Construction and Practice of Digital Twin Operation and Maintenance Platform for Water Supply and Drainage Systems in New Energy Power Stations

Abstract

To address the critical challenges of low operational reliability, high energy consumption, and inefficient fault response in water supply and drainage (WSD) systems of new energy power stations (NEPS), this study proposes a digital twin (DT)-enabled operation and maintenance (O&M) platform with multi-dimensional monitoring, cross-protocol adaptation, and intelligent decision-making capabilities. First, a real-time monitoring model integrating pipeline flow, water quality, and equipment status was constructed, leveraging high-precision sensing networks and machine learning algorithms—specifically, a Support Vector Machine (SVM)-based water quality prediction model (mean absolute error, MAE = 0.03 pH units; root mean square error, RMSE = 0.05 NTU for turbidity) and a Convolutional Neural Network (CNN)-driven equipment fault diagnosis model (accuracy = 98.2%, F1-score = 0.978 for pump bearing faults). Second, a cross-protocol data adaptation framework was designed to achieve seamless integration with Vestas wind turbine equipment, supporting Modbus, IEC 61850, and ThingWorx protocols, with data transmission latency reduced to 87 ± 5 ms and packet loss rate < 0.1%. Finally, the platform was validated in the Inner Mongolia 6000 MW Wind Power Demonstration Station, a world-class onshore wind project. Field test results showed that the platform shortened fault response time from 4.0 ± 0.5 hours to 28 ± 3 minutes (a 91.7% reduction), reduced O&M costs by 32.4% (from (12.6 million/year to )8.5 million/year), and improved annual power generation by 41.2% (from 12.7 GWh to 18.0 GWh) by optimizing anti-freezing energy consumption and reducing equipment downtime. This study provides a scalable technical paradigm for intelligent O&M of NEPS WSD systems, with significant implications for advancing the decarbonization and digitalization of the global energy sector.