An error compensation method for remote sensing measurement of mobile source emissions

Abstract

The remote sensing measurement of mobile source emissions is often performed in a complex and changeable outdoor environment. Thus, the remote sensing measurement process is easily affected by environmental interferences. To address this problem, a new error compensation method based on the transfer entropy - artificial neural network - adaptive weighted fusion (TE-ANN-AWF) model is proposed. The method combines the transfer entropy causality analysis and adaptive fusion estimation. Firstly, the original observation sequence is decomposed into multiple virtual sequences by the proposed virtual measurement method. Then, each virtual sequence is compensated by ANN error prediction model. Finally, the compensated multiple virtual sequences are reconstructed by TE and AWF. In the process of reconstruction, the multiple virtual sequences are fused adaptively by AWF. Meantime, TE is used to estimate multi-interference imbalance degree and improve the dynamic performance of the error compensation process. The experiments and simulations show that the method can effectively compensate the measurement error and improve the instrumental environment adaptability.