Power equipment defect reports contain critical information for maintenance decisions, but increasing data volume and human inconsistencies limit manual processing effectiveness. While general text correction methods address basic errors, they cannot resolve domain-specific semantic ambiguities and structural flaws in defect descriptions. This study introduces a comprehensive quality assessment framework based on three metrics: completeness, accuracy, and redundancy. Using word embeddings, semantic similarity computation, and structured sequence analysis, we develop and validate a text enhancement pipeline through multiple case studies. Results demonstrate significant improvements in defect text clarity and reliability, enabling more precise equipment condition assessment and maintenance planning.