To make up for the intrinsic flaws in evolutionary computation,Sun et al. [22] proposed the Mind Evolutionary Algorithm (MEA) in1998, which emulates activities of the human mind [23] and exhibitsan excellent optimising performance and convergence ratio.The Back Propagation Neural Network (BPNN) is a multilayerfeed-forward neural network, and it possesses a strong nonlinear mapping ability. The objective of the present study is to design anovel range hood and optimise its geometric configuration to meetthe requirements of exhaust airflow rate with high grease separationefficiency. To realise this purpose, an inverse design method isproposed based on the 3D CFD model coupled with MEA-BPNN,where the initial weights and thresholds of BPNN are optimisedby MEA; furthermore, MEA is coupled with the optimised BPNNto find the optimal structural parameters of the novel range hood.The optimal objective of MEA is to maximise the exhaust airflowrate with good grease separation; the considered design variablesare seven structural parameters, including blade inlet and outletangles, blade number and height, guide vane number and height,and diffuser diameter. The typical working conditions of trainingand validating BPNN are determined by the method of orthogonalexperiment. The results of the typical working conditions areobtained by CFD to establish the relationship between optimal valuesand design variables. MEA and Mono-Factor Analysis (MFA)and Half Octave Method (HOM) are adopted to enhance the efficiencyof optimisation. The results show that the exhaust airflowrate and grease separation performance of the proposed novelrange hood is satisfactory.