A Survey of AI-Driven Report Generation Technologies in the Power Sector

To enhance the intelligence level of report generation in the power industry, this paper systematically reviews four mainstream technical approaches: template-based generation, information extraction, neural network-based generation, and large language models (LLMs). These methods are compared in terms of structure controllability and language generation characteristics. Application patterns and platform architectures are analyzed in typical scenarios such as dispatching operations and equipment inspection. Based on this, future directions including multimodal fusion, knowledge graph integration, edge deployment, and LLM collaboration are proposed to support the development of efficient and trustworthy report generation systems in the power sector.