Research On Acoustic-based Ceramic Structure Defect Detection Method

Ceramic culture is an important part of Chinese culture,reflecting the ingenuity of our working people.Since the reform and opening up,new breakthroughs were made in abundant colors and diversified functions during ceramic long development.In order to improve the economic benefits of the ceramic industry,it is necessary to strictly control its quality.Therefore,an in-depth exploration of the acoustics-based non-destructive testing methods for structural defects of ceramics is needed.The self-made experimental device is designed to collect the ceramic tapping signal and select the appropriate tapping strength,tapping position and sampling time to ensure the experimental effect.The IIR digital filter and wavelet denoising were used to eliminate the field noise.The experimental results prove that the experimental parameters such as percussion strength and percussion position have great influence on the quality of experimental data.Compared with IIR digital filter,wavelet denoising is more suitable for experimental data and has better effect.The time and frequency domains of the knocking signals are analyzed to find the classification basis of whether there are defects in ceramics.After comparing common eigenvalues such as waveform factor,skewness,and duration in the time domain,the duration was determined as the eigenvalue for classifying defective and non-defective ceramics.In the frequency domain analysis,a weighted overlapping average method is used to calculate the estimated power spectrum of the signal,and the peak point is the place where the most energy is concentrated in the knocked signal,and the difference between the two is obvious.So the peak data in the power spectrum is used as the eigenvalue for defect detection,and the main peak of the spectrum is also used as the frequency domain eigenvalue.Since manual processing of large amount of feature data is time-consuming and laborious,the nearest KNN algorithm,decision tree,and BP neural network were used to classify the two types of ceramics with and without structural defects.The experimental results show that all three classification algorithms have high accuracy,which confirms the feasibility of this non-destructive testing method,and the BP has the best classification effect.Figure 30;Table 16;Reference 60...