An informer-algorithm-assisted MOF-derived CuO nanoparticle/CeO2 nanorod heterostructure sensor for early rapid detection of CO leakage

This study proposes a synergistic approach combining high-performance sensing materials with a deep learning prediction algorithm for accelerated detection. The metal-organic frameworks (MOFs)-derived CuO nanoparticles and CeO 2 nanorods were synthesized by co-precipitation and solvothermal methods. After annealing treatment, the CuO/CeO 2 nanocomposite was prepared by mechanical mixing. The successful synthesis of CuO/CeO 2 was confirmed by characterization techniques such as XRD, SEM, TEM, and XPS. The CuO/CeO 2 nanocomposite exhibited excellent CO sensing performance, including a low operating temperature (160°C), fast response and recovery times (100 ppm, 8 s/14 s), a low detection limit (1 ppm), high selectivity, as well as good repeatability and long-term stability. The excellent CO sensing performance is primarily attributed to the p-n heterojunction formed between CuO and CeO 2 , as well as the structure of CeO 2 nanorod, which effectively suppresses the aggregation of CuO nanoparticle. The sensing mechanism of the CuO/CeO 2 nanocomposite towards CO gas was further investigated through density functional theory (DFT) calculations. To improve signal quality and detection speed, the sliding window adaptive Kalman filter (SW-AKF) is adopted to suppress noise, and a time series prediction model is constructed based on the Informer architecture (using multi-head probabilistic sparse self-attention and self-attention distillation) to achieve advanced prediction of CO resistance signals.