For noise uncertainty linear systems with time delay, a P-radius based zonotopic filter for time delay systems is proposed. First, the state feasible set is parameterized by zonotopes. Subsequently, depending on whether the current system state is affected by time delay or not, P-radius based zonotopic filters are designed respectively. By establishing zonotopic space expansion models, the state estimation problem is converted into a parameter optimization problem. Furthermore, by implementing Schur complement lemma, the optimization problem is solved by adopting linear matrix inequalities. Then, by means of dimensionality reduction operation, the tightest feasible set of polyhedron state space is calculated. Finally, a case simulation of a wind turbine system is applied, demonstrating the effectiveness of the proposed algorithm.