Optimization And Transfer Analysis Of Input Value Combinations For Multivariable Systems

There are a lot of input combinations which lead to the same main outputs in multivariable systems. Taking it all round, we can find that the superiority of different combinations are usually different. In general, the bigger the possible dynamic range of the system input is, the greater the ability to process and compensate system inputs will be when disturbances occur with different types, and so do the higher superiority of the corresponding input combination. To get a dynamic range as large as possible so that it can respond to all possible disturbances, we suggest two multiobjective optimization models for selecting and determining the optimal inputs combination. Firstly, a function is presented to represent the flexibility of decision variable in static situation. And two functions are proposed in dynamic situation. One is employed to describe the adjustable dynamic range of the inputs; another can show the level of participation in the control process. And then the new functions are integrated into the original multi-objective optimization control model with two ways that one is hierarchical and another non-hierarchical. Finally, two new models are theoretically analyzed and compared.Because of the complexity of the multivariable systems, in the control process, people often focus only on the mostly target for some reason while ignore the balance of control input combinations. That will result in the increase of all kinds of loss in the system, or constraining the ability to prevent dynamic jamming. Thus, it is necessary to transfer a combination of input values to one which can balance factors in indifferent kinds while the transferring process is smoothly and the main output targets are kept in an allowed range. Based on neighborhood system theory, it is proved that a system process in limited period of time can be completely covered by finite feasible neighborhoods in some time points. Based on this point, methods are given for obtaining transferring channel in different combinations of input values, and every channel was covered by overlapping neighborhood of finite number of different input point. When the control points transfer in the channel, it can ensure the output of the system within a predetermined range. For three different cases in control systems, i.e. setpoint control, process control and servo control, we respectively give transfer method of control input. Finally, the presented methods are used to the matching control of shift and speed of an intelligent car. Simulation results confirm the validity of the method.