An important feature of the deep learning algorithm is that the hidden layer of the neural network is more dependent on more computing resources and a larger amount of data. In this paper, we use Triplet GAN method to...
详细信息
An important feature of the deep learning algorithm is that the hidden layer of the neural network is more dependent on more computing resources and a larger amount of data. In this paper, we use Triplet GAN method to identify human body images with tattoos using fewer data sample labels. The model achieved better results(0.9462) than the only Triplet(0.8352) and only GAN(0.9178) in the MNIST dataset, it also achieved higher recognition accuracy(77-82%) under the premise of saving the amount of data, compared with the traditional machine learning algorithm(54-84%). We provide a new idea for tattoo image recognition applications in embedded computing units.
In the modern power system, overvoltage phenomenon occurs frequently. Overvoltage generation will cause electrical equipment breakdown, explosion, and other accidents. These accidents not only cause unnecessary econom...
详细信息
In the modern power system, overvoltage phenomenon occurs frequently. Overvoltage generation will cause electrical equipment breakdown, explosion, and other accidents. These accidents not only cause unnecessary economic losses but also seriously affect the operation of power system. The popularization and application of on-line monitoring technology for UHV AC and DC transmission lines provide a guarantee for the safe operation of the power grid. Monitoring the overvoltage intrusion of UHV substations and accurately recording the parameters of overvoltage intrusion is of great significance to the study of overvoltage and the protection of over-voltage in substations. Based on the analysis of the principle of a monitoring system, this paper puts forward the mathematical model of a monitoring system. According to the basic functions of a monitoring system, it puts forward the overall design plan combining the upper computer and the lower computer, and the key problems to be solved in the monitoring system. Including the board design and the high speed data acquisition board based on FPGA design of controlsystem based on STM32, and the automatic variable rate compression technology, GPS timing and Timekeeping Technology, overvoltage invasion wave trigger judgment, high speed data acquisition and cache technology and based on TCP protocol according to the number of batch transmission.
This paper investigates the problem of the strictly(Q,S,R)-γ-dissipativity analysis for Markovian jump neural networks with a time-varying *** employing an appropriate Lyapunov-Krasovskii functional and using the ext...
详细信息
This paper investigates the problem of the strictly(Q,S,R)-γ-dissipativity analysis for Markovian jump neural networks with a time-varying *** employing an appropriate Lyapunov-Krasovskii functional and using the extended relaxed integral inequality to estimate its derivative,a delay-dependent and mode-dependent condition that guarantee the considered Markovian jump neural networks strictly(Q,S,R)-γ-dissipative is ***,a numerical example is provided to illustrate the effectiveness of the proposed method.
In the slope monitoring based on image detection,the main work is to process the acquired slope *** landslide occurred mostly in the rain,fog and other complex weather *** we can process effectively and fast fog image...
详细信息
In the slope monitoring based on image detection,the main work is to process the acquired slope *** landslide occurred mostly in the rain,fog and other complex weather *** we can process effectively and fast fog images according to the fog horizon slope vision *** would be helpful for subsequent image segmentation,object extraction,positioning,and improving the accuracy and efficiency of detection of slope *** on the visual technology in slope monitoring,we compared two kinds of defogging algorithm of slope *** is a kind of image enhancement method of non-physical model,mainly including:equalization algorithm and homomorphic filtering algorithm,McCann Retinex algorithm and multi-scale Retinex algorithm and a global *** other is the image restoration method based on physical model,including the dark channel prior bilateral filtering algorithm,and combined with the theory of dark channel prior to fog *** experimental results show that the histogram equalization method has the advantage of fast imaging quality in slope visual image processing,and is more suitable for slope monitoring.
An improved spectral reflectance reconstruction method is developed to transform camera RGB to spectral reflectance by inserting white balance and link function during the training-based method. The novelty in our met...
详细信息
An improved spectral reflectance reconstruction method is developed to transform camera RGB to spectral reflectance by inserting white balance and link function during the training-based method. The novelty in our method is the use of whitebalancing to normalize the scene illumination and link function to transform the reflectance, we use a radial basis function network to model the mapping between camera-specific RGB values and specific reflectance spectra. Experimental results indicate that the proposed method significantly outperforms currently existing methods in terms of spectral error and shape especially under the illumination not present in the training process.
During the drilling process, accurate prediction of drilling efficiency and safety plays a key role in timely adjustment of drilling process state. In general, surface parameters rate of penetration(ROP) and mud pit...
详细信息
During the drilling process, accurate prediction of drilling efficiency and safety plays a key role in timely adjustment of drilling process state. In general, surface parameters rate of penetration(ROP) and mud pit volume(MPV) are often used as important parameters to judge drilling safety and efficiency due to the bad bottom hole environment and unreliable detection devices. However, most drilling systems are underground, the structure is complex and exists many disturbances, so the state of drilling process is difficult to accurately predict. In this paper, an online support vector regression(OSVR) model is proposed to predict the ROP and MPV. First, the parameters of the model are determined by simple drilling process analysis. Then, the fast fourier transform filtering method is used to filter the high frequency disturbances of the data. Finally, the prediction model is established by support vector regression(SVR) method and the model is continuously updated by the model update method. The simulation results of industrial data show that the proposed model has a good prediction effect.
In this paper,the problem on guaranteed H∞ performance state estimation for static neural networks with a timevarying delay is investigated and the corresponding criterion is ***,a novel augmented Lyapunov-Krasovskii...
详细信息
In this paper,the problem on guaranteed H∞ performance state estimation for static neural networks with a timevarying delay is investigated and the corresponding criterion is ***,a novel augmented Lyapunov-Krasovskii functional(LKF) is *** the derivative of the LKF is estimated by the relaxed integral ***,the state estimator can be calculated by solving a set of linear matrix ***,an example is used to illustrate the effectiveness of the proposed method.
Collision avoidance is the primary problem to be solved in formation flight of multiple Unmanned aerial vehicles(UAVs). Firstly, a cooperative collision avoidance architecture of multiple UAVs is designed according to...
详细信息
ISBN:
(数字)9781728158594
ISBN:
(纸本)9781728158600
Collision avoidance is the primary problem to be solved in formation flight of multiple Unmanned aerial vehicles(UAVs). Firstly, a cooperative collision avoidance architecture of multiple UAVs is designed according to the requirement of autonomous collision avoidance of single UAV. Then a new cooperative collision avoidance method of multiple UAVs based on Kalman filter and model predictive control(MPC) is proposed. In this method, extended Kalman filter(EKF) is used to estimate the state of obstacles and target points in uncertain environment space, and to predict the trajectory of obstacles and target points. At the same time, relevant performance index functions and constraints are set up. On the basis of sharing environmental information, model predictive control strategy is used to guide and make cooperative collision avoidance decisions for multiple UAVs. The simulation results show that the proposed method is effective in uncertain environment perception and UAV collision avoidance, and the cooperative mechanism has obvious advantages.
Accurate and timely assessment of drilling system is key for achieving safety and efficiency in deep drilling. In this paper, an online assessment model is proposed by applying online sequential extreme learning mach...
详细信息
Accurate and timely assessment of drilling system is key for achieving safety and efficiency in deep drilling. In this paper, an online assessment model is proposed by applying online sequential extreme learning machine(OS-ELM). The model has been tested through the actual drilling data for drilling system safety assessment and accidents early warning. By analyzing the mechanism characteristics of accidents, well logging parameters are chosen as the input and accident types are chosen as the output. Owing to the OS-ELM is capable of updating network parameters based on new arriving data without retraining historical data, the model can be updated online for specific formation accidents information to make it more adaptable to a particular environment. The numerical test results show that, comparing with other widely used assessment techniques like support vector machines(SVM) and back propagation(BP), the proposed model has a higher accuracy and shorter recognition time.
Considering the difficulties in estimating depth from single image, in this paper, we propose a method to obtain the absolute scale depth map by combining the convolution neural network and depth filter. We compute re...
详细信息
Considering the difficulties in estimating depth from single image, in this paper, we propose a method to obtain the absolute scale depth map by combining the convolution neural network and depth filter. We compute relative transformation between consecutive frames by direct tracking features, which are extracted from RGB images and whose depthes are predicted by deep network, and then optimize relative motion by searching for a better feature alignment in epipolar line, and finally update every pixel depth of the reference frame by depth filter. We evaluate the proposed method on the open dataset comparison against the state of the art in depth estimation to evaluate our method.
暂无评论