A novel improved hybrid approach for order reduction of high order physical systems

The paper presents a novel hybrid approach for the simplification of higher order models to obtain the lower order model. The proposed method presented in this paper is used to reduce the single and multivariable systems by amalgamation of pole clustering technique and genetic algorithm. The improved pole clustering technique is achieved by Lehmer measure which is used to obtain the denominator of reduced model and the coefficients of reduced order numerator are obtained by genetic algorithm approach. To compare the newly developed hybrid approach with the previous literature of model order reduction, relevant examples are illustrated here. Three examples of single variable and one example of multivariable physical systems are tested by using MATLAB 2017a software and its control system toolbox. The performance parameters like integral square error, integral time absolute error, integral absolute error, overshoot, peak time, settling time and steady state error are used. Gain margin and phase margin is also obtained from bode plot analysis to study the stability measures of the reduced system. The comparative analysis shows that the proposed method performs far better in achieving the reduced value of all mentioned errors, better approximation of time domain characteristics than the previously developed techniques as reported in literature and hence helps to obtain more accurate reduced approximation of higher scale physical systems.