As　a　new　service-oriented　computing　paradigm,　cloud　computing　facilitates　users　to　share　and　use　resources.　However,　due　to　the　dynamic　and　openness　of　its　operating　environment,　only　relying　on　traditional　identity　authentication　technology　can　no　longer　fully　meet　the　security　requirements　of　cloud　computing.　The　trust　evaluation　of　user　behavior　has　become　the　key　to　improve　the　security　of　cloud　computing.　Therefore,　in　view　of　some　problems　existing　in　our　current　research　on　user　behavior　trust,　this　paper　optimizes　and　improves　the　construction　of　the　evaluation　index　system　and　the　calculation　of　trust　value,　and　proposes　a　cloud　end-user　behavior　trust　evaluation　model　based　on　sliding　window.　Finally,　the　model　is　proved　to　be　scientific　and　effective　by　simulation　experiments,　which　has　certain　significance　for　the　security　protection　of　cloud　resources.　©　2021　IEEE.