When the bubbles of medical plastic bottles are detected, the arbitrariness of the bubble position in the bottle body, the uncertainty of the bubble size, and the similarity between the bubble characteristics and the bottle body characteristics increase the difficulty of detecting the bubble defects. In order to solve the above problems in the detection of bubble defects, a real-time segmentation algorithm based on improved Fast Segmentation Convolutional Neural Network (Fast-SCNN) was proposed. The basic framework of the segmentation algorithm is the Fast-SCNN. In order to make up for the lack of robustness of the original network segmentation scale, the ideas of the usage of the information between the channels of Squeeze-and-Excitation Networks (SENet) and the multi-level skip connection were adopted. Specifically, the deep features were extracted by further down-sampling of the network, the up-sampling operation was merged with SELayer module in the decoding stage, and the skip connections with the shallow layer of the network were increased two times at the same time. Four sets of experiments were designed for comparison on the bubble dataset with the Mean Intersection over Union (MIoU) and the segmentation time for single image of the algorithm used as evaluation indicators. The experimental results show that the comprehensive performance of the improved Fast-SCNN is the best, this network has the MIoU of 97.08%, the average segmentation time for a medical plastic bottle of 24.4 ms, and the boundary segmentation accuracy 2.3% higher than Fast-SCNN, which improves the segmentation ability of tiny bubbles, and this network has the MIoU improved by 0.27% and the time reduced by 7.5 ms compared to U-Net, and the comprehensive detection performance far better than Fully Convolutional Networks (FCN-8s). The proposed algorithm can effectively segment smaller bubbles with unclear edges and meet the engineering requirements for real-time segmentation and detection of bubble defects.

Focusing on the problems such as low inspection speed and low accuracy caused by subjective factors in the manual inspection method of ampoule packaging quality, an inspection algorithm based on machine vision and lightweight neural network was proposed. First, threshold processing, tilt correction and cutting of ampoule regions were performed on the images to be inspected by using the threshold segmentation and affine transformation methods in machine learning. Second, the network structure of the classification algorithm was designed according to the characteristics of images and the requirements of defect recognition. Finally, the ampoule packaging defect dataset was constructed by collecting the images of the production site. After that, the proposed ampoule packaging defect identification network was verified, and the accuracy and inspection speed of the algorithm deployed on the Jetson Nano embedded platform were tested. Experimental results show that, taking the product of five ampoules each box as the example, the proposed ampoule packaging quality inspection algorithm takes 70.1 ms/box averagely, that is up to 14 boxes/s, and has the accuracy of 99.94%. It can achieve online high-precision ampoule packaging quality inspection on the Jetson Nano embedded platform.

In order to describe the Radio Frequency (RF) power amplifier with memory effect more quickly and accurately, a new X-parameter power amplifier modeling method was proposed based on the traditional X-parameter model combined with the memory effect and load-pulling of the power amplifier. Firstly,the load reflection coefficient was introduced into the new scheme. Secondly,the two-memory path model was used to extract the nonlinear function to represent the memory effect instead of the kernel function. Three variables including amplitude, load reflection coefficient and frequency were regarded as output signals to build the new Feed-Forward (FF) structure. In the end, the step signal was used instead of the original two-tone signal to simplify the model extraction method and improve the feasibility of model extraction. By using the proposed new X-parameter modeling program to model power amplifier, the simulation results of the power amplifier CGH40045F data showed that compared with the traditional X-parameter model, FF structure X-parameter model and FeedBack (FB) structure X-parameters model, the relative error was reduced; compared with the FF model and the FB model, the simulation time was reduced by 4.08 s and 1.64 s, respectively. The results prove that the model modeled by the proposed mehtod can characterize the amplifier with non-linear memory effect more quickly and effectively

In view of the shortcomings of the initial value limitation of traditional algorithms and slow convergence speed of intelligent algorithms in harmonic balance analysis, an adaptive bee colony algorithm based on local search strategy of Broyden-Fleteher-Goldfarl-Shanno (BFGS) algorithm was proposed. Based on the basic bee colony algorithm, nonlinear dynamic adjustment factor was introduced to replace the random variables in the formula, thus improving the adaptability of searching. Meanwhile, BFGS algorithm was applied to the later period of bee colony algorithm to speed up the local search capability. Simulation results show that compared with the standard bee colony algorithm, the number of iterations of the improved algorithm was reduced by 51.9%, and the proposed algorithm has better convergence performance compared with the traditional BFGS algorithm and some other improved intelligent algorithms.

To measure Chinese microblog, a framework of Credibility of Chinese Microblog based on Information Fusion (CCM-IF) was proposed by analyzing impact factors of Chinese microblog and their pedigree. Firstly, different evaluating methods were implemented for three particular features, such as text message, user, and information propagation. Secondly, a method based on Dempster-Shafer (D-S) theory was proposed to combine the features from the fuzzy nature of the credibility. Thirdly, a series of experimental validations involving two real datasets from Sina Weibo were conducted. Experimental results show that the accuracy of CCM-IF is 10%-20% higher than that of the classical ranking algorithm named LMJM (Language Modeling with Jelinek-Mercer smoothing). So, as a static indicator of quality assessment, CCM-IF can be used for microblog retrieval ordering and garbage microblog filtering.

To quickly design a two-sided dipole antenna, which is used to Wireless Local Area Network (WLAN), with small size, dual-band and broadband characteristics, the dipole patch were printed on opposite sides of the dielectric substrate, and balun feed mode was used to achieve better broadband matching. Two arms of the dipole were slotted to achieve the small and dual-band characteristics, and to meet the dual-band requirements of WLAN of 2.45GHz and 5.49GHz. The size of the whole antenna is 28mm×44mm×1.6mm. And the Neural Network (NN) with electromagnetic simulation software named High Frequency Structure Simulator (HFSS) was combined to optimize the key size of the antenna and speed up the design process. The simulation results show that, when the S11 is less than -10dB, the bandwidth of the antenna in low frequency and high frequency can reach 470MHz (2.29—2.76GHz) and 3650MHz (4.96—8.61GHz) respectively; When the S11 is less than -14dB, the bandwidth of the antenna can reach 210MHz (2.36—2.57GHz) and 770MHz (5.13—5.9GHz) respectively. The pattern has good omni-direction, and the measurement and simulation results are in good consistency. The antenna can meet the requirements of WLAN.

By separating CE from FE, ForCEs IP routers can easily configure out flexible and intelligent network. A mechanism for which FE Module can be implemented in Linux was brought forward. Based on the mechanism, A ForCEs IP router was constructed and corresponding result was given after testing. ForCEs IP routers implemented by the mechanism can meet all the requirements of ForCEs achitecture well.