Two inverse design optimisations are involved in this study, viz.inverse design optimisation for BPNN and the novel range hood. Toimprove prediction accuracy, MEA is introduced in these inversedesign optimisation processes. To optimise BPNN, the initial weights and thresholds were used as design variables and the minimummean square error (MSE) was set as the optimal objective.The design variables used to optimise the range hood include thefollowing structural parameters – blade inlet and outlet angles,number and height of blades, number and height of guide vanes,and diffuser diameter; in addition, the optimal objective is toachieve the maximum exhaust airflow rate at a satisfactory greaseseparation level.