In complex vehicular cyber physical systems (CPS) network environment, there exist trust-based recommendation schemes that could effectively filter most of the false data. Though, these schemes may exhaust vehicular network resources, including energy, computation ability, and storage, causing a network outage. To ensure real-time data transmission and security in a vehicular CPS network, a novel trust-based recommendation scheme (TBRS) is proposed in this research. The main contributions presented in this paper are as follows: (1) The isomerism of vehicular sensor nodes in CPS networks, where the differences of mobility between normal nodes and selfish/malicious nodes are analyzed. Besides, a trust model is designed based on delivery credibility and position intimacy of nodes. This model can adjust the weight coefficient of direct trust parameters, which can be utilized to analyze the secure and trustable tasks in data transmission, and (2) To address attacks caused by selfish/malicious nodes and sparsity issues of nodes in vehicular CPS, a secure filtering algorithm based on K-Nearest Neighbor (KNN) cooperative computing is proposed. The trust value is calculated by the proposed trust model. The cooperative computing-based filtering algorithm is utilized to filter false recommendation trust values from selfish/malicious nodes, which greatly reduces interference of selfish/malicious nodes on the performance of vehicular CPS network. The way of calculating trust value cooperatively and recommend trust value makes the TBRS model more secure and reliable than previous ones. Experimental results show that the TBRS scheme is superior to the existing schemes in terms of delivery rate, transmission delay and reliability. Besides, the resistance against illegal eavesdropping attacks has increased by an average of 32.53% when compared to other algorithms.