改进骆驼算法辅助随机森林模型预测TBM油温

PREDICTION OF TBM OIL TEMPERATURE BY IMPROVED CAMEL ALGORITHM ASSISTED RANDOM FOREST MODEL

为获取对盾构机(Tunnel Boring Machine,TBM)油温的分类预测性能,基于自然天气现象,提出一种骆驼行走阻力与行走耐力策略改进骆驼算法优化随机森林的预测模型。首先,采用提出的策略对传统骆驼算法进行改进,结果表明,改进后的骆驼算法具有良好的收敛速度和收敛精度;其次,利用改进骆驼算法对随机森林建立的盾构机油温预测模型进行参数优化,获得最优模型;最后,在此基础上,对测试数据集进行分类预测研究分析。实验结果表明,提出的模型预测准确率达到97.71%,相比于传统随机森林模型在准确率上提升了6.38%,可以达到避免油温过高引起盾构机故障的目的,为未来的整机材料-结构-控制多学科协同优化设计和性能预测提供基础。

In order to obtain the classification prediction performance of TBM oil temperature,a camel walking resistance and walking endurance strategy based on natural weather phenomena was proposed to improve the camel algorithm to optimize the prediction model of random forest.Firstly,the traditional camel algorithm was improved by using the proposed strategies.The results showed that the improved camel algorithm has good convergence speed and convergence accuracy.Secondly,the improved camel algorithm was used to optimize the parameters of the TBM oil temperature prediction model established by the random forest to obtain the best model.Finally,on this basis,the classification prediction research and analywis of the test data set was carried out.The experimental results show that the prediction accuracy of the proposed model reaches 97.71%,compared with the traditional random forest model,the accuracy is mproved 6.38%,which can achieve the purpose of avoiding the failure of the shield machine caused by high oil temperature.It provides the basis for the whoce machine material-structure-control future multidisciplinary cooperative optimization design and performance prediction.