In　order　to　further　improve　the　artistic　style　transfer　effect　of　traditional　painting,　this　paper　proposes　an　artistic　style　transfer　method　of　Chinese　painting　based　on　generative　adversarial　network.　Where,　the　cyclic　generative　adversarial　network　(CycleGAN)　is　selected　as　the　basic　style　transfer　algorithm,　and　the　loss　function　in　the　network　is　improved,　so　as　to　further　enhance　the　style　transfer　effect.　The　experimental　results　show　that　compared　with　the　CycleGAN　before　improvement,　the　improved　CycleGAN　has　better　stability.　Compared　with　other　transfer　algorithms,　the　designed　Chinese　painting　artistic　style　transfer　algorithm　based　on　improved　CycleGAN　has　better　transfer　effect,　and　the　FID　score,　PSNR　and　SSIM　of　the　designed　algorithm　are　162.09,　99.61　and　0.7,　respectively.　In　conclusion,　the　transfer　effect　of　the　designed　style　transfer　algorithm　is　good,　and　the　designed　style　transfer　algorithm　can　be　applied　to　the　actual　Chinese　painting　artistic　style　transfer　with　high　reliability.　©　2023　IEEE.