In the context of industrial cyber-physical systems (ICPS) that are subject to hybrid asynchronous DoS attacks and actuator faults, and driven by a dual-endadaptive discrete event-triggered communication scheme (ADETCS), this study investigates a multi-objective constrained multimodal comprehensive security control method for ICPS. Firstly, a dual-end ADETCS is designed to select the transmitted data at both the sensor side and the actuator side, with triggering parameters that adaptively change according to the system's motion behavior. Then, a clustering algorithm based on the improved sparrow search algorithm-FCM (ISSA-FCM) is used to classify different control modes offline division and online identification. Then, at the sensor side, a data-model linkage method is applied to handle DoS attacks of varying energy levels. At the actuator side, a Bayes-LSTM (long short-term memory) network is used to reconstruct and compensate for the lost data due to DoS attacks, while active fault tolerance is employed to address actuator faults. Furthermore, by introducing $ \alpha {\text{-}} $stability and $ {H_\infty }{\text{-}} $multi-objective constraints, the solution for the robust observer and comprehensive security controller of ICPS is derived based on the constructed Lyapunov-Krasovskii function, and the online switching fusion method of multimodal controllers is studied under different control modes corresponding to different trigger parameters. Finally, the effectiveness and practicality of the proposed method are validated through simulation experiments, achieving bidirectional adaptation between the network service quality and control system performance of ICPS.