Chen, Yan
A New Computerized Data Acquisition and Analysis System for KES-FB Instruments1
ABSTRACT
The Kawabata evaluation system for fabrics (KES-FB) has been commonly used to measure fabric mechanical behaviors related to hand. However, the data acquisition and analysis technique for this system is obsolete and time-consuming. The purpose of this study is to develop a new data acquisition and analysis system for the KES-FB instruments using LabVIEW(TM) software and corresponding hardware (National Instruments). Windows-- interface programs for each KES test method are created with the LabVIEW system. In order to validate this data acquisition system, twenty apparel fabrics are tested using both the original KES-FB data recording method and the new computerized data processing method. Linear regression and variance analysis are used to compare tested results and evaluate repeatability and variability within the two data acquisition methods. The R^sup 2^ values of most linear regression models are close to 1.00, indicating that the KES data system has been successfully updated.
The instruments of the Kawabata evaluation system for fabrics (KES-FB), including the tensile and shear tester (KES-FB1), bending tester (KES-FB2), compression tester (KES-FB3), and friction and roughness tester (KES-FS4), have been widely used to measure fabric mechanical properties since the 1970s [2]. Other industrial applications of these instruments were also developed [1]. However, the data acquisition method used for the KES-FB instruments has a serious drawback that jeopardizes the ability of KES-FB users to expand industrial applications. The KES-FB instruments provide two options for data acquisition. One option is to use a KES-specific auto data processing unit, which is controlled by a Pc with cpu Intel 486 33MHz or lower and Dos operation system. The software used for this data processing unit produces a special format that cannot be read by other IBM Pcs, and the unit is neither cost-effective nor compatible with general-purpose desktop computers. The other option is to use an X-Y pen recorder to record testing curves and manually calculate Kawabata parameters [3]. This data processing method is time-consuming, and the lack of digitized data hinders research. For example, a large volume of signal noise is produced when the KES-FB bending tester is used to measure very floppy fabrics with high sensitivity. High signal noise makes X- Y chart recording extremely irregular, so it is impossible to infer a fabric responding signal separate from the noise. We have developed a new data acquisition and analysis system (DAQ) to overcome these disadvantages. Existing commercially available hardware and software ensure that the new DAQ is cost-effective and affordable for most KES-FB end-users.
Constructing the Data Acquisition System
The new data acquisition and analysis system consists of a Pentium Pc, a plug-in DAQ interface board, and LabVIEW software version 5.1 [4]. Figure 1 illustrates the structure of this system. Analog signal output from each KS-FB tester, which is usually fed to an X-Y recorder, is input to an interface board through a connector block. This interface board is a product of E series multifunction I/O hardware manufactured by National Instruments, featuring 16 single-ended analog inputs, 16 bits resolution, 100 kS/s sampling rate, and easy plug-in to a PCI slot in the PC. Any brand of IBM desktop PC now available on the market could be used to install and run the DAQ hardware and software.
The LabVIEW software used for DAQ processing control provides a graphic programming environment with the G language, allowing users to build their own customized virtual instruments (VI) for testing, implementation, and control in engineering systems. Because LabVIEW relies on graphic symbols rather than textual languages to compose a program, it is suitable for end-users with little programming experience. There are only two primary steps for LabVIEW programming-design of a block diagram and execution of a virtual instrument. According to a specific end-use application, a block diagram (also called a dataflow diagram) is designed using a series of icons such as functions, structures, control terminals, and indicators, etc., to create a graphic source code for running aVI. After completion of this block diagram, running theVI program will produce a graphic user interface on the front panel of LabVIEW.
As shown in Figure 2, the block diagram programmed in this study produces a windows interface, including a curve display, Kawabata parameter display, filter type adjustment, frequency adjustment, and scan rate selection. This Windows interface enables auto recording and computing when running a KES-FB tester. After completion of a measurement, recorded data are saved in an ASCII-format file that can be conveniently input into spreadsheet software for later analyses.
Results and Discussion
To verify the processing and computing accuracy of the DAQ system, we randomly selected twenty commercial apparel fabrics and measured their mechanical properties using both the KES-FB X- Y recorder method and the computerized DAQ method. We used linear regression to evaluate the strength of the linear correlation between these two data processing methods. R^sup 2^ (Table I) of all linear regression models for tensile properties (LT, WT, RT, and EMT), shear properties (G, 2HG, and 2HG5), bending properties (B and 2HB), and surface properties (MIU, MMD, and SMD) are close to 1.00, reflecting agreement in data recorded by the two methods. Figures 3-5 illustrate the liner correlations for the KES tensile properties.
For the compressive properties, i.e., compressive linearity (LC), compressive energy (WC), compressive resilience (RC), and maximum compressive strain (EMC), the linear regression models obtained indicate low R^sup 2^ values except for WC and To. This creates a need for further analysis of the consistency between the two data acquisition methods in terms of the compression test. First, we use variance analysis [5] to determine if the instrumental data obtained by the computerized DAQ method are significantly different from those obtained by the X-Y recorder. The result of variance analysis, as listed in Table II, indicates that for all the calculated values of each F statistic referred to for each compressive parameter, probabilities of a larger F are all above 0.01. The statistical inference is therefore that there is no significant difference between the two data acquisition methods at the confidence level of 99%.
Table II shows that the variance source of error accounts for a large portion of the grand variance of each parameter. This error may result from non-uniformity of fabric materials, irregularities of fabric surface conditions (such as different hairiness and wrinkling in different locations of a fabric specimen), and other uncontrolled random errors. In this circumstance, we further use variance analysis to estimate repeatability and variability within the X-Y recorder method and the DAQ method. Repeated compression tests are executed by testing six fabric specimens cut from the same piece of polyester suiting fabric. Each specimen is measured with both the X-Y recorder and DAQ system. Variance within each single data acquisition method is listed in Table III. The estimation indicates that when using the DAQ method, the variance of LC and EMC will be two times and RC three times that from the X-Y recorder.
Referring to the X-Y chart recording method, deviation of the compression test using the DAQ method can be estimated by the following equation:
Conclusions
We have used National Instruments' hardware and software to construct a computerized data acquisition system for KES-FB testers. This system provides a user-- interface window that simplifies data recording and allows real-time calculation. Measured instrumental data are automatically stored in a computer and easily output for data manipulation. The signal noise produced by testing low-stiffness fabrics using the KES-FB bending tester with high sensitivity is also eliminated by this DAQ system. Therefore, the testing performance and efficiency of the KES-FB instruments are improved to meet the need for testing diverse textile and nontextile materials using our updated computer techniques that are inexpensive and available commercially. Statistical analysis for the comparative fabric testing shows that no significant difference exists between the X-Y chart recording method and the DAQ method. Repeatability and variability of the DAQ method are estimated by a repeat test of fabric compression on the same fabric. With reference to the X-Y chart recording method, the maximum error rate of instrumental data recorded by the DAQ system is below +/-8.9%.
1 Approved by the Louisiana Agricultural Experiment Station as manuscript no. 00-25-0326.
Literature Cited
1. American Association of Chemists and Colorists, Sueo Kawabata to Receive The Millson Award, Textile Chem. Color. 24(9), 52, 72 (1992).
2. Hearle, J. W. S., Can Fabric Hand Enter the Dataspace? Textile Horizons 6, 16-20 (1993).
3. Kawabata, S., and Niwa, M., Fabric Performance in Clothing Manufacture, J. Textile Inst. 80(1), P19-P50 (1989).
4. National Instruments Corporation, "LabVIEW(TM) User Manual," National Instruments Co., Austin, TX, 1998.
5. Sokal, Robert R., and Rohlf, James F., "Biometry, the Principles and Practice of Statistics in Biological Research," W. H. Freeman and Company, San Francisco, CA, 1969.
Manuscript received August 8, 2000; accepted November 22, 2000.
YAN CHEN AND TAO ZHAO
School of Human Ecology, Louisiana State University Agricultural Center, Baton Rouge, Louisiana, 70803, U.S.A.
BENNY TURNER
Technical Center, Albemarle Corporation, Baton Rouge, Louisiana, 70820, U.S.A.
