The paper discusses methods, primarily those using frequency-domain techniques, for the analysis and design of nonlinear feedback control systems. The behavioural properties peculiar to nonlinear feedback systems are first discussed. This is followed by a review and discussion of the applicability of absolute stability criteria and describing-function methods for single-variable and multivariable systems. The calculation of limit cycles in relay systems, time-domain methods of analysis and simulation techniques are then considered. Finally, a few applications are considered which highlight both the applicability and limitations of the available analytical techniques for nonlinear systems.
IntroductionThe aim of this paper is to give an overview of methods available for the analysis and design of nonlinear systems. As a consequence of the distinct behaviour of nonlinear systems, most methods of analysis are directed at solutions to specific problems, such as system stability or the existence of limit cycles. Unlike the case for linear systems, the approach may be of limited usefulness for answering other questions of concern, such as, for example, the time-response behaviour. Further, in several methods, the assumptions made are based on the expected form of the solution, so that some knowledge of the possible forms of nonlinear behaviour is a requirement for the analyst. Because of the limited applicability and the approximations often involved in methods of nonlinear analysis, it is important for the engineer to be familiar with a variety of techniques, to appreciate their limitations, and to know how to combine their useful attributes. The nonlinear features of a system may originate inherently or intentionally. Inherent nonlinearity is an inseparable characteristic of the laws governing the operation of the system to be controlled. This situation has frequently been handled by considering the linearised performance about one or more system operating points, often resulting in loss of accuracy, unreliable predictions and the need for costly testing. Intentional nonlinearity, on the other hand, is deliberately introduced into the design of the system as the best way to achieve the desired performance. Typical examples of this are the use of on/off time-optimal position controls [1 ] and the application of jet thrusters in the attitude control loops of current communication [2], and some scientific [3], satellites. The increasing use of microprocessor controllers will, without doubt, result in the more frequent use of nonlinear control strategies as, in contrast to analogue controllers, they allow complex nonlinear algorithms to be easily implemented. The requirements placed on process-controller algorithms to achieve both satisfactory control of set-point changes and regulation are rarely compatible, and thus an opportunity for the use of nonlinear techniques clearly exists. Some nonlinear controllers which have recently become available [4,5] have been designed primarily on the basis of mimicking the behaviour ...