As cascading failure is difficult to predict， it will be helpful to prevent or mitigate cascading failure propagation if it could be detected in its initial stage. To identify the occurrence of cascading failure， a hierarchy model for complex event processing（CEP） and correlation analysis is developed based on CEP technology. For a typical cascading failure pattern and one with hidden failure in its initial stage， the temporal， spatial and electrical correlation between several successive failure events during the initial stage is analyzed. Different correlation rules are formulated among events. A case study shows that the method proposed is effective.