Abstract: Alarm flooding is recognized as a serious problem in modern process industry. To find root alarms and reduce invalid alarms so as to tackle the low efficiency of the traditional methods, we propose an alarm correlation analysis approach based on alarm time information mining. Taking advantage of the block method of alarm time series, we convert the alarm data sequences to the time node sequences, and then we utilize the matching degree between the temporal blocks as the evaluation criterion of the alarm correlations, which can reduce the calculation burden of the correlation analysis. Additionally, we calculate the temporal information associated with the correlated alarm time series through contrasting sliding windows. The Tennessee Eastman (TE) process are employed to illustrate the benefits of the proposed method.