Research On Dual Control With Probing For The Minimum Variance Problems

In the control field, the control parameters changed constantly with the change of the control environment and the impact of outside interference, which makes the controller uncertainty. For the uncertain parameter stochastic system, the stochastic optimal control strategy which obtained by optimize the performance index contained uncertain measurement has the dual property. On the one hand, the control signal can keep expectations on the System outputs and states, called regulatory function. On the other hand, it can be utilized for probing future information, to reduce the uncertainty of the system, called learning or probing function. However, to solving the dual control problems, it need to solve the Bellman equation which the analytical solution can not be obtained commonly, even the numerical solution can only be obtained in a few simple instance. From this it generated many sub-optimal solutions, but how to designing dual controller problems were not solved entirely; solving the problem of dual control still needs a lot of effort.Firstly, the basic knowledge of adaptive dual control were introduce, analyzed the basic principles of adaptive control and provided an overview of the classification and research of adaptive control.Secondly, a dual adaptive control based on innovation was introduced. Aiming at the learning factor cannot be adjusted in the process of the innovation dual control; a new strategy that the learning factor can regulate with variance of the parameters was advanced. Simulation results show the advantage of the strategy, and it has the better dual property.Finally, a dual control strategy that add additional perturbation in caution controller were advanced. It can change the perturbation signals with the variance of parameters, achieved the balance between the identification of system parameters and following control objects, avoided the "turn off phenomenon which occurred in using caution control singly. Compare with other sub-optimal dual control. The strategy has advantages of small amount of calculation, easy to implement, flexible designed dynamic performance and tracking parameters.