| 基于DPSO-ANFIS的大坝变形预测模型 |
| |
| 引用本文： | 马广臣,杨杰,程琳,宋锦焘.基于DPSO-ANFIS的大坝变形预测模型[J].水利水运工程学报,2021,0(6):116-123. |
| |
| 作者姓名： | 马广臣 杨杰 程琳 宋锦焘 |
| |
| 作者单位： | 1.西安理工大学 水利水电学院，陕西 西安 710048 |
| |
| 基金项目： | 国家自然科学基金资助项目（51809212） |
| |
| 摘 要： | 变形预测模型是大坝结构安全性态分析的关键技术支撑。针对现有大坝变形预测模型在精确度、泛化性等方面的不足，将自适应模糊神经网络引入到大坝变形预测模型中，利用动态权重粒子群算法对自适应模糊神经网络中模糊层的适应度值进行参数寻优，形成可以寻找最优适应度值的自适应模糊神经网络，进而建立基于DPSO-ANFIS的大坝变形预测模型。根据大坝原型监测数据，代入训练好的模型得到输出值，并将其与实际监测数据进行对比分析。工程实例应用表明，基于DPSO-ANFIS的大坝变形预测模型输出值与实测值偏差最大为0.0516 mm，均方根误差为0.0351 mm，平均绝对误差为0.0320 mm，各项指标精度均优于基于PSO-ANFIS、ANFIS和BP神经网络的大坝变形预测模型。针对不同位置测点、预测时间段，基于DPSO-ANFIS的大坝变形预测模型输出值接近实测值，预测趋势符合真实值走向，整体预测性能稳定。该模型具有较高的精确度、良好的泛化性与可靠的稳定性，工程实用综合性能较优。
|
| 关 键 词： | 自适应模糊神经网络 动态权重粒子群算法 大坝变形预测 适应度 |
| 收稿时间： | 2020-12-13 |
The dam deformation monitoring model based on DPSO-ANFIS |
| |
| Affiliation: | 1.Institute of Water Resources and Hydropower, Xi’an University of Technology, Xi’an 710048, China2.State Key Laboratory Base of Eco-Hydraulic Engineering in Arid Area, Xi’an 710048, China |
| |
| Abstract: | Deformation prediction model provides an essential support for the analysis of dam structure safety. However, the existing dam deformation prediction models suffer from low precision and insufficient generalization. To solve this problem, an adaptive fuzzy neural network was introduced into the dam deformation prediction model in this study. Specifically, the particle swarm optimization (PSO) algorithm with dynamic weights was applied to optimize the parameter of fitness values in the fuzzy layer of the adaptive fuzzy neural network. Based on this, an adaptive fuzzy neural network called DPSO-ANFIS was formed, to search for the optimal fitness value. And a dam deformation prediction model based on DPSO-ANFIS was established. Experiments were conducted to verify the effectiveness of the proposed deformation prediction model. The monitoring data from the dam prototype was applied to the trained model to obtain the output values, and these values were then compared with the actual monitoring data. The engineering practice showed that the maximum deviation between the output values of the dam deformation prediction model based on DPSO-ANFIS and the measured values was 0.051 6 mm; the RMSE (root mean square error) was 0.035 1, and the average absolute error was 0.032 0. All the values of these performance indicators were better than those of the dam deformation prediction models based on PSO-ANFIS, ANFIS, and BP neural network. As for the measuring points and time, the predicted values from the dam deformation prediction model based on DPSO-ANFIS exhibited a trend much close to that of measured values. Also, the proposed deformation prediction model achieved a stable overall prediction performance. Therefore, the dam deformation prediction model based on DPSO-ANFIS achieves high accuracy, good generalization, strong stability, and excellent comprehensive performance in practical engineering applications. |
| |
| Keywords: | |
|
| 点击此处可从《水利水运工程学报》浏览原始摘要信息 |
|
点击此处可从《水利水运工程学报》下载全文 |
|
