Target Identification With Uwb Based On S Transform And Improved Artificial Bee Colony Algorithm

Ultra-wideband (UWB) is a personal area network communication technology with low power consumption and high-speed transmission, the applications are very common. Generally used in personal wireless networks,home networks, and short-range radar field. At this stage, UWB radar has the ability to detect targets and analyze targets. However, ultra-wideband radar in real applications is to analyze the radar echo signal, cannot complete the communication requirements. With the development of modern UWB detection and identification technology, new requirements have been put forward for the fusion of UWB communication and target detection. At the same time, in the UWB radar recognition process, due to the influence of the data feature dimension and the classifier parameters, the recognition process is slow and the recognition accuracy is low.In view of the above problems and subject requirements, this paper proposes a UWB target detection and recognition technology combined with S-transform and improved artificial bee colony algorithm. Through the above methods, we can simultaneously achieve the goal of target detection and ultra-wideband communication. Firstly, this paper analyzes the practical significance of UWB communication combined with target detection.Secondly, according to the actual requirements, this paper introduces the time-frequency analysis and optimization algorithms, and analyze the feasibility of each algorithm. Using the time-frequency analysis to transform the signal to frequency domain. And we obtain the time-frequency matrix of the signal as the input matrix for recognition. The study shows that the recognition task can be completed by using the time-frequency matrix as input directly, but it will lead to longer classification time and lower the efficiency of the whole system. In this paper, we analyze the different commonly used dimensionality reduction algorithms to reduce the time-frequency transformed matrix, and compare the classification time and accuracy. When using support vector machine (SVM) to classify the targets,the parameters of the kernel function of SVM still have an effect on the recognition results. In this paper, we use the improved artificial bee colony algorithm (ABC) to optimize the SVM. The basic ABC algorithm adopts the roulette method to obtain the optimization strategy in the onlooker bee phase.This will make the diversity of the colony be destroyed, so that the bee colony algorithm becomes premature convergence. By analyzing the selection strategy of the algorithm, we propose the probability selection curve based on the quadratic function, and keep the diversity of the evolution process.Experimental results show that the improved bee colony algorithm can obtain better target recognition effect. The new target detection and identification method proposed in this paper has certain practical value and research significance.