I. INTRODUCTIONPPLICATION of fractional calculus based control system designs have gained impetus in recent times due to the flexibility and effectiveness that can be gained through such methodologies [1][2]. These controllers use the FO integrodifferential operators as extra tuning knobs which can be tuned to meet additional design constraints and provide additional robustness to the design [3][4]. Merging the concepts of computational intelligence techniques with FO control designs are also being recently explored with encouraging results [5].However applications of fractional order controllers for electrical power and energy systems are still largely unexplored. A few studies have been done for the application of the FOPID controller to the design of the AVR in a power system. In [6][7], the FOPID parameters are tuned for an AVR system with swarm based optimization algorithms which show better time domain performance over that with conventional PID structure. In [8] a multi-objective formalism has been Manuscript received November 17, 2013; revised February 18, 2014 developed and the time domain performance trade-offs of various design objectives have been investigated. The obtained results are mixed, with the PID performing better under some objectives and the FOPID performing better at others. However, all these investigations suffer from a common problem. The objective functions are framed in time domain and the evolutionary algorithms minimize some time domain performance index. Therefore, the obtained results give no indication of other important design focuses like robust stability, disturbance rejection capability etc. To alleviate these issues, the problems in power system control can be framed in frequency domain and the system performance can be expressed in terms of system norms which need to be minimized or maximized. In [9], the FOPID design for an AVR system has been done in frequency domain. The conflicting objectives being maximization of phase margin (for better oscillation damping) and gain crossover frequency (for higher speed of operation) are studied by coupling them with multi-objective evolutionary algorithms. In this paper, we extend this concept to an exhaustive set of frequency domain design criteria for the AVR system and show the achievable set of performances with the FOPID controller structure.The rest of the paper is organized as follows. Section II briefly introduces the theoretical background of the problem with the basics of FO control using system norms. Section III shows the design trade-offs for several combinations of conflicting objectives and compares the best compromise solutions for each controller structure. The paper ends with the conclusion as section IV.
II. THEORETICAL BACKGROUND OF FRACTIONAL ORDER CONTROL DESIGN FOR AVR SYSTEM
A. Fractional Calculus Based Control SystemsFractional calculus extends the common notion of integer order integration or differentiation to any arbitrary real number.
B. AVR System and Controller StructureIn the conventional AVR loop in...