This paper introduces Attribute-Adaptive Noise Injection (AANI), a novel approach to enhance differential privacy in Cyber-Physical Systems (CPS). AANI addresses the privacy-utility trade-off by dynamically adjusting noise injection based on individual data attribute sensitivity, correlation, and utility needs. This tailored approach allows for fine-grained privacy control, adapting to the diverse data generated by CPS components. The paper outlines AANI's framework, proposes efficient algorithms for attribute-specific noise calculation, and demonstrates its effectiveness through simulations. Results show AANI outperforms traditional differential privacy methods by improving both privacy protection and data utility in CPS.