صفحه محصول - پاورپوینت تشخیص الگوی نمودار کنترل با استفاده از شبکه عصبی بهینه شده و ویژگی های کارآمد

پاورپوینت تشخیص الگوی نمودار کنترل با استفاده از شبکه عصبی بهینه شده و ویژگی های کارآمد (pptx) 45 اسلاید

دسته بندی : پاورپوینت

نوع فایل : PowerPoint (.pptx) ( قابل ویرایش و آماده پرینت )

تعداد اسلاید: 45 اسلاید

قسمتی از متن PowerPoint (.pptx) :

بِسم الله الرحمن الرحیم Control Chart Pattern Recognition Using an Optimized Neural Network and Efficient features عنوان پروژه : a r t i c l e i n f o  Article history:Received 1 November 2009Received in revised form20 March 2010Accepted 24 March 2010Available online 18 April 2010 Keywords: Control chart pattern recognitionWavelet decomposition entropiesNeural networksLearning algorithmParticle swarm optimization 2 Automatic recognition of abnormal patterns in control charts has seen increasing demands nowadays in manufacturing processes. This study investigates the design of an accurate system for control chart pattern (CCP) recognition from two aspects. First, an efficient system is introduced that includes two main modules: the feature extraction module and the classifier module. The feature extraction module uses the entropies of the wavelet packets. These are applied for the first time in this area. In the classifier module several neural networks, such as the multilayer perceptron and radial basis function, are investigated. Using an experimental study, we choose the best classifier in order to recognize the CCPs. a b s t r a c t 3 Second, we propose a hybrid heuristic recognition system based on particle swarm optimization to improve the eneralization performance of the classifier. The results obtained clearly confirm that further improvements in terms of recognition accuracy can be achieved by the proposed recognition system.' 2010 ISA. Published by Elsevier Ltd. All rights reserved. 1. Introduction Control chart patterns (CCPs) are important statistical process control tools for determining whether a process is run in its intended mode or in the presence of unnatural patterns. CCPs can exhibit six types of pattern: normal (NR), cyclic (CC), increasing trend (IT), decreasing trend (DT), upward shift (US) and downward shift (DS) [1]. All patterns other than normal patterns indicate that the process being monitored is not functioning correctly and requires adjustment 4 . Over the years, numerous supplementary rules known as zone tests or run tests [2] have been proposed to analyze control charts. Interpretation of the process data still remains difficult because it involves pattern recognition tasks. It often relies on the skill and experience of the quality control personnel to identify the existence of an unnatural pattern in the process. An efficient automated control chart pattern (CCP) recognition system can compensate this gap and ensure consistent and unbiased inter- pretation of CCPs, leading to a smaller number of false alarms and better implementation of control charts. With this aim, several approaches have been proposed for CCP recognition. Some researchers have used expert systems [2], and others have used artificial neural networks (ANNs) [313]. ANNs can be simply classified into two main categories: supervised ANNs and unsupervised ANNs. A literature review shows that the techniques which use multilayer perceptron (MLP) neural networks as the classifier have 5 high performances. The advantage with a neural network is that it does not require the provision of explicit rules or templates. Rather,it learns to recognize patterns directly by being presented with typical example patterns during a training phase. Among the ANNs,the multilayer perceptron (MLP) with back-propagation learning algorithm is perhaps the most widely used neural network model,being easy to understand and easy to implement. Also some of the researchers have used fuzzy-clustering for recognition of CCPs [14].A decision tree (DT) based classifier is also popular for the problemof CCP recognition [15]. 6 Most the proposed techniques used the unprocessed data as the input of the CCP recognition system. The use of unprocessed CCP data has many additional problems, such as the amount of data to be processed being large. On the other hand, a feature-based approach is more flexible to deal with a complex process problem, especially when no prior information is available. If the features represent the characteristics of patterns explicitly and if their components are reproducible with the processconditions, the classifier recognition accuracy will increase [11]. 7 Further, if the feature is amenable to reasoning, it will help in understanding how a particular decision was made, and this makes the recognition process a transparent process. Features could be obtained in various forms, including principal component analysis shape features [9,11], correlation between the input and various reference vectors [16], and statistical correlation coefficients [17].Based on the published papers, there exist some important issues in the design of automatic CCP recognition system which, if suitably addressed, may lead to the development of more robust and efficient recognizers. One of these issues is the selection of the features. In this paper, we propose using the entropies of the 8 wavelet packet decomposition as the prominent characteristics of the received pattern. These features are presented in Section 2. Another issue is related to the choice of the classification approach to be adopted. Here, in the classifier module, several neural networks, such as the multilayer perceptron and radial basis function, are investigated. Section 3 explains the classifier.Then,we propose a novel recognition system based on particle swarm optimization to improve the generalization performance of the classifier. Section 4 describes the particle swarm optimization and optimization method. 9