Approximate models in multivariable process control. An inverse Nyquist array and robust tuning regulator interpretation

Using the widely accepted philosophy that simple control design procedures are of great value in producing acceptable controllers for plant with a degree of dynamic and/or structural simplicity, particularly in the situation where closed-loop performance requirements are modest, recent work on the use of approximate models and plant step data in multivariable design is applied to produce simple, robust design algorithms for process plant which, in the presence of suitable precompensation, has `small¿¿ interaction effects and/or simple loop dynamics that can be approximated by delay-lag models. In one form, the results have a structure and generate a procedure identical to the inverse Nyquist array design technique, with Gershgorin circles replaced by `confidence circles¿¿ deduced from simple graphical operations on plant step data. Other interpretations of the work lead to estimates of the maximum gain required to retain stability in Davison´s robust feedback regulator. An optimisation procedure for choice of precompensator using step data is also outlined with strong connections with the method of pseudodiagonalisation.