Chatbot Dedicated to Supporting the Operation of Small and Medium-Sized Wastewater Treatment Plants

Abstract

This paper presents the results of research on applying language model technologies to develop a low-cost chatbot to support operators of small- and medium-sized wastewater treatment plants. The aim of the study was to prepare a tool in the form of a specialized chatbot useful in the operator’s day-to-day work. A chatbot prototype was developed as a web application written in JavaScript, with GPT-3.5-Turbo functionality implemented in Python. Localization of the chatbot for a selected treatment plant involved loading the model with data such as the plant’s operation manual, technical and commissioning documentation of equipment, and time series of operational parameters previously exported from the SCADA system. The primary benefit of implementing the chatbot was rapid access to precise information relevant to the operator, especially in emergency situations. In addition, the chatbot demonstrated significant potential to present data collected from the SCADA system and to perform basic statistical analyses of these data