Adaptive Two-Stage Fuzzy Temperature Control for an Electroheat System

to many temperature analyses as a result of its simplicity. In [2], a PID controller is used for temperature control of a 20 KW industrial electric resistance furnace where the plant is modeled by a first-order transfer function with a time delay. Temperature control of a heated barrel with electric heaters and water coolers was proposed in [3] which used the discrete-time variable structure system (VSS) control scheme. In this paper, we construct a simple adaptive two-stage fuzzy temperature tracking controller for a commercial electroheat system. Being not well isolated from the environment, it is hard to build an analytical model for the system due to heat convection caused by a fan for heat mixing in the chamber, heat leakage to the environment, variation of the environment temperature, and uncertain nonlinear heating dynamics. Also, as a commercial product by using cheap heater coils and without equipping a cooling system, this system usually has a bad transient response such as a long rise time, a large overshoot, and a long settling time. Moreover, temperature tracking at the steady-state phase is not easy to maintain due to heat interaction with the environment. Here, in the presence of the unknown system dynamics, we use a two-stage fuzzy controller to improve the transient response. Furthermore, a simple fine-tuning adaptive control scheme is proposed to overcome environmental influence and ensure tracking of the temperature setting. Simulation study and experimental results show good performance of the adaptive fuzzy temperature control system. It is noted that a model based on the general energy balance equation does not have a good fit for a large chamber, because the distribution of heat is radiating to the space. With the modelfree approach, fuzzy temperature controllers were introduced in [4][5] which utilize two inputs (error, error change) to infer the fixed fuzzy rules and produce an output (duty cycle) to actuate the process. For crown temperature control of a TV glass furnace, a combination of PI control and fuzzy logic control has been proposed in [6] where the fuzzy system is used to determine the crown temperature setting. For superheated steam temperature of a power plant, a nonlinear generalized predictive controller based on a neuro-fuzzy network, which consists of local GPCs models of the plant, is proposed in [7]. A neural fuzzy inference network is proposed in [8] to construct a water-bath temperature control system which can automatically generate the fuzzy rules. The work in [8] is further improved in [9][10] by adopting a recurrent fuzzy network for improvement of the learning rate of the neural network.