A Method of Time Series Data Compression and Washing-Cycle Based Data Parsing for Smart Washing Machines

The massive volumes of time series data produced by multi-sensory, intelligent washing machines show complex relationships, robust temporal dependencies, and substantial redundancy. These issues lead to low data quality, high time costs, and analysis difficulties. This research proposes a novel approach that combines periodicity-related compression and trend pattern filtering to compress data and washing-cycle based data parsing. The method employs a periodic grouping strategy to remove duplicative data and caching to support lossless restoration. Then, it finds trend patterns using sequence pattern mining to filter the redundant data. Finally, this method uses a finite state machine to abstract the washing-cycle state transitions, resulting in adaptive washing-cycle based data parsing that significantly enhances data utilization efficiency. Experiments with real applications demonstrate that our time-series data compression method improves the maximum compression ratio to 8.19. The adaptive washing cycle parsing method outperforms the traditional method on cycle parsing and statistics query speed. Thus, our method guarantees correctness and integrity while preventing errors and information loss.