Józef Korbicz

Soft Computing Approaches to Fault Diagnosis for Dynamic Systems

Recent approaches to fault detection and isolation for dynamic systems using methods of integrating quantitative and qualitative model information, based upon soft computing (SC) methods are surveyed and studied in some detail. SC methods are considered an important extension to the quantitative model-based approach for residual generation in fault detection and isolation (FDI). When quantitative models are not readily available, a correctly trained neural network (NN) can be used as a non-linear dynamic model of the system. The paper describes some powerful NN methods, taking into account the dynamic as well as non-linear system behaviour. Sometimes, further insight is required as to the explicit behaviour of the model-involved and it is then that fuzzy and even neurofuzzy methods come to their own in data-driven FDI applications. The paper also discusses the use of evolutionary programming tools for observer and NN design. The paper provides many powerful examples of the use of SC methods for achieving good detection and isolation of faults in the presence of uncertain plant behaviour, together with their practical value for fault diagnosis of real process systems.

Computer-aided diagnosis of breast cancer based on fine needle biopsy microscopic images

Prompt and widely available diagnostics of breast cancer is crucial for the prognosis of patients. One of the diagnostic methods is the analysis of cytological material from the breast. This examination requires extensive knowledge and experience of the cytologist. Computer-aided diagnosis can speed up the diagnostic process and allow for large-scale screening. One of the largest challenges in the automatic analysis of cytological images is the segmentation of nuclei. In this study, four different clustering algorithms are tested and compared in the task of fast nuclei segmentation. K-means, fuzzy C-means, competitive learning neural networks and Gaussian mixture models were incorporated for clustering in the color space along with adaptive thresholding in grayscale. These methods were applied in a medical decision support system for breast cancer diagnosis, where the cases were classified as either benign or malignant. In the segmented nuclei, 42 morphological, topological and texture features were extracted. Then, these features were used in a classification procedure with three different classifiers. The system was tested for classification accuracy by means of microscopic images of fine needle breast biopsies. In cooperation with the Regional Hospital in Zielona Góra, 500 real case medical images from 50 patients were collected. The acquired classification accuracy was approximately 96-100%, which is very promising and shows that the presented method ensures accurate and objective data acquisition that could be used to facilitate breast cancer diagnosis.

A GMDH neural network-based approach to passive robust fault detection using a constraint satisfaction backward test

This paper proposes a new passive robust fault detection scheme using non-linear models that include parameter uncertainty. The non-linear model considered here is described by a group method of data handling (GMDH) neural network. The problem of passive robust fault detection using models including parameter uncertainty has been mainly addressed by checking if the measured behaviour is inside the region of possible behaviours based on the so-called forward test since it bounds the direct image of an interval function. The main contribution of this paper is to propose a new backward test, based on the inverse image of an interval function, that allows checking if there exists a parameter in the uncertain parameter set that is consistent with the measured system behaviour. This test is implemented using interval constraint satisfaction algorithms which can perform efficiently in deciding if the measured system state is consistent with the GMDH model and its associated uncertainty. Finally, this approach is tested on the servoactuator being a FDI benchmark in the European Project DAMADICS.

A novel genetic programming approach to nonlinear system modelling: application to the DAMADICS benchmark problem

Abstract Nonlinear system modelling is a diverse research area where different kinds of methodologies can be employed. However, due to the large variety of this field, no approach imposes itself as the best one. The difficulty of system modelling consists in the necessity of approximating both the structure and the parameters of a system. That is why the choice of the approach to be used usually depends on a specific application. This paper presents a modified genetic programming approach for model structure selection combined with a classical technique for parameter estimation. In particular, various combinations of parameterised fixed length trees are proposed as candidate model structures. The algorithms that can be used to obtain a suitable form of these structures are proposed as well. The final part of the paper justifies the possibility of using this approach in practice, i.e. a comprehensive empirical study is performed with the data acquired from an industrial actuator.

Neuro-fuzzy networks and their application to fault detection of dynamical systems

The paper tackles the problem of robust fault detection using Takagi-Sugeno neuro-fuzzy (N-F) models. A model-based strategy is employed to generate residuals in order to make a decision about the state of the process. Unfortunately, such an approach is corrupted by model uncertainty due to the fact that in real applications there exists a model-reality mismatch. In order to ensure reliable fault detection, the adaptive threshold technique is used to deal with the problem. The paper focuses also on the N-F model design procedure. The bounded-error approach is applied to generate rules for the model using available data. The proposed algorithms are applied to fault detection in a valve that is a part of the technical installation at the Lublin sugar factory in Poland. Experimental results are presented in the final part of the paper to confirm the effectiveness of the method.

Nonlinear model predictive control of a boiler unit

Nonlinear model predictive control of a boiler unit: A fault tolerant control study This paper deals with a nonlinear model predictive control designed for a boiler unit. The predictive controller is realized by means of a recurrent neural network which acts as a one-step ahead predictor. Then, based on the neural predictor, the control law is derived solving an optimization problem. Fault tolerant properties of the proposed control system are also investigated. A set of eight faulty scenarios is prepared to verify the quality of the fault tolerant control. Based of different faulty situations, a fault compensation problem is also investigated. As the automatic control system can hide faults from being observed, the control system is equipped with a fault detection block. The fault detection module designed using the one-step ahead predictor and constant thresholds informs the user about any abnormal behaviour of the system even in the cases when faults are quickly and reliably compensated by the predictive controller.

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 Networks and their Application in Fault Detection and Diagnosis

Abstract The solution of fault detection and localization problems for complex objects and technological processes needs the application of modern methods and techniques. In this paper some problems of neural networks approach are studied. Especially, the design problems of different architectures of neural networks for static and dynamic processes are considered. In the final section of this paper the so-called GMDH neural network is applied to design a diagnostic knowledge-based system for the measuring electronic system of the dustmeter. Effectiveness of the diagnostic system is illustrated by some simulation results of the typical faults appearing in the dustmeter.

Design of a fault-tolerant control scheme for Takagi-Sugeno fuzzy systems

In this paper, a new active FTC strategy is proposed. First, it is developed in the context of linear systems and then it is extended to Takagi-Sugeno fuzzy systems. The key contribution of the proposed approach is an integrated FTC design procedure of the fault identification and fault-tolerant control schemes. Fault identification is based on the use of an observer. While, the FTC controller is implemented as a state feedback controller. This controller is designed such that it can stabilize the faulty plant using Lyapunov theory and LMIs.

Optimal sensor allocation for parameter estimation in distributed systems

Abstract - Some fundamental results of the modern theory of optimum experimental design are extended here to address the problem of determining the optimal measurement scheduling, encountered while estimating unknown parameters in mathematical models described by partial differential equations from observations of the underlying physical phenomenon being modelled. Special emphasis is put on measurements realized by optimal motion of spatially-movable sensors for which we generalize the approach advanced by Rafajłowicz in his seminal paper [16] to the case of minimizing a general performance index defined on the Fisher information matrix related to the parameters to be identified. Since only the measurability of the resulting trajectories can be guaranteed, we also show how to ease this inconvenience by introducing a suitable parametrization of the set of admissible solutions. In the latter case, we also detail how to adapt standard sequential numerical algorithms of optimum experimental design so that they could be employed for computation of trajectories in particular situations.