Classic load model is conservative and inaccurate by only considering constant power load model, but actual situation can be described more precisely by using ZIP load model. The different load rate based on ZIP load model has a serious impact on static voltage stability. Due to the timevarying load rate, time-sharing modeling, which is identified for every 20 minutes in a day, can describe the variation of different load rate for the whole day more accurate than allday modeling, which is identified once by using all data in a day. In this paper, all-day ZIP load model and time-sharing ZIP load model are established respectively in order to justify the effectiveness of time-sharing ZIP load modeling. By using Linear-BP network, the model parameters can be identified more easily and precisely. A case study is performed by using data from Dalian University of Technology. The results verify that this network is more suitable to ZIP model parameters identification.