Research On Hybrid Intelligent System And ITS Application Based On Artificial Neural Network |
Posted on:2010-07-24 | Degree:Doctor | Type:Dissertation |
Country:China | Candidate:W Xuan | Full Text:PDF |
GTID:1118360302987638 | Subject:Communication and Information System |
Abstract/Summary: | PDF Full Text Request |
The core content and research method of this paper can be summarized as follows:analyzing the non-stationary characteristics of ship self noise to explore solutions of disadvantages of ANC system; designing and improving adaptive noise cancellation system based on artificial neural networks; importing fuzzy logic reasoning thought to design more intelligent network structure; further improving the network's approximation capacity and adaptive performance using the idea of multi-resolution analysis; importing genetic evolution theory combined with strong local structure algorithm to construct adaptive mixed-learning strategies; verifying the conclusions of the paper by Simulation and the real sea trial data-processing systemically.The target signal's acoustic field environment decided the non-linear features of ANC system. Artificial neural network adaptive processor was designed based on traditional non-linear filter against the non-stationary signals. Its strong functions similar to the human brain were used to solve the inherent flaws of conventional filtering mode. Neural network transformed the problem of non-linear mapping into the problem of optimization solving in system's adaptive learning. The optimization process can be realized by simple learning algorithm relying on advantages such as its self-learning and parallel processing. The introduction of RBF neural network enhanced system's capacity of solving complex and highly non-linear problem greatly. The model designed in this paper gave a good foundation for solution of non-linear system problem. It also had middle computing complexity and high application value.T-S fuzzy reasoning model was used to construct sub-linear model. It was used as the network infrastructure and divided the input space into a number of fuzzy subspaces. Through fuzzy systems and artificial neural network equivalent features, fuzzy reasoning model performed as an adaptive neural network. This made network's all the nodes and the corresponding parameters have practical significance. At the same time, every fuzzy rules was equal to a wavelet sub-network by combining traditional T-S and MRA and putting WNN into after-pieces of fuzzy reasoning. Then new adaptive hybrid genetic algorithm was imported for network and parameters amendments to improve the operating efficiency of the system and solve quality. Data processing results showed that the modified HIS make the system evolution easier to understand and effectively improve the system performance. |
Keywords/Search Tags: | adaptive noise canceling (ANC), nonlinear, artificial neural network (ANN), fuzzy logic inference (FLI), wavelet neural network (WNN), genetic algorithm (GA) |
PDF Full Text Request |
Related items |