• Home
  • System Theory
  • Artificial Neural Networks for the Modelling and Fault by Krzysztof Patan

Artificial Neural Networks for the Modelling and Fault by Krzysztof Patan

By Krzysztof Patan

An unappealing attribute of all real-world platforms is the truth that they're susceptible to faults, malfunctions and, extra quite often, unforeseen modes of - haviour. This explains why there's a non-stop desire for trustworthy and common tracking platforms in line with appropriate and e?ective fault prognosis ideas. this is often very true for engineering systems,whose complexity is completely turning out to be as a result of inevitable improvement of recent in addition to the data and communique expertise revolution. certainly, the layout and operation of engineering structures require an elevated realization with recognize to availability, reliability, defense and fault tolerance. hence, it really is average that fault prognosis performs a basic function in glossy keep an eye on concept and perform. this can be re?ected in lots of papers on fault prognosis in lots of control-oriented c- ferencesand journals.Indeed, a largeamount of knowledgeon version basedfault analysis has been gathered via scienti?c literature because the starting of the Nineteen Seventies. for this reason, a large spectrum of fault analysis ideas were constructed. an important classification of fault prognosis ideas is the version established one, the place an analytical version of the plant to be monitored is thought to be available.

Show description

Read or Download Artificial Neural Networks for the Modelling and Fault Diagnosis of Technical Processes PDF

Best system theory books

Mathematical Systems Theory I. Modelling, State Space Analysis, Stability and Robustness: Pt. 1

This booklet provides the mathematical foundations of platforms concept in a self-contained, entire, precise and mathematically rigorous means. this primary quantity is dedicated to the research of dynamical platforms, while the second one quantity should be dedicated to keep an eye on. It combines good points of an in depth introductory textbook with that of a reference resource.

Nonholonomic Manipulators (Springer Tracts in Advanced Robotics)

This concentrated monograph builds upon an expanding curiosity in nonholonomic mechanical platforms in robotics and keep an eye on engineering. It covers the definition and improvement of recent nonholonomic machines designed at the foundation of nonlinear regulate concept for nonholonomic mechanical structures.

New Trends in Nonlinear Dynamics and Control, and their Applications

A range of papers exploring a large spectrum of latest tendencies in nonlinear dynamics and regulate, comparable to bifurcation regulate, nation estimation and reconstruction, research of habit and stabilities, dynamics of nonlinear neural community types, and numerical algorithms. The papers specialize in new rules and the newest advancements in either theoretical and utilized learn themes of nonlinear keep watch over.

System Engineering Analysis, Design, and Development: Concepts, Principles, and Practices

Compliment for the 1st version: “This first-class textual content can be precious to each process engineer (SE) whatever the domain. It covers ALL correct SE fabric and does so in a truly transparent, methodical fashion. The breadth and intensity of the author's presentation of SE rules and practices is exceptional.

Additional info for Artificial Neural Networks for the Modelling and Fault Diagnosis of Technical Processes

Sample text

9) as well as different types of synaptic connections, different dynamic neuron models can be obtained. 9) may be expanded into various other representations. 1. Neural networks composed of dynamic neuron units have a recurrent structure with lateral links between neurons, as depicted in Fig. 6(b). A different approach providing dynamically driven neural networks is used in the so-called Locally Recurrent Globally Feed-forward (LRGF) networks [38, 40]. LRGF networks have an architecture that is somewhere inbetween a feedforward and a globally recurrent architecture.

The dynamic neuron unit systematized by Gupta and co-workers in [77] as the basic element of neural networks of the dynamic type is presented in Fig. 6(a). The neuron receives not only external inputs but also state feedback signals from itself and other neurons in the network. The synaptic links in this model contain a self-recurrent connection representing a weighted feedback signal of its state and lateral connections which constitute state feedback from other neurons of the network. The dynamic neuron unit is connected to other (n − 1) models of the same type forming a neural network (Fig.

Bn ]T . Assuming that the parameter vector θ has physical meaning, the task consists in detecting faults in a system by measuring the input u(k) and the output y(k), and then giving ˆ If the fault is modelled the estimate of the parameters of the system model θ. 2) where θnom represents the nominal (fault-free) parameter vector, then the parameter estimate θˆ indicates a change in the parameters as follows: ˆ Δθ = θ − θ. 3) is greater than a predefined threshold value. The methods of threshold determining are widely presented in Chapter 7.

Download PDF sample

Rated 4.79 of 5 – based on 40 votes