Research And Application For Intelligent Fusion Algorithm Based On Artificial Immune System

With the development of information age, multi-source information fusion technology has been deeply applied to many areas. Artificial intelligence is an important mean to develop information fusion, and artificial immune system algorithm has become a new research focus in artificial intelligence since neural network and genetic algorithm. The thesis mainly studies intelligent fusion algorithm based on artificial immune system.Considering that information fusion still lacks guiding by systematic basic theories, for discussing the generally applicable fusion problem model sake, the thesis puts forward multi-objective fusion theory and its unified model description by referencing multi-objective optimization theory and method. The fusion of information can be converted into the fusion of several optimization results by estimating proper optimization criterion. Based on the multi-objective fusion theory, the thesis can be divided into two parts of the theoretical methods and application researches.The theoretical methods include:1) Artificial immune system and ICSA research. By analyzing the idea of ICSA and clone operator the flow of ICSA is implemented and the algorithm convergence is proved. The theoretical analysis and simulation results show that the immune algorithm system has more advantages than the evolution algorithm system.2) The RBF network design based on immune system. In order to improve the traditional RBF learning strategy, a three-level RBF network learning algorithm based on immune system is proposed, which can calculates the number of the hidden-layer neurons in the first level as immune vaccine, the network can be established and adjusted by itself, and the complexity of search space in the second level can be reduced. The global optimum hidden-layer nonlinear parameters are searched for in the second level by parallel searching with artificial immune algorithm. The output-layer linear parameters are estimated in the third level with least square method, which makes the design dimension of the second level decreased and the algorithm efficiency improved. The application and the experiment of Hermit polynomial approximation both show that the performance of the RBF network trained by the algorithm is superior.The application researches include:1) Research on fusion model for locomotive secondary spring load adjustment. A two-level fusion algorithm for locomotive secondary spring load adjustment based on IDCMA is presented with multi-objective fusion theory. The priori knowledge of the spring load adjustment problem is designed as immune dominance in the algorithm model, and the model is built as two-level structure with different goal and feasible set according to the preference of the criterion. Real locomotive test data shows that the algorithm has superior performance over the others.2) Research on error compensation for track scale based on multi-sensor information fusion. By analyzing factors that affect the weighing error of track scale, an error model is established and an error compensation algorithm for track scale based on immune RBF network is proposed. The RBF network is designed and trained to achieve the error compensation successfully with the above immune three-level algorithm. Experiments show that the immune RBF network has superiority in the error compensation over the weighted fusion algorithm. The above applications both have significant engineering value.