Automated detection of anomalies in high-frequency water quality sensor data using machine learning

Wastewater treatment facilities are increasingly installing more and more high-frequency water quality sensors, as high-quality data is essential for plant operation and optimization. The sheer volume of data being collected and the necessity to avoid the collection of erroneous data, has created a need for automated tools to assess the quality of that data and signal for maintenance as the need arises. As these datasets have increased in size and complexity, it has become difficult to identify problems in a timely manner either manually or to use simple rules that might have been sufficient previously. A software solution is thus developed to provide a quick analysis of fault detection. The anomaly detection algorithm is developed based on deep learning technology, where the detection model is derived solely from the data and no prior knowledge required.