Heterogeneous Modeling and Design of Control Systems

Complex control systems integrate a variety of functions and capabilities, which will in general rely on different computational mechanisms. The plant model may be represented as a set of ordinary differential equations, the mode switching logic may be expressed as a finite state machine, and dataflow models may be used to capture the architecture of a sensor processing subsystem, for example. Design tools are needed that can support these heterogeneous models of computation and their integration within a single control system. This chapter describes Ptolemy II, a component-based design environment that allows different models of computation to be hierarchically composed. Individual components are called actors; these can include simple operators such as an AND gate and, through compositionality in the form of arbitrarily nested actor hierarchies, complex functions such as a Kalman filter. Different models of computation can then Ž . be realized by imposing a possibly partial execution order and a communiation mechanism on the actors that comprise a composite actor. Ptolemy II can be particularly useful for the design of hybrid dynamical systems systems that combine discrete event mode logic and continuous time dynamics within each mode. By using Ptolemy II to model a hybrid system, we can choose whichever model of computation is most appropriate for a particular task while ensuring the consistency and integrity of the overall model. An application of Ptolemy II to the control of the Furuta inverted pendulum is also described in this chapter, with an emphasis on implementation issues relevant to real-time operating systems. As noted in Chapter 4, Ptolemy II has been interfaced with the OCP and used to model a nonlinear actuator as part of an overall vehicle simulation. Chapter 5 also notes an OCP-related application of the Ptolemy project, in this case to support the validation of reconfiguration strategies and other hybrid control aspects.