A dual-kernel independent component analysis (DKICA) algorithm for chemical process fault diagnosis based on kernel principal component analysis (KPCA) and kernel independent component analysis（KICA） was proposed. First, this algorithm uses nonlinear kernel function of KPCA to whiten preprocessing data by mapping the original space into the high-dimension feature space. Then, the KICA algorithm deals with the data while statistical indices of fault monitoring are obtained and control confidence limits are calculated in the feature space. The proposed algorithm was applied to the continuous stirred tank reactor (CSTR) process. The results indicate that the algorithm can effectively increase the accuracy and reduce the false negative rate and false positive rate of fault diagnosis for nonlinear chemical process.