Development of an aid tool for fault detection in ball mills

"It was developed in this work a computational tool to aid in anomaly detection present in the operation of ball mills. The tool uses a torque estimator that is based on measurements taken exclusively in the drive motor. The ball mill equipment widely used in the mining industry has, inherent to its principle of operation, high energy consumption and low efficiency. These characteristics are accentuated when the mill operates outside the ideal conditions of velocity and loading. Direct measurement of these variables is extremely difficult because it requires the installation of sensors in moving parts of the mill. Thus, systems that can help monitoring the operating conditions and predict anomalies are useful because they contribute to improving the performance of the process, help to reduce the energetic consumption, wear on the mill surface and frequency of maintenance shutdowns. The detection strategy implemented in this work is based on the fact that operational anomalies such as an inadequate percentage of balls and incorrect angular velocity manifest itself as a change in spectral components of the torque signal in the mill driven by an electric motor. The mill torque was estimated by processing motor variables measured by sensors commonly used for control purpose and installed exclusively in the motor. The mill was operated at various fault operating conditions and signals from electric motor and vibration on the mill shaft were measured. The spectral components present in the estimated torque signal were then compared with signals from a vibration sensor in order to establishing a correlation with anomalous operation condition of the mill. The experimental results showed that the signals obtained has an excellent correlation with other variables obtained by direct measurement and have potential to be used to construct a system for automatic fault detection in the operation of ball mills."