目的 探索基于情感感知的助眠音乐推荐设计的实践效果，通过测试其在情感改善与睡眠诱导方面的效用，验证其可行性并进一步洞察优化空间。方法 从感知情感对诱发情感的引导出发，以智能床为硬件载体，兼具音乐助眠场景下的信息提供、序列选择引导及播放设置等功能，对音乐序列推荐逻辑及交互界面进行设计。开展原型评估实验，以测试并验证其在情感调节及助眠方面的效果、分析不同序列设计间的相互差异。此外，结合访谈数据反馈，进一步优化界面设计，开展可用性验证。结果 在情感调节方面，将唤醒度和愉悦度按序排列的音乐序列分别在其对应维度有着最佳调节效果，基于喜爱度及熟悉度权重的音乐序列的综合调节表现最优。在助眠效果方面，唤醒度降序排列和基于喜爱度及熟悉度权重的音乐序列表现较优。结论 从情感感知视角提出的助眠音乐推荐设计策略的应用原型，能有效地调节用户情感并产生睡眠诱导效果，同时启发了对应交互设计的功能体验优化，具有较好的实际应用价值。

The work aims to explore the practical effect of sleep-aid music recommendation design based on emotional perception and verify its feasibility and gain further insight into the optimization space by testing its effectiveness in emotion regulation and sleep induction. According to the guidance of perceived emotion on induced emotion, the music sequence recommendation logic and interaction interface were designed with the smart bed as the hardware carrier characterized by music information, sequence selection guidance and playback settings in the music sleep-aid scenario. Prototype evaluation experiments were carried out to verify its effect in emotion regulation and sleep aid and analyze its mutual differences in different sequences of design. In addition, the interface design was further optimized combined with the interview data feedback and the usability was verified. In terms of emotion regulation, the music sequences arranged according to arousal and valence in order had the best regulation effect in their corresponding dimensions. Sequences based on the weight of favoritism and familiarity performed best in regulating comprehensive emotions. In terms of sleep-aid effect, the music sequence arranged in descending order of arousal and based on the weight of favoritism and familiarity performed best. The application prototype of the sleep-aid music recommendation design strategy proposed from the perspective of emotional perception can effectively regulate users' emotion and produce sleep-aid effects. It also inspires functional experience optimization of corresponding interaction design, and has good practical application value.