This research applies Advanced Process Control (APC)-based injection molding system to improve the quality of products. The relationships between the input parameters (injection velocity, injection pressure, injection time and barrel temperature) and a single output variable (the weight of product) were found through an experimental design method. Moreover, the injection molding process model was built via a multiple regression analysis. A dynamic model turning minimum variance (DMTMV) method is utilized to control the process. Quantified improvements were further obtained from experiments. Keywords-advanced process control; design of experiment; injection molding; dynamic model turning; minimum variance I. INTRODUCTION The replacing glass, steel, leather, and wood gradually, the plastic material with features of small specific gravity, high mechanical strength, good sound and heat insulation, optimal chemical and electrical isolation, and easy-processing, has been extensively applied in electronic components, motor vehicle parts, textile fabrics, and transportation vehicles. Among methods used for processing of plastic molding, the injection molding has not only the most popular applications in virtue of its advantageous features of high production rate and good precision, mass production for complex objects with same dimensions, low costs of production, and various appropriate materials, but its crucial technology includes the control of injection molding and mold design. The three stages for injection molding are filling, packing, and cooling [1], with material features, mold structure and processing conditions affecting consequences of each stage. Moreover, the process parameters for the injection molding consist of cooling temperature, inlet and heater temperatures, injection velocity, held pressure, pressure-holding time, cooling time, and injection pressure. On the whole, the improper adjustment for parameters may lead to defects of injected products, such as flaws on appearance and faults of mechanical characteristics by means of all process parameters affecting quality of products. In this regard, the flaws on appearance include short-shot, flash, deformation, flow mark, etc. The mechanical characteristics deserving to be considered include tensile strength, extension ratio, crystallinity, and residue stress, etc. [2, 3]. Based on these factors, it is indispensible to acquire adequate processing conditions along with a qualified mold for acceleration of a production cycle and improvement in quality of molded products. However, the relationships between the status of products’ quality and configurations of process parameters may be affected for the general processes due to wear of equipments for injection molding, loss of materials, variation around the environment, and periodical maintenance. With process parameters kept constant as original settings, the output values for quality of injected products may deviate from the default target values and lead to low yield rates of outputs. Based on this perception, the status of machinery needs to be further adjusted in order to match demands for product outputs. Nevertheless, depending on engineer’s experiences in adjustment as regards settings for process parameters after occurrence of warnings for Statistical Process Control (SPC), the traditional control techniques usually miss the appropriate opportunity to rectify shortcomings in virtue of unavailable optimal values for real situations when an engineer relying on personal experiences dominates the process parameters that may fail to materialize effects in quality control of products. Recently, lots of researches regarding process control have been published by many scholars. MacGregor [4], Box & Kramer [5], and Box & Jenkins [6] offered the EWMA (Exponentially Weighted Moving Average) controller theory. Using differences between the target value and the output values at the various corresponding time points versus current time points, Box and Jenkins assigned weights with decreasing geometric distribution to the EWMA controller, i.e., data nearer current time points have larger weights. Furthermore, the fact that the EWMA controller is a MMSE (Minimum Mean Squared Error) controller in case of IMA (1, 1) (Integrated Moving Average) turbulence of process parameters has been demonstrated by Box & Jenkins. Applying the batch control theory into the semiconductor first, Guldi et al [7], Texas Instrument, explored growth of thickness of an oxide layer controlled by the time interval of oxidation. Next, the research groups in MIT such as Sachs et al [8], Boning et al [9], and Smith & Boning [10] published the batch controller theory and experimental results. In addition, the similar research was also offered by Leang et al [11] & Moyne et al [12], and Stefani et al [13] of T.I. Applied to research of processes with shift tendencies, the PCC (Predictor Corrector Control) method proposed by Butler & Stefani [14] modifies the average value of a process toward a target value. The Age Based DEWMA Control method offered by Chen et al [15] is used for estimation of real shift in a process and applied to intermittent sampling measurement. In the last research, the paper published by Tseng et al [16] indicated that EWMA is inappropriate due to significant variance for initial several batches when processes are adjusted based on historical data, and further suggested the Variable EWMA to rectify shortcomings of the conventional EWMA. Tseng et al [17] employed the initial intercept iteratively adjusted (IIIA) controller to improve variance of initial unstable several runs. Integrating the Kalman filter and time series, Chen et al [18, 19] used the time series and the recursive least squares (RLS) for the batch control in thickness of deposited films during the metal sputter deposition process. The experimental results indicate better quality of products than those by the EWMA controller. The objective of this paper is to enhance product quality of injection molding using advanced process control (APC). The first experiment performs the fractional factorial experiment to pick out critical factors affecting quality of products. The next is the RSM (Response Surface Methodology) full factorial experiment plus the regression analysis for constructing the relationships of input process parameters and output variable regarding quality of product. The recursive least squares (RLS) is used to dynamically tune the process model and then the minimum variance controller (MVC) is used to predic the next process input to match the output with a desired target value. Finally, the performance of a controller is assessed during a real production process of ADSL lower-cap plastics.
I am Frank,works as sales manager of Bestar China Co.,Ltd.which is a reputable
China plastic injection mold maker of plastic injection molding with over 15 years.
If you are looking for a China molder maker to make your injection mold,you can contact me at any time.
My company website:
http://www.bestar-china.com/
