Fire　evacuation　simulations　are　essential　for　assessing　evacuation　schemes　in　subway　stations.　However,　most　studies　and　practical　solutions　have　primarily　focused　on　the　effects　of　fire　smoke　on　the　evacuation　behaviour　of　evacuees.　Moreover,　these　solutions　overlook　the　effects　of　the　movement　of　evacuees　on　fire-smoke　diffusion;　hence,　they　do　not　adequately　reflect　fire-pedestrian　interaction.　This　study　addresses　this　problem　by　formulating　a　fire-smoke　evolution　model　using　the　lattice　Boltzmann　method　to　simulate　fire　smoke　from　a　mesoscopic　perspective.　The　proposed　model　can　reflect　fire-smoke　diffusion,　including　the　effects　of　evacuees　on　the　evacuation　scheme.　The　model　easily　exchanges　data　with　the　microscopic　pedestrian　simulation　model.　An　agent-based　evacuation　model　is　formulated　using　the　social　force　method.　The　agent　can　regulate　the　evacuation　states　using　perceptive　environment　information.　Finally,　the　models　are　integrated　into　one　framework　to　evaluate　fire-pedestrian　interaction.　A　simplified　fire　evacuation　scenario　based　on　a　two-dimensional　subway　station　platform　is　simulated　using　the　proposed　integration　model.　The　results　show　that　the　proposed　integration　model　can　simulate　the　effect　of　the　disturbance　of　the　movement　of　evacuees　on　fire-smoke　diffusion　and　evaluate　fire-pedestrian　interaction.