Machine learning based soft sensor model for BOD estimation using intelligence at edge

Abstract Real-time water quality monitoring is a complex system as it involves many parameters to be monitored, the nature of these parameters, and non-linear interdependence between themselves. Intelligent algorithms crucial in building intelligent systems are good candidates for reliable convenient system. To analyze quality, we need understand, model, monitor pollution real time using different online sensors through an Internet things framework. However, cannot easily measured due several reasons such high-cost sensors, low sampling rate, multiple processing stages by few heterogeneous requirement frequent cleaning calibration, spatial application dependency among bodies. A soft sensor efficient alternative approach monitoring. In this paper, propose machine learning-based model estimate biological oxygen demand (BOD), time-consuming challenging process measure. We also architecture implementing both on cloud edge layers, so that device can make adaptive decisions water. comparative study computational performance nodes terms prediction accuracy, learning time, decision (ML) presented. This paper establishes BOD efficient, less costly, reasonably accurate with example real-life application. Here, IBK ML technique proves most predicting BOD. The experimental setup uses 100 test readings STP samples evaluate technique, statistical measures reported correlation coefficient = 0.9273, MAE 0.082, RMSE 0.1994, RAE 17.20%, RRSE 37.62%, response 0.15 s only.