Flood water level modelling using Multiple Input Single Output (MISO) ARX structure and cascaded Neural Network for performance improvement

摘要

Flood water level prediction using system identification technique is still new area for most of the researchers. This is due to the dynamics of the flood water level itself that is often characterized as highly nonlinear. Thus, it is quite a challenging task to represent the flood water level behavioural in mathematical expressions. This paper presents flood water level modelling using MISO (Multiple Input Single Output) ARX (Autoregressive Exogenous Input) structure and cascaded Neural Network model for performance improvement. In this paper, the transfer function relating the input parameters and output parameter was identified with the aid of MISO ARX model. The input and output parameters are based on real time data obtained from Department of Irrigation and Drainage Malaysia. However, the MISO ARX performance result is not quite impressive to look into. Hence, Neural Network model is cascaded to the MISO ARX model to improve the result. Simulation results show that the proposed cascaded model provides improved prediction performance.