A new method of gearbox fault diagnosis based on SVM（Support vector machine） and GA（Genetic algorithm）which is used to optimize parameters is presented.Firstly

the raw vibration signal is preprocessed by Time Synchronous Average algorithm.Then

the signal wavelet packet decomposition is carried out

standard deviation of wavelet packet coefficients of the signals is considered as the fault feature vector

and the normalization process of the fault feature vector is carried out.In the end

the fault feature vector is used as the input of SVM.In this process

the Daubechies order

wavelet packet decomposition level

c and g of SVM are optimized by GA.After that

the optimized parameter is used in training model which will be used for fault diagnosis.The experimental result shows that SVM and GA can be used to effectively diagnose faults of gearbox.