Implantação e avaliação do controle avançado em um processo de flotação com o auxílio de instrumentos de visão artificial em tempo real

Abstract

Flotation performance depends on the interaction of many physical and chemical factors. Therefore, flotation circuits are subjected to diverse types of process disturbances, some of which are caused by changes in mineral characteristics and others by variations in operating conditions. Besides, measurement errors and damaged equipment add further difficulties. Thus, the demand for simple and reliable control systems is essential. Therefore, flotation control based on froth characteristics, using a combination of artificial vision and intelligent control with fuzzy logic techniques were used in this work, which aimed to implement and evaluate the Advanced Process Control in a flotation circuit, containing a rougher stage, a cleaner stage, and a scavenger stage. Also, another objective of the work was to present the correlation between the data of concentrate grade generated by real-time image analysis and the data of laboratory analysis. The methodology used in this work involved the installation of cameras and lights on the columns, the sampling and calibration of the image analysis software, the implementation of Advanced Process Control, and the impact evaluation of this system on the grade and recovery parameters. In general, the methodology consisted of using the information provided in real-time by the image analysis software to optimize the flotation process, manipulating the operational variables to seek greater stability and greater performance of the circuit in terms of recovery. The accomplishment of this work met the proposed objectives. The evaluation of the advanced process control system showed that, even with the flotation feed having similar characteristics for the two scenarios evaluated, the concentration performance was better with the activated system, with greater recovery and less mineral grade in the tailings. The average concentrate grade of 53.8% was virtually unchanged, keeping the product in the same quality specification. On the other hand, the average tailings grade fell 8.9%. Mass recovery increased by an average of 3.9%, while the metallurgical recovery was, on average, 2.5% higher with Advanced Process Control operating in the circuit. This corroborates that real-time image analysis systems associated with advanced control systems bring better metallurgical performance to column flotation concentration processes. The results of the grade prediction system by real-time image analysis showed a moderate correlation, R² = 0.45, with the data from the laboratory.