Arun K. Samantaray

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.

Bicausal bond graphs for supervision : From fault detection and isolation to fault accommodation

Model-based fault detection and isolation (FDI) requires an analytical system model from which fault indicators can be derived by assigning proper computational causalities. Many bond graph (BG) model-based techniques for FDI have been developed in recent past. Furthermore, many other advances have been made in the field of control engineering applications of BG modelling. Supervision systems not only perform FDI, but also take the necessary steps for fault accommodation. Fault accommodation is done either through system reconfiguration or through fault tolerant control (FTC). In this paper, it is shown that bicausal BG modelling proves to be a unified approach for sensor placement from the FDI and FTC viewpoint, identification of hardware redundancies for system reconfiguration, generation of fault indicators, estimation of fault parameters for fault accommodation, inversion of systems and actuator sizing for FTC, etc. It is shown that the use of bicausalled BG helps to integrate many of the recently developed advances made in the field of control engineering into development of complex supervision systems.

Evaluation of antilock braking system with an integrated model of full vehicle system dynamics

Abstract Antilock braking system (ABS), traction control system, etc. are used in modern automobiles for enhanced safety and reliability. Autonomous ABS system can take over the traction control of the vehicle either completely or partially. An antilock braking system using an on–off control strategy to maintain the wheel slip within a predefined range is studied here. The controller design needs integration with the vehicle dynamics model. A single wheel or a bicycle vehicle model considers only constant normal loading on the wheels. On the other hand, a four wheel vehicle model that accounts for dynamic normal loading on the wheels and generates correct lateral forces is suitable for reliable brake system design. This paper describes an integrated vehicle braking system dynamics and control modeling procedure for a four wheel vehicle. The vehicle system comprises several energy domains. The interdisciplinary modeling technique called bond graph is used to integrate models in different energy domains and control systems. The bond graph model of the integrated vehicle dynamic system is developed in a modular and hierarchical modeling environment and is simulated to evaluate the performance of the ABS system under various operating conditions.

Improvements to Single-Fault Isolation Using Estimated Parameters

A method for finer fault isolation or localization in the model-based fault detection and isolation (FDI) paradigm is developed using parallely computed bond graph models. Many of the existing modelbased FDI methods are based on the evaluation of model consistency expressed in terms of analytical redundancy relations (ARR). These evaluations lead to residuals, and a number of sensors are to be installed in the plant to generate independent signatures needed for fault isolation. However, all the possible faults may not be isolable with the available instrumentation, and it is sometimes expensive or technically impossible to install necessary sensors in the plant to physically measure each and every state. In such situations, all component faults may not be uniquely isolated. However, a unique fault parameter subspace can be identified. One of the possible solutions, as proposed in this article, is to estimate parameters of that subspace from the ARR by assuming a single-fault hypothesis and then to incorporate the estimated values in separate models to run parallel with the plant during the fault. Thereafter, comparison of model behaviors leads to localization of the faulty parameters. This method is applied to an example system.

Bond graph model-based evaluation of a sliding mode controller for a combined regenerative and antilock braking system

Combined regenerative and antilock braking in electric/hybrid-electric vehicles provides higher safety in addition to an energy storing capability. Development of a control law for this type of braking system is a challenging task. The antilock braking system (ABS) uses a control strategy to maintain the wheel slip within a predefined range. A sliding mode controller (SMC) for ABS is developed to maintain the optimal slip value. The braking of the vehicle, performed by using both regenerative and antilock braking, is based on an algorithm that decides how to distribute the braking force between the regenerative braking and the antilock braking in emergency/panic braking situations as well as in normal city driving conditions. Detailed bond graph models of a quarter car and four-wheeled vehicles are used in this article to implement and test the control laws. It is found that with combined regenerative and antilock braking, the vehicle’s safety increases (in terms of stopping distance and manoeuvrability) and some amount of kinetic energy can be recovered and stored in the regenerative battery pack. The passenger comfort is improved when a sliding mode ABS controller is used in place of a standard ABS controller for the mechanical braking part. Moreover, the influence of load transfer on the wheels during braking was evaluated on a four-wheeled vehicle model.

Sensitivity bond graph approach to multiple fault isolation through parameter estimation

Abstract In model-based quantitative multiple fault detection and isolation (FDI), fault disambiguation is based on parameter estimation. In this paper, the fault hypothesis is generated by evaluating a set of analytical redundancy relations (ARRs) and parameter values corresponding to the unstructured part of the fault subspace are estimated by minimizing a function of the ARRs. Process and measurement uncertainties are handled by using a passive approach for robust FDI. Bond graph modelling is used to describe process models and to derive the ARRs. The bond graph model of the process is differentially causalled and it is then converted into a diagnostic bond graph form. The diagnostic bond graph is further converted into its corresponding sensitivity bond graph form, which gives the residual sensitivity to parametric changes. The developed algorithm provides quicker fault isolation because only a few parameters are estimated and it does not need several model simulations, thereby making it suitable for real-time process supervision.

Modelling of basic induction motors and source loading in rotor–motor systems with regenerative force field

Abstract In this article a short route to arrive at bondgraph model of induction motor is discussed. Creating a model of an induction motor with electrical, magnetic and mechanical ports is a considerably puzzling issue. The usual approach is to view induction motor as an extended transformer with a resistor varying as a function of slip. Such models are inadequate when motors are driving a general class of loads and when electrical input is not a pure harmonic. Though there are several models with intertance fields derived in classical manner they obscure the physics of the system and are often intractable. In this article a general principle of modelling induction motors is presented by viewing the fluctuation of magnetic field from two coordinates, one fixed to stator and other to the rotor. The mechanical port emerges out naturally and the model is computationally efficient and compact. An interesting phenomenon of such a motor driving a rotor with material damping beyond its threshold speed of instability is simulated and discussed. Rotors with internal damping tend to become unstable when driven beyond a speed which is entirely determined by a ratio of external and internal dampings and the critical speed. When such rotors are driven by an induction motor with supply frequency beyond the threshold speed of instability very interesting phenomena are observed in the coupled system like entrainment of rotor speed, existence of limiting oribit of the rotor, small fluctuation of angular speed caused by unbalance and seemingly chaotic behaviour. This example shows the power of bondgraph modelling of induction motor with a mechanical port which can be coupled to a general class of loads.

Modeling and analysis of preloaded liquid spring/damper shock absorbers

Abstract Preloaded liquid spring/damper based shock isolation systems are suitable for heavy load military applications. In this paper, mathematical models are developed for passive liquid spring shock absorbers. The preloading is achieved by mounting the load between two liquid spring/dampers. Dynamics of such shock absorbers involve coupled hydrodynamic and thermodynamic phenomena. The energy dissipated through orifice due to hydrodynamic losses heats up the working fluid and consequently the heat is dissipated to environment. Such multi-energy domain interaction is well represented in this paper by using bond graph models. Moreover, the developed model accounts for the strain-rate dependent damping offered by the compressible working fluid in the liquid spring. The results show that proper choice of preloading and geometric parameters (spring dimensions and orifice sizes) can, respectively, reduce the thermodynamic and strain-rate dependent damping phenomena.

Radial basis function neural network model based prediction of weld plate distortion due to pulsed metal inert gas welding

Abstract Welding shrinkage and distortion affect the shape, dimensional accuracy and strength of the finished product. This work concerns the prediction of welding distortion in a pulsed metal inert gas welding (PMIGW) process. Six different types of radial basis function network (RBFN) models have been developed to predict the distortion of welded plates. Six process parameters, namely, pulse voltage, background voltage, pulse duty factor, pulse frequency, wire feed rate and the welding speed, along with the root mean square (RMS) values of two sensor signals, namely, the welding current and the voltage signals, are used as input variables of these models. The angular distortion and the transverse shrinkage of the welded plate are considered as the output variables. Inclusion of sensor signals in the models, as developed in this work, results in better output prediction.