ECT Image Recognition of Pipe Plugging Flow Patterns Based on Broad Learning System in Mining Filling

The process of mining filling, when the slurry is transported to goaf by filling pipeline, very important find location and size caking in pipeline time for safe stable operation mine pipeline. It an research work detect different flow patterns after two-dimensional section reconstruction closed based on ECT (electrical capacitance tomography) visualization method. Slurry regarded as a two-phase flow, multishape distribution was reconstructed into images intelligently recognized broad learning system (BLS) algorithm. BLS feedforward neural network with few optimization parameters fast training speed. In this paper, three features sample images, number regional blocks, roundness barycenter are combined structure recognize patterns. Through simulation, recognition accuracy fillback image more than 99%. This conclusion indicates effectiveness predict patterns; it also provides new solution pattern