Simulation of reactor system situational control

Abstract

Development of conventional automated systems with a centralized structural organization, satisfying the applicable operational requirements in terms of an accurate solution to the control problem, speed and the required computing resources is usually difficult or impossible for such industrial facilities as complex reactor systems. This is explained by the fact that the emerging control problems are extremely complex due to the necessity of taking into account a large number of parameters that remain in diverse, numerous and complex relations. Hence, the creation of automated control systems based on the centralized principle of construction and holistic representation of the control object is often cost-intensive due to the need for significant computing resources to ensure its effective functioning. The elimination of these difficulties can be achieved by a structured representation of the reactor system and structuring the control problem. In such a case, this problem is assumed to be solved in the decentralized control system consisting of a set of local control systems that operate autonomously, and the coordinating body which correlates their functioning. The desired optimum for the reactor system as a whole is achieved as a result of information interchange between local control systems and the coordinating body. In the process of this interchange, local control systems solve optimization problems for individual elements of the reactor system taking into account the impacts of the coordinating body, while the coordinating body solves the global problem of the reactor system optimization as a whole by means of proper coordination of local problem solutions. The arising local optimization problems and the global coordination problem are much simpler than the initial problem of the reactor system control. Therefore, their solution is less expensive in terms of consumed computing resources. To realize the decentralized approach in this class of control objects, it seems appropriate to use a method of situational decomposition. This method provides ample opportunities for varying the control scheme, and it does not impose special requirements on the structure of the initial optimization problem as opposed to the classical decomposition methods. Verification of this assumption is performed through experimental research in the form of model reactor system control simulation with situational decomposition of the control problem.