An engineering cost data sensitivity index intelligent analysis method and system

Abstract

The present application relates to a kind of engineering cost data sensitivity index intelligent analysis method and system, belong to engineering cost intelligent analysis technical field, a kind of engineering cost data sensitivity index intelligent analysis method, comprising: collection historical project engineering data and constructs historical index library;Based on historical index library, constructs sensitivity analysis model;Using the model, the sensitivity analysis of historical project engineering data is carried out, and sensitive index is obtained;Based on sensitive index, the target value of new project limit is calculated;Based on sensitive index, the analysis of new project engineering data is carried out, and engineering optimization point is obtained;According to limit target value and engineering optimization point, generate intelligent analysis result.The problem that traditional engineering cost analysis is not accurate, and the rationality of cost prediction is insufficient is solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent analysis technology for engineering cost, and in particular to an intelligent analysis method and system for sensitivity indicators of engineering cost data. Background Technology

[0002] Construction cost management is a core component of the entire construction project management process, and its accuracy directly impacts project investment decisions and cost control effectiveness. With the rapid development of Building Information Modeling (BIM) technology and big data analytics, the ability to collect and store engineering data has significantly improved, and the massive amounts of engineering data accumulated from historical projects provide a rich data foundation for cost analysis. Currently, construction cost analysis primarily relies on traditional methods such as quota-based pricing and bill of quantities pricing, calculating and comparing cost indicators through manual experience or simple statistical analysis.

[0003] Existing engineering cost analysis methods lack the ability to deeply explore the complex nonlinear relationship between design parameters and cost indicators, making it difficult to automatically identify key factors affecting costs from massive historical data; insufficient utilization of historical project data makes it impossible to effectively screen historical cases similar to the current project for accurate benchmarking, resulting in large deviations in cost estimation; quota targets are mostly based on static experience values ​​or simple averages, lacking dynamic, data-driven scientific prediction mechanisms, making it difficult to adapt to the determination of reasonable ranges under different design conditions; engineering optimization directions rely on subjective judgment and lack quantitative optimization basis, resulting in weak targeting of cost control measures. Summary of the Invention

[0004] Based on the above analysis, the embodiments of the present invention aim to provide an intelligent analysis method and system for sensitivity indicators of engineering cost data, in order to solve the problems of inaccurate cost analysis and insufficient rationality of cost prediction in traditional engineering cost analysis.

[0005] On the one hand, embodiments of the present invention provide an intelligent analysis method for sensitivity indicators of engineering cost data, including the following steps: Collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects; A sensitivity analysis model was constructed based on a historical indicator database; Obtain the target cost index, and perform sensitivity analysis on the engineering data of the historical project based on the sensitivity analysis model to obtain the sensitivity index of the target cost index; Calculate the target limit value for new projects based on the aforementioned sensitive indicators; Based on the aforementioned sensitive indicators, the engineering data of the new project is analyzed to obtain the engineering optimization points of the new project; Based on the target limit value of the new project and the engineering optimization points of the new project, intelligent analysis results are generated.

[0006] Furthermore, the collection of engineering data from historical projects includes: Using individual building numbers from historical projects as the basic unit, we collect engineering data and design condition parameters for each building number; Based on the engineering data for each building number, calculate the cost index for each building number; A historical index database was built based on all cost indicators and design condition parameters.

[0007] Furthermore, the process of obtaining the target cost index involves performing sensitivity analysis on the engineering data of the historical projects based on the sensitivity analysis model, resulting in sensitive indicators for the target cost index, including: Based on the aforementioned sensitivity analysis model, the global sensitivity of the design condition parameters related to the target cost index is calculated respectively. Based on the design condition parameters and their corresponding building number's target cost index, calculate the local sensitivity of the design condition parameters related to the target cost index; Identify specific sensitivities for particular building types; Sensitivity indicators are determined based on the global sensitivity, the local sensitivity, and the specific sensitivity factors.

[0008] Furthermore, calculating the target limit value for new projects based on the aforementioned sensitive indicators includes: Obtain the design condition parameters for the target building number of the new project; Historical projects with a similarity to the design condition parameters of the target building number greater than a preset threshold are selected from the historical indicator database. For each sensitive indicator, the difference method is used to make independent predictions to obtain independent predicted values; A comprehensive forecast value is obtained based on the independent forecast values ​​of each sensitive indicator; The target limit value is obtained by combining the cost indicators corresponding to the selected historical projects with the comprehensive forecast value.

[0009] Furthermore, the target limit value obtained by combining the cost indicators corresponding to the selected historical projects and the comprehensive forecast value includes: Obtain the discrete distribution of cost indicators for each building in the selected historical projects; Calculate the standard deviation of the discrete distribution; Based on the comprehensive forecast value and the standard deviation, the lower limit and upper limit of the fluctuation range are obtained; The fluctuation range will be used as the target limit for the new project.

[0010] Furthermore, the analysis of the engineering data of the new project based on the aforementioned sensitive indicators to obtain the engineering optimization points of the new project includes: Based on the local sensitivity of each design condition parameter, the design condition parameters are sorted from largest to smallest. The design condition parameter with the highest local sensitivity in the sorting is determined as the engineering optimization point.

[0011] Furthermore, the calculation of the global sensitivity of the design condition parameters related to the target cost index based on the sensitivity analysis model includes: For each design condition parameter, random sampling is performed within its value range to generate multiple sets of different design condition parameter values; By inputting multiple sets of design condition parameter values ​​into the sensitivity analysis model, the corresponding cost index prediction values ​​are obtained. The variance-based Sobol index method is used to calculate the total influence index of each design condition parameter; The global sensitivity of the design condition parameter is obtained based on the total impact index.

[0012] Furthermore, the calculation of the local sensitivity of the design condition parameters related to the target cost index based on the design condition parameters and their corresponding building numbers includes: The design condition parameters related to the target cost index for each building number in the historical index database are input into the sensitivity analysis model to obtain the predicted value of the target cost index for each building number. For each building number, calculate the SHAP value of the predicted target cost index for that building number for each design condition parameter associated with the target cost index; The local sensitivity of a design condition parameter is obtained by averaging the absolute values ​​of all SHAP values ​​for a given building number.

[0013] Furthermore, based on the engineering data for each building number, the cost index for each building number is calculated, including: Calculate the total amount of concrete based on the total volume of concrete of all components within the building. Calculate the total amount of steel reinforcement based on the total weight of the steel reinforcement in all components within this building; Calculate the total amount of formwork based on the total contact area of ​​all components within the building. Calculate the weight ratio of reinforced concrete based on the total weight of steel bars and the total volume of concrete of all components in this building. Calculate the total amount of steel reinforcement in the frame columns based on the total weight of the steel reinforcement in all frame columns within the building. Calculate the total amount of raft foundation concrete based on the concrete volume of all raft foundations within this building.

[0014] On the other hand, embodiments of the present invention provide an intelligent analysis system for sensitivity indicators of engineering cost data, the system comprising: The data acquisition module is used to collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects. The module is used to build sensitivity analysis models based on a historical indicator library; The analysis module is used to obtain the target cost index, and to perform sensitivity analysis on the engineering data of the historical project based on the sensitivity analysis model to obtain the sensitivity index of the target cost index. The calculation module is used to calculate the target limit value for new projects based on the aforementioned sensitive indicators; The optimization module is used to analyze the engineering data of the new project based on the aforementioned sensitive indicators to obtain the engineering optimization points of the new project. The generation module is used to generate intelligent analysis results based on the target limit value of the new project and the engineering optimization points of the new project.

[0015] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects: 1. This invention constructs a historical index database and, based on a sensitivity analysis model, comprehensively calculates the global and local sensitivity of design condition parameters. At the same time, it identifies specific sensitive factors for specific building types, thereby generating comprehensive and accurate sensitivity indicators, which solves the problem of inaccurate cost analysis in existing technologies.

[0016] 2. This invention selects similar historical projects from a historical indicator database based on sensitive indicators, and identifies the most sensitive parameters that have the greatest impact on the target cost indicator. After independently predicting each most sensitive parameter using the difference method, the predicted value is obtained by combining the predictions. Then, the fluctuation range is calculated by combining the discrete distribution of the cost indicators of historical projects, and finally the limit target value of the new project is formed. This invention overcomes the limitations of existing technologies that rely on expert experience or simple analogy for limit prediction, and improves the rationality and credibility of cost prediction.

[0017] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0018] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Figure 1 is a flowchart illustrating an intelligent analysis method for sensitivity indicators of engineering cost data according to an embodiment of the present invention; Figure 2This is a schematic diagram of the structure of an intelligent analysis system for sensitivity indicators of engineering cost data, provided in an embodiment of the present invention. Detailed Implementation

[0019] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0020] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating an intelligent analysis method for sensitivity indicators of engineering cost data provided in an embodiment of the present invention. Figure 1 As shown, an intelligent analysis method for sensitivity indicators of engineering cost data includes: S1. Collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects; Engineering data is collected from the original design drawings, bills of quantities, and settlement documents of completed electronics industrial plant projects, and a historical index database is constructed based on the collected engineering data. Specifically, S1 includes: S11. Using a single building number from historical projects as the basic unit, collect engineering data and design condition parameters for each building number; Each historical project includes multiple building numbers with different functions, such as the main plant, power station, warehouse, and R&D building. The engineering data includes the total volume of concrete, the total weight of steel reinforcement, the total contact area of ​​formwork, the weight of steel structure, and the concrete and steel reinforcement usage for each component type, such as independent foundations, pile caps, raft slabs, foundation beams, frame columns, beam slabs, DECK slabs, waffle slabs, lattice beams, shear walls, parapet walls, etc.

[0021] Design parameters include structural form, seismic resistance rating, live load in the production area, grid dimensions, and story height. Structural form refers to the main type of load-bearing structure of the building, including frame structures, frame-shear wall structures, and steel structures. Seismic resistance rating is a design level classified according to national seismic codes based on factors such as the building's seismic fortification category, site category, and structural type. Common seismic resistance ratings are Level I, II, III, and IV, with Level I being the highest. Live load in the production area refers to the variable loads that may be applied to the floors or roofs of an industrial plant during its use, including equipment loads, personnel loads, and material loads. Common values ​​range from 5 kN / m² to 20 kN / m², for example, 5~8 kN / m² for light industrial workshops and 15~20 kN / m² for heavy machinery workshops. Grid dimensions refer to the axial spacing of the building's column or wall grid, including transverse and longitudinal axial spacing, such as 8m×8m, 9m×6m, etc., directly affecting beam and slab spans and the amount of structural materials used. Floor height refers to the vertical distance from the ground floor of a floor to the floor above it, affecting the height of structural columns and the material consumption of walls.

[0022] S12. Based on the engineering data for each building number, calculate the cost index for each building number, including: Calculate the total amount of concrete based on the total volume of concrete of all components within the building. Total concrete volume refers to the sum of the concrete volumes of all concrete components within a building, including foundations, columns, beams, slabs, and walls, expressed in cubic meters. The total concrete volume is typically related to design parameters such as structural type, grid dimensions, floor height, and number of floors. For example, the concrete usage per square meter is usually higher for frame structures than for steel structures; larger grid dimensions and larger beam / slab cross-sections result in higher concrete usage; and higher floor heights require more concrete for columns.

[0023] Calculate the total amount of steel reinforcement for each component based on the total weight of the steel reinforcement in all components within that building. The total amount of steel reinforcement in a structural member refers to the sum of the weights of all structural members within that building, including the steel reinforcement in concrete members and steel structural members, expressed in tons. The total amount of steel reinforcement is typically related to design parameters such as seismic resistance level, live load in the production area, grid dimensions, and structural form. For example, a higher seismic resistance level requires more stringent steel reinforcement measures and thus a larger amount of reinforcement; a greater live load in the production area necessitates more reinforcement in beams and slabs; and an increase in grid dimensions leads to a larger beam span and a significant increase in steel reinforcement usage.

[0024] Calculate the total amount of formwork based on the total contact area of ​​all components within the building. Formwork is used to shape concrete components during pouring, and its quantity directly affects the cost. The total formwork volume refers to the sum of the surface areas of all concrete components in contact with the formwork within a building, measured in square meters. The total formwork volume is related to design parameters such as grid dimensions, floor height, structural form, and component type. For example, the higher the floor height, the larger the column formwork area; the formwork area for beam-slab structures is usually larger than that for flat slab structures.

[0025] Calculate the weight ratio of reinforced concrete based on the total weight of steel bars and the total volume of concrete of all components in this building. The reinforced concrete weight ratio is an important indicator for evaluating the economic efficiency of structural design, representing the amount of steel reinforcement per cubic meter of concrete. Its calculation formula is as follows:

[0026] in, The weight ratio of reinforced concrete; This represents the total weight of the reinforcing steel bars in all components within that building. This represents the total concrete volume of all components within that building. The reinforced concrete weight ratio reflects the proportional relationship between the amount of steel reinforcement and concrete used, and is related to design parameters such as seismic resistance level, live load in the production area, and structural form. For example, buildings with high seismic resistance levels typically have a larger steel reinforcement ratio.

[0027] Calculate the total amount of steel reinforcement in the frame columns based on the total weight of the steel reinforcement in all frame columns within the building. Frame columns are the main load-bearing components that bear vertical loads. The total amount of steel reinforcement in frame columns refers to the sum of the steel reinforcement weights of all frame columns within that building. The total amount of steel reinforcement in frame columns is related to design parameters such as grid dimensions, story height, seismic resistance level, and live load in the production area. For example, a larger grid dimension and a larger column spacing may result in larger column cross-sections and reinforcement; a higher seismic resistance level requires more stirrups in the column end reinforcement zone.

[0028] Calculate the total amount of raft foundation concrete based on the concrete volume of all raft foundations within this building. A raft foundation is a type of large-area concrete foundation, often used in soft soil or where overall uplift resistance is required. The total concrete volume of the raft foundation refers to the sum of the concrete volumes of all raft foundations within a given building. The total concrete volume of the raft foundation is related to design parameters such as soil bearing capacity, groundwater level, superstructure load, and grid dimensions. For example, a thicker raft foundation is needed when the soil bearing capacity is low; and the raft foundation requires increased counterweight when the groundwater level is high and uplift resistance is required.

[0029] S13. Construct a historical index database based on all cost indicators and design condition parameters; A database of cost indicators (total concrete, total steel reinforcement in structural members, total formwork, reinforced concrete weight ratio, total steel reinforcement in frame columns, and total concrete in raft slabs) and design condition parameters for all building numbers is constructed. Each building number corresponds to one data record in the historical database. This record contains all design condition parameter fields and all cost indicator fields for that building number. The database is stored in a database file such as SQLite or CSV and indexed for fast retrieval.

[0030] S2. Construct a sensitivity analysis model based on a historical indicator database; Sensitivity analysis is a machine learning regression model used to express the nonlinear mapping relationship between design condition parameters and cost indicators. The input of the model is the design condition parameters, which are encoded into different values. The output of the model is the cost indicator, such as the total amount of steel reinforcement (tons) or the total amount of concrete (cubic meters).

[0031] During model construction, design condition parameters for each building number are extracted from a historical indicator database as input, and the corresponding cost indicators are output. An ensemble learning algorithm based on decision trees, such as Random Forest or XGBoost, can be used. During training, the data in the historical indicator database is randomly divided into a 70% training set, a 15% validation set, and a 15% test set. The model is initialized with default hyperparameters on the training set, and then optimized using the validation set. Adjusted parameters include the maximum depth of the decision tree (e.g., grid search from 3 to 10), learning rates (e.g., 0.01, 0.05, 0.1), and subsampling ratios (e.g., 0.6, 0.8, 1.0). The model uses mean squared error as the loss function, and early stopping can be employed to prevent overfitting. Model accuracy is evaluated on the test set, for example, by calculating the coefficient of determination and root mean square error.

[0032] Finally, the trained, optimized, and evaluated model is saved as a sensitivity analysis model. This model can predict the construction cost index of a building based on the design condition parameters corresponding to that building number. Furthermore, the sensitivity analysis model can also be implemented using other existing regression algorithms such as neural networks and support vector regression, which will not be elaborated upon here.

[0033] S3. Obtain the target cost index. Based on the sensitivity analysis model, perform sensitivity analysis on the engineering data of the historical project to obtain the sensitivity index of the target cost index. Sensitivity analysis is used to identify which design condition parameters have the greatest impact on cost indices. During implementation, a target cost index is first selected. The selection method can be determined based on actual analysis needs, such as a user-specified cost index, or an index with the greatest impact on the total project cost, such as the index that represents the largest proportion of structural steel reinforcement in the structural cost. Specifically, sensitivity analysis is conducted to obtain sensitive indices including: S31. Based on the aforementioned sensitivity analysis model, calculate the global sensitivity of the design condition parameters related to the target cost index respectively; Different cost indices are associated with different design condition parameters. For example, the total amount of steel reinforcement in a structural member is usually related to design condition parameters such as seismic resistance level, live load in the production area, grid dimensions, and structural form; the total amount of concrete in a raft foundation is related to foundation bearing capacity, groundwater level, superstructure load, and grid dimensions. Therefore, the design condition parameters related to the target cost index should first be determined based on engineering experience or user input, and the global sensitivity of each design condition parameter related to the target cost index should be calculated.

[0034] Global sensitivity measures the average impact of a single design condition parameter on the cost index output as it varies across its entire value range. Further, based on the design condition parameter and its corresponding building number's target cost index, the local sensitivity of the design condition parameter related to the target cost index is calculated, including: S311. For each design condition parameter, random sampling is performed within its value range to generate multiple sets of different design condition parameter values. The design condition parameters related to the target cost index are randomly sampled within their respective ranges. The range of values ​​for each design condition parameter is the interval between the minimum and maximum values ​​for all building numbers in the historical index database. For example, if the seismic resistance level is at least level 1 and at most level 3 in historical data, the range is [1,3]; if the live load in the production area is at least 5 kN / m² and at most 20 kN / m², the range is [5,20]. The Latin hypercube sampling function from the SciPy library in Python can be used to generate M sets of random values ​​for each design condition parameter between its historical minimum and maximum values. M is typically between 500 and 1000.

[0035] When determining the values ​​of design condition parameters, you can either independently and randomly sample each design condition parameter in the group within its value range each time, or you can change only one target design condition parameter in each group, that is, randomly sample the target design condition parameter within its value range, while fixing the other design condition parameters to the reference value, such as the median or average value.

[0036] S312. Input the values ​​of multiple sets of design condition parameters into the sensitivity analysis model to obtain the corresponding target cost index prediction values; Input multiple sets of design condition parameters into the trained sensitivity analysis model, output the predicted values ​​of the target cost index for each set, and record the parameter values ​​and their predicted values ​​for each set.

[0037] S313. Using the variance-based Sobol index method, calculate the total influence index of each design condition parameter, and obtain the global sensitivity of the design condition parameter based on the total influence index. The Sobol index method is a global sensitivity analysis method based on variance decomposition. It calculates the total variance of the predicted value of the target cost index output by the model and decomposes the total variance into the sum of the variances generated by the individual effects and interactions of each input parameter. It can be implemented in Python using `scipy.stats.qmc.LatinHypercube`. First, determine the number `d` of design condition parameters related to the current cost index and the value range of each parameter. Create a Latin hypercube sampler, for example, `sampler = qmc.LatinHypercube(d)`. Call `sampler.random(n=N)` to generate N sampling points. Each sampling point is a d-dimensional vector, with values ​​in the range [0,1]. Map each value to the actual parameter value to obtain matrix A.

[0038] Similarly, create a new Latin hypercube sampler independently, call sampler.random(n=N) to generate another set of N normalized sampling points, and map them to obtain matrix B. Matrix A and B are both of size N×d, and the number of samplings N is usually between 1000 and 5000.

[0039] For each design condition parameter i (i=1,…,d), construct a mixture matrix Ci, with the same size as A. Ci is obtained by replacing the i-th column of A with the i-th column of B. Input each row of matrices A, B, and Ci into the sensitivity analysis model to obtain the corresponding model output values ​​YA, YB, and YCi. The formula for the total influence index is as follows: Total Impact Index:

[0040] in, Let N be the estimated total influence index of the i-th design condition parameter, where N is the number of samplings; j represents the j-th sampling point, i.e., the j-th row. This represents the model output value corresponding to the j-th row of matrix A; This represents the model output value corresponding to the j-th row of matrix B; Let be the model output value corresponding to the j-th row of matrix Ci.

[0041] The total impact index is taken as the global sensitivity of the design condition parameters. The total impact index includes all the interaction effects of the design condition parameters and can more comprehensively reflect the importance of the parameters.

[0042] S32. Based on the design condition parameters and their corresponding building number's target cost index, calculate the local sensitivity of the design condition parameters related to the target cost index; local sensitivity measures the contribution of a certain design condition parameter to the predicted value of the cost index on a specific historical sample. Calculating the local sensitivity of the design condition parameters related to the target cost index includes: S321. Input the design condition parameters related to the target cost index for each building number in the historical index database into the sensitivity analysis model to obtain the predicted value of the target cost index for each building number. For the j-th building number in the historical indicator database, input the design condition parameters related to the target cost indicator into the sensitivity analysis model, and the model outputs the corresponding predicted cost indicator value.

[0043] S322. For each building number, calculate the SHAP value of the predicted target cost index for that building number for each design condition parameter related to the target cost index; Using the TreeExplainer or KernelExplainer from Python's shap package, the SHAP value is calculated for each target cost index. Specifically, for each building number j and each design condition parameter i related to the target cost index, the SHAP value of that design condition parameter on the predicted value of its associated target cost index is calculated. The SHAP value satisfies an additive interpretation, meaning that the predicted cost index value for each building number can be decomposed into the sum of the SHAP contributions of all parameters and the model's average predicted value, as shown in the following formula:

[0044] in, d represents the predicted target cost index for the j-th building; d is the total number of design condition parameters. The SHAP contribution value of the i-th design condition parameter to the j-th building number on the predicted value of the target cost index; This represents the average predicted value of the target cost index across all building numbers by the sensitivity analysis model.

[0045] S323. Take the absolute value of all SHAP values ​​for a building number and then average them to obtain the local sensitivity of the design condition parameter. For the design condition parameter i, its local sensitivity is as follows:

[0046] in, Let be the local sensitivity of the i-th design condition parameter to the target cost index; N is the total number of building numbers in the historical index database. Local sensitivity reflects the average absolute contribution of the design condition parameter to the cost index. At the same time, the direction of influence can be determined according to the positive and negative distribution of SHAP values. If most SHAP values ​​are positive, then the parameter is positively correlated with the cost index; if they are negative, then it is negatively correlated.

[0047] S323. Take the absolute value of all SHAP values ​​for a building number and then average them to obtain the local sensitivity of the design condition parameter. For design condition parameter i, its local sensitivity to the target cost index is as follows:

[0048] in, Let be the local sensitivity of the i-th design condition parameter to the target cost index; N is the total number of building numbers in the historical index database. Local sensitivity reflects the average absolute contribution of the design condition parameter to the cost index. At the same time, the direction of influence can be determined according to the positive and negative distribution of SHAP values. If most SHAP values ​​are positive, then the parameter is positively correlated with the cost index; if they are negative, then it is negatively correlated.

[0049] S33. Identify specific sensitive factors for a particular building type based on the local sensitivity of each design condition parameter; Specific sensitivity factors refer to design condition parameters that exhibit high sensitivity only for specific building types. For example, for Class A warehouses, the proportion of explosion-proof walls is a specific sensitivity factor; for power stations, the layout of large equipment on the roof and the floor where the water tank is located are specific sensitivity factors. The identification method involves grouping building numbers in the historical indicator database according to building type, such as main plant, power station, warehouse, R&D building, etc., calculating the local sensitivity of each design condition parameter within each group, and extracting the top 3 parameters as specific sensitivity factors for that building type.

[0050] S34. Determine the sensitivity index based on the global sensitivity, the local sensitivity, and the specific sensitivity factor; The global and local sensitivities of each design condition parameter related to the target cost index are normalized, for example, by using min-max normalization to map to the [0,1] interval. For parameters marked as specific sensitivity factors, their comprehensive score is multiplied by a preset amplification factor, such as 1.2. The comprehensive score of each design condition parameter is calculated as follows:

[0051] in, The i-th design condition parameter represents the comprehensive score of the target cost index. and These are weighting coefficients, for example, all taken as 0.5; The normalized global sensitivity; This represents the normalized local sensitivity. The overall scores are sorted from highest to lowest, and parameters exceeding a preset threshold, such as the top 30%, are identified as sensitive indicators.

[0052] S4. Calculate the target limit value for the new project based on the aforementioned sensitive indicators; The new project is an electronics industrial plant construction project that has not yet been constructed and is in the conceptual design or preliminary design stage. The target cost limit refers to the reasonable range within which key cost indicators should be controlled for this new project under given design conditions; it is usually expressed as a fluctuation range. The calculation of the target cost limit for the new project based on the aforementioned sensitive indicators includes: S41. Obtain the design condition parameters for the target building number of the new project; The design parameters for the target building number of the new project will be provided by the designers based on process requirements and specifications.

[0053] S42. Select historical projects from the historical index database whose design condition parameters are more similar to those of the target building number than a preset threshold. Similarity can be calculated using weighted Euclidean distance or cosine similarity. Taking cosine similarity as an example, the formula for calculating cosine similarity is:

[0054] in, Let be the similarity between the k-th historical project and the target building number; Design condition parameters for the target building number. This is the design condition parameter for the k-th historical item in the historical indicator database. The preset threshold is usually set to 0.85, which means that historical items with a similarity greater than 85% are selected.

[0055] S43. For each sensitive indicator, use the difference method to make independent predictions and obtain independent predicted values; The differential method is a linear extrapolation method based on historical benchmarks. It first calculates the arithmetic mean of the i-th sensitive indicator and the average of the target cost indicator. Then, it obtains the values ​​of the design condition parameters corresponding to the new project and calculates the independent predicted value as shown in the following formula.

[0056] in, Let be the independent predicted value of the i-th sensitive indicator; This represents the average target cost index across historical projects. For local sensitivity; The values ​​for this design condition parameter for the new project; This is the average value of the design condition parameter in historical projects.

[0057] S45. A comprehensive forecast value is obtained based on the independent forecast values ​​of each sensitive indicator; The comprehensive predicted value is shown in the following formula:

[0058] in, This represents the overall forecast value; N is the number of sensitive indicators. Let be the local sensitivity of the i-th sensitive index; Let be the independent predicted value of the i-th sensitive indicator.

[0059] S46. Combining the cost indicators corresponding to the selected historical projects with the comprehensive forecast value, the target limit value is obtained, including: S461. Obtain the discrete distribution of cost indicators for each building in the selected historical projects; If the selected historical projects have a total of K building numbers, and each building number corresponds to a target cost index value, then sort all the target cost index values ​​in ascending order to obtain a sequence. , ,..., .

[0060] S462. Calculate the standard deviation of the discrete distribution; The standard deviation is calculated as shown in the following formula;

[0061] in, denoted as standard deviation; K represents the total number of historical project building numbers selected. Let the target cost index value be the j-th building number; This represents the average of the target cost index. Alternatively, the upper and lower quartiles can be used instead of the standard deviation; the lower quartile Q1 can be the 25th quartile, and the upper quartile Q3 can be the 75th quartile.

[0062] S463. Based on the comprehensive forecast value and the standard deviation, the lower limit and upper limit of the fluctuation range are obtained, and the fluctuation range is used as the target limit value for the new project. The lower limit L and upper limit U of the fluctuation range are shown in the following formula;

[0063] in, This is a comprehensive forecast value; The standard deviation is denoted as .

[0064] If quartiles are used, the lower limit L and upper limit U of the fluctuation range are as follows;

[0065] Q1 and Q3 are the lower quartile and upper quartile, respectively.

[0066] The target limit is the range [L, U], which represents the reasonable cost range for a new project under the design conditions. Designers should control the final design results within this range.

[0067] S5. Analyze the engineering data of the new project to obtain the engineering optimization points of the new project; Engineering optimization points refer to the aspects of a new project where costs can be significantly reduced by adjusting design condition parameters. Specifically, obtaining engineering optimization points for a new project includes: sorting each design condition parameter from largest to smallest based on its local sensitivity; and identifying the design condition parameter with the highest local sensitivity in the sorted list as the engineering optimization point. Select the top-ranked design condition parameter, such as the live load of the production area or the grid dimensions, as the optimization point for the project. The top-ranked design condition parameter has the greatest impact on cost and should be optimized first. Specific adjustment suggestions can also be given based on the direction of this parameter's influence; for example, if the local sensitivity is positive, reducing this parameter can lower the cost.

[0068] S6. Generate intelligent analysis results based on the target limit value of the new project and the engineering optimization points of the new project; The intelligent analysis results are a comprehensive report that includes target limits and recommendations for engineering optimization. This report can be output in text, table, or chart format for use by designers and managers in decision-making.

[0069] like Figure 2 As shown in some embodiments, an intelligent analysis system for sensitivity indicators of engineering cost data is provided. The system includes: The data acquisition module 201 is used to collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects. Module 202 is used to build a sensitivity analysis model based on a historical indicator library; Analysis module 203 is used to obtain the target cost index, and to perform sensitivity analysis on the engineering data of the historical project based on the sensitivity analysis model to obtain the sensitivity index of the target cost index. Calculation module 204 is used to calculate the target limit value for the new project based on the sensitive indicators; The optimization module 205 is used to analyze the engineering data of the new project based on the sensitive indicators to obtain the engineering optimization points of the new project; The generation module 206 is used to generate intelligent analysis results based on the target limit value of the new project and the engineering optimization points of the new project.

[0070] It is understandable that the modules and references recorded in the intelligent analysis system for sensitivity indicators of project cost data are... Figure 1 The steps described in the intelligent analysis method for sensitivity indicators of engineering cost data correspond to each other. Therefore, the operations, characteristics, and beneficial effects described above for the intelligent analysis method for sensitivity indicators of engineering cost data also apply to the intelligent analysis system for sensitivity indicators of engineering cost data and its included modules, and will not be repeated here.

[0071] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0072] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent analysis of sensitivity indicators of engineering cost data, characterized in that, Includes the following steps: Collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects; A sensitivity analysis model was constructed based on a historical indicator database; Obtain the target cost index, and perform sensitivity analysis on the engineering data of the historical project based on the sensitivity analysis model to obtain the sensitivity index of the target cost index; Calculate the target limit value for new projects based on the aforementioned sensitive indicators; Based on the aforementioned sensitive indicators, the engineering data of the new project is analyzed to obtain the engineering optimization points of the new project; Based on the target limit value of the new project and the engineering optimization points of the new project, intelligent analysis results are generated.

2. The method according to claim 1, characterized in that, The collection of engineering data from historical projects includes: Using individual building numbers from historical projects as the basic unit, we collect engineering data and design condition parameters for each building number; Based on the engineering data for each building number, calculate the cost index for each building number; A historical index database was built based on all cost indicators and design condition parameters.

3. The method according to claim 2, characterized in that, The process of obtaining the target cost index involves performing sensitivity analysis on the engineering data of the historical projects based on the sensitivity analysis model. The resulting sensitivity indicators for the target cost index include: Based on the aforementioned sensitivity analysis model, the global sensitivity of the design condition parameters related to the target cost index is calculated respectively. Based on the design condition parameters and their corresponding building number's target cost index, calculate the local sensitivity of the design condition parameters related to the target cost index; Identify specific sensitivities for particular building types; Sensitivity indicators are determined based on the global sensitivity, the local sensitivity, and the specific sensitivity factors.

4. The method according to claim 1, characterized in that, The calculation of the target limit value for new projects based on the aforementioned sensitive indicators includes: Obtain the design condition parameters for the target building number of the new project; Historical projects with a similarity to the design condition parameters of the target building number greater than a preset threshold are selected from the historical indicator database. For each sensitive indicator, the difference method is used to make independent predictions to obtain independent predicted values; A comprehensive forecast value is obtained based on the independent forecast values ​​of each sensitive indicator; The target limit value is obtained by combining the cost indicators corresponding to the selected historical projects with the comprehensive forecast value.

5. The method according to claim 4, characterized in that, The target limit value obtained by combining the cost indicators corresponding to the screened historical projects and the comprehensive forecast value includes: Obtain the discrete distribution of cost indicators for each building in the selected historical projects; Calculate the standard deviation of the discrete distribution; Based on the comprehensive forecast value and the standard deviation, the lower limit and upper limit of the fluctuation range are obtained; The fluctuation range will be used as the target limit for the new project.

6. The method according to claim 3, characterized in that, The analysis of the engineering data of the new project based on the aforementioned sensitive indicators, resulting in the following engineering optimization points: Based on the local sensitivity of each design condition parameter, the design condition parameters are sorted from largest to smallest. The design condition parameter with the highest local sensitivity in the sorting is determined as the engineering optimization point.

7. The method according to claim 3, characterized in that, The calculation of the global sensitivity of design condition parameters related to the target cost index based on the sensitivity analysis model includes: For each design condition parameter, random sampling is performed within its value range to generate multiple sets of different design condition parameter values; By inputting multiple sets of design condition parameter values ​​into the sensitivity analysis model, the corresponding cost index prediction values ​​are obtained. The variance-based Sobol index method is used to calculate the total influence index of each design condition parameter; The global sensitivity of the design condition parameter is obtained based on the total impact index.

8. The method according to claim 3, characterized in that, The calculation of the local sensitivity of the design condition parameters related to the target cost index, based on the design condition parameters and their corresponding building numbers, includes: The design condition parameters related to the target cost index for each building number in the historical index database are input into the sensitivity analysis model to obtain the predicted value of the target cost index for each building number. For each building number, calculate the SHAP value of the predicted target cost index for that building number for each design condition parameter associated with the target cost index; The local sensitivity of a design condition parameter is obtained by averaging the absolute values ​​of all SHAP values ​​for a given building number.

9. The method according to claim 1, characterized in that, Based on the engineering data for each building number, the cost indicators for each building number are calculated as follows: Calculate the total amount of concrete based on the total volume of concrete of all components within the building. Calculate the total amount of steel reinforcement based on the total weight of the steel reinforcement in all components within this building; Calculate the total amount of formwork based on the total contact area of ​​all components within the building. Calculate the weight ratio of reinforced concrete based on the total weight of steel bars and the total volume of concrete of all components in this building. Calculate the total amount of steel reinforcement in the frame columns based on the total weight of the steel reinforcement in all frame columns within the building. Calculate the total amount of raft foundation concrete based on the concrete volume of all raft foundations within this building.

10. An intelligent analysis system for sensitivity indicators of engineering cost data, characterized in that, The systems included are: The data acquisition module is used to collect engineering data from historical projects and build a historical indicator database based on the engineering data from historical projects. The module is used to build sensitivity analysis models based on a historical indicator library; The analysis module is used to obtain the target cost index, and to perform sensitivity analysis on the engineering data of the historical project based on the sensitivity analysis model to obtain the sensitivity index of the target cost index. The calculation module is used to calculate the target limit value for new projects based on the aforementioned sensitive indicators; The optimization module is used to analyze the engineering data of the new project based on the aforementioned sensitive indicators to obtain the engineering optimization points of the new project. The generation module is used to generate intelligent analysis results based on the target limit value of the new project and the engineering optimization points of the new project.

Images