This paper describes two different control principles using simple combinations between basic control laws. The objective is to improve the performance (time response, overshoot, robustness…), without strongly increasing the complexity. These principles are tested on a d.c.-d.c. power converter. Linear and/or non linear controllers are mixed depending on the optimisation objective and on the decision data. The basic idea is that in the world of control, as elsewhere, “unity is strength”, i.e. complex functions could be carried out through basic object combinations. The first principle is based on the combination of two linear controllers. Only two models of the system are taken into account in this Multi Model Control (MMC) method. A new and specific procedure for model validity is put forward. It is based on the distance between two extreme models of the controlled system, with unknown severe disturbances. The second principle suggests the combination of a linear controller and a non-linear controller (a Bang Bang and an Integral and Proportional controller). This combination is created through the fuzzy fusion of state variables. The tuning of the fuzzy logic parameters is made through Hooke and Jeaves optimisation procedure. The performance is verified through experimental results on a d.c./d.c. converter application. They are significantly improved with respect to standard controllers, with only one voltage sensor and without any adaptive gains.
| Filename: | Unnamed file |
| Filesize: | 784.8 KB |
