Aiming at the complexity of Substation Scene and the serious interference of electromagnetic noise and natural climate on foreground target detection, an intelligent identification system and method for abnormal event detection oriented to Substation Scene optimization are proposed. Combined with the substation monitoring objects and equipment, the abnormal condition monitoring strategy is designed. Based on the system architecture, combined with the improved Gaussian mixture background model and wavelet change principle, the abnormal target events are extracted and denoised. The spatiotemporal gradient model is used to improve the efficiency of anomaly target feature extraction, and the hidden Markov model is used to monitor the anomaly of surveillance video. Through field experiments and application verification, it is proved that the proposed method has higher accuracy and better detection effect than the traditional Gaussian mixture model method.