Statistical Properties and Design Criterions for AI-Based Fault Isolation

IEEE Trans. Syst., Man, Cybern. A

2010

Self Cite

112

Abstract-An essential step in the design of a model-based diagnosis system is to find a set of residual generators fulfilling stated fault detection and isolation requirements. To be able to find a good set, it is desirable that the method used for residual generation gives as many candidate residual generators as possible, given a model. This paper presents a novel residual generation method that enables simultaneous use of integral and derivative causality, i.e., mixed causality, and also handles equation sets corresponding to algebraic and differential loops in a systematic manner. The method relies on a formal framework for computing unknown variables according to a computation sequence. In this framework, mixed causality is utilized, and the analytical properties of the equations in the model, as well as the available tools for algebraic equation solving, are taken into account. The proposed method is applied to two models of automotive systems, a Scania diesel engine, and a hydraulic braking system. Significantly more residual generators are found with the proposed method in comparison with methods using solely integral or derivative causality.

Section: Dae Systemsmentioning

confidence: 99%

“…In this paper, we have the view that the design of diagnosis systems is a two-step approach, as elaborated in [3] and [4]. In Manuscript received April 8, 2009.…”

mentioning

confidence: 99%

Residual Generators for Fault Diagnosis Using Computation Sequences With Mixed Causality Applied to Automotive Systems

IEEE Trans. Syst., Man, Cybern. A

2010

Self Cite

112

“…Let M = (E, X, Z, F ) be a model, M a residual generation method, and F an isolability requirement. Also, let S M be the set of all candidate equation sets for M and I the set of all isolability classes of S M for F, defined according to (3) and (4). Then, for each S ⊆ S M where all S ∈ S are realizable with M it holds that F is fulfilled with…”

Section: Formalization Of the Selection Problemmentioning

confidence: 99%

“…In this work, as in for example [2], [3], [4], [5], design of the residual generation sub-system is considered to be a two-step approach. In the first step, a large set of candidate residual generators are found.…”

Section: Introductionmentioning

confidence: 99%

Realizability Constrained Selection of Residual Generators for Fault Diagnosis With an Automotive Engine Application

IEEE Trans. Syst. Man Cybern, Syst.

Frisk

2013

Self Cite

Abstract-This paper considers the problem of selecting a set of residual generators for inclusion in a model-based diagnosis system, while fulfilling fault isolability requirements and minimizing the number of residual generators. Two novel algorithms for solving the selection problem are proposed. The first algorithm provides an exact solution fulfilling both requirements and is suitable for small problems. The second algorithm, which constitutes the main contribution, is suitable for large problems and provides an approximate solution by means of a greedy heuristic and by relaxing the minimal cardinality requirement. The foundation for the algorithms is a novel formulation of the selection problem which enables an efficient reduction of the search-space by taking into account realizability properties, with respect to the considered residual generation method. Both algorithms are general in the sense that they are aimed at supporting any computerized residual generation method. In a case study the greedy selection algorithm is successfully applied in an industrial sized automotive engine system.

Automotive engine FDI by application of an automated model-based and data-driven design methodology

Control Engineering Practice

Frisk

et al. 2013

Self Cite

Fault detection and isolation (FDI) in automotive diesel engines is important in order to achieve and guarantee low exhaust emissions, high vehicle uptime, and efficient repair and maintenance. This paper illustrates how a set of general methods for model-based sequential residual generation and data-driven statistical residual evaluation can be combined into an automated design methodology. The automated design methodology is then utilized to create a complete FDI-system for an automotive diesel engine. The performance of the obtained FDI-system is evaluated using measurements from road drives and engine testbed experiments. The overall performance of the FDI-system is good in relation to the required design effort. In particular no specific tuning of the FDI-system, or any adaption of the design methodology, were needed. It is illustrated how estimations of the statistical powers of the fault detection tests in the FDI-system can be used to further increase the performance, specifically in terms of fault isolability.