Multi-objective particle swarm optimization-based adaptive neuro-fuzzy inference system for benzene monitoring

Air pollutants such as benzene (\(\text {C}_6\text {H}_6\)) have accelerated the rate of cancer among human beings. Currently, atmospheric contamination is measured using spatially separated networks with limited sensors. However, the expenses involving multiple sensors with varying sizes limit the operational efficiency. Therefore, in this paper, a novel multi-objective regression model is proposed to predict benzene concentration in the ambient air pollution data, without need to deploy actual sensors for benzene detection. It is possible because there is a relation among various atmospheric gasses and thus regression can be performed to measure \(\text {C}_6\text {H}_6\) if the concentration level of other gasses is known. Proposed technique utilizes adaptive neuro-fuzzy inference system (ANFIS) and particle swarm optimization (PSO) to predict \(\text {C}_6\text {H}_6\) density in the air. PSO is employed to enhance the accuracy of ANFIS for runtime parameter tuning by calculating multi-objective fitness function which involves accuracy, root mean squared error and correlation (r). The proposed technique is tested on well-known publicly available air pollution datasets and on real-time primary dataset for quantitative analysis. Experimental results indicate that the proposed method consistently outperforms over available methods to predict \(\text {C}_6\text {H}_6\) concentration in the atmosphere. Thus, it is well suitable to build self-dependable time and cost-effective benzene prediction model.