Modeling And Optimizing For Real Time Communication Of Industrial Ethernet And System Maintenance

For safety and stability of a networked control system (NCS), real-time communication and equipment maintenance are two important issues. In Media Access Control (MAC) layer, NCS retains the CSMA/CD protocol, in which the transmission of frames have a large time-delay after a collision, so that the stability of NCS is not guaranteed. Therefore, how to reduce the time-delay and improve real-time communication for guaranteeing the stability of system working are worth doing the research in depth. In addition, most NCS are maintainable rather than scraped regularly. Hence, an effective maintenance for equipment of NCS can decease the failure probability of equipment and increase the working life of NCS significantly, which is very importance for system working stably. The main work is as follows.Different buffer sizes should be allocated for data frames in different network loads of industrial Ethernet. To that end, the model of industrial Ethernet buffer size required is constructed based on queue theory. In this model, the transmission process of frames in buffer queue of industrial Ethernet is analyzed and the queuing time-delay of frames is identified as the key factor affecting the network time-delay. Then, the distribution of process which frames entry buffer queue is determined. On the basis of these, the model of queuing time-delay of frames is established by queue theory and the objective function regarding communication cost is given. Last, through solving the objective function, the optimal buffer queue length is obtained. The experiment regarding the optimal solution is performed, which validate the corrective and effective of the model constructed. In addition, the experiment result is comparison with the result obtained by WFQ method. The comparison result demonstrates our model is better than the WFQ method in terms of communication cost.Transmission rates of data frames in sending and receiving ends of an industrial Ethernet is statistic equilibrium or not can affect the efficiency of industrial Ethernet communication directly. In order to improve the communication performance of industrial Ethernet, the optimal model for industrial Ethernet based on statistic distribution is constructed using Markov theory. In this model, the transmission state of the industrial Ethernet is analysed. Moreover, the transition probabilities between any two different Ethernet states are detemined, which constitute transition probability matrix. Then, the probability of each state is derived by solving the steady state equation. Subsequently, the probability of frames successful retransmitted is calculated when collision occurs in each time. The maximal successful rate of data frame transmission as objective function is identified, and optimizing the transmission rate of sending end is performed. Finally, the simulation experiment is performed, where the experiment results demonstrate that the constructed model is correct and effective. Meanwhile, the experiment results is compared with the SRTC approach. The comparison results demonstrate that the proposed model is better than the SRTC approach in terms of successful transmission rate, throughout and average time-delay of end-to-end.In order to improve the working life of redundant control system, the optimal maintenance method based on semi-Markov theory is proposed. In this method, the system dynamics are analyzed under some necessary assumptions. On the basis, transition probabilities between system states, which are the kernel of semi-Markov, are characterized by semi-Markov chain and renew point technology. Then the image function of system states distribution are obtained by Laplace transform on Markov kernel. And according to image function, the expected working life function of system is obtained. By analyzing the affection of preventive maintenance cycle in system expected working life, the relation between preventive maintenance cycle and the longest expected working life is obtained under different conditions. The optimal maintenance method is given. Finally, a numerical example and simulation experiment are performed to validate the effectiveness of the policy.In order to reduce the cost caused by the failure of NCS equipment, the non-priority and priority corrective maintenance models are proposed. For the non-priority maintenance model, the procedure of system maintenance is given. The system transition state and state space are defined repectively, and the transition probabilities between any two states are analyzed. The transition process of system states is identified as a special Markov process, namely a birth-death process. Through calculation results, two important parameters, birth rate and death rate, are obtained. On the basis, the expected failure time of controller is computed by queue theory and convolution. Then considering the affection of failure controller in the whole system, we give the objective function subjecting to system reliability based on system loss. In priority maintenance model, the priority of each class of controllers is specified, and the mean failure time of highest priority controllers is calculated. Afterwards, we could obtain the relation between virtual failure time and actual failure time of other priority controllers. The actual failure times of each class of controllers are calculated by recurrence, and the objective function based on system loss is given. Finally, two numeral examples verify the correctness and effectiveness of these two proposed models.In the final part, we test the models of industrial Ethernet buffer size and optimizing the industrial Ethernet communication performance on the platform of integrated monitoring system of DCCE. The results demonstrate that constructed models can improve the network performance. Regarding to optimal maintenance, we also give the experiments on the redundant sub-system in integrated monitoring system and the NCS.