The digitization of the pump characteristic curve between lift and flow rate is crucial for state detection, fault diagnosis and optimal control of large watering and drainage system in modern industry. In this paper, the characteristic equation of pump is analyzed, an adline neural network is designed for parameters identification for the pump characteristic equation. The proposed scheme solves the question how to measure the characteristic equation in arbitrary speed, raises the identification precise with adaptive filtering, learning, and approaching function of adline neural network. The experiment results confirm its feasibility and effectiveness.