Data analysis has many facets, ranging from statistics to engineering. In this paper basic models and algorithms for data analysis are discussed. Novel uses of cluster analysis, precedence analysis, and data mining methods are emphasized. The software for the cluster analysis algorithm and the triangularization is presented.
Machine controllers built from standardized software components have the greatest potential to reap open architecture benefits--including plug-and-play, reusability and extensibility. A challenge to component-based controllers relates to standardizing behavior in a non- restrictive manner to accommodate component packaging and component integration. Control component packaging requires behavior to be dependable, well-defined, and well-understood among a variety of users to help ensure the reusability of the component, the reliability of the component, and the correctness of the system built using the component. Integration of control components requires that the behavior model is consistent not just within a single component, but across all components in a system so that the components interoperate correctly. At the same time, the component behavioral model must be reasonably flexible to accommodate all behavioral situations and not be restrictive to a single programming methodology. Further, not all the behavior in the system may be pre-packaged as part of a component. Thus, another issue is the suitability of the standard behavior model for programming and integration of new control logic. Ideally, we need a vendor-neutral, tool-neutral, controller- neural behavior model to allow the export/import of any and all types of control logic programs. This paper will analyze the requirements of component-based, machine controller behavior, then offer a refinement of a Finite State Machine as the basis of a behavior model to satisfy these requirements. Examples will be presented based on the behavioral model the efforts of the Open, Modular, Architecture Controller User's Group Application Programming Interface for standardized, interchangeable machine controller components.
In recent years a growing number of government and university las, non-profit organizations and even a few for- profit corporations have found that making their source code public is good for both developers and users. In machine tool control, a growing number of users are demanding that the controllers they buy be `open architecture,' which would allow third parties and end-users at least limited ability to modify, extend or replace the components of that controller. This paper examines the advantages and dangers of going one step further, and providing `open source' controllers by relating the experiences of users and developers of the Enhanced Machine Controller. We also examine some implications for the development of standards for open-architecture but closed-source controllers. Some of the questions we hope to answer include: How can the quality be maintained after the source code has been modified? Can the code be trusted to run on expensive machines and parts, or when the safety of the operator is an issue? Can `open- architecture' but closed-source controllers ever achieve the level of flexibility or extensibility that open-source controllers can?
Virtual objects in a web-based environment can be interfaced to and controlled by external real world controllers. A Virtual Reality Modeling Language (VRML) inspection cell was created that models a real-time inspection system. The tested consists of a Cordax Coordinate Measuring Machine (CMM), a vision system for determining the part position and orientation, and an open architecture controller. Because of the open architecture, data such as the probe position and the part position and orientation, can be obtained from the controller to drive a VRML model of the system. The VRML CMM is driven using a socket connection between the collaborator's web browser and the real world controller. The current probe position, which is stored in a world model buffer in the controller, is collected by a Java applet running on the web page. The applet updates the VRML model of the CMM via the External Authoring Interface of the VRML plug-in. The part position and orientation is obtained from the vision system and the part is updated in the VRML model to represent the part's real world position and orientation. The remote access web site also contains a client-controlled pan/tilt/zoom camera, which sends video to the client allowing them to monitor a remote inspection with a PC and an Internet connection.
Real time defect detection on fine cloth is an urgent problem to solve: detecting a long and serious defect on a roll, as soon as it is produced, can reduce damages to the roll, and the consequent decrement of price. The paper describes the work performed at the Department of Energy Engineering `Sergio Stecco' of the University of Florence, in collaboration with well-known high quality wool cloth manufacturers (Marzotto) and machine builders (Sulzer, Benninger). The main goal has been to obtain a new and innovative production line, endowed with a system (based on image processing techniques) for detecting defects in real- time and thus for controlling the production process. The system is based on image processing techniques with a special attention to the real-time constraints. An architecture separating an on-line defect detection and an off-line classification has been proposed. An intelligent optical head, assembled on the loom, acquires images and detects the defects in real-time. A server has the offline task to classify each defect detected by the head. The system has been tested on a real loom, with good results in terms of reliability, false alarms and stability.
Fiber optic sensors have been established as one of the best available technologies for acquiring measurements in harsh environments. These sensors are tolerant to extreme temperature, EMI, shock and vibration, and offer reduced weight and increased accuracy over conventional instrumentation. As a result, these sensors have begun to replace conventional sensors in harsh environment applications.
A very high resolution 8,002-pixel trilinear image sensor is under development to meet customer requirement as they progress toward the next generation graphic arts scanning and industrial inspection systems. High-performance features will include an enhanced color filter scheme providing improved blue and green responsivity; better filter uniformity; lower dark current; improved, uncooled dynamic range to 15 bits; and will provide over 400,000 electrons of charge capacity. This sensor maintains a common optical length to Kodak's current line of long trilinear imagers.
This paper presents a technique for reconstructing smooth closed Bezier surfaces from coordinate measurements based on a Bernstein Basis Function (BBF) network. While various neural networks, such as the backpropagation network and radial basis function networks, have been effective in functional approximation and surface fitting these neural networks produce system dependent solutions that are not easily transferable to commercially available design software. The BBF network has an advantage over other networks by directly employing the same Bernstein polynomial basis functions that are used in describing Bezier surfaces. The BBF network is capable of implementing a close approximation to any continuous nonlinear mapping by forming a linear combination of nonlinear Bernstein polynomial basis functions. Changing the number of basis neurons in the network architecture is equivalent to modifying the degree of the Bernstein polynomials. Complex smooth surfaces can be reconstructed by using several simultaneously updated networks, each corresponding to a separate surface patch. A smooth transition between adjacent Bezier surface patches can be achieved by imposing additional positional C0 and tangential C1 continuity constraints on the weights during the adaptation process. Once adapted, the final weights of the networks correspond to the control points of the Bezier surface, and can therefore be used directly in commercial CAD software packages that utilize parametric modelers.
In this paper, an analysis of the dynamic characteristics of machine tool spindle-bearing systems is presented. The research utilized the force impact-response testing method. The results are applied to the analysis and modeling of the dynamic performance of machine tool spindle-bearing systems. As an indicator of dynamic performance, the impulse response matrices are experimentally obtained. Two types of impulse response matrices are considered: (1) with respect to (wrt) acceleration; which describes the space-coupled relationship between the vectors of the force (impact) and measured acceleration (response) and (2) wrt displacement; which describes the space-coupled relationship between vectors of the force and simulated displacement. The results indicate an interrelation between different directions of displacements, and lay a foundation for the dynamic modeling of spindle-bearing systems in view of the transfer matrix with nonzero non-diagonal elements. From an engineering point of view, the transfer function matrix can be considered a `dynamic imprint', or `signature' of system performance. As a practical example, the dynamic properties (the impulse and frequency response matrices) of the spindle-bearing system of a Barer-Proteo D/94 high precision machining center are obtained, identified and investigated. The developed approach for modeling and parameter identification appears promising for a wide range of industrial applications, including rotary systems.
Modern computer controlled manufacturing systems demand tight integration between machines distributed across the factory floor. Despite wide industry demand for simple and powerful integration between heterogeneous machines, such capabilities are still missing from current systems. The objective of this paper is to describe a toolkit, the Distributed Open Manufacturing Environment (DOME), that will allow developers to easily create such a system. The toolkit builds upon the standardized Java Platform and Java Jini distributed system. A reference implementation will be created to both prove the DOME toolkit, as well as to investigate novel distributed control methods.
A strategy has been developed to computationally accelerate the response time of a generic electronic sensor. The strategy can be deployed as an algorithm in a control system or as a physical interface (on an embedded microcontroller) between a slower responding external sensor and a higher- speed control system. Optional code implementations are available to adjust algorithm performance when computational capability is limited. In one option, the actual sensor signal can be sampled at the slower rate with adaptive linear neural networks predicting the sensor's future output and interpolating intermediate synthetic output values. In another option, a synchronized collection of predictors sequentially controls the corresponding synthetic output voltage. Error is adaptively corrected in both options. The core strategy has been demonstrated with automotive oxygen sensor data. A prototype interface device is under construction. The response speed increase afforded by this strategy could greatly offset the cost of developing a replacement sensor with a faster physical response time.