Application Of Extreme Learning Machine In Radar Target Recognition

Radar target recognition plays an important role in both military and civilian fields.The contradiction between classification accuracy and real-time performance of traditional recognition methods restricts its practical applications.Feature extraction is a key step in radar target recognition.The quality of the extracted features determines the performance of target recognition.While the features extracted by the traditional methods are mostly artificially designed and they need to rely on the experience of the researchers.If we do not have sufficient prior knowledge,the extracted features would be incomplete.Although the traditional deep learning models can mine the deep feature representation of the target,their common problem is that the real-time performance is not high.In addition,training deep learning models requires a large number of data samples.In order to solve these problems,this paper studies the application of extreme learning machine in radar target recognition.Extreme learning machine(ELM),as a new learning algorithm for single-hidden-layer feedforward neural networks(SLFNs),has attracted great concerns from various fields for its fast learning speed and good generalization performance.Besides,the training of the deep extreme learning machine can be completed quickly under the condition of small samples.In chapter 1,this paper introduces the background and significance of this dissertation,and illustrates the development of radar target recognition system and methods.We also review the development of neural network,especially the deep learning model,and its application in radar target recognition.The advantages and disadvantages of ELM algorithm in radar target recognition are also presented in this section.In chapter 2,the similarities and differences between ELM algorithm and traditional BP algorithm and SVM algorithm are analyzed theoretically.The performance of the ELM algorithm is verified by regression experiment and classification experiment.However due to the random selection of input weights and hidden biases,ELM tends to need a large number of hidden nodes for better generalization,which may increase the complexity of the network.In chapter 3,a novel evolutionary extreme learning machine(ELM)based on improved quantum-behaved particle swarm optimization(IQPSO)for radar target classification is presented.Quantum-behaved particle swarm optimization(QPSO)has been used in ELM to solve the problem that ELM needs more hidden nodes than conventional tuning-based learning algorithms due to the random set of input weights and hidden biases.But the method for calculating the characteristic length of Delta potential well of QPSO may reduce the global search ability of the algorithm.To solve this issue,a new method to calculate the characteristic length of Delta potential well is proposed in this chapter.The proposed algorithm is evaluated by using radar data,the experimental results indicate that the proposed algorithm is effective in terms of real-time and accuracy.In chapter 4,a novel radar target recognition method based on a stacked autoencoder(SAE)and extreme learning machine(ELM)is presented.As a key component of deep structure,the SAE does not only learn features by making use of data,it also obtains feature expressions at different levels of data.However,with the deep structure,it is hard to achieve good generalization performance with a fast learning speed.In this chapter,SAE and ELM are combined to make full use of their advantages and make up for each of their shortcomings.The effectiveness of the proposed method is demonstrated by experiments with measured radar data.The experimental results show that the proposed method can achieve good performance in the two aspects of real-time and accuracy.In chapter 5,dropout constrained deep extreme learning machine(DCDELM)is proposed to solve the problem that the deep extreme learning machine easily falls into overfitting when the training sample is limited.The experimental results show that the algorithm can effectively solve the problem of overfitting.Furthermore,a deep complex extreme learning machine based on manifold regularization(MR?DCELM)is proposed.The deep complex extreme learning machine(DCELM)extends the deep extreme learning machine(DELM)from the real domain to the complex domain,and can effectively capture the higher-level underlying structure of the complex data.Introducing manifold regularization(MR)in network training can make the network better preserve neighborhood relationships.The effectiveness of the proposed method is demonstrated by experiments with darkroom measured radar data.The experimental results show that the proposed method can achieve good performance.Chapter 6 summarizes the dissertation and discusses the further work.