Intelligent Monitoring and Diagnostic System for Improving the Reliability of Power Generation, Storage and Distribution Systems in Solar Plants

Research relevance consists in the application of renewable energy sources for power generation. Due to the fact that frequent disturbances in the power system and improper operation of some modules lead to disruptions in the continuity of power supply and as a result cause economic and financial losses, the research represents an effective solution under conditions of energy system instability. Energy generation systems are being widely introduced at large industrial facilities as well as in private households. However, they remain sensitive to both adverse weather conditions and physical damage to control components. The main goal of this research is to develop an intelligent system for monitoring and predictive diagnostics of power generation system components, distribution, and storage systems of solar power plants. The subject of this study is the enhancement of power system survivability – a property that enables the system to maintain operational capability or perform critical functions during component failures. The functionality of the developed system ensures the detection of anomalies, the optimisation of load distribution, and the adaptation of control algorithms to operate under conditions of partial failures. A classic client-server architecture has been employed, which involves dividing the application into client and server components that interact via a network and ensure the full cycle of user request processing. The research methods are based on advanced machine learning techniques for anomaly detection (Isolation Forest) and parameter forecasting (Random Forest Regressor), redundancy methods to ensure uninterrupted power supply, and dynamic and statistic methods for load balancing between system modules to ensure system stability. Research results. Based on the obtained results of simulation modelling, high system survivability rates of S in the range of 80–90% were achieved. This demonstrates the effectiveness of the proposed solutions despite unstable power generation and significant temperature variations. The developed method allows the system parameter to be maintained within a higher operability range and at the same time minimizes the impact of critical failures of individual components on general survivability of the energy sector.