Disclosed is a state estimation method for a power battery formation process based on convex space filtering, belonging to the technical field of power battery manufacturing. The method performs state estimation on a time delay system by a filtering method, and an iterative replacement method is provided for converting the state quantity at a time k to the state quantity at a time k−h and subsequent items, so as to combine time delay items, thereby avoiding the problem that the dimension is increased when a state matrix A and a state matrix Ah of a time-delay state quantity are subsequently combined into a new state matrix, and reducing the computation complexity and computation time in subsequent computations. Moreover, the estimation accuracy is also improved to a certain extent because of the cancellation of the same items in the iterative replacement. In addition, the method of this application uses two times of update when obtaining an update step, so that the obtained convex space is wrapped more compactly, so as to improve the state estimation accuracy for the battery formation process.