Multivariate Sensing Control
in Polymer Processing

Multivariate Sensing Control

Intelligent manufacturing (IM) is characterized by advanced production technologies that automatically adapt to changing environments and varying requirements, with the capability to manufacture a variety of products with little assistance from operators. Mechatronics plays an important role in this rapidly advancing field, with synergistic integration of precision mechanical engineering with advanced sensing, control and computer theories, and technologies in the design of more intelligent manufacturing processes and equipment. Advancement in multivariate sensing control has continually expanded the contribution of mechatronics to IM, enabling functionalities that were not feasible before in terms of in-situ state monitoring and process control. New sensor technologies not only acquire higher resolution data at faster rates, but can also provide local computing resources for autonomously analyzing the acquired data for intelligent decision support.

A multivariate sensor is described for intelligent polymer processing (with validation of use in injection molding) that incorporates a piezoelectric ring to acquire melt pressure as well as a thermopile to acquire melt temperature and mold temperature. The process states are analyzed according to mechanistic relations to provide estimates of the melt velocity and melt viscosity. Validation experiments are implemented to characterize the sensor’s performance against an array of commercial sensors. Models of product quality with the described sensor far outperform those based on data from commercial sensors. Multivariate sensors have also been developed to measure in-situ material shrinkage for monitoring and control of injection molded part dimensions.

Multivariate in-mold sensor for measurement of melt pressure, melt temperature, melt velocity, and melt viscosity

Multivariate Process and Quality Control

An on-line model development and quality control methodology is presented for manufacturers in the process industries with the goal of enabling automated quality assurance. Given appropriate process instrumentation, the methodology starts with the characterization of statistical variation for the process while operating in steady state. Significant process conditions are then perturbed by six standard deviations to bound the expected long term process variation including lot to lot variability of feedstock materials. If the process is found to be robust, acquired process data is used to model the process behavior using principle components analysis (PCA). The PCA model is then used to accept and reject manufactured parts given real-time process data. This methodology is applied to an instrumented injection moulding process that is subjected to twelve common process faults. The results indicate that the methodology was able to detect every one of thirty three defective molding cycles caused by eight of the imposed faults as well as two additional faults that did not result in observable defective products. The quality controller did not detect the two remaining imposed process faults that did not produce defects, and also rejected three molding cycles (2% of the total) that appeared to produce acceptable products.

The methodology is highly rigorous and provides significant benefits. The benefits include the characterization of the manufacturing process’ consistency, validation of the manufacturing process’ robustness, automatic quality control, causal diagnosis of the rejected manufacturing cycles, and associated operator education. There are two significant issues. First, the methodology requires an instrumented process and characterization experiments to develop a capable process model; some manufacturers may not be able or willing to properly instrument or characterize their manufacturing processes. Second, the developed models are susceptible to extrinsic variables due to their high sensitivity; the development of robust models requires significant extrinsic variables to be brought into the characterization experiment or otherwise tightly controlled at a constant value. These issues may seem insurmountable to some manufacturers. Yet, the implementation of methodologies similar to that described herein is central to the achievement of fully automated manufacturing plants and enduring competitiveness in the global marketplace.

Loadings Scatter Plot of Principal Components

Hotellings T-Squared Scores Across Characterization and Fault Experiment

Related Publications

Check the archive for all our publications. Specific related publications will be subsequently added once links to the publications are completed.