An engineering cost deviation identification method based on bill item time sequence evolution modeling

Abstract

This invention discloses a method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items. The method includes: extracting core parameters of temporal evolution through an intelligent bill of quantities cost analysis system; constructing an evolutionary model using a progressive deviation iteration algorithm; mining fluctuation correlation and tracing algorithms to uncover fluctuation relationships; generating evolutionary trend curves using an adaptive cost trend inference model; and fusing multiple models to construct a comprehensive identification model. Through phased data processing, parameter optimization, and multi-dimensional fusion calculations at each step, the method achieves accurate identification and tracing of cost deviations at each temporal node. This method is adapted to the temporal evolution characteristics of bill of quantities items. Through standardized data processing procedures, dynamic parameter adjustments, and multi-model collaboration, it improves the comprehensiveness, accuracy, and timeliness of deviation identification, providing a reliable technology for refined management of engineering costs. It is applicable to various scenarios involving cost deviation management throughout the entire lifecycle of construction projects.

Description

Technical Field

[0001] This invention relates to the field of engineering cost deviation identification technology, and in particular to an engineering cost deviation identification method based on the time-series evolution modeling of bill of quantities items. Background Technology

[0002] In the field of construction engineering, cost control of bill of quantities items runs through the entire project lifecycle. As project scale expands and implementation cycles lengthen, the dynamic changes in cost components become increasingly complex. Factors such as fluctuations in quantities over time and the interrelationships of cost elements make traditional static control models insufficient for accurate deviation identification. Changes in quantities, resource price fluctuations, and adjustments to construction procedures at each stage of project implementation can all cause cost deviations. If these deviations cannot be identified and traced in a timely manner, they will lead to cost overruns and schedule delays. Therefore, a systematic deviation identification solution adapted to the characteristics of time-series evolution is urgently needed. Currently, cost deviation identification in engineering projects mostly uses bill of quantities standard data as a benchmark, comparing the actual cost data with the standard value through manual verification or basic software assistance. Its core implementation logic relies on established quota standards and historical data to compare quantities and cost components one by one. Some technologies integrate single-dimensional data analysis methods to perform simple statistical processing on discrete cost data, thereby determining whether deviations exist. This approach is essentially a post-hoc verification based on static data, failing to take into account the sequential nature of project implementation and failing to form a systematic correlation analysis of factors affecting cost.

[0003] The existing technology has two key drawbacks: First, it lacks the ability to adapt to the temporal evolution characteristics of the bill of quantities items, focusing only on the static comparison of single-point data. It fails to capture the dynamic changes in engineering quantity and cost elements as the construction progresses, and cannot reflect the cumulative impact of the evolutionary characteristics of different stages on the formation of deviations, resulting in insufficient comprehensiveness in deviation identification. Second, the ability to trace the source of deviations is weak. It can only discover the surface differences in cost data, failing to clarify the internal correlation mechanism between various cost components. It is difficult to accurately locate the root cause of deviations, making subsequent corrective measures lack specificity and unable to avoid the risk of deviation expansion from the root, thus restricting the level of precision in cost control. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of existing technologies, this invention provides a method for identifying engineering cost deviations based on the temporal evolution modeling of list items.

[0005] The technical solution adopted in this invention is a method for identifying engineering cost deviations based on the time-series evolution modeling of bill of quantities items, comprising the following steps: S1, extracting time-series evolution parameters of bill of quantities items based on an intelligent bill of quantities cost analysis system, wherein the parameters include dynamic change data of engineering quantities during the project implementation stage, correlation data of cost components, and time-series node feature data; S2, performing multi-dimensional iterative calculations on the engineering quantity data in the time-series evolution parameters using an iterative algorithm for progressive deviation of engineering quantities, and constructing a progressive time-series evolution model of engineering quantities; S3, using a cost fluctuation correlation tracing algorithm to mine the correlation of cost components. S4. Establish a cost fluctuation source correlation model based on the fluctuation correlation in the data; S5. Use the cost trend adaptive inference model to perform trend inference on the characteristic data of time series nodes and generate the cost time series evolution trend curve; S6. Based on the time series evolution modeling framework of the bill of quantities items, integrate the engineering quantity time series progressive evolution model, the cost fluctuation source correlation model and the cost time series evolution trend curve to construct a comprehensive model for identifying engineering cost deviations; S7. Use the comprehensive model for identifying engineering cost deviations to calculate and identify the cost data of each time series node of the bill of quantities items, and output the cost deviation results and evolution trajectory data of each node.

[0006] Furthermore, the calculation process of the iterative algorithm for the time-series deviation of the quantity of work in S2 includes: obtaining the observed values ​​of the quantity of work for each time-series node of the bill of quantities item. and corresponding time series weights ,in Indicates the item number in the list. This represents the time-series node number, and the initial value for iteration is determined based on the dynamic changes in the engineering quantity in the time-series evolution parameters. Constructing an iterative expression for the time-series deviation of engineering quantities. ,in This is the iteration step size coefficient. For time sequence nodes The corresponding timestamp parameter, For the iterative stability coefficient, The iteration count is set to [number]; the iteration termination condition is set to the absolute value of the difference between two consecutive iteration results being less than a preset threshold. The converged time series baseline value of the engineering quantity is obtained through multiple iterative calculations. Based on the phase division parameters of the time-series evolution modeling of the bill of quantities items, the quantity evolution stages are divided; and the time-series baseline values ​​of the quantity of each stage are used as a basis. Based on the dynamic change data of the engineering quantity in the corresponding stage, a time-series progressive evolution model of the engineering quantity is constructed, which includes stage characteristic parameters, time-series progressive coefficients and deviation thresholds. The stage characteristic parameters are determined by the distribution characteristics and time-series weight ratio of the engineering quantity observations in each stage.

[0007] Furthermore, the calculation process of the cost fluctuation correlation tracing algorithm in S3 includes: extracting the correlation data of cost components in the time-series evolution parameters of the list items, and determining the set of cost influencing factors. ,in Indicates the first Several cost-influencing factors were identified, including material price fluctuation data, labor cost changes data, and machinery usage fee adjustment data; a cost fluctuation correlation and source tracing expression was constructed. ,in Indicates list items Total cost Represents time sequence nodes Next The values ​​of each influencing factor, Impact Factor The time-series average value, For time sequence nodes Next project Cost data, For the project The time-series average cost, The correlation strength coefficient, Impact Factor For the project The specific correlation correction coefficient; the correlation strength value between each cost influencing factor and the total cost is calculated based on the correlation source tracing expression. By combining the priority parameters of time-series nodes in the time-series evolution model of the list items, a cost fluctuation source tracing correlation model is established with the correlation strength value as the core and the time-series nodes as the dimensions, so as to locate the source of cost fluctuation.

[0008] Furthermore, the modeling process of the cost trend adaptive extrapolation model in S4 includes: extracting the basic cost data of each time series node from the time series node feature data. Temporal evolution rate and trend influence coefficient The trend influence coefficient is obtained by training historical data from the time-series evolution model of the list items; an initial expression for the time-series evolution trend of cost is constructed. ,in For the first Cost prediction values ​​for each time series node The time interval between adjacent time nodes. For the front The average cost of each node. The attenuation coefficient is introduced; an adaptive adjustment factor is introduced. ,in Describe the standard deviation calculation function and construct the optimized adaptive extrapolation expression for cost trends. ,in For adaptive correction coefficients, The change in the temporal evolution rate; adaptive correction coefficients for different stages based on the stage division results of the temporal evolution modeling of the inventory items. and attenuation coefficient Dynamic adjustments are made to adapt the model to the cost change characteristics at different evolution stages; the cost prediction values ​​at each time series node are... By comparing the model with actual observations and optimizing the model parameters through a feedback adjustment mechanism, a cost trend adaptive extrapolation model that combines time-series adaptability and prediction accuracy is finally constructed, generating a continuous cost time-series evolution trend curve.

[0009] Furthermore, the intelligent cost analysis system adopts a multi-core parallel processing architecture, including an FPGA chipset and a GPU acceleration module. The FPGA chipset is used for rapid extraction and preliminary classification of engineering quantity data, while the GPU acceleration module supports parallel computation of multiple algorithms. The system's analysis speed is no less than [a certain value]. The sampling frequency of the time-series evolution parameters of the listed items is synchronized with the project construction schedule, and the sampling interval is dynamically adjusted according to the construction stage. The parameter storage adopts a distributed database architecture, supporting real-time data writing and querying; the iteration step size coefficient of the engineering quantity time-series progressive deviation iterative algorithm. The value range is 0.01-0.15, and the iterative stability coefficient is... The value ranges from 0.001 to 0.05, representing the correlation strength coefficient of the cost fluctuation correlation tracing algorithm. The attenuation coefficient of the cost trend adaptive extrapolation model has a value range of 0.8-1.2. The value range is 0.1-0.3, and the adaptive correction coefficient is used. The value range is 0.9-1.1, and all parameters support custom configuration based on project type.

[0010] Further, S2 includes the following sub-steps: S21, based on the dynamic change data of the engineering quantity output by the intelligent analysis system for bill of quantities cost, classify the data according to the sub-project categories of the bill of quantities items, extract the time sequence data of the engineering quantity under each category, and clarify the construction process association information corresponding to each time sequence; S22, identify outliers in the classified engineering quantity time sequence data, mark data points that exceed the reasonable change range based on the local change trend of the time sequence and the correlation of adjacent data, and retain valid data for algorithm calculation; S23, import the valid data into the engineering quantity time sequence progressive deviation iteration algorithm in the order of time sequence nodes, set the initial iteration parameters of the algorithm, the initial iteration parameters include the upper limit of the number of iterations, the convergence threshold, and the time sequence weight allocation rule, the time sequence weight allocation rule is determined based on the construction importance parameters of the bill of quantities item time sequence evolution model; S24, obtain the engineering quantity iteration results of each time sequence node through algorithm calculation, combine the construction process association information, construct the engineering quantity time sequence progressive evolution model including process connection characteristics, and output the engineering quantity benchmark value and deviation fluctuation range of each time sequence node.

[0011] Further, S3 includes the following sub-steps: S31, performing hierarchical processing on the cost component data, dividing it into material, labor, and machinery cost data according to component type, and extracting the time-series change characteristics and inter-category correlation data of each type of data; S32, based on the hierarchically processed data, determining the time-series fluctuation amplitude and frequency of each cost component, and marking the significant fluctuation time-series nodes by combining the time node division parameters of the time-series evolution model of the list items; S33, inputting the cost data and fluctuation characteristic parameters of the marked time-series nodes into the cost fluctuation correlation tracing algorithm, obtaining the correlation degree data between each component and the total cost through algorithm calculation, and establishing an element correlation matrix; S34, based on the element correlation matrix and the location results of significant fluctuation nodes, constructing a cost fluctuation tracing correlation model, which supports outputting possible fluctuation sources by correlation degree and marking the influence range and propagation path of each source.

[0012] Further, S4 includes the following sub-steps: S41, expanding the dimensionality of the time-series node feature data by extracting, in addition to the basic cost data, the construction progress completion rate, resource input intensity, and external environmental impact parameters corresponding to each node, forming a multi-dimensional feature dataset; S42, dividing the multi-dimensional feature dataset into a training set and a validation set according to time sequence, wherein the training set is used for parameter training of the cost trend adaptive extrapolation model, and the validation set is used for model performance verification. The division ratio is dynamically determined based on the total number of nodes in the time-series evolution modeling of the list items; S43, optimizing and adjusting the initial parameters of the model using the training set data to enable the model to accurately fit the evolution trend of historical time-series data, and verifying the prediction error of the model using the validation set data. If the error exceeds the preset range, the parameters are readjusted; S44, extrapolating the cost trend of the time-series nodes based on the trained model to generate a cost time-series evolution trend curve including predicted values, confidence intervals, and trend change rates. The curve data is stored in the order of the time-series nodes for fusion modeling.

[0013] Further, S5 includes the following sub-steps: S51, establishing a unified data interface for the time-series evolution modeling of bill of quantities items, converting the output data of the time-series progressive evolution model of engineering quantities, the cost fluctuation source correlation model, and the cost time-series evolution trend curve into a unified format to ensure the consistency of the data structure; S52, based on the unified format data, extracting the output parameters of each model, including the benchmark value of engineering quantities, the correlation strength of cost influencing factors, and the cost prediction value, and constructing a comprehensive feature matrix; S53, setting the fusion weight of each output parameter, the fusion weight is dynamically allocated based on the recognition accuracy of each model in historical data, and the allocation process is combined with the stage characteristics of the time-series evolution modeling of bill of quantities items; S54, performing calculations on the comprehensive feature matrix through a weighted fusion algorithm to construct a comprehensive model for identifying engineering cost deviations that includes multi-dimensional identification indicators, and the model supports adjusting the priority of identification indicators according to the characteristics of different engineering types.

[0014] A method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items is disclosed. This method is implemented through an engineering cost deviation identification platform based on the temporal evolution modeling of bill of quantities items, comprising: an intelligent extraction module for time-series parameters of bill of quantities items, a progressive evolution modeling module for engineering quantities, a cost fluctuation correlation tracing analysis module, an adaptive cost trend extrapolation module, a multi-model fusion deviation identification module, and a deviation result visualization output module. The intelligent extraction module for time-series parameters of bill of quantities items establishes a data communication connection with the intelligent bill of quantities cost analysis system to extract the temporal evolution parameters of the bill of quantities items and transmit them to the progressive evolution modeling module for engineering quantities and the cost fluctuation correlation tracing analysis module. The progressive evolution modeling module for engineering quantities processes the received parameters based on a progressive deviation iteration algorithm for engineering quantities, generates a progressive evolution model for engineering quantities, and sends it to multiple... The system comprises three modules: a model fusion deviation identification module; a cost fluctuation correlation tracing analysis module; a cost fluctuation correlation model generation module; a cost trend adaptive inference module; a cost trend adaptive inference module; a cost trend time-series evolution trend curve; a multi-model fusion deviation identification module; a multi-model fusion deviation identification module; a cost trend adaptive inference module; a cost trend trend model ...

[0015] Beneficial Effects: This invention proposes a method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items. By deeply integrating the temporal evolution characteristics of bill of quantities items, a systematic deviation identification framework is constructed. It is no longer limited to static data comparison but dynamically captures the evolutionary patterns of engineering quantities and cost elements at each stage of project implementation. The dynamic changes in the temporal dimension are integrated into the entire process of deviation identification, comprehensively reflecting the cumulative impact of evolutionary characteristics at different stages on deviation formation. This solves the problems of insufficient adaptation to temporal evolution characteristics and lack of comprehensive deviation identification in existing technologies. Simultaneously, it integrates related technical means from multiple dimensions. Establishing a correlation analysis mechanism among cost components can not only accurately identify surface cost deviations, but also deeply explore the root causes of these deviations, clearly clarify the internal logic of each component, and provide a reliable basis for developing targeted corrective measures. This effectively compensates for the shortcomings of existing technologies in tracing the source of deviations and the lack of targeted correction, improving the accuracy and timeliness of cost deviation identification. It realizes the transformation from passive post-event verification to proactive pre-event prediction and in-event control, significantly improving the level of refinement in engineering cost control, and providing an efficient and feasible technical solution for cost control in large and complex engineering projects. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention. Figure 2 This is a flowchart of method step S2 of the present invention; Figure 3 This is a flowchart of method step S3 of the present invention; Figure 4 This is a flowchart of method step S4 of the present invention; Figure 5 This is a flowchart of step S5 of the method of the present invention. Detailed Implementation

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] like Figure 1 As shown, a method for identifying engineering cost deviations based on the time-series evolution modeling of bill of quantities items is characterized by the following steps: S1, extracting time-series evolution parameters of bill of quantities items based on the intelligent analysis system of bill of quantities cost, the parameters including dynamic change data of engineering quantities during the project implementation stage, correlation data of cost components, and time-series node feature data; S2, performing multi-dimensional iterative calculations on the engineering quantity data in the time-series evolution parameters through the engineering quantity time-series progressive deviation iterative algorithm to construct the engineering quantity time-series progressive evolution model; S3, using the cost fluctuation correlation tracing algorithm to mine the correlation data of cost components. S4. Establish a cost fluctuation source correlation model based on the fluctuation relationship in the data; S5. Use the cost trend adaptive inference model to perform trend inference on the characteristic data of time series nodes and generate the cost time series evolution trend curve; S6. Based on the time series evolution modeling framework of the bill of quantities items, integrate the engineering quantity time series progressive evolution model, the cost fluctuation source correlation model and the cost time series evolution trend curve to construct a comprehensive model for identifying engineering cost deviations; S7. Use the comprehensive model for identifying engineering cost deviations to calculate and identify the cost data of each time series node of the bill of quantities items, and output the cost deviation results and evolution trajectory data of each node.

[0019] Step S1 involves extracting the core parameters of the time-series evolution of the bill of quantities items through the intelligent cost analysis system. During implementation, the system's multi-core parallel processing architecture must be activated first, utilizing the collaborative working mode of the FPGA chipset and GPU acceleration module to ensure that the data extraction and processing rate remains above 10GB / s. The system will automatically connect to the engineering bill of quantities database, filtering target data according to the sub-project category. The extracted core parameters specifically include dynamic change data of the project quantity during the project implementation phase, correlation data of cost components, and time-series node characteristic data. The dynamic change data of the project quantity includes information such as the completed quantity and changes of each construction process; the correlation data of cost components includes the proportion and mutual influence data of various costs such as materials, labor, and machinery; and the time-series node characteristic data corresponds to the time node identifier and progress weight of each construction stage. The parameter sampling frequency is synchronized with the project construction schedule, and the sampling interval is dynamically adjusted according to the construction stage. Specifically, it is 1 day for the basic construction stage, 3 days for the main construction stage, and 7 days for the decoration and finishing stage. During the sampling process, a distributed database architecture is used to realize real-time writing and querying of data, ensuring the integrity and timeliness of the parameters. This provides comprehensive and accurate basic data support for the subsequent calculation and modeling process of each algorithm. The implementation quality of this step directly determines the accuracy of subsequent deviation identification and is the data foundation guarantee link of the entire method.

[0020] Step S2 uses a progressive deviation iterative algorithm to perform multi-dimensional iterative calculations on the quantity data and construct an evolutionary model. During implementation, the dynamic quantity change data extracted in S1 is first categorized by the list item number, and valid data is filtered out and imported into the algorithm's calculation module. Initial parameters need to be set when the algorithm starts, including an iteration step size coefficient ranging from 0.01 to 0.15, an iteration stability coefficient ranging from 0.001 to 0.05, an upper limit of 100 iterations, and a convergence threshold of 0.0001. During the calculation, the algorithm processes the data step-by-step according to the time sequence nodes, adjusting the calculation results through multiple iterations until the absolute value of the difference between two adjacent iterations is less than the convergence threshold, thus obtaining the converged time-series baseline value of the quantity. Simultaneously, combined with the stage division parameters for the time-series evolution modeling of the list items, the project implementation process is divided into 3 to 5 evolutionary stages. The division of each stage is determined based on the construction progress completion rate, i.e., stage division is triggered when the progress completion rate reaches 20%, 40%, 60%, and 80%. Finally, based on the time-series baseline values ​​of the engineering quantity at each stage and the dynamic change data of the engineering quantity at the corresponding stage, a time-series progressive evolution model of the engineering quantity is constructed, which includes stage characteristic parameters, time-series progression coefficients and deviation thresholds. This model can accurately reflect the dynamic change law of the engineering quantity over time and provide a quantitative analysis basis for subsequent deviation identification. During implementation, it is necessary to ensure the synergy between algorithm operation and stage division to ensure the time-series adaptability of the model.

[0021] Step S3 employs a cost fluctuation correlation tracing algorithm to mine the fluctuation correlations in the cost component data and establish a tracing model. During implementation, the cost component correlation data extracted in S1 is first processed hierarchically, categorized into four main types: materials, labor, machinery, and other expenses. Each type of data requires separate extraction of time-series variation characteristics and inter-category correlation data. Subsequently, the time-series fluctuation amplitude and frequency of each cost component are determined. The fluctuation amplitude is calculated as the deviation ratio based on the historical average value of the component. The fluctuation frequency is counted as the number of times the fluctuation amplitude exceeds 5% per unit time. Combined with the time node division parameters of the bill of quantities project time-series evolution model, time nodes with fluctuation amplitudes exceeding 10% or fluctuation frequencies greater than 3 times per month are marked as key time nodes. The cost data and fluctuation characteristic parameters of the key time nodes are input into the algorithm. During the algorithm's operation, the correlation strength coefficient ranges from 0.8 to 1.2. By calculating the correlation strength value between each cost component and the total cost, an element correlation matrix is ​​established. Based on the location results of the element association matrix and key time-series nodes, a cost fluctuation source tracing association model is constructed with the association strength value as the core and the time-series nodes as the dimensions. This model can clearly present the degree of influence of each cost component on the total cost. During the implementation process, it is necessary to ensure the accuracy of data layer processing and the rationality of key node marking, so as to provide technical support for accurately locating the source of cost fluctuations and ensure the pertinence of subsequent deviation cause analysis.

[0022] Step S4 utilizes an adaptive cost trend extrapolation model to extrapolate the trend of time-series node feature data and generate a cost time-series evolution trend curve. During implementation, the basic cost data, time-series evolution rate, and trend influence coefficient for each time-series node are first extracted from the time-series node feature data extracted in S1. The trend influence coefficient is obtained through training on time-series data from 30 similar historical projects. The selection of training samples must ensure coverage of project types with different project scales and construction environments. In the initial stage of model operation, the attenuation coefficient ranges from 0.1 to 0.3, and the adaptive correction coefficient ranges from 0.9 to 1.1. The time interval between adjacent time-series nodes is determined according to the construction stage: 1 day for the foundation construction stage, 3 days for the main structure construction stage, and 7 days for the decoration and finishing stage. After generating the initial cost prediction value through model operation, an adaptive adjustment factor is introduced. This factor is calculated based on the ratio of the standard deviation to the mean of the cost data of the preceding nodes, and is used to optimize and correct the initial prediction value. Subsequently, based on the phase division results of the time-series evolution modeling of the bill of quantities items, the adaptive correction coefficients and attenuation coefficients for different phases are dynamically adjusted. For example, the attenuation coefficient for the foundation construction phase is 0.3, and the adaptive correction coefficient is 1.1; the attenuation coefficient for the main construction phase is 0.2, and the adaptive correction coefficient is 1.0; and the attenuation coefficient for the decoration and finishing phase is 0.1, and the adaptive correction coefficient is 0.9. The optimized cost prediction values ​​for each time-series node are compared with the actual observed values, and the model parameters are further optimized through a feedback adjustment mechanism. Finally, a continuous cost time-series evolution trend curve is generated. This curve can accurately predict the cost change trend of subsequent time-series nodes. During implementation, it is necessary to ensure the adaptability of model parameter adjustments to the phase characteristics.

[0023] Step S5, based on the time-series evolution modeling framework of the bill of quantities, integrates the three models generated in the previous steps to construct a comprehensive model for identifying engineering cost deviations. During implementation, a unified data interface is first established. Data such as the baseline values ​​and deviation fluctuation ranges output by the time-series progressive evolution model of engineering quantities, the element correlation matrix and fluctuation source data output by the cost fluctuation tracing and correlation model, and the cost prediction values ​​and confidence intervals output by the time-series evolution trend curve of cost are converted into a unified data format to ensure data structure consistency. Subsequently, the core output parameters of each model are extracted to construct a comprehensive feature matrix comprising three dimensions: engineering quantity, cost influencing factors, and cost prediction values. Each dimension includes 10 to 15 specific indicators, the selection of which must be determined in conjunction with the core requirements of deviation identification. The fusion weights of each core output parameter are set. These fusion weights are dynamically allocated based on the identification accuracy of each model in 20 historical projects. The accuracy calculation is based on the degree of agreement between the actual deviation data and the model's identification results. If the accuracy of a model reaches 90% or higher, the fusion weight of its corresponding parameter is set to 0.4, and the fusion weights of the corresponding parameters of the other two models are set to 0.3 respectively. By using a weighted fusion algorithm to operate on the comprehensive feature matrix, a comprehensive model for identifying engineering cost deviations is constructed, which includes multi-dimensional identification indicators. The priority of each identification indicator in the model is adjusted according to the type of project. For example, the priority of material cost-related indicators is set to the highest in building construction projects, and the priority of machinery usage-related indicators is set to the highest in municipal engineering projects. During implementation, it is necessary to ensure the accuracy of data format conversion and the rationality of fusion weight allocation to guarantee the identification accuracy of the comprehensive model.

[0024] Step S6 uses the comprehensive cost deviation identification model to calculate and identify deviations in the cost data of each time-series item in the bill of quantities, outputting deviation results and evolution trajectory data. During implementation, the actual cost data for each time-series item is first imported into the comprehensive model. The model analyzes the data one by one according to a preset identification index system. Deviation calculation is based on the cost benchmark value generated by the model, calculating the deviation ratio between the actual cost data and the benchmark value. A deviation ratio within ±3% is considered no deviation, a deviation ratio between 3% and 5% is considered a slight deviation, a deviation ratio between 5% and 10% is considered a moderate deviation, and a deviation ratio exceeding 10% is considered a serious deviation. Simultaneously, combining the evolution model and trend curve generated in the previous steps, the changing patterns of deviation data at each time-series item are tracked to form deviation evolution trajectory data. The trajectory data must include key information such as deviation value, deviation level, deviation duration, and deviation impact range. When outputting results, the deviation results and evolution trajectory data must be organized sequentially according to the time sequence nodes to ensure data completeness and organization. The output information for each time sequence node must be correlated with the engineering quantity change data, cost component fluctuation data, and trend prediction data for that node, providing a comprehensive basis for the formulation of subsequent corrective measures. During implementation, the deviation judgment criteria must be strictly followed to ensure the accuracy of deviation level classification, while also ensuring the readability and usability of the output data, meeting the cost deviation control needs of engineering management personnel, and achieving closed-loop management from deviation identification to data application.

[0025] Preferably, the calculation process of the time-series progressive deviation iterative algorithm for quantities in S2 includes: obtaining the quantity observation values ​​of each time-series node of the bill of quantities item. and corresponding time series weights ,in Indicates the item number in the list. This represents the time-series node number, and the initial value for iteration is determined based on the dynamic changes in the engineering quantity in the time-series evolution parameters. Constructing an iterative expression for the time-series deviation of engineering quantities. ,in This is the iteration step size coefficient. For time sequence nodes The corresponding timestamp parameter, For the iterative stability coefficient, The iteration count is set to [number]; the iteration termination condition is set to the absolute value of the difference between two consecutive iteration results being less than a preset threshold. The converged time series baseline value of the engineering quantity is obtained through multiple iterative calculations. Based on the phase division parameters of the time-series evolution modeling of the bill of quantities items, the quantity evolution stages are divided; and the time-series baseline values ​​of the quantity of each stage are used as a basis. Based on the dynamic change data of the engineering quantity in the corresponding stage, a time-series progressive evolution model of the engineering quantity is constructed, which includes stage characteristic parameters, time-series progressive coefficients and deviation thresholds. The stage characteristic parameters are determined by the distribution characteristics and time-series weight ratio of the engineering quantity observations in each stage.

[0026] Specifically, the iterative algorithm for progressive deviation of quantities first collects the relevant data and corresponding time-series weights of the quantities for each time-series node of the bill of quantities. Based on the dynamic change data of quantities in the core parameters of time-series evolution, the initial value is derived using a weighted average method. This derivation method can integrate the basic data of each time-series node, ensuring that the initial value fits the actual implementation of the project. The time-series weights are allocated according to the importance of the construction stage: 30% for the foundation construction stage, 40% for the main construction stage, and 30% for the decoration and finishing stage. The constructed iterative expression is derived by introducing an iteration step size coefficient and an iteration stability coefficient based on the initial value, combined with the fourth power of the timestamp parameter of the time-series node. The fourth power of the timestamp parameter can amplify the influence of key time-series nodes. The iteration step size coefficient ranges from 0.01 to 0.15, and the iteration stability coefficient ranges from 0.001 to 0.05. This formula design can avoid oscillations during the iteration process and ensure convergence stability. The iteration termination condition is set based on the absolute value of the difference between two adjacent iterations being less than a preset threshold, which is set to 0.0001. Through multiple iterations, a converged time-series baseline value for the project quantity is obtained. Simultaneously, combined with the stage division parameters of the bill of quantities project time-series evolution modeling, 3 to 5 evolution stages are defined according to progress completion rates of 20%, 40%, 60%, and 80%, ensuring that stage division is synchronized with construction progress. Based on the baseline values ​​and dynamic change data of each stage, an evolutionary model is constructed, including stage characteristic parameters, time-series progression coefficients, and deviation thresholds. The stage characteristic parameters are determined by the data distribution characteristics of each stage, the time-series progression coefficients are calculated based on the data change rate between stages, and the deviation threshold is set at 5% of the baseline value for each stage. This model can accurately reflect the time-series change pattern of the project quantity, providing a quantitative basis for subsequent deviation identification. The entire implementation process must ensure the accuracy of data transmission at each step and guarantee the synergy between algorithm calculation and modeling.

[0027] Preferably, the calculation process of the cost fluctuation correlation tracing algorithm in S3 includes: extracting the correlation data of cost components in the time-series evolution parameters of the list items, and determining the set of cost influencing factors. ,in Indicates the first Several cost-influencing factors were identified, including material price fluctuation data, labor cost changes data, and machinery usage fee adjustment data; a cost fluctuation correlation and source tracing expression was constructed. ,in Indicates list items Total cost Represents time sequence nodes Next The values ​​of each influencing factor, Impact Factor The time-series average value, For time sequence nodes Next project Cost data, For the project The time-series average cost, The correlation strength coefficient, Impact Factor For the project The specific correlation correction coefficient; the correlation strength value between each cost influencing factor and the total cost is calculated based on the correlation source tracing expression. By combining the priority parameters of time-series nodes in the time-series evolution model of the list items, a cost fluctuation source tracing correlation model is established with the correlation strength value as the core and the time-series nodes as the dimensions, so as to locate the source of cost fluctuation.

[0028] Specifically, the cost fluctuation correlation tracing algorithm extracts correlation data of cost components, identifying a set of influencing factors including materials, labor, machinery, and other expenses. Data for each influencing factor comes from core parameters extracted by the intelligent cost analysis system. The extraction process is layered by factor type to ensure accurate data classification, laying the foundation for subsequent correlation analysis. The constructed correlation tracing expression first calculates the partial derivatives of each influencing factor with respect to the total cost, reflecting the direct impact of a single factor. Then, it introduces the ratio of the product of covariance and standard deviation to quantify the linear correlation strength between the factor and the total cost. Finally, it adds a correlation strength coefficient and a specific correlation correction coefficient. The correlation strength coefficient ranges from 0.8 to 1.2, while the specific correlation correction coefficient is adjusted according to different project types: 0.9 for building construction, 1.0 for municipal engineering, and 1.1 for installation engineering. This formula design comprehensively considers both direct and indirect influences, accurately uncovering fluctuation correlations. A cost fluctuation tracing correlation model is established based on correlation strength values. Factors with correlation strength values ​​greater than 0.8 are identified as core influencing factors. Combining the time-series node priority parameters of the bill of quantities project time-series evolution model, nodes are sorted according to their importance. Key nodes are set as level one, secondary nodes as level two, and ordinary nodes as level three. The model, with correlation strength values ​​as the core and time-series nodes as the dimension, can clearly present the degree of influence of each factor on the total cost and the fluctuation propagation path. During implementation, it is necessary to ensure the accuracy of data layering and the rationality of key node marking. The key node marking standard is a fluctuation amplitude exceeding 10% or a fluctuation frequency greater than 3 times per month to ensure the accuracy of fluctuation source location.

[0029] Preferably, the modeling process of the cost trend adaptive extrapolation model in S4 includes: extracting the basic cost data of each time series node from the time series node feature data. Temporal evolution rate and trend influence coefficient The trend influence coefficient is obtained by training historical data from the time-series evolution model of the list items; an initial expression for the time-series evolution trend of cost is constructed. ,in For the first Cost prediction values ​​for each time series node The time interval between adjacent time nodes. For the front The average cost of each node. The attenuation coefficient is introduced; an adaptive adjustment factor is introduced. ,in Describe the standard deviation calculation function and construct the optimized adaptive extrapolation expression for cost trends. ,in For adaptive correction coefficients, The change in the temporal evolution rate; adaptive correction coefficients for different stages based on the stage division results of the temporal evolution modeling of the inventory items. and attenuation coefficient Dynamic adjustments are made to adapt the model to the cost change characteristics at different evolution stages; the cost prediction values ​​at each time series node are... By comparing the model with actual observations and optimizing the model parameters through a feedback adjustment mechanism, a cost trend adaptive extrapolation model that combines time-series adaptability and prediction accuracy is finally constructed, generating a continuous cost time-series evolution trend curve.

[0030] Specifically, the cost trend adaptive extrapolation model extracts the basic cost data, time-series evolution rate, and trend influence coefficient for each time-series node. The trend influence coefficient is obtained through training on time-series data from 30 similar historical projects. The training samples cover different project scales and construction environments to ensure the universality of the coefficient. The time-series evolution rate is calculated as the ratio of the change in data between adjacent nodes to the time interval, which is determined to be 1 day, 3 days, or 7 days depending on the construction stage. The constructed initial trend expression is derived based on the current node cost data, combined with the time-series evolution rate and time interval to calculate the basic predicted value. Then, an exponential decay term is introduced based on the difference between the previous node data and the average value. The decay coefficient ranges from 0.1 to 0.3, which can weaken the outdated influence of distant nodes and highlight the reference value of recent data. The introduced adaptive adjustment factor, calculated based on the ratio of the standard deviation to the mean of the cost data from preceding nodes, dynamically adjusts the predicted values ​​according to the degree of data dispersion. The optimized extrapolation expression adds an adaptive correction coefficient and a product term of the evolution rate change to the initial predicted values. The adaptive correction coefficient ranges from 0.9 to 1.1, and this formula design enhances the model's adaptability to data changes. Based on the phase division results of the time-series evolution modeling of the bill of quantities items, the adaptive correction coefficient and attenuation coefficient are dynamically adjusted for different stages: 0.3 for the foundation construction stage and 1.1 for the adaptive correction coefficient; 0.2 for the main construction stage and 1.0 for the adaptive correction coefficient; and 0.1 for the attenuation coefficient and 0.9 for the decoration and finishing stage, ensuring the model adapts to the cost change characteristics of different stages. By comparing the predicted values ​​with the actual observed values, the model parameters are optimized through a feedback adjustment mechanism. When the prediction error exceeds 3%, parameter adjustment is triggered. The final model can generate a continuous cost time series evolution trend curve and accurately predict the cost changes at subsequent nodes. During implementation, it is necessary to ensure the timeliness of parameter adjustment at each step and the accuracy of data comparison to guarantee the prediction accuracy and time series adaptability of the model.

[0031] Preferably, the intelligent cost analysis system adopts a multi-core parallel processing architecture, including an FPGA chipset and a GPU acceleration module. The FPGA chipset is used for rapid extraction and preliminary classification of engineering quantity data, while the GPU acceleration module supports parallel computation of multiple algorithms. The system's analysis speed is no less than [a certain value]. The sampling frequency of the time-series evolution parameters of the listed items is synchronized with the project construction schedule, and the sampling interval is dynamically adjusted according to the construction stage. The parameter storage adopts a distributed database architecture, supporting real-time data writing and querying; the iteration step size coefficient of the engineering quantity time-series progressive deviation iterative algorithm. The value range is 0.01-0.15, and the iterative stability coefficient is... The value ranges from 0.001 to 0.05, representing the correlation strength coefficient of the cost fluctuation correlation tracing algorithm. The attenuation coefficient of the cost trend adaptive extrapolation model has a value range of 0.8-1.2. The value range is 0.1-0.3, and the adaptive correction coefficient is used. The value range is 0.9-1.1, and all parameters support custom configuration based on project type.

[0032] Preferred, such as Figure 2 As shown, S2 includes the following sub-steps: S21, based on the dynamic change data of the engineering quantity output by the intelligent analysis system of bill of quantities cost, the data is classified according to the sub-project categories of the bill of quantities items, and the time sequence data of the engineering quantity under each category is extracted to clarify the construction process association information corresponding to each time sequence; S22, outlier identification is performed on the classified engineering quantity time sequence data, and based on the local change trend of the time sequence and the correlation of adjacent data, data points that exceed the reasonable change range are marked, and valid data is retained for algorithm calculation; S23, the valid data is imported into the engineering quantity time sequence progressive deviation iteration algorithm in the order of time sequence nodes, and the initial iteration parameters of the algorithm are set. The initial iteration parameters include the upper limit of the number of iterations, the convergence threshold, and the time sequence weight allocation rule. The time sequence weight allocation rule is determined based on the construction importance parameters of the bill of quantities item time sequence evolution model; S24, the engineering quantity iteration results of each time sequence node are obtained through algorithm calculation, and combined with the construction process association information, an engineering quantity time sequence progressive evolution model including process connection characteristics is constructed. The model outputs the engineering quantity benchmark value and deviation fluctuation range of each time sequence node.

[0033] Specifically, in step S2, sub-step S21 first utilizes the parallel processing capabilities of the intelligent cost analysis system to classify the extracted dynamic change data of engineering quantities according to the sub-project categories, simultaneously extracting the time-series data of engineering quantities under each category, clarifying the construction process association information corresponding to each time-series, and setting the data filtering accuracy to 99.5% during the classification process to ensure the validity of the data for subsequent calculations. Sub-step S22 performs outlier identification on the classified time-series data. Based on the local change trend of the time-series and the correlation of adjacent data, a sliding window analysis method is used to mark data points that exceed the reasonable change range. The reasonable change range is set at ±15% of the historical data of similar projects, and only valid data is retained for subsequent calculations to avoid outliers interfering with the algorithm accuracy. Sub-step S23 imports the valid data into the corresponding algorithm in the order of time-series nodes, sets the initial iteration parameters, with the upper limit of the number of iterations set to 100 and the convergence threshold set to 0.0001. The time-series weight allocation rule is determined based on the construction importance parameters of the time-series evolution model of the bill of quantities items, with the weight of core processes being 20% ​​higher than that of non-core processes. Step S24 obtains the iterative results of the engineering quantity at each time node through algorithm calculation. Combined with the construction process association information, an evolutionary model including process connection characteristics is constructed. The model outputs the engineering quantity benchmark value and deviation fluctuation range for each time node. The deviation fluctuation range is set at ±5% of the benchmark value. The entire implementation process must ensure the continuity of data transmission in each step to ensure that the model can accurately reflect the correlation between the engineering quantity time sequence change and the process connection.

[0034] Preferred, such as Figure 3 As shown, step S3 includes the following sub-steps: S31, performing hierarchical processing on the cost component data, dividing it into material, labor, and machinery cost data according to component type, and extracting the time-series change characteristics and inter-category correlation data of each type of data; S32, based on the hierarchically processed data, determining the time-series fluctuation amplitude and frequency of each cost component, and marking the significant fluctuation time-series nodes by combining the time node division parameters of the time-series evolution model of the list items; S33, inputting the cost data and fluctuation characteristic parameters of the marked time-series nodes into the cost fluctuation correlation tracing algorithm, obtaining the correlation degree data between each component and the total cost through algorithm calculation, and establishing an element correlation matrix; S34, based on the element correlation matrix and the location results of significant fluctuation nodes, constructing a cost fluctuation tracing correlation model, which supports outputting possible fluctuation sources by correlation degree sorting, and marking the influence range and propagation path of each source.

[0035] Specifically, step S3 involves in-depth processing and modeling of cost fluctuation correlation data. Sub-step S31 first performs layered processing on the correlation data of cost components, clearly dividing them into four categories based on component type: materials, labor, machinery, and other expenses. During the layering process, the independence of each type of data is ensured, and cross-data is categorized according to the category with the highest influence weight, with a categorization accuracy rate of no less than 98%. Simultaneously, the temporal variation characteristics of each type of data and inter-category correlation data are extracted. Sub-step S32 calculates the temporal fluctuation amplitude and frequency of each cost component based on the layered data. The fluctuation amplitude is calculated as the deviation ratio based on the historical average value of the component. The fluctuation frequency is counted as the number of times the fluctuation amplitude exceeds 5% per unit time. Combined with the time node division parameters of the bill of quantities project temporal evolution model, time nodes with fluctuation amplitudes exceeding 10% or fluctuation frequencies greater than 3 times per month are marked as key time nodes. Step S33 inputs the cost data and fluctuation characteristic parameters of key time-series nodes into the corresponding algorithm. The algorithm calculates the correlation between each component and the total cost. The correlation calculation uses a multi-dimensional weighted summation method to construct a complete element correlation matrix. The matrix dimensions are consistent with the number of cost components. Step S34, based on the element correlation matrix and the location results of significant fluctuation nodes, constructs a cost fluctuation source tracing correlation model. The model supports sorting possible fluctuation sources from high to low correlation and outputting the influence range and propagation path of each source. The influence range is defined by the set of elements with a correlation higher than 0.7, ensuring the accuracy and comprehensiveness of fluctuation source location.

[0036] Preferred, such as Figure 4 As shown, step S4 includes the following sub-steps: S41, expanding the dimensionality of the time-series node feature data by extracting, in addition to the basic cost data, the construction progress completion rate, resource input intensity, and external environmental impact parameters corresponding to each node, forming a multi-dimensional feature dataset; S42, dividing the multi-dimensional feature dataset into a training set and a validation set according to time sequence, wherein the training set is used for parameter training of the cost trend adaptive extrapolation model, and the validation set is used for model performance verification. The division ratio is dynamically determined based on the total number of nodes in the time-series evolution modeling of the list items; S43, optimizing and adjusting the initial parameters of the model using the training set data to enable the model to accurately fit the evolution trend of historical time-series data, and verifying the prediction error of the model using the validation set data. If the error exceeds the preset range, the parameters are readjusted; S44, extrapolating the cost trend of the time-series nodes based on the trained model to generate a cost time-series evolution trend curve including predicted values, confidence intervals, and trend change rates. The curve data is stored in the order of the time-series nodes for fusion modeling.

[0037] Specifically, step S4 consists of four sub-steps focusing on the construction and optimization of the cost trend projection model. Sub-step S41 expands the dimensionality of the time-series node feature data. In addition to the basic cost data, it extracts the construction progress completion rate, resource input intensity, and external environmental impact parameters corresponding to each node, forming a multi-dimensional feature dataset. The construction progress completion rate is statistically analyzed daily, the resource input intensity is statistically analyzed weekly, and the external environmental impact parameters are quantified into values ​​of 0-1 according to the degree of impact. Sub-step S42 divides the multi-dimensional feature dataset into a training set and a validation set according to the time sequence. The division ratio is dynamically determined based on the total number of nodes in the time-series evolution model of the bill of quantities project. When the total number of nodes is less than 50, the training set accounts for 70% and the validation set accounts for 30%. When the total number of nodes is not less than 50, the training set accounts for 80% and the validation set accounts for 20%, ensuring that the training set can cover the feature data of different stages. Step S43 optimizes the initial parameters of the model using the training set data, employing gradient descent to minimize the prediction error, enabling the model to accurately fit the evolution trend of historical time-series data. The model performance is then tested using validation set data, with a preset error threshold of 3%. If the error exceeds the threshold, the parameters are readjusted until the error meets the requirements. Step S44, based on the trained model, extrapolates the cost trend for subsequent time-series nodes, generating a cost time-series evolution trend curve including predicted values, confidence intervals, and trend change rates. The confidence interval is set at a 95% confidence level, and the trend change rate is calculated based on the difference between adjacent prediction nodes. The curve data is stored sequentially by time-series node to ensure the data can be directly used for subsequent fusion modeling, guaranteeing the practicality and accuracy of the trend extrapolation.

[0038] Preferred, such as Figure 5 As shown, S5 includes the following sub-steps: S51, establishing a unified data interface for the time-series evolution modeling of the bill of quantities items, converting the output data of the time-series progressive evolution model of engineering quantities, the cost fluctuation source correlation model, and the cost time-series evolution trend curve into a unified format to ensure the consistency of the data structure; S52, based on the unified format data, extracting the output parameters of each model, including the benchmark value of engineering quantities, the correlation strength of cost influencing factors, and the cost prediction value, and constructing a comprehensive feature matrix; S53, setting the fusion weight of each output parameter, the fusion weight is dynamically allocated based on the recognition accuracy of each model in historical data, and the allocation process is combined with the stage characteristics of the time-series evolution modeling of the bill of quantities items; S54, performing calculations on the comprehensive feature matrix through a weighted fusion algorithm to construct a comprehensive model for identifying engineering cost deviations that includes multi-dimensional identification indicators, and the model supports adjusting the priority of identification indicators according to the characteristics of different engineering types.

[0039] Specifically, step S5 involves multi-model data fusion and comprehensive model construction. Step S51 first establishes a unified data interface for the time-series evolution modeling of the bill of quantities items. The interface uses a standardized data transmission protocol to ensure data compatibility. The output data from the previous three models is converted to a unified format using a structured data table. Fields include parameter names, values, time-series nodes, and error ranges, ensuring data structure consistency and integrity. Step S52, based on the unified format data, extracts the core output parameters of each model. Specifically, the quantity evolution model extracts the benchmark value of the quantity, the cost tracing model extracts the correlation strength value, and the trend extrapolation model extracts the cost prediction value. This is then integrated to construct a comprehensive feature matrix including multi-dimensional identification indicators. The matrix's row dimension represents the number of time-series nodes, and the column dimension represents the number of core parameters. Step S53 sets the fusion weights for each core output parameter. These weights are dynamically allocated based on the identification accuracy of each model in 20 similar historical projects. For every 10% increase in accuracy, the corresponding parameter's fusion weight increases by 0.1, with a total weight of 1. The allocation process considers the stage characteristics of the time-series evolution modeling of the bill of quantities items, with the weight allocation deviation between different stages not exceeding 0.2. Step S54 uses a weighted fusion algorithm to calculate the comprehensive feature matrix and construct a comprehensive model for identifying engineering cost deviations, which includes multi-dimensional identification indicators. The priority of each identification indicator in the model is adjusted according to the engineering type. For industrial building engineering, the priority of mechanical cost-related indicators is increased by 15%, and for civil building engineering, the priority of material cost-related indicators is increased by 15%, ensuring that the comprehensive model can adapt to the cost deviation identification needs of different engineering types and improve the targeting and accuracy of identification.

[0040] A method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items is disclosed. This method is implemented through an engineering cost deviation identification platform based on the temporal evolution modeling of bill of quantities items, comprising: an intelligent extraction module for time-series parameters of bill of quantities items, a progressive evolution modeling module for engineering quantities, a cost fluctuation correlation tracing analysis module, an adaptive cost trend extrapolation module, a multi-model fusion deviation identification module, and a deviation result visualization output module. The intelligent extraction module for time-series parameters of bill of quantities items establishes a data communication connection with the intelligent bill of quantities cost analysis system to extract the temporal evolution parameters of the bill of quantities items and transmit them to the progressive evolution modeling module for engineering quantities and the cost fluctuation correlation tracing analysis module. The progressive evolution modeling module for engineering quantities processes the received parameters based on a progressive deviation iteration algorithm for engineering quantities, generates a progressive evolution model for engineering quantities, and sends it to multiple... The system comprises three modules: a model fusion deviation identification module; a cost fluctuation correlation tracing analysis module; a cost fluctuation correlation model generation module; a cost trend adaptive inference module; a cost trend adaptive inference module; a cost trend time-series evolution trend curve; a multi-model fusion deviation identification module; a multi-model fusion deviation identification module; a cost trend adaptive inference module; a cost trend trend model ...

[0041] A method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items is proposed. This method uses the entire project implementation process as its core framework, incorporating key information such as changes in quantities and dynamic adjustments to cost elements at each stage into a unified identification framework. It moves beyond isolated comparisons of single-point data, systematically capturing the correlation and evolution patterns between different temporal nodes, clearly presenting the cumulative effects of changes in quantities and costs as construction progresses. By constructing a dynamic identification mechanism covering the entire lifecycle, potential deviation trends can be predicted in advance, shifting from passive post-event verification to proactive in-process control. This significantly improves the foresight and comprehensiveness of deviation identification, effectively avoiding the problems of deviation omissions or misjudgments caused by neglecting temporal evolution characteristics.

[0042] This method enhances the ability to trace and precisely control deviations, successfully overcoming the shortcoming of existing technologies that struggle to pinpoint the root cause of deviations. By integrating multi-dimensional technical means, it establishes a deep correlation analysis system among cost components, enabling not only rapid identification of surface cost deviation data but also layer-by-layer breakdown of the interaction logic of each element to accurately pinpoint the core causes of deviations. This end-to-end technical design, from deviation identification to tracing its root causes, provides a solid basis for developing targeted corrective measures, completely changing the current situation where corrective technologies lack specificity. Simultaneously, the systematic technical architecture significantly improves the accuracy and efficiency of deviation identification, enabling rapid response to the cost control needs of large and complex projects, significantly enhancing the refinement of engineering cost control, and providing efficient and reliable technical support for project cost control.

[0043] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for identifying engineering cost deviations based on the temporal evolution modeling of bill of quantities items, characterized in that, Includes the following steps: S1. Extract time-series evolution parameters of the bill of quantities items based on the intelligent analysis system for bill of quantities cost. These parameters include dynamic change data of engineering quantities during the project implementation phase, correlation data of cost components, and time-series node feature data. S2. Perform multi-dimensional iterative calculations on the engineering quantity data in the time-series evolution parameters using the engineering quantity time-series progressive deviation iteration algorithm to construct a time-series progressive evolution model for engineering quantities. S3. Employ a cost fluctuation correlation tracing algorithm to mine the fluctuation correlation relationships in the correlation data of cost components and establish a cost fluctuation tracing correlation model. S4. Utilize a cost trend adaptive inference model to perform trend inference on the time-series node feature data, generating a cost time-series evolution trend curve. S5. Based on the bill of quantities item time-series evolution modeling framework, integrate the engineering quantity time-series progressive evolution model, the cost fluctuation tracing correlation model, and the cost time-series evolution trend curve to construct a comprehensive model for identifying engineering cost deviations. S6. Calculate and identify deviations in the cost data of each time-series node of the bill of quantities item using the comprehensive model for identifying engineering cost deviations, and output the cost deviation results and evolution trajectory data for each node.

2. The engineering cost deviation identification method based on the time sequence evolution modeling of the bill items according to claim 1, characterized in that, The calculation process of the time-series progressive deviation iterative algorithm for quantities in S2 includes: obtaining the quantity observation values ​​of each time-series node of the bill of quantities item. and corresponding time series weights ,in Indicates the item number in the list. This represents the time-series node number, and the initial value for iteration is determined based on the dynamic changes in the engineering quantity in the time-series evolution parameters. Constructing an iterative expression for the time-series deviation of engineering quantities. ,in This is the iteration step size coefficient. For time sequence nodes The corresponding timestamp parameter, For the iterative stability coefficient, The iteration count is set to [number]; the iteration termination condition is set to the absolute value of the difference between two consecutive iteration results being less than a preset threshold. The converged time series baseline value of the engineering quantity is obtained through multiple iterative calculations. Based on the phase division parameters of the time-series evolution modeling of the bill of quantities items, the quantity evolution stages are divided; and the time-series baseline values ​​of the quantity of each stage are used as a basis. Based on the dynamic change data of the engineering quantity in the corresponding stage, a time-series progressive evolution model of the engineering quantity is constructed, which includes stage characteristic parameters, time-series progressive coefficients and deviation thresholds. The stage characteristic parameters are determined by the distribution characteristics and time-series weight ratio of the engineering quantity observations in each stage.

3. The engineering cost deviation identification method based on the time sequence evolution modeling of the bill items according to claim 1, characterized in that, The calculation process of the cost fluctuation correlation tracing algorithm in S3 includes: extracting the correlation data of cost components in the time-series evolution parameters of the list items, and determining the set of cost influencing factors. ,in Indicates the first Several cost-influencing factors were identified, including material price fluctuation data, labor cost changes data, and machinery usage fee adjustment data; a cost fluctuation correlation and source tracing expression was constructed. ,in Indicates list items Total cost Represents time sequence nodes Next The values ​​of each influencing factor, Impact Factor The time-series average value, For time sequence nodes Next project Cost data, For the project The time-series average cost, The correlation strength coefficient, Impact Factor For the project The specific correlation correction coefficient; the correlation strength value between each cost influencing factor and the total cost is calculated based on the correlation source tracing expression. By combining the priority parameters of time-series nodes in the time-series evolution model of the list items, a cost fluctuation source tracing correlation model is established with the correlation strength value as the core and the time-series nodes as the dimensions, so as to locate the source of cost fluctuation.

4. The method for identifying engineering cost deviations based on the temporal evolution modeling of list items as described in claim 1, characterized in that, The modeling process of the cost trend adaptive extrapolation model in S4 includes: extracting the basic cost data of each time series node from the feature data of the time series nodes. Temporal evolution rate and trend influence coefficient The trend influence coefficient is obtained by training historical data from the time-series evolution model of the list items; an initial expression for the time-series evolution trend of cost is constructed. ,in For the first Cost prediction values ​​for each time series node, The time interval between adjacent time nodes. For the front The average cost of each node. The attenuation coefficient is introduced; an adaptive adjustment factor is introduced. ,in Describe the standard deviation calculation function and construct the optimized adaptive extrapolation expression for cost trends. ,in For adaptive correction coefficients, The change in the temporal evolution rate; adaptive correction coefficients for different stages based on the stage division results of the temporal evolution modeling of the inventory items. and attenuation coefficient Dynamic adjustments are made to adapt the model to the cost change characteristics at different evolution stages; the cost prediction values ​​at each time series node are... By comparing the model with actual observations and optimizing the model parameters through a feedback adjustment mechanism, a cost trend adaptive extrapolation model that combines time-series adaptability and prediction accuracy is finally constructed, generating a continuous cost time-series evolution trend curve.

5. The engineering cost deviation identification method based on the time sequence evolution modeling of the bill items according to claim 1, characterized in that, The intelligent cost analysis system for the bill of quantities adopts a multi-core parallel processing architecture, including an FPGA chipset and a GPU acceleration module. The FPGA chipset is used for rapid extraction and preliminary classification of engineering quantity data, while the GPU acceleration module supports parallel computation of multiple algorithms. The system's analysis speed is no less than [a certain value]. The sampling frequency of the time-series evolution parameters of the listed items is synchronized with the project construction schedule, and the sampling interval is dynamically adjusted according to the construction stage. The parameter storage adopts a distributed database architecture, supporting real-time data writing and querying; the iteration step size coefficient of the engineering quantity time-series progressive deviation iterative algorithm. The value range is 0.01-0.15, and the iterative stability coefficient is... The value ranges from 0.001 to 0.05, representing the correlation strength coefficient of the cost fluctuation correlation tracing algorithm. The attenuation coefficient of the cost trend adaptive extrapolation model has a value range of 0.8-1.

2. The value range is 0.1-0.3, and the adaptive correction coefficient is used. The value range is 0.9-1.1, and all parameters support custom configuration based on project type.

6. The engineering cost deviation identification method based on bill item timing evolution modeling according to claim 1, characterized in that, S2 includes the following steps: S21, based on the dynamic change data of engineering quantities output by the intelligent analysis system for bill of quantities cost, classify the data according to the sub-project categories of the bill of quantities items, extract the time-series data of engineering quantities under each category, and clarify the construction process association information corresponding to each time-series; S22, identify outliers in the classified engineering quantity time-series data, mark data points that exceed the reasonable change range based on the local change trend of the time-series and the correlation of adjacent data, and retain valid data for algorithm calculation; S23, import the valid data into the engineering quantity time-series progressive deviation iteration algorithm in the order of time-series nodes, set the initial iteration parameters of the algorithm, the initial iteration parameters include the upper limit of the number of iterations, the convergence threshold, and the time-series weight allocation rules, the time-series weight allocation rules are determined based on the construction importance parameters of the bill of quantities item time-series evolution model; S24, obtain the engineering quantity iteration results of each time-series node through algorithm calculation, combine the construction process association information, construct the engineering quantity time-series progressive evolution model including process connection characteristics, and output the engineering quantity benchmark value and deviation fluctuation range of each time-series node.

7. The method for identifying engineering cost deviations based on the temporal evolution modeling of list items as described in claim 1, characterized in that, S3 includes the following steps: S31, perform hierarchical processing on the cost component data, classifying it into material, labor, and machinery cost data according to component type, and extracting the time-series change characteristics and inter-category correlation data of each type of data; S32, based on the hierarchically processed data, determine the time-series fluctuation amplitude and frequency of each cost component, and mark the significant fluctuation time-series nodes by combining the time node division parameters of the time-series evolution model of the list items; S33, input the cost data and fluctuation characteristic parameters of the marked time-series nodes into the cost fluctuation correlation tracing algorithm, and obtain the correlation degree data between each component and the total cost through algorithm calculation, and establish a component correlation matrix; S34, based on the component correlation matrix and the location results of significant fluctuation nodes, construct a cost fluctuation tracing correlation model. The model supports sorting and outputting possible fluctuation sources by correlation degree, and marking the influence range and propagation path of each source.

8. The engineering cost deviation identification method based on bill item timing evolution modeling according to claim 1, characterized in that, S4 includes the following sub-steps: S41, expanding the dimensionality of the time-series node feature data by extracting, in addition to the basic cost data, the construction progress completion rate, resource input intensity, and external environmental impact parameters corresponding to each node, forming a multi-dimensional feature dataset; S42, dividing the multi-dimensional feature dataset into a training set and a validation set according to time order, wherein the training set is used for parameter training of the cost trend adaptive extrapolation model, and the validation set is used for model performance verification. The division ratio is dynamically determined based on the total number of nodes in the time-series evolution model of the list items; S43, optimizing and adjusting the initial parameters of the model using the training set data to enable the model to accurately fit the evolution trend of historical time-series data, and verifying the prediction error of the model using the validation set data. If the error exceeds the preset range, the parameters are readjusted. S44, based on the trained model, extrapolates the cost trend of time series nodes and generates a cost time series evolution trend curve including predicted values, confidence intervals and trend change rates. The curve data is stored in the order of time series nodes for fusion modeling.

9. The engineering cost deviation identification method based on bill item timing evolution modeling according to claim 1, characterized in that, S5 includes the following steps: S51, establishing a unified data interface for the time-series evolution modeling of the bill of quantities items, converting the output data of the time-series progressive evolution model of engineering quantities, the cost fluctuation source correlation model, and the cost time-series evolution trend curve into a unified format to ensure the consistency of the data structure; S52, based on the unified format data, extracting the output parameters of each model, including the benchmark value of engineering quantities, the correlation strength of cost influencing factors, and the cost prediction value, and constructing a comprehensive feature matrix; S53, setting the fusion weight of each output parameter, the fusion weight is dynamically allocated based on the recognition accuracy of each model in historical data, and the allocation process is combined with the stage characteristics of the time-series evolution modeling of the bill of quantities items; S54, performing calculations on the comprehensive feature matrix through a weighted fusion algorithm to construct a comprehensive model for identifying engineering cost deviations that includes multi-dimensional identification indicators, and the model supports adjusting the priority of identification indicators according to the characteristics of different engineering types.

10. The construction cost deviation identification method based on bill item timing evolution modeling according to any one of claims 1-9, characterized in that, This method is implemented through an engineering cost deviation identification platform based on the time-series evolution modeling of bill of quantities items. It includes: an intelligent extraction module for time-series parameters of bill of quantities items, a time-series progressive evolution modeling module for engineering quantities, a cost fluctuation correlation tracing analysis module, a cost trend adaptive inference module, a multi-model fusion deviation identification module, and a deviation result visualization output module. The intelligent extraction module for time-series parameters of bill of quantities items establishes a data communication connection with the intelligent bill of quantities cost analysis system to extract the time-series evolution parameters of the bill of quantities items and transmit them to the time-series progressive evolution modeling module for engineering quantities and the cost fluctuation correlation tracing analysis module. The time-series progressive evolution modeling module for engineering quantities processes the received parameters based on the time-series progressive deviation iterative algorithm for engineering quantities, generates a time-series progressive evolution model for engineering quantities, and sends it to the multi-model fusion deviation identification module. The cost fluctuation correlation and source analysis module generates a cost fluctuation correlation model through a cost fluctuation correlation and source analysis algorithm, and transmits the model data to the multi-model fusion deviation identification module. The cost trend adaptive inference module processes the time-series node feature data using the cost trend adaptive inference model, generates a cost time-series evolution trend curve, and sends it to the multi-model fusion deviation identification module. The multi-model fusion deviation identification module receives the output data from the above three modules, performs fusion calculations based on the time-series evolution modeling framework of the bill of quantities items, constructs a comprehensive model for engineering cost deviation identification, and performs deviation identification. The deviation result visualization output module is connected to the multi-model fusion deviation identification module and is used to output the cost deviation results and evolution trajectory data in the form of time-series charts, correlation heatmaps, and trend curves for intuitive display of deviation data.

Images