Taking standard floor of high-rise office buildings as a research object in five cities of northern coastal region, Lianyungang, Qingdao, Yantai, Tianjin and Dalian included, model simulation and performance calculation are carried out with the help of Ladybug + Honeybee (L + H) platform. Principal component analysis (PCA) is used to solve the multicollinearity problem that 15 design variables including a building's body (B, DR, FH, C and RE), window form (WWR, WH, SH, WHD and WVD) and envelope property (K, SHGC, VT, EW_R and F_R) were reduced to 12 and divided into 4 principal components (PCs), namely PC1 (buildings' appearance), PC2 (window position), PC3 (window property) and PC4 (exterior-wall property), and then weights of PC1~PC4 were obtained by weight analysis to be 0.41, 0.268, 0.188 and 0.134 respectively, and the principal component linear evaluation function was proposed as Q = 0.41 PC1 + 0.268 PC2 + 0.188 PC3 + 0.134 PC4. On the other hand, with the help of genetic algorithm (GA), buildings' energy consumption (BEC), thermal discomfort (PPD, Predicted Percentage Dissatisfied) and natural daylighting (sDA, Spatial Daylight Autonomy) were coupled to optimize standard floor shapes to find out excellent solutions. In summary, the paper proposes a performance optimization design process of high-rise office buildings' standard floor.