Development of Soft Computing Models for Noise Prediction in Opencast Mines

Noise is generated by almost all opencast mining operations from different stationary, mobile and impulsive sources, thereby becoming an integral part of the mining environment. Increase in mechanisation of opencast mines has resulted in increase in noise levels, resulting in noise pollution in mining and its allied industries. Prolonged exposure of miners to the high levels of noise can cause noise induced hearing loss (NIHL) besides several non-auditory health effects. Directorate General of Mines Safety (DGMS) Technical Circular No. 18 of 1975 has prescribed the permissible noise level of 90 dB (A) as the “danger limit” and 85 dB (A) as the “warning limit” in a shift of 8 hours for unprotected ear and Circular No.5 of 1990 on “Protection of Workers against noise” is emerging as an important and challenging health hazard for mine workers. Occupational Safety and Health Administration (OSHA) general industry standard, 29 CFR 1910.95 (c)(2) and The National Institute for Occupational Safety and Health (NIOSH) recommends a hearing loss prevention program (HLPP) when employee noise exposure exceeds an 8 hour TWA sound level of 85 dB measured on the A scale (dB (A)) and 5 dB exchange rate. To ascertain the impact of noise in mines, appropriate noise survey of opencast mining machineries ought to be conducted and noise levels need to be studied. Noise exposure from workplace can cause an effect of NIHL which initially occurs at high frequencies (3kHz,4kHz, 6kHz) and gradually spreads to the low frequencies (0.5 kHz, 1 kHz, 2 kHz). NIHL has a greater impact on affected personnel and substantially decreases quality of working life. The requirement of precisely measuring the noise levels at different 1/3 octave band frequencies of all the noise emitting sources in the opencast mines is well established. The measurement of sound pressure level using sound measuring devices is not accurate due to instrumental error and various attenuation factors. Some of the generally used mathematical noise prediction models like frequency independent (VDI-2714) and frequency dependent (CONCAWE, ENM, OCMA, ISO-9613-2) have been applied in mining and allied industries. All the noise prediction models treat noise as a function of distance, sound power level (SWL) and different forms of attenuations factors. These parameters are measured in the mines and best fitting models are applied to predict noise. All these models are used to calculate the noise levels from various opencast mining machineries by considering all the attenuation factors. VDI – 2714 is independent from the frequency domain and is the simplest noise prediction model as compared with the other noise prediction models. Mathematical models were very complex to utilize each time and fail to predict the future parameters from current and past measurements. To overcome the drawbacks of the mathematical models and considering the successful application of soft computing techniques in the real time applications, soft computing models were used in the development of predicting far field noise levels due to operation of opencast mining machineries. Soft computing models like Fuzzy Inference System (Mamdani and Takagi Sugeno Kang (TSK) fuzzy inference systems), Artificial Neural Networks (Multilayer perceptron, feedforward, backpropagation and radial basis function network (RBFN)), Adaptive neuro fuzzy inference system (ANFIS) and evolutionary algorithms (genetic algorithm (GA) and differential evolutionary algorithm (DEA) were used to predict the far field noise levels due to operation of opencast mining machineries in the two opencast mines. The proposed soft-computing models were designed for both frequency and non-frequency based noise prediction models. After successful application of all proposed soft-computing models, comparative studies were made considering Root Mean Square Error (RMSE) as the performance parameter. It was observed that proposed soft-computing models give good prediction results with accuracy. However, DEA model gives better noise prediction with better accuracy than other proposed soft-computing noise prediction models. Distribution of noise levels in any mining area depends not only on the stationary or moving sources but also on the complex geographical conditions. It is essential to measure the noise levels being produced from the specified sources and to share the information with the people exposed to those sources. This assists the people to understand the levels of noise that they are normally exposed to. Noise maps depict the spatial distributions of noise levels and allow a proficient representation of the noise distributions in the selected areas sensitive to noise. This study also aimed to produce noise map using ArcGIS by considering the noise levels in the different parts of the opencast mine with their positions to define how the noise sources affects the mine and its surrounding areas