摘要
为了对城区地表沉降量进行准确预测,提出基于蜣螂优化算法(DBO)改进深度置信网络(DBN)的城区地表沉降量预测模型。首先,针对深度置信网络存在参数调整困难的问题,引入蜣螂优化算法对参数进行优化设置;其次,以小基线集干涉测量技术(SBAS-InSAR)获取淮南特征区的地表沉降量作为原始序列进行预测计算,采用k折交叉验证以避免过拟合风险,并对比分析了反向传播神经网络、DBN和DBO-DBN模型预测结果。结果表明:(1)DBO-DBN模型预测的准确率为96.30%,均方根误差为0.840 mm,R^(2)值为0.9926,相较于BP神经网络和DBN模型,改进后的DBO-DBN预测精度提高,地表沉降预测值与真实值绝对误差趋势表现最好。(2)P1、P2两个特征点未来12个月的沉降量进行预测,结果显示P1点未来沉降量在一定范围内波动,P2点未来沉降量基本稳定,二者均未呈现明显的沉降趋势。
In order to accurately predict the subsidence of urban areas,a deep belief network(DBN)prediction model for urban subsidence based on the optimization algorithm of dung beetle is proposed.Firstly,to address the difficulty of parameter adjustment in deep belief networks,the Dung Beetle Optimizer(DBO)algorithm is introduced to optimize the parameter settings;Secondly,the Small Baseline Subset InSAR(SBAS InSAR)technique was used to obtain the surface subsidence of the Huainan feature area as the original sequence for prediction and calculation.K-fold cross validation was employed to avoid overfitting risks,and the prediction results of backpropagation neural network(BP),DBN,and DBO-DBN models were compared and analyzed.The results showed that:(1)the accuracy of the DBO-DBN model prediction was 96.30%,with a root mean square error of 0.840 mm and a value of 0.9926.Compared with the BP neural network and DBN model,the improved DBO-DBN model improved the prediction accuracy,and the absolute error trend between the predicted surface subsidence value and the true value was the best.(2)the prediction of the settlement amount of the two feature points P1 and P2 in the next 12 months shows that the future settlement amount of point P1 fluctuates within a certain range,while the future settlement amount of point P2 is basically stable,and neither of them shows a significant settlement trend.
作者
刘明众
王子祎
LIU Mingzhong;WANG Ziyi(School of Geology and Construction Engineering,Anhui Technical College of Industry and Economy,Hefei 230051,China;School of Civil and Hydraulic Engineering,Hefei University of Technology,Hefei 230009,China)
出处
《齐鲁工业大学学报》
2025年第5期61-69,共9页
Journal of Qilu University of Technology
基金
国家自然科学基金(42004001)
安徽省自然科学基金(2308085MD125)
安徽省高校自然科学研究项目(2024AH050179,2023AH052675)
省级质量工程虚拟仿真实训基地项目(2023fzsx023)。
