Statistical Process Control Using On-Machine Probing Data

Product quality plays a major role in the success of every manufacturing organization. The popular way to study and analyze the quality is through the use of a set of Statistical Process Control (SPC) tools. The correct application of these tools in a manufacturing scenario is fundamental to good process management and reduces process variation.Coordinate Measuring Machines (CMM) and on-machine probing are being extensively used in the inspection of mechanical components for statistical process and quality control in manufacturing processes. SPC uses certain process performance indicators and statistical methods to monitor for changes that might affect the quality of the product.  It is important that we need to understand the true reason behind process variability; and simply not whether a process is in control or parts being manufactured have been accepted or rejected. A preliminary step towards understanding inherent process variation present in cutting process is to dwell into an SPC monitoring system that deals with raw probing data. The MTConnect standard has facilitated extensions to its XML tags to integrate sensory data from the on-machine probes along with control data, which is readily available across the shop floor network. MTConnect adapters are developed with customized XML tags that successfully collect raw data from the on-machine probes. This is accomplished by indirectly establishing communication between the probe sensor and the MTConnect Agent via the machine controller. An SPC monitoring application based upon the data collected through such an MTConnect implementation is presented. The application is used to collect real machining data under multiple cutting process conditions, thereby demonstrating how certain SPC performance indicators (trending, shifting) are related to avoidable and inherent variations in the cutting process (tool wear, tool macro-geometry disparities). Improved SPC monitoring methods that incorporate knowledge of variations in estimating performance indicators are discussed.

Reference: Statistical Process Control Using MTConnect; Atluru S. Deshpande A.; Proceedings of the ASME 2012 International Manufacturing Science and Engineering Conference (MSEC2012); June 4-8, 2012, Notre Dame, Indiana, USA.

A Study of Machining Process Power Monitoring and Product Quality Prediction

The adoption of power sensor and power data analysis techniques has been expanding in the area of machine condition monitoring. Besides typical power usage analytics, machine health status and component degradation are the emerging merits of power data to provide more insights in the machine and process performance. This paper presents a methodology to monitor power consumption of a milling process and predict part quality based on a correlation model developed. A power sensor is instrumented at the main power supply of a three-axis horizontal milling center to manufacture a batch of typical aerospace components having a circular boss and bore features. A batch of 48 components is produced and the tool wear, product quality, power consumption and real-time machining parameters are monitored. The tool change is performed based on quality requirements and tolerance information. The boss and bore diameter is measured for each part using on-machine probing and compared with its nominal value, wherein the difference is used as the part quality metric. Effective power data in kilowatt from all cycles is analyzed and meaningful features are extracted from the power signal. The feature deviations from the baseline are used to interpret the performance degradation of each tool over cycles. The deviation trend is successfully correlated with the change in the part quality, verifying that the power data can be used to infer the part quality using the correlation model developed. In the future, the presented work can be validated with further testing and improved to be adaptive with multiple manufacturing process regimes. To conclude, the framework of using power data to predict machine performance in terms of health condition and part quality is highly beneficial to manage maintenance, information and product quality.

Reference: A Study of Machining Process Power Monitoring and Product Quality Prediction, Zhao W., Deshpande A., Lee J.,  The Prognostics and Health Management Solutions Conference MFPT 2012, 24-26 April 2012.