Research On Monitoring Algorithm Of Coal Mine Compressed Air Water Supply Network Based On Time Frequency Image Feature Extraction

In the process of coal mining,the most important problem is mining safety.In order to maintain water supply and ventilation,many pipelines are laid underground.These pipelines crisscross,once the leakage occurs,it is difficult to find it manually.If the leakage can not be detected in time,it will seriously threaten the safety of underground workers.Therefore,it is very necessary to detect the leakage of pressurized air water supply pipeline in coal mine.Most of the traditional pipeline detection methods deal with one-dimensional vibration or pressure signal to obtain the partial characteristics of the leakage signal,so as to determine whether the pipeline has leakage.Compared with one-dimensional signal,two-dimensional signal can highlight the micro characteristics of the signal,so in this study,the two-dimensional image of the signal is obtained by using time-frequency analysis method,and whether the pipeline leakage occurs is judged by extracting the characteristics of the two-dimensional image.Firstly,the time-frequency analysis of the collected pipeline vibration signal is carried out.Through the study of traditional time-frequency analysis methods,it is found that the short-time Fourier transform is limited by the time and frequency resolution,and the wavelet basis of wavelet transform can not be simply selected.Wigner Ville distribution is the most basic and widely used method in time-frequency analysis.Compared with other methods,Wigner Ville distribution can display the energy distribution of signals.The cross term interference can be suppressed and eliminated by adding kernel function,that is,smoothing pseudo Wigner Ville distribution.Therefore,based on the Wigner Ville distribution,the smooth pseudo Wigner Ville distribution is used for time-frequency analysis of pipeline vibration signals.Secondly,gray level co-occurrence matrix is used to extract the features of time-frequency image.The time-frequency image is transformed into gray-scale image by grayscale processing.In order to diagnose whether the pipeline leakage occurs,the feature extraction of the image is needed.In this paper,the gray level co-occurrence matrix method is selected.Then,the energy,correlation and contrast of gray level co-occurrence matrix are calculated and input into support vector machines(SVM)for training and diagnosis.Finally,the minimum entropy deconvolution(MED)algorithm and cross-correlation algorithm are combined to locate the pipeline leakage point.On the basis of the traditional pipeline location method,signal noise reduction is added.In this paper,a variety of signal denoising methods are studied,and it is found that compared with other algorithms,Med algorithm can not only reduce the noise interference,but also highlight the effective components of the signal,making the pipeline positioning more accurate.According to the underground pipe network structure of coal mine,the pipe network system model is built.Through the simulation of pipeline leakage,the detection algorithm and positioning method are tested,and the test data are analyzed.The experimental results show that the system has good performance and can meet the requirements of leakage detection and location of coal mine fluid pipe network.