Marcin Mrugalski

An unscented Kalman filter in designing dynamic GMDH neural networks for robust fault detection

This paper presents an identification method of dynamic systems based on a group method of data handling approach. In particular, a new structure of the dynamic multi-input multi-output neuron in a state-space representation is proposed. Moreover, a new training algorithm of the neural network based on the unscented Kalman filter is presented. The final part of the work contains an illustrative example regarding the application of the proposed approach to robust fault detection of a tunnel furnace.

Confidence estimation of GMDH neural networks and its application in fault detection systems

This article deals with the problem of determination of the model uncertainty during the system identification via application of the self-organising group method of data handling (GMDH) neural network. In particular, the contribution of the neural network structure errors and the parameter estimates inaccuracy to the model uncertainty were presented. Knowing these sources and applying the Outer Bounding Ellipsoid (OBE) algorithm it was possible to calculate the uncertainty of the parameters and the model output. The mathematical description of the model uncertainty enabled designing the robust fault detection system, whose effectiveness was verified by the DAMADICS benchmark.

Neural network-based robust actuator fault diagnosis for a non-linear multi-tank system

The paper is devoted to the problem of the robust actuator fault diagnosis of the dynamic non-linear systems. In the proposed method, it is assumed that the diagnosed system can be modelled by the recurrent neural network, which can be transformed into the linear parameter varying form. Such a system description allows developing the designing scheme of the robust unknown input observer within H∞ framework for a class of non-linear systems. The proposed approach is designed in such a way that a prescribed disturbance attenuation level is achieved with respect to the actuator fault estimation error, while guaranteeing the convergence of the observer. The application of the robust unknown input observer enables actuator fault estimation, which allows applying the developed approach to the fault tolerant control tasks.

Towards Robust Neural-Network-Based Sensor and Actuator Fault Diagnosis: Application to a Tunnel Furnace

The paper shows a unified approach for designing both sensor and actuator fault diagnosis with neural networks. In particular, a general scheme of the group method of data handling neural networks is recalled. Subsequently, a unscented Kalman filter approach for designing the network and determining its uncertainty is briefly portrayed. The achieved results are then used to obtain the so-called robust sensor fault diagnosis scheme. The main contribution of this paper is to show how to use the above-mentioned results for actuator fault diagnosis. In particular, the obtained neural model is used to obtain the input estimates. The achieved estimates are then compared with the original input signals to formulate the diagnostics decisions. The input estimation scheme is based on a chain of robust observers, which guaranties that the input estimates are obtained with a prescribed disturbance attenuation level while ensuring the convergence of the observers. The final part of the paper shows a comprehensive case study regarding the laboratory tunnel furnace, which exhibits the performance of the proposed approach.

Least Mean Square vs. Outer Bounding Ellipsoid Algorithm in Confidence Estimation of the GMDH Neural Networks

The paper deals with the problem of determination of the model uncertainty during the system identification with the application of the Group Method of Data Handling (GMDH) neural network. The main objective is to show how to employ the Least Mean Square (LMS) and the Outer Bounding Ellipsoid (OBE) algorithm to obtain the corresponding model uncertainty.

Dynamic GMDH Type Neural Networks

This paper presents a new identification method based on Artificial Neural Networks (ANNs) which can be used for both static and dynamic systems. In particular, a Group Method of Data Handling (GMDH) type neural network with dynamic neurons is considered. The final part of this work contains an illustrative example regarding an application of the proposed approach to the real system identification task.

ROBUST FAULT DETECTION VIA GMDH NEURAL NETWORKS

Abstract This paper presents a new parameter and confidence estimation techniques for dynamic Group Method of Data Handling Neural Networks (GMDHNNs). The main objective is to show how to employ the bounded error approach to solve such a challenging task that occurs in many practical situations. In particular, the proposed approach can be easily applied in robust fault detection schemes.

An H∞ approach to fault estimation of non-linear systems: Application to one-link manipulator.

The paper is focused on the problem of robust fault estimation of non-linear discrete-time systems. The general unknown input observer scheme and the H∞ framework are applied to design a robust fault estimation methodology. The main advantage of the proposed approach is its simplicity resulting from the boiling down of designing methodology to solving a set of linear matrix inequalities, which can be efficiently done by the application of modern computational packages. The resulting approach guaranties that a prescribed disturbance attenuation level is achieved with respect to the fault estimation error while guaranteeing the convergence of the observer with a possibly large decay rate of the state estimation error. The final part of the paper presents an illustrative example regarding the application of the proposed approach to faults estimation of the one-link manipulator.

Robust unknown input filter for fault diagnosis of non-linear systems

The paper concerns the task of robust fault diagnosis of actuators in non-linear discrete-time systems. The general unknown input observer strategy and the H∞ framework are utilised to design a robust fault detection scheme. Moreover, the proposed approach enables to perform the isolation of the faulty actuators. The final part of the paper presents an illustrative example which exhibits the performance of the proposed approach.

A GMDH NEURAL NETWORK BASED APPROACH TO PASSIVE ROBUST FAULT DETECTION USING A CONSTRAINTS SATISFACTION BACKWARD TEST

Abstract This paper focus on the problem of passive robust fault detection using nonlinear models that include parameter uncertainty. The non-linear model considered here is described by a Group Method of Data Handling Neural Network (GMDHNN). The problem of passive robust fault detection using models including parameter uncertainty has been mainly addressed checking if the measured behaviour is inside the region of possible behaviours following what will be called in the following a forward test. In this paper, a backward test based on checking if there exists a parameter in the uncertain parameter set that is consistent with the measured behaviour is introduced. This test is implemented using interval constraint satisfaction algorithms which can perform efficiently in deciding if the measured state is consistent with the GMDHNN model and its associated uncertainty. Finally, this approach is tested on the servoactuator proposed as a FDI benchmark in the European Project DAMADICS.