基于PSO-BP神经网络的Savonius型叶轮阵列消波性能优化
Optimization of Wave Absorption Performance of Savonius Type Impeller Array Based on PSO-BP Neural Network
盛勇 1宋瑞银 2杨状状 1刘博宇 3吴瑞明 4任聪杰2
作者信息
- 1. 浙江科技大学机械与能源工程学院,杭州 310023;浙大宁波理工学院机电与能源工程学院,浙江宁波 315100;智慧海洋牧场装备技术浙江省工程研究中心,浙江宁波 315100
- 2. 浙大宁波理工学院机电与能源工程学院,浙江宁波 315100;智慧海洋牧场装备技术浙江省工程研究中心,浙江宁波 315100
- 3. 浙大宁波理工学院机电与能源工程学院,浙江宁波 315100;浙江大学机械工程学院,杭州 310023;智慧海洋牧场装备技术浙江省工程研究中心,浙江宁波 315100
- 4. 浙江科技大学机械与能源工程学院,杭州 310023
- 折叠
摘要
为了提高Savonius型(S型)叶轮的消波性能,提出一种S型叶轮阵列装置.通过试验记录不同的叶轮间距和叶轮相对入水深度等5个参数下波浪经过叶轮阵列后的透射系数Kt,建立基于粒子群优化(PSO)算法和反向传播(BP)神经网络的S型叶轮阵列消波性能预测模型.将采用该模型与采用BP网络模型和GA-BP网络模型得到的平均绝对误差、均方根误差和决定系数R2指标进行对比,结果表明,采用PSO-BP神经网络模型优化能得到误差更小、更精准的预测结果.当相邻叶轮间距分别为0.62m和0.41m、各叶轮入水深度分别为0.15 m、0.18m和0.19m时,S型叶轮阵列具有相对最佳的消波性能.
Abstract
In order to improve the wave absorption performance of the Savonius-type(S-type)impeller,a S-type impeller array device is proposed.By conducting experiments and recording the transmission coefficient Kt of waves passing through the impeller array under five parameters,including different impeller spacing and relative submergence depth,a prediction model for the wave absorption performance of the S-type impeller array is established based on particle swarm optimization(PSO)algorithm and back-propagation(BP)neural network.The model is compared with the BP neural network model and the GA-BP neural network model using the average absolute error,root mean square error,and determination coefficient R2 as indicators.The results show that the PSO-BP neural network model optimization can achieve smaller and more accurate prediction errors.When the adjacent impeller spacing is 0.62 m and 0.41 m,and the submergence depths of each impeller are 0.15 m,0.18 m,and 0.19 m respectively,the S-type impeller array exhibits relatively optimal wave absorption performance.
关键词
Savonius型叶轮/消波性能/粒子群优化(PSO)算法/反向传播(BP)神经网络Key words
Savonius-type impeller/wave absorption performance/particle swarm optimization(PSO)/back propagation(BP)neural network引用本文复制引用
基金项目
浙江省"领雁"重大专项计划(2023C03122)
宁波市公益性科技计划项目(2022S095)
出版年
2024
NETL
NSTL
万方数据