Research on Multi-Index Temperature Monitoring Data Forecasting of Natural Gas Compressor Unit Based on Dense-Informer

Forecasting and alarm of early abnormal monitoring data in natural gas compressor units is crucial for the safe and reliable operation. This article conducts forecasting research on temperature monitoring data for compressor units, proposing a multi-index time series forecasting model based on a novel Dense-Informer framework. Dense connection strategy and channel compression mechanism are introduced to optimize the traditional Informer forecasting model structure, thereby further enhancing the model's learning capacity. We collected field temperature monitoring data to conduct comparative analysis of the proposed model and traditional methods in forecasting multi-index monitoring data for compressor units. Additionally, comparative validation of model hyperparameters are conducted to further explore their impact on the feature propagation efficiency and predictive performance. The validation results indicate that the proposed Dense-Informer model outperforms other traditional models in terms of forecasting performance. In practical production processes, it can accurately forecast temperature monitoring data for natural gas compressors, providing strong support for early warning and timely maintenance and repair. It has significant implications for production in engineering practice.