Application of AI to Filter Anomalous Data from Sensors in an Online Water Quality Monitoring System

Abstract

The multiprobe sensor technology used in online water quality monitoring systems can produce water quality measurement data from several parameters at once quickly and in large quantities. The accuracy of the data is highly dependent on the quality of the river water being monitored and the performance of the probe on the sensor used. The worse the water quality and the decreased performance of the sensor probe, cause the reading of data by the sensor can produce inappropriate anomalous data. Anomalous data can cause water quality analysis to be invalid, therefore we need an Artificial Intelligence (AI) application to filter anomalous data by developing computational algorithms that can provide learning for computers to identify data generated and sent to data center servers. The algorithm method was developed using several mathematical logic models using real conditions and existing needs both in terms of regulations and historical data on water quality monitoring. By applying this algorithm, the computer can have artificial intelligence in analyzing data more accurately with monitoring data that matches the range of sensor measurement capabilities, so this method can guarantee the quality of online monitoring data.