Smart case-based indexing in worsted roving process: Combination of rough set and case-based reasoning

Abstract

Rough set data analysis (RSDA) has been primarily studied in order to obtain knowledge rules. Taking one group of data as an example, according to the simplified attributes and extractive rules a case was established; using case-based reasoning (CBR), another case was established based on the parameters influencing on the roving quality. In addition, RSDA was combined with CBR to build the case library. This allowed unimportant parameters to be removed and the case library to be simplified; in turn allowing easier, more efficient searching of attributes from the simplified case library. The union of Rule-Based Reasoning (RBR) and CBR means that complicated calculations around similar cases and the associated error can be avoided. Using RSDA one is able to reveal characteristic attributes and deduce knowledge rules associated with the model/problem in order to build a case library directly from historical data. So using the above procedure, allows machine settings to be defined and the prediction and control of the end-product quality to be easier.

Introduction

The entire processing procedures in a worsted mill from wool top to end product generally include spinning preparation, worsted spinning, weaving, fabric-finishing and so on. The corresponding half-finished product and control target are roving, worsted yarn, grey cloth and fabric. Obviously, as the first step of the whole process, spinning preparation is very important. The quality in spinning preparation inevitably influences the quality and efficiency of processes en-route to the final product.

According to the previous experiments, yarn evenness and yarn end-breaks on the spinning frame have well defined relationships with roving quality [1]. Controlling the evenness of intermediate products in spinning preparation, especially of roving, is an extremely important quality monitoring measure for the wool mill [2]. Therefore, roving quality becomes an important part of the yarn quality’s control. It is well known that spinning preparation process is a complex manufacturing system with uncertainty and imprecision, in which raw materials, processing methodologies, and equipment all influence roving quality. In actual production, the parameters that influenced the roving quality include the oil content of tops, moisture regain of tops, fiber mean diameter, coefficient of diameter variability, fiber mean length, coefficient of length variability, short fiber content and so on, as many as 13 variables. At present, quality control of roving in a mill depends on the traditional survey, record and adjust approach, utilizing all of the aforementioned quality parameters. Accordingly production managers and operators will test and adjust, based on historic data, material and equipment specifications, processing conditions and so on, in order to control or rectify a quality problem. Often however, the large amount of data accumulated and the time required to generate and analyse, it mean that proper objective judgments are unable to be applied to quality problems in the mill [3]. Therefore, these data are useless and production managers cannot get right decisions from them. To determine the spinnability and roving quality for the given material and processing conditions, empirical models [4], [5], [6], [7] and prediction packages [2], [8], [9], [10], [11] have been published. However, these empirical and predictions model only are used to forecast corresponding product’s quality. The mill can not get any case for reference and case library for indexing the most similar case. On the other hand, it is difficult to develop some universal practical models that can accurately predict roving quality for different mills [8].

Case-based reasoning (CBR) is an important aspect of artificial intelligence, which uses cases or accumulated experiences as stored knowledge to learn and solve problems [12]. Obviously, how to construct the appropriate case library and how to search the case fast and effectively are extremely important. They influence the efficient of solving problems directly [13], [14]. Rough set (RS) theory, proposed by Polish scholar Pawlak in 1982 [13], is a mathematical approach used for dealing with imprecision, vagueness and uncertainty in data analysis. This theory has been demonstrated to have its usefulness and versatility in successfully solving a variety of problems [14], [15], [16]. By using the concepts of lower and upper approximations in rough set theory, knowledge hidden in information systems may be unraveled and expressed in the form of decision rules [17], [18], [19], [20]. Another application of rough set theory is that of attribute reduct in databases. Given a dataset with discretized attribute values, it is possible to find a subset of the original attributes that contain the same information as the original one. The concept of attributes reduct can be viewed as the strongest and the most important result in rough set theory to distinguish itself from other theories. In recent years it has received great attention of researchers around the world and has been successfully applied in many areas, such as artificial intelligence (AI), knowledge discovery in database (KDD), pattern recognition, fault diagnosis and expert system. Based on RS, the rough set data analysis (RSDA) can be used to extract the correlations, dependencies and rules associated with a data set.

In this paper data gathered from a worsted mill were actualized by RSDA and the extraction rules were introduced into a CBR dialogue to simplify redundant attributes and build the case library with RBR. It is proposed that this is faster and more effective for searching the most similar case from the reduced case library. Meanwhile the complicated calculation of case’s similarity and the error’s fluctuation aroused by experiences were avoidable. Therefore, indexing the case libraries combined with the workers’ actual operating experiences is used to guide the roving production.