Search

This paper presents a new approach to synchronize spatio-temporal systems including parametric identification as goal. The paper concentrates on generalized synchronization with a so-called Master-Slave (unidirectional) topology. The approach deals with the construction of a nonlinear observer to estimate the parameters of a Master system, such that the estimates can be used on-line by a Slave system to synchronize with the Master. Both, the Master and the Slave system, are approximated by the Finite Difference Method.

International audience; This work proposes an approach to detect single and multiple leaks. The task is carry out by identifying the parameters of ¯nite models associated with the leak events. The identification problem is attacked by using the Prediction Error Method (PEM). In addition, a frequency evaluation is realized to check the conditions for implementing the PEM or any other method which require an excitation condition.

In this paper a solution for the multileak reconstruction issue is introduced using a sliding mode observer scheme. Since fluid model in a pressurized single pipeline with only known pressures and flows at the ends does not satisfy the sliding mode observer conditions for leak reconstruction, a cascade scheme to satisfy such conditions is proposed. The key of the proposal is the estimation of pressures at the leak points through successive estimations of internal states up to those in which faults affect. The performance of the technique and its robustness with respect to operation point changes after the leak occurrences are illustrated by simulated and experimental results.

This work deals with the issue of faults classification on line for non linear Hessenberg systems whose faults are not isolable in steady state conditions. A stepwise procedure is proposed. The first step consists of the parameters estimation associated with one fault model when abnormal conditions are detected. As second step, a residual is calculated considering the identified model together with a periodic auxiliary input to the system. The auxiliary signal added to the system allows the discrimination of two faults scenarios via features of the residual. The novel procedure allows: (1) the identification of the fault parameters, if only one occurs and (2) the evaluation of some boundaries for the faults parameters, if two faults occurs simultaneously. To show the viability of the method, the leaks diagnosis in pipeline is considered as example.

This paper deals with issues of actuator fault tolerant PID controllers. It is studied, how residual generation based on a linear model is affected by the robust properties of a PID used in feedback. To get the sensitivity of the residual with respect to faults and to keep the robustness of the controller, a scheme is proposed with a non linear residual generator and a family of PIDs interconnected in such a way that the stability is held even with actuator faults. The conditions reported by Bhattacharyya for a family of stabilizing PID controllers are used to select the fault tolerant PID parameters considering actuator faults. Using the benchmark of the three tank system with two pumps, simulations and experimental results show the advantages of the fault tolerant PID scheme.

The paper deals with practical issues on adjusting and implementation of leaks diagnosis systems taking into account standards of the American Petroleum Industry. Considering a pilot pipeline developed specially for leak detection tests which is 132m long with 10.4cm of diameter and it is instrumented with different kinds of flow and pressure sensors at the ends of the line, we share our experience about how to cope with practical aspects of the design and adjusting of multi-leaks decision systems including open issues. The study also shows the results of three detection methods tested with the same pipeline: a model-based observer, an identification approach and a data driven method.

This work proposes a method for fault isolation by means of the structured model characterization with isolation capability, and the residual generation through dynamic principal component analysis. Specifically, the characterization is obtained using graph theory tools, and is expressed in terms of known variables and subsets of constraints. Thus, in the absence of analytical explicit models, the fault isolation task can be solved if the structured models satisfy isolability conditions and a set of nominal historical data from the process is available to carry out the dynamic principal component analysis based monitoring with adaptive standardization. Simulation results for the three tanks system show the effectiveness of the solution for fault isolation tasks.

This paper discusses some issues associated to the generation of the residual structure for fault detection and isolation using graph theory tools. It is shown that the residual structure of a causal dynamic system is as well causal, since a complete matching is only manipulation of internal variables of the graph. Then, no additional conditions are required during the matching process to get causal analytical redundancy relations. Moreover, the consideration that at least a pair of known variables, cause-effect , is necessary for each matching reduces substantially the number of cases which allows to span the residual structure.

Abstract This paper shows the significance of the structural models to solve diagnosis issues for concurrent faults in large scale systems. Considering that the capability to solve a particular fault diagnosis problem is a structural property between internal and external variables of the process with redundancy it is suggested to investigate the structural model of a process as previous work to the formulation of a fault diagnosis task. This previous study guarantees a successful residual generator task and simplifies the fault tolerance control reconfiguration issue. The generic structure of a system is analyzed in term of graph theory tools, since diverse software toolboxes are today available to straightforward handle relations between variables and their redundancy. Real applications cases, as the concurrent leaks isolation in a pipeline and faults in a gas turbine, illustrate the advantages of the structural model to formulate proper diagnosis tasks.

The Dynamic Principal Component Analysis is an adequate tool for the monitoring of large scale systems based on the model of multivariate historical data under the assumption of stationarity, however, false alarms occur for non-stationary new observations during the monitoring phase. In order to reduce the false alarms rate, this paper extends the DPCA based monitoring for non-stationary data of linear dynamic systems, including an on-line means estimator to standardize new observations according to the estimated means. The effectiveness of the proposed methodology is evaluated for fault detection in a interconnected tanks system.

This paper presents results that enable monitoring of an aerobic sequencing batch wastewater treatment (WWT) process based on online dissolved oxygen (DO) concentration measurement. It introduces the sequential process and the DO dynamics, presents a formal discrete event system (DES) model for the traditional SBR process and a modification that enables online monitoring of the respiration rate during the reaction phase and of the oxygen transfer parameter K La over process cycles. Finally a test method for verifying the presence of substrate in the reactor is introduced. Parts of the results have been validated with experimental data from the EOLI project.

This paper deals with generic properties of dynamic systems structure. The major part of the paper discusses conditions for the solvability of the Fault Detection and Isolation, FDI, issues using the Structural Analysis as framework. From a generic paths of the output to the unknown variables, one shows how to extract available inherent redundant information of the system in terms of its graph. The key of the analysis is the concept of monitorability which is related to the Analytical Redundancy Relations and can be checked by means of direct graphs. We review results concerning the FDI and Fault Tolerance Control problems and show with an illustrative example of a transport phenomena process with 14 constraints, the simplicity and advantages of the Structural Analysis.

This work studies under what conditions there exists a solution to the multi-leak detection problem in a pipeline when only measurements of pressure and flow at its extremes are available. A fluid model of finite dimension is considered with leaks distributed equally throughout the pipeline. The analysis begins reformulating the multi-leak detection problem in the following way; each residual must be robust to one leak but sensitive to the others. The conclusion of the analysis shows that this reformulation fails in general if one assumes more than two leaks with fixed positions in the pipeline. This fact is proved using geometric tools and extending the necessary and sufficient conditions presented in (De-Persis and Isidori, 2001) to get a residual generator assuming one set of faults and one set of disturbances. In particular, the residual generators designed considering a nonlinear model and simulated data with two leaks show that both leaks are perfectly isolated. However, the residuals in the case of experimental data shown that a decision logic is required to isolate the leak, because of the leak uncertain position.

This paper deals with the problem of fault detection and isolation for nonlinear Hessenberg systems. In particular, the first part of the FDI problem associated with the disturbance decoupling transformation is considered. The approach is based on a relation between the relative degree from the unknown input to the output variables for nonlinear Hessenberg systems. The major advantage for this class of systems is that a decoupled system can be obtained analytical without solving partial difference equations. Moreover this transformation can be used to define the structure of a reduced observer for the system where different observer design procedures can be used, since the subspace is uniform observable.

The modern multi-input multi-output (MIMO) control design techniques could significantly contribute to the reduction of the development time and cost of a fly-by-wire flight control system also for a small commercial aircraft. Moreover, they allow taking into account system robustness properties for stability and performance with respect to the unavoidable model uncertainties. In this paper the potential benefits of the μ analysis and synthesis theory to the flight control law design are evaluated by means of a practical benchmark problem.

This paper introduces a set of laboratory experiments to study practical issues of faults detection and identification. The laboratory installation consists of a pilot water pipeline, pressure and flow sensors located at the end of the line, a data acquisition system and the software package MATLAB and Labview. The main topics of the experiments are parameter identification and unknown input observer for residual generator. The general goal of the tasks is to force students to associate the fluid mechanics phenomena with the theoretical concepts of the fault detection and isolation problem at each step of the training.

The innovation introduced by technological growth and the need for new wind tunnel testing capabilities, as test parameter trajectory tracking, have made possible to considerably improve performance and capabilities of Wind Tunnel Automation and Control Systems. Specifically this paper deals with design and development of a real time computer system for automatic control of the main test parameters in a high enthalpy facility. The results achieved by using scheduled gain PID controllers, tuned and optimised essentially by means of a facility dynamic simulation model, show the feasibility of an automatic control system when dealing with risky and innovative processes.

This paper deals with design and implementation of a distributed PID pure feedback control system for the regulation of critical test parameters of the CIRA Injector Powered Transonic Wind Tunnel in its whole operating envelope. Fine tuning was performed by using a combined numerical/experimental approach which allows to reduce, with respect to classical approaches, costs, risks and time needed to perform control system experimental tuning. The achieved results in terms of optimised dynamic and steady performances and the low number of experimental test needed for final tuning justified the major effort spent in developing a detailed simulation model.

This paper deals with the multi-leak detection and location in pipelines. The proposed approach is based on the continuous parity equations which describe the fluid dynamic behaviour in n sections when only the states at the pipeline extremes are measured. Since the perfect decoupling conditions for multi-leak are not satisfied, a criterium with equality constraint is formulated. This allows us to minimize the effects of the state and of one leak in the residual and at the same time to maximize the effect of the rest of the leaks. To address the location problem of multi-leak a comparative scheme of the residuals with bias is proposed. It is shown by simulation that even small leak can be located and if two leaks appear, the scheme can identify both of them.

This paper introduced some experiments based on a magnetic suspension system, MATLAB and Matrixx which can be used to study the practical issues of nonlinear systems and the advantages and drawbacks of the methods to design PID controllers. Important to remark is that the weD known Ziegels-Nichols rules to adjust the controller can not be used, since the plant is unstable and it has not oscillatory behaviour for a proportional controller. In particular the physical interpretation of the dynamic behaviour and limitation of the real system are important aspects in the set of experiments. Moreover to illustrate the difference between theoretical and practical issues, students must compare results by simulation and the magnetic system in the laboratory.

This paper introduces some experiments based on an inverted pendulum and MATLAB package which can be used to study the advantages and drawbacks of two of the well recognized methods to design robust controllers; the linear quadratic regulator and its extension with robust observer if the state is not measurable (LQR/LQE-LTR) and the optimal controller based on the H∞ theory. In particular, the lack of appropriated friction nonlinear model for the pendulum makes interesting the task of design a robust controller. Therefore the inverted pendulum is considered a suitable test-case for controller design methods and demands good preparation of graduate students to achieve satisfactory control system designs. Moreover to illustrate the difference between theory and practice the students can compare and investigate both design methods (LQG/LTR and H∞) by simulation and with the inverted pendulum. The main objectives of each experiment are described here and the experiments will be presented in the IFAC'96 Congress together with a video in which versatility and simplicity of the set are shown.