Local feature scale decomposition（Local characteristic scale decomposition

LCD）is a new adaptive time-frequency analysis method

which can adaptively decompose a complex signal into a number of ISC（Intrinsic scale component

ISC）components.SVM（support vector machine

SVM）is a kind of intelligent classification methods of machine learning

Shannon entropy is a nonlinear statistical learning methods.The LCD and SVM are introduced to the fault classification of mechanical transmission system.The Hilbert demodulation algorithm is used to strike the envelope signal of ISC components

through which the Shannon entropy is structured.After an in-depth analysis on the basis

it is combined with SVM in good relationship.The result is input vectors of SVM classifier

the trained SVM is used to determine the fault location

the type or degree of the test sample.Through the experimental analysis and verification

using this combining method

the recognition rate of four operating states which contains the normal state

the root crack of tooth

the wear of tooth surface

the composite fault is up to 97.5%.