TY - JOUR AU - Chowdhury, Tarun AU - Sinha, Suranjan AU - Roy, Sudip Kumar PY - 2022/01/24 Y2 - 2024/03/29 TI - Artificial Neural Network And Fuzzy Rule Based Two-stage System For Estimating Mine Haul Road Performance JF - Journal of Mines, Metals and Fuels JA - jmmf VL - 69 IS - 10 SE - Articles DO - 10.18311/jmmf/2021/29403 UR - https://informaticsjournals.com/index.php/jmmf/article/view/29403 SP - 347-356 AB - This paper presents a two-stage scheme for estimating the performance of mine haul road. In the first stage, we design a feed forward artificial neural network (ANN) model to predict three important attributes (namely speed, fuel cost, and dust) of a haul road that eventually define the performance of the haul road. The ANN model is trained with eight input variables that are basically eight important parameters for determining speed, fuel cost, and dust, using gradient descent optimization technique. Moreover, a sensitivity analysis is carried out for examining the effects (or for determining the relative contribution and importance) of inputs on the outputs of the proposed ANN model. Further, the 3D response graphs are drawn for investigating the influences of inputs on the outputs of ANN. While the second stage of our method introduces a fuzzy rule based approach for estimating the performance (or condition) of a haul road. The database of two mines is used for training by back propagation and refining the model output so that prescribed error limit by the user is attained. R value obtained is 0.95. Next in a third mine using the mine database without further training the ANN model architecture the model is run and the output is compared to the target value 0.17. Average error between the predicted and the target value is. Next, ANN model output is further analysed for knowledge gain about the otherwise imprecise system by using sensitivity analysis and 3D plots being identified as a input variable with highest relative importance is a major knowledge gain. Few expert rules are then evaluated by following fuzzy rule based approach. Finally, there is some knowledge gain about the system by extraction of rules which will help in initiating timely maintenance of the haul roads so that productivity gain in the mine can be maintained. ER -