For the Industrial Cyber-Physical System (ICPS) with Denial of Service (DoS) attacks and actuator failures, the integrated mechanism analysis explores the collaborative design challenge of ICPS comprehensive security control and communication and data-driven methods based on a novel Adaptive Discrete Event-triggered Communication Scheme (ADETCS). Firstly, a new kind of ADETCS is presented, and its trigger parameter can adaptively change according to system operation conditions, which can save network communication resources more efficiently. Secondly, a detection mechanism that relies on energy levels is implemented to tackle DoS attacks, facilitating the identification and differentiation of DoS attacks with varying energy levels, whether high or low.. Thirdly, during the course of active intrusion tolerance, a data-driven Long Short-Term Memory (LSTM) neural network is adopted to reconstruct and compensate for data lost due to high-energy DoS attacks, which is then combined with the elastic passive intrusion tolerance based on mechanism analysis for DoS attacks with low energy levels, achieving a mixed active-passive intrusion tolerance against DoS attacks. Then, based on time-delay system theory and novel Bessel-Legendre inequalities, the analytical solutions for robust observers and comprehensive security controllers are derived to achieve a trade-off balance between comprehensive security and the effectiveness of communication performance. Ultimately, the effectiveness and accuracy of the suggested method are validated through simulation examples.