As software complexity increases, quality issues become more prominent. Reducing early errors, shortening development cycles, and improving software reliability are key challenges facing software companies. To address these challenges, this study introduces an automated requirements analysis and modeling method based on deep learning technology, which aims to optimize the requirements acquisition and processing steps in the software design and development process. First, this paper proposes a structured requirements analysis framework that uses deep learning technology to automatically extract and classify user requirements. Subsequently, the requirements are modeled in combination with feature models to ensure that the constructed model can accurately reflect the real needs of users. To further enhance the effectiveness of requirements modeling, this study integrates business process modeling methods, Rational unified process, and BP (Back Propagation) neural networks to provide a multi-level modeling strategy. This strategy optimizes the software development life cycle by dynamically adjusting the requirements analysis process to adapt to the changing market and user needs. The experimental part verifies the effectiveness and accuracy of the proposed method by collecting data sets from actual software projects, including text descriptions of user requirements and functional specifications. The experimental results show that the proposed automated requirements analysis and modeling method has significant advantages in improving the accuracy of requirements analysis, reducing development cycles, and improving the consistency between requirements and the market.