The Applied Artificial Intelligence to the ControlElectronics Improving the Energy Efficiency ofVentilation Systems in the Industry

Abstract

In the furnaces of multiple hearths there is a problem related to the inefficiency of the current control of the combustion air system and the energy reserves present, since manually the excess air flow is regulated, while the fans work at full capacity all the time what produces an unnecessary expenditure of electrical energy. The experimental identification of the air supply system through artificial neural networks and fuzzy logic is executed for the future implementation of a control architecture with a view to increasing the energy efficiency of said process. Several experiments were carried out: first, in a laboratory, on an air flow test bench simulating in miniature the real process with a possible closed loop that links the input variables (sensors) with the speed of the motor (actuator: speed variator) ) and second, field, on one of the actual slabs of the plant, varying the number of revolutions of the motors in a staggered way around an operating point, alternating the operation of the fans with manual adjustment of the variable speed drives that they were installed previously. Of all the configurations tested, the most energy efficient variant resulted in the use of the two fans farthest from the common manifold (seven and eight), keeping the reserve nine. The proposed artificial neural model predicts the air pressure in the collector with 98 % correlation and 71 % in its generalization.