CN115994784A - Price determination model and construction method thereof - Google Patents

Abstract

The invention discloses a price determination model construction method, which comprises the following steps: s1, selecting influencing factors based on an index calculation process and actual application scene factors, respectively extracting factors influencing equipment material pricing from three aspects of production cost, market factors and policy factors, and analyzing a price forming mechanism; s2, data collection and data processing: collecting influence factor data and actual price data, constructing a data preprocessing model based on an improved particle swarm optimization clustering model, and preprocessing data to form a data set; and S3, constructing a neural network electric power engineering material pricing model based on genetic algorithm optimization, optimizing the neural network model by utilizing the weight and the threshold value of the genetic algorithm optimization BPNN, and further obtaining a model for determining the price of the electric transmission and transformation engineering iron tower material, so that the price is intelligently determined.

Description

A price determination model and its construction method

technical field

The invention specifically relates to a price determination model and a construction method thereof, which is specifically applied to price determination of iron tower materials in power transmission and transformation projects, and belongs to the technical field of data processing.

Background technique

Faced with the continuous development of social economy and the increasingly complex external construction environment, it poses greater challenges to the management of power grid infrastructure projects. In the process of engineering review, the prices of equipment and materials are generally calculated and listed according to the contract price, market information price, and information price of power grid engineering equipment and materials in sequence; if there are some new materials and new equipment, there is no necessary basis for pricing, and it is generally necessary to refer to similar equipment in recent projects The listing of the winning bid price or market inquiry often leads to a large deviation between the valuation and the actual cost of the project, which affects the optimal allocation and use efficiency of the company's funds. Under market economy conditions, the price of new equipment and new materials is not only affected by its own cost, but also affected by factors such as regional economic development level, industrial policy, industry monopoly and other factors. Therefore, the pricing method is more difficult, and the traditional pricing method is more difficult. Most rely on manual experience or simple statistical methods, and the scientificity of pricing needs to be improved.

Contents of the invention

In order to solve the problems existing in the prior art, the present invention provides a method for constructing a price determination model, which explores the factors affecting the price of equipment and materials from various aspects such as the market, macroeconomics and policies, and uses the mining and analysis technology of influencing factors to determine the price of equipment and materials. In-depth analysis of various factors, combined with data processing technology, constructs a neural network prediction analysis model based on genetic algorithm optimization, and evaluates the prediction effect of the model, so as to realize the correct grasp of the changing trend of equipment and material prices, and strengthen the analysis and analysis of equipment and material prices. Pricing work to help reasonably estimate the cost of power projects, optimize investment strategies, and reduce project investment risks.

Technical scheme of the present invention is as follows:

The invention provides a method for constructing a price determination model, comprising the following steps:

S1. The processing system receives the influencing factors that can affect the price of power transmission and transformation project tower materials obtained after the identification of price influencing factors;

S2. Collect the influencing factor data corresponding to the influencing factors and the actual price data of the iron tower materials of the power transmission and transformation project, build a data preprocessing model based on the improved particle swarm optimization clustering model to find out and delete the influencing factor data and all the influencing factors. outliers and repeated values in the actual price data to form a data set;

S3. Input the data in the data set into the genetic algorithm, and use the genetic algorithm to optimize the weight and threshold of the BP neural network model to obtain a price determination model for determining the price of the iron tower material for the power transmission and transformation project.

Further, in the step S2, the variable mean vector and the variance-covariance matrix are used as prior information to construct a Markov chain, and the Markov chain is repeatedly simulated by sampling to obtain a stable posterior distribution and generate missing data. Estimates include the following steps:

S21. Receive the continuous data vector set Y _c =[Y ₁ ,Y ₂ ,...,Y _n ], wherein the i-th data vector is Y(i)=[y _i (1), y _i (2) ,...,y _i (D)], i=1,2,...,N, wherein, the data in the data vector set is the price influencing factors obtained after identification and analysis in step S1, Y _c includes observed data Y _wz and missing data Y _qs ;

S22. Set the Gaussian model according to the i-th item data, wherein the parameter space of the Gaussian model is θ, and calculate the probability of missing data occurrence p(Y _qs丨Y _wz , θ ^g ) according to the estimated value θ ^g of the parameter space θ ;

以及对高斯模型的参数空间θ的估计值进行更新，直到得到的马尔科夫链

收敛时，估计缺失数据，缺失数据的计算公式如式(Ⅰ)所示：S23. Calculate the occurrence probability of the parameter space θ according to the current occurrence probability of complete data and missing data

And update the estimated value of the parameter space θ of the Gaussian model until the obtained Markov chain

When it converges, the missing data is estimated, and the formula for calculating the missing data is shown in formula (I):

为缺失数据；Among them, N _sample is the total number of samples, N _Burn-in is the number of missing samples,

for missing data;

S24. Deleting outliers and repeated values, and finally obtaining a processed data set.

Further, the method further includes: performing a correlation analysis on the data in the data set obtained after the processing in step S2, and analyzing the existence and strength of the correlation between the influencing factors and the material price dependent variable based on the bivariate correlation analysis model. relationship, and get the main influencing factors according to the strength of the relationship as the input object of the price determination model.

Further, the bivariate correlation analysis model adopts Pearson simple correlation coefficient or hypothesis test for analysis.

Further, the specific steps of optimizing the neural network model in the step S3 are as follows:

S31. In the BP neural network model, determine the network topology, obtain the initial weight and initial threshold length of the BP neural network model, substitute the initial weight and initial threshold length into the genetic algorithm, and use the genetic algorithm to The initial weight and initial threshold length are encoded;

S32. Input the data in the data set obtained after step S2 preprocessing into the genetic algorithm, and pass the data in the data set, the encoded initial weight, and the encoded threshold length through the BP neural network model The error obtained after training is used as the fitness value, and the fitness value is calculated through the selection operation, crossover operation, and mutation operation in the genetic algorithm in turn, and the fitness value that does not meet the end condition is cycled into the selection operation , crossover operation and mutation operation;

S33. Substituting the fitness value that satisfies the end condition after the genetic algorithm operation in step S32 is resubstituted into the BP neural network model to obtain the optimal weight and the optimal threshold length, and then through error calculation, the weight and threshold are updated until the end condition is met. Finally, the final prediction result, that is, the final price, is obtained; the weight and threshold length that do not meet the end condition are cycled into error calculation and weight and threshold update until the end condition is met.

The present invention also provides a method for determining the price of iron tower materials for power transmission and transformation projects, the method comprising:

Obtain the data set of the power transmission and transformation project tower materials to be priced, where the data set is price influencing factor data and actual price data;

The data set is input into the price determination model constructed according to the above-mentioned price determination model construction method to obtain the price of the iron tower material of the power transmission and transformation project to be priced.

The present invention also provides a price determination model construction device, including:

The price influencing factor collection module is used to receive the influencing factors that can affect the price of the iron tower material of the power transmission and transformation project obtained after the price influencing factors are identified;

The data preprocessing model building module is used to collect the influencing factor data corresponding to the influencing factors and the actual price data of the iron tower materials of the power transmission and transformation project, and construct a data preprocessing model based on the improved particle swarm optimization clustering model to find out and delete The influence factor data and the abnormal values and repeated values in the actual price data form a data set, and the data in the data set are input into a genetic algorithm, and the weight of the BP neural network model is optimized by using the genetic algorithm and the threshold value, the price determination model used to determine the price of tower materials for power transmission and transformation projects is obtained.

The present invention also provides a price determination device, including:

The obtaining module is used to obtain the data set of the iron tower material of the power transmission and transformation project to be priced, wherein the data set is price influencing factor data and actual price data;

The input module is configured to input the data set into the price determination model constructed according to the price determination model construction method described in any one of the above embodiments, to obtain the price of the power transmission and transformation engineering tower material to be priced.

The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the program, the above-mentioned method for determining the price of iron tower materials in power transmission and transformation projects is realized. .

The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned method for determining the price of an iron tower material in a power transmission and transformation project is realized.

Compared with the prior art, the beneficial effects of the present invention are:

1. The present invention combines relevant research results and sample data to intelligently determine the price of new materials as the research goal, comprehensively considers the market, macroeconomics and policies, and other factors that affect the price of iron tower materials for power transmission and transformation projects. The analysis technology conducts an in-depth analysis of the factors that affect the price of equipment materials, and then combines data processing technology to construct a neural network predictive analysis model based on genetic algorithm optimization. After the optimized model of the present invention is substituted into the influencing factors through analysis and calculation, the final price of equipment materials Confirmation can correctly grasp the price change trend of equipment and materials, strengthen the price analysis and pricing of equipment and materials, and help to reasonably estimate the cost of electric power projects, optimize investment strategies, and reduce project investment risks.

2. In the present invention, the calculation of equipment and material prices is carried out by constructing a model, which can overcome the unclear pricing basis, unreasonable price adjustment, and large deviation between pricing and actual engineering costs caused by manual pricing or simple statistical methods in the prior art. To solve the problem, improve the scientificity and rationality of pricing, and form a scientific and reasonable new material price determination model, which can further improve the accuracy of cost control, and then achieve the rational allocation of resources and the sustainable development of enterprises.

Description of drawings

Fig. 1 is the schematic diagram of genetic algorithm optimization neural network in the embodiment 1 of the present invention;

Fig. 2 is the fitness curve in the embodiment of the present invention 1;

Fig. 3 is the comparison chart of the results of pricing between the optimized GA-BPNN model and the BPNN model in Example 1 of the present invention;

Fig. 4 is a comparison diagram of the relative error in pricing between the optimized GA-BPNN model and the BPNN model in Example 1 of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and preferred embodiments, and the given embodiments are only for clarifying the present invention, not for limiting the scope of the present invention.

Example 1

A method for building a price determination model, comprising the steps of:

S1. The processing system receives the influencing factors that can affect the price of power transmission and transformation project tower materials obtained after the identification of price influencing factors; among them, the identification process of price influencing factors is: based on the index calculation process and actual application scene factors to select the influencing factors, from The factors affecting the pricing of equipment and materials are extracted from the three aspects of production cost, market factors and policy factors, and the price formation mechanism is analyzed;

S2. Data collection and data processing: collect influencing factor data and actual price data, build a data preprocessing model based on the improved particle swarm optimization clustering model, perform data preprocessing, find and delete outliers and duplicate values in the data, form a data set;

The data set formed after preprocessing includes labor cost, raw material price, product life cycle, supply and demand relationship, inflation degree, industry monopoly degree, regional economic development level, industrial policy, monetary policy, population, temperature and topography, etc. Factors affecting the pricing of equipment and materials (the specific data are mainly selected bidding manufacturers, data published by the Bureau of Statistics, etc.)

S3. Construct a neural network power engineering material pricing model based on genetic algorithm optimization, input the data in the data set obtained through step S2 into the genetic algorithm, and use the genetic algorithm to optimize the weight and threshold of BPNN, and perform a neural network model Optimization, and then obtain a model for determining the price of tower materials for power transmission and transformation projects, making the price intelligently determined.

Among them, in this embodiment, a correlation coefficient verification step can be added in step S2 and step S3, and the data in the data set obtained after the processing of step S2 is subjected to correlation analysis, and the influencing factors are analyzed based on the bivariate correlation analysis model Whether or not the correlation with the material price dependent variable is strong or weak, and the main influencing factors are obtained. The main influencing factors are to eliminate factors such as population, temperature, topography and other factors that are weakly correlated with the predicted structure, and finally serve as the input object of the intelligent determination model .

In this embodiment, the method of fishbone diagram is used in step S1 to identify and analyze the influencing factors, and distinguish them according to major factors, important factors and minor factors, including labor cost, raw material price, product life cycle, supply and demand relationship, Inflation, industry monopoly, regional economic development level, industrial policy, monetary policy, population, temperature and topography, a total of 12 influencing factors.

In this embodiment, in step S2, the variable mean vector and the variance-covariance matrix are used as prior information to construct a Markov chain to ensure that the distribution of its elements can converge to a stable distribution, and the Markov chain is repeatedly simulated by sampling Chain, get a smooth posterior distribution, generate estimates of missing data, and estimate missing data as missing data estimates (missing data that cannot be directly statistically collected or cannot provide complete data, for example: knowing the years or months before and after data, and then estimate the data of the current year or month), which specifically includes the following steps: S21, receiving a continuous data vector set Y _c =[Y ₁ , Y ₂ ,...,Y _n ], wherein the i-th data vector is Y(i)=[y _i (1), y _i (2),..., y _i (D)], i=1, 2,..., N, wherein, the data vector set The data are price influencing factors obtained after the identification and analysis in step S1, and Y _c includes observed data Y _wz and missing data Y _qs ;

S22. Set the Gaussian model according to the i-th item data, wherein the parameter space of the Gaussian model is θ, and calculate the missing data occurrence probability p(Y _qs |Y _wz , θ ^g ) according to the estimated value θ ^g of the parameter space θ;

以及对高斯模型的参数空间θ的估计值进行更新，直到得到的马尔科夫链

收敛时，估计缺失数据以满足相关影响因素的数据完整性，缺失数据的计算公式如式(Ⅰ)所示：S23. Calculate based on the current complete data (data that can be collected directly and can be fully obtained is complete data) and the probability of missing data, for example: knowing the data of several years or months before and after, and then estimating the data of the current year or month) The probability of occurrence of the parameter space θ

And update the estimated value of the parameter space θ of the Gaussian model until the obtained Markov chain

When converging, the missing data is estimated to meet the data integrity of the relevant influencing factors, and the calculation formula of the missing data is shown in formula (I):

为缺失数据；Among them, N _sample is the total number of samples, N _Burn-in is the number of missing samples,

for missing data;

S24. A method of deleting both outliers and duplicate values is adopted. Outliers refer to abrupt values in the data, and duplicate values refer to two data with a nesting relationship, and finally a processed data set is obtained.

In this embodiment, the bivariate correlation analysis model can be analyzed using Pearson simple correlation coefficient or hypothesis testing;

Among them, if the Pearson simple correlation coefficient is used to measure the linear correlation relationship between fixed-distance variables, the calculation formula is shown in (II):

Among them, n is the number of samples, and x _i and y _i are the values of two variables in different samples respectively. Because the calculation formula of Pearson's simple correlation coefficient happens to be in the form of matrix product, it is also called product-distance correlation coefficient. After changing the formula, it is found that the correlation coefficient can be expressed as x _i and y _i are standardized and multiplied, and then the average of n products is calculated;

After the correlation coefficient is obtained according to the variable characteristics, it can be analyzed; when r=0, it means that there is no linear correlation between the two variables; when 0<|r|≤0.3, it means that the two variables are weakly correlated; when 0.3 When <|r|≤0.5, the two are lowly correlated; when 0.5<|r|≤0.8, the two are significantly correlated; when 0.8<|r|<1, the two are highly correlated; when r=1, The two are completely linearly related;

In the analysis of factors affecting the cost of power transmission and transformation projects, it is possible to determine whether there is a significant linear relationship between the factors to lay the foundation for subsequent multi-factor dimensionality reduction and key factor screening.

Among them, if the additional test is used for correlation analysis, the joint distribution of the two variables is pre-set to be a two-dimensional normal distribution: when X takes any value, the conditional distribution of Y is a normal distribution; when Y takes any value , the conditional distribution of X is a normal distribution, and due to the randomness of sampling, small sample size, etc., the results obtained based on the sample selection cannot be directly used to explain the overall population, and it needs to be inferred by the method of hypothesis testing. The steps are as follows:

(1) Putting forward the null hypothesis, there is no significant linear correlation between the two variables;

Construct the test statistic; the test statistic of Pearson correlation coefficient is T statistic, T～t(n-2);

Calculate the observed value of the test statistic, look up the table to get the significance (Sig) corresponding to the observed value, and compare it with the significance level; if it is less than the significance level, reject the null hypothesis and consider that there is a significant linearity between the two variables correlation; anyway, accept the null hypothesis;

(2) Gray relational cluster analysis

There are n observation objects, and each object observes m characteristic data, and the obtained sequence is as follows:

如下

其中，

令

则X_i与X_j的灰色绝对关联度为：

从而得到上三角矩阵A：X _i ＝( _xi (1), _xi (2),…, _xi (n)), and the initial zeroing image generated by Xi _and X _j

as follows

in,

make

Then the gray absolute correlation degree between X _i and X _j is:

Thus, the upper triangular matrix A is obtained:

Where, ε _ij =1, i=1,2,...,m;

The size of the critical value τ (0<τ≤1) can be determined according to the needs of practical problems, generally τ>0.5; the closer τ is to 1, the finer the classification and the fewer features in each category; the smaller the value of τ The coarser the classification, the more features there are in each category; when ε _ij ≥ τ, then regard Xi _{and Xi} as the same kind of features under the level τ, so that the features X ₁ , X ₂ , ..., a classification of X _n at level τ;

Since when _Xi and _Xi are positively correlated, their corresponding S values have the same sign (both positive or negative), and |s _i -s _j | is smaller, and the correlation between _Xi and _Xi is larger; when When Xi _i is negatively correlated with Xi _i , the corresponding S values have different signs, |s _i -s _j | is larger, and the correlation between Xi _i and Xi _i is small. Therefore, the relationship between Xi _i and Xi _i is The two can be considered to be positively correlated when they have the same characteristics;

After analyzing the two variables of influencing factors and price according to any of the above correlation analysis methods, the corresponding main influencing factors are selected, and the above influencing factors are used as the input data of the subsequent genetic algorithm, as shown in the following table:

Wherein, in this embodiment, the specific steps of optimizing the neural network model in the step S3 are shown in Figure 1:

S31. In the BP neural network model, determine the network topology, obtain the initial weight and initial threshold length of the BP neural network model, substitute the initial weight and initial threshold length into the genetic algorithm, and use the genetic algorithm to The initial weight and initial threshold length are encoded;

S32. Input the data in the data set obtained after step S2 preprocessing into the genetic algorithm, and pass the data in the data set, the encoded initial weight, and the encoded threshold length through the BP neural network model The error obtained after training is used as the fitness value, and the fitness value is calculated through the selection operation, crossover operation, and mutation operation in the genetic algorithm in turn, and the fitness value that does not meet the end condition is cycled into the selection operation , crossover operation and mutation operation;

Among them, the end condition is when the fitness value tends to be stable, that is, the weight and threshold of the optimal BP neural network are selected, which can be brought into the next step of the BP neural network; for example, use the genetic algorithm with parameters to optimize BP The weights and thresholds of the neural network, and the changes in the appropriate value of the genetic algorithm during the optimization process are shown in Figure 2. It can be seen from Figure 2 that when the evolution reaches the 19th generation, the average fitness level tends to be stable, and its value is close to 1.3 , to obtain the optimized BP neural network weights and thresholds, the results are as follows:

Connection weights from the input layer to the hidden layer:

Connection weight from hidden layer to output layer: W _jk ＝[-1.1520 -0.3231…-0.2231];

Threshold from input layer to hidden layer: b ₁ =[1.1935 -2.7450...-2.6296];

Threshold from hidden layer to output layer: b ₂ =2.6825;

S33. Substituting the fitness value that satisfies the end condition after the genetic algorithm operation in step S32 is resubstituted into the BP neural network model to obtain the optimal weight and the optimal threshold length, and then through error calculation, the weight and threshold are updated until the end condition is met. After that, the final prediction result is obtained, that is, the final price; the weight and threshold length that do not meet the end condition are cycled into error calculation and weight and threshold update until the end condition is met;

S33. Substituting the fitness value that satisfies the end condition after the genetic algorithm operation in step S32 into the BP neural network operation to obtain the optimal weight and threshold length, and then undergoes error calculation in turn, and the weight and threshold are updated until the end condition is met. Get the final prediction result, that is, the final price; the weight and threshold length that do not meet the end condition will cycle into error calculation and weight and threshold update until the end condition is met. At this time, the end condition is to meet the error target value set before the calculation ;

Specifically, in step S4, apply the genetic algorithm to optimize the BP neural network to price power engineering materials; first, set the parameters of the genetic algorithm and the BP neural network, and set the average The variance is used as the fitness value of the genetic algorithm; the number of input layer nodes of the BP neural network is set to 14, the number of hidden layer nodes is set to 10, the number of output layer nodes is set to 1, and the structure of the BP neural network is constructed as 14-10-1 BP neural network structure; other parameter settings are shown in Table 1;

Table 1 Other parameter settings

In this embodiment, in order to verify the effectiveness of the model, 35 groups of iron tower prices were selected as sample data, the first 25 groups were selected as the training data of the pricing model, the last 10 groups were used as the test data of the model, and 14 influencing factors were used as input to test the validity of the model;

Substitute the parameters optimized above into the BPNN pricing model for pricing of power engineering materials. In order to show the effectiveness of the optimized GA-BPNN model, the pricing results of the BPNN model are introduced as a comparison. The results are shown in Table 2 and Figure 3. It can be obtained from the table and the figure that for the 10 test sample projects, GA-BPNN The average absolute percentage error of the result is 7.55%, which is better than the average absolute percentage error of 14.63% for the BPNN result.

Table 2 Results comparison table

It can be seen from Table 3 and Figure 4 that the relative errors of the GA-BPNN model in samples 4, 7, and 9 are inferior to those of the BPNN model; the relative errors of the results of the remaining 7 projects are all better than those of BPNN. In conclusion, GA-BPNN The accuracy of the BPNN model in the pricing results of power engineering materials is obviously better than that of the BPNN model, which has certain practicability.

Example 2

This embodiment provides a method for determining the price of iron tower materials for power transmission and transformation projects, and the method includes:

Obtain the data set of the power transmission and transformation project tower materials to be priced, where the data set is price influencing factor data and actual price data;

The data set is input into the price determination model constructed according to the price determination model construction method in any of the above embodiments, and the price of the power transmission and transformation engineering tower material to be priced is obtained.

Example 3

This embodiment provides a price determination model construction device, including:

The price influencing factor collection module is used to receive the influencing factors that can affect the price of the iron tower material of the power transmission and transformation project obtained after the price influencing factors are identified;

The data preprocessing model building module is used to collect the influencing factor data corresponding to the influencing factors and the actual price data of the iron tower materials of the power transmission and transformation project, and construct a data preprocessing model based on the improved particle swarm optimization clustering model to find out and delete The influence factor data and the abnormal values and repeated values in the actual price data form a data set, and the data in the data set are input into a genetic algorithm, and the weight of the BP neural network model is optimized by using the genetic algorithm and the threshold value, the price determination model used to determine the price of tower materials for power transmission and transformation projects is obtained.

Example 4

This embodiment provides a price determination device, including:

The obtaining module is used to obtain the data set of the iron tower material of the power transmission and transformation project to be priced, wherein the data set is price influencing factor data and actual price data;

The input module is configured to input the data set into the price determination model constructed according to the price determination model construction method described in Embodiment 1, to obtain the price of the iron tower material of the power transmission and transformation project to be priced.

Example 5

This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the program, the input and transformation as described in Embodiment 1 is realized. Method for determining the price of iron tower materials for electrical engineering.

Example 6

This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method for determining the price of iron tower materials in a power transmission and transformation project as described in Embodiment 1 is implemented.

Those skilled in the art can understand that all or part of the process of implementing the methods of the above embodiments can be completed by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium; wherein, the computer can The read storage medium is a magnetic disk, an optical disk, a read-only memory or a random access memory, and the like.

The above is only an embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process conversion made by using the content of the description of the present invention, or directly or indirectly used in other related technical fields, shall be The same reasoning is included in the patent protection scope of the present invention.

Claims (10)

1. The price determining model construction method is characterized by comprising the following steps:

s1, a processing system receives influence factors which are obtained after price influence factor identification and can influence the price of iron tower materials of power transmission and transformation engineering;

s2, collecting influence factor data corresponding to the influence factors and actual price data of the power transmission and transformation engineering iron tower material, constructing a data preprocessing model based on an improved particle swarm optimization clustering model, finding out and deleting the influence factor data and abnormal values and repeated values in the actual price data, and forming a data set;

s3, inputting the data in the data set into a genetic algorithm, and optimizing the weight and the threshold value of the BP neural network model by using the genetic algorithm to obtain a price determining model for determining the price of the power transmission and transformation engineering iron tower material.

2. The method for constructing a price determining model according to claim 1, wherein in the step S2, a markov chain is constructed by using a variable mean vector and a variance-covariance matrix as prior information, and the markov chain is repeatedly simulated by sampling to obtain a stable posterior distribution, and the method for constructing the estimate of missing data comprises the steps of:

s21, receiving continuous data vector set Y _c ＝[Y ₁ ,Y ₂ ,....,Y _n ]Wherein the ith data vector is Y (i) = [ Y ] _i (1),y _i (2),.....,y _i (D)]I=1, 2,) N, wherein the data in the data vector set is the influencing factor in step S1, Y _c Including observed data Y _wz And absence ofData Y _qs ；

S22, setting a Gaussian model according to the ith item data, wherein a parameter space of the Gaussian model is theta, and estimating a value theta according to the parameter space theta ^g Calculating the occurrence probability p (Y) _qs Y (Y) _wz ，θ ^g )；

S23, calculating the occurrence probability of the parameter space theta according to the occurrence probability of the current complete data and the missing data

And updating the estimated value of the parameter space theta of the Gaussian model until the obtained Markov chain

And during convergence, estimating missing data, wherein a calculation formula of the missing data is shown as a formula (I):

wherein N is _sample N is the total number of samples _Burn-in In order to determine the number of samples to be deleted,

is missing data.

S24, deleting the abnormal value and the repeated value to finally obtain the processed data set.

3. A price determination model construction method as claimed in claim 1, characterized in that the method further comprises: and (2) carrying out correlation analysis on the data in the data set obtained after the processing in the step (S2), analyzing whether the correlation between the influence factors and the material price dependent variables is in a strong-weak relation or not based on a bivariate correlation analysis model, and obtaining main influence factors according to the strong-weak relation to be used as an input object of a price determination model.

4. A price determining model construction method according to claim 3, characterized in that the bivariate correlation analysis model is analyzed by Pearson's simple correlation coefficient or hypothesis test.

5. The method for constructing a price determining model according to claim 1, wherein optimizing the BP neural network model in step S3 comprises:

s31, in a BP neural network model, determining a network topology structure, obtaining an initial weight and an initial threshold length of the BP neural network model, substituting the initial weight and the initial threshold length into a genetic algorithm, and encoding the initial weight and the initial threshold length by using the genetic algorithm;

s32, inputting data in the data set obtained after preprocessing in the step S2 into the genetic algorithm, taking the data in the data set, the coded initial weight and the error obtained after the coded threshold length are trained by the BP neural network model as adaptive values, sequentially carrying out selection operation, crossover operation and mutation operation in the genetic algorithm on the adaptive values, carrying out fitness value calculation, and circularly entering the adaptive values which do not meet the end conditions into the selection operation, crossover operation and mutation operation;

s33, substituting the fitness value which meets the end condition after the genetic algorithm operation in the step S32 into the BP neural network model again to obtain an optimal weight value and an optimal threshold length, and then sequentially carrying out error calculation, updating the weight value and the threshold value until the end condition is met, so as to obtain a final prediction result, namely a final price; and (5) circulating the weight and the threshold length which do not meet the end condition into error calculation and weight and threshold updating until the end condition is met.

6. The method for determining the price of the material of the iron tower of the power transmission and transformation project is characterized by comprising the following steps:

acquiring a data set of a power transmission and transformation engineering iron tower material to be priced, wherein the data set is price influence factor data and actual price data;

inputting the data set into a price determination model constructed according to the price determination model construction method of any one of claims 1-5, so as to obtain the price of the power transmission and transformation engineering iron tower material to be priced.

7. A price determining model construction apparatus, comprising:

the price influence factor collection module is used for receiving influence factors which are obtained after price influence factor identification and can influence the price of the iron tower material of the power transmission and transformation project;

the data preprocessing model construction module is used for collecting influence factor data corresponding to the influence factors and actual price data of the power transmission and transformation engineering iron tower materials, constructing a data preprocessing model based on an improved particle swarm optimization clustering model, finding and deleting abnormal values and repeated values in the influence factor data and the actual price data to form a data set, inputting data in the data set into a genetic algorithm, and optimizing weights and thresholds of a BP neural network model by utilizing the genetic algorithm to obtain a price determination model for determining the price of the power transmission and transformation engineering iron tower materials.

8. A price determining device, comprising:

the acquisition module is used for acquiring a data set of the power transmission and transformation project iron tower material to be priced, wherein the data set is price influence factor data and actual price data;

the input module is used for inputting the data set into a price determination model constructed according to the price determination model construction method according to any one of claims 1-5, so as to obtain the price of the power transmission and transformation engineering iron tower material to be priced.

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 6 when the program is executed by the processor.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any one of claims 1 to 6.

Images