A single image de-raining algorithm based on low rank matrix completion is proposed. The algorithm adopts the three-stage strategy of detection, repair and optimization. In the rain detection stage, the rain intensity information is used as a prior to build a detection model. In the repair stage, a similar patch matching algorithm is first used to construct a similar patch matrix, and then the problem of de-raining is transformed into the task of low-rank matrix completion based on its low-rank attribute. In the optimization stage, a correction strategy is adopted to further improve the de-raining effect and objective measurements. The algorithm is verified on synthetic rain images and real rain images. Experimental results show that the algorithm shows a good rain removal effect, and the processing of heavy rain images is also satisfactory. Compared with other methods, it has certain advantages in both objective metrics and subjective visual quality.

In order to solve the problem of low recognition accuracy of current vehicle color and model recognition methods based on machine learning, a vehicle color and model recognition method based on convolutional neural network is proposed. The method uses YOLOv3(You Only LookOnce Version 3) algorithm in Darknet network to detect and locate the vehicle face, and then the vehicle color and model recognition algorithm based on convolutional neural network is used to identifythe vehicle color and model. This is a multi-attribute recognition method for vehicle, it is different from the recognition method of single vehicle attribute. On public traffic data collection of Peking University Vehicle Datasets experiment, the experimental results show that the vehicle color and model recognition accuracy of 93.75% at the same time, the recognition accuracy of vehicle color is 94.98%, the recognition accuracy of vehicle model attribute recognition is 98.38%, It is obviously better than the vehicle attribute recognition algorithm based on machine learning, the algorithm is proved to be feasible and effective. Finally, the vehicle color and model recognition technology is applied to the intelligent parking fee system.

With the development of online shopping, the demand for order printing has increased, and the cluster printing system can effectively improve efficiency. However, the cluster system requires high robustness and reliability, so order monitoring and handling of printing equipment failures have become the core issues of the cluster printing system. The SDG technology with the characteristics of real-time monitoring node data and revealing the fault propagation path is applied to the cluster printing system. The fault diagnosis reasoning rules of the cluster printing system are established, and the “If-Then” form of diagnosis rule base is formed. In addition, an order-full-life tracking model is constructed, and fault tasks are identified and managed in combination with a diagnostic rule base to implement fault task transfer and self-recovery of the cluster printing system.

In view of the unsteady motion and low overall running speed of the hand by hand teaching spray robot, an optimization method based on spline curve fitting and motion planning is proposed. By using the non-uniform B-spline curve to fit the teaching track, the position and attitude fitting curve is obtained. The position curve and attitude curve are discretized separately. The position curve is discretized based on S-type acceleration and deceleration, and the two-way interpolation method is used to find the deceleration point. The problem of attitude and position synchronization is solved by establishing the mapping relationship between the parameters of attitude curve and position curve. Simulation and experiment show that the algorithm can effectively improve the stability and running speed of the teaching robot.

A dual-aspect item attention network (DIAN) for item-based recommendation is proposed, which jointly takes into account the aspects of importance of historical items in a user profile to the target items and the underlying relations among these items. DIAN consists of two main modules, a neural attentive model for item similarity between historical and target items (NAIS), and a dual normalization self-attention item similarity model for item similarity underlying historical items (SAIS). On one hand, the neural attentive model is introduced to distinguish the different contribution of the historical items in a user profile to the perdition on the target item. On the other hand, a self-attention network is proposed to infer the item-item relationship from users’ historical interactions, which is able to estimate the relative weights of each item in user interaction trajectories, in order to learn better representations for users’ interests. Furthermore, a self-attention network is proposed using a dual normalization mechanism, consisting of a layer focusing on extracting users’ representation from historical items, and a layer making it unaffected by the number of users’ historical items. Extensive experiments conducted on two public benchmarks demonstrate the proposed method outperforms the state-of-the-art recommendation models.

To solve the problem of boundary blurring of current unsupervised monocular depth estimation method, a network architecture is proposed based on dual attention module. This architecture can effectively solve the problem of boundary blurring of depth estimation by using long-range context information of image features. The model framework that includes depth estimation network and pose estimation network is trained by an unsupervised method based on view synthesis and estimation depth and camera pose transformation at the same time. The dual attention module is embedded in the depth estimation network, including position attention module and channel attention module. This module can represent the long-range spatial location and the context information between different feature maps, so that the network can estimate the depth information with better details. The experimental results on the KITTI dataset and the Make3D dataset show that our method can effectively improve the accuracy of the monocular depth estimation and can solve the depth estimation boundary blur problem.

The special features of microblog text cause difficulties in building microblog user interest portrait. To address the problem, an ensemble algorithm based on NWD-Bi-LSTM-XGBoost is proposed. Firstly, a new word discovery algorithm from the perspective of support is raised to deal with the informality of microblog text, exploring the ubiquitous internet phrases and achieving more accurate word segmentation and semantic understanding. Then, a Simhash algorithm is introduced to mitigate the information overload of microblog text. To improve the feature sparsity caused by microblog text’s conciseness, bidirectional long short-term memory networks are used to extract semantic features. Finally, the XGBoost model is trained by combining the static features of microblog users with the semantic features of the blog text for constructing the multi-granularity microblog user interest portrait efficiently. The experimental results show that the macro-average F1 score and AUC value of coarse-granularity (primary) interest tag model are up to 83.6% and 79.7% and that of fine-granularity (secondary) interest tag model are 70.4% and 63.6%, respectively. Compared with other benchmark models, the macro-average F1 score and AUC value of the models increase by 3%~5% due to ensemble of the NWD algorithm, which is superior to the existing new word discovery methods.

With the access of distributed generation(DG), the direction of fault overcurrent in the distribution network is no longer unique, and hence the traditional method about fault section location will not be applicable. A nonlinear complementary constrained smoothing model is proposed with a solution method for fault section location in distribution networks with DG. Considering the deficiencies in numerical stability and decision efficiency based on logical relationship of the indirect fault location method, a switch function is constructed based on algebraic relationships that can adapt to multiple DG switching situations. Two simulation algorithms and contrast experiments show that nonlinear complementary constrained smoothing method of fault section location in distribution networks with DG has a high fault tolerance performance and can achieve exact orientation under multi-information distortion, and so it can be seen that our algorithm is superior to intelligence algorithms in numerical stability.

It is a common investment decision-making method for investors to comprehensively consider expert advice. By aggregating active expert advice, a new online portfolio strategy is proposed. First, considering all constant rebalanced portfolio strategies as experts, an active expert set is constructed by eliminating the worst recent performing expert. Second, using the weak aggregating algorithm to aggregate all active expert advice, an online portfolio strategy is then constructed. The proposed strategy is numerically analyzed by using actual stock data. The results show that the strategy has a more competitive performance.

Let G be a finite group, H is a subgroup of $G$. H is S-semipermutable in $G$, if $H{G_p} = {G_p}H$ for any prime divisor p of ${\rm{|}}G{\rm{|}}$ and Sylow p-subgroup ${G_p}$ with $(p, \left| H \right|) = 1$. The influence of the S-semipermutability of subgroups on the p-supersolubility of a finite group is investigated. Some interesting results are obtained which generalize some known results.

Oscillation of numerical solutions is studied with regard to a class of delayed Gompertz equations, which have been widely used in description of the population dynamics and tumour growth. Firstly, the linearized equations are obtained by Taylor formula and its corresponding difference equations by linear θ method. Secondly, the oscillation theory is applied to analyze those obtained equations. In the process, the oscillation is primarily discussed through studying the properties of roots for the corresponding characteristic equations. For requirement, the oscillation of numerical solutions is discussed while the variable θ belongs in different scopes. Accordingly, the sufficient conditions under which numerical solutions oscillate are acquired. To verify the results, some numerical experiments are given. The first three experiments validate the conditions of the delayed Gompertz equation which has three delay terms. And the rest of the experiments check on another which has two delay terms.

Aiming at the problem of electrode wear in electrical discharge machining(EDM), a drilling method for axial compensation of electrode wear is proposed. A series of blind holes of different intended depths are drilled on the micro-EDM machine (Sarix SX-200hpm) using the proposed method and they are compared with holes drilling without compensation. The brass and copper are employed as tool electrode and the workpiece is Titanium alloy (TC4). As a result the depth of blind hole processed by this method can be close to the expected and the morphology of blind hole processed with the copper electrode is better.

A novel solar energy solution regenerator is presented which uses parabolic trough collector to heat the LiBr-H₂O solution. The solution evaporates in the regenerator and completes the regeneration. The performance of heat absorption and regeneration of the regenerator with or without the pre-heated progress are analyzed. The results indicate that LiBr-H₂O solution in the regenerator can produce steam constantly and steadily in both situations. The daily vapor generating efficiency under the situation of pre-heating is 0.186 which is 3.8 times higher compared with the situation without pre-heating.

A facile method is developed to modify the surface of nano-CuO by absolute ethanol quenching was developed, which results in enhancing the photocatalytic performance. The nano-CuO is heated to a high temperature (800 ℃) and immediately quenched by submersion in absolute ethanol. The photocatalytic decomposition of rhodamine B demonstrates that, under ultraviolet light irradiation, a better photocatalytic performance is achieved with our modified CuO. The characterization of the samples indicates that by absolute ethanol quenching, the interaction between hot CuO and absolute ethanol leads to the introduction of a high concentration of oxygen vacancies on the surface of the Nano-CuO. The CuO-Cu₂O heterojunctions and Cu-CuO/Cu₂O Schottky heterojunctions may be formed as well because of the Cu₂O and Cu reduced by absolute ethanol. This further illustrates that the ethanol quenching method can effectively modify the surface of metal oxides, increase surface oxygen vacancies on the oxides, even form heterojunctions, and improve the photocatalytic performance of the materials.

Neuroinflammation runs through the whole course of neurodegenerative diseases. In normal physiological state, nerve inflammation contributes to the repair of nervous system damage, while excessive inflammation can cause cellular damage, accelerating the deterioration of neurodegenerative diseases. When nerve inflammation occurs, microglia cells are activated, which makes them a sensitive and specific quantitative indicator to reflect the pathophysiological changes of microglia cells. Nuclide and visible light imaging technologies were used to detect neuroinflammatory targets, and the research progress of molecular probes targeting TSPO targets in recent years was introduced, including nuclide imaging and fluorescence imaging. Finally, the application prospect and market value of the neuroinflammatory molecular probes are summarized and forecasted, which is of certain reference significance for the development of novel neuroinflammatory molecular probes.

Based on the background of urban renewal and the current policies on the encouragement for young people’s innovation and entrepreneurship, the Youth Apartment and Hackerspace which have been more and more developed in recent years are focused on. With several field studies on the representative cases in Guangzhou and Shenzhen, doing some investigations on the salient issues such as social challenges, living conditions, community cultures, and regional identities of the youth groups, the design idea and constructive strategy of the symbiotic between the Youth Apartment and Hackerspace are put forward. which has the values for promoting community sharing, embracing contemporary needs, and fitting the trend of the urban development.

Aiming at the development bottlenecks of many new R&D institutions, such as inefficient innovation and ineffective innovation, based on SNM theory, a path to improving the development quality of new R&D institutions is designed from five dimensions: vision building, technology selection, network construction, results transformation and interactive learning, and then through condensing international high-level R&D. Successful practices of R&D institutions are proposed to enhance the high quality development of new R&D institutions, aiming at providing a new perspective and reference for improving the quality and efficiency of new R&D institutions in China.