Abstract: Aiming at the misdetection or missed detection of multi-range household appliances during event detection, a sliding window event detection method based on density-based local outlier factor detection and coefficient of variation is proposed. First, the load characteristic sample space is constructed. The time domain characteristics of the steady-state current waveform after the load switching and the frequency domain characteristics of the load current after fast Fourier transform are extracted, and these sample sets are input to the support vector machine model for training, and finally the classification and recognition of household appliances is realized. Experimental results prove that the accuracy of the method for detecting switching events of multi-range household appliances reaches 97.6%, and the accuracy of load identification reaches 92%. In actual application scenarios, this method will have a better detection effect considering the simultaneous working of multiple household appliances.