G. Dauphin-Tanguy

Diagnostic bond graphs for online fault detection and isolation

Abstract Analytical redundancy relations (ARR) are symbolic equations representing constraints between different known process variables (parameters, measurements and sources). ARR are obtained from the behavioural model of the system through different procedures of elimination of unknown variables. Numerical evaluation of each ARR is called a residual, which is used in model based fault detection and isolation (FDI) algorithms. For processes and systems with complex non-linearity, eliminating all unknown variables is not trivial, e.g. in the presence of algebraic loops, implicit equations, non-invertible functions, etc. However, most symbolically non-resolvable relationships can be numerically solved; and then, it becomes possible to maximise the number of structurally independent residuals. Bond graph modelling is used in this paper to derive ARR and to obtain the computational model in the case of non-resolvability of equations. A set of sub-graph substitutions in the bond graph model are developed. These substitutions directly lead to a form, where known variables (measurements, sources and parameters) are the inputs and the residuals are the outputs. Such a model is then called a diagnostic bond graph (DBG) model. It is shown that DBG models can be used for online residual computation as well as for offline verification using process data from a database. A method for the coupling of the bond graph model, used to generate the residuals, with a bond graph model, used to describe the process behaviour, is presented. The coupled model allows simulation of process behaviour both in the presence and in the absence of the faults, which is consequently used to obtain residual responses and validate the fault signatures.

Component-Based Modelling of Thermofluid Systems for Sensor Placement and Fault Detection

Modelling of thermofluid systems is complex because of the coupling of thermal and hydraulic energies. Bond graph language is a suitable tool for modelling such nonlinearmultienergy domain systems along with their control systems. Structural control properties of the system can be determined using the causal properties of bond graphs, without resorting to any formal computation. An ontology for classifying thermofluid process components is presented in this article, along with connection syntax and model validation algorithms. Then the theory for structural analysis is used to design sensor placements for observability and also for component fault detection and isolation. This study also deals with developing a bond graph–based model representation mechanism to perform structurallevel linking of submodels.The methodology developed is applied to two example systems. Simulation of process faults is used to validate the theoretically obtained fault signatures for a thermofluid system in the second example.

Bond-Graph Analysis of Structural FDI Properties in Mechatronics System

Abstract This paper deals with structural fault detection and isolation (FDI) properties using bond graph approach and residual tree. Two graphical methods to generate analytical redundancy relations (RRA), used for FDI are presented. The first one is based on the bond graph using. The second one is interested in the exploitation of residual trees. Based on the limits of exposed methods, is proposed an oriented residual tree to generate directly the relations which lead to the RRAs.

Fault Detection of Backlash Phenomenon in Mechatronic System with Parameter Uncertainties Using Bond Graph Approach

This paper presents a method for fault detection and residuals evaluation, applied on electromechanical test bench system in the presence of backlash phenomenon and parameter uncertainties. The analytical redundancy relations (ARRs) are generated using uncertain bond graph model, in linear fractional transformation form (LFT). Through the presented method, one can distinguish between backlash default and system parameter variation. Simulation tests presented in this paper show the influence of dead zone magnitude and system parameters variations on residuals evaluation

Robust FDI based on LFT BG and relative activity at junction

It is seen that methodology for diagnosis of uncertain systems using the Bond Graph (BG) model in Linear Fractional Transformation (LFT) form is an effective way to detect the fault by systematic generation of robust adaptive thresholds over the nominal residuals. There are limitations associated, many of which are imposed due to the manner uncertain parameters can be treated. In this work, a new method of generating robust and adaptive thresholds is developed, where uncertain parameters are treated as intervals that vary between the upper and lower bounds. The developed method proposes to generate the envelopes, using the interval extension form of uncertain part of Analytic redundancy relations (ARR) that are derived from BG model in LFT form. The method is applied on an uncertain two tank system to show its viability. Issue of missed detection is discussed where not all of the considered uncertain parameters may deviate outside their allowed limits. To diagnose such a case, generation of multiple thresholds is proposed by selection of parameters based on their increasing activity. The algorithm for generation of set of robust envelopes is presented and applied over the hydraulic system to alleviate the missed alarm problem. Results are validated through simulations.

Bond Graph Model of Wind Turbine Blade

Abstract The design of a wind turbine system (sizing, control) according to specifications in terms of quality of produced electricity, stability and safety, needs the use of performing simulation tools able to change from simple static calculations assuming a constant wind to dynamic simulation that, from the unsteady aerodynamic loads, models the aero elastic response of the entire wind turbine system, including blades, tower, drive train, rotor and control system. In the paper, a bond graph model of a blade is proposed. It consists in considering the blade as a Rayleigh beam composed of a number de sections submitted to aerodynamic forces calculated using the BEM theory. The model is validated using data from a NACA 4415 blade profile.

Analysis of Structural Properties of Thermodynamic Bond Graph Models

Abstract The present paper deals with the structural properties of thermodynamic systems using the bond graph approach. The monitoring ability and observability are directly determined from the linearized multiport bond graph with any need to generate the analytical redundancy relations or the signature faults. This approach is illustrated by an application to a three-tank system.

Flat control of a torsion bar with unknown input estimation

The main purpose of this paper is to design a flat control of a real torsion bar system subject to unknown input variables using the concept of unknown input observer. The bond graph approach is used due to its structural and causal properties. Three main points are developed: modeling of the dynamical system with model simplification using an integrated approach, application of an unknown input observer with estimation of physical variables and application of a flat control on the experimental system.

Optimal sensor placement for fault diagnosis

Ability to detect and to isolate faults which may affect the system depends essentially on instrumentation architecture. This is why, before designing an industrial supervision system, determination of monitoring ability based on technical specifications is important. Used methods in the consulted literature are based on a model given as a set of collected data in different modes (faulty and normal) or under complex differential equations. In the present paper it will be shown how the behavioral, structural and causal properties of the bond graph model can be used for monitoring ability analysis (which part of the system is over, just or under constrained) with no need of calculation. The developed method is applied to the designing a real time monitoring of an electromechanical system.

Matroid algorithm for monitorability analysis of bond graphs

The monitorability analysis (ability to detect and isolate faults) using a model is presented. A foundation of this structural property using bond-graph model is proposed. The present paper deals with independent residual generation based on binary matroid, especially chain groups over GF(2). The methodology is applied to bond-graph model, and from the structural point of view, for this task, a matroid intersection algorithm is used to compute the maximum matching.