Abstract:The megawatt level offshore wind turbine gearbox is easily affected by the time-varying non-torque load. Aiming at this problem, this paper proposed a multi-wind speed condition optimization method for the planetary load sharing performance of the gearbox. By establishing the dynamic model of a 5 MW wind turbine gearbox system, the influence of wind speed conditions on the internal excitation and load sharing performance of the planetary was analyzed. Then, the support vector regression method was used to reconstruct the mapping relationship among wind speed conditions, gear modification parameters and load sharing coefficient. The optimization model of load sharing coefficient of the planetary considering the influence of different wind speed conditions was built to realize the optimization design of gear modification parameters. The results show that the time-varying non-torque load caused by wind speed conditions will produce the partial load on the tooth surface and bearing of the planetary stage, and it will significantly reduce the load sharing performance of the planetary stage at low wind speed conditions. Under different wind speeds, the internal excitation of the optimized wind turbine gearbox at the planetary stage is significantly reduced, and the tooth surface and bearing load distribution are more uniform, and the load sharing performance is significantly improved.

Abstract:The high-precision path tracking control of unmanned rollers relies on the accurate measurement of the vehicle's location information by sensors. However, in the harsh plateau dam environment, the positioning sensors' signals are prone to sudden changes and loss, which poses a great threat to the safety and stability of the system. For this reason, a fault-tolerant control method based on signal compensation by using self-learning model was proposed. First, according to the drift characteristics of the steering system reflected in the operation of the unmanned roller, the complex hydraulic steering system was simplified, and a linear steering model with flow loss parameter was constructed. Then, an online learning algorithm was proposed for mainly learning the flow loss characteristics of the system. Finally, a cascaded disturbance rejection controller was used to realize the path tracking control. Both simulation and real vehicle experiments were carried out. The results show that the vehicle can continue high-precision running within 40 s after the sensor fails. Compared with the cases of no fault-tolerant control and fault-tolerant control without model self-learning, the running time is extended by 18.7 times and 2.7 times respectively, which greatly improves the safety and efficiency of the system.

Abstract:In order to solve the problem of target loss due to size change in visual tracking, an improved particle filter vehicle tracking algorithm fusing visual and millimeter wave radar is proposed. First, genetic algorithm is used to improve particle degradation and resampling induced by standard particle filtering. The dynamic adaptive genetic cross probability is calculated according to particle degradation degree, and instead of mean distribution Gaussian function is used to calculate fitness. Then, the HSV histogram features are combined with the improved particle to achieve vehicle multi-target tracking. Finally, the location and size of the tracking bounding boxes are modified by the range information from radar. The experimental results show that compared with the standard particle filter, the improved particle filter algorithm significantly improves the multi-object tracking accuracy (MOTA) and multi-object tracking precision (MOTP) by 22.1% and 21.1% respectively. Compared to visual tracking algorithms, the tracking algorithm that fused radar data can improve the precision by 9.2% again.

Abstract:Overall equipment effectiveness (OEE) is a key indicator to measure the operation status of machine tool, attracting much attention from manufacturing enterprises. However, the conventional approaches have the disadvantages of inconvenience, high cost and poor universality in calculating the availability, performance and quality rate indicators of OEE. Hence, an intelligent identification approach of OEE based on power information was proposed. Firstly, the time-frequency characteristics of machine tool power information were obtained, and the sampling period feature vector was established. Then, the principal component analysis was employed to construct the status matching library. Combined with the nearest neighbor algorithm, the running status was identified, and its duration was quantified to calculate the availability. In addition, the sliding window was applied to develop the processing period feature vector, and the distance matching was used to obtain the actual number of processing pieces. Combining the number of processing pieces with the data obtained by the MES system, the performance and quality rate were calculated. To verify the feasibility and practicability of the approach, the experimental study of the milling was performed, and the relative error between the theoretical value and the actual value is 4.99%.

Abstract:The mounting head needs to move frequently between the feeder and the printed circuit board when the mounting machine works. Shortening the moving distance of the mounting head is an important way to improve the mounting efficiency. Optimization of the mounting path for arch style mounting machine is studied. Variation of the mounting path that the mounting head moves along the feeder is analyzed, and the formula of moving distance is proposed. An optimization model is established with the goal of minimizing moving distance of the mounting head. The sequence of pick-and-place is coded by component mounting positions, and a genetic algorithm is used to optimize the sequence of pick-and-place as well as the moving direction of the mounting head. The differences and characteristics of the mounting path corresponding to different picking up methods is analyzed. The results show that there are significant differences between two-way and near side picking up especially when the distribution of points is more on one side of PCB. A shorter mounting path can be obtained by two-way picking up than by near side picking up. Therefor the two-way picking up is more useful to improve mounting efficiency.

Abstract:Single-phase grounding faults frequently occur in low and medium voltage (form 6 kV to 66 kV) distribution network of China. The fault characteristics caused by single-phase ground fault are weak, and the characteristics of different types are not very distinguished, which makes it difficult to identify their types. Thus this paper proposes a single-phase ground fault type identification method based on feature decomposition and deep learning. Firstly, this method performs preliminary processing on the fault record data collected by the distribution network using Hilbert-Huang Transform (HHT) to highlight the characteristics of different fault types; then designs a deep learning model ResNet18 to learn the complex non-linear characteristics of the fault event and identifies the specific fault type. The verification through the recorded wave data collected by a domestic true test site proves that the method proposed in this paper can accurately identify multiple types of single-phase grounding faults, which can provide a reliable basis for the subsequent formulation of targeted fault handling measures.

Abstract:It is difficult to obtain wake galloping forces in wind tunnel experiments, but it is easy to acquire them by Fluent numerical simulations. In this paper, wake galloping responses of a spring-mounted downstream circular cylinder, with two degree-of-freedom in the wake of a stationary one, were simulated in staggered arrangement at streamwise spacing ratio L/D=2 and cross-stream spacing ratio T/D=1. In combination of structured meshing methods in ICEM and dynamic mesh techniques, the simulation was performed by the proposed unsteady fluid-structure coupling method, which embeds user defined code into Fluent by adopting the unsteady SST k-ω model for the 2D Reynolds-Averaged Navier-stokes (2D RANS) model. The study was carried out with reduced velocities V_r varying from 5 to 60 and Reynolds numbers Re varying from 2.4×10³ to 2.82×10⁴. Vibration responses obtained by the proposed unsteady fluid-structure coupling method were validated by experimental data and the aerodynamic forces at V_r=50 were compared to that calculated by the quasi-steady calculation method. Results indicate that the dimensionless amplitude A_y/D increases nearly linearly with the increase of V_r as a typical wake galloping phenomenon. The amplitudes have an excellent agreement with the experimental data. The wake depresses the random vortex shedding of downstream cylinder in vortex-induced resonance region. The trajectory of the downstream cylinder appears like a counterclockwise tilted oval with a clear directivity and self-limitation in the wake galloping region. Additionally, the quasi-steady calculation method has insufficient consideration on higher-order forces and vortex-shedding forces. The displacements obtained by the two methods have an excellent agreement. The self-excitation forces of primary four order frequency multiplication play a major role in controlling the displacement response undergoing wake galloping, indicating that the wake galloping is a kind of self-induced vibration.

Abstract:In order to facilitate the application of the screw pile in civil engineering, this paper deduces the expressions of the cross-sectional area and perimeter of the screw pile, and establishes the numerical calculation methods of the inertia moment based on four key sectional parameters (inside radius, screw pitch, height and thickness of thread). Then, the section geometric characteristics of the screw pile influenced by sectional parameters are discussed and compared with those of the circular pile within the scope of the size in engineering application. The results show that the screw pile cross-sectional area and perimeter increase with the increase of inside radius, height and thickness of the screw thread, but decrease with the increase of screw pitch. What's more, area increases more significantly than perimeter. The inside radius plays the most important part in influencing the cross-sectional area and perimeter. When the inside radius varies from 0.2 m to 0.5 m, the area increases 4.4 times and the perimeter increases 1.3 times. The perimeter ratio between the screw pile and the circular pile with the same sectional area is 1.00 to 1.05, and the area ratio between the screw pile and the circular pile with the same sectional perimeter is 0.91 to 0.99. Therefore, the perimeter of screw pile is only 0% to 5% larger than that of the circular pile with the same pile length and the same amount of concrete. The screw section has two main inertia moments:I_xc relative to symmetry axis of section and I_yc relative to axis perpendicular to symmetry axis and intersects with centroid of section, and I_xc is 13% to 72% larger than I_yc. They increase exponentially with inside radius, increase in contradictory trend with thickness of the screw thread, increase linearly with height of the screw thread, and decrease linearly with screw pitch. Compared with circular piles, the main inertia moment I_yc is smaller, while the other one I_xc is larger. However, compared with the circular pile with the same perimeter, the main inertia moment I_xc is smaller when thread thickness is less than 0.06m.

Abstract:The diversity of urban environment and design parameters makes the shopping center's heating and cooling loads have multiple uncertainties. Through the sensitivity analysis of three typical shopping centers in two urban environments, this paper aims to determine the key design parameters in cold region. Latin Hypercube Sampling was used to obtain 200 design combinations. Based on uncertainty and sensitivity analysis, the cooling and heating load fluctuations of the shopping center and the sensitivity ranking of 16 design parameters were obtained. Uncertainty analysis shows that the design parameters cause the HVAC (heating, ventilation and air conditioning) load to fluctuate up to 62.3%, and the change in heating load is 5 times of that in cooling load. Sensitivity analysis shows that the entrance wind speed, the ratio of skylight area to roof area, and roof U-value are three important design parameters that affect the building HVAC load of shopping centers in cold region. The urban environmental impact shows that the demand for cooling load in the high-density area is lower than that in the low-density of the city. Meanwhile, the sensitivity of the window solar heat gain coefficient is significantly higher in the open areas than in the high-density built areas.

Abstract:The dislocation etch of the crystal grown by frustum platform and ordinary flat platform was observed by optical microscope. It was found that the dislocation distribution in the upper half of the crystal grown by the frustum platform was similar to that of the crystal grown by the flat platform, and its dislocation density in some areas was lower than that of the crystal grown on the flat platform. For the lower half of the crystal grown on the frustum platform, there were few dislocation defects on the prismatic faces, and the dislocations concentrated on the lower pyramidal surface that was in direct contact with the frustum. The existence of the lower pyramidal faces prevented the formation of new dislocations at the bottom of the prismatic faces and provided better growth conditions for the growth of the prismatic faces. Hardness analysis showed that the use of the frustum platform contributed to the crystalline perfection of the prismatic faces partly.

Abstract:In order to solve the problem that the dual stack traffic can not be communicated when any protocol stack link fails in software defined wide area network, a dual stack traffic scheduling algorithm for software defined wide area network is proposed. The algorithm is based on fast rerouting mechanism. By combining segment routing over IPv6, it schedules the dual stack traffic of different parks when one of the wide area network dual stack link fails. Through the simulation in mininet, the feasibility of the algorithm is verified, and the link performance is measured by using the bandwidth measurement tool iPerf. The data shows that dual stack traffic scheduling algorithm greatly improves the quality of service and utilization rate of dual stack links between different parks.

Abstract:In the process of parking space allocation, the plane mobile stereo garage has the limitations of long waiting time for customers and low efficiency of vehicle entry and exit. In order to reduce customer waiting time and improve the service efficiency of the stereo garage, a dynamic location allocation strategy is proposed. The BP neural network model optimized by the dragonfly algorithm predicts the parking time of the vehicle, and selects the parking space of the garage according to the parking time of the vehicle, and finally uses the tabu search algorithm to search for the most suitable parking space in the area. At the same time, the average waiting time of customers, average waiting queue length, average service time and garage operation energy consumption are used as evaluation indicators to compare and analyze with the existing optimization methods of location allocation. The simulation results show that this strategy can more effectively reduce customer waiting time and improve the overall operating efficiency of the stereo garage.

Abstract:In order to improve the accuracy of facial expression recognition, a facial expression recognition based on data enhancement strategy is proposed. The training data set used in the experiment is enhanced by adding different weight allocation strategies. The weights of each training are randomly generated to ensure the diversity of the training data, and which weight distribution strategy is suitable for the enhancement of facial expression recognition data by comparing the experimental results. At the same time, it solves the problem of facial expression recognition due to the lack of diversity and low accuracy of facial expression recognition, and improves the accuracy and robustness of facial expression recognition. In addition, it also uses VGG19 feature extraction network to achieve high-precision facial expression recognition by learning robustness and distinguishing features from the data. The experimental results show that the model trained by the enhanced data in this way can improve the recognition rate of seven kinds of facial expressions on Fer2013 and extended Cohn-Kanade (CK+) database compared with the original data set.