Introduction

Generally, “solving” the controller design problem meansfinding the proper mathematical representation of a control action thatmeets a set of desired performance criteria. In reality, this is only onepart of the solution (albeit an important part); the control-systemsdevelopment process also includes steps for selecting the correct hardware– loaded with the proper software – for the controller module,which is the real end-product of this process (Figure 2.1).

The control-system development process begins by first developing thehigh-level system requirements, which are generally verbal and abstract andrarely point to a recognizable control design problem such as those thattraditional engineering students would see in their control classes. Theformal and technical requirements documents can be described as “wishlists” regarding the overall system features and performance. Theresult of the process is the controller module, which is to be deployed inbulk to the end product. The purpose of studying the control-systemsdevelopment process is to provide a reliable, robust, and repeatablesequence of actions to develop ECUs.

In recent years, computer-aided design and analysis tools (e.g., MATLAB andSimulink) have improved the efficiency of design processes and increased theapplication of the model-based controller design and development process(Chrisofakis et al. 2011; Mahapatra et al. 2008; Michaels et al. 2010;Powers and Nicastri 2000). Figure 2.2 is a general outline of themodel-based controller design and deployment process. The major componentsof this process (i.e., design, implementation, and testing) are discussed inthe next section. The process outline in Figure 2.2 is based on developmentand testing portions that progress in parallel and continuously interactthroughout the development cycle. This is, in fact, one of the mostimportant features of model-based design, which enables debugging andvalidation of the current work while minimizing changes from the previousphase. Therefore, as the control development evolves, so does the testingplatform for its debugging and validation.