王如会

（ 山东东方监理咨询有限公司 ， 山东 济宁 272000)

摘   要 ：针对建筑工程造价样本中特征量较多、难以提取导致的数值预测精准度不高的问题，提出基于改  进灰色Markov 模型的预测研究，以实现有效解决 。首先采用线性回归分析法获得建筑造价样本中人工费 、机械  使用费 、材料费以及总造价等子明细特征的先验数据 ，并建立数据矩阵 。然后通过前后相邻数值求得概率均值， 并初值化处理均值数据得到对应子明细特征的先验模型 。基于灰色Markov 模型一阶代表一个变量的特点 ，建立微分预测模型 ，计算每次数据更新生成的特征值变化，并与初始值对比，生成特征向量变化序列 。最后采用最小二乘积算法求得数据状态转移概率，结合待预测点的向量和转移概率，通过查找对比得出有效工程造价预测。实验数据证明，所提方法针对建筑工程造价的特征量分布情况预测精准度高 ，造价预测值与真实值间的差距较小，预测性能表现优异。

关键词 ：改进灰色Markov 模型 ；建筑工程造价 ；最小二乘积算法 ；数据状态转移概率 ；特征向量

中图分类号 ：TU789           

文献标志码 ：A            

文章编号 ：1005- 8249  (2023)   06- 0122- 06

DOI : 10. 19860/j . cnki . issn1005- 8249. 2023 . 06. 020

WANG Ruhui

( Shandong Oriental Supervision Consulting Co. ，Ltd. ，Jining 272000 ,   China)

Abstract: In response to the problem of low numerical prediction accuracy caused by the large number of feature quantities and difficulty in extracting in construction project cost samples, a prediction study based on an improved grey Markov model is proposed to effectively solve the problem. This method first uses linear regression analysis to obtain prior data on sub detailed features such as labor cost, machinery usage cost, material cost, and total cost in the construction cost sample, and establishes a data matrix. Then, the probability mean is obtained by adjacent values before and after, and the mean data is initialized to obtain a prior model of the corresponding sub detailed features. Based on the first order representation of a variable in the grey Markov model, a differential prediction model is established to calculate the changes in eigenvalues generated by each data update and compare them with the initial values to generate a sequence of eigenvector changes. Finally, the least squares integration method is used to obtain the probability of data state transition, and combined with the vector and transition probability of the predicted points, effective engineering cost prediction is obtained through search and comparison. The experimental data shows that the proposed method has high prediction accuracy for the distribution of characteristic quantities of construction project costs, and the difference between the predicted cost value and the actual value is small. The prediction performance is excellent.

Keywords: improving the grey markov model; construction project cost; minimum quadratic product algorithm; probability of data state transition; feature vector

作者简介 ：王如会  (1974—) ,  男 ，本科 ，高级工程师 ，研究方向 ：建筑工程。

收稿日期 ：2023- 10-31