利用组合核函数建立VSV系统回归预测模型 |
点此下载全文 |
引用本文:夏存江,王文博.利用组合核函数建立VSV系统回归预测模型[J].计算技术与自动化,2021,(2):106-109 |
摘要点击次数: 373 |
全文下载次数: 0 |
|
|
中文摘要:利用支持向量回归机(SVR)建立了飞机巡航阶段发动机可调静子叶片系统(VSV)的回归预测模型。在利用SVR进行建模时,核函数的选用尤为关键,核函数有局部核函数和全局核函数,利用单一核函数训练模型易出现过拟合或欠拟合问题。为解决核函数的选用难题,避免训练过程中出现模型过拟合或欠拟合问题,提出了组合核函数。通过对单一核函数的组合,组合核函数兼具全局核函数和局部核函数的优点。最后,利用粒子群算法(PSO)对模型进行参数寻优优化,结果表明:相较于单一核函数,采用组合核函数的模型训练时间更短,模型精度更高。 |
中文关键词:SVR 过拟合 欠拟合 组合核函数 PSO 回归预测模型 |
|
Regression Prediction Model Of VSV System Established By Using Combination Kernel Function |
|
|
Abstract:A support vector regression machine (SVR) was used to establish a regression prediction model for the engine adjustable stator blade system (VSV) during the cruise phase. When using SVR for modeling, the selection of kernel functions is especially critical. Kernel functions include local kernel functions and global kernel functions. Using a single kernel function to train a model is prone to overfitting or underfitting. In order to solve the problem of choosing the kernel function and avoid the problem of over-fitting or under-fitting of the model during training, a combined kernel function is proposed. By combining single kernel functions, the combined kernel functions have the advantages of both global kernel functions and local kernel functions. Finally, the particle swarm optimization (PSO) is used to optimize the parameters of the model. The results show that compared with the single kernel function, the training time of the model using the combined kernel function is shorter and the model accuracy is higher. |
keywords:SVR overfitting underfitting combinatorial kernel function PSO regression prediction model |
查看全文 查看/发表评论 下载pdf阅读器 |