Flexible load adjustable interval prediction combines non-intrusive load identification and big data mining, big data mining obtains the probability characteristic curve of flexible load power, and then based on non-intrusive method to identify the actual opening period of flexible load in flexible load. The probabilistic characteristic curve of power intercepts the curve located in the time window of the period, and finally aggregates the multi-family households to obtain a flexible load-adjustable interval prediction utility model. Compared with the traditional method of installing the detection device, the non-intrusive identification pays more attention to the privacy of the user and also saves the cost of installing the device. Then, the relationship between the energy habits and the influencing factors is obtained by Apriori rule quantification, and finally the probability characteristic curve of the flexible load power of the resident users is obtained. Combining the two can obtain the flexible load adjustable interval prediction model.