No abstract

This paper extends the stage-based sub-PCA modeling method originally proposed by the authors to the monitoring of batch processes with durations of uneven lengths. Two models for each stage are developed, one for the stage division and the other for process monitoring. The purposes of the stage division are two-fold, to enhance process understanding and to provide stage-division information necessary for the development of PCA monitoring models. With the proposed method, batch processes with durations of uneven lengths can be effectively monitored for fault detection and diagnosis.

No abstract

Proper control of packing is critical to the quality of injection molded parts. To control the packing phase, nozzle pressure was chosen as a better indicator of the degree of packing than cavity pressure. This allows one nozzle pressure transducer to be used for different molds and is more suitable for industrial applications. The dynamics of nozzle pressure during the packing stage were studied and found to be both nonlinear and time varying. Therefore, an adaptive self‐tuning controller was designed and implemented. Several improvements—an anti‐windup estimate, an adaptive feedforward, and cycle‐to‐cycle adaptation—were incorporated to produce excellent packing pressure control results. The controller works well for a wide range of different conditions, such as set point profiles, barrel temperatures, molds, and materials.

No abstract

Proper settings of key process variables are critical to the product quality control of mass-producing batch processes. The most widely used method in searching for the optimal process condition is the model-based optimization method (MBO). However, model development could be a challenging task in many cases. The accuracy of the model may deteriorate when the process conditions are changed. A systematic, model-free optimization method (MFO) for a type of batch process with a short cycle time and low operational cost is proposed to improve the efficiency of quality control. Instead of building a quality model, a direct search for the optimum process condition using experimental measurements is applied. Optimization algorithm is implemented as well to improve the search efficiency; both the gradient-based and the gradient-free optimization methods are discussed. The simultaneous perturbation stochastic approximation (SPSA) and the simplex search algorithm are incorporated in the MFO. The MFO method was applied to the quality control of injection molding process for demonstration using the part weight, part dimension, and focal length of molded products as quality measurements. A comparison with the Kriging modeling and optimization technique is also presented. The experimental results proved the effectiveness of the MFO technique.

In this paper, an active fault-tolerant control scheme is developed for nonlinear batch processes with sensor faults. This scheme contains three modules:  a fault detection and isolation (FDI) module, an output estimation module, and a controller. A batchwise neural-network-based FDI and a batch-wise neural-network-based soft sensor are proposed as the first and second modules, respectively, together with an iterative learning controller. In the nominal case, the measured output is used in the iterative learning control. After a fault is detected and isolated, the estimation produced by the soft sensor is used. The FDI and estimation modules are not model-based; therefore, most types of sensor faults can be addressed. To illustrate the effectiveness and practicability of this method, two examples are given:  the first one is simulation study on a three-tank system, and the second one is an experimental application on the injection molding process.

Weight is an important quality characteristic of injectionmolding products. The current work focuses on the online prediction and closed-loop control of the product weight. Previous researchers used the process setpoints as the inputs to establish weight prediction model. These models cannot reflect the weight variations at a given setting. In this study, an online weight prediction model has been developed, with the process variable trajectories as the inputs, using a principal component regression (PCR) model. A nonlinear enhancement has been made to improve the prediction accuracy of the PCR weight model. Based on such an online prediction, a closed-loop weight control system has been developed and tested experimentally. POLYM. ENG. SCI., 46:540 -548, 2006.