Search

Mining knowledge from data is an important way to solve scheduling problems. Adequate samples are a prerequisite for mining effective scheduling knowledge. However, it is difficult and time-consuming to obtain high-quality samples from the production process. To solve this problem, we propose a data mining method for smart shop floor scheduling based on GAN (Generative Adversarial Network). First, GAN is used to learn the distribution of initial samples and generate enough simulation samples to meet the data requirements for scheduling knowledge mining. Then, SVR (Support Vector Regression) is applied to perform scheduling knowledge mining, that is to establish the mapping relationship between shop floor production state and optimal scheduling strategy. The proposed method has been validated on the production system model MiniFab. Compared with the traditional sample acquisition method, the proposed method can effectively shorten the sample acquisition time and ensure the validity of the mined scheduling knowledge.

In order to improve the self-learning and adaptive capabilities of smart shop floor, this paper proposes an adaptive scheduling method based on deep Q network (DQN). In this study, a dual network scheduling model with a scheduling experience pool is established to improve the efficiency and stability of the scheduling model convergence. Combined with other various functional modules that cooperate with each other, it completes real-time interaction among smart shop floor data, and realizes online monitoring on the scheduling model completely without supervision. The proposed method is verified on the MiniFab semiconductor production shop floor model, and results have shown that the proposed scheduling method is more adaptable to changes in the production environment than simple scheduling rules, ensuring stable performance.

In the big data age, data-driven dynamic scheduling is an effective method for manufacturing systems. Through data mining, the acquired scheduling knowledge can guide production scheduling. But the knowledge may be invalid with the production scene change due to the scene relevance or timeliness of data. In order to ensure the validity of the knowledge, this paper studies the data-driven scheduling knowledge life-cycle management for the smart shop floor. It includes learning, application, online evaluation, and online updating of scheduling knowledge. Based on the combined scheduling rules, the extreme learning machine (ELM) is applied to learn knowledge. The quality control theory is used to evaluate the effectiveness of knowledge. And the online sequential extreme learning machine (OS-ELM) is adopted to update the knowledge. In the iterative update of the knowledge, the life-cycle management of the knowledge is implemented. This method is validated in minifab, a model of the semiconductor production line. The results show that the effectiveness of scheduling knowledge is improved and the performance of the smart shop floor is continuously optimized.

An investigation of deformation properties of square concrete columns confined with CFRP sheets at low temperatures were carried out experimentally. The experimental results showed that the CFRP sheets can increase the bearing and deformation capacity of concrete columns effectively at low temperatures. And the experimental curves show that area of plastic deformation increases, the curves become more stable and slow, and the peak strain increases with increasing usage of the CFRP sheets. The curves of N/N u vs. strain of steel and CFRP at different temperatures are almost the same, which indicates that the temperature has little effect on the deformation of steel and CFRP. However, the curves of N/N u vs. strain of concrete are different, which indicates that the temperature has an obvious effect on the deformation of concrete.