The Research Of Mine Microseismic Signal Classification And Recognition Method Based On SVM

Mine microseismic monitoring has been widely used in production as a means of safety warning and post disaster rescue.For excavate information as far as possible and know the characteristics of different earthquake disasters,in this paper,the classification and recognition of microseismic signals in mines are studied by combining theory with practice.The main contents include the following points:1)The applicability of wavelet and HHT method is compared from three aspects of decomposition and reconstruction,denoising and filtering and spectrum analysis.The results show that the two methods can effectively decompose and reconstruct the signal.The error is very small and can be neglected.In the noise elimination aspect,both of them can restore the test signal of adding white noise,but the reduction effect of the mutation point is poor.n the spectrum analysis,the two can get the time-frequency characteristics.The wavelet time-frequency spectrum contains a large amount of harmonic components,and has the phenomenon of energy leakage.The HHT transform is not affected by it,and frequency range is small.2)Relying on the experiment of Jixian coal mine in Heilongjiang,monitoring mine excavation activities by way of laying sensors on the ground.the influence of different loading speed and different gas concentration on the mechanical properties and microseismic characteristics of coal and rock is studied by laboratory test.The experimental results show that the strength of the coal and rock increases with the increase of loading speed,and the center frequency and dominant frequency of the signal is decress.When the gas concentration or the void pressure increases,the strength of the coal and rock will be weakened to some extent,but the central frequency and the dominant frequency will increase.3)A new method is proposed to accurately pick up the initial arrival of microseismic signals by introducing the concept of entropy.The sensitivity of the method is evaluated by sensitivity to amplitude,frequency,phase change,and sensitivity.The results show that the new method is sensitive to both amplitude,frequency and seismic phase changes.it is strong on anti noise effect In terms of parameter influence,the new method has many parameters,but some of them have no influence on accuracy and have little influence.4)The establishment method of mine signal database is put forward,and the concept of EMD energy entropy and marginal entropy is introduced into the feature extraction of mine vibration signal for the first time.Taking into account the S wave,P wave propagation speed is different,the energy is also different,the proposed method of △T characteristics of the pickup.The results show that the three methods can extract the signal characteristics better.5)Classification and identification of vibration signals in mine monitoring by SVM.The results show that the support vector function with EMD energy entropy,HHT marginal entropy and △T value method as the characteristic parameters can achieve more accurate classification,and the effect is ideal,and can be used in engineering practice.