A TOOLBOX FOR DESIGN OF DIAGNOSIS SYSTEMS Erik Frisk, Mattias Krysander, Mattias Nyberg, and Jan ˚ Aslund Department of Electrical Engineering Link¨ opings universitet, 581 83 Link¨ oping, Sweden {frisk,matkr,matny,jaasl}@isy.liu.se Abstract: Design of diagnosis systems is a complex task that involves many different steps. Full understanding of all different parts of the design procedure requires deep knowledge on theory from a wide variety of subjects. Thus, to encourage the use of results from diagnosis research it is highly desirable to have software support in the design process. This paper describes ongoing work for determining an architecture for such a toolbox. The paper also describes software solutions in the toolbox. In industry as well as in universities, Matlab is probably the most widespread tool used by control engineers. Therefore the toolbox is primarily based upon Matlab but also some computer algebraic tools such as Mathematica and Maple are used. Copyright c  2006 IFAC. Keywords: Fault diagnosis, software, toolbox, Matlab 1. INTRODUCTION Design of diagnosis systems is a complex task that in- volves many different steps. Full understanding of all different parts of the design procedure requires deep knowledge on theory from a wide variety of subjects. Thus, to encourage the use of results from diagnosis research it is highly desirable to have software support in the design process. This paper describes ongoing work for determining an architecture for such a tool- box and also software solutions to support a diagnosis systems engineer. The work has been carried out partly as a collaboration between Link¨ oping University and Scania CV. 2. USING THE TOOLBOX IN THE DESIGN PROCESS The process of designing a diagnosis system contains several steps. Important steps are the following: • Importing and converting models. • Isolability and detectability analysis. • Selection of submodels to be used in residual generator design. • Residual generator design. • Threshold, possibly adaptive, design for the residual generators. • Selection of which diagnostic tests to use for optimal fault isolation performance. When using the toolbox to design a diagnosis system, the first step is to import the model. Models can be de- scribed in many ways: Simulink models, linear state- space or DAE-models in the Matlab control toolbox format. For all these model descriptions, the toolbox has support for handling information of how different faults affect the process. Several fault types can be handled such as parameter faults and additive faults. This topic is briefly discussed in Section 4. When the model is loaded, there is a possibility to an- alyze fault isolability and fault detectability properties of the model. Depending on the model description, different tools are available. If the model is linear, an- alytical and exact methods are available. If the model is nonlinear, structural and simulation based methods are available. This is discussed in Sections 3 and 5 The next step is to select submodels to be used in the design of residual generators. Here, the user can specify different fault scenarios, for which the residual design should be based upon. This includes to decide which faults that should be decoupled in each residual.