The principal constituents of soft computing are the fuzzy logic (FL), artificial neuralnetworks (NN) and probabilistic reasoning (PR). It is generally regarded that FL primarily deals with imprecision, NN with learn...
详细信息
The principal constituents of soft computing are the fuzzy logic (FL), artificial neuralnetworks (NN) and probabilistic reasoning (PR). It is generally regarded that FL primarily deals with imprecision, NN with learning, and PR with uncertainty. They have, however, overlapping boundaries and are known to be complementary rather than competitive to each other in many applications. Here, two control algorithms, one implemented by fuzzy logic and the other by a neural network, are used as the basis to highlight salient features of soft computing. A DC motor servo system with the proposed soft computing based algorithms is discussed. The fuzzy logic control employs the principles of fuzzy logic to calculate an optimal output action based on input conditions, and a knowledge base expressed in linguistic forms, thereby performing a parallel operation to control the output with a high degree of robustness against parameter change. In the neural network control, focus is on how neuralnetworks can overcome deadzone-plus-saturation nonlinearity commonly found in the power driver of a DC servo motor. Simulation results have been performed to establish the validity of these control algorithms.< >
neural concepts and methods for process control and their use in the paper industry are discussed and illustrated. A basic problem in the paper industry is to produce pulp of optimal quality. A neural net has been use...
详细信息
neural concepts and methods for process control and their use in the paper industry are discussed and illustrated. A basic problem in the paper industry is to produce pulp of optimal quality. A neural net has been used to predict the optimal time for the cooking process. The neural net prediction model was found to give more accurate predictions than the originally used analytic model.
The paper outlines a general scheme for implementing neurofuzzy algorithms and emphasizes their desirable features for on-line and off-line learning. Examples will be used to show how these techniques can be used for ...
详细信息
The paper outlines a general scheme for implementing neurofuzzy algorithms and emphasizes their desirable features for on-line and off-line learning. Examples will be used to show how these techniques can be used for nonlinear modelling, reinforcement learning and on-line control.
The modeling abilities of radial basis function networks were tested on a real plant. A training algorithm which allows an on-line adaptation of the neural plant model was developed. The results obtained during contro...
详细信息
The modeling abilities of radial basis function networks were tested on a real plant. A training algorithm which allows an on-line adaptation of the neural plant model was developed. The results obtained during control experiments were satisfactory in spite of the inverse model inaccuracy.
The paper describes the application of the Kohonen feature map (KFM) for analyzing and splitting extensive load databases. The objective is to get separate clusters of load shapes for making short term load forecast (...
详细信息
The paper describes the application of the Kohonen feature map (KFM) for analyzing and splitting extensive load databases. The objective is to get separate clusters of load shapes for making short term load forecast (STLF) models with high accuracy. The forecast models are based on feedforward neuralnetworks.
Hypersonic aircraft require a high degree of system integration. Design tools are needed to provide rapid, accurate calculations of complex fluid flows. The paper discusses the application of neuralnetworks to the ca...
详细信息
Hypersonic aircraft require a high degree of system integration. Design tools are needed to provide rapid, accurate calculations of complex fluid flows. The paper discusses the application of neuralnetworks to the calculation of fluid flow and heat transfer in a heat exchanger panel for the National AeroSpace Plane (NASP).
The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-bas...
详细信息
The training speeds of Batch Backpropagation using steepest descent, Conjugate Gradient and Quasi-Newton algorithms for a feedforward neural network are compared. Results illustrating the advantages of the Hessian-based techniques are given and issues affecting speed discussed.
The Local Model network extension to Basis Function networks is especially important for practical application of learning systems to real modelling problems. This is because of the architecture's representational...
详细信息
The Local Model network extension to Basis Function networks is especially important for practical application of learning systems to real modelling problems. This is because of the architecture's representational ability, as well as for its transparency and its compatibility with conventional modelling methods - the model is often already in a form which engineers can understand by using conventional analysis to examine individual local models.
The TACOMA (TAsk decomposition, COrrelation Measures and local Attention neurons) learning architecture is proposed for growing complex artificial neuralnetworks. The algorithm generates a feed forward network bottom...
详细信息
The TACOMA (TAsk decomposition, COrrelation Measures and local Attention neurons) learning architecture is proposed for growing complex artificial neuralnetworks. The algorithm generates a feed forward network bottom up by cyclically inserting cascaded hidden layers. Inputs of a hidden layer unit are locally restricted with respect to the input space by using a new kind of activation function, combining the local characteristics of radial basis function units with sigmoid units.
暂无评论