Design of an analytical aggregation of rules of a diffuse controller and its application in the model of a nuclear research reactor

As they have gone being managed complex systems that fulfill tasks inside industrial or nuclear processes, it becomes necessary the development of technical novel of control, in which can incorporate heuristic knowledge of operation without to necessarily use the theories of classic control based mainly in mathematical models. One of the control techniques that allows to carry out this is the control based on diffuse logic. For the case of a model of the nuclear research reactor Triga Mark III of the National Institute of Nuclear Research have been developed diverse algorithms of diffuse control that have as objective the regulation of the neutron power in the nucleus. The aggregation stages and desdifussification in these algorithms discretize the universe of values of the control variable, being required a high number of operations for their execution. With the purpose of reducing this number of operations and to obtain results more exact in the generation of the aggregated group in each cycle of control and in the determination of the center of gravity of this added group, it is presented the development of an analytical method for these calculations. The main objectives outlined in this entitled thesis Design of an analytical aggregate of a diffuse controller rules and their application in the pattern of a nuclear research reactor, they are: to improve the behavior of control systems in closed knot based on diffuse logic by means of the development of an analytical method that determines an aggregated group resultant of the activation of rules in the diffuse controller and the obtaining of the exit variable using an exact solution of the technique of the center of gravity; and to compare the operation of these methods with those traditionally used ones that consider the discretization of the universe of the exit variable so much for the aggregation like for the desdiffusification. The chapters 1 and 2 present an introduction at two fundamental themes of the project. The first gives a general reference on the types of reactors that exist in the field of the nuclear industry, their main components, and finally, information more detailed on the Triga Mark III reactor. The second chapter explains the concepts that more are used in control systems based on diffuse logic, as the use of linguistic variables, diffuse groups and lastly, a description of diverse types of diffuse controllers. In the chapter 3 it is presented the design of a diffuse controller, denominated ''discreet'' type Mamdani that carries out a discretization of the exit variable so much for the stage of aggregation like for the desdiffusification stage. These stages have been incorporate to the control algorithms whose objective is the ascent and regulation of the neutron power of the reactor. In the same way the chapter 4 it presents with detail the algorithms developed of the diffuse controller, call ''exact'' that defines analytically the lines which conform the continuous aggregated group in each control cycle and the one calculation with exact formulas of the signal control that should be applied to the system in closed knot. At the end of each chapter they are presented graphics obtained in ambient Matlab that represent the obtained results in the variables ne (error of neutron power) and T (period of the reactor) in the simulations of both controllers. The chapter 5 presents the design of the communication interface created in Matlab for the remittance of a signal obtained inside of the main program of the diffuse controller. This signal arrives to an electronic system, presented in the chapter 6, that it controls the movement of two motors of steps to simulate the ascent or descent of the control bars of the reactor. Finally, the chapter 7 it presents all the comparisons carried out among the diffuse controller ''discreet'' and the diffuse controller ''exact'' through charts and graphic, which teach the results of this stage like they are: Count of instructions carried out by each controller, count of operations of floating type, time of execution of each function called by the control program and accuracy in the obtained values of the variables of interest of the system during the simulation in closed knot. (Author)