Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - An improved algorithm for locating a gas source using inverse methods
AU - Thomson, L.C.
AU - Hirst, B.
AU - Gibson, G.
AU - Gillespie, S.
AU - Jonathan, P.
AU - Skeldon, K.D.
AU - Padgett, M.J.
PY - 2007
Y1 - 2007
N2 - We apply an inverse problem approach to locating a known gas source in a desert setting from simultaneous measurements of gas concentration and wind data. We use a random search algorithm with simulated annealing to generate candidate distributions of source strengths and positions. These distributions are then assessed by means of a cost function, which quantifies the degree to which the postulated source distribution accounts for the measured gas concentrations. We present results from using three cost functions with differing regularisation terms. We assess the robustness of these and the differing regularisation terms by the progressive addition of random noise and systematic offsets to the concentration data. We show that for our application, the best reconstructions are obtained by using a multiplicative regularisation parameter defined to minimise the total gas emissions. © 2006 Elsevier Ltd. All rights reserved.
AB - We apply an inverse problem approach to locating a known gas source in a desert setting from simultaneous measurements of gas concentration and wind data. We use a random search algorithm with simulated annealing to generate candidate distributions of source strengths and positions. These distributions are then assessed by means of a cost function, which quantifies the degree to which the postulated source distribution accounts for the measured gas concentrations. We present results from using three cost functions with differing regularisation terms. We assess the robustness of these and the differing regularisation terms by the progressive addition of random noise and systematic offsets to the concentration data. We show that for our application, the best reconstructions are obtained by using a multiplicative regularisation parameter defined to minimise the total gas emissions. © 2006 Elsevier Ltd. All rights reserved.
KW - Cost function
KW - Gas dispersion
KW - Inverse problem
KW - Random search
KW - Simulated annealing
KW - Algorithms
KW - Concentration (process)
KW - Inverse problems
KW - Robustness (control systems)
KW - Spurious signal noise
KW - Candidate distributions
KW - Cost functions
KW - Tracking (position)
KW - algorithm
KW - concentration (composition)
KW - cost-benefit analysis
KW - gas
KW - measurement method
KW - monitoring
KW - simulated annealing
KW - article
KW - atmosphere
KW - atmospheric dispersion
KW - exhaust gas
KW - noise
KW - priority journal
KW - problem solving
KW - simulation
KW - wind
U2 - 10.1016/j.atmosenv.2006.10.003
DO - 10.1016/j.atmosenv.2006.10.003
M3 - Journal article
VL - 41
SP - 1128
EP - 1134
JO - Atmospheric Environment
JF - Atmospheric Environment
SN - 1352-2310
IS - 6
ER -