In real-world recommendation systems, user interactions inherently include behavioral information (e.g., clicks, cart). Multi-behavior sequential recommendations focus on predicting the next item a user will interact with under the target behavior. Incorporating behavioral information enhances the diversity of user interaction data. However, existing multi-behavior recommendation methods perform unified modeling without distinguishing the interest categories reflected by different data. Hence, we propose a novel method called TDBSR (Trend-Detection and Diversity-Aware for Multi-Behavior Sequential Recommendation) which separately captures interest trend and dispersed diversity before performing a unified integration. We design a mask generation module that decouples the data, employing MLP-based modules: HITM (heterogeneous interest trend module) and ABIP (auxiliary behavior intent perception module) to extract user interest tendencies. Concurrently, we leverage MLP and max-pooling layers to extract user interest diversity. Experiments on two public datasets validate that our proposed TDBSR outperforms state-of-the-art methods.