Application of modified conditional simulation and artificial neural networks to open pit optimization.

摘要

Design and optimization of open pit limits are of paramount importance because they provide information for evaluating economic potential of a mineral deposit and for developing short- and long-range mine plans. Many algorithms and their modifications have been used to design and optimize open pit However, they do not address the random field properties associated with the ore grades and reserves and commodity prices, and thus, fail to yield the truly optimized pit limits in any time horizon. Also, in mine design and valuation, commodity price forecasts are required to assess the economic viability of the project. The forecast must cover relevant period to capture the trend and volatility in prices within a mining business cycle.; In this study, a new algorithm, CS/MFNN, which overcomes these limitations of forecasting is proposed and used to optimize open pit limits. The random field properties of the ore grade and reserves have been modelled using the modified conditional simulation based on the best linear unbiased estimation and turning bands method. Artificial neural networks are used to classify the blocks into classes based on their conditioned values. The error back propagation algorithm, in the neural networks, is used to optimize the pit limits by minimizing the desired and actual outputs error in a multilayer perceptron under the wall slope constraints. Comparing the Lerchs-Grossmann's and CS/MFNN algorithms, it can be said that both yield the same optimum pit value in the absence of grid blocks with zero economic block values. However, in the presence of grid blocks with zero economic block values they may portray different pit outlines. The stochastic gold price is modelled via two main models, namely; multiple regressional model (MRM) and multilayer feedforward neural networks (MFNN) model. The MFNN model is used to predict the average-annual monthly high gold price. World annual gold production, annual gold consumption, average-annual monthly high gold price, average-annual monthly low gold price, socio-politico-economic condition, interest rate and inflation rate are identified as the most important factors and/or parameters which determine world average annual gold prices. Analysis of the results shows that the mineral price model predicts average-annual gold price with negligible error. The main novelty of this methodology is the solution of the randomness property associated with mineral prices using multiple regression and artificial neural network to reduce the mineral price forecasting error. (Abstract shortened by UMI.)