An Application Of Tree Classification Method In Analysis Of Longwall Shield Stability

Longwall mining systems are extremely complex and expensive and require great caution for reliable and efficient operation. In many cases, ground control problems are the primary concern in longwall mining. For example, unexpected threats such as roof falls or shield damage can cause sudden interruption of coal production and safety problems, which can originate from many parameters such as shield capacity, setting load, shield type, conditions of immediate and main roofs, pillar stability, geological anomalies, etc. However, it is difficult to correlate ground-control problems at a longwall face with the related mining and geological factors. In this paper, a method of correlation analysis, called tree classification, which permits the use of a large and variable number of features, is introduced The method provides an effective approach to produce an accurate classifier or uncover a knowledge structure, which will be used to predict future consequences depending on past and current data sets. The application of the tree classification method in longwall stability analysis is discussed, and the major procedures of constructing a knowledge structure are included as an example utilizing monitored leg pressure data in a longwall face. Also, the importance and ranking of the variables related lo longwall shield stability is discussed This is a major step towards developing an artificial intelligence system for real-time interpretation of shield monitoring data and hazard prevention.