System Identification Error Computation a Plant with Neural Network

Today, there are many systems in the industry which their performances can be detected by using input -output of systems. To do this, they use system identification algorithms. n this paper, the differential equation of input and output a plant is investigated. We used three methods to identify the target system; Levenberg Marquardt, Recursive Batch Back Propagation and Memory-saving implementation. Finally, it was observed that the Memory- saving implementation method has the lowest value of MSE and Final Prediction Error and Recursive Batch Back Propagation method has the lowest Error in last iteration.

Keywords:

System Identification , Plant , Neural Network