No abstract

With the drastic development of the globally advanced manufacturing industry, transition of the original production pattern from traditional industries to advanced intelligence is completed with the least delay possible, which are still facing new challenges. Because the timeliness, stability and reliability of them is significantly restricted due to lack of the real-time communication. Therefore, an intelligent workshop manufacturing system model framework based on digital twin is proposed in this paper, driving the deep inform integration among the physical entity, data collection, and information decision-making. The conceptual and obscure of the traditional digital twin is refined, optimized, and upgraded on the basis of the four-dimension collaborative model thinking. A refined nine-layer intelligent digital twin model framework is established. Firstly, the physical evaluation is refined into entity layer, auxiliary layer and interface layer, scientifically managing the physical resources as well as the operation and maintenance of the instrument, and coordinating the overall system. Secondly, dividing the data evaluation into the data layer and the processing layer can greatly improve the flexible response-ability and ensure the synchronization of the real-time data. Finally, the system evaluation is subdivided into information layer, algorithm layer, scheduling layer, and functional layer, developing flexible manufacturing plan more reasonably, shortening production cycle, and reducing logistics cost. Simultaneously, combining SLP and artificial bee colony are applied to investigate the production system optimization of the textile workshop. The results indicate that the production efficiency of the optimized production system is increased by 34.46%.

Multi-sensory configuration enables the collection of comprehensive information relating to the operating condition of machinery in use. However, the complex properties of multi-sensory monitoring data create serious challenges for data modelling and analysis. This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations. To utilize the synergistic information of multiple sensors, a prediction model is established based on MRA. By virtue of this model, the residual between model output and sensor observation is defined as a dynamic stability indicator, for the purpose of characterising the running status of a motor. A commonly used testing hypothesis is performed, based on the use of such an indicator for decision-making. As such, the motor speed condition can be inspected in a continuous manner on-line. The proposed approach is applied to both simulated and real-life engineering applications, where encouraging results have been demonstrated. Moreover, a comparison of this approach with representative and competing methods suggests its great potential for industry applications.

Semisupervised learning is an idea that addresses how to use a large number of unlabeled samples and a limited number of labeled samples to learn decision knowledge together. In this paper, we propose a multitask multiview semisupervised learning model based on partial differential equation random field and Hilbert independent standard probability image genus attribute model, i.e., shared semantics. In the framework of the image-like genus attribute model, data from different data sources are generated by their shared hidden space representation. Different from the traditional model, this paper uses the Hilbert independence criterion to inscribe the shared relationship of hidden expressions. Meanwhile, to exploit the correlations between labels in the label space as well, this paper uses the partial differential equation random field to inscribe the correlations between different kinds of labels in the label space and the correlations between hidden features and labels. Using the variational expectation-maximization algorithm, the whole generative process model can be inferred. To verify the effectiveness of the model, two artificial datasets and three real datasets are tested in this paper, and the experimental results verify the effectiveness of the algorithm in the paper. On the one hand, it not only improves the classification accuracy of the multiclassification problem and the multilabel problem; it also outputs the association structure between different kinds of labels and between hidden features and labels.

No abstract