Abstract
A case study on continuous process control based on fuzzy logic and supported by expert knowledge is proposed. The aim is to control the coal-grinding operations in a cement manufacturing plant. Fuzzy logic is based on linguistic variables that emulate human judgment and can solve complex modeling problems subject to uncertainty or incomplete information. Fuzzy controllers can handle control problems when an accurate model of the process is unavailable, ill-defined, or subject to excessive parameter variations. The system implementation resulted in productivity gains and energy consumption reductions of 3% and 5% respectively, in line with the literature related to similar applications.
Highlights
According to Wang (1999), the fuzzifier-defuzzifier controller is the most suitable for industrial applications, and this paper focuses exclusively on this controller
The execution and transfer of the machine decisions to the process are not instantaneous and the expert system considers the natural inertia of the system it controls
The technology holder highlighted several other applications with gains from 6% to 10% in productivity and up to 3% in energy efficiency
Summary
According to Legg and Hutter (2007), intelligence is the demonstration of coherent principles used to adapt to a complex environment with verifiable results. CBR solves new problems by adapting previously confirmed solutions using the knowledge generated in the past experiences. CBR solves a new problem by the adaptation of the solution of an old case. The use of FL to control industrial processes was first proposed by Mamdani (1974), using fuzzy sets, linguistic terms, conditional sentences, and inference rules. Fuzzy controllers have been demonstrated to have an adaptive capacity to handle control problems when an accurate model of the process is unavailable, ill-defined, or subject to excessive parameter variations (Bose, 1994). According to a block of rules representing the expertsknowledge, an inference engine calculates incremental values ∆u to impose on the input variable u, achieving an expected result on x according to an established control strategy. A routine of selection, mixing, and improving the rules provides a new solution that increments the database
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have