Implementation of Transfer Function From System Identification Tools For PID Controller Design For Ammonium Removal In Sand Biofilter

This paper discusses the design of a PID controller for ammonium removal from sand biofilter using transfer function estimated from system identification tools in MATLAB. Determining the transfer function of a dynamic system is challenging because it requires the development of an accurate mathematical model. The ammonium removal from sand biofilter is a nonlinear process with multiple input parameters, namely the initial concentration of ammonium (X1), initial concentration of manganese (X2), retention time (X3), and aeration rate (X4) with one output parameter, i.e., ammonium removal (Y1). The sensitivity analysis was used to choose the most influential input parameter to develop a one-input and one-output PID controller. An artificial neural network (ANN) with a back-propagation training algorithm was used, and the result shows that X1 was the influential input parameter based on the R2 and MSE values of 0.7851 and 30.4121, respectively. The transfer function was selected in system identification tools based on best fittings (98.55), i.e., FPA1. The feedback closed loop PID controller was developed in a Simulink environment, and the optimum controller setting obtained showed that the output response was able to reach a set point and was stable. The developed PID control resulted in a stable output response resulting from a stable controller design. The minimum IAE and ISE values for the set point test were 2.217 x 10
¹
and 0.612 x 10
¹
, respectively. The load disturbance test values were IAE= 1.505 x 10
²
and ISE= 6.426 x 10
¹
.