摘要
针对舰船装备维修费影响因素复杂、预测结果不稳定等问题,构建了AdaBoost-PSO-BP的舰船维修费用预测模型。引入机器学习中的方法,针对BP神经网络在较小样本规模的情况下精度不高的问题,采用粒子群算法和早期停止法,确定了最佳参数,消除了过拟合现象,并通过AdaBoost将BP神经网络优化集成,提升了模型的准确度和稳定性。实例验证了该模型的实用性、科学性和有效性。
Aiming at the problems of complex influencing factors and unstable prediction results of ship equipment maintenance cost,an AdaBoost-PSO-BP model for ship maintenance cost prediction was constructed.Methods in machine learning were introduced for the situation that the accuracy of BP neural network is not high in the case of smaller sample size,the particle swarm algorithm and early stop method were used to determine its optimal parameters and eliminate the phenomenon of over-fitting,and the BP neural network was optimally integrated through AdaBoost to improve the accuracy and stability of the model.The scientificity and effectiveness of the model were verified through examples.
作者
蒋国萍
杨洋
訾书宇
高坤
JIANG Guoping;YANG Yang;ZI Shuyu;GAO Kun(Dept.of Management Engineering and Equipment Economics,Naval Univ.of Engineering,Wuhan 430033,China)
出处
《海军工程大学学报》
CAS
北大核心
2023年第1期81-86,共6页
Journal of Naval University of Engineering
