CN112801687A - Overhead line engineering-based cost prediction model construction method - Google Patents
Overhead line engineering-based cost prediction model construction method Download PDFInfo
- Publication number
- CN112801687A CN112801687A CN202011455746.3A CN202011455746A CN112801687A CN 112801687 A CN112801687 A CN 112801687A CN 202011455746 A CN202011455746 A CN 202011455746A CN 112801687 A CN112801687 A CN 112801687A
- Authority
- CN
- China
- Prior art keywords
- cost
- prediction model
- engineering
- overhead line
- construction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 82
- 238000013528 artificial neural network Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 26
- 238000004519 manufacturing process Methods 0.000 claims abstract description 7
- 238000012847 principal component analysis method Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 21
- 238000012795 verification Methods 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 10
- 238000007781 pre-processing Methods 0.000 claims description 9
- 239000011248 coating agent Substances 0.000 claims description 8
- 238000000576 coating method Methods 0.000 claims description 8
- 238000012417 linear regression Methods 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 8
- 230000003068 static effect Effects 0.000 claims description 7
- 238000004140 cleaning Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 5
- 230000015271 coagulation Effects 0.000 claims description 3
- 238000005345 coagulation Methods 0.000 claims description 3
- 238000009430 construction management Methods 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 claims description 3
- 238000013439 planning Methods 0.000 abstract description 2
- 238000009434 installation Methods 0.000 description 14
- 238000000556 factor analysis Methods 0.000 description 8
- 238000012216 screening Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 238000007726 management method Methods 0.000 description 6
- 238000010606 normalization Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 229910000831 Steel Inorganic materials 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000007728 cost analysis Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 239000010959 steel Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 238000013501 data transformation Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004220 aggregation Methods 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000012550 audit Methods 0.000 description 1
- 239000000571 coke Substances 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0206—Price or cost determination based on market factors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a construction method of a cost prediction model based on overhead line engineering, which comprises the following steps: selecting and analyzing historical construction cost data of the completed and put-into-production overhead line project within 2-3 years; on the basis of analyzing main influence factors of overhead line construction cost, a data base is provided for model construction through a principal component analysis method; forming a combined prediction model by training sample data through an algorithm of support vector regression W2 and an artificial neural network W1; if the combined prediction model is analyzed through the prediction result, a cost prediction model is obtained; and in the initial stage of project construction, inputting key factors of the project to be predicted through a cost prediction model, so that a cost prediction value of the project can be obtained. The invention verifies the optimized intelligent prediction model, has certain prediction capability on the prediction model of each subentry cost of the newly-built project, and has more accurate prediction result. The method can assist the investment prediction work in project front-end investment planning.
Description
Technical Field
The invention belongs to the field of construction cost prediction model construction of overhead line engineering, and particularly relates to a construction method of a cost prediction model based on overhead line engineering.
Background
In the construction of overhead line engineering projects, the influence of the management and control of engineering cost on the whole engineering management is significant. The level, the structural change, the influence factors and the development trend of the construction cost are analyzed, the effective control on the construction cost is realized, and the method has great significance for reasonably checking the transmission cost, improving the investment benefit of a power grid and strengthening the project investment management.
However, because the overhead line construction cost has many factors, the construction environment is complex, and the construction level of the line engineering is different, the overhead line construction cost under the same voltage class is different, so that for a certain engineering, the amount of the comparable overhead line construction cost sample which can be referred to is small, and the related auditors of the construction engineering are difficult to audit and estimate the construction cost according to experience and a conventional statistical estimation model.
Therefore, it is an important task to develop a cost prediction model construction method based on overhead line engineering.
Disclosure of Invention
The invention aims to solve the defects of the background technology and provide a construction method of a cost prediction model based on overhead line engineering.
The technical scheme adopted by the invention is as follows: the construction method of the overhead line engineering-based cost prediction model is characterized by comprising the following steps of: the method comprises the following steps:
the method comprises the following steps: selecting and analyzing historical construction cost data of the completed and put-into-production overhead line project within 2-3 years; factors influencing the construction cost level of the overhead line are combed, the main influence factors of construction cost are identified, and the main factors influencing investment cost are determined;
step two: on the basis of analyzing main influence factors of overhead line engineering construction cost, analyzing and identifying the main factors of the overhead line engineering by a principal component analysis method to form a key factor library and provide a data basis for model construction;
taking data in one part of projects in the key factor library as verification sample data, and taking data in the other part of projects as training sample data;
step three: forming a combined prediction model by training sample data through an algorithm of support vector regression W2 and an artificial neural network W1; the combined prediction model comprises a neural network cost intelligent prediction model and a support vector regression cost intelligent prediction model;
step four: analyzing the prediction result of the combined prediction model by verifying the sample data, and if the combined prediction model does not pass through the combined prediction model, re-training the training sample, namely returning to the step two until the training sample passes through the step three;
step five: if the combined prediction model is analyzed through the prediction result, a cost prediction model is obtained;
step six: and in the initial stage of project construction, inputting key factors of the project to be predicted through a cost prediction model, so that a cost prediction value of the project can be obtained.
In the technical scheme, the artificial neural network is an intelligent neural network cost prediction model which is constructed by dividing sample data into training samples and verification samples and continuously correcting loss functions after analyzing and identifying main factors and key factors of overhead line engineering.
In the technical scheme, the support vector regression is a support vector regression cost intelligent prediction model constructed by analyzing historical cost data, changing kernel functions and parameters and continuously correcting loss functions.
In the above technical solution, the method for constructing the overhead line engineering-based cost prediction model specifically includes the following steps:
A. identifying main factors influencing the expense of each project of the overhead line project, and obtaining the corresponding relation between the expense of each subdivided project and the influencing factors;
B. collecting overhead line engineering information and preprocessing the overhead line engineering information;
C. respectively identifying influence factors of the single-circuit line engineering cost and the double-circuit line engineering cost, and respectively establishing a linear regression relationship between the influence factor variables of the single-circuit line engineering cost and the double-circuit line engineering cost and the engineering cost; simplifying the input influence factor variables according to the analysis results screened out by the software respectively, and finally determining the irreplaceable influence factor variable input values of the single-circuit line engineering and the double-circuit line engineering;
D. substituting the finally determined irreplaceable influence factor variable input value of the single-circuit engineering cost into a neural network prediction model of the construction engineering cost to obtain a prediction result of the single-circuit engineering cost;
substituting the irreplaceable influence factor variable input value of the finally determined double-circuit line engineering cost into a neural network prediction model of the double-circuit line engineering cost to obtain a prediction result of the double-circuit line engineering cost;
E. generating body cost through the prediction results of the single-circuit engineering cost and the double-circuit engineering cost, and predicting other costs by combining the body cost; manually inputting a site building fee; and adding the body cost, other costs and the construction cost to obtain the static investment forecast cost. Other costs include construction site acquisition and cleaning costs, project construction management costs, project construction technical service costs, and production preparation costs, among others.
In the above technical solution, the step B specifically includes the following steps: deleting overhead line type factors;
the total (single-folding) s of the line length is kept in the technical index field of the line length, and the comprehensive loop number is calculated, wherein the length s of the single loop is1Length of double loop s2Three loop length s3Length of four loops s4Length of wire hanging s only5Length s of single loop for double loop tower hanging6The comprehensive loop number s is calculated by the following formula:
the tower reserves the basic number of the total tower and classifies the information according to the material;
deleting the wire type factors only having unique values from the wires and the wires;
the ice-coated wind speed line length is subjected to statistical arrangement again, and is divided into two natural conditions of ice coating and wind speed, and the two natural conditions are expressed by main ice coating and main wind speed;
the landform is mainly divided into flat ground, hills, river network mud and marsh, mountains and mountains, and the total proportion of the landform is 100 percent.
Retaining the basic coagulation total amount factors and deleting the proportion information of various types;
the earthwork type comprises a foundation pit, a grounding part, a base surface and a peak, and the total amount of the four earthwork types is taken as a comprehensive factor;
and (5) unifying dimensions.
In the above technical solution, the step C specifically includes the following steps: respectively introducing the influence factor variables of each engineering cost into a multiple linear regression model, and selecting all possible correlation factors; evaluating the quality according to the adjusted R side and the KMO value, and continuously deleting the influencing factor variable with the largest significance value to improve the adjusted R side and the KMO value until the two values are not obviously changed any more; and finding out the missing key influence factor variables, adding the influence factor variables which are not added one by one, and keeping the influence factor variables which are added or do not influence the adjusted R side and the KMO value.
In the above technical solution, in the step D, a neural network prediction model of the single-circuit line engineering cost is constructed by using the influencing factor variable of the single-circuit line engineering cost as an input quantity, wherein the model constructs a neural network of ten hidden layers, and the number of neurons in each layer of the hidden layers is 16, 8, 4; adopting a RELU function; 10000 training rounds and 0.01 learning rate.
In the above technical solution, in the step D, a neural network prediction model of the single-circuit line engineering cost is constructed by using an influencing factor variable of the double-circuit line engineering cost as an input quantity, wherein the model constructs a neural network of ten hidden layers, and the number of neurons in each layer of the hidden layers is 16, 8, 4; adopting a RELU function; 10000 training rounds, 0.01 learning rate
The construction method of the overhead line engineering-based cost prediction model has the following advantages: the invention screens the influence factors of the item cost prediction model by adopting a linear regression model and a factor analysis theory, and screens out the factors with smaller influence factors and insignificant weight coefficients.
For newly-built overhead line engineering, key influence factors influencing installation engineering cost and other cost are established.
The invention uses the optimized neural network intelligent prediction model to independently model the overhead line engineering investment prediction model by dividing static cost prediction into installation single-loop engineering cost and double-loop engineering cost.
And summarizing a prediction model for the item cost, and finally establishing a static investment prediction model for the overhead line engineering to meet the requirement of construction cost prediction dimension refinement.
The method carries out model verification on the optimized neural network intelligent prediction model, has certain prediction capability on the prediction model of each subentry cost of the newly-built overhead line engineering, and has accurate prediction result. The method can assist the investment prediction work in project front-end investment planning.
Drawings
FIG. 1 is a schematic diagram of a method of constructing an overhead line engineering based cost prediction model according to the present invention;
FIG. 2 is a schematic flow chart of a method for constructing an overhead line engineering-based cost prediction model according to the present invention;
FIG. 3 is a schematic diagram of overhead line installation project cost division;
FIG. 4 is a normal P-P diagram of an overhead single-circuit line installation project as a dependent variable;
FIG. 5 is a predicted neural network topology for an overhead line installation project;
FIG. 6 is a graph of installation engineering loss values and trends.
Detailed Description
The invention will be further described in detail with reference to the drawings and specific examples to facilitate a clear understanding of the invention, but they are not to be construed as limiting the invention, but merely as exemplifications, and the advantages thereof will be more clearly understood and appreciated by illustrating the invention.
With reference to the accompanying drawings: the technical scheme adopted by the invention is as follows: the construction method of the overhead line engineering-based cost prediction model comprises the following steps:
the method comprises the following steps: selecting and analyzing historical construction cost data of the completed and put-into-production overhead line project within 2-3 years; factors influencing the construction cost level of the overhead line are combed, the main influence factors of construction cost are identified, and the main factors influencing investment cost are determined;
step two: on the basis of analyzing main influence factors of overhead line engineering construction cost, analyzing and identifying the main factors of the overhead line engineering by a principal component analysis method to form a key factor library and provide a data basis for model construction;
taking data in one part of projects in the key factor library as verification sample data, and taking data in the other part of projects as training sample data;
step three: forming a combined prediction model by training sample data through an algorithm of support vector regression W2 and an artificial neural network W1; the combined prediction model comprises a neural network cost intelligent prediction model and a support vector regression cost intelligent prediction model;
step four: analyzing the prediction result of the combined prediction model by verifying the sample data, and if the combined prediction model does not pass through the combined prediction model, re-training the training sample, namely returning to the step two until the training sample passes through the step three;
step five: if the combined prediction model is analyzed through the prediction result, a cost prediction model is obtained;
step six: and in the initial stage of project construction, inputting key factors of the project to be predicted through a cost prediction model, so that a cost prediction value of the project can be obtained. As shown in fig. 1.
The artificial neural network is an intelligent neural network cost prediction model constructed by dividing sample data into training samples and verification samples and continuously correcting loss functions after analyzing and identifying main factors and key factors of overhead line engineering.
The support vector regression is a support vector regression cost intelligent prediction model constructed by analyzing historical cost data, changing kernel functions and parameters and continuously correcting loss functions.
As shown in fig. 2, the optimization of the invention for the overhead line project is mainly to model and predict the single-circuit line project and the double-circuit line project in the project separately, and then predict other expenses by summarizing, and predict the construction cost of the overhead line project by summarizing construction expenses.
The construction method of the overhead line engineering-based cost prediction model specifically comprises the following steps: as shown in fig. 2.
A. Identifying main factors influencing the expense of each project of the overhead line project, and obtaining the corresponding relation between the expense of each subdivided project and the influencing factors;
B. collecting overhead line engineering information and preprocessing the overhead line engineering information;
C. respectively identifying influence factors of the single-circuit line engineering cost and the double-circuit line engineering cost, and respectively establishing a linear regression relationship between the influence factor variables of the single-circuit line engineering cost and the double-circuit line engineering cost and the engineering cost; simplifying the input influence factor variables according to the analysis results screened out by the software respectively, and finally determining the irreplaceable influence factor variable input values of the single-circuit line engineering and the double-circuit line engineering;
D. substituting the finally determined irreplaceable influence factor variable input value of the single-circuit engineering cost into a neural network prediction model of the construction engineering cost to obtain a prediction result of the single-circuit engineering cost;
substituting the irreplaceable influence factor variable input value of the finally determined double-circuit line engineering cost into a neural network prediction model of the double-circuit line engineering cost to obtain a prediction result of the double-circuit line engineering cost;
E. generating body cost through the prediction results of the single-circuit engineering cost and the double-circuit engineering cost, and predicting other costs by combining the body cost; manually inputting a site building fee; and adding the body cost, other costs and the construction cost to obtain the static investment forecast cost.
The overhead line engineering is mainly divided into two aspects of body engineering and other expenses. By means of the further subdivision it is possible to,
overhead line body engineering generally comprises foundation engineering, tower engineering, grounding engineering, stringing engineering, accessory engineering and auxiliary engineering; other costs typically include construction site acquisition and cleanup costs, project construction management costs, project construction technical service costs, and production preparation costs. As shown in fig. 3.
And excavating main influence factors of the body engineering cost through the subdivision of the overhead line body engineering project to obtain the corresponding relation between the body engineering cost subdivision project and the influence factors. Such as: for tower engineering, the tower construction is influenced by main factors such as wire section, wind speed, icing, strain proportion, tower usage, terrain and the like. The specific analysis is shown in the following table.
TABLE 1 overhead line engineering cost major impact factor identification
For other expenses, except that the expropriation and cleaning expenses of the construction site are greatly influenced by factors such as project scale, terrain, geographical position and the like, other items are mainly related to project main body project expenses, and the related influence factors of the project main body project expenses can be referred.
In the specific embodiment, typical engineering selected by the neural network model modeling training sample is real overhead line engineering data of 110kV-500kV in Hubei province from 2016 to 2018, and the engineering range of the real overhead line engineering data covers 14 city units in the asset jurisdiction area of Hubei companies. Through statistics, 421 items of samples used in training are counted, including investment estimation, initial setting approximate calculation, engineering settlement and completion settlement data of engineering with three voltage levels of 500kV, 220kV and 110kV, wherein: 9 items of 500kV sample engineering, 101 items of 220kV sample engineering and 311 items of 110kV sample engineering. Because the modeling is in the verification and test stage, the data preprocessing is only carried out on the samples affecting the overhead line engineering in the stage. There are 263 overhead line projects in total in the training samples.
In 263 samples of the overhead line project collected at this time, 5 500kV overhead line projects, 66 220kV overhead line projects and 192 kV overhead line projects are included.
In the process of counting and sorting the historical construction cost data of the overhead line project, not all the construction cost data are directly imported and calculated, most of the data have the problems of irregularity, unclear concept level, different orders of magnitude and the like, and cannot be used for data analysis and data mining in the next stage. Therefore, the development of data preprocessing work is particularly important, and the data preprocessing is carried out by adopting four means such as data cleaning, integration, transformation and specification, so as to ensure the authenticity and the accuracy of training sample data of an input model.
Data cleaning: and filling or neglecting data omission, denoising impurity data, and checking and correcting inconsistent data.
Data integration: integrating a plurality of original data into one data storage to check and process data redundancy and data conflict values.
Data transformation: the data transformation mainly comprises means of smoothing, aggregation, data generalization, normalization, attribute construction and the like. The normalization modes include maximum and minimum normalization, normalization with 0-1 value, decimal scaling normalization and the like.
And (3) data reduction: the data reduction technology is mainly used for reducing the data set, reducing the time required by analysis without losing the integrity of the data, and ensuring the consistency of results before and after reduction.
In the data preprocessing stage, the factors influencing the construction cost of the overhead line project are various and have differences, so that the data preprocessing is carried out on the sample set in the project. The main preprocessing case contents are as follows:
the 'overhead line type' is conventional, so the overhead line type factor is eliminated.
② the technical index column of 'line length' retains 'line length total (single-folding) s', at the same time, it calculates comprehensive loop number, in which the single-loop length s1Length of double loop s2Three loop length s3Length of four loops s4Length of wire hanging s only5Length s of single loop for double loop tower hanging6The comprehensive loop number s is calculated by the following formula:
and thirdly, the basic number of the total tower is reserved by the tower, and information is classified according to materials.
And fourthly, deleting the wire type factors only having unique values from the wires and the wires.
Fifthly, the ice-coating wind speed line length is counted again and sorted, the ice-coating wind speed line length is divided into two natural conditions of ice coating and wind speed, and the two natural conditions are expressed by main ice coating and main wind speed.
Sixthly, the terrain is mainly divided into flat ground, hilly ground, river network mud and marsh, mountain land and mountain land, and the total proportion of the terrain is 100 percent.
Retaining the basic coagulation total amount factor and deleting the proportion information of various types.
The types of the earthwork comprise a foundation pit, a grounding part, a base plane and a peak, and the total amount of the four types of earthwork is taken as a comprehensive factor.
Dimension is unified in the coke.
All collected factors do not have obvious influence on the cost result, some factors cannot cause great influence on the whole cost level within a certain range, and if all variables are included in the model, the precision of the model is influenced. Further key influencing factor analysis can be performed from the collected factors through a linear regression model and a factor analysis method.
The raw data is imported into SPSS (statistical Product and Service solutions) software, and the SPSS is used for carrying out correlation analysis of various factors with the cost of construction, equipment and installation. All engineering variables are introduced during the initial analysis by using a multiple linear regression model, all possible correlation factors are selected, but the result of the 'R square after adjustment' is about 0.5, which indicates that the correlation of some factors with the cost of buildings, equipment and installation is not obvious, and the factors cannot be used as input attributes during the establishment of a neural network. There are also a few factors that do not take into account the input variables added to the model at the time of initial screening, and should be taken into account in the input attributes that ultimately build the neural network.
The main factors are screened respectively by using a one-by-one comparison experiment method. Firstly, the method comprises the following steps: the evaluation of the quality is performed according to the values of the adjusted R-square and the KMO. The "adjusted R-square" and "KMO" values are increased by continually eliminating the factor by which the "significance" value is greatest until the values no longer change significantly. Secondly, the method comprises the following steps: finding out the missing key factors, adding the factors which are not added one by one, and keeping the factors which increase or do not influence the values of the 'adjusted R square' and the 'KMO'. Because the more neural network input attributes are established, the more accurate the trained model is, and more factors with good correlation are reserved as far as possible.
The irreplaceable influence factor selection process of the overhead single-circuit installation project is as follows:
the factors initially selected were: OPGW price element, tower material amount, single index of wire amount (t) folding, the number of high-voltage side circuit breakers, single index of concrete beam folding, strain proportion (%), tower material price (element/t), single index of tower base folding, total line length (folding), single index of base steel amount (t) folding and high-strength steel tower material amount ratio.
TABLE 2-1 construction engineering fee factor analysis preliminary screening results
TABLE 2-2 overhead single-circuit installation engineering cost factor analysis preliminary screening results
TABLE 3 overhead single-circuit installation engineering cost factor analysis significance coefficient preliminary screening results
| Factor(s) | Normalized coefficient | t | Significance of |
| High side outgoing line number | 163365.278 | .260 | 1.460 |
| Number of outgoing lines at low voltage side | 45881.427 | 1.037 | 2.778 |
| Main unit price of transformer equipment | .213 | -.133 | -1.191 |
| Number of high-voltage side circuit breakers | 92416.035 | .014 | .072 |
| Number of main transformer in this period | 486233.593 | -.364 | -1.236 |
| Number of medium voltage side circuit breakers | 47691.594 | .488 | 2.551 |
| Number of low-voltage side circuit breakers | 32543.421 | .067 | .195 |
| High-pressure side breaker unit price (Wanyuan/table) | 7643.769 | .068 | .433 |
| Low-voltage side circuit breaker unit price | 70139.060 | .006 | .046 |
The main factors for the final screening are: the method comprises the following steps of (1) folding a single index for the quantity (t) of basic steel, a single index for the square of basic earth and stone, a single index for the quantity (t) of lead wires, a strain ratio (%), a single index for concrete beams, a single index for the base number of towers, the price (unit/t) of tower materials, the total length of lines (folding unit), a single index for the quantity (t) of tower materials and a single index for the number of towers.
TABLE 4 overhead single-circuit installation engineering fee factor analysis final screening results
TABLE 5 overhead single-circuit installation engineering cost factor analysis significance coefficient final screening results
The process of double-loop cost factor screening and the irreplaceable influence factor selection process of the overhead single-loop installation project are the same.
The optimization model for neural network prediction of the overhead line engineering selects 28 pieces of 220kV data in the overhead line as data samples (a section of line below 3km without a relocation line), randomly selects 4 pieces of data as a verification sample set, and the rest 24 pieces of data as a training sample set, and builds a neural network with eleven hidden layers, wherein the number of neurons in each layer of the hidden layers is 16, 8, 4, 8 and 4 respectively.
Because overhead lines have less training data available, the model error has been above 50% despite the correlation of 0.98. The initial design uses an activation function which is a sigmoid function, the sigmoid is an activation function which is not suitable for a regression model and is most commonly used as a classification function, and then the activation function is used as a RELU function, the RELU function is the best parameter of the current training model in the industry, and the error is reduced to about 35 percent after the use. The number of training layers for overhead lines is initially 4, but the effect is not always good. The deeper neural network training effect is better than the wider neural network, so the number of layers is increased from 4 layers to 11 layers, and finally the model error is controlled to be about 15%.
10000 training rounds are continuously used, the learning rate is 0.01, and the unit capacity cost prediction neural network topological structure is shown in figure 6:
and (3) verifying the 4 verification set data by using the trained model, inputting the scales and technical parameters of the 4 220kV overhead line projects to be predicted into the model, and outputting verification results as shown in a table 6.
TABLE 6220 kV overhead line prediction result table
According to the chart, there are 2 predicted static investment and real static investment deviations within 10%, 1 deviation in 10% -20%, and 1 deviation in 20% -25%, all within a reasonable range.
The average deviation of investment of 4 prediction projects is 10.25%.
Through the construction of an automatic construction cost analysis model and a big data mining calculation means, proper construction cost analysis data is selected to correspond to a service platform for data calculation training, an automatic construction cost analysis function is formed, and sample classification and comparison analysis, measurement and calculation of investment balance rate and deviation analysis, and setting and analysis of reasonable construction cost intervals can be automatically completed; the method can automatically generate post-investment evaluation analysis and conclusion for a certain overhead transmission line engineering project, deeply excavate factors which have large influence on investment, and form measures for controlling different investment in each stage.
Project investment management units should make prior control, process management and post closed loop. And each risk point which possibly influences the engineering cost in the project period is strictly controlled, the precaution consciousness is enhanced, and when the construction cost is changed, a countermeasure is taken in time to control the engineering cost within a reasonable range, so that the lean management of the engineering cost of the power distribution network is promoted, the transformation and the upgrade of the power distribution network construction are promoted, the modernization construction development process of the power enterprises is met, and the social livelihood is better served.
Those not described in detail in this specification are within the skill of the art.
Claims (8)
1. The construction method of the overhead line engineering-based cost prediction model is characterized by comprising the following steps of: the method comprises the following steps:
the method comprises the following steps: selecting and analyzing historical construction cost data of the completed and put-into-production overhead line project within 2-3 years; factors influencing the construction cost level of the overhead line are combed, the main influence factors of construction cost are identified, and the main factors influencing investment cost are determined;
step two: on the basis of analyzing main influence factors of overhead line engineering construction cost, analyzing and identifying the main factors of the overhead line engineering by a principal component analysis method to form a key factor library and provide a data basis for model construction;
taking data in one part of projects in the key factor library as verification sample data, and taking data in the other part of projects as training sample data;
step three: forming a combined prediction model by training sample data through an algorithm of support vector regression W2 and an artificial neural network W1; the combined prediction model comprises a neural network cost intelligent prediction model and a support vector regression cost intelligent prediction model;
step four: analyzing the prediction result of the combined prediction model by verifying the sample data, and if the combined prediction model does not pass through the combined prediction model, re-training the training sample, namely returning to the step two until the training sample passes through the step three;
step five: if the combined prediction model is analyzed through the prediction result, a cost prediction model is obtained;
step six: and in the initial stage of project construction, inputting key factors of the project to be predicted through a cost prediction model, so that a cost prediction value of the project can be obtained.
2. The overhead line engineering-based cost prediction model construction method according to claim 1, characterized in that: the artificial neural network is an intelligent neural network cost prediction model constructed by dividing sample data into training samples and verification samples and continuously correcting loss functions after analyzing and identifying main factors and key factors of overhead line engineering.
3. The overhead line engineering-based cost prediction model construction method according to claim 1, characterized in that: the support vector regression is a support vector regression cost intelligent prediction model constructed by analyzing historical cost data, changing kernel functions and parameters and continuously correcting loss functions.
4. The method of constructing an overhead line engineering based cost prediction model according to claim 1, 2 or 3, wherein: the construction method specifically comprises the following steps:
A. identifying main factors influencing the expense of each project of the overhead line project, and obtaining the corresponding relation between the expense of each subdivided project and the influencing factors;
B. collecting main factors of overhead line engineering information, forming a factor library, and preprocessing the main factors;
C. respectively identifying influence factors of the single-circuit line engineering cost and the double-circuit line engineering cost, and respectively establishing a linear regression relationship between the influence factor variables of the single-circuit line engineering cost and the double-circuit line engineering cost and the engineering cost;
then simplifying the input influence factor variables by the respectively screened analysis results, and finally determining the input values of the influence factor variables of the single-circuit line engineering and the double-circuit line engineering;
D. substituting the finally determined influence factor variable input value of the single-circuit engineering cost into a neural network prediction model of the construction engineering cost to obtain a prediction result of the single-circuit engineering cost;
substituting the finally determined influence factor variable input value of the double-circuit line engineering cost into a neural network prediction model of the double-circuit line engineering cost to obtain a prediction result of the double-circuit line engineering cost;
E. generating body cost through the prediction results of the single-circuit engineering cost and the double-circuit engineering cost, and predicting other costs by combining the body cost; and manually inputting the site building fee;
adding the body cost, other costs and the site building cost to obtain static investment forecasting cost;
other costs include construction site acquisition and cleaning costs, project construction management costs, project construction technical service costs, and production preparation costs, among others.
5. The overhead line engineering-based cost prediction model construction method according to claim 4, wherein: the step B specifically comprises the following steps: deleting overhead line type factors; keeping the total s of the length of the line in the technical index field of the length of the line, and calculating the number of comprehensive loops, wherein the length s of the single loop1Length of double loop s2Three loop length s3Length of four loops s4Length of wire hanging s only5Double-return tower hangerThe single loop length s6, the overall loop number s, is calculated by the following equation:
s＝s1+1+s2+2+s3+3+s4+4+s5+s6
the tower reserves the basic number of the total tower and classifies the information according to the material;
deleting the wire type factors only having unique values from the wires and the wires;
the ice-coated wind speed line length is subjected to statistical arrangement again, and is divided into two natural conditions of ice coating and wind speed, and the two natural conditions are expressed by main ice coating and main wind speed;
the landform is mainly divided into flat ground, hills, river network mud and marsh, mountains and mountains, and the total proportion of the landform is 100 percent;
retaining the basic coagulation total amount factors and deleting the proportion information of various types;
the earthwork type comprises a foundation pit, a grounding part, a base surface and a peak, and the total amount of the four earthwork types is taken as a comprehensive factor;
and (5) unifying dimensions.
6. The overhead line engineering-based cost prediction model construction method according to claim 4, wherein: the step C specifically comprises the following steps: respectively introducing the influence factor variables of each engineering cost into a multiple linear regression model, and selecting all possible correlation factors; evaluating the quality according to the adjusted R side and the KMO value, and continuously deleting the influencing factor variable with the largest significance value to improve the adjusted R side and the KMO value until the two values are not obviously changed any more; and finding out the missing key influence factor variables, adding the influence factor variables which are not added one by one, and keeping the influence factor variables which are added or do not influence the R side and the KMO value after adjustment.
7. The overhead line engineering-based cost prediction model construction method according to claim 4, wherein: in the step D, a neural network prediction model of the single-circuit line engineering cost is constructed by taking influencing factor variables of the single-circuit line engineering cost as input quantities, wherein the model constructs a neural network of eleven hidden layers, and the number of neurons in each layer of the hidden layers is 16, 8, 4, 8 and 4 respectively; adopting a RELU function; 10000 training rounds and 0.01 learning rate.
8. The overhead line engineering-based cost prediction model construction method according to claim 4, wherein: in the step D, a neural network prediction model of the single-circuit line engineering cost is constructed by taking influencing factor variables of the double-circuit line engineering cost as input quantities, wherein the model constructs a neural network of eleven hidden layers, and the number of neurons in each layer of the hidden layers is 16, 8, 4, 8 and 4 respectively; adopting a RELU function; 10000 training rounds and 0.01 learning rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011455746.3A CN112801687A (en) | 2020-12-10 | 2020-12-10 | Overhead line engineering-based cost prediction model construction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011455746.3A CN112801687A (en) | 2020-12-10 | 2020-12-10 | Overhead line engineering-based cost prediction model construction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112801687A true CN112801687A (en) | 2021-05-14 |
Family
ID=75806621
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011455746.3A Pending CN112801687A (en) | 2020-12-10 | 2020-12-10 | Overhead line engineering-based cost prediction model construction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112801687A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037538A (en) * | 2021-11-15 | 2022-02-11 | 国网湖北省电力有限公司经济技术研究院 | Power grid infrastructure project investment balance control method and system |
| CN114187047A (en) * | 2021-12-13 | 2022-03-15 | 苏州方兴信息技术有限公司 | Project cost determination method and device, electronic equipment and storage medium |
| CN114329691A (en) * | 2021-11-25 | 2022-04-12 | 国网河南省电力公司经济技术研究院 | Foundation construction floor area calculation method, computer equipment and storage medium |
| CN114819417A (en) * | 2022-06-28 | 2022-07-29 | 北京航空航天大学 | Aero-engine overhaul cost prediction method and system based on correlation between costs |
| CN115034598A (en) * | 2022-06-02 | 2022-09-09 | 国网经济技术研究院有限公司 | Overhead line engineering cost data analysis, review and analysis method |
| CN115526393A (en) * | 2022-09-23 | 2022-12-27 | 国网冀北电力有限公司经济技术研究院 | Construction cost prediction method based on transformer project key influence factor screening |
| CN118469768A (en) * | 2024-07-11 | 2024-08-09 | 中铁四局集团有限公司 | Engineering component consumption data analysis processing method, equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160224951A1 (en) * | 2004-09-10 | 2016-08-04 | Steven M. Hoffberg | Game theoretic prioritization system and method |
| US20190188584A1 (en) * | 2017-12-19 | 2019-06-20 | Aspen Technology, Inc. | Computer System And Method For Building And Deploying Models Predicting Plant Asset Failure |
| CN110264260A (en) * | 2019-06-17 | 2019-09-20 | 国网新疆电力有限公司经济技术研究院 | A kind of intelligent Forecasting of grid equipment material price |
| CN110992113A (en) * | 2019-12-23 | 2020-04-10 | 国网湖北省电力有限公司 | Neural network intelligent algorithm-based project cost prediction method for capital construction transformer substation |
| CN111127246A (en) * | 2019-11-21 | 2020-05-08 | 国网福建省电力有限公司经济技术研究院 | Intelligent prediction method for transmission line engineering cost |
-
2020
- 2020-12-10 CN CN202011455746.3A patent/CN112801687A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160224951A1 (en) * | 2004-09-10 | 2016-08-04 | Steven M. Hoffberg | Game theoretic prioritization system and method |
| US20190188584A1 (en) * | 2017-12-19 | 2019-06-20 | Aspen Technology, Inc. | Computer System And Method For Building And Deploying Models Predicting Plant Asset Failure |
| CN110264260A (en) * | 2019-06-17 | 2019-09-20 | 国网新疆电力有限公司经济技术研究院 | A kind of intelligent Forecasting of grid equipment material price |
| CN111127246A (en) * | 2019-11-21 | 2020-05-08 | 国网福建省电力有限公司经济技术研究院 | Intelligent prediction method for transmission line engineering cost |
| CN110992113A (en) * | 2019-12-23 | 2020-04-10 | 国网湖北省电力有限公司 | Neural network intelligent algorithm-based project cost prediction method for capital construction transformer substation |
Non-Patent Citations (1)
| Title |
|---|
| 唐定曾 唐海: "《建筑工程电气概算（第3版）》", 31 October 1997, 中国建筑工业出版社 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037538A (en) * | 2021-11-15 | 2022-02-11 | 国网湖北省电力有限公司经济技术研究院 | Power grid infrastructure project investment balance control method and system |
| CN114037538B (en) * | 2021-11-15 | 2023-11-03 | 国网湖北省电力有限公司经济技术研究院 | Method and system for controlling investment balance of power grid infrastructure project |
| CN114329691A (en) * | 2021-11-25 | 2022-04-12 | 国网河南省电力公司经济技术研究院 | Foundation construction floor area calculation method, computer equipment and storage medium |
| CN114187047A (en) * | 2021-12-13 | 2022-03-15 | 苏州方兴信息技术有限公司 | Project cost determination method and device, electronic equipment and storage medium |
| CN115034598A (en) * | 2022-06-02 | 2022-09-09 | 国网经济技术研究院有限公司 | Overhead line engineering cost data analysis, review and analysis method |
| CN114819417A (en) * | 2022-06-28 | 2022-07-29 | 北京航空航天大学 | Aero-engine overhaul cost prediction method and system based on correlation between costs |
| CN115526393A (en) * | 2022-09-23 | 2022-12-27 | 国网冀北电力有限公司经济技术研究院 | Construction cost prediction method based on transformer project key influence factor screening |
| CN115526393B (en) * | 2022-09-23 | 2023-09-29 | 国网冀北电力有限公司经济技术研究院 | Manufacturing cost prediction method based on screening of key influencing factors of power transformation engineering |
| CN118469768A (en) * | 2024-07-11 | 2024-08-09 | 中铁四局集团有限公司 | Engineering component consumption data analysis processing method, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112801687A (en) | Overhead line engineering-based cost prediction model construction method | |
| CN112508275B (en) | Power distribution network line load prediction method and equipment based on clustering and trend indexes | |
| CN112541631A (en) | Expense prediction method for transformer substation engineering | |
| CN105930931A (en) | Electric power engineering cost management method | |
| CN110992113A (en) | Neural network intelligent algorithm-based project cost prediction method for capital construction transformer substation | |
| Enke et al. | A hybrid neuro-fuzzy model to forecast inflation | |
| Zhang et al. | Impact of structure on unified efficiency for Chinese service sector—A two-stage analysis | |
| CN110490409B (en) | DNN-based low-voltage transformer area line loss rate benchmarking value setting method | |
| CN108074004A (en) | A kind of GIS-Geographic Information System short-term load forecasting method based on gridding method | |
| CN109034511A (en) | Based on the power distribution network investment decision analysis model for improving Topsis method | |
| CN108021773A (en) | The more play flood parameters rating methods of hydrological distribution model based on DSS data base read-writes | |
| CN106599417A (en) | Method for identifying urban power grid feeder load based on artificial neural network | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN112085156A (en) | Method and system for predicting shunt line loss rate and evaluating loss reduction potential | |
| CN104573857A (en) | Power grid load rate prediction method based on intelligent algorithm optimization and combination | |
| CN112347670A (en) | Rockfill material creep parameter prediction method based on neural network response surface | |
| CN109872252A (en) | A kind of electricity provider integrated evaluating method based on MATLAB algorithm | |
| CN109657846A (en) | Power grid alternative subsidy scale impact factor screening technique | |
| CN108009684A (en) | A kind of micro-grid connection state energy management method comprising short-term load forecasting | |
| CN108537581A (en) | Based on the GMDH energy consumption Time Series Forecasting Methods selectively combined and device | |
| CN115511543A (en) | Power transmission and transformation project cost measuring and calculating method | |
| CN116611785A (en) | Power transmission and transformation project cost model construction method, system, equipment and medium based on big data | |
| CN114037538B (en) | Method and system for controlling investment balance of power grid infrastructure project | |
| CN115936792A (en) | Power transmission and transformation project cost prediction method | |
| Han et al. | Predicting Growth of City's Built-up Land Based on Scenario Planning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210514 |