Application of Decentralized Control to a Laser Alignment System

Light propagating through a large laser system is subject to numerous perturbations which can cause deviations from the desired path. Closed loop alignment controls are typically located at key points to compensate for such errors. We have tested a decetralized approach to the alignment control of such a laser system. The decentralized control system has a modular structure in which the control command for each subsystem is a function of only two errors, the error detected by that subsystem's own sensor and the error detected by the sensor in the preceding subsystem. Therefore, although the controller uses local information, the whole alignment system can operate in a coordinated fashion. On the other hand, any design change of a subsystem will cause redesign of only one additional controller, the one for the subsystem immediately downstream. The rest of the system remains unchanged, as we verified in the laboratory by disabling one control loop to simulate a change in the subsystem. The control accuracy of the whole system remained the same after the controller downstream of the disabled loop was modified. This modular approach increases design flexibility and facilitates future expansion of the system.