摘要
根据空中目标威胁估计的特点,分析了基于BP神经网络的空中目标威胁估计方法的不足。运用蚁群优化算法(ACO)的全局寻优能力,对BP神经网络的初始权值进行优化,建立了改进的BP(ACOBP)空中目标威胁估计方法,解决了BP神经网络初始权值的随机性和网络易陷入局部极小值的问题,提高了算法的收敛速度。并采用30组训练样本数据及8组测试数据,对算法的性能进行了仿真分析。仿真结果表明,该算法估计结果准确合理,收敛速度和收敛精度均优于BP算法,证明了该方法的有效性。
On the basis of the characteristics of aerial targets threat assessment,the weaknesses of BP neural network for aerial targets threat assessment are analyzed. By using the ant colony optimization（ ACO） algorithm seeking global excellent result to optimize the random of BP algorithm,a new aerial targets threat assessment method is established and the ACOBP algorithm is achieved by the method,which overcomes the randomness of BP network initial weights,solves the problem lost in local minimum,and improves convergence speed of the network. Finally,the performance of the algorithm is analyzed. Simulation results show the ACOBP algorithm can estimate threat degree accurately and appropriately with faster convergence and better performance than the BP algorithm,proving that the ACOBP algorithm is an effective approach to threat assessment.
出处
《电子科技》
2016年第7期33-35,39,共4页
Electronic Science and Technology
关键词
威胁估计
BP神经网络
全局优化
threat assessment
BP neural network
global optimization
