The work aims to explore the emotional design of virtual assistants for the elderly and conduct a design study based on the B-W matrix to enhance the happiness of emotional companionship for the elderly living at home. Firstly, 31 design elements were obtained through interviews with relevant elderly populations, and the obtained design elements were validated through methods such as skewness analysis, T-test, and homogeneity test. Five main factors were identified through factor analysis, namely appearance, interaction, emotion, sound, and scene factors. Secondly, the element classification was constructed through the B-W matrix, and TOPSIS was used for further weight correction. According to the Kano attribute elimination rule, 29 virtual assistant design elements for the elderly population and their weight ranking were obtained, and the design strategy of virtual assistants for the elderly was inferred. Finally, based on the design strategy, case studies and evaluations were conducted on the virtual assistant accompanying APP to verify the feasibility of the design strategy for virtual assistants for the elderly. The research findings provide relevant suggestions for the design of virtual assistants for the elderly in China, and can provide reference for the relevant field of virtual assistant design.