【摘要】：利用粒子群優(yōu)化算法(PSO)較強的魯棒性和全局搜索能力等優(yōu)點,將PSO算法與BP神經(jīng)網(wǎng)絡相結(jié)合,優(yōu)化了BP神經(jīng)網(wǎng)絡分類時的初始權(quán)值和閾值�；谥榻涌谌侵薜膫�(cè)掃聲吶圖像數(shù)據(jù),提取了海底聲吶圖像中砂、礁石、泥3類典型底質(zhì)的6種主要特征向量,利用PSO-BP方法對海底底質(zhì)進行分類識別。實驗表明,3類底質(zhì)分類精度均大于90%,高于BP神經(jīng)網(wǎng)絡70%左右的分類精度,表明PSO-BP方法可有效應用于海底底質(zhì)的分類識別。
[Abstract]:Based on the strong robustness and global searching ability of particle swarm optimization (PSO) algorithm, the initial weights and thresholds of BP neural network classification are optimized by combining PSO algorithm with BP neural network. Based on the side scan sonar image data of the Pearl River estuary delta, six main characteristic vectors of sand, reef and mud are extracted from the seabed sonar image, and the seabed sediment is classified and recognized by PSO-BP method. The experimental results show that the classification accuracy of the three types of sediments is more than 90%, which is higher than that of BP neural network about 70%. It shows that the PSO-BP method can be applied to the classification and recognition of seabed sediment effectively.
【作者單位】： 國家海洋局第二海洋研究所;國家海洋局海底科學重點實驗室;浙江大學地球科學學院;
【基金】：國家自然科學基金(41476049) 科技基礎性工作專項(2013FY112900) 海洋公益項目(201105001)
【分類號】：P733.2
，

本文編號：2194559