محمد جواد خسروجردی

In many applications, especially military applications, the inertial navigation system (INS) needs to achieve a high level of accuracy in a short time. For alignment, recursive estimator filters and, in non-linear cases, the Extended Kalman Filter (EKF) is often used. The dynamics of a real, continuous system and the output of the sensors are available discretely. Therefore, a hybrid filter has been used. In addition, a robust filter is used to increase the reliability of system operation. In this paper, a Hybrid Extended Kalman Filter (HEKF) is presented and then upgraded to the Hybrid Robust Extended Kalman Filter (HREKF). By running the algorithm on the data of a real system, it was observed that the speed of convergence increased especially in the yaw direction. By running the algorithm on the data of a real system, it was observed that the speed of convergence has increased especially in the yaw direction. Finally, using the impulsive system approach, a new stability analysis of the proposed algorithms is presented, which guarantees the boundedness of the error estimation, which is unique.

Degradation due to the long time operation at the gas turbine may cause a fall in the combustor and compressor efficiency and characteristics, which makes a decrease in total efficiency and, an increase in fuel consumption with growth in the production of the pollutants. These phenomena are not distinguishable by conventional diagnostic systems. Although instrumental protection of gas turbine may be profited by the control system but, it depends to components damage result revealation. So, in this research, a special method using the Multiple Model approach, based on multi-operating points is developed to continuously estimation these kinds of faults, to sustain the major defects in an industrial gas turbine. The object is done by defining the main components health indicator variables. First nonlinear thermodynamic static and dynamic models are constructed and linearized. By a combination of those in a dynamic convex set the alternate adaptive model is developed that, could cover the dynamic of the gas turbine. By decoupling the adaptive filter residuals, estimation of the faults and operating point determination are executed. Finally, the efficiency of the proposed approach is demonstrated in a simulation environment.

A flight control system (FCS) as an example of nonlinear systems should be stabilizing the airplane in healthy and faulty modes with the desired performance. In this paper, first, the nonlinear model of the flight system (FS) is linearized in predefined operating points, and the airplane’s healthy and faulty modes convert to a multi-model system. To eliminate the undesired effect of multi-modeling and the controller robustness that usually causes to mask the faults in FSs, the problem of integrated design of FCS based on active fault detection (AFD) is formulated. The proposed problem concluded to design the optimal auxiliary signal of AFD and a static, fixed, and full-order controller with the capability in reduced-order design. The FCS guarantees to stabilize all of the healthy and faulty models and to satisfy the performance constraints. In the following, to solve the formulated problem, a numerical solution based on the genetic algorithm is proposed. To evaluate the proposed method, two full-order and reduced-ordered controllers are designed for an aircraft in the case of actuator degradation conditions. The simulation results show the ability of the proposed method to meet the control objectives and design the auxiliary signal for AFD.

In this paper, the effect of different sensors on the observer performance of vehicle suspension system is investigated. For this purpose, the concept of observable degree analysis is used to quantitatively measure the observability for different sensor choices. A new method, for determining the observable degree of linear time invariant (LTI) systems has been developed on the basis of distance of system from set of similar unobservable systems. A long distance is equivalent to a strong observability and a short distance is equivalent to a weak observability. The zero distance means that the system is unobservable. Since the distance to different unobservable modes can be determined separately, a comprehensive investigation of system observability and the effect of different sensor choices on the observer performance can be provided. In the following, the observable analysis of the suspension system was performed based on the proposed method and the effect of different outputs on the observer performance has been investigated. The results show that when the observable degree is increased for a specific sensor, the observer gain is decreased and consequently the sensitivity of observer relative to the noise and measurement errors is decreased. The increased accuracy of observer demonstrates a good conformity between observable degree analysis and observer performance. Also, a comparative study showed that, contrary to previous criteria that only considered a certain aspect of observability, the proposed method is more comprehensive and realistic, and the results obtained from the previous criteria can easily be achieved through the proposed method.

In this paper, a data driven method for fault estimation in control systems is proposed. The fault estimation problem is formulated as a time domain control problem and the H∞ technique is applied for solving it. The mathematical model is assumed to be unknown and only input/output (I/O) data are used for detector design. Minimizing the fault estimation error is considered as the fault detection performance measure. Also, since the estimated fault plays the role of a virtual control input, a cost is defined on it to avoiding the H∞ control problem being singular. These measures are achieved by tuning a scalar parameter and some weighting matrices. An easily implementable design algorithm summarizes the methodology presented in the paper. The proposed algorithm is applied to a numerical example in order to illustrate its effectiveness.

This paper presents a new approach for fault tolerant control system (FTCs) design based on robust nonlinear model predictive control (NMPC) for multivariable affine systems. The proposed FTCs uses an estimation scheme that is based on adaptive extended kalman filter (AEKF) for the state estimation of plant and loss of effectiveness factors of actuators. A supervisor module also uses the fault modeling and correction of plant model per sampling time to accommodate bias and loss of effectiveness of actuators. In addition by feedback compensation in NMPC، the proposed controller is robust through plant uncertainties. The most important advantage of the proposed approach is its ability to deal with the constraints and simultaneous fault in actuators and it is practical. Simulation results of the proposed method on automotive engine show the effectiveness of the proposed method.

In this paper، a novel technique is presented for determining of operating and surge point in compressor via sensor fault tolerant control. The analytical redundancy method and the dynamic neural network (DNN) based on robust identification scheme is presented to determine of compressor surge point accurately، even in the presence of uncertainty in the compressor and noise in the sensor. Generally، the main drawback of DNN method is the lack of systematic law for selecting of initial Hurwitz matrix. Therefore، in this paper the subspace identification method is proposed for selecting this matrix. The required data is obtained from compressor Moore-Greitzer simulated model. In the residual evaluation block، a specified algorithm is proposed for obtaining fault properties. Virtual sensor idea is utilized for fault tolerant control system. A number of simulation results are carried out to demonstrate and illustrate the advantages، capabilities، and performance of our proposed fault toleant control scheme.

Active fault detection consists of finding an auxiliary input signal the use of which allows detection of the masked faults using a multi-model framework in continuous or discrete- time cases. In this paper, a modified approach to optimal auxiliary signal design in robust fault detection based on a multi-model formulation of healthy and faulty systems is used to study the problem of active fault detection for a class of systems with nonlinear coupled continuous state-space equations in the presence of uncertainties and disturbances. Due to the nonlinearity in the state-space equations, the traditional active fault detection approach is not straightforward to be employed. To overcome this difficulty, a modified solution is proposed in order to design an optimal auxiliary signal to guarantee robust fault detection for this class of nonlinear systems in the presence of uncertainties and disturbances. Finally, the proposed solution for optimal auxiliary signal design is applied to a Lumped Tire-Road Friction system.

In this paper, a model-based active fault tolerant control system (FTCs) is proposed for three phase induction motor (IM) drives subjected to the mechanical faults caused by both stator and rotor failures. FTCs structure consists of two main parts. The first part is a nominal controller based on feedback linearization for fault-free case to achieve control objectives (rotor flux and speed control). The second part is a sliding mode observer (SMO) in order to estimate additive faults which model mechanical faults in the state space model of IM. This observer has been used not only for fault reconstruction and production of additional control inputs for compensating their undesirable influences on performance of IM, but also for online estimation of axial fluxes in any operating conditions. The simulations results are shown to illustrate the effectiveness of the proposed approach to compensate the mechanical faults in IM.

Anti lock braking system (ABS) is one of the most important safety devises in vehicles during the severe braking. In this system, by regulating the wheel longitudinal slip at its optimum value, the maximum braking force is generated and therefore the minimum stopping distance for the vehicle is achieved. The hard nonlinearity due to the saturation of tire forces and modeling uncertainties are the main difficulties arising in the design of ABS. Also, the system states including the longitudinal speed and the wheel slip are not directly measurable and have to be estimated. In this paper, a nonlinear optimization based controller is analytically designed for ABS and is combined with a nonlinear estimator based on Unscented Kalman Filter (UKF). This estimation algorithm directly uses nonlinear equations of the system and does not require the linearization and differentiation. Here, the performance of the designed controller in the presence of states estimation and parametric uncertainties is analytically investigated. The simulation results indicate the efficiency of the proposed controller in tracking the optimal longitudinal wheel slip.