.

[0027] In one embodiment, if Figure 1 and Fig.10 As shown, the present application discloses a method for detecting the welding direction of a microphone, which specifically includes the following steps: S10: Acquiring acoustic feedback information of the microphone to be tested.

[0028] Specifically, by applying a set of preset acoustic excitation signals to the microphone to be tested, the acoustic excitation signals can be test signals covering the operating frequency range of the microphone, and a sinusoidal wave scanning signal or a white noise signal in the range of 200 Hz to 300 Hz is selected. The sinusoidal wave scanning signal or the white noise signal is used as an input signal, and the acoustic signal is converted into a corresponding electrical signal through the microphone. At the same time, the electrical signal is sampled and processed by a signal acquisition device, and the acquisition device digitizes the analog electrical signal into discrete time series data. By setting an appropriate sampling frequency, it is ensured that all valid frequency components of the acoustic feedback signal can be retained, thereby obtaining the acoustic feedback information of the microphone to be tested.

[0029] S20: Process the acoustic feedback information to remove high-frequency interference and background noise to obtain filtered information.

[0030] Specifically, the acoustic feedback information is band-pass filtered to remove interference signals outside the target frequency range. The upper and lower cut-off frequencies of the band-pass filter are set to 200 Hz to 300 Hz, and the input signal is filtered point by point using digital filtering technology. In specific implementation, an IIR filter or a FIR filter can be used. By setting the frequency response function of the filter, the signal strength of the high-frequency component and the low-frequency component is weakened, and only the signal components in the range of 200 Hz to 300 Hz are retained. In addition, in order to further reduce the impact of random noise on the signal, the filtered signal can be smoothed. For example, a sliding average method is used. By setting a time window of a fixed size, the signal around each sampling point is averaged, thereby smoothing the sharp fluctuations in the signal and obtaining the filtered information with high-frequency interference and background noise removed.

[0031] S30: extracting frequency spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information.

[0032] In this embodiment, the preset frequency range refers to a target frequency range selected according to the working characteristics of the microphone and the welding direction detection requirements, wherein the preset frequency range is 200 Hz to 300 Hz.

[0033] Specifically, the time domain signal is converted into a frequency domain signal by performing a fast Fourier transform operation on the filtered information. In the specific implementation, the FFT algorithm is performed on the discrete sampling points of the signal, and the amplitude and phase information of the frequency component are generated by repeated complex addition, subtraction and multiplication operations; in the generated frequency domain signal, the frequency components in the range of 200Hz to 300Hz and their corresponding amplitudes are selected as the spectrum feature information, and the energy size or average amplitude information of the specific frequency component can be calculated by statistically analyzing the amplitude distribution in the target frequency range; further, the spectrum features can be normalized by dividing the amplitude of each frequency component by the total energy of the frequency domain signal to generate standardized spectrum feature information, thereby obtaining the target frequency band information in the preset frequency range of 200Hz to 300Hz.

[0034] S40: Obtaining actual phase value information by performing phase calculation on the target frequency band information.

[0035] Specifically, the frequency components in the target frequency band information are represented in complex form. By extracting the real signal and the imaginary signal of each frequency component, the phase angle of the corresponding frequency is calculated using the formula, and the phase information of each frequency point is integrated into the initial phase information of the time series. In order to solve the discontinuity caused by the phase jump, the initial phase information can be phase-unwrapped by detecting whether the phase difference between two adjacent frequency points exceeds the range of −π to π. If so, 2π is added or subtracted to achieve continuous phase unwrapping. Furthermore, the information after phase unwrapping is weighted to generate actual phase value information, for example, the actual phase value is calculated by taking the weighted average of the phase values ​​of the frequency components according to the weights of the corresponding amplitudes.

[0036] Further, Fig.10 The microphone phase test in Figure 1 shows the phase response under different welding directions. The light-colored line represents the forward welding phase response, and the dark-colored line represents the reverse welding phase response. The difference between these two lines provides key information for judging the welding direction. By calculating the phase information of each frequency point, and through phase unwrapping and weighting processing, the actual smooth phase value information is finally obtained.

[0037] S50: Inputting the actual phase value information into the welding direction determination model to generate a characteristic difference value, and based on the characteristic difference value, determining the microphone with the wrong welding direction.

[0038] Specifically, the actual phase value information is divided into multiple time windows of fixed length, and the actual phase value in each time window is sent as input data to the welding direction determination model. The welding direction determination model performs feature analysis on the input actual phase value according to pre-trained parameters, specifically including standardizing the input features, adjusting the input values ​​to within the range accepted by the model, and then processing the input features layer by layer through the computing layer of the model. For example, in a neural network model, the input features are linearly combined and nonlinearly transformed by setting the activation function and weight parameters of the hidden layer to generate a difference value between the actual phase value and the normal welding direction standard features pre-stored in the model. After the feature difference value is output, the difference value is compared with the set deviation threshold. If the feature difference value is less than or equal to the threshold range, the welding direction is determined to be correct. If the feature difference value is greater than the threshold range, the welding direction is determined to be wrong, thereby completing the identification and judgment of the microphone with the wrong welding direction.

[0039] Further, according to Fig.10 In the microphone welding direction phase test, there is a significant difference in the phase response between forward welding and reverse welding. By calculating these phase difference values, it is possible to effectively determine whether the welding direction is correct. If the difference value exceeds the preset threshold, it is determined that the welding direction is wrong.

[0040] In one embodiment, if Figure 2 and Fig.10 As shown, in step S40, the actual phase value information is obtained by performing phase calculation on the target frequency band information, including: S401: Divide the target frequency band information into time domain signals of multiple fixed time windows, and perform Fourier transform on the time domain signals of the multiple fixed time windows to obtain multiple frequency domain signals.

[0041] Specifically, by dividing the time domain signal of the target frequency band information into time windows, first, a fixed window length and overlap rate are set according to the sampling rate of the target frequency band information, for example, the length of each window is set to 1024 sampling points, and the windows overlap by 50%, to ensure that the signal in each time window has sufficient frequency resolution and reduce time domain loss. Then, windowing processing is applied to the signal in each time window to reduce spectrum leakage caused by signal truncation. A Hanning window or a Hamming window function can be used to perform weighted processing on the signal samples in the window. By multiplying the sample points by the corresponding weights of the window function, the influence of boundary effects on the accuracy of the frequency domain signal is reduced. Subsequently, a fast Fourier transform operation is performed on the windowed signal, and the time domain signal in each window is converted into a frequency domain signal through a hierarchical iterative complex operation process. The fast Fourier transform process generates amplitude and phase information in the frequency domain by performing complex addition and subtraction operations on the real and imaginary parts of the sample points, and finally obtains multiple frequency domain signals, and the frequency domain signals correspond to the spectrum characteristics of each time window.

[0042] Further, in Fig.10 There are two phase response curves in total. The light-colored line represents the phase response of the microphone in the forward welding direction, while the dark-colored line represents the phase response of the microphone in the reverse welding direction. The two lines show the phase difference of the microphone in the forward and reverse welding directions. This difference is very important for the subsequent welding direction determination. The frequency domain signal can be obtained through Fourier transform, which provides the basis for phase difference analysis.

[0043] S402: Acquire corresponding complex forms based on multiple frequency domain signals, and calculate initial phase information according to the complex forms.

[0044] Specifically, by extracting the complex form of each frequency point within the target frequency range in the frequency domain signal, the complex form is represented by the real and imaginary parts of the frequency component, and by directly reading the complex value of the corresponding frequency from the frequency component data of the frequency domain signal, including the real part Re(f) and the imaginary part Im(f), and then, according to the mathematical properties of the complex form, the initial phase information of the corresponding frequency component is calculated.

[0045] Further, from Fig.10 From the phase response information in , we can see that in the range of 200Hz to 300Hz, the phase values ​​of forward welding and reverse welding show obvious differences. By calculating the complex form of these frequency points, the phase information of each frequency point is obtained.

[0046] S403: Perform phase unwrapping processing on the initial phase information to generate information after phase unwrapping.

[0047] Specifically, the difference between adjacent points of the phase value in the initial phase information is calculated to detect whether there is a discontinuity caused by a phase jump. First, the phase difference between each two adjacent frequency points in the initial phase information is calculated in turn. When the absolute value of the phase difference is greater than π, it is determined that the phase of the frequency point has jumped. If the phase difference is positive and exceeds π, 2π is subtracted from the phase value of the current frequency point and all subsequent frequency points. If the phase difference is negative and less than −π, 2π is added to the phase value of the current frequency point and all subsequent frequency points. In this way, the phase jump is compensated to ensure that the phase value is continuous in the frequency dimension. The phase unwrapping process can be performed point by point within the entire target frequency band. The phase information is recursively unwrapped starting from the first frequency point until the entire initial phase information set is traversed, and finally the information after phase unwrapping is generated.

[0048] Further, from Fig.10 From the phase response results in Figure 1, the phase response curve of reverse welding (dark line) shows a larger offset than the phase response curve of forward welding (light line). In order to eliminate this jump and discontinuity, phase unwrapping processing can smooth the phase data and eliminate the phase discontinuity caused by the wrong welding direction, thereby ensuring that subsequent analysis is more accurate.

[0049] S404: Perform time series weighting processing on the phase-unwrapped information to generate actual phase value information.

[0050] Specifically, actual phase value information is generated by performing weighted averaging calculations on the phase-unwrapped information within a time window. The specific operations include first calculating the weight value of each time window. The weight value can be calculated based on the signal energy or amplitude. For example, the signal energy is obtained by summing the squares of the signal amplitudes of each time window, and the signal energy of each time window is used as a weight factor. Then, the phase unwrapping information within the time window is multiplied by the corresponding weight factor, the weighted phase values ​​of all time windows are accumulated, and divided by the sum of the weight values ​​of all time windows to generate the final actual phase value information.

[0051] In one embodiment, if Figure 3 As shown, in step S402, the initial phase information is calculated according to the complex form, including: S4021: Obtain a real signal and an imaginary signal from the complex form.

[0052] Specifically, the complex value of each frequency point is read from the complex data of the frequency domain signal, and the cosine component is obtained by directly extracting the real part of the complex number, and the sine component is obtained by extracting the imaginary part of the complex number. The real signal and the imaginary signal correspond to the cosine and sine contributions of the amplitude information of the frequency component in the time dimension respectively. The real signals and imaginary signals obtained in all target frequency ranges are respectively organized into array representations of the corresponding frequency points for subsequent phase calculations.

[0053] S4022: Calculate the initial phase information using the following formula: , where θ represents the initial phase information, arctan represents the inverse tangent function, Re(f) and Im(f) represent the real signal and imaginary signal in complex form, respectively.

[0054] Specifically, by extracting the real and imaginary signals in complex form within the target frequency range, the formula , calculate the initial phase information of each frequency point, where Re(f) represents the real signal in complex form, and Im(f) represents the imaginary signal in complex form; in the calculation process, for each target frequency point, the extracted real signal and imaginary signal are input respectively, and the phase angle is calculated by the inverse tangent function. For example, assuming that the complex form corresponding to a certain target frequency point f1=250Hz is f1=3+j4, where the real signal Re(f1)=3 and the imaginary signal Im(f1)=4, then the initial phase information θ1 of this frequency point can be calculated as , the calculation result is θ1≈0.93 radians. By performing the same calculation operation on all frequency points in the target frequency range in turn, a complete set of initial phase information is generated.

[0055] In one embodiment, if Figure 4 As shown, in step S403, the initial phase information is subjected to phase unwrapping processing to generate information after phase unwrapping, including: S4031: Perform a continuity check on the initial phase information to identify phase jump points between adjacent time points.

[0056] Specifically, by calculating the phase difference between adjacent time points in the initial phase information, it is determined whether there is a phase jump phenomenon. The specific operations include: extracting the phase values ​​θ(t) and θ(t+1) of adjacent time points t and t+1 in turn from the time series of the initial phase information, calculating the phase difference Δθ=θ(t+1)−θ(t), and when the absolute value of the phase difference exceeds π, marking the current point t+1 as a phase jump point; at the same time, judging the direction of the jump according to the positive or negative nature of the phase difference, for example, when the phase difference is positive and exceeds π, it is marked as a positive jump point, and when the phase difference is negative and less than −π, it is marked as a negative jump point, and calculating the phase difference of all adjacent time points in the initial phase information point by point.

[0057] S4032: Correct the phase jump point based on the following formula: , where θ 展开 (t) is the information after phase expansion, θ(t) is the phase jump point, and k is the adjustment coefficient at the jump point.

[0058] Specifically, by applying the formula to the identified phase jump point , where k represents the adjustment coefficient at the jump point, the adjustment coefficient k for the positive jump point is −1, and the adjustment coefficient k for the negative jump point is +1. The specific operations include: for each time point t marked as a phase jump point, first calculate its adjustment direction according to the phase difference value of the jump point. When the phase difference value of the jump point is greater than π, the adjustment coefficient k is set to −1, and −2π is added to the current phase value θ(t). When the phase difference value of the jump point is less than −π, the adjustment coefficient k is set to +1, and +2π is added to the current phase value θ(t). Then, update the adjusted phase value to the phase time series, and recursively process all time points after the jump point to ensure the continuity of the phase value. For example, if the phase jump point θ(5)=3.5π is detected at the time point t=5, and the adjustment coefficient k=−1, then the corrected phase value θ 展开 (5)=3.5π−2π=1.5π. After the correction is completed, the result is integrated into the entire time series to generate continuously changing phase-unwrapped information.

[0059] In one embodiment, if Figure 5 As shown, in step S404, the information after phase expansion is subjected to time series weighting processing to generate actual phase value information, including: S4041: Calculate the time weight according to the signal energy value in each time window after the phase expansion. The calculation formula of the time weight is: , where w(t) represents the weight of the t-th time window, E(t) represents the square of the signal energy of the t-th time window, ∑ tE(t) represents the sum of the signal energies in all time windows.

[0060] Specifically, the time weight is generated by calculating the signal energy value in each time window. First, the signal amplitude data of each time window is extracted from the information after phase expansion, and the square of the signal amplitude of the time window is summed to obtain the signal energy value E(t). The square of the signal energy is calculated by summing the square of the discrete signal sample value in each time window. For example, for the signal sample point [x1, x2, ..., xn] in the t-th time window, the calculation formula of the signal energy value is: , and then calculate the sum of the signal energy of all time windows ∑ t E(t) is used to normalize the energy value of each time window. The square of the signal energy E(t) of each time window is divided by the sum of the signal energies of all time windows to obtain the weight value w(t) of the time window, ensuring that the sum of the weight values ​​of all time windows is 1. This calculation process is performed on all time windows one by one to generate a set of time weights [w(1), w(2), …, w(T)].

[0061] S4042: Based on the time weight, weighted processing is performed on the information after phase expansion. The weighted calculation formula is: , where θ 加权 represents the actual phase value information, w(t) is the weight of the tth time window, θ 展开 (t) is the information after phase unwrapping.

[0062] Specifically, the information after phase expansion is weighted based on the time weight. First, the phase expansion information θexpand(t) in each time window is multiplied point by point with the corresponding time weight w(t), and the product of the phase expansion value and the weight is used as the weighted phase value; then, the weighted phase values ​​of all time windows are summed to obtain the sum of the weighted phases. , and directly use the result as the actual phase value information θ 加权 , which can ensure that the phase value after weighted calculation fully considers the differences in signal energy contributions of each time window; for example, if the signal energy of a time window is higher, the corresponding weight is larger, and the phase contribution of this window will account for a larger proportion during calculation, while the time window with lower signal energy will contribute less to the overall phase value, and the actual phase value information will be finally generated.

[0063] In one embodiment, if Figure 6 As shown, in step S50, the actual phase value information is input into the welding direction determination model to generate a characteristic difference value, including: S501: The welding direction determination model extracts features from the actual phase value information to obtain actual phase features.

[0064] Specifically, by inputting the actual phase value information into the welding direction determination model, the welding direction determination model performs layer-by-layer feature processing on the input actual phase value information. First, the actual phase value information is preprocessed, including denoising and standardizing the phase value, eliminating abnormal fluctuations in the data and unifying the numerical range to ensure the consistency and accuracy of the model input data; then, the feature extraction unit of the welding direction determination model is used to perform in-depth analysis on the input data. For example, the first layer of the model extracts global characteristics in the time series and generates primary feature vectors through weighted summation and bias adjustment. The second layer uses convolution operations to extract local change characteristics. For example, convolution kernels are used to detect the changing trends of phase values ​​in different time periods to capture key characteristic points; subsequently, the high-order feature layer of the model further abstracts the phase information and extracts more advanced features with judgment significance, such as the phase consistency features of the time series or the phase offset features of the frequency distribution. Finally, all the extracted features are integrated into a unified feature vector as the actual phase feature.

[0065] S502: comparing the actual phase feature with the standard feature of the normal welding direction pre-stored in the welding direction determination model point by point, and calculating the feature difference value between the actual phase feature and the standard feature.

[0066] Specifically, the standard feature of the normal welding direction pre-stored is extracted from the welding direction judgment model. The standard feature is a feature template obtained by training based on the normal welding direction and is expressed as a feature vector with the same dimension as the actual phase feature. The actual phase feature is compared point by point with the standard feature according to the same dimension. During the comparison, the feature difference value is generated by calculating the difference in each dimension. The specific operations include: for each dimension iii of the feature vector, the component value xi of the actual phase feature and the component value yi of the standard feature are taken out, the difference Δi=xi−yi is calculated, and the difference is stored as the feature difference value. Subsequently, in order to quantify the overall feature difference, all the differences are squared, summed and squared, and the Euclidean distance representation of the feature difference value is calculated. The formula is: , where n is the total number of dimensions of the feature vector, which reflects the overall deviation between the actual phase feature and the standard feature; finally, the calculated point-by-point difference value and the overall difference value are used as the result of the feature difference value for the subsequent judgment of the correctness of the welding direction.

[0067] In one embodiment, if Figure 6 As shown, in step S50, based on the characteristic difference value, determining the microphone with the wrong welding direction includes: S503: extracting a deviation threshold of a normal welding direction from the welding direction determination model, where the deviation threshold of the normal welding direction is used to define an allowable range of characteristic differences in the normal welding direction.

[0068] Specifically, the deviation threshold of the normal welding direction is read from the preset parameters of the welding direction judgment model. The deviation threshold is a numerical range obtained by statistics based on a large number of normal welding direction sample data during the training process. The numerical range reflects the upper and lower limits of the characteristic difference value of the normal welding direction. When extracting the deviation threshold, first match the applicable welding direction model version according to the target frequency band and the specific application scenario, and directly read the corresponding deviation threshold parameters from the storage parameters of the model, such as the upper and lower limits [Δmin, Δmax]. According to the judgment requirements of the welding direction, the deviation threshold is defined as a single numerical value in a specific form. For example, a fixed threshold Δthreshold is set based on the average deviation of the characteristic difference value to determine whether the characteristic difference value is within the allowable range of the normal welding direction. The deviation threshold is part of the model parameters and is directly extracted for subsequent comparison operations.

[0069] S504: Compare the feature difference value with the deviation threshold of the normal welding direction. If the feature difference value is less than or equal to the deviation threshold of the normal welding direction, the welding direction is determined to be correct. If the feature difference value is greater than the deviation threshold of the normal welding direction, the welding direction is determined to be incorrect.

[0070] Specifically, by comparing the calculated feature difference values ​​with the extracted normal welding direction deviation threshold one by one, the feature difference value is first read and compared with the deviation threshold. When the feature difference value is less than or equal to the deviation threshold Δthreshold, it is determined that the degree of deviation of the actual phase feature from the normal welding direction is within the allowable range, and the welding direction is determined to be correct; when the feature difference value is greater than the deviation threshold Δthreshold, it is determined that the deviation of the actual phase feature exceeds the allowable range of the normal welding direction, and the welding direction is determined to be wrong; in a specific implementation, a numerical judgment can be directly performed through a comparison operator, for example, the judgment result of the welding direction is determined by judging the conditions Δ≤Δthreshold or Δ>Δthreshold; in addition, in order to improve the judgment accuracy, further analysis can be performed in combination with the distribution characteristics of the feature difference value, such as statistically analyzing the deviation concentration of the feature difference value in each dimension, and generating the final judgment output in combination with the comparison result of the deviation threshold.

[0071] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0072] In one embodiment, a microphone welding direction detection device is provided, and the microphone welding direction detection device corresponds one-to-one to the microphone welding direction detection method in the above embodiment. Figure 8As shown, the microphone welding direction detection device includes an acoustic signal acquisition module, a signal preprocessing module, a spectrum feature extraction module, a phase calculation and feature generation module and a welding direction determination module. The functional modules are described in detail as follows: An acoustic signal acquisition module, used to obtain acoustic feedback information of the microphone to be tested; The signal preprocessing module is used to process the acoustic feedback information, remove high-frequency interference and background noise, and obtain filtered information; A spectrum feature extraction module is used to extract spectrum feature information within a preset frequency range from the filtered information and generate target frequency band information; The phase calculation and feature generation module is used to obtain the actual phase value information by performing phase calculation on the target frequency band information; The welding direction determination module is used to input the actual phase value information into the welding direction determination model, generate a characteristic difference value, and determine the microphone with the wrong welding direction based on the characteristic difference value.

[0073] Optionally, the phase calculation and feature generation module includes: The time domain signal segmentation and frequency domain transformation submodule is used to divide the target frequency band information into multiple time domain signals of fixed time windows, perform Fourier transform on the time domain signals of multiple fixed time windows, and obtain multiple frequency domain signals; A complex form extraction and initial phase calculation submodule, used to obtain corresponding complex forms based on multiple frequency domain signals, and calculate initial phase information according to the complex forms; The phase unwrapping processing submodule is used to perform phase unwrapping processing on the initial phase information to generate information after phase unwrapping; The time series weighting and actual phase value generation submodule is used to perform time series weighting processing on the information after phase expansion to generate actual phase value information.

[0074] Optionally, the complex form extraction and initial phase calculation submodule includes: A real and imaginary signal extraction submodule, used for obtaining a real signal and an imaginary signal from a complex number form; The initial phase calculation submodule is used to calculate the initial phase information using the following formula: , where θ represents the initial phase information, arctan represents the inverse tangent function, Re(f) and Im(f) represent the real signal and imaginary signal in complex form, respectively.

[0075] Optionally, the phase unwrapping processing submodule includes: A phase jump detection submodule is used to perform continuity check on the initial phase information and identify phase jump points between adjacent time points; The phase jump correction submodule is used to correct the phase jump point based on the following formula: , where θ 展开 (t) is the information after phase expansion, θ(t) is the phase jump point, and k is the adjustment coefficient at the jump point.

[0076] Optionally, the time series weighting and actual phase value generation submodule includes: The time weight calculation submodule is used to calculate the time weight according to the signal energy value in each time window after phase expansion. The calculation formula of the time weight is: , where w(t) represents the weight of the t-th time window, E(t) represents the square of the signal energy of the t-th time window, ∑ t E(t) represents the sum of the signal energies in all time windows; The weighted phase calculation submodule is used to perform weighted processing on the information after phase expansion based on the time weight. The weighted calculation formula is: , where θ 加权 represents the actual phase value information, w(t) is the weight of the tth time window, θ 展开 (t) is the information after phase unwrapping.

[0077] Optionally, the welding direction determination module includes: The actual phase feature extraction submodule is used for the welding direction determination model to extract the feature of the actual phase value information to obtain the actual phase feature; The feature comparison and difference calculation submodule is used to compare the actual phase feature with the standard feature of the normal welding direction pre-stored in the welding direction determination model point by point, and calculate the feature difference value between the actual phase feature and the standard feature.

[0078] Optionally, the welding direction determination module also includes: A normal welding direction deviation threshold extraction submodule is used to extract the deviation threshold of the normal welding direction from the welding direction determination model. The deviation threshold of the normal welding direction is used to define the allowable range of characteristic differences in the normal welding direction. The feature difference value determination submodule is used to compare the feature difference value with the deviation threshold of the normal welding direction. If the feature difference value is less than or equal to the deviation threshold of the normal welding direction, the welding direction is determined to be correct; if the feature difference value is greater than the deviation threshold of the normal welding direction, the welding direction is determined to be incorrect.

[0079] For the specific definition of the microphone welding direction detection device, please refer to the definition of the microphone welding direction detection method above, which will not be repeated here. Each module in the above-mentioned microphone welding direction detection device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0080] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig. 9 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a microphone welding direction detection method is implemented.

[0081] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: Obtaining acoustic feedback information of the microphone under test; Process the acoustic feedback information, remove high-frequency interference and background noise, and obtain filtered information; Extracting spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; By performing phase calculation on the target frequency band information, the actual phase value information is obtained; The actual phase value information is input into the welding direction determination model to generate a characteristic difference value, and based on the characteristic difference value, the microphone with the wrong welding direction is determined.

[0082] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Obtaining acoustic feedback information of the microphone under test; Process the acoustic feedback information, remove high-frequency interference and background noise, and obtain filtered information; Extracting spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; By performing phase calculation on the target frequency band information, the actual phase value information is obtained; The actual phase value information is input into the welding direction determination model to generate a characteristic difference value, and based on the characteristic difference value, the microphone with the wrong welding direction is determined.

[0083] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0084] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0085] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting microphone welding direction, characterized in that: The microphone welding direction detection method comprises: Obtaining acoustic feedback information of the microphone under test; Processing the acoustic feedback information to remove high-frequency interference and background noise to obtain filtered information; Extracting spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; By performing phase calculation on the target frequency band information, actual phase value information is obtained; The actual phase value information is input into a welding direction determination model to generate a characteristic difference value, and based on the characteristic difference value, a microphone with a wrong welding direction is determined.

2. The method for detecting microphone welding direction according to claim 1, characterized in that: The step of performing phase calculation on the target frequency band information to obtain actual phase value information includes: Dividing the target frequency band information into time domain signals of multiple fixed time windows, performing Fourier transform on the time domain signals of the multiple fixed time windows to obtain multiple frequency domain signals; Acquire corresponding complex forms based on multiple frequency domain signals, and calculate initial phase information according to the complex forms; Performing phase unwrapping processing on the initial phase information to generate phase unwrapped information; The phase-unwrapped information is subjected to time series weighting processing to generate actual phase value information.

3. The method for detecting microphone welding direction according to claim 2, characterized in that: The calculating the initial phase information according to the complex form comprises: obtaining a real signal and an imaginary signal from the complex form; The initial phase information is calculated by the following formula: , wherein the θ represents the initial phase information, the arctan represents the inverse tangent function, and the Re(f) and the Im(f) respectively represent the real signal and the imaginary signal in the complex form.

4. The method for detecting microphone welding direction according to claim 2, characterized in that: The performing phase unwrapping processing on the initial phase information to generate phase unwrapped information includes: Performing a continuity check on the initial phase information to identify phase jump points between adjacent time points; The phase jump point is corrected based on the following formula: , where the θ 展开 (t) is the information after the phase is expanded, θ(t) is the phase jump point, and k is the adjustment coefficient at the jump point.

5. The method for detecting microphone welding direction according to claim 2, characterized in that: The step of performing time series weighted processing on the phase-unwrapped information to generate actual phase value information includes: The time weight is calculated according to the signal energy value of the phase-unwrapped information in each time window. The calculation formula of the time weight is: , wherein w(t) represents the weight of the t-th time window, E(t) represents the square of the signal energy of the t-th time window, and ∑ t E(t) represents the sum of the signal energies in all time windows; Based on the time weight, the information after the phase expansion is weighted, and the weighted calculation formula is: , where the θ 加权 represents the actual phase value information, w(t) is the weight of the t-th time window, and θ 展开 (t) is the information after the phase is unfolded.

6. The method for detecting microphone welding direction according to claim 1, characterized in that: The step of inputting the actual phase value information into a welding direction determination model to generate a characteristic difference value comprises: The welding direction determination model extracts features from the actual phase value information to obtain actual phase features; The actual phase feature is compared point by point with the standard feature of the normal welding direction pre-stored in the welding direction determination model, and the feature difference value between the actual phase feature and the standard feature is calculated.

7. The method for detecting microphone welding direction according to claim 1, characterized in that: The method of determining the microphone with the wrong welding direction based on the characteristic difference value comprises: Extracting a deviation threshold of a normal welding direction from the welding direction determination model, wherein the deviation threshold of the normal welding direction is used to define an allowable range of characteristic differences in the normal welding direction; The characteristic difference value is compared with the deviation threshold of the normal welding direction. If the characteristic difference value is less than or equal to the deviation threshold of the normal welding direction, the welding direction is determined to be correct; if the characteristic difference value is greater than the deviation threshold of the normal welding direction, the welding direction is determined to be incorrect.

8. A microphone welding direction detection device, characterized in that: The microphone welding direction detection device comprises: An acoustic signal acquisition module, used to obtain acoustic feedback information of the microphone to be tested; A signal preprocessing module, used to process the acoustic feedback information, remove high-frequency interference and background noise, and obtain filtered information; A spectrum feature extraction module, used to extract spectrum feature information within a preset frequency range from the filtered information to generate target frequency band information; A phase calculation and feature generation module, used to obtain actual phase value information by performing phase calculation on the target frequency band information; The welding direction determination module is used to input the actual phase value information into the welding direction determination model to generate a characteristic difference value, and determine the microphone with the wrong welding direction based on the characteristic difference value.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the microphone welding direction detection method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the microphone welding direction detection method according to any one of claims 1 to 7 are implemented.

Images