The batch process data is a kind of special time series, in engineering. Because of the discontinuity of the observation data, the fault detection technology for batch processes data has always been a difficult problem. In order to solve this problem, this paper constructed a fault detection algorithm based on the batch process model, and particularity of this type of data was fully considered in the algorithm. At the same time, in order to prevent abnormal data's influence on the algorithm, the algorithm fault-tolerant processing. Finally, an example was analyzed in detail, and the result shows that the model is correct.