Abstract: Based on the principles of Relevant Vector Machine (RVM), appropriate type of kernel function is selected, Quantum Particle Swarm Optimization (QPSO) algorithm is used to optimize the kernel parameters, and the QPSO-RVM model is established. The QPSO-RVM model is applied to predict dam deformation monitoring data, and comparison between the prediction results and the results from multivariate statistical regression analysis is carried out. The result shows that the fitting and prediction accuracy of QPSO-RVM model is obviously better than that of multivariate statistical regression model, which has practical application value.

Key words: RVM, kernel parameter, QPSO, dam deformation, prediction

摘要： 根据相关向量机（RVM）原理，选择合适的核函数类型，运用量子粒子群（QPSO）算法对核参数进行优化运算，建立QPSO-RVM模型。运用QPSO-RVM模型对大坝变形监测数据进行预测，并将预测成果与多元统计回归分析成果进行对比研究。结果表明，QPSO-RVM模型的拟合及预测精度明显优于多元统计回归模型，具有工程运用价值。

关键词: 相关向量机, 核参数, 量子粒子群, 大坝变形, 预测