Recognition of technological states of drum mills on the basis of neuron networks of adaptive resonance

Abstract

Purpose.The purpose of this article is to establish the most effective methods of using indirect estimates. These estimates relate to the spectral composition for instantaneous values in two key areas: active power consumed by the drive motor and artificial intelligence (in the form of adaptive resonance neural networks). These are both necessary in order to monitor the technical parameters involved in filling drums for wet self-grinding mills. Methodology.The method we use is as follows. First, we examine the spectral characteristics (as well as the composition of instantaneous values) consumed by the drive motor. Second, we apply the basic algorithms of adaptive resonance neural networks. This is in order to analyze the “information sensitivity” properties of real power signals generated within the drum mill motors thus allowing us to identify and predict the technological states of the latest, and largest, drums filled with crushed ore. Findings.Based on the research we conducted using the software simulation model, there is the fundamental possibility of implementing an indirect control of the technological parameters when filling drum mills. It is based on the sensitivity of the characteristics of the spectral composition of fluctuations contained within the instantaneous values for the active power consumed by the drive motor. In addition, it realizes the recognition properties of the adaptive resonance neural network algorithms for drum mills of MMS type 90*30А. We expect to conduct additional experimental research in production to confirm the reliability of the proposed methodology. Originality.We formed a new method of indirect recognition of the technological parameter operational values when filling the drum with ore in wet self-grinding mills. It is based on a simulation model using a basic algorithm of adaptive resonance neural networks. Practical value.Insertion of new indirect informative algorithms to the structure of automated control systems for drum mills allows the increase in the reliability of identifying their current operational status.