CAJ | 학술논문

在相干光学目标识别技术研究中,为了更好地判读相关峰,把遗传算法和人工神经网络相结合,以反向传播神经网络和遗传算法为基础,建立了以遗传算法优化神经网络初始权值和阈值的相关峰判读系统,既避免了神经网络训练中容易陷于局部最小值和收敛速度较慢的缺点,又克服了遗传算法局部精确搜索能力的不足,实现其优势互补,从而有利于更好地解决相关峰判读问题.结果表明,改进后的算法充分发挥遗传算法和反向传播算法的优点,达到了较好的判读效果.
재상간광학목표식별기술연구중,위료경호지판독상관봉,파유전산법화인공신경망락상결합,이반향전파신경망락화유전산법위기출,건립료이유전산법우화신경망락초시권치화역치적상관봉판독계통,기피면료신경망락훈련중용역함우국부최소치화수렴속도교만적결점,우극복료유전산법국부정학수색능력적불족,실현기우세호보,종이유리우경호지해결상관봉판독문제.결과표명,개진후적산법충분발휘유전산법화반향전파산법적우점,체도료교호적판독효과.
In order to identify correlation peak better in the research of target recognition technology based on coherent optics,combining the genetic algorithm (GA) and the artificial neural network(ANN),based on GA and back propagation(BP)neural network,a correlation peak identification system was built with GA optimizing the initial weights and thresholds of the ANN.The optimized identification system could not only avoid the tendency of local minimum and slow convergence speed in ANN training,but also overcome the shortage of local precise searching capacity in GA.It realizes the superiority complementation between the both the methods,and is helpful to solve the problem of recognizing correlation peak.The testing results show that the improved method makes full use of the advantages of GA and BP algorithm,and gets much better interpretation effect.