摘要
地铁轨道结构变形是影响地铁安全运营的重要因素,尤其是在沉降变形方面,因此针对地铁轨道沉降的变形监测,同时根据监测结果对轨道的沉降变形趋势进行准确判断具有重要意义。本文以某市地铁2号线轨道监测数据为例,发挥小波分析与极限学习机(Extreme Learning Machine, ELM)模型在数据处理、数据预测中的优势,将粒子群优化(Particle Swarm Optimization, PSO)算法用于ELM模型参数优化中,构建基于小波去噪的PSO-ELM组合预测模型,实现地铁轨道的沉降变形预测研究。通过小波分析进行监测数据去噪,解决了监测数据不稳定带来的预测结果的干扰问题;通过构建PSO-ELM组合预测模型,解决了模型参数选取随机性带来的预测精度受限问题。将本文提出的小波去噪PSO-ELM模型与单一ELM模型、小波去噪ELM模型沉降预测结果进行对比分析,结果表明本文提出组合预测模型预测精度最高,同时预测误差不会随预测期数的增加产生明显变化,具有较高的稳健性与适应性。
The deformation of subway track structure is an important factor affecting the safe operation of subway, especially in terms of settlement deformation. Therefore, it is of great significance to monitor the deformation of subway track settlement and accurately judge the trend of track settlement deformation based on the monitoring results. This article takes the monitoring data of the subway line 2 in a certain city as an example, and leverages the advantages of wavelet analysis and Extreme Learning Machine (ELM) models in data processing and prediction. The Particle Swarm Optimization (PSO) algorithm is applied to optimize the ELM model parameters, and a PSO-ELM combined prediction model based on wavelet denoising is constructed to achieve the prediction of subway track settlement deformation. By using wavelet analysis to denoise monitoring data, the interference problem of prediction results caused by unstable monitoring data has been solved;By constructing a PSO-ELM combined prediction model, the problem of limited prediction accuracy caused by the randomness of model parameter selection has been solved. The proposed wavelet denoising PSO-ELM model was compared and analyzed with the settlement prediction results of a single ELM model and wavelet denoising ELM model. The results showed that the proposed combined prediction model had the highest prediction accuracy, and the prediction error did not change significantly with the increase of prediction periods, demonstrating high robustness and adaptability.
出处
《测绘科学技术》
2024年第3期231-239,共9页
Geomatics Science and Technology
