Non-intrusive power waveform modeling and identification of air conditioning load

•Unsupervised autonomous AC waveform model construction and self-update method is proposed based the general electrical characteristics.•An AC waveform extraction method based on dynamic guided filtering is designed for the multi-appliance mixed operation scenarios.•An online AC state identification and power disaggregation method via waveform matching is proposed.•Comparison tests on real measured datasets demonstrate the proposed method outperforms the existing solutions in state identification and power estimation.•The proposed method can be potentially used as an embedded module in other NILM systems to improve their overall performance.