Prediction of blast induced ground vibrations and frequency in opencast mine: A neural network approach

https://doi.org/10.1016/j.jsv.2005.02.044Get rights and content

Abstract

This paper presents the application of neural network for the prediction of ground vibration and frequency by all possible influencing parameters of rock mass, explosive characteristics and blast design. To investigate the appropriateness of this approach, the predictions by ANN is also compared with conventional statistical relation. Network is trained by 150 dataset with 458 epochs and tested it by 20 dataset. The correlation coefficient determined by ANN is 0.9994 and 0.9868 for peak particle velocity (PPV) and frequency while correlation coefficient by statistical analysis is 0.4971 and 0.0356.

Introduction

The increasing development of opencast mines due to the enhanced demand for coal, and other minerals has lead to usage of huge amounts of explosives for blasting particularly in India. Till now, explosives are the efficient source of energy required for breakage and excavation of rocks. When an explosive detonates in a blast hole, instantaneously huge amount of energy in forms of pressure and temperature liberates. Although significant developments have taken place in explosive technology, the explosive energy utilization has not made much progress due to complexity of various rock parameters [1], [2], [3]. Only a small proportion of this total energy is utilized for actual breakage and displacement of rock mass and the rest of the energy is spent in undesirable side effects like ground vibrations, air blasts, noises, back breaks, etc. [4].

The ground vibration is literally a wave motion, spreading outward from the blast like ripples spreading outwards due to impact of a stone dropped into a pond of water. As the vibration passes through the surface structures, it induces vibrations in those structures also. These vibrations induce a resonance in the structures if the frequency of ground vibration matches with the frequency of the structure and due to this, amplitude of the vibration may exceed the amplitude of the initial ground vibrations [3]. Frequency and peak particle velocity (PPV) are most commonly used parameters for assessment of ground vibrations. Dowding [5] underlies the importance of frequency because structural responses depend on the frequency of ground vibrations. Ground vibration is influenced by a number of parameters such as physico-mechanical properties of rock mass, explosive characteristics and blast design. It is essential to know the effect of these parameters on blasting for efficient utilization of explosive energy in a given rock mass vis-à-vis minimization of blast induced side effects. The design parameters like maximum charge per delay, delay time, burden, spacing, charge length, initiation sequence and decoupling charges considerably alter dispersion of the seismic energy. Rock characteristics also often vary greatly from place to place in a mine or even from one end to another of a single face. Hence, blast design parameters and explosives characteristics need to be optimized based on rock mass properties, e.g. strength, density, porosity, longitudinal wave velocity, impedance, stress–strain response and presence of structural discontinuities [6].

The artificial neural network (ANN) is a new branch of intelligence science and has developed rapidly since 1980s. Nowadays, ANN is considered to be one of the intelligent tools to understand the complex problems. Neural network has the ability to learn from the pattern acquainted before. Once the network has been trained, with sufficient number of sample datasets, it can make predictions, on the basis of its previous learning, about the output related to new input dataset of similar pattern [7]. Due to its multidisciplinary nature, ANN is becoming popular among the researchers, planners, designers, etc., as an effective tool for the accomplishment of their work. Therefore, ANN is being successfully used in many industrial areas as well as in research area also. Maulenkamp and Grima [8] developed a model by which uniaxial compressive strength can be predicted from Equotip hardness. It has been reported that the prediction of uniaxial compressive strength by ANN is closer from the measured values. It is indicated by the consistency of the correlation coefficient for the different test set. Yang and Zhang [9] investigated the point load testing with ANN. Cai and Zhao [10] used ANN for tunnel design and optimal selection of the rock support measure and to ensure the stability of the tunnel. Singh et al. [11] predicted the strength property of schistose rocks by neural network. The stability of waste dump from dump slope angle and dump height is investigated by Khandelwal and Singh [12]. They found very realistic results as compared to the other analytical approach. Maity and Saha [13] assessed the damage in structures from changes in static parameters by neural network. Singh et al. [14] predicted the P-wave velocity and anisotropic properties of rocks by neural network. These applications demonstrate that neural network model have superiority in solving problems in which many complex parameters influence the process and results, when process and results are not fully understood and where historical or experimental data are available. The prediction of blast induced ground vibrations is also of this type.

In the present investigation, an attempt has been made to predict the PPV and its corresponding frequency with the help of ANN by using relevant parameters of rock mass, explosive characteristics and blast design.

Section snippets

Geology of the area

The study was conducted at Northern Coalfields Limited (NCL), which is a subsidiary company of Coal India Limited and it is located at Singrauli, Dist. Sidhi (M.P.). It is one of the biggest coal producing company at about 2202 km2. The area of NCL lies geographically between latitudes of 24°0′–24°12′ and longitudes 82°30′–82°45′ and comprises Gondwana rocks.

The coalfield can be divided into two basins, viz. Moher sub-basin (312 km2) and Singrauli Main basin (1890 km2). It is divided into 11

Factors affecting ground vibrations and frequency

The nature and intensity of blast induced ground vibrations and frequency is largely dependent upon many factors [16]. The most important influencing factors are shown in Fig. 1.

All the above-mentioned parameters are dependent upon each other and mostly interrelated. If a particular variable is changed, others parameters will also be changed.

The surrounding rock types have moderate influence on ground vibration behavior [17]. While designing any blast, geo-physical properties should be

Artificial neural network

ANN is a branch of the ‘Artificial Intelligence’, other than, Case Based Reasoning, Expert Systems, and Genetic Algorithms. The Classical statistics, Fuzzy logic and Chaos theory are also considered to be related fields. The ANN is an information processing system simulating the structure and functions of the human brain. It attempts to imitate the way in which a human brain works in processes such as studying, memorizing, reasoning and inducing with a complex network, which is performed by

Network training

A network first needs to be trained before interpreting new information. Several different algorithms are available for training of neural networks but the back-propagation algorithm is the most versatile and robust technique, which provides the most efficient learning procedure for multilayer neural networks. Also, the fact that back-propagation algorithms are especially capable to solve predictive problems which make them so popular. The feed forward back propagation neural network (BPNN)

Dataset

The range of values of different input parameters has been decided by the detailed field investigations by the authors as well as from the published literatures by the various researchers [6], [21], [22], [23], [24] (Table 1).

Rock density is also a critical parameter for prediction of PPV and frequency of ground vibration. The range of rock density of Singrauli area lies between 2.05 and 2.97 t/m3. The variation of rock density is not so high, that is why it has not taken as an input parameter

Network architecture

Feed forward network is adopted here as this architecture is reported to be suitable for problem based on problem identification. Pattern matching is basically an input/output mapping problem. Closer the mapping, better performance of the network.

The objective of the present investigation was to predict PPV and its corresponding frequency from relevant parameters like physico-mechanical properties of the rock mass, explosive properties and blast design. It is difficult to determine all the

Testing and validation of ANN model

To test and validate the ANN model, the new datasets have been chosen. These data are not used while training the network. It will validate the use of ANN in a more versatile way. However, all available vibration predictors proposed by different researchers have site-specific equations [25], [26], [27], [28], [29]. They are not able to use any equation for even other similar geo-mining conditions. The constants which are called site-specific constants and attenuation factor varied once the

Multivariate regression analysis (MVRA)

The purpose of multiple regressions is to learn more about the relationship between several independent or predictor variables and a dependent or criterion variable. The goal of regression analysis is to determine the values of parameters for a function that cause the function to best fit a set of data observations provided. In linear regression, the function is a linear (straight-line) equation. When there is more than one independent variable, then multivariate regression analysis (MVRA) is

Results and discussion

Fig. 4, Fig. 5 show that predictions of PPV and frequency by ANN are very nearer to measured PPV in field but predictions by MVRA have shown very high errors (Fig. 6, Fig. 7), even though, most of the researchers used simply regression analysis or conventional method for prediction of blast induced ground vibration. MVRA is not able to predict the PPV and frequency upto an acceptable limit. ANN demonstrates the superiority over MVRA technique particularly when variables are more. After

Conclusions

Using Bayesian regulation and optimum number of neurons in the hidden layer, the MAPE for PPV and frequency are 4.76 and 6.99, respectively, by ANN. The corresponding coefficients of correlation are 0.9994 and 0.9868, respectively. The prediction by MVRA has very high error. The coefficient of correlation for PPV and frequency by MVRA are 0.4971 and 0.0356, respectively, and MAPE is 343.98 and 140.40. Considering the complexity of the relationship among the inputs and outputs, the results

References (30)

  • V.K. Singh et al.

    Prediction of strength properties of some schistose rock

    International Journal of Rock Mechanics and Mining Sciences

    (2001)
  • G. Cheng et al.

    Analysis of ground vibration caused by open pit production blast

  • Suggested method for blast vibration monitoring

    International Journal of Rock Mechanics and Mining Sciences & Geomechnical Abstract

    (1992)
  • C. McKenzie

    Quarry blast monitoring technical and environmental perspective

    Quarry Management

    (1990)
  • T.N. Hagan, Rock breakage by explosives, in: Proceedings of the National Symposium on Rock Fragmentation, Adelaide,...
  • C.H. Dowding

    Blast Vibration Monitoring and Control

    (1985)
  • D.P. Singh et al.

    Rock fragmentation by blasting influence of joint filling material

    Journal of Explosive Engineering

    (1986)
  • M. Khandelwal et al.

    Application of artificial neural network in mining industry

    Indian Mining & Engineering Journal

    (2004)
  • F. Maulenkamp et al.

    Application of neural networks for the prediction of the unconfined compressive strength (UCS) from Equotip Hardness

    International Journal of Rock Mechanics and Mining Sciences

    (1999)
  • Y. Yang, Q. Zhang, Analysis for the results of point load testing with artificial neural network, in: Proceedings of...
  • J.G. Cai, J. Zhao, Use of neural networks in rock tunneling, in: Proceedings of the Computer Methods and Advances in...
  • M. Khandelwal, T.N. Singh, Prediction of waste dump stability by an intelligent approach, in: National Symposium on New...
  • D. Maity et al.

    Damage assessment in structure from changes in static parameters using neural networks

    Sadhana

    (2004)
  • T.N. Singh et al.

    Prediction of P-wave velocity and anisotropic properties of rock using Artificial Neural Networks technique

    Journal of Scientific and Industrial Research

    (2004)
  • V.K. Singh

    Northern coalfields ltd: Surging ahead with time

    Journal of Mines, Metals and Fuels

    (2004)
  • Cited by (259)

    View all citing articles on Scopus
    View full text