No abstract

This study aims to develop a novel automated computer vision algorithm for quality inspection of surfaces with complex patterns. The proposed algorithm is based on both an autoencoder (AE) and a fully convolutional neural network (FCN). The AE is adopted for the self-generation of templates from test targets for defect detection. Because the templates are produced from the test targets, the position alignment issues for the matching operations between templates and test targets can be alleviated. The FCN is employed for the segmentation of a template into a number of coherent regions. Because the AE has the limitation that its capacities for the regeneration of each coherent region in the template may be different, the segmentation of the template by FCN is beneficial for allowing the inspection of each region to be independently carried out. In this way, more accurate detection results can be achieved. Experimental results reveal that the proposed algorithm has the advantages of simplicity for training data collection, high accuracy for defect detection, and high flexibility for online inspection. The proposed algorithm is therefore an effective alternative for the automated inspection in smart factories with a growing demand for the reliability for high quality production.

This paper presents a sliding mode observer (SMO) with new reaching law (NRL) for observing the real-time linear speed of a controllable excitation linear synchronous motor (CELSM). For the purpose of balancing the dilemma between the rapidity requirement of dynamic performance and the chattering reduction on sliding mode surface, the proposed SMO with NRL optimizes the reaching way of the conventional constant rate reaching law (CRRL) to the sliding mode surface by connecting the reaching process with system states and the sliding mode surface. The NRL is based on sigmoid function and power function, with proper options of exponential term and power term, the NRL is capable of eliminating the effect of chattering on accuracy of the angular position estimation and speed estimation. Compared with conventional CRRL, the SMO with NRL achieves suppressing the chattering phenomenon and tracking the transient process rapidly and accurately. The stability analysis is given to prove the convergence of the SMO through the Lyapunov stability theory. Simulation and experimental results show the effectiveness of the proposed NRL method.

In order to solve the problem that the test time is long and the test efficiency is affected in the process of IC test. With the increase in the complexity of integrated circuits, it is difficult now to diagnose the faults. To overcome this situation, there is a need to upgrade the test strategies. Based on the fault probability model, the order of test types and test vector is being adjusted. To improve the test efficiency, the high-quality test types and test vectors are loaded first, and the fault circuits are hit earlier. A hierarchical dynamic method for IC test flow is proposed. The Bayesian probability model was established by counting the failure rates of each test type and each test vector in the sample integrated circuit, and the loading sequence of each test vector was adjusted according to the probability of hitting the fault point. As the test progresses, the test data are collected constantly, the test failure rates of test type and test vector are dynamically updated, and the loading sequence of test type and test vector is adjusted synchronously. It is proved that the final circuit test time is reduced to 32.172s by the dynamic adjustment method, and the test time is reduced by 53.9%. The use of dynamically adjusted test process can find the fault circuit earlier, significantly reduce the test time of the fault circuit, and improve the test efficiency.