• Proposes a novel multi-timescale feature extraction method using adaptive entropy to handle varying sampling frequencies in wastewater treatment.
• Integrates expert knowledge graph with multiscale entropy analysis and partial least squares to enhance data representation and monitoring.
• Demonstrates improved water quality data representation and monitoring capabilities through experimental validation.
• Addresses the challenge of information loss and errors from data interpolation in multi-timescale wastewater data.