Abstract：EH–4 is the update numeral instrument of continuous conductivity imaging system with electromagnetic method. The numeral signals transferred by EH–4 must be transformed into frequency ones. Fourier transform is a kind of full analysis in frequency domain without any analysis function of time domain，and fails to consider the problems of unstable signals caused by noises. It is necessary to cull the noises. The original method culling noises is surveying and its mathematical expectation is used. Its drawback is obvious，that is ，all the signals are not examined and therefore some unstable morbid signals are joined into the statistical analysis. The wavelet analysis method means analyzing each signal segment using wavelet function in different scales，and culling or revising unstable morbid signals. This method has full analysis function in both frequency domain and time domain. Through the study of discrete-time orthogonal wavelet decomposition，the signals observed by EH–4 are decomposed into different scale elements and analyzed in both time domain and frequency domain；subsequently the stability and quality of signals are ascertained in different scales. The study shows that the wavelet analysis method can analyze signals in different scales，detect unstable morbid signals and revise or cull them. It is one of effective means to analyze and detect signals in different frequency scales，and is useful to interpret signals all-round and accurately.
