To　meet　the　demand　for　standardized　design　of　high-performance　terahertz　functional　devices,　this　paper　applies　intelligent　inverse　design　methods　to　the　design　and　optimization　of　terahertz　bandpass　filters.　A　subwavelength　metasurface　equivalent　model　is　established　with　a　digital　space　mapping,　starting　from　the　target　function　and　constraint　conditions　of　the　device.　Intelligent　algorithms　are　used　to　explore　all　possible　structures　in　the　entire　solution　space　and　iteratively　optimize　them　to　the　optimal　structure　pattern.　This　paper　uses　the　established　inverse　design　framework　to　design　a　terahertz　bandpass　filter　with　a　center　frequency　of　0.51　THz,　bandwidth　of　41.5 GHz,　and　insertion　loss　of　−0.1071 dB.　Compared　with　traditional　manual　forward　design,　the　inverse　design　method　can　deconstruct　a　bandpass　filter　that　is　narrow-band,　low-insertion　loss,　strong　out-of-band　suppression,　and　has　strong　polarization　stability.　©　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.　2024.