Output consistency evaluation for Class D amplifiers during distribution network fault based on multi-domain feature fusion

Purpose For distribution network single-phase ground faults, class D amplifiers simulate fault characteristic signals for protection testing and analysis. Traditional single-index consistency evaluation methods fail to fully capture steady-state and transient signal features, leading to inaccurate fault diagnosis. This paper aims to propose a multilayer, multi-domain feature fusion strategy to improve consistency assessment reliability and support fault diagnosis. Design/methodology/approach An Analytic Hierarchy Process based three-layer evaluation system is constructed, covering five criterion dimensions. A hybrid weighting method is adopted, where AHP determines subjective weights for criteria layer, while correlation coefficient analysis calculates objective weights for indicator layer. A five-step process yields the comprehensive consistency index finally. Findings The hybrid weighting method balances expert knowledge and data characteristics. Field tests and load variation experiments show that proposed method provides a more comprehensive assessment compared to traditional single-index signal similarity approaches, making it suitable for applications in distribution network fault diagnosis and protection. Originality/value A multi-domain evaluation framework and hybrid weighting strategy are established, breaking the limitations of single-domain indicators. The highly operable evaluation process is validated by field data, providing a practical tool for distribution network fault diagnosis and amplifier performance verification.