A method for analyzing and displaying engineering cost data

Abstract

A method for analyzing and displaying engineering cost data, which first collects data of completed projects in various regions at different stages and levels and then inputs them into a constructed big data sharing platform, which then classifies and integrates them into a data set, and then inputs the data of the total project and each sub-project of the project to be constructed. The big data sharing platform then matches completed projects of the same type according to the type of the project to be constructed, and then uses the matched completed project data as samples for training to form a neural network prediction model, and inputs the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction results of the project to be constructed, and finally, after obtaining the user's query request command and screening conditions, retrieves the prediction results, filters out the results that meet the user's query requirements, collects the screening results and displays them in charts. This design not only ensures the accuracy of cost prediction, but also realizes the in-depth display of data.

Description

Technical Field

[0001] The invention belongs to the technical field of engineering cost, and in particular relates to an analysis and display method for engineering cost data. Background Art

[0002] Engineering cost is a comprehensive application of knowledge and skills in management, economics and engineering technology. Based on engineering cost analysis and overall data integration, engineering cost is a relatively complex system. It includes many influencing factors, and there are also many uncertain factors in the construction process of the project. The relationship between these factors is intricate, and it is difficult to determine the impact of a single factor on the cost. Therefore, it affects the accurate and rapid prediction of the estimated project cost. At the same time, a large amount of data calculation work is involved in the process of data collection and engineering cost calculation. The consideration of multiple influencing factors will affect the accuracy of the engineering cost, and it is easy to have a large deviation between the judgment of the engineering cost and the actual cost. Summary of the invention

[0003] The purpose of the present invention is to solve the above problems existing in the prior art and to provide a method for analyzing and displaying engineering cost data which can effectively improve the accuracy of cost prediction.

[0004] To achieve the above objectives, the present invention provides the following technical solutions:

[0005] A method for analyzing and displaying engineering cost data comprises the following steps in sequence:

[0006] Step A: First collect the data of completed projects in various regions at different stages and levels and input them into the constructed big data sharing platform, which then classifies and integrates the above data to form a data set;

[0007] Step B: first input the data of the total project and each sub-project of the project to be constructed into the big data sharing platform, and then the big data sharing platform matches the completed projects of the same type according to the type of the project to be constructed;

[0008] Step C: using the matched data of completed projects as samples for training to form a neural network prediction model;

[0009] Step D: inputting the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction result of the project to be constructed;

[0010] Step E: After obtaining the user's query request command and screening conditions, retrieve the prediction results of step D, screen out the results that meet the user's query requirements, aggregate the screening results and generate corresponding charts for display.

[0011] In step A, the data of the completed project at different stages and levels include engineering cost data, project terrain proportions, meteorological conditions and construction technology solutions;

[0012] The method for forming a data set by classification and integration is as follows: first, the terrain data in the completed project data is divided into five terrain data sets, namely, mountain, plateau, plain, hill and basin, according to the geographical location, and the range values ​​of the five terrain data sets are encoded, and then the standard coding value range of all data in each terrain data set is determined according to the classification standard of the terrain data and the range value of each terrain data set, and then the other data in the completed project data is divided into a general data set, a design data set, a process processing data set and a management information data set, and the values ​​of the general data set, the design data set, the process processing data set and the management information data set are encoded to complete the classification and integration of the data.

[0013] The step B first identifies and encodes the data of the project to be constructed, and then matches it with the coding values ​​of various terrain data sets, general data sets, design data sets, process processing data sets and management information data sets based on fuzzy mathematical theory to determine completed projects of the same type.

[0014] In step C, the neural network prediction model is:

[0015] p(Y′|Y,α * ,σ *2 )=∫p(Y′|w,σ *2 )p(w|Y,α * ,σ *2 )dw

[0016] In the above formula, p(Y′|Y,α * ,σ *2 ) is the likelihood function, Y′ is the input sample at the next moment, Y is the input sample at the current moment, α * is the hyperparameter that determines the prior distribution of node weight w, σ *2 To optimize the parameters.

[0017] The step B also includes automatic inspection of the construction project data, specifically:

[0018] First, the maximum and minimum values ​​of each data in the completed projects of the same type are statistically matched and used as the reasonable range of the data. Then, the data of the project to be constructed are compared with its corresponding reasonable range. If the data of the project to be constructed is within the reasonable range, it is input into the neural network prediction model as valid data for prediction calculation. If the data of the project to be constructed is not within the reasonable range, a warning will be issued.

[0019] In step E, the process of aggregating and filtering results includes: first classifying the filtered data into levels, then aggregating the data according to the project code, and finally automatically summarizing and outputting the collected data according to the project structure, wherein the levels include construction projects, single projects, sub-items, lists, quotas, consumption, and labor, material and machinery data.

[0020] The method further comprises:

[0021] Step F: First, input the actual data of each sub-project of different projects into the big data sharing platform. Then, the big data sharing platform will compare the displayed result data with the actual data, and analyze the reasons for the large differences between the two, and use them as influencing factors. Then, input the determined influencing factors into the neural network prediction model for optimization.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. A method for analyzing and displaying engineering cost data of the present invention first collects data of completed projects in various regions at different stages and levels and then inputs them into a constructed big data sharing platform. The big data sharing platform then classifies and integrates the above data to form a data set, and then inputs the data of the total project and each sub-project of the project to be constructed into the big data sharing platform. The big data sharing platform then matches completed projects of the same type according to the type of the project to be constructed, and then uses the matched data of the completed project as a sample for training to form a neural network prediction model, and inputs the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction result of the project to be constructed. Finally, after obtaining the user's query request command and screening conditions, the prediction results are retrieved, and the results that meet the user's query requirements are screened out, the screening results are aggregated and corresponding charts are generated for display. The method realizes the collection and sharing of the data of completed projects in various regions through the big data sharing platform, matches the data of the project to be constructed with the completed projects in the big data sharing platform, selects completed projects of similar types to train and test the neural network model, so as to ensure the accuracy of the engineering cost of the neural network model, and sorts out the data of different stages, different levels, total projects, and sub-projects respectively, so as to realize the in-depth display of data. Therefore, the present invention not only ensures the accuracy of cost prediction, but also realizes the in-depth display of data.

[0024] 2. In a method for analyzing and displaying engineering cost data of the present invention, the method adopted for classifying and integrating the completed project data is as follows: first, the terrain data in the completed project data is divided into five terrain data sets, namely, mountain, plateau, plain, hill and basin, according to the geographical location, and the range values ​​of the five terrain data sets are encoded, and then the standard coding value range of all data in each terrain data set is determined according to the classification standard of the terrain data and the range value of each terrain data set, and then the other data in the completed project data is divided into a general data set, a design data set, a process processing data set and a management information data set, and the values ​​of the general data set, the design data set, the process processing data set and the management information data set are encoded. The method classifies the completed project data according to the geographical location obtained. On the one hand, the method can not only effectively avoid the changes in construction technology, construction difficulty and construction progress caused by the terrain, so that the final analysis result is greatly different from the actual situation, but also effectively promotes the final cost prediction result to be more in line with the actual situation, more accurate and reliable, and facilitates the retrieval of relevant data in the subsequent use process. On the other hand, the method effectively saves the space required for storage by encoding the data. Therefore, the present invention not only further improves the accuracy of the prediction results, but also facilitates the subsequent retrieval of relevant data and saves storage space. DETAILED DESCRIPTION

[0025] The present invention will be further described below in conjunction with specific implementation modes.

[0026] A method for analyzing and displaying engineering cost data comprises the following steps in sequence:

[0027] Step A: First collect the data of completed projects in various regions at different stages and levels and input them into the constructed big data sharing platform, which then classifies and integrates the above data to form a data set;

[0028] Step B: first input the data of the total project and each sub-project of the project to be constructed into the big data sharing platform, and then the big data sharing platform matches the completed projects of the same type according to the type of the project to be constructed;

[0029] Step C: using the matched data of completed projects as samples for training to form a neural network prediction model;

[0030] Step D: inputting the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction result of the project to be constructed;

[0031] Step E: After obtaining the user's query request command and screening conditions, retrieve the prediction results of step D, screen out the results that meet the user's query requirements, aggregate the screening results and generate corresponding charts for display.

[0032] In step A, the data of the completed project at different stages and levels include engineering cost data, project terrain proportions, meteorological conditions and construction technology solutions;

[0033] The method for forming a data set by classification and integration is as follows: first, the terrain data in the completed project data is divided into five terrain data sets, namely, mountain, plateau, plain, hill and basin, according to the geographical location, and the range values ​​of the five terrain data sets are encoded, and then the standard coding value range of all data in each terrain data set is determined according to the classification standard of the terrain data and the range value of each terrain data set, and then the other data in the completed project data is divided into a general data set, a design data set, a process processing data set and a management information data set, and the values ​​of the general data set, the design data set, the process processing data set and the management information data set are encoded to complete the classification and integration of the data.

[0034] The step B first identifies and encodes the data of the project to be constructed, and then matches it with the coding values ​​of various terrain data sets, general data sets, design data sets, process processing data sets and management information data sets based on fuzzy mathematical theory to determine completed projects of the same type.

[0035] In step C, the neural network prediction model is:

[0036] p(Y′|Y,α * ,σ *2 )=∫p(Y′|w,σ *2 )p(w|Y,α * ,σ *2 )dw

[0037] In the above formula, p(Y′|Y,α * ,σ *2 ) is the likelihood function, Y′ is the input sample at the next moment, Y is the input sample at the current moment, α * is the hyperparameter that determines the prior distribution of node weight w, σ *2 To optimize the parameters.

[0038] The step B also includes automatic inspection of the construction project data, specifically:

[0039] First, the maximum and minimum values ​​of each data in the completed projects of the same type are statistically matched and used as the reasonable range of the data. Then, the data of the project to be constructed are compared with its corresponding reasonable range. If the data of the project to be constructed is within the reasonable range, it is input into the neural network prediction model as valid data for prediction calculation. If the data of the project to be constructed is not within the reasonable range, a warning will be issued.

[0040] In step E, the process of aggregating and filtering results includes: first classifying the filtered data into levels, then aggregating the data according to the project code, and finally automatically summarizing and outputting the collected data according to the project structure, wherein the levels include construction projects, single projects, sub-items, lists, quotas, consumption, and labor, material and machinery data.

[0041] The method further comprises:

[0042] Step F: First, input the actual data of each sub-project of different projects into the big data sharing platform. Then, the big data sharing platform will compare the displayed result data with the actual data, and analyze the reasons for the large differences between the two, and use them as influencing factors. Then, input the determined influencing factors into the neural network prediction model for optimization.

[0043] The construction process of the neural network prediction model of the present invention is as follows:

[0044] First, establish the following neural network initial model:

[0045]

[0046] In the above formula, x and y are the values ​​of the input sample before and after normalization, respectively. min 、x max are the minimum and maximum values ​​of the collected samples respectively;

[0047] The Bayesian theorem is used to optimize the above model, and the cost prediction values ​​are expressed by variables as follows:

[0048] y i =y(x i ,w)+α i +σ 2

[0049] In the above formula, y(x i ,w) is the relationship between the weight of the input sample value before normalization and the value of the input sample after normalization of the i-th node, y i The normalized value of the input sample for the i-th node, x i is the value of the input sample before normalization of the i-th node, w is the node weight, α i is the noise of the ith node, σ 2 To optimize the parameters;

[0050] The likelihood function of the training sample is:

[0051]

[0052] In the above formula, m is the number of output nodes, y is the normalized value of the input sample, is a random value;

[0053] The distribution of node weight w is defined as normal distribution, as follows:

[0054]

[0055] In the above formula, α is the hyperparameter that determines the prior distribution of the node weight w, and n is the number of input nodes;

[0056] Based on the Bayesian theorem, the posterior distribution of the node weight w is obtained as follows:

[0057]

[0058] It can be seen that the posterior distribution also conforms to the following Gaussian distribution:

[0059] p(w|Y,α,σ 2 )=N(u,∑)

[0060] In the above formula, u is the mean of all nodes;

[0061] Through repeated iterations, we can obtain:

[0062]

[0063]

[0064] In the above formula, are α before and after iteration respectively. i ,∑ii is the sum of the values ​​of the i nodes, u i is the mean value of the i-th node;

[0065] The final optimized neural network prediction model is:

[0066] p(Y′|Y,α * ,σ *2 )=∫p(Y′|w,σ *2 )p(w|Y,α * ,σ *2 )dw

[0067] In the above formula, Y′ is the input sample at the next moment, Y is the input sample at the current moment, and α * is the hyperparameter that determines the prior distribution of node weight w, σ *2 To optimize the parameters.

[0068] Embodiment 1:

[0069] A method for analyzing and displaying engineering cost data is performed in the following steps:

[0070] 1. First, collect data of completed projects in different stages and at different levels in various regions, and then input them into the constructed big data sharing platform. The data of completed projects at different stages and at different levels include engineering cost data, project terrain proportions, meteorological conditions and construction technical solutions;

[0071] 2. The big data sharing platform classifies and integrates the completed project data into a data set according to the following methods:

[0072] First, the terrain data in the completed project data are divided into five terrain data sets according to the geographical location: mountain, plateau, plain, hill and basin, and the range values ​​of these five terrain data sets are coded. Then, according to the classification standard of terrain data and the range value of each terrain data set, the standard coding value range of all data in each terrain data set is determined. Then, the other data in the completed project data are divided into general data sets, design data sets, process processing data sets and management information data sets, and the values ​​of the general data sets, design data sets, process processing data sets and management information data sets are coded to complete the classification and integration of data;

[0073] 3. First collect the data of the total project and each sub-project of the project to be constructed and input them into the big data sharing platform, then identify and encode the data of the project to be constructed, and then match it with the encoding values ​​of the terrain data set, general data set, design data set, process processing data set and management information data set based on fuzzy mathematical theory to determine the completed projects of the same type, wherein the matching steps of the fuzzy mathematical theory are: first extract the features of the data of the completed projects to obtain feature data with statistical significance, then construct a fuzzy parameter matrix based on the extracted feature data, and then match the completed project data that meets the features according to the features of the data of the project to be constructed and the fuzzy parameter matrix;

[0074] 4. First, statistically match the maximum and minimum values ​​of each data in the completed projects of the same type, and use them as the reasonable range of the data. Then compare the data of the project to be constructed with its corresponding reasonable range. If the data of the project to be constructed is within the reasonable range, it will be input into the neural network prediction model as valid data for prediction calculation. If the data of the project to be constructed is not within the reasonable range, it will be warned.

[0075] 5. Use the matched data of completed projects as samples to train the following neural network prediction model:

[0076] p(Y′|Y,α * ,σ *2 )=∫p(Y′|w,σ *2 )p(w|Y,α * ,σ *2 )dw

[0077] In the above formula, p(Y′|Y,α * ,σ *2 ) is the likelihood function, Y′ is the input sample at the next moment, Y is the input sample at the current moment, α * is the hyperparameter that determines the prior distribution of node weight w, σ *2 To optimize the parameters;

[0078] 6. Input the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction result of the project to be constructed;

[0079] 7. After obtaining the user's query request command and filtering conditions, retrieve the project cost forecast results, filter out the results that meet the user's query requirements, and then classify the filtered data into levels, and then group the data according to the project code. Finally, automatically summarize and output the grouped data according to the project structure, and display it in the form of a chart, where the levels include construction projects, single projects, sub-items, lists, quotas, consumption, and labor, material and machinery data.

[0080] Embodiment 2:

[0081] The difference from Example 1 is that:

[0082] The method further comprises:

[0083] 8. Input the actual data of each sub-project of different projects into the big data sharing platform. Then the big data sharing platform will compare the displayed result data with the actual data, analyze the reasons for the large differences between the two, and use them as influencing factors. Then input the determined influencing factors into the neural network prediction model for optimization.

Claims

1. A method for analyzing and displaying engineering cost data, characterized in that: The method comprises the following steps in sequence: Step A: First collect the data of completed projects in various regions at different stages and levels and input them into the constructed big data sharing platform, which then classifies and integrates the above data to form a data set, wherein the data of completed projects at different stages and levels include engineering cost data, project terrain ratio, meteorological conditions and construction technology solutions; The method for forming a data set by classification and integration is as follows: first, for the terrain data in the completed project data, the terrain data are divided into five types of terrain data sets, namely, mountain, plateau, plain, hill and basin, according to the geographical location, and the range values ​​of the five terrain data sets are encoded, and then the standard coding value range of all data in each terrain data set is determined according to the classification standard of the terrain data and the range value of each terrain data set, and then the other data in the completed project data are divided into a general data set, a design data set, a process processing data set and a management information data set, and the values ​​of the general data set, the design data set, the process processing data set and the management information data set are encoded to complete the classification and integration of the data; Step B: first input the data of the total project and each sub-project of the project to be constructed into the big data sharing platform, and then the big data sharing platform matches the completed projects of the same type according to the type of the project to be constructed; Step C: using the matched data of completed projects as samples for training to form a neural network prediction model, wherein the construction process of the neural network prediction model is as follows: First, establish the following neural network initial model: In the above formula, x and y are the values ​​of the input sample before and after normalization, respectively. min 、x max are the minimum and maximum values ​​of the collected samples respectively; The Bayesian theorem is used to optimize the above model, and the cost prediction values ​​are expressed by variables as follows: y i =y(x i ,w)+a i +s 2 In the above formula, y(x i ,w) is the relationship between the weight of the input sample value before normalization and the value of the input sample after normalization of the i-th node, y i The normalized value of the input sample for the i-th node, x i is the value of the input sample before normalization of the i-th node, w is the node weight, α i is the noise of the ith node, σ 2 To optimize the parameters; The likelihood function of the training sample is: In the above formula, m is the number of output nodes, y is the normalized value of the input sample, is a random value; The distribution of node weight w is defined as normal distribution, as follows: In the above formula, α is the hyperparameter that determines the prior distribution of the node weight w, and n is the number of input nodes; Based on the Bayesian theorem, the posterior distribution of the node weight w is obtained as follows: It can be seen that the posterior distribution also conforms to the following Gaussian distribution: p(wY,α,σ 2 )=N(u,∑) In the above formula, u is the mean of all nodes; Through repeated iterations, we can obtain: In the above formula, are α before and after iteration respectively. i ,∑ii is the sum of the values ​​of the i nodes, u i is the mean value of the i-th node; The final optimized neural network prediction model is: p(Y′Y,α * ,s *2 )=∫p(Y′w,σ *2 )p(wY,α * ,s *2 )dw In the above formula, Y′ is the input sample at the next moment, Y is the input sample at the current moment, and α * is the hyperparameter that determines the prior distribution of node weight w, σ *2 To optimize the parameters; Step D: inputting the data of the project to be constructed into the neural network prediction model to obtain the engineering cost prediction result of the project to be constructed; Step E: After obtaining the user's query request command and screening conditions, retrieve the prediction results of step D, screen out the results that meet the user's query requirements, aggregate the screening results and generate corresponding charts for display.

2. The method for analyzing and displaying construction cost data according to claim 1, characterized in that: The step B first identifies and encodes the data of the project to be constructed, and then matches it with the coding values ​​of various terrain data sets, general data sets, design data sets, process processing data sets and management information data sets based on fuzzy mathematical theory to determine completed projects of the same type.

3. The method for analyzing and displaying construction cost data according to claim 1, characterized in that: The step B also includes automatic inspection of the construction project data, specifically: First, the maximum and minimum values ​​of each data in the completed projects of the same type are statistically matched and used as the reasonable range of the data. Then, the data of the project to be constructed are compared with its corresponding reasonable range. If the data of the project to be constructed is within the reasonable range, it is input into the neural network prediction model as valid data for prediction calculation. If the data of the project to be constructed is not within the reasonable range, a warning is issued.

4. The method for analyzing and displaying construction cost data according to claim 1, characterized in that: In step E, the process of aggregating and filtering results includes: first classifying the filtered data into levels, then aggregating the data according to the project code, and finally automatically summarizing and outputting the collected data according to the project structure, wherein the levels include construction projects, single projects, sub-items, lists, quotas, consumption, and labor, material and machinery data.

5. The method for analyzing and displaying engineering cost data according to claim 1, characterized in that: The method further comprises: Step F: First, input the actual data of each sub-project of different projects into the big data sharing platform. Then, the big data sharing platform will compare the displayed result data with the actual data, and analyze the reasons for the large differences between the two, and use them as influencing factors. Then, input the determined influencing factors into the neural network prediction model for optimization.

Images