A method, device and storage medium for estimating construction project cost
By obtaining and analyzing the internal and external feature data of the housing construction project, matching similar project engineering groups, and implementing indicator collection calculations and prediction and difference adjustment models, the problem of rapid and accurate prediction in the traditional housing construction project cost estimate method is solved, and the problem of rapid and accurate prediction in the traditional housing construction project cost estimate method is improved. The accuracy and efficiency of cost estimates achieve efficient management and investment control.
Patent Information
- Application Number
- CN202410088952.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-22
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-01-22
AI Technical Summary
It is difficult to conduct cost estimation methods for traditional housing construction projects quickly and accurately in the early stage of the project, especially when the design drawings are insufficient or the depth requirements are not met, it is difficult to achieve rapid estimation and review, and the cost indicator estimate efficiency is low. The data screening process is large and it is difficult to establish indicators.
By obtaining the internal and external characteristic engineering data of the project to be predicted, inputting the database analysis model, matching the project similar to the project, performing indicator collection and calculation, obtaining the meta-engineering quantity index data, and generating the cost prediction of the housing construction project through the prediction and difference adjustment model Estimate reports and use computers to quickly and accurately generate and predict project cost indicators.
Improve the accuracy and efficiency of cost estimates of housing construction projects, solve the problems of low standardization of engineering cost data, huge data screening processing volume, and difficult to establish indicators, realize rapid and accurate cost indicator generation, improve management efficiency and Investment control.
Smart Images

Figure CN117934035B_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present application relate to the field of computer technology, and in particular to a method, device and storage medium for estimating the cost of a building construction project. Background Art
[0002] The building construction project budget refers to the process of rationally planning and estimating the various resources required for the project before the construction project begins. It is also an important means of project cost control.
[0003] The results of traditional building construction project cost consulting rely on the consultant's experience summary based on past historical data, or an average of historical data. However, there is a demand for cost consulting companies to quickly estimate, review, and calculate when the design drawings and materials are insufficient or do not meet the depth requirements in the early consultation stage of the project. The traditional building construction project cost estimation method is difficult to meet the above needs.
[0004] Therefore, in the face of the growing and more complex demand for cost indicator estimation, it is necessary to change the traditional processing methods, further improve processing efficiency and ensure that valuable cost indicator data is obtained. Summary of the invention
[0005] The embodiment of the present application provides a method, device and storage medium for estimating the cost of a building construction project to solve the demand for estimating cost indicators. The technical solution is as follows:
[0006] On the one hand, a method for estimating the cost of a building construction project is provided, the method comprising:
[0007] Obtain internal and external feature engineering data of the project to be predicted;
[0008] Input the internal and external characteristic engineering data of the project to be predicted into the database analysis model, determine a similar project engineering group matching the internal and external characteristic engineering data of the project to be predicted, wherein the similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree, wherein each project engineering indicates the meta-project quantity indicator data contained in the meta-project data;
[0009] Execute the index collection calculation to obtain the meta-project quantity index data of each project engineering;
[0010] Combine and calculate the various meta-engineering data contained in the various project engineering and the corresponding project engineering level matching degree to obtain the initial data of various meta-engineering quantity indicators contained in the project engineering to be predicted;
[0011] Inputting each meta-engineering data contained in each project engineering and each meta-engineering quantity index initial data contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted;
[0012] A building construction project cost estimation report corresponding to the project to be predicted is generated according to the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project to be predicted and the initial data of each meta-engineering quantity index contained in the project to be predicted.
[0013] Optionally, before obtaining the internal and external feature engineering data of the project to be predicted, the method further includes:
[0014] Acquire historical engineering project data, and construct the database analysis model based on the historical engineering project data;
[0015] The meta-engineering data is constructed from the historical engineering project data according to the measurement and pricing rules.
[0016] Optionally, constructing the meta-engineering data from the historical engineering project data according to the measurement and pricing rules includes:
[0017] According to the measurement and pricing rules, each historical engineering project is divided into various meta-projects from the historical engineering project data;
[0018] The bill of quantities items belonging to the same meta-project are automatically aggregated to obtain the meta-project data corresponding to each meta-project.
[0019] Optionally, after inputting the internal and external feature engineering data of the project to be predicted into the database analysis model, the method further includes:
[0020] Obtaining the names of main internal and external feature items and corresponding values in the internal and external feature engineering data of the project to be predicted, wherein the names of the main internal and external feature items are obtained according to a preset selection;
[0021] According to the names of the main internal and external characteristic items and the corresponding values, a corresponding similar project engineering group is matched from the historical engineering project data and a collection calculation is performed;
[0022] After the calculation is completed, the meta-project data of each project in the similar project group is fed back.
[0023] Optionally, after inputting the internal and external feature engineering data of the project to be predicted into the database analysis model, the method further includes:
[0024] Adding a custom collection label to each meta-project in the similar project group;
[0025] According to the consistency requirements between the meta-project and its corresponding measurement unit, the correctness of the added custom collection label is judged and prompted;
[0026] The similar project engineering group is updated and exported.
[0027] Optionally, the method further includes:
[0028] Obtaining input project key feature information about the engineering data of the project to be predicted;
[0029] The step of inputting the internal and external characteristic engineering data of the project to be predicted into a database analysis model to determine a group of similar project engineering that matches the project to be predicted data includes:
[0030] Standardizing the key information of the project characteristics and inputting it into the database analysis model;
[0031] The database analysis model is used to match similar project engineering groups corresponding to the project feature key information.
[0032] Optionally, the step of inputting each meta-engineering data contained in each project engineering and each meta-engineering quantity index initial data contained in the project engineering to be predicted into the prediction difference adjustment model to obtain the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted includes:
[0033] Inputting the meta-project quantity index data of each project in the similar project engineering group and the initial data of each meta-project quantity index contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the meta-project level matching degree of each meta-project quantity index data contained in each project engineering in the similar project engineering group;
[0034] The matching degree of the meta-engineering level of each meta-engineering quantity index data contained in each project engineering in the similar project engineering group is averaged to obtain the engineering quantity index correction coefficient of each meta-engineering quantity data contained in the project engineering to be predicted.
[0035] Optionally, the engineering quantity index correction coefficient is recorded as Δw x , the calculation formula is:
[0036]
[0037] Among them, w ji (n) is the synaptic weight connecting neuron i to neuron j, Δw ji (n) is the synaptic weight w after N trainings ji (n) is the correction value, N is a positive integer, 1≤n≤N.
[0038] On the other hand, a device for estimating the cost of a building construction project is also provided, the device comprising:
[0039] The data acquisition module is used to obtain the internal and external feature engineering data of the project to be predicted;
[0040] A data input module, used for inputting the internal and external characteristic engineering data of the project to be predicted into a database analysis model, and determining a similar project engineering group matching the internal and external characteristic engineering data of the project to be predicted, wherein the similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree, wherein each project engineering indicates the meta-project quantity indicator data contained in the meta-project data;
[0041] A collection and calculation module, used to perform index collection and calculation to obtain the meta-project quantity index data of each project engineering;
[0042] A combined calculation module is used to perform combined calculation on each meta-engineering data contained in each project engineering and its corresponding project engineering level matching degree, so as to obtain initial data of each meta-engineering quantity indicator contained in the project engineering to be predicted;
[0043] A prediction and adjustment module is used to input the various meta-engineering data contained in the various project engineering and the initial data of the various meta-engineering quantity indicators contained in the project engineering to be predicted into the prediction and adjustment model to obtain the engineering quantity indicator correction coefficient corresponding to the various meta-engineering data contained in the project engineering to be predicted;
[0044] The cost estimation module is used to generate a building construction project cost estimation report corresponding to the project to be predicted based on the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project to be predicted and the initial data of each meta-engineering quantity index contained in the project to be predicted.
[0045] On the other hand, a server is provided, comprising a processor and a memory; the memory stores at least one instruction, and the at least one instruction is used to be executed by the processor to implement the method for estimating the cost of a building construction project as described in the above aspect.
[0046] On the other hand, a computer-readable storage medium is provided, wherein the storage medium stores at least one instruction, and the at least one instruction is used to be executed by a processor to implement the method for estimating the cost of a building construction project as described in the above aspect.
[0047] On the other hand, a computer program product is also provided, which stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the method for estimating the cost of building construction projects described in the above aspect.
[0048] The building construction project cost estimation method provided in this application can at least provide the following technical effects.
[0049] The method is mounted on a cost estimation platform at a server or terminal. When the method is used for prediction, the internal and external characteristic engineering data of the project to be predicted are obtained and input into a database analysis model to obtain a similar project engineering group that accurately matches it. The index aggregation calculation is performed to obtain the meta-engineering quantity index data of each project engineering. The meta-engineering data contained in each project engineering and the corresponding project engineering level matching degree are combined and calculated to obtain the initial data of each meta-engineering quantity index contained in the project engineering to be predicted. The meta-engineering quantity index data of each project engineering are combined and input into a prediction adjustment model to obtain the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted. The method can solve the problems of low standardization of engineering cost data, huge data screening and processing volume, and difficulty in establishing indicators in related technologies by accurately matching similar project engineering groups and analyzing and calculating meta-engineering data. Compared with relying on human experience, computers can be used to quickly and accurately generate predictions for project cost indicators, which is helpful to improve management efficiency and project investment control, and is also conducive to saving resource input and achieving social and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 A flow chart showing a method for estimating the cost of a building construction project provided by an exemplary embodiment of the present application is shown;
[0051] Figure 2 Shows the corresponding Figure 1 A schematic diagram of a building construction project cost estimation system;
[0052] Figure 3 An interface diagram showing the matching degree between several similar project works in the database and the project works to be predicted is shown;
[0053] Figure 4 A flow chart showing a method for estimating the cost of a building construction project provided by another exemplary embodiment of the present application is shown;
[0054] Figure 5 The induced local domain v generated at the activation function input of neuron j provided by an exemplary embodiment of the present application is shown. j (n) Schematic diagram;
[0055] Figure 6 shows the total instantaneous error E of the network provided by an exemplary embodiment of the present application n Schematic diagram of the resulting process;
[0056] Figure 7 The schematic diagram of the structure of three samples of the prediction adjustment model is shown;
[0057] Figure 8 A structural block diagram of a building construction project cost estimation device provided by an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION
[0058] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.
[0059] The term "multiple" as used herein refers to two or more than two. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the related objects are in an "or" relationship.
[0060] Based on the existing cost estimation problem raised by the background technology, the applicant analyzed that the accumulation and application of historical project data is one of the key factors in solving the problem. It is necessary to effectively utilize the historical project data and corresponding project information accumulated by the enterprise itself to form valuable cost indicator data.
[0061] At the same time, it is also an active response to the Ministry of Housing and Urban-Rural Development's project cost reform work plan. Accelerate the establishment of a database of project costs invested by state-owned funds, publish cost indexes for labor, materials, projects, etc. by region, project type, building structure, etc., use big data, artificial intelligence and other information technologies to provide a basis for the preparation of budget estimates, and encourage capable project cost consulting companies to establish and improve their project cost databases.
[0062] In addition, the design ideas disclosed in the housing construction project cost estimation method provided by this application can be widely used in scenarios where cost estimation is required, such as in other production fields. In order to facilitate the description of the following embodiments, each embodiment of this application is schematically described by taking the estimation of housing construction project cost indicators as an example, but it does not limit this application.
[0063] Please refer to Figure 1 , which shows a flow chart of a method for estimating the cost of a building construction project provided by an exemplary embodiment of the present application. In addition, Figure 2 Shows the corresponding Figure 1 A schematic diagram of a building construction project cost estimation system. Figure 1 The methods shown include:
[0064] Step 101, obtaining internal and external feature engineering data of the project to be predicted.
[0065] The internal and external characteristic engineering data of the project to be predicted are the internal and external characteristic engineering data of the project to be predicted corresponding to the current building construction project to be cost evaluated. In order to facilitate the input of the database analysis model, the internal and external characteristic engineering data of the project to be predicted can be processed data that matches the model input requirements, or it can be data files related to the internal and external characteristic engineering of the project to be predicted. The internal and external characteristic engineering data of the project to be predicted that matches the input database analysis model can be obtained through technical personnel sorting or electronic document sorting.
[0066] In one possible implementation, the model input includes various project characteristics, such as construction area, fortification intensity, structure type, foundation form, building height, base area, building area, number of above-ground building floors, etc. The output is the estimated cost of the project. The model is trained with historical data to ensure its accuracy on new projects. The database analysis model is a key component of the entire housing construction project cost estimation method. The following is an exemplary description of the construction, training and technical features of the database analysis model.
[0067] In the process of building a database analysis model, model selection is first performed. Select a database analysis model that is suitable for the problem. You can consider using a machine learning model, such as a regression model, to predict the cost of the project. Another option is to use a deep learning model, such as a neural network, to handle more complex data patterns. Further, feature selection is performed. Determine the features to be used to build the model. Features can include internal and external features of the project, such as construction area, fortification intensity, structure type, foundation form, building height, base area, building area, number of above-ground building floors, etc. Ensure that the selected features have an impact on the project cost and can provide sufficient information for accurate estimation. In addition, data preparation is performed. Collect and prepare a database for training. This may include detailed information on historical projects, including the quantity, unit price, total price, etc. of each meta-project. Ensure the quality of the data and perform necessary cleaning and preprocessing.
[0068] During the training of the database analysis model, the data is first segmented into training and test sets to ensure that the model fully learns the data patterns. Subsequently, the database analysis model is selected to train the training set so that it can understand the patterns and associations of historical engineering project data. Finally, the model is simply tuned, mainly by adjusting hyperparameters, to improve the generalization performance of the model for new projects. It should be noted that the current module only implements two functions: automatic data aggregation into meta-engineering data and project engineering-level matching calculation, which simplifies the implementation of model training and tuning. In actual situations, researchers can further optimize as needed to fully realize the potential of the database analysis model.
[0069] In one example, model training and tuning can be further refined, such as dividing the dataset into a training set and a test set. The training set is used to train the model, while the test set is used to evaluate the performance of the model. Typically, an 80-20 or 70-30 split can be used. Next, the model is trained, and the training set is trained using the selected database analysis model. During the training process, the model will learn patterns and associations in the data so that it can accurately predict new items later. Finally, the model is tuned: the hyperparameters of the model are adjusted based on the performance of the training set. This can be achieved through techniques such as cross-validation to ensure the generalization performance of the model.
[0070] In an embodiment of the present application, the database analysis model has the characteristics of a machine learning model, a deep learning model, and has certain interpretability and transparency, as well as generalization performance and other characteristics.
[0071] Machine learning models use algorithms such as linear regression, decision trees, and random forests. These models are generally suitable for problems with clear feature-target relationships, but may require more feature engineering data to improve accuracy. The above is an exemplary description of the construction, training, and features of the database analysis model in the embodiment of this application. The specific selection and implementation can be adjusted by the technical staff according to specific needs.
[0072] Step 102, inputting the internal and external characteristic engineering data of the project to be predicted into the database analysis model, and determining a similar project engineering group matching the internal and external characteristic engineering data of the project to be predicted.
[0073] The similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree. Among them, each project engineering indicates the meta-project quantity indicator data contained in the meta-project data.
[0074] First, based on the concept of engineering cost indicators and other related theoretical foundations, combined with the measurement and pricing rules, the smallest basic unit of engineering cost indicator application is established, which is recorded as meta-project, and the data of each level contained in the meta-project is recorded as meta-project data. In a possible implementation, when extracting meta-project data from the "Specifications for Calculation of Engineering Quantities for Housing Construction and Decoration Engineering", the following content explains how to map the engineering quantity calculation rules in the specifications to specific meta-project data items to ensure that the data extracted from the specifications matches the actual project.
[0075] Secondly, interpret the specification. Carefully read the "Specifications for Calculation of Construction Quantities for Housing Construction and Decoration Engineering" to understand the definitions, calculation rules, and application rules of measurement units for different levels of engineering quantities. Pay special attention to the relevant application regulations for each engineering quantity list item involved in the specification.
[0076] Next, define the meta-engineering. Determine the definition of the meta-engineering in the specification. The meta-engineering can be a type of engineering classification clearly defined in the specification, a basic building unit, or a type of common engineering component. Ensure that the definition of the meta-engineering is clearly understood.
[0077] Further mapping of engineering quantity calculation rules is performed. The engineering quantity calculation rules in the specification are mapped to specific meta-engineering data items. This may require the establishment of a rule mapping table to correspond the calculation rules of each bill of quantities item in the specification to the meta-engineering data item. Ensure the accuracy and comprehensiveness of the rule mapping. In this solution, mapping rules are established in combination with the existing provisions of the specification, and the first 9 digits of the 12-digit code corresponding to the minimum measurement unit bill of quantities item defined in the specification are mapped to the same meta-engineering data item.
[0078] The specific steps include identifying meta-engineering, developing a mapping table, considering detailed calculations, unit standardization, and verifying the mapping relationship. Identify clearly defined meta-engineering from the specification, involving the chapters or appendices in the specification, and classify and describe the meta-engineering; establish a mapping table that lists each engineering quantity calculation rule in the specification and their corresponding specific meta-engineering data items. The mapping table includes detailed information such as calculation formulas, units, and calculation rules in the specification; consider the details of engineering quantity calculation during the mapping process. For example, if the specification involves the calculation rules for a certain type of building material, ensure that the mapped meta-engineering data item can also accurately express the usage of the material; ensure that the mapped meta-engineering data items meet the corresponding unit standards. The specification usually stipulates the standard measurement units for various engineering quantities, and ensure that the mapped meta-engineering data items meet these standards; use actual project data for verification. Select some actual projects, apply the mapping relationship to calculate the meta-engineering data, and compare them with the engineering quantity data of the actual projects to ensure the accuracy and reliability of the mapping.
[0079] In combination with the existing provisions of the specification, mapping rules are established. One of the rules is to map the first 9 digits of the 12-digit code corresponding to the minimum measurement unit bill of quantities defined in the specification to the same meta-engineering data item. The mapping process is introduced through the following example.
[0080] Assume that the practical pricing document lists the following two minimum measurement unit bill of quantities: 1. There is a beam and slab, coded as 010505001004, with a project quantity of 103.55m 3 The amount is RMB 79563.3464; 2. The inclined slab with beams, code 010505001005, the project volume is 2.77m 3 The amount is 1745.4921 yuan.
[0081] In these two codes, the first 9 digits 010505001 are the same. According to the mapping rules, they will be mapped to the same meta-engineering data item. During the mapping process, the system integrates the relevant information of these two minimum measurement unit bill of quantities items into one meta-engineering data item.
[0082] The mapped meta-project data items include at least the following information: Meta-project name: beam-slab; coding range: 010505001 (the first 9 digits are the same); related measurement information: the project volume is 106.32m 3 The amount is 81308.8385 yuan.
[0083] Through such mapping rules, the system can effectively classify the minimum measurement unit bill of quantities items with matching degrees into the same meta-engineering data item, thereby providing a more organized data structure for subsequent analysis and prediction. This helps to simplify data management and improve efficiency. It should be noted that in order to ensure the effectiveness of the mapping relationship, it is necessary to regularly check the update of the specification to ensure that the mapping relationship is still applicable to the latest version of the specification. In addition, the mapping relationship can be adjusted according to the specific characteristics of the project. Ensure the flexibility of the mapping relationship to adapt to different types of housing construction projects.
[0084] In the application embodiment, the measurement and pricing rules include the current "Specifications for Calculation of Construction and Decoration Engineering Quantities" GB 50854-2013 and "Specifications for Pricing of Construction Engineering Quantities List" GB 50500-2013, etc., which are not limited here.
[0085] It should be noted that the concept of meta-project is not completely consistent with the concept of hierarchical project in the existing engineering cost field. The hierarchical project in the existing engineering cost field includes bill of quantities items, sub-projects, divisional projects, unit projects and single projects in order of grade from small to large.
[0086] Furthermore, an intelligent algorithm is used to automatically match and determine several similar project engineering groups that accurately match the project engineering data to be predicted from a large number of project engineering that have been entered into the database and completed information maintenance. In one example, the above intelligent algorithm selects a classification algorithm in machine learning, such as a support vector machine (SVM), and applies it to the project matching scenario. First, a support vector machine (SVM) is selected as the classification algorithm for project matching. SVM is suitable for high-dimensional data sets and can find the optimal hyperplane in the feature space to achieve data classification. Secondly, key features are extracted from the project engineering data, which may include engineering quantities, costs, engineering durations, etc. at various levels. Use principal component analysis (PCA) to reduce the dimensionality of features and reduce redundant information. The features are further standardized to ensure that they are on the same scale. This helps to improve the training effect of the algorithm.
[0087] In addition, training data preparation is required. Prepare historical project engineering data as training data sets, and annotate their matching degrees as training labels. Make sure that the data sets contain a combination of various project engineering features to improve the generalization ability of the algorithm. Use support vector machine to train the model on the training data. SVM will learn the relationship between historical project engineering features and matching degrees. Use the validation data set to evaluate the trained model and check its performance on unseen data. Evaluation indicators can include accuracy, recall, precision, etc. Perform hyperparameter tuning based on the evaluation results to further improve model performance. Apply the trained support vector machine model to real-time project matching. Input the feature data of the project to be predicted, and output the matching degree through the model. Collect actual matching results regularly and compare them with the model output. Based on the comparison results, update and improve the model to adapt to the new data distribution. Finally, deploy the trained and evaluated support vector machine model to the actual system to automate the project matching process.
[0088] In another example, the above intelligent algorithm selected a clustering algorithm in machine learning, such as the k-means algorithm, and applied it to the project matching scenario. Matching degree calculation is the core of cluster analysis. The matching degree calculation method varies according to the clustering object. The matching degree between individuals is measured by distance. Common distance calculation methods include Euclidean distance, absolute value distance, and Minkowski distance.
[0089] Considering the diversity of construction projects, it is necessary to process various types of data. In order to more accurately represent and process these data, different types of variables are introduced in the embodiments of the present application. First, in the training process of the database analysis model, historical engineering project data are collected and sorted. These data include rich information, some of which are interval variables, such as building height, assembly rate, etc. These interval variables help the model learn patterns and associations in the data during model training. On the other hand, it also involves binary variables, that is, variables with only two possible values. In project data, there are some features that are binary, such as whether the construction nature is new or rebuilt. By introducing binary variables, the impact of these binary features on the cost can be captured more accurately. In addition, categorical variables are also considered to describe the category or group of the project. For example, the green building grade can be regarded as a categorical variable, including one star, two stars, three stars, etc. It helps operators to conduct more detailed analysis of construction projects with different green building grades. Finally, the concept of ordinal variables is introduced, which is a variable that describes ordered categories. In housing construction projects, an example is the seismic grade of a building, which is an ordered categorical variable. Taking these ordered relationships into account, operators can better understand the impact of different seismic resistance levels on construction costs.
[0090] Therefore, by considering these different types of variables, operators can more fully understand the characteristics of project data and provide more accurate and valuable information for model training and prediction. The selection and use of these variables can better adapt to the characteristics and complexity of different projects. The following is an explanation of the specific calculation content of interval variables, binary variables, categorical variables, and ordinal variables.
[0091] 1. Interval variables. These are continuous variables, such as length, width, height, etc. Suppose there are two individual objects, a and b, and the n interval variables describing individual object a have values x a1 , x a2 , x a3 …, x an , describing the n interval variables of b individual objects with values x b1 , x b2 , x b3 …, x bn , the distance (difference) between individual objects a and b is d ab . Then the Euclidean distance formula between the two is shown in formula (Formula 1).
[0092]
[0093] 2. Two-state variables. It is a discrete variable, and its value is generally 0 and 1. The calculation method of the difference degree of two-state variables is different from that of interval variables. Usually, the values of two-state variables need to be counted before the difference degree calculation. Still taking the two individual objects a and b in the above text as an example, at this time, the objects a and b are described by n two-state variables. Suppose the value statistics of the two-state variables in a and b are shown in Table 1 below.
[0094] Table 1
[0095]
[0096] There are c cases when the binary variable in object a takes the value of 1 and the binary variable in object b takes the value of 0, d cases when the binary variable in object a takes the value of 1 and the binary variable in object b takes the value of 1, e cases when the binary variable in object a takes the value of 0 and the binary variable in object b takes the value of 1, and f cases when the binary variable in object a takes the value of 0 and the binary variable in object b takes the value of 0. The sum of c, d, e, and f is n, and the formula for calculating the difference between objects a and b is as follows. The difference between a and b is the ratio of the number of different cases of binary variables in a and b to the total number of binary variables.
[0097]
[0098] 3. Categorical variables. Categorical variables are also discrete variables, but they have 3 or more values. The idea of categorical variables is consistent with that of binary variables. Take two individual objects a and b as an example. The objects a and b are described by n categorical variables. Let c be the number of cases where the categorical variables in object a and object b have the same value. The difference calculation formula for objects a and b is shown in (Formula 3). In most cases, categorical variables can be converted into asymmetric binary variables and calculated using the difference calculation formula for binary variables.
[0099]
[0100] 4. Ordinal variables. It can be a discrete variable or a continuous variable. If the ordinal variable is a discrete variable, it is generally classified, but there is a certain order relationship between the categories. The calculation of the difference of the ordinal variable is to standardize the rank value of the variable, convert it into an interval variable, and use the calculation method of the interval variable difference. Take the kth variable of object a as an example. The formula for the standardized calculation of the rank value is as follows. Where r ak is the rank value of the kth variable of object a, R k Z is the maximum value of the level value divided by the kth variable of all objects. ak is the grade value after standardization, and the corresponding calculation formula is shown in Formula 4.
[0101]
[0102] In practical applications, individual objects usually contain several different attribute variables, and the values of each attribute variable need to be standardized so as to perform a unified difference calculation. In most cases, the importance of each attribute variable is also different, which is taken into consideration in the difference calculation. The idea of variable value standardization is to convert all categorical variables into binary variables and all ordinal variables into interval variables. If the attribute variables of individual objects have been standardized accordingly, the difference calculation between objects can be performed. Taking m objects as an example, after variable standardization, the m objects are only composed of interval variables and binary variables. At this time, the formula for calculating the difference between objects a and b in the m objects is shown in Formula 5, Formula 6 and Formula 7. The variable in object a is x a1 , x a2 , x a3 …, x an , the variable in object b is x b1 , x b2 , x b3 …, x bn , the number of variables is n, and the value of the kth variable of all objects is x 1k , x 2k ,x 3k …, x mk ,w k is the weight of the k variable. When the k variable is an asymmetric two-dimensional variable, and x ak =x bk =0, the variable is usually considered unimportant and thus has a weight of 0.
[0103]
[0104] In another possible implementation, the project engineering to be predicted is first divided into projects to obtain several meta-projects, and the above method is implemented by inputting the data of the project engineering to be predicted into a database analysis model to first obtain several divided meta-projects, and then determine a similar project engineering group that matches the project engineering to be predicted.
[0105] like Figure 3 As shown, Figure 3 An interface diagram representing the degree of match between several similar project projects in the database and the project projects to be predicted is shown. The interface diagram represents the degree of match between several similar project projects in the database and the project projects to be predicted. In an example, the value range is set as follows: if the calculated value is closer to 1, the greater the degree of match between the two project projects, and vice versa, and the projects are sorted by the degree of match.
[0106] pass Figure 3It is obtained that a method for estimating the cost of a building construction project in an embodiment of the present application corresponds to the following Figure 1 The building construction project cost estimation system and its software shown in the figure are implemented. Figure 3 In the unit engineering interface shown, the first column is the project column, the second column is the total cost, the third column is the project format, the fourth column is the engineering category, the fifth column is the professional classification, the sixth column is the building area, the seventh column is the above-ground building area, the eighth column is the number of above-ground floors, the ninth column is the standard floor height, the tenth column is the building height, the eleventh column is the base area, the twelfth column is the unit cost, and the last thirteenth column is the matching degree. The column names displayed in this interface are the various project features that can be included in the input of the above-mentioned model, but there is no limitation on this. The feature names can be modified according to the actual parameter requirements and data classification requirements.
[0107] Step 103, executing indicator aggregation calculation to obtain the meta-project quantity indicator data of each project.
[0108] In this step, data collection, data processing and meta-project quantity indicator calculation are involved. Data collection, collect the original data of each major indicator (such as steel bars, cement, commercial concrete, etc.). This may include information such as the quantity and cost used in each project; data processing, clean and pre-process the collected data as needed to ensure the consistency and accuracy of the data. Map the first 9 digits of the 12-digit code corresponding to the minimum measurement unit bill of quantities defined in the specification to the same meta-project data item. Further standardize according to the measurement unit of the indicator so that it can be unified in subsequent calculations; meta-project quantity indicator calculation, use the corresponding calculation method to calculate the meta-project quantity indicator of each project, obtain the specific values of various indicators such as steel bars and cement, and apply the corresponding measurement methods and formulas to process and calculate these data to obtain meta-project quantity indicator data.
[0109] Step 104, combining and calculating the various engineering quantity data contained in each project and its corresponding project level matching degree, and obtaining the initial data of various engineering quantity indicators contained in the project to be predicted.
[0110] In a possible implementation, step 104 includes the following content.
[0111] Content 1: Ensure the accuracy of combined calculations. Before combined calculations, perform data consistency checks to ensure that the data sources and calculation methods of each meta-project in the target meta-project group are consistent. If there are differences in the measurement units of the meta-project data, ensure that they are standardized before combined calculations to maintain data consistency. Detect and process outliers in the target meta-project group to avoid the impact of abnormal data on the combined calculation results; Content 2: Use the project engineering level matching algorithm. Consider using the project engineering level matching algorithm, and combine the hierarchical structure of the project engineering to calculate the matching of each project engineering in a similar project engineering group. When calculating the matching, consider the weight of the project engineering features, and make the matching calculation closer to the actual situation by correcting the weight; Content 3: Real-time feedback and verification of matching. During combined calculations, feedback the matching information of each project engineering in real time so that users can understand the specific situation of the matching. Establish a matching verification mechanism to ensure that the matching calculation meets expectations and avoid errors caused by calculation errors; Content 4: Data update strategy. Regularly check historical engineering project data and matching calculation algorithms, and update data and algorithms according to actual conditions to adapt to changes in business and market. When problems are found in the data or algorithm, an automatic correction mechanism is designed to make corrections without affecting the overall process; Content 5: Refine the combined calculation process. Consider dividing the combined calculation process into multiple stages and gradually combine meta-engineering data at different levels to reduce the calculation complexity and improve the calculation efficiency. Visualize the intermediate results of the combined calculation to facilitate users to monitor and analyze the calculation process and results.
[0112] Therefore, when calculating the initial data of the meta-project quantity indicators of each project in a similar project engineering group, the accuracy of the calculation and the consistency of the data are guaranteed. At the same time, the algorithm of project engineering level matching is adopted to feedback the matching information in real time, making the matching process more transparent and controllable.
[0113] Step 105, inputting each meta-engineering data contained in each project engineering and each meta-engineering quantity index initial data contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted.
[0114] In one possible implementation, the following scheme is given for the construction of the prediction difference adjustment model. First, select a model suitable for prediction difference adjustment. You can consider using a regression model, such as linear regression, support vector regression, or a more complex neural network model. The specific choice depends on the complexity of the data and the performance requirements of the model; second, determine the input characteristics of the model. In this scenario, the input of the model should include each meta-engineering data and the initial data of each meta-engineering quantity indicator contained in the project to be predicted. These data will be used to predict the engineering quantity indicator correction coefficient corresponding to each meta-engineering data contained in the project to be predicted; further, clearly define the output of the model. In this scenario, the output of the model is the engineering quantity indicator correction coefficient corresponding to each meta-engineering data; further, prepare a training data set, including each meta-engineering data, the initial data of each meta-engineering quantity indicator contained in the project to be predicted, and the corresponding actual correction coefficient. Ensure the diversity and coverage of the data set; further, use the training data to train the model. According to the selected model type, adjust the parameters of the model to minimize the error between the prediction result and the actual correction coefficient; further, use the validation data set to evaluate the trained model to check its performance on unseen data. Evaluation indicators may include mean square error (MSE), etc. Further, hyperparameters are tuned according to the evaluation results to improve model performance; further, the trained model is applied to actual prediction. Input the initial data of each element of the project to be predicted and each element of the engineering quantity index, and output the engineering quantity index correction coefficient through the model.
[0115] In addition, actual correction coefficients are collected regularly and compared with model outputs. Based on the comparison results, the model is updated and improved to adapt to the new data distribution. Finally, the trained and evaluated prediction and adjustment model is deployed to the actual system for real-time engineering quantity indicator correction.
[0116] Step 106, generating a building construction project cost estimation report corresponding to the project to be predicted based on the engineering quantity index correction coefficient corresponding to each meta-engineering data included in the project to be predicted and the initial data of each meta-engineering quantity index included in the project to be predicted.
[0117] In addition, this step is a multi-level, hierarchical pricing process. For each element project, multiple pricing is performed based on its revised engineering quantity index and corresponding price (element project comprehensive unit price). Then, the final forecast price (project engineering cost) is formed by summarizing the results level by level.
[0118] The report may include specific values of each major indicator, total cost and other information, which can be formatted and visualized as needed.
[0119] In a possible implementation, the following scheme for generating a building construction project cost estimate report is provided.
[0120] First, design the structure of the report to ensure that it can clearly display the revised engineering quantity indicators, comprehensive unit prices, price combination process and final project engineering cost of each element project. The structure of the report should include but not be limited to the following information.
[0121] Basic project information: project name, project number, construction location, unit cost, project business type, etc.
[0122] Detailed information of meta-projects: revised quantity indicators, comprehensive unit prices, etc. of each meta-project.
[0123] Pricing process: The pricing process for each meta-project, including multiple and hierarchical pricing.
[0124] Level-by-level summary: level-by-level summary of the price of each level of sub-project.
[0125] Final forecast price: The final forecast price of the entire project.
[0126] Secondly, regarding the report content, determine the specific information that needs to be included in the report. In this scenario, you can consider including the following report content.
[0127] Revised engineering quantity indicators of each sub-project: including quantity, measurement unit, revised value, etc.
[0128] Meta-project comprehensive unit price: the comprehensive unit price of each meta-project, reflecting price changes.
[0129] Price composition process records: record the multiple and hierarchical price composition processes of each meta-project for easy review and traceability.
[0130] Level-by-level summary information: level-by-level summary information of each level of meta-project price, including the total price of each level.
[0131] Final project cost: The final predicted price of the project, which may be the sum of all the component prices.
[0132] Furthermore, charts or other visualization methods can be used to enhance the readability of the report. For example, a bar chart can be used to represent the revised engineering quantity indicators of different element projects, and a line chart can be used to represent the changing trend of the pricing process. In addition, data quality prompts are added to the report to indicate the credibility and confidence level of the correction coefficient and the pricing process. This helps users understand the accuracy of the forecast results. Finally, a report export function is provided to allow users to save or share the generated report. Select common export formats, such as Excel and PDF.
[0133] For a more user-friendly design, you can also add user customization options, so that users can choose to display specific information or adjust the presentation of the report as needed, and add data tracing functions to the report, allowing users to trace the source of each meta-engineering data, the calculation method of the correction coefficient, and the detailed record of the price combination process. In addition, record the review history of the report, including review time, reviewer and other information, to ensure the traceability and transparency of the review.
[0134] In the embodiment of the present application, by clearly defining the structure and content of the report, adding visual elements to improve the user experience, and adding data quality prompts and traceability functions, it is ensured that the generated building construction project cost estimate report meets user needs and is reliable and operational.
[0135] In summary, the cost estimation method for building construction projects can be carried on a cost estimation platform at a server or terminal. When the method is used for prediction, the internal and external characteristic engineering data of the project to be predicted are obtained and input into a database analysis model to obtain a similar project engineering group that accurately matches it. The index aggregation calculation is performed to obtain the meta-engineering quantity index data of each project engineering. The meta-engineering data contained in each project engineering and the corresponding project engineering level matching degree are combined and calculated to obtain the initial data of each meta-engineering quantity index contained in the project engineering to be predicted. The meta-engineering quantity index data of each project engineering are combined and input into a prediction adjustment model to obtain the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted. By accurately matching similar project engineering groups and analyzing and calculating meta-engineering data, the problems of low standardization of engineering cost data, huge data screening and processing volume, and difficulty in establishing indicators in related technologies can be solved. Compared with relying on human experience, computers can be used to quickly and accurately generate predictions for project cost indicators, which is helpful to improve management efficiency and project investment control, and is also conducive to saving resource input and achieving social and economic benefits.
[0136] Please refer to Figure 4 , which shows a flow chart of a method for estimating the cost of a building construction project provided by another exemplary embodiment of the present application. The method includes:
[0137] Step 401, obtaining historical engineering project data, and building a database analysis model based on the historical engineering project data.
[0138] In the background technology, there is a need for cost consulting companies to quickly estimate, review, and calculate when the design drawings and materials are insufficient or do not meet the depth requirements in the early consultation stage of the project. Therefore, the accumulation and application of historical project data is particularly important. In order to solve the pain points of the above business scenarios, it is necessary to effectively use the historical project data and corresponding project information accumulated by the company itself to form valuable cost indicator data. Collect, analyze, and enter historical engineering project data.
[0139] Step 402, constructing meta-project data from historical project data according to measurement and pricing rules.
[0140] In a possible implementation, each historical engineering project is divided into different contents from historical engineering project data according to measurement and pricing rules to obtain meta-projects with different project contents.
[0141] Furthermore, the bills of quantities belonging to the same meta-project are automatically aggregated to obtain meta-project data corresponding to each meta-project, and the meta-project data is used to indicate meta-project quantity indicator data of the corresponding meta-project.
[0142] In a more detailed implementation, step 402 may include the following.
[0143] Content 1: Data extraction of historical engineering projects.
[0144] Consider the source of data and extract the required information from the accumulated historical project data, including bill of quantities, pricing rules, project characteristics, etc. Ensure the integrity of historical project data, including accurate quantity data, project characteristics, costs and other information.
[0145] Content 2: Mapping measurement and pricing rules to meta-engineering.
[0146] Design mapping rules to map the various contents of the measurement and pricing rules to the meta-projects to ensure that each meta-project can accurately reflect the corresponding measurement and pricing rules. Emphasize accuracy during the mapping process, consider the complexity and diversity of the rules, and ensure that the mapping results are consistent with the actual project situation.
[0147] Content 3: Meta-engineering division and collection.
[0148] According to the measurement and pricing rules, historical engineering projects are divided into different meta-projects to ensure the rationality of the division and the construction process of the actual project. For the bill of quantities items belonging to the same meta-project, an automated collection mechanism is introduced to ensure the efficiency and accuracy of the collection.
[0149] Content 4. Meta-engineering data indicates meta-engineering quantity indicators.
[0150] The collected meta-project data is associated with the meta-project quantity indicator data to ensure that each meta-project has corresponding quantitative indicators. The accuracy of the association is verified to ensure that the meta-project data can accurately indicate the meta-project quantity indicator data of the corresponding meta-project.
[0151] The above scheme ensures the accuracy and effectiveness of a series of steps, including extraction from historical engineering project data, mapping to meta-projects, and the division and automatic collection of meta-projects. It helps to build representative meta-projects and provides a reliable basis for subsequent engineering cost estimation.
[0152] Step 403, obtaining internal and external feature engineering data of the project to be predicted.
[0153] Please refer to step 101 for this step, and the embodiment of the present application will not be repeated here.
[0154] Step 404, input the internal and external characteristic engineering data of the project to be predicted into the database analysis model, and determine a similar project engineering group that matches the internal and external characteristic engineering data of the project to be predicted.
[0155] Please refer to step 102 for this step, and the embodiment of the present application will not be repeated here.
[0156] In a possible implementation manner, step 404 also includes the following content.
[0157] Obtain the names and corresponding values of the main internal and external feature items in the internal and external feature engineering data of the project to be predicted, and the names of the main internal and external feature items are obtained according to the preset selection; match the corresponding similar project engineering groups from the historical engineering project data according to the names and corresponding values of the main internal and external feature items and perform aggregation calculations; after the calculation is completed, feed back the meta-engineering data of each project engineering in the similar project engineering group.
[0158] Among them, the names of the main internal and external characteristic items are: steel bars, cement, commercial concrete, mortar, wood, gravel, blocks, bricks, doors, windows, insulation materials, waterproof membranes, waterproof coatings, prefabricated panels, etc.
[0159] In another possible implementation, step 404 also includes the following content.
[0160] Add custom collection labels to each meta-project in the similar project engineering group; judge and give prompts on the correctness of the added custom collection labels based on the consistency requirements between the meta-project and its corresponding measurement unit; update and export the similar project engineering group.
[0161] Among them, the custom collection labels can be steel bars, primary structure, secondary structure, PC components, formwork, masonry, ALC wall panels, floor, interior wall plastering, exterior wall plastering, ceiling plastering, waterproof membrane, etc.
[0162] Further implementation of this content includes adding custom collection labels to similar project engineering groups. According to preset needs, such as adding labels as "one structure", "two structures", etc., it is used to more accurately distinguish the characteristics of meta-engineering. According to the consistency requirements of meta-engineering and measurement elements, the correctness of the added custom collection labels is judged to ensure that the added labels meet the engineering pricing specifications and the requirements of the actual project. Export similar project engineering groups. Export the calculated and updated similar project engineering groups for use in subsequent steps. The exported similar project engineering groups need to maintain real-time to ensure that the data used in the prediction process is up-to-date and accurate.
[0163] Through the above scheme, the operations of adding and updating custom tags are considered to meet the needs of more flexible and accurate engineering feature matching.
[0164] Step 405, executing indicator aggregation calculation to obtain the meta-project quantity indicator data of each project.
[0165] Please refer to step 103 for this step, and the embodiment of the present application will not be repeated here.
[0166] Step 406, combining and calculating the various engineering quantity data contained in each project and its corresponding project level matching degree, and obtaining the initial data of various engineering quantity indicators contained in the project to be predicted.
[0167] Please refer to step 104 for this step, and the embodiment of the present application will not be repeated here.
[0168] Step 407, inputting each meta-engineering data contained in each project engineering and each meta-engineering quantity index initial data contained in the project engineering to be predicted into the prediction adjustment model, and obtaining the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted.
[0169] Please refer to step 105 for this step, and the embodiment of the present application will not be repeated here.
[0170] Step 408, generating a building construction project cost estimation report corresponding to the project to be predicted based on the engineering quantity index correction coefficient corresponding to each meta-engineering data included in the project to be predicted and the initial data of each meta-engineering quantity index included in the project to be predicted.
[0171] Please refer to step 106 for this step, and the embodiment of the present application will not be repeated here.
[0172] In summary, in the embodiment of the present application, the above content describes a technical process of estimating the cost of a building construction project, which involves multiple steps such as the collection of indicators, the calculation of matching degree, and the application of the prediction adjustment model. This method has the following technical effects:
[0173] By considering the meta-quantity data contained in the project and its corresponding project level matching, the prediction error model is used to correct the meta-quantity indicators, which is expected to improve the accuracy of the prediction. The model training process can be carried out according to the actual sample size, which helps to capture potential rules and trends.
[0174] Taking into account multiple main indicators, such as steel bars, cement, etc., makes the forecast more comprehensive. Through the combined calculation of indicators, the characteristics of the meta-engineering can be more fully understood, thereby improving the comprehensiveness of the forecast.
[0175] By using models such as BP artificial neural network and correcting model parameters through feedback, the model has the ability to learn and adapt. In this way, the model can be continuously optimized in subsequent predictions to improve the robustness of the prediction.
[0176] Before prediction, Z-score data standardization is performed to help eliminate the dimensional effects between different indicators and make the model more universal.
[0177] The weight influence of meta-engineering features is taken into account, that is, the influence of the weights of different feature factors on the matching value is considered. Through weight correction, the matching calculation is more in line with the actual situation and the matching accuracy is improved.
[0178] The final building construction project cost estimate report combines the revised meta-quantity indicators and provides a comprehensive view of the cost estimate, which is helpful for decision-making and planning.
[0179] The above embodiment provides an implementation method of the building construction project cost estimation method. Combined with the specific algorithm, the algorithm formulas involved in the main steps are described below.
[0180] In step 102, a similar project engineering group uses a clustering algorithm as an example for illustration, but this is not limited to this. The corresponding principle is introduced as follows.
[0181] K-means clustering method, given a sample set of similar project engineering groups Ω = {x1, x2, ..., x m}, first we need to determine the cluster center (mean vector) of each class {c1,c2,…,c k}, where c i It is cluster S i The mean vector of is calculated as:
[0182]
[0183] Among them, |S i | represents the number of samples in the i-th class. Then calculate each sample x i =(x i1 , x i2 ,…,xm ) to each cluster center c j =(c j1 ,c j2 ,…,c jn ) n (x it -c jt ), taking Euclidean distance as an example, the calculation formula is as follows:
[0184]
[0185] According to the calculated distance, the sample is divided into the category of the cluster center closest to it, and the cluster division S = {S1, S2, ..., S k The indicator used to evaluate the quality of the k-means algorithm classification results is the average error E, and its calculation formula is:
[0186]
[0187] in Represents sample x and mean vector c i The square of the L2 norm of the difference. The average error E measures the closeness of samples in the class around the cluster center. The smaller E, the higher the matching degree of samples in the cluster. Therefore, the smallest average error E corresponds to the best classification scheme.
[0188] The above formula is an optimization problem for different vector combinations. It is a non-deterministic problem with polynomial complexity and requires an iterative solution. The k-means algorithm uses an iterative method to optimize the problem. The specific process of the algorithm is as follows:
[0189] 1. Initialization: Given the number of classes k, randomly select k samples from the sample set as cluster centers.
[0190] Second, in the division stage, the distance between each sample and the k cluster centers is calculated one by one, and each sample is divided into the category to which the cluster center closest to it belongs, forming k classes.
[0191] Third, in the update phase, the cluster center of each class is recalculated, that is, the mean vector of all samples in each class is calculated as the new cluster center.
[0192] 4. Iteration end judgment: when the distance between the current cluster center of each class and the cluster center of the previous iteration is less than the set threshold, the iteration is terminated and the classification result is output; otherwise, repeat steps 2 to 4.
[0193] In the above algorithm flow, the initial cluster centers are k randomly selected samples. In practical applications, all samples can also be randomly assigned to k categories, and the centers of each cluster are calculated according to the randomly formed classes as the initial cluster centers.
[0194] When determining the number of clusters, there are situations where the initial number of clusters cannot be avoided. Through extensive reading and analysis of literature on practical problems, it is found that the initial number of clusters is often also the final number of clusters and the number of clusters is proportional to the number of samples. Therefore, researchers need to preset the number of clusters when performing cluster analysis.
[0195] Hierarchical clustering algorithm, the key to the aggregation clustering algorithm is the calculation of the distance between classes. The choice of calculation method directly affects the clustering results. Therefore, it is necessary to choose a suitable algorithm according to different problems. There are many methods for calculating the distance between classes. The commonly used ones are the minimum distance d min (C i ,C j ), maximum distance d max (C i ,C j ) and the average distance d avg (C i ,C j ).
[0196] Minimum distance d min (C i ,C j ), given two classes C i and C j , minimum distance d min (C i ,C j ) refers to the distance between the closest samples in two classes, that is:
[0197] d min (C i ,C j )=min{dist(x,z)x∈C i ,z∈C j}
[0198] Where dist is the generalized distance function label. The aggregation clustering algorithm that uses the minimum distance as the distance calculation method between classes can handle non-spherical clusters.
[0199] Maximum distance d max (C i ,C j ), maximum distance d max (C i ,C j ) refers to the distance between the farthest samples in two classes, that is:
[0200] d max (C i ,C j )=max{dist(x,z)x∈C i ,z∈C j}
[0201] The aggregation clustering algorithm that uses the maximum distance as the method for calculating the distance between classes is not easily affected by noise and outliers, but it is easy to destroy larger clusters and tends to form spherical clusters.
[0202] Average distance d avg (C i ,C j ), the average distance is the average of the distances between all samples in the two classes, that is:
[0203]
[0204] The agglomerative clustering algorithm that uses average distance as the distance calculation method between classes is also not easily affected by noise and outliers, but tends to form spherical clusters. In view of the type and distribution of sample data in this scheme, this clustering method is proposed for agglomerative clustering analysis.
[0205] In the process of completing the preliminary screening results of the similar project engineering group in step 102, the internal and external characteristic engineering data involved can be further screened to further process the screening of the similar project engineering group.
[0206] Since different characteristic factors of internal and external characteristic engineering data may have different influences on the estimated engineering quantity indicators in practice, the AHP hierarchical analysis method can be used to determine the weight of each characteristic factor item during clustering operation.
[0207] First, we build a hierarchical model, first determine the evaluation goal, then clarify the evaluation criteria, and then construct a hierarchical model with the goal, evaluation criteria and action plan. In the model, the goal, evaluation criteria and action plan are at different levels, and the relationship between them is represented by line segments.
[0208] Secondly, the factor pairwise comparison scale, after the hierarchical model is made, the evaluator, based on his own knowledge, experience and judgment, starts from the first criterion layer and gradually determines the importance weights of the factors in each layer relative to the factors in the previous layer. Using the pairwise comparison method, the importance is measured by the scale a. ij express.
[0209] Scale a ij When it is 1, it means that factor i is as important as factor j; scale a ij When it is 3, it is defined that factor i is slightly more important than factor j; scale a ij When it is 5, it means that factor i is more important than factor j; scale a ij When it is 7, it is defined that factor i is much more important than factor j; scale a ij When it is 9, it is defined that factor i is much more important than factor j; scale a ijWhen 2, 4, 6 and 8, the comparison result of the importance of factors i and j is defined to be in the middle of the above results; scale a ij When is the inverse, the comparison result of the importance of factors j and i is the inverse of the comparison result of the importance of factors i and j. Secondly, the judgment matrix used for the pairwise comparison method can specifically determine the importance weights of different factors at each level through scaling. Furthermore, the calculation of the eigenvector of the judgment matrix. After the judgment matrix is written out, the eigenvector of the judgment matrix needs to be calculated.
[0210] Take a 3×3 judgment matrix A as an example: the first step is to calculate the sum of each row of A; the second step is to calculate the average value of each row. Because A has 3 columns, divide by 3 when calculating the average value; the third step is standardization, that is, divide each row by the sum of three rows to obtain the eigenvector of A. Finally, for the consistency test of the judgment matrix, inconsistent judgments may occur when comparing two by two, so it is also necessary to check whether the judgments obtained by pairwise comparison in the judgment matrix A are consistent. The eigenvector calculated based on the judgment matrix with inconsistent judgments will lead people to wrong conclusions. Therefore, it is necessary to check the consistency of the judgment matrix. To this end, the consistency index CI is calculated, which is defined as:
[0211]
[0212] In the above formula, n is the order of the judgment matrix, λ max is the maximum eigenvalue of the judgment matrix. Then, check the randomness index CR from the table below and calculate the ratio CI / CR. When CI / CR < 0.1, it can be considered that the consistency of the judgment matrix has met the requirements. Otherwise, it is necessary to re-judge and write a new judgment matrix.
[0213] Calculating λ max The method is as follows.
[0214] The first step is to calculate the vector A w , A w = AW; Step 2, calculate λ max ,
[0215] The above W i is the corresponding judgment matrix (A w ) i Characteristic vector, in the initial stage, this scheme assumes that the weights of different characteristic factors are consistent. In the subsequent research stage, the specific weight values ω of different characteristic factors corresponding to different types of engineering project business types can be further determined.
[0216] Step 102 is explained in which the project engineering is arranged in order according to the project engineering level matching degree, and the matching value λ between the similar project engineering group x and the internal and external feature engineering data y of the predicted project engineering is calculated. xy(i.e., cosine angle method), the calculation formula is as follows, where a kx , a ky Refers to the kth data in two data samples (i.e. the kth internal and external feature engineering data of the project).
[0217]
[0218] Before the cost data is analyzed, the Z-score data is standardized and normalized. The calculation formula is as follows.
[0219] After considering the specific weights of different characteristic factors to evaluate the matching value of similar meta-projects, the matching value λ is corrected according to the weight value ω corresponding to each meta-project feature item. The corrected matching value λ' xy修 The calculation formula is as follows, where ω k Refers to the weight value corresponding to the kth internal and external feature item of the project.
[0220]
[0221]
[0222] In combination with the method of calculating construction project cost indicators in the "Standard for Classification and Measurement of Construction Project Cost Index" GB / T 51290-2018, in an example, the calculation of predicted engineering quantity / labor consumption / material consumption / machinery consumption indicators is taken as an example.
[0223] The calculation formula for the initial data of each engineering quantity index included in the project to be predicted in step 104 is:
[0224]
[0225] P x The initial data of the xth element engineering quantity indicator included in the project to be predicted;
[0226] S n The building area data for the nth similar project;
[0227] λ' 修n It is the matching degree correction value of each meta-project data contained in each project in the similar project group and its corresponding project level;
[0228] P n The engineering quantity indicator data of each element corresponding to the nth project in the similar project group in the database;
[0229] ^ is the ellipsis mark.
[0230] Through this formula, the meta-engineering data contained in each project engineering and its corresponding project engineering level matching degree can be combined and calculated to obtain the initial data of each meta-engineering quantity indicator contained in the project engineering to be predicted.
[0231] The prediction adjustment model involved in step 105 can also be compared with P x The actual value corrects the correlation coefficient of the prediction adjustment model according to the feedback result, using the BP artificial neural network algorithm. This application takes a two-layer neural network as an example, such as Figure 7 As shown, a corresponding specific example including three meta-project quantity indicators is illustrated, wherein R, S, and T represent different meta-project quantity indicators respectively. Figure 7 The two-layer neural network includes an input layer, a hidden layer and an output layer. The output layer corresponds to the initial data of each engineering quantity indicator contained in the project to be predicted. The number of neurons in the hidden layer is x, the number of neurons in the output layer is 1, and Figure 7 In this example, there are 3 samples, and there are 3 neurons in the input layer and hidden layer. The numbers in the circles represent the number. xy The value is assigned as the initial weight w of each neuron in the hidden layer O (n), introduce the engineering cost index to calculate the correction coefficient Δw, even if E n The gradient descent method is used for minimization, and the synaptic weight w obtained after training is ji Correction value Δw of (n) ji (n).
[0232] like Figure 5 As shown, the induced local domain v generated at the activation function input of neuron j j (n) is defined as follows.
[0233]
[0234] is the activation function, y i (n) is the input signal, and the function signal y at the output of neuron j j (n) is defined as follows.
[0235]
[0236] In the figure, y j (n) and d j (n) are the actual output and expected output of neuron j respectively. The error signal generated by the output of neuron j is defined as follows.
[0237] e j (n) = d j (n)-y j (n)(5-3)
[0238] Among them, d j (n) is the jth element of the expected response vector d(n). Define the instantaneous error E of neuron j as j (n) is defined as follows, where e j (n) is the error signal generated by the output of neuron j.
[0239]
[0240] Add up the errors of all output layer neurons to get the total instantaneous error E of the entire network n The definitions are as follows.
[0241]
[0242] The set C includes all neurons in the output layer. The BP algorithm repeatedly modifies the weights to make E n Minimize the synaptic weight w using the gradient descent method ji (n) Apply a correction value Δw ji (n), which is proportional to the partial derivative δE(n) / δw ji (n). According to the chain rule of differentiation, this gradient is expressed as follows.
[0243]
[0244] Partial derivative δE(n) / δw ji (n) represents a sensitive factor that determines w ji The search direction in weight space.
[0245] On both sides of equation (4-5), j (n) Taking the derivative, we get:
[0246]
[0247] In equation (4-3), both sides of y j (n) Taking the derivative, we get:
[0248]
[0249] On both sides of formula (4-2), j (n) Taking the derivative, we get:
[0250]
[0251] Finally, on both sides of equation (4-1) ji (n) Taking the derivative, we get:
[0252]
[0253] Substituting equations (4-7) to (4-10) into equation (4-6), we obtain:
[0254]
[0255] Applied to w ji Correction Δw of (n) ji (n) is defined as:
[0256]
[0257] Where η is the learning rate for error back propagation, and the negative sign indicates gradient descent in the weight space.
[0258] Substituting formula (5-11) into formula (5-12), we get:
[0259] Δw ji (n) = ηδ j (n)y i (n)(5-13)
[0260] Where δj(n) is the local gradient defined as:
[0261]
[0262] The above derivation process assumes that neuron j is in the output layer node. Now, since it is a two-layer neural network, neuron j is a hidden layer node. When neuron j is in the hidden layer, there is no specified expected response to the input neuron. The error signal of the hidden layer is determined recursively based on the error signals of all neurons directly connected to the hidden layer neurons. Considering neuron j as a hidden layer node, the local gradient δj(n) of the hidden layer neuron is redefined according to formula (5-14) as:
[0263]
[0264] like Figure 6 As shown in the figure, subscript j represents the hidden layer neurons, and subscript k represents the output layer neurons.
[0265] The total instantaneous error of the network E n for:
[0266]
[0267] On both sides of equation (5-17), the function signal y j (n) Taking partial derivative, we get:
[0268]
[0269] Among them, v k(n) is the induced local domain generated at the input of the activation function of neuron k (i.e., the input of neuron k), d k (n) is the expected output of neuron k, e k (n) is the error signal generated by the output of neuron k.
[0270] For the output layer neuron k, it induces a local domain:
[0271]
[0272] Solve equation (5-21) for y j Differentiating (n) yields:
[0273]
[0274] Substituting equations (5-20) and (5-22) into equation (5-18) yields:
[0275]
[0276] Substituting formula (5-23) into formula (5-16), we get the local gradient δj(n) of the hidden layer neuron j:
[0277]
[0278] Based on the above, by substituting equation (5-13) into equation (5-24), the modified value Δwji(n) of the synaptic weight connecting neuron i to neuron j is defined as follows:
[0279] Δw ji (n) = ηδ j (n)y i (n)(5-25)
[0280] The activation function uses the sigmoid function, as shown below; the learning rate is 0<η<1; training is based on the actual sample capacity (the number of training times is the number of meta-engineering of sample meta-engineering in the library, N, generally N≧50).
[0281]
[0282] In step 105, the calculation of the correction coefficient of the engineering quantity index is to take the calculation correction coefficient Δw of the xth most similar element engineering cost index x value and is the synaptic weight w after training N times ji Correction value Δw of (n) ji The average value of (n), that is:
[0283]
[0284] like Figure 7As shown in FIG. 1 , after three trainings, the correction coefficient is the average of the three times. In summary, the calculation formula for the predicted value of the corrected element engineering quantity / labor consumption / material consumption / machinery consumption in step 106 is:
[0285]
[0286] Among them, P n The engineering quantity indicator data of each element corresponding to the nth project in the similar project group in the database;
[0287] S x The building area data for the nth similar project;
[0288] Δw n It is the engineering quantity index correction coefficient corresponding to each element engineering data contained in the project to be predicted;
[0289] It is the matching degree correction value of each meta-project data contained in each project in the similar project group and its corresponding project level;
[0290] P 预测 Estimated cost report data for the meta-project quantity indicators of the xth meta-project of the project to be predicted.
[0291] Please refer to Figure 8 , which shows a structural block diagram of a building construction project cost estimation device provided by an exemplary embodiment of the present application, the device comprising:
[0292] The data acquisition module 801 is used to acquire the internal and external feature engineering data of the project to be predicted;
[0293] The data input module 802 is used to input the internal and external characteristic engineering data of the project to be predicted into the database analysis model, and determine a similar project engineering group that matches the internal and external characteristic engineering data of the project to be predicted, wherein the similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree, wherein each project engineering indicates the meta-project quantity indicator data contained in the meta-project data;
[0294] The aggregation calculation module 803 is used to perform index aggregation calculation to obtain the meta-project quantity index data of each project engineering;
[0295] The combined calculation module 804 is used to perform combined calculation on each meta-engineering data contained in each project engineering and its corresponding project engineering level matching degree, so as to obtain initial data of each meta-engineering quantity indicator contained in the project engineering to be predicted;
[0296] The prediction and adjustment module 805 is used to input the initial data of each meta-engineering data contained in each project engineering and each meta-engineering quantity index contained in the project engineering to be predicted into the prediction and adjustment model, and obtain the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted;
[0297] The cost estimation module 806 is used to generate a building construction project cost estimation report corresponding to the project to be predicted based on the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project to be predicted and the initial data of each meta-engineering quantity index contained in the project to be predicted.
[0298] Optionally, before the data acquisition module 801, the device further includes:
[0299] A first acquisition module, used to acquire historical engineering project data and construct the database analysis model according to the historical engineering project data;
[0300] The second acquisition module is used to construct the meta-engineering data from the historical engineering project data according to the measurement and pricing rules.
[0301] Optionally, the second acquisition module includes:
[0302] A first acquisition unit is used to divide the content of each historical engineering project from the historical engineering project data according to the measurement and pricing rules to obtain each meta-project;
[0303] The second acquisition unit is used to automatically collect the bill of quantities items belonging to the same meta-project to obtain the meta-project data corresponding to each meta-project.
[0304] Optionally, after the data input module 802, the device further includes:
[0305] Obtaining the names of main internal and external feature items and corresponding values in the internal and external feature engineering data of the project to be predicted, wherein the names of the main internal and external feature items are obtained according to a preset selection;
[0306] According to the names of the main internal and external characteristic items and the corresponding values, a corresponding similar project engineering group is matched from the historical engineering project data and a collection calculation is performed;
[0307] After the calculation is completed, the meta-project data of each project in the similar project group is fed back.
[0308] Optionally, after the data input module 802, the device further includes:
[0309] Adding a custom collection label to each meta-project in the similar project group;
[0310] According to the consistency requirements between the meta-project and its corresponding measurement unit, the correctness of the added custom collection label is judged and prompted;
[0311] The similar project engineering group is updated and exported.
[0312] Optionally, the device further comprises:
[0313] A first processing module is used to obtain input project key feature information about the engineering data of the project to be predicted;
[0314] The second processing module is used to input the internal and external characteristic engineering data of the project to be predicted into the database analysis model to determine a similar project group matching the project data to be predicted, including:
[0315] A third processing module is used to perform standardization processing on the key information of the project characteristics and input it into the database analysis model;
[0316] The fourth processing module is used to match the similar project engineering group corresponding to the project feature key information through the database analysis model.
[0317] Optionally, the prediction and adjustment module 805 includes:
[0318] The first adjustment unit is used to input the meta-project quantity index data contained in each project in the similar project engineering group and the initial data of each meta-project quantity index contained in the project engineering to be predicted into the prediction adjustment model to obtain the matching degree of the meta-project level of each meta-project quantity index data contained in each project engineering in the similar project engineering group;
[0319] The second adjustment unit is used to calculate the arithmetic average of the matching degrees of the meta-engineering level of each meta-engineering quantity index data contained in each project engineering in the similar project engineering group, and obtain the engineering quantity index correction coefficient of each meta-engineering data contained in the project engineering to be predicted.
[0320] Optionally, the engineering quantity index correction coefficient is recorded as Δw x , the calculation formula is:
[0321]
[0322] Among them, w ji (n) is the synaptic weight connecting neuron i to neuron j, Δw ji (n) is the synaptic weight w after N trainings ji (n) is the correction value, N is a positive integer, 1≤n≤N.
[0323] An embodiment of the present application also provides a computer-readable storage medium, which stores at least one instruction, at least one program, a code set or an instruction set. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the method for estimating the cost of a building construction project provided in the above-mentioned embodiments.
[0324] Optionally, the computer readable storage medium may include: a read-only memory (ROM), a random access memory (RAM), a solid state drive (SSD), or an optical disk, etc. Among them, the random access memory may include a resistance random access memory (ReRAM) and a dynamic random access memory (DRAM).
[0325] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.
[0326] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.
[0327] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. A method for estimating the cost of a building construction project, characterized in that: The method comprises: Obtain internal and external feature engineering data of the project to be predicted; Input the internal and external characteristic engineering data of the project to be predicted into the database analysis model, determine a similar project engineering group matching the internal and external characteristic engineering data of the project to be predicted, wherein the similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree, wherein each project engineering indicates the meta-project quantity indicator data contained in the meta-project data; Execute the index collection calculation to obtain the meta-project quantity index data of each project engineering; Combining and calculating the various meta-engineering data contained in the various project engineering and the corresponding project engineering level matching degree, and obtaining the initial data of various meta-engineering quantity indicators contained in the project engineering to be predicted; Inputting each meta-engineering data contained in each project engineering and each meta-engineering quantity index initial data contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted, including: Inputting the meta-project quantity index data of each project in the similar project engineering group and the initial data of each meta-project quantity index contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the meta-project level matching degree of each meta-project quantity index data contained in each project engineering in the similar project engineering group; The matching degree of the meta-project level of each meta-project quantity index data contained in each project engineering in the similar project engineering group is averaged to obtain the engineering quantity index correction coefficient of each meta-project data contained in the project engineering to be predicted, and the engineering quantity index correction coefficient is recorded as Δw x , the calculation formula is: Among them, w ji (n) is the synaptic weight connecting neuron i to neuron j, Δw ji (n) is the synaptic weight w after N trainings ji (n) is the correction value, N is a positive integer, 1≤n≤N; A building construction project cost estimation report corresponding to the project to be predicted is generated according to the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project to be predicted and the initial data of each meta-engineering quantity index contained in the project to be predicted.
2. The method for estimating the construction cost of a building project according to claim 1, characterized in that: Before obtaining the internal and external feature engineering data of the project to be predicted, the method further includes: Acquire historical engineering project data, and construct the database analysis model based on the historical engineering project data; The meta-engineering data is constructed from the historical engineering project data according to measurement and pricing rules.
3. The method for estimating the construction cost of a building according to claim 2, characterized in that: The step of constructing the meta-project data from the historical project data according to the measurement and pricing rules includes: According to the measurement and pricing rules, each historical engineering project is divided into various meta-projects from the historical engineering project data; The bill of quantities items belonging to the same meta-project are automatically aggregated to obtain the meta-project data corresponding to each meta-project.
4. The method for estimating the construction cost of a building project according to claim 1, characterized in that: After inputting the internal and external feature engineering data of the project to be predicted into the database analysis model, the method further includes: Obtaining the names of main internal and external feature items and corresponding values in the internal and external feature engineering data of the project to be predicted, wherein the names of the main internal and external feature items are obtained according to a preset selection; According to the names of the main internal and external characteristic items and the corresponding values, a corresponding similar project engineering group is matched from the historical engineering project data and a collection calculation is performed; After the calculation is completed, the meta-project data of each project in the similar project group is fed back.
5. The method for estimating the construction cost of a building according to claim 1, characterized in that: After inputting the internal and external feature engineering data of the project to be predicted into the database analysis model, the method further includes: Adding a custom collection label to each meta-project in the similar project group; According to the consistency requirements between the meta-project and its corresponding measurement unit, the correctness of the added custom collection label is judged and prompted; The similar project engineering group is updated and exported.
6. The method for estimating the construction cost of a building project according to claim 1, characterized in that: The method further comprises: Obtaining input project key feature information about the engineering data of the project to be predicted; The step of inputting the internal and external characteristic engineering data of the project to be predicted into a database analysis model to determine a group of similar project engineering that matches the project to be predicted data includes: Standardizing the key information of the project characteristics and inputting it into the database analysis model; The database analysis model is used to match similar project engineering groups corresponding to the project feature key information.
7. A device for estimating the cost of a building construction project, characterized in that: The device comprises: The data acquisition module is used to obtain the internal and external feature engineering data of the project to be predicted; A data input module, used for inputting the internal and external characteristic engineering data of the project to be predicted into a database analysis model, and determining a similar project engineering group matching the internal and external characteristic engineering data of the project to be predicted, wherein the similar project engineering group is composed of at least one matching project engineering and each project engineering is arranged in sequence according to the project engineering level matching degree, wherein each project engineering indicates the meta-project quantity indicator data contained in the meta-project data; A collection and calculation module, used to perform index collection and calculation to obtain the meta-project quantity index data of each project engineering; A combined calculation module is used to perform combined calculation on each meta-engineering data contained in each project engineering and its corresponding project engineering level matching degree, so as to obtain initial data of each meta-engineering quantity indicator contained in the project engineering to be predicted; The prediction and adjustment module is used to input the initial data of each meta-engineering data contained in each project engineering and each meta-engineering quantity index contained in the project engineering to be predicted into the prediction and adjustment model to obtain the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project engineering to be predicted, including: Inputting the meta-project quantity index data of each project in the similar project engineering group and the initial data of each meta-project quantity index contained in the project engineering to be predicted into the prediction difference adjustment model, and obtaining the meta-project level matching degree of each meta-project quantity index data contained in each project engineering in the similar project engineering group; The matching degree of the meta-project level of each meta-project quantity index data contained in each project engineering in the similar project engineering group is averaged to obtain the engineering quantity index correction coefficient of each meta-project data contained in the project engineering to be predicted, and the engineering quantity index correction coefficient is recorded as Δw x , the calculation formula is: Among them, w ji (n) is the synaptic weight connecting neuron i to neuron j, Δw ji (n) is the synaptic weight w after N trainings ji (n) is the correction value, N is a positive integer, 1≤n≤N; The cost estimation module is used to generate a building construction project cost estimation report corresponding to the project to be predicted based on the engineering quantity index correction coefficient corresponding to each meta-engineering data contained in the project to be predicted and the initial data of each meta-engineering quantity index contained in the project to be predicted.
8. A computer-readable storage medium, characterized in that: The storage medium stores at least one instruction, and the at least one instruction is used to be executed by the processor to implement the building construction project cost estimation method as described in any one of claims 1 to 6.
Citation Information
Cited By
Engineering cost control method based on BIM
CN121094910A
