As the training process of back-propagation neural networks converges slowly and immerses in local vibration frequently, an algorithm named parallel training algorithm is proposed, which is based on the Master/Slave m...
详细信息
ISBN:
(纸本)7312012035
As the training process of back-propagation neural networks converges slowly and immerses in local vibration frequently, an algorithm named parallel training algorithm is proposed, which is based on the Master/Slave model and training learning samples in each search subspace at the same time. The experiment results show that this algorithm converges at high rate and reaches global minimum soon.
For a long time, PID-like controllers have been successfully used in academic and industrial tasks. This is thanks to its simplicity and suitable performance in linear or linearized plants, and under certain condition...
详细信息
ISBN:
(纸本)9781509015276
For a long time, PID-like controllers have been successfully used in academic and industrial tasks. This is thanks to its simplicity and suitable performance in linear or linearized plants, and under certain conditions, in nonlinear ones. A number of PID controller gains tuning approaches have been proposed in the literature in the last decades;most of them off-line techniques. However, in those cases wherein plants are subject to continuous parametric changes or external disturbances, online gains tuning is a need. This is the case of modular underwater ROVs (Remotely Operated Vehicles) where parameters (weight, buoyancy, added mass, among others) change according to the tool they are fitted with. In practice, some amount of time is dedicated to tune the PID gains of a ROV. Once the best set of gains has been achieved the ROV is ready to work. However, when the vehicle changes its tool or it is subject to ocean currents, its performance deteriorates since the fixed set of gains is no longer valid for the new conditions. Thus, an online PID gains tuning algorithm should be implemented to overcome this problem. In this paper, an auto-tuned PID-like controller based on Neural Networks (NN) is proposed. The NN plays the role of automatically estimating the suitable set of PID gains that achieves stability of the system. The NN adjusts online the controller gains that attain the smaller position tracking error. Simulation results are given considering an underactuated 6 DOF (degrees of freedom) underwater ROV. Real time experiments on an underactuated mini ROV are conducted to show the effectiveness of the proposed scheme.
A prediction model based on Rough Set and Neural Network is *** the model,we remove some redundant *** attributes that are necessary for rule discovery are kept as the input units of Neural *** input dimensions of Neu...
详细信息
A prediction model based on Rough Set and Neural Network is *** the model,we remove some redundant *** attributes that are necessary for rule discovery are kept as the input units of Neural *** input dimensions of Neural NetWork are *** the meantime,we give some improvement to tradition back-propagation Neural *** last,we present an applied instance.
Information fusion is an important research field, one major theory and technology is neural networks especially back-propagation (BP) neural network. Meanwhile BP neural network has been applied in many fields. But t...
详细信息
Information fusion is an important research field, one major theory and technology is neural networks especially back-propagation (BP) neural network. Meanwhile BP neural network has been applied in many fields. But traditional BP neural network has some faults, such as bad convergence rate and low learning rate aiming at huge date sets, poor generalization, poor ability of error weight update and batch learning. Aiming at these faults, some improved methods are proposed to solve these problems, one of methods is BP neural network based on Kalman Filter which can solve beforementioned faults partly. But present methods of BP neural network based on Kalman Filter can not do batch processing and study multi-sample conditions. Improved BP neural network based on Kalman Filter is proposed depending on present BP neural network based on Kalman Filter. The idea of new method includes two steps, firstly we obtain the update of estimation weight, secondly we use the obtained results to mend the Kalman Gain for new update of time and measurement, at the some time the new algorithm can adopt batch processing to learning neural network. Experiments show the new algorithm can solve high-dimensional, large computation problem, keeping robustness and improving the learning efficiency.
BP feed-forward network is the most widely applied neural network. There are a number of algorithms *** respective strengths and weaknesses of 8 kinds of BP algorithm provided by the neural network toolbox in MATLA...
详细信息
BP feed-forward network is the most widely applied neural network. There are a number of algorithms *** respective strengths and weaknesses of 8 kinds of BP algorithm provided by the neural network toolbox in MATLAB are studied in the paper in order to choose a more appropriate and faster algorithms under different conditions .Based on this, the measurement of vacuum level with the method of magnetron-discharge is taken as an example to carry on the simulation, the convergence steps of a variety of BP algorithm are compared in different situations, the fast convergence property of trainlm is confirmed, the conclusion is obtained that BP algorithm can forecast the vacuum level.
In this paper,the theory of artificial neural network with back-propagation algorithm (BPN) is presented,and the BPN model is used to predict the accumulated temperature for Northeast China,North China,and the Huang-H...
详细信息
In this paper,the theory of artificial neural network with back-propagation algorithm (BPN) is presented,and the BPN model is used to predict the accumulated temperature for Northeast China,North China,and the Huang-Huai-Hai Plain.A total of 235 records collected from 235 meteorology stations were fed into the BPN model for training and *** latitude,longitude and elevation of each station were used as input variables of BPN,and the accumulated temperature as output *** key network parameters, such as learning rate,momentum,the number of hidden nodes,and the learning iterations, were optimized using a trial and error *** optimized BPN model was compared with the multiple linear regression(MLR) *** summary,BPN model was generally more accurate than MLR *** infers that artificial neural network models are more applicable than regression models when predicting accumulated temperature.
Artificial neural networks back-propagation algorithm was applied to the prediction of vibration frequencies of v1 and v2 modes of octahedral hexahalide (MX6(n-)). Three-layer networks with one and two output nodes we...
详细信息
Artificial neural networks back-propagation algorithm was applied to the prediction of vibration frequencies of v1 and v2 modes of octahedral hexahalide (MX6(n-)). Three-layer networks with one and two output nodes were used. Two inertia terms and training step controlling scheme were adopted to weights adjustment. The result of one output node networks has little difference from that of two output nodes networks. The frequencies of [Mof6]2-, [BiF6]- and [AuF6]- from literature are a great deal different from those calculated or predicted values.
In general, we describe three different methods to select an appropriate distribution form:bistogram, probability plots, and hypothesis test. The life distribution is recognized by a neural network method. The relatio...
详细信息
In general, we describe three different methods to select an appropriate distribution form:bistogram, probability plots, and hypothesis test. The life distribution is recognized by a neural network method. The relationship among life distribution with life data is described through threshold and weight of neural networks. The method is convenient to use. An example is presented to validate this method, and the results are satisfactory.
暂无评论