Vasanth Krishnaswami

Fault detection in IC engines using nonlinear parity equations

This paper presents a novel nonlinear parity equation residual generation scheme for fault detection and isolation. The scheme is applied to the problem of health monitoring of an automobile internal combustion engine during execution of a standard US Environmental Protection Agency inspection and maintenance test cycle. The fault detection and isolation scheme is shown to be sensitive to actuator and sensor faults and the scheme is robust to the unmeasured load torque acting on the engine.

Model based health monitoring of vehicle steering system using sliding mode observers

This paper presents a solution to the problem of diagnosing faults in a vehicle steering system using nonlinear observers constructed by sliding mode observer design techniques. A bicycle vehicle model using small angle approximations is used to represent vehicle behavior. The model reproduces the in plane motions (longitudinal velocity, lateral velocity and yaw rate) of the vehicle. A bank of observers each utilizing different measurements to estimate the system states are constructed using sliding mode theory. The sliding observers are used in a binary phase detection filter configuration to isolate sensor faults. Further the design of the sliding observers is extended to estimate the inputs acting on the system and these estimates are then used for isolation of actuator faults. The effectiveness of the diagnostic scheme is demonstrated through the use of simulations. Further, experimental implementation currently being scheduled in collaboration with the Transportation Research Center of Ohio.

Robust Residual Generation for Nonlinear System Fault Detection and Isolation

Abstract This paper presents the development of a fault detection and isolation scheme for nonlinear dynamic systems. The paper presents in detail the problem formulation, explores the existence conditions for fault isolation and quantifies the computational needs of the isolation scheme. A detailed procedure is presented for the construction of residual generators from models of the system to be monitored. The formulation presented here is completely-new and extremely general, being applicable to almost all kinds of faults, including input, output and component faults

Vehicle steering system state estimation using sliding mode observers

The increasing sophistication of the electronics package available on board the modern automobile, has made viable a number of new technologies that can greatly enhance the driveability and safety of the vehicle. One area that has received much attention is the use of drive by wire steering systems that partially or in some cases (such as in some envisaged IVHS systems) completely takes over the tasks of vehicle directional control. The vehicle handling dynamics are quite nonlinear (especially for severe maneuvers and over large variations in longitudinal speed). Most steering control algorithms require knowledge of the vehicle states and in many cases are quite sensitive to the values of various vehicle parameters such as tire cornering stiffness, tire-road coefficient of friction etc. Measuring the states is usually expensive and direct measurement of the parameters is often impossible. Thus, some means of estimating these variables becomes imperative. Recent advances in sliding mode control/estimation theory have demonstrated the effectiveness of such schemes when applied to nonlinear systems. This paper presents some extensions to the theory of sliding mode observers and an application to the problem of estimating vehicle steering system states.

A new model order identification algorithm with application to automobile oxygen sensor modeling

A method of determining the model order of MIMO nonlinear dynamic systems was proposed using the concept of the coherence function. It was demonstrated that the proposed scheme is extremely successful even in the presence of noise, clearly outperforms the Lipschitz criterion scheme in determining the model order accurately and unambiguously and identifies the order correctly even in the case when the Lipschitz scheme produces an erroneous result. The effectiveness of the order determination scheme in conjunction with the NARMAX identification technique, for identifying a severely nonlinear system such as an automobile exhaust gas oxygen (EGO) sensor was also studied. It was shown that the proposed scheme is suitable for practical implementation using simulation results.

Determination of influencing inputs and their orders for nonlinear dynamic systems-application to IC engine modeling

This paper presented the development of a technique for the identification of influencing inputs for nonlinear dynamic systems from input-output data. The technique was compared against two previously proposed schemes through application to a chaotic system. The applicability of the technique to identifying the time delays of the exhaust system of an automobile with a binary oxygen sensor was demonstrated through a simulation example.