الفهرس | Only 14 pages are availabe for public view |
Abstract The elastic coupled multi-motor systems (ECMMS), and the elastic coupled twomass systems (ECTMS), elastic coupled multi-mass systems are common examples of the nonlinear, multi-input / multi-output (MIMO) complicated systems used in industry. In such system mechanical vibrations are resulted due to mechanical coupling through shafts which are coupling among multi-motors and single or multiloads of the system. One source of nonlinearity in such type of systems is mechanical coulomb friction which may be associated to the operation of ECMMS or ECTMS. Another source of nonlinearity is the limitations of electrical or mechanical actuators. Also, the backlash in the gear box coupling the motor shaft to the load is another source of nonlinearity. Moreover, the torsional oscillations and torsional resonance which can arise within the MMS can cause various damages in the system. The source of such torsional oscillations is the elastic long and / or short shafts of the mechanically coupled motors and the rotating large or variable inertias included in the system. So, this thesis presents proposed control algorithms in order to equalize against or reduce the resulted damaging resonance oscillation by designing a decentralized control system for speed control of the multi-mass systems (MMS). The core controllers employed in the proposed decentralized control system is the self-adaptive, intelligent, and simple structure proportional, integral, and derivative (PID) controllers. The thesis presents a proposed algorithm to auto-tune the parameters of the included PID controllers. The proposed algorithm depends on searching a whole defined space of the control parameters in order to acquire the best values for optimum control effect and maintain better performance index as possible. One of the algorithms is off-line; it is applied on the model of the ECMMS system. The main objective is to perform a searching mechanism in the predefined subspace of the entire space of controller parameters and computing the cost function for each vector of experienced parameters. The final emerged output of the algorithm is to deliver the vector of controller parameters for which the value of 7 the is minimized, over the total period of operation. The values of the cost function obtained for each subspace are collected in a table to be compared and, then to get the optimal values of the controller parameters over the entire space. This off-line algorithm is modified to the optimal parameters of controller parameter vector depending on the initial value of the error at the start of system excitation or re-excitation, the controller parameters are clustered in a reference table. The reference table is passed to be employed by one of the proposed practical algorithms to be referenced automatically for the optimal parameters depending on the measured values of the reference and the actual variable of the system. Two additional proposed practical algorithms which are intelligent and selfadaptive are presented through this thesis, with the practical results obtained from their application on the system. These algorithms combine altogether the advantages of using the common used PID and the intelligent neural network based controller (neuro -controller) or (NC). An additional algorithm of adaptation of an important parameter of the employed (NC) is applied practically to improve the transient and the steady state response of the system. The adapted parameter is the learning factor of the NC. The proposed adaptation algorithm is based on Lyapunov stability analysis in order to guarantee bounded characteristic of system response. The results emerged from adopting the proposed control techniques either the simulation or the practical results have supported each other towards the success to suppress the evolution of oscillatory frequencies and mechanical vibration modes. The results also supported the capability of the proposed control algorithms to effectively compensate against the load disturbances which are introduced to the related ECMMs. Although the results provided are better for some techniques rather the others, but all of those proposed techniques won to restrict the effect of mechanical elasticity and stiffness on the ECMMS. The proposed techniques have shared the common characteristic of having simpler structure, processed rapidly to conform to the short time constant motors of ECMMS. |