Some Improved Control Charts Based On Ranked Set Sampling Schemes For Monitoring The Process Location

Quality is the cornerstone of every competitive and successful business whether it is manufacturing or service.The level of success depends upon the quality of products or services being provided.According to modern definition,quality is inversely proportional to variability,thus reduction in the variability results in quality improvement.Statistical process control(SPC)plays a fundamental role in quickly identifying assignable causes of variation which may undermine the manufacturing or service delivery process,and thus aids in reducing the variability.Quality control charts are a powerful online process monitoring tools of SPC,which are being used in numerous manufacturing and service industries to monitor designed processes to prevent possible deviations from the standard parameters.The process monitoring and construction of control charts involve the periodic selection of rational subgroups/samples from the process to measure the quality characteristic(s)of the manufactured products or services under certain constraints.Good representative samples from the process can help to better understand the behavior of the variable of interest being monitored and the parameters can be estimated more efficiently.Therefore,the choice of sampling scheme becomes pivotal to select good representative subgroups.To this end,simple random sampling(SRS)scheme is,generally,used in the existing SPC literature.There are many processes where some auxiliary information correlated with the variable of interest is readily available or can be obtained cheaply.This additional information can be advantageously used to select more representative subgroups from the process being monitored.The SRS scheme neglects such auxiliary information,but there are other sampling schemes such as ranked set sampling(RSS)scheme,and its several types which effectively utilize this kind of information.The RSS estimators,under symmetric parent population,are unbiased and efficient,with lower mean square errors,as compared to their SRS counterparts.Moreover,the RSS schemes also encompass the SRS scheme as their special case.Thus,RSS schemes can be helpful in designing more flexible and efficient control charts for improved process monitoring.This thesis presents some improved control charting strategies which make advantageous use of different ranked set sampling schemes for more flexible and efficient monitoring of the process location.Firstly,two optimal Shewhart-type control charts are designed by integrating two types of RSS schemes,the traditional RSS and median RSS(MRSS),with the generalized multiple dependent state(GMDS)plan.These charts are optimized to detect mean shift of a specified magnitude and consists of two pair of control limits to consider both the current and some previous samples information to make decision about state of the process.Secondly,a new type of RSS scheme,namely neoteric ranked set sampling(NRSS),is considered to design Shewhart,cumulative sum(CUSUM),and exponentially weighted moving average(EWMA)control charts to enhance their mean shift detection abilities.Thirdly,combined application of Shewhart-CUSUM(CSCUSUM)and Shewhart-EWMA(CSEWMA)under the NRSS scheme is proposed for efficient detection of both the small and large mean shifts in the process.Moreover,robustness of these charts is assessed against non-normal and contaminated process environments.Lastly,a homogeneously weighted moving average(HWMA)control charting structure under four different variations of RSS schemes(namely RSS,MRSS,extreme RSS(ERSS),and NRSS)is proposed as an alternate to EWMA control chart for quick detection of process mean shifts.The run length profiles of the proposed control charts are obtained through numerical simulations and their performance is compared with their competitors by using various performance indices.In addition,illustrative examples are provided by using real data of industrial processes which demonstrated the effectiveness of proposed techniques.Based on the findings of this study,these charting strategies are recommended for practical implementation with reliability,especially for those processes where auxiliary information related to the variable of interest is easily available.