Source term estimation with deficient sensors: A temporal augment approach

Abstract

Air pollution of chemical industrial parks (CIPs) is becoming increasingly severe, which has major impacts on the health of local residents. Therefore, source term estimation (STE) is of high importance to find the source locations and back-calculate the source emission rates based on the ambient concentration measurements and meteorological information. However, the number of the ambient sensors is far less than the one of pollution sources for CIPs. This issue of sensor deficiency makes the unknown source parameters untraceable. In this paper, the concept of traceability is introduced to propose a condition to explain when the STE problem has a unique solution. Then an approach using measurements at multiple time points for estimating the emission rate of each source is proposed. In order to satisfy the condition of traceability, the coefficient matrix can be expanded by augmenting measurement samples collected at different time instances with different wind directions when the emission rates keep unchanged. Furthermore, based on the rank of the coefficient matrix, the problem can be classified as fully traceable, partially traceable, and untraceable. Then, the regularized least squares method is applied to estimate the source rates in real-time. Some test results with a simulated scenario demonstrate that the source rate can be reliably estimated with the method proposed. Finally, the limitations and conclusions of the method are stated.