Marine Power Generator Based On Rbf Neural Network Modeling

Synchronous generator is the heart of the whole power system, The modeling and analyze of the synchronous generator is an challenging research, and it is difficult to set up an accurate mathematic model, and the simplified mathematic model is no simple at all. There are abilities of learning and batch operation with artificial neural networks (ANN). It is reality to use ANN to build a fast running speed model of marine diesel generator for real time marine power simulator. ANN has aroused more attention in dynamic modeling. There is ability to approximation nonlinear model with radial basis function neural networks (RBF NN). RBF NN is a forward-propagation network of excellent property. There is no BP NN part optimization. Furthermore, its learning algorithm is simple, its learning efficiency high.This research is supported by the project (Marine large power generator modeling and analysis based on chaotic neural network). And this project is supported by fund of science and technology development of Shanghai municipal education committee (03IK06). The dissertation launched a probe into modeling of marine diesel engine generator by RBF NN. The result of my research shows that RBF NN is a new tool to build the model of synchronous generator. The dissertation still puts forward the way of realize NN modeling with general DSP chips. The data of NN learning and testing is gathered by running the marine power simulation system of ship YUNHE. The dissertation includes six chapters:Chapter 1 Briefly introduced simulation system of marine diesel generatorand modeling by ANN.Chapter 2 Briefly introduced the knowledge of ANN. Chapter 3 Introduced simulation marine power system of ship YUNHE and modified it in part. The data of NN learning and testing is gathered by running the simulation system.Chapter 4 Build the model of simulation YUNHE diesel engine generator by RBF NN under the development environment of MATLAB, and compared two ways of modeling.Chapter 5 Introduced a way of realize ANN modeling by general DSP chips, and analyzed the time that ANN modeling spared underthe environment of TMS320LF2407 DSP.Chapter 6 Summarize this dissertation. It also put forward some problems existing at present and the trends of development in this field.