Architecture for AI-Based Validation of Wastewater Management Using Open Data Exchange Technique

To processing plant-wide data, it is not uncommon for there to be issues with sensors. Some examples of these issues include sensor failures, sensor calibration difficulties, sensor fouling and blockage, and connection problems between sensors, actuators, and the data management system. These problems may have an effect on the quality of the data, which will make it more difficult to implement sophisticated data-driven monitoring and control of (crucial) water operations. A method for the intelligent validation of data is provided in this article. The sensor data coming from a wastewater treatment plant are validated and verified by this process. The application for data validation has been integrated, tested, and put into production at the water utility known as Water Net. The method of validation is predicated on an anomaly detector that, in the event that an anomaly is found, combines (statistical) threshold-based procedures, (AI-based) autoencoder model predictions, and a component for data reconciliation. This occurrence repeats itself whenever an anomaly of some kind is found. In spite of the signal's volatility, the autoencoder models displayed outstanding levels of accuracy and great reconciliation performance. In addition, signals for both verified and raw data are provided in a way that is extremely near to real time to a dashboard. This keeps the dashboard up to date with the most recent information. Validated data may also be utilized as a screening tool for data that is being ingested by another AI-based model in order to monitor and intelligently manage the wastewater treatment plant, as well as to minimize greenhouse gas emissions and energy consumption while maintaining effluent water quality requirements. These goals may be accomplished while still meeting all applicable criteria. If this were done, it would be in compliance with the regulations for maintaining the effluent water's quality.