No abstract

No abstract

No abstract

No abstract

The paper proposes a method for the optimal synthesis of planar mechanisms, where a combination of cams and linkages is used in order to obtain a precise path generation. As a first step, based on Gruebler’s mobility criterion, a linkage mechanism is considered, with as many degrees of freedom as required by the generation task. One or more disk cams are then synthesized in order to reduce the system’s mobility and to obtain a single-input combined mechanical system. The final combined mechanism is able to guide a coupler point through any number of precision positions. A strategy for the global optimization of the synthesis process, based on evolutionary theory, is also proposed. A goal function is defined on the basis of dimensional and kinematic constraints and performance criteria, while a genetic algorithm is employed as an optimum searching procedure. An industrial application of the proposed methodology is described, where a path generation problem with time prescription is dealt with. The objective of the generation task is to guide a coupler point along a figure-eight trajectory, with a constant tangential velocity. Such a task is required by polishing machines for fiber optic connectors and similar components. A kinematic simulation of the optimal mechanism is used to validate the proposed synthesis methodology.

No abstract

No abstract

Convolutional neural networks with skip connections have shown good performance in music source separation. In this work, we propose a denoising Auto-encoder with Recurrent skip Connections (ARC). We use 1D convolution along the temporal axis of the time-frequency feature map in all layers of the fully-convolutional network. The use of 1D convolution makes it possible to apply recurrent layers to the intermediate outputs of the convolution layers. In addition, we also propose an enhancement network and a residual regression method to further improve the separation result. The recurrent skip connections, the enhancement module, and the residual regression all improve the separation quality. The ARC model with residual regression achieves 5.74 siganl-to-distoration ratio (SDR) in vocals with MUSDB in SiSEC 2018. We also evaluate the ARC model alone on the older dataset DSD100 (used in SiSEC 2016) and it achieves 5.91 SDR in vocals.