Higher switching frequency can suppress low frequency harmonics, but at the same time will introduce a large number of high frequency harmonics. The LCL filter can satisfy the low frequency gain and attenuate the high frequency harmonics, but this filter is a high order system with a resonant peak. In order to solve these problems, a control strategy based on state feedback is proposed. The state observer is used to observe the state variables of the system and the state feedback technique is used to improve the robustness of the system. Furthermore, the control strategy reduces the number of system sensors. The simulation and experimental results show that the control strategy is feasible.

The multi-band antenna met the demand of the multi-functional communication systems, corresponding with the trend of the system integration. A four-band compact microstrip patch antenna was proposed by adding a slot and a pin to a square microstrip patch. A fast design procedure was also examined, by which the center frequencies of each operational band as well as the input impedances could be altered by modifying the dimension of the slots and the locations of the pins. The full wave simulations of a prototype were made by commercial software. The simulated and the measured S₁₁ and the radiation patterns during the operating bandwidths were obtained, testifying the design concept and the antenna performance. This model might be used as a multi-system compatible antenna.

In order to attain the optimal ventilation performance for solar chimney under the different solar radiation intensity, a rotating collector plate solar chimney model is proposed and the fluent software is used to simulate its ventilation performance. The ventilation performance of the two plates with the bottom width of 100 mm-300 mm, the solar radiation intensity of 200-400 W/m² and the rotating angle of the collector plate at 0°-14° was studied. And the distribution of temperature and velocity field inside the chimney was analyzed. Simulation results show that the ventilation performance of solar chimney can be effectively improved by increasing the rotating angle of the collector plate, the bottom width of two plates and the radiation intensity within a certain range.

Using KMnO₄ as the source of Mn and the oxidant, the α-MnO₂ nanowires with different length were synthesized by controlling the amount of CH₃COOH under hydrothermal conditions. The structure and redox properties of the α-MnO₂ were characterized by X-ray powder diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), Raman and hydrogen temperature-programmed reduction (H₂-TPR) techniques. The catalytic activity of the α-MnO₂ in dimethyl ether (DME) combustion was also investigated. The results indicated that the length of α-MnO₂ nanowire was determined by the amount of CH₃COOH, and the length shortened with the rise of the CH₃COOH concentration. The length of the α-MnO₂ nanowire affected its redox property and catalytic combustion activity. The α-MnO₂ nanowire with the moderate length of 4-8 μm synthesized using 1.4 ml of CH₃COOH showed the best catalytic activity with the initial conversion temperature of 167℃ and the total conversion temperature of 240℃ for DME combustion. The time-on-stream 20 h-test shows that the prepared α-MnO₂ nanowire has relatively good stability.

All-acrylic poly(2-hydroxyethyl acrylate)-poly(isobornyl acrylate) (PHEA₂₅-PIBOA_n) block copolymer nano-objects were prepared by photoinitiated RAFT dispersion polymerization of IBOA in ethanol/water at 40℃ using a PHEA macromolecular chain transfer agent (macro-CTA). Polymerizations proceeded rapidly upon exposure to visible light irradiation (λ_max=405 nm, 0.4 mW/cm²) with high monomer conversion (≥ 95%) being achieved within 30 min. The kinetics of polymerization indicated that the polymerization could be separated into two stages. Gel permeation chromatography (GPC) analysis demonstrated that good control was maintained throughout the photo-PISA process, and the final block copolymers exhibited relatively low polydispersities (M_w/M_n ≤ 1.50). Spherical micelles and vesicles were prepared by this method.

The development process and practice of domestic and foreign manufacturing industries have proved that vigorously promoting the mutual development of the manufacturing industry, especially the advanced manufacturing and producer service, is the key to optimizing the industrial structure, helping the industrial transformation and upgrading, and improving the industrial competitiveness.Therefore, the vector autoregressive VAR model is used to conduct an empirical research on Guangzhou City, with the relevant data of the advanced manufacturing and production service industries during the "12th Five-Year Plan" period, into the current situation of the interactive integration of advanced manufacturing and producer service, and the relevant and related countermeasures are proposed, in order to provide a scientific basis for grasping the current state of integration of advanced manufacturing and producer service in Guangzhou, and to provide decision-making reference for government departments and industry associations.

Aiming at accurate simulation of multifactor infectious diseases and using Gauss-pressure model to simulate death in a central epidemic area, using the recovery rate and isolation intensity based on the sigmoid function to simulate the passive transformation of the infected individual, with the probability of the occurrence of its behavior obeying the dynamic probability, a multi-factor cellular automata infections disease model is established. Some comparison experiments are carried out on the model, and the results show that the model is more stable and more able than the current model to accurately simulate the actual situation of epidemic disease transmission.

A study is conducted on a mathematical model of retina oxygen distribution and the role of neuroglobin, which contains four sets of mutually coupled reaction diffusion equations. The existence and uniqueness of the model in the local solution is proved by using the Banach Fixed Point Theorem, applying L^p-theory of parabolic equation. And then the existence and uniqueness of the global solution is obtained by using the extension method.

The protection management and active utilization of urban historical and cultural heritage is an important component of urban construction management. The focus of work promoted by UNESCO in recent years is to help the protection of historical and cultural heritage by digital technology. As a new discipline direction, management information system (MIS) has been increasingly valued and promoted in the field of cultural heritage protection, and has become a common concern of the world. From the perspective of demand analysis, system design, system implementation steps, a comprehensive and in-depth research on the construction of UHCHMIS is conducted, and the technical problems to be encountered in the system construction and its solutions are discussed. Effective guidance and reference will be provided for the current digital protection of urban historical and cultural heritage.

The issue of reduction dimension about the correlation of multivariable financial assets in financial big data is analyzed. A simple search algorithm is developed to compute conditionally uncorrelated volatility models, which greatly improves the speed and precision of the estimation parameters. In order to verify the validity of the algorithm, the conditional uncorrelation between stock market, bond market, fund market, foreign exchange market and futures market is tested. The results show that the algorithm provided is very effective to solve the CUC model, and the correlations between the stock market and the other markets is negative or positive. The research provides a new method for financial big data correlation analysis, which has important theoretical significance and application value.

The detection of loop closure is a very important part of visual slam. Successful detection of loop closure can reduce the accumulated mileage drift generated by positioning algorithms. In view of the superior performance of deep convolutional neural networks in classification, the network of VGG16-Places 365 is used, which is widely used in image classification to the area of loop closure detection for the first time. The registration data are input into a trained convolutional neural network, and the output of each hidden layer corresponds to the image feature representation. Then, experiments are implemented to get an intermediate layer with higher matching accuracy, which is used to complete scene feature extraction, and then the loop closure region is obtained by calculating the similarity of the scene feature; finally, experimental tests are performed on loop closure detection dataset. Test results show that the accuracy rate of the VGG16-Places 365 convolutional neural network model is about 3% higher than the traditional ways under the same recall rate; and the the feature extraction time is about 5 to 10 times faster on the CPU and 100 times on the GPU.

Nowadays, as the patent information is huge, patent avoidance design has important application value for the majority of enterprises to set up patent barriers hindering the later competition. The combination design of different patents is an important means to realize product innovation. A method of design around is proposed for heterogeneous product patents as heterogeneous product patents contain strong heuristic knowledge. Through retrieving patent database obtaining patents and extraction patent information, structure and function analysis is fulfilled, to construct a similarity matrix and a function element graph. After that avoidance objects are determined and a suitable strategy is chosen, using the basic principles of patent infringement as a guide, the design of the new circumvention plan is obtained. This process is applied on a multifunctional household finally.

Aiming at the influence of illumination in face recognition process, a method of illumination normalization based on CycleGAN is presented. By using the Generative Adversarial Nets and the principle of image translation, the illumination style of the darker image was shifted to the brighter image, while the surface and the structure of the face kept smooth at the same time. Unpaired data sets without labels are used, in order to achieve unsupervised removal of illumination, and greatly simplify the work of data preprocessing. Finally, the CroppedYale data set is used to train a deep learning face recognition model, using this CycleGAN model to process the test set, and comparing the accuracy before and after processing. Experiments show that this method has a strong ability to reduce the influence of human face illumination on recognition rate while basically not changing the face structure, and therefore is helpful to improve the accuracy of face recognition.

An analysis is conducted into element's conduction process element and the process element set. The dependence and restriction relationship between process elements is analyzed by using the principle of correlation analysis. When the process element at a time is changed by extension transformation, the correlation between the process will lead to the conduction process, related to element transform another time, so that it gives concept of the conduction process and its conduction process element transmission probability function. By combining the process element and the set theory, the concept of process element extension set is given. On this basis, the application of it in the field of engineering is analyzed and studied.

In view of the problem that the traditional feature selection algorithm can not be applied to the multi-label learning context, a MML-RF algorithm is presented. The MML-RF improves the way of defining and searching nearest neighbor on the basis of the ReliefF, and introduces a new parameter to consider the contribution values of different labels. The improved weighting formula enables MML-RF to be used to the multi-label dataset. MML-RF algorithm makes use of mutual information as the measure of feature redundancy, and puts forward a solution to redundancy, which can get smaller subset of features. Experiments show that the feature subset of MML-RF is smaller, and has good classification effect on multi-label dataset, which can further enhance the efficiency of subsequent multi-label learning and data mining.