IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING

Abstract – This paper presents an improved micro artificial immune (IMAI) algorithm utilizing basic concepts from swarm intelligence. In particular, to enhance the searching capability of the recently developed micro artificial immune system (Micro-AIS) algorithm, employed honey bees are recruited to provide high-quality antibodies for the working population of the IMAI algorithm. The proposed algorithm is used to find the optimal kernel values for the Volterra series model to identify nonlinear systems. To demonstrate the efficiency of the proposed method, three different types of nonlinear systems are considered, including a highly nonlinear rational system, a heat exchanger, and a continuous stirred tank reactor (CSTR). For all these systems, the IMAI algorithm has achieved accurate modelling results and fast convergence rates. Moreover, a comparative study was conducted with other optimization methods, namely the original Micro-AIS algorithm, the improved particle swarm optimization (IPSO), the real-coded genetic algorithm (GA), the least mean squares (LMS), the least mean p-norm (LMP), and the least mean absolute deviation (LMAD). From this comparative study, the proposed IMAI algorithm has achieved the best modelling performance compared to the other methods.

Abstract – The Recursive Least Squares (RLS) is usually utilized in control
applications as in self-tuning strategy to estimate the plant discrete-time transfer
function. Furthermore, it can be used as a tool to continuously monitoring the operating
condition of the plant under control. However, in such applications, the RLS should be
always in a “wake up” state so that it can estimate, in a few sampling time, the plant
transfer function after any abrupt change in its dynamic.
In this work, two modifications to the standard RLS are presented. The first
modification is called the “switching forgetting factor” while the other is called the”
resetting covariance matrix”. The two modifications are applied, under LabVIEW
environment, on-line to estimate the proper transfer function of a DC motor as an
example to show their capabilities to monitor the motor operation. It is found that with
these modifications, the RLS can estimate the plant transfer function much faster in
comparison to the standard RLS algorithm.