In this paper semi-active control of the suspension of an all-terrain vehicle is considered. A seven degree of freedom suspension model is presented first. A fuzzy approach for controller synthesis is then proposed. Expert knowledge is stored in the form of IF-THEN rules. The Takagi-Sugeno inference system is employed, with triangle membership functions. The fuzzy system output is the damper coefficient. In contrary to many other control algorithms, the presented fuzzy algorithm does not require inverse modelling of the magnetorheological damper. Instead, some scaling parameters are set. They can be chosen experimentally, or a bio-inspired strategy can be applied. The fuzzy control is then compared with the Skyhook control in simulations, in terms of road holding and driving comfort indicators. Obtained results are similar. However, lack of necessity to use an MR inverse model allows the fuzzy system to provide successful performance in case of different operating conditions, what is an important benefit.

For many adaptive noise control systems the Filtered-Reference LMS, known as the FXLMS algorithm is used to update parameters of the control filter. Appropriate adjustment of the step size is then important to guarantee convergence of the algorithm, obtain small excess mean square error, and react with required rate to variation of plant properties or noise nonstationarity. There are several recipes presented in the literature, theoretically derived or of heuristic origin.This paper focuses on a modification of the FXLMS algorithm, were convergence is guaranteed by changing sign of the algorithm steps size, instead of using a model of the secondary path. A TakagiSugeno-Kang fuzzy inference system is proposed to evaluate both the sign and the magnitude of the step size. Simulation experiments are presented to validate the algorithm and compare it to the classical FXLMS algorithm in terms of convergence and noise reduction.

The Filtered-Reference LMS, also known as FxLMS algorithm, is one of the most commonly used adaptive algorithms for noise control systems. It is appreciated due to its simplicity, low computational complexity, and performance efficiency. For its convergence, responses of the acousto-electric secondary path and its model should not differ by more than pi/2 for any frequency contributing to the noise being controlled. Good amplitude matching is, in turn, responsible for convergence rate. Thus, without an acceptable model, the FXLMS algorithm is vulnerable to high excess mean square error or even divergence due to updating control filter parameters in improper direction. The literature presents several recipes, of heuristic or theoretic origin, to cope in circumstances of inaccurate secondary path modeling, or when the path is subject to change during control system operation and thus differs from the model. They usually require additional wideband random excitation to allow for on-line modeling during control system operation. Such excitation deteriorates, however, overall noise reduction results, and the acoustic impression perceived by the user is poor. This paper focuses on an active noise control approach, which requires no secondary path modelling. The convergence is guaranteed by switching sign of the algorithm step size and, in this way, the sign of the parameter update term. It is combined with on-line tunable delay of the reference signal to significantly improve convergence properties of the algorithm. Theoretical justification for this approach is shown for a tonal noise. Then, the method is extended for the case of a narrowband noise. Theoretical consideration is validated by simulations based on data acquired from a real application.

No abstract

An active casing made of appropriately controlled vibrating plates can be used to reduce noise propagating from the mechanism enclosed in the casing. Since a practical vibrating casing can behave in a nonlinear way, the performance quality strongly depends on the ability of control filters to compensate for the nonlinearity. The classical approach to nonlinear active control, e.g. based on the Volterra filters, can deal with harmonics generated by the nonlinearity. However, when a complex structure is considered, neural networks have a higher potential. Although, they are much more computationally demanding, for some cases they can be simplified and still provide acceptable performance.In this paper, results of control obtained for a real casing with multiple actuators exciting each wall are presented and discussed.