The invention discloses an electric load model constructing method, device and equipment. A random Gaussian measurement matrix is used to perform compressed sensing sampling on an electric load signal, the amount of collected data can be greatly reduced, therefore, storage and communication resources are saved, and the transmission speed is accelerated. After receiving a low-dimensional signal after the compressed sensing sampling, the signal is accurately recovered by using a Fourier-based sparse matrix recovery algorithm and a wavelet-based sparse matrix recovery algorithm according to a condition whether the low-dimensional signal belongs to a steady-state signal or a transient signal, a load static model and a load dynamic model are respectively constructed according to restored signalparameters, the models have the advantages of small storage, fast transmission and accurate modeling, and the technical problem of the lack of a precise electric load model with small storage and fast transmission is solved.