Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Blasting is the most popular method for rock fragmentation in open-pit mines. However, the side effects caused by blasting operations include ground vibration, air overpressure (AOp), fly rock, back-break, dust, and toxic are the critical factors which have a significant impact on the surrounding environment, especially AOp. In this paper, a robust artificial intelligence system was developed for predicting blast-induced AOp based on artificial neural networks (ANNs) and random forest (RF), code name ANNs-RF. Five ANN models were developed first; then, the RF algorithm was used to combine them. An empirical technique, ANN, and RF models were also developed to predict and compare with the ANNs-RF model. For this aim, 114 blasting events were recorded at the Nui Beo open-pit coal mine, Vietnam. The maximum explosive charge capacity, monitoring distance, vertical distance, powder factor, burden, spacing, and length of stemming were used as the input variables for predicting AOp. The quality of the models is evaluated by root-mean-square error (RMSE), determination coefficient (R2), mean absolute error (MAE), and a simple ranking method. The results indicated that the proposed ANNs-RF model was the most superior model with RMSE of 0.847, R2 of 0.985, MAE of 0.620 on testing dataset, and total ranking of 40. In contrast, the best ANN model yielded a slightly lower performance with RMSE of 1.184, R2 of 0.960, MAE of 0.809, and a total ranking of 39; the RF model yielded a performance with RMSE of 1.550, R2 of 0.939, MAE of 1.222, and total ranking of 22; the empirical model provided the lowest accuracy level with RMSE of 5.704, R2 of 0.429, MAE of 5.316 on the testing dataset, and total ranking of 6.

Predicting Blast-Induced Air Overpressure: A Robust Artificial Intelligence System Based on Artificial Neural Networks and Random Forest

Ground vibration (PPV) is one of the hazard effects induced by blasting operations in open-pit mines, which can affect the surrounding structures, particularly the stability of benches and slopes in open-pit mines, and impact underground water, railway, highway, and puzzling for neighboring communities. Therefore, controlling, accurate prediction, and mitigating blast-induced PPV are necessary. This study contributed a new computational model in predicting blast-induced PPV for the science community and practical engineering with high accuracy level. In this study, a novel hybrid artificial intelligence model based on the hierarchical k-means clustering algorithm (HKM) and artificial neural network (ANN), namely a HKM–ANN model, was considered and proposed for predicting blast-caused PPV in open-pit mines. Accordingly, input data were first clustered by the HKM algorithm, and then, the ANN models were developed based on the obtained clusters. For this aim, 185 blasting events were collected and analyzed. A hybrid model based on fuzzy c-means clustering (FCM) and support vector regression (SVR), i.e., FCM–SVR model, which was proposed by previous authors was also applied for comparison of results with our proposed HKM–ANN model. In addition, an empirical method, several ANN and SVR models (without clustering), FCM–ANN, and HKM–SVR were developed for comparison purposes. For measuring the performance of the improved models, coefficient determination (R2), root-mean-square error, and variance account for were used as the performance indicators. The results show that the HKM algorithm played a significant role in improving the accuracy of the ANN models. The proposed HKM–ANN model was the most superior model in estimating PPV caused by blasting operations in this study.

Prediction of Blast-Induced Ground Vibration in an Open-Pit Mine by a Novel Hybrid Model Based on Clustering and Artificial Neural Network

Analysis of rate of penetration (ROP) prediction in drilling using physics-based and data-driven models

Use of machine learning and data analytics to increase drilling efficiency for nearby wells

Using Trees, Bagging, and Random Forests to Predict Rate of Penetration During Drilling

Evaluation of coupled machine learning models for drilling optimization

\n Real-time drilling optimization is a topic of significant interest because of its economic value, and its importance increases particularly during periods of low oil prices. This paper evaluates different optimization strategies and algorithms for real-time optimization of an objective function (function to be optimized) specific to drilling. The objective function optimized here is derived from a data-driven (or machine-learning) model with an unknown functional form. A data-driven model has been used to calculate the objective function [rate of penetration (ROP)] because it has been shown to be more efficient in ROP prediction relative to deterministic models (Hegde and Gray 2017). The data-driven ROP model is built using machine-learning algorithms; measured drilling parameters [weight on bit (WOB), revolutions per minute (rev/min), strength of rock, and flow rate] are used as inputs to predict the ROP.\n Real-time drilling optimization that is data-driven is challenging because of run-time constraints. This is perceived as a handicap for data-driven models because their functional form is unknown, making them more difficult to optimize. This paper evaluates algorithms depending on their ability to best maximize the objective (ROP) and their time effectiveness. Two simple yet robust algorithms, the eyeball method and the random-search method, are presented as plausible solutions to this problem. These methods are then compared with popular metaheuristic algorithms, evaluating the tradeoff between improvement in the objective (search for a global optimal) and the computational time of run.\n Using results from the simulations conducted in this paper, we concluded that data-driven models can be used for real-time drilling despite their computational constraints by choosing the right optimization algorithm. The best tradeoff in terms of ROP increase as well as computational efficiency evaluated in this paper is the simplex algorithm. The ROP was improved by 30% on average with a variance of 2.5% in the test set over 14 formations that were tested.

Chiranth Hegde

Papers

Analysis of rate of penetration (ROP) prediction in drilling using physics-based and data-driven models

Use of machine learning and data analytics to increase drilling efficiency for nearby wells

Using Trees, Bagging, and Random Forests to Predict Rate of Penetration During Drilling

Evaluation of coupled machine learning models for drilling optimization

Performance Comparison of Algorithms for Real-Time Rate-of-Penetration Optimization in Drilling Using Data-Driven Models