In this paper, an AUV decentralized control system approach is investigated, this will open a possibility to enable control system component to interact between various control components on the simulation network infrastructure. During the course of theoretical studies to simulation platform development, an OCP has emerges some promises to overcome any boundary for both in control system domain and network infrastructure domain.
3.1 Simulation workstation
To implement the proposed AUV depth control system simulation, information may needs to be rerouted between AUV subsystems or control components. In this situation, sometimes a certain data may became temporarily very important and at other time not needed at all. In figure 5, a simulation system consist of two node connected with a general 10 Mb Ethernet, PC1 as a server and PC2 as client, every node will consists of two blocks, first block consist of vehicle model and control algorithm, and second block is consist realtime network components that support hard control reconfiguration.
The two PCs as server and client are connected via middleware communication using TAO-CORBA Event channel, which is described more details in (Schmidt D. C. et al., 2000). A PC1 running a Matlab simulation of vehicle model and control algorithm, while PC2 running a Matlab simulation as sensor source and mission control station to allows a user to dynamically modify any parameters during a runtime simulation.
Generally, in the development step, mostly control engineer test the new control algorithm in Matlab environment. Matlab is a convenient tool for graphical plotting; it is relatively difficult to use C++ to plot system responses in multiple dimensions. However, C++ is widely used in real-time data acquisition and control in industrial applications. The interface between C++ and Matlab offers a significant improvement in data acquisition and control system analysis. This makes the analysis for complicated systems possible in the real world. Using the interface method, it is much more convenient to perform matrix operations with real-time controllability.
Obtain 8, tv, q, and z values t ~ Apply K into AUV control t ~ Calculate E
t ~ Calculate/1 using ARE
t ~ Obtain Q and R values
Another point is, for real-time systems, especially for the multiple variables control system, a state variable matrix has to be used to make the real-time analysis based on the state feedback from the system outputs. By using the interface between C++ and Matlab, a lot of data analysis and real-time control tasks for actual systems are possible. Another advantage to using the interface between C++ and Matlab is to handle multi dimension matrix operations and continuous plotting of system responses. Almost all data acquisition and control processes need time response of trajectory in real-time.
In figure 6, illustrates the block diagram of operation principle C++ and Matlab interface. Via this interface, C++ program should collect the data from PC2 through CORBA and create the data variables in ASCII format. Matlab first picks up the data from the data variables have stored by C++ and then performs the matrix operations based on the data. The results can be sent back to C++ by Matlab in the ASCII variables, while C++ program continue to executing the communication task to send a Matlab results.
The interface between Matlab and C++ in order to transfer a data through CORBA event channel is not so complicated, although for control engineers, this method offers a significant improvement in data acquisition and control system analysis; this makes the analysis for complicated systems possible in the real world.
Was this article helpful?