With the increasing necessities for reliable PCB product, there has been a considerable demand for high speed, high precision vision system to place the electric parts on PCB automatically. To identify the electric chips with high accuracy and reliability with obtained images, a classification algorithm is needed to identify the type of parts and their defects. In this paper, we design a learning vector quantization (LVQ) neural network to achieve this. From the images obtained under the versatile lighting system, characteristic features for classification are extracted, from which type of chip is identified through the neural network based classification algorithm.
12 Inspection and shape measurement of 3D objects are widely needed in industries for quality monitoring and control. A number of visual or optical technologies have been successfully applied to measure 3D surfaces. However, those conventional visual or optical methods have inherent shortcomings, which are occlusion problem and variant surface reflection problem. X-ray vision system can be a good solution to these conventional problems, since we can extract the volume information including both the surface geometry and the inner structure of the object. In the x-ray system, the surface condition of an object, whether it is lambertian or specular, does not affect the inherent characteristics of its x-ray images. In this paper, we propose a 3D x-ray imaging method to reconstruct a 3D structure of an object out of 2D x-ray image sets.
In hot plate mills the slabs from incoming reheat furnace are reduced to the desired width and thickness, being rolled out with considerable accuracy. The process of changing the plate width is controlled by a pair of edge rolls, which is called edger. The objectives of this edging process are to meet tight width tolerances of plates and to reduce the yield loss caused by trimming when irregular width is formed at the plate edge. There are several factors that result in complexity and uncertainty in width control. These include inaccurate edger set-up model, degradation of various mill equipment, variation of operation conditions, environments and variation of the dimension of incoming cast slabs. In this paper, a genetic algorithm-based PID control is proposed to ensure the control of the desired width at the exit of the mill. The approach adopted here is essentially optimization of the PID controller gains in order to minimize the error between the desired and actual slab width. Since the design parameters associated with genetic algorithm affect convergence performance, the effects of these parameters are investigated in detail. In addition, the control performance is also evaluated for various process parameters such as initial width of the incoming slab and temperature of the slab. Based on the result obtained from a series of simulations, the proposed control method is found to yield satisfactory performance for various process conditions.