基于IPSO-SVM对尾矿坝变形预测的研究

Prediction of Tailings Dam Deformation Based on Improved Particle Swarm Optimization and Support Vector Machine

针对支持向量机在参数模型选择上的敏感性,以及在理论上无法直接实现的问题,在标准粒子群算法的基础上对粒子速度与位置更新策略进行改进,通过改进的粒子群算法对支持向量机模型参数进行选择优化,进而提出了一种改进粒子群优化支持向量机(IPSO-SVM)算法模型。根据尾矿坝实测数据,建立了基于IPSO-SVM算法的对尾矿坝坝体位移预测模型,同时与经典的SVM算法以及PSO-SVM算法进行比较分析。结果表明,3种算法在坝体变形预测中都具有较好的可行性,但IPSO-SVM算法在训练效率上有较大优势,而且具有较高的预测精度,更适合在变形预测中应用。

In view of the sensitivity of support vector machine to parameter model selection,and also that it can not be realized directly in theory,the particle velocity and position updating strategy is improved on the basis of standard particle swarm optimization,the model parameters of support vector machine are selected and optimized by improved particle swarm optimization and an improved particle swarm optimization support vector machine(ISPO-SVM)algorithm model is proposed.According to the measured data of the tailings dam,a prediction model of the displacement of the tailings dam based on the IPSO-SVM algorithm is established,and it is compared with the classical SVM algorithm and the PSO-SVM algorithm.The results show that the three algorithms all have good feasibility in the dam deformation prediction,but the IPSO-SVM algorithm has a great advantage in the training efficiency,and higher prediction accuracy,and so it is more suitable for the application of deformation prediction.