Neural Network–Based Multi-Feature Disturbance Analysis for Power Line Installation in Renewable Energy Systems

Abstract

Power line communication and installation in renewable energy–based off-grid and microgrid systems are highly sensitive to electrical disturbances, environmental variability, and appliance-induced non-stationary noise. This paper proposes an RBF neural network–based multi-feature disturbance analysis framework for power line installation assessment. The model integrates power, voltage, frequency, and environmental (P–V–F–E) features into a unified learning architecture to predict disturbance severity and transmission power losses. MATLAB-based simulations were conducted for rural, urban, and industrial environments under seasonal and appliance-induced noise conditions. Compared with traditional and static assessment methods, the proposed model achieved up to 74% reduction in power losses and improved voltage stability under high-disturbance scenarios. The results demonstrate that multi-feature neural learning enables robust disturbance characterization and provides actionable decision support for disturbance-aware power line planning in renewable energy systems.