Citas bibligráficas
Kobashicawa, J., (2006). Diseño de un controlador por lógica difusa para una planta de trituración secundaria y terciaria [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/2402
Kobashicawa, J., Diseño de un controlador por lógica difusa para una planta de trituración secundaria y terciaria [Tesis]. : Universidad Nacional de Ingeniería; 2006. http://hdl.handle.net/20.500.14076/2402
@mastersthesis{renati/705702,
title = "Diseño de un controlador por lógica difusa para una planta de trituración secundaria y terciaria",
author = "Kobashicawa Chinen, Juan Antonio",
publisher = "Universidad Nacional de Ingeniería",
year = "2006"
}
In this thesis, a fuzzy logic controller has been designed for the secondary and tertiary crushing plant of Southern Peru Copper Corporation (SPCC) at Toquepala’s division. In this plant it is required to control the product size distribution fixed in 15 % in weight over half inch with maximum production (i.e. tons per hour) taking care of not overloading crushers (i.e. amperes must be under an operation limit) and remaining mineral levels constant on crusher’s bowls. This control has been done under disturbing events, which consist on the stochastic variation of the feed size distribution and Work Index (related with mineral hardness). To accomplish this, all the plant was mathematically modeled. Conic crushers were based on Whiten’s model, vibrating screens on Ferrara-Pretti-Schena’s model, mineral feeders on a non-linear relationship (i.e. sigmoid), charge distributors on a linear relationship. Disturbing events were modeled assuming that the value of in a determined moment may vary in a range of ±5 % respect to the value of the previous moment. In feed case gamma distribution was used, which was determined after making regressions of feed size distributions, for this, ten kinds of functions or distributions of common use in mineral processing were applied. Crusher parameters were estimated by non-linear regression, while Genetic Algorithms (GA) were used for vibrating screens because 30 parameters had to be estimated (for each screen). In the controller design, fuzzy sets and rule base were developed under autor’s criteria based on Mamdarni’s system. Simulations were executed under different conditions with bowl levels Set Points constant (to 70 %) and product size Set Point varying according a square function between values of 15 and 20. In those simulations, different methods of fnzzyfication were evaluated (singleton and no-singleton), fuzzy operators, implication and dcfuzzyfication methods and Takagi- Sugeno controller were also evaluated, this controller was based on Mamdani’s system an whose parameters were estimated by m^2sug of MATLAB 7,0 (zero order system) and by recursive least squares (RLS) method in which a first order system has been obtained. Finally simulation was executed following the Mamdani’s system with a Set Point equal to % + 1/2” = 15 winch is what is required in plant. Moreover, it has tried to optimize the control system (i.e. rules consequents) using Genetic Algorithms. Lamentably the system could not control bowl levels and crusher were overloaded. This is an important point due to weighting factors in fitness function, genetic operators and/or mechanisms, coding, etc. A bad choice in any of such factors will produce low performance and wasting time in optimization process.
Este ítem está sujeto a una licencia Creative Commons Licencia Creative Commons