As industrial systems become more complex and monitoring sensors for everything from surveillance to our health become more ubiquitous, multivariate time series prediction is taking an important place in the smooth-running of our society. A recurrent neural network with attention to help extend the prediction windows is the current-state-of-the-art for this task. However, we argue that their vanishing gradients, short memories, and serial architecture make RNNs fundamentally unsuited to long-horizon forecasting with complex data. Temporal convolutional networks (TCNs) do not suffer from gradient problems and they support parallel calculations, making them a more appropriate choice. Additionally, they have longer memories than RNNs, albeit with some instability and efficiency problems. Hence, we propose a framework, called PSTA-TCN, that combines a parallel spatio-temporal attention mechanism to extract dynamic internal correlations with stacked TCN backbones to extract features from different window sizes. The framework makes full use parallel calculations to dramatically reduce training times, while substantially increasing accuracy with stable prediction windows up to 13 times longer than the status quo.

No abstract

Nowadays, the plastic injection molding industry is ever-growing, crucial, and its plastic products can be seen everywhere. However, the mold damage problem still frustrates operators because of its high maintenance price and time-consuming maintenance process. This damage is commonly caused by foreign bodies in mold area, and the conventional mold protection method is insufficient for high-performance injection molding machines because of the uncertainty from many setting parameters. To improve detection precision of mold protection driven by a toggle mechanism ( T M ), this paper puts forward E M P , i.e., an extended Kalman filter ( E K F ) based self-adaptive mold protection method, wherein the E K F is used in current curve optimization, and the self-adaptive method ( S A M ) is proposed to gain an safety range of current curve. The E M P was verified in a 140-ton electric injection molding machine. Compared with a general method, the proposed method decreases the detected distance of mold protection by 22% under different thickness foreign bodies.

No abstract

No abstract