Vehicular crowdsensing networks play a critical role in the Internet of Vehicles by enabling efficient information services, but challenges like on-demand message authentication and privacy protection persist. We propose a secure crowdsensing scheme, TRAMS, based on multi-authority attribute-based signatures, enabling fine-grained policies for participant authentication while safeguarding vehicle privacy. TRAMS incorporates a multi-authority key management system to enhance sensing efficiency, achieving superior message authentication compared to single-authority systems. To address malicious behaviors in VANETs, we introduce HDRS, a hybrid reputation system where vehicles and roadside units independently evaluate reputations and cross-reference results. HDRS leverages a reliability module and a dynamic adjustment mechanism to counter intelligent attacks, achieving up to 30% higher detection rates for collusion and 16% for adaptive threats compared to existing solutions.