摘要：为提高架空输电线路工程造价预测精度，建立PSO（粒子群优化算法）-LSSVM（最小二乘支持向量机）-KDE（核密度估计）混合架空输电线路工程造价预测模型。该模型首先通过PCA实现数据降维，PSO优化LSSVM参数构建造价点预测模型，再创新性引入KDE对预测误差进行估算，得到不同置信度下的架空输电线路工程造价区间。结果表明，本文选取的68个架空输电线路施工数据的PSO-LSSVM最优参数C=271.6，σ2=0.53，模型平均相对误差6.547%，该模型预测精度高于其他模型。在85%的置信水平下，造价预测区间覆盖率88.89%，具有较高的可靠性，为架空输电线路工程投资决策提供可靠依据。

关键词：工程造价；粒子群优化算法；最小二乘支持向量机；核密度估计；架空输电线路工程

Abstract：To improve the prediction accuracy of overhead transmission line project cost，this paper constructs a hybrid PSO-LSSVM-KDE prediction model for overhead transmission line project cost. Firstly，PCA is adopted for data dimensionality reduction，and PSO is used to optimize the parameters of LSSVM to establish a point prediction model for project cost. Furthermore，KDE is innovatively introduced to estimate prediction errors，so as to obtain the cost intervals of overhead transmission line projects under different confidence levels. The results show that for the 68 sets of construction data of overhead transmission lines selected in this paper，the optimal parameters of PSO-LSSVM are C=271.6 and σ2 = 0.53，and the average relative error of the model is 6.547%. The proposed model achieves higher prediction accuracy than other models. At the 85% confidence level，the coverage rate of the cost prediction interval reaches 88.89%，indicating favorable reliability，which has relatively high reliability，providing a reliable basis for the investment decision making of overhead transmission line projects.

Keywords：project cost；particle swarm optimization algorithm；least squares support vector machine；kernel density estimation；overhead transmission line project

参考文献

[1] 俞敏，王愿翔，闫园，等.架空线路改造工程造价的组合预测方法[J].电力科学与技术学报，2020（1）：24-30.

[2] 陈典，钟海旺，夏清，等.基于有效输电成本的输电网扩展规划[J].中国电机工程学报，2020（1）：30-38+371.

[3] 朱学文，董士波，徐慧声，等.电力工程定额动态管理研究[J].建筑经济，2022（3）：67-73.

[4] 卢文飞，袁竞峰，张嘉澍，等.基于机器学习算法的输电线路工程投资预测[J].科学技术与工程，2022（17）：6992-7001.

[5] 凌云鹏，阎鹏飞，韩长占，等.基于BP神经网络的输电线路工程造价预测模型[J].中国电力，2012（10）：95-99.

[6] 常鹏，荣高升，妙旭娟，等.架空输电线路工程造价影响因素分析[J].中国电力企业管理，2022（3）：54-56.

[7] Cheng M Y，Tsai H C，Sudjono E. Conceptual cost estimates using evolutionary fuzzy hybrid neural network for projects in construction industry[J].Expert systems with applications，2010（6）：

4224-4231.

[8] Zhongfu Q，Xiaolong L，Dong Z，et al. Forecasting the costs of residential construction based on support vector machine and least squares-support vector machine[J]. Journal of Zhejiang University（Science Edition），2016（43）：357-363.

[9] Guo X，Li D，Zhang A. Improved support vector machine oil price forecast model based on genetic algorithm optimization parameters[J].Aasri procedia，2012，1（Complete）：525-530.

[10] Tongyao L，Tao Y，Chao Z，et al. Intelligent prediction of the construction cost of substation projects using Support Vector Machine optimized by Particle Swarm Optimization[J]. Mathematical problems in engineering，2019（2019）：1-10.

[11] Sun W，Sun J . Daily PM2.5 concentration prediction based on principal component analysis and LSSVM optimized by cuckoo search algorithm[J]. Journal of environmental management，2016：188-144.

[12] Grounds A M，Joslyn S，Otsuka K. Probabilistic interval forecasts：An individual differences approach to understanding forecast communication[J]. Advances in meteorology，2017：1-18.

[13] Amara F，Agbossou K，Dubé Y，et al. Household electricity demand forecasting using adaptive conditional density estimation[J]. Energy buildings，2017（156）：271-280.

建筑经济，2026（9）：78-86