A road construction material cost control system

Abstract

The present application relates to the technical field of cost management, in particular to a road construction material cost control system, the system comprises: material consumption data acquisition module, material dynamic consumption benchmark generation module, material consumption cost anomaly discrimination module, material space risk linkage identification module.In the present application, the material consumption data is collected in real time by the field equipment, combined with the progress and equipment operation and other multi-dimensional elements, the weighted moving average and the multivariate linear regression model are used to dynamically correct the consumption benchmark, the accurate correction of the material plan consumption is realized, the actual consumption cost and the dynamic benchmark interval are used to judge and find the abnormality in time, the space aggregation analysis is used to identify the material consumption risk in the same area, the single-point monitoring is changed to multi-dimensional dynamic judgment and space linkage identification, the scientificity and timeliness of cost control are improved, the potential abnormality is actively warned, the human error and data lag are reduced, the reasonable allocation of resources and the fine management of cost are assisted, and the efficiency of fund utilization and the risk prevention and control ability are improved.

Description

Technical Field

[0001] This invention relates to the field of cost management technology, and in particular to a road construction material cost control system. Background Technology

[0002] Cost management technology involves monitoring, accounting for, and optimizing various resource consumption and economic expenditures of enterprises or engineering projects. It mainly includes core aspects such as cost accounting, cost budgeting, cost forecasting, cost analysis, and cost control, covering the entire process from cost data collection, aggregation, and allocation to decision support. It is widely used in manufacturing, construction, transportation, and other industries, aiming to improve the efficiency of capital utilization and achieve optimal resource allocation. Among these, the traditional road construction material cost control system refers to a management system that statistically controls the costs of procurement, transportation, storage, and consumption of materials required during road construction. Traditional road construction material cost control typically involves manually compiling material inbound and outbound data, manually preparing cost ledgers, and periodically summarizing material consumption and comparing budgets with actual expenditures.

[0003] Existing technologies rely heavily on manual statistics and ledger compilation in the process of material cost management. Data collection and summarization are often delayed, making it difficult to reflect the actual consumption status in a timely manner. Manual operations are prone to omissions or misjudgments, which can lead to a decrease in data accuracy. They cannot dynamically adjust consumption benchmarks, lack the ability to provide real-time feedback and adjustments based on actual site conditions, and have a weak ability to identify the spatial correlation between abnormal material consumption and risks. In situations with large material usage, long project cycles, or dispersed work sites, problems such as material overconsumption and cost deviations from the budget are likely to occur, affecting management efficiency and increasing project risks. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a road construction material cost control system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a road construction material cost control system comprising:

[0006] The material consumption data acquisition module is used to obtain the actual material consumption, planned consumption, progress percentage and equipment runtime through field equipment, correct the planned material consumption using a weighted moving average model, generate the weighted corrected material consumption and transmit it to the material dynamic consumption benchmark generation module.

[0007] The material dynamic consumption benchmark generation module is used to obtain the weighted and corrected material consumption, meteorological parameters, and material unit price, correct the material consumption using a multiple linear regression model, generate a material dynamic consumption cost benchmark range, and transmit it to the material consumption cost anomaly detection module.

[0008] The material consumption cost anomaly detection module calculates the actual material consumption cost based on the actual material consumption and the material unit price, calls the material dynamic consumption cost benchmark range for range judgment, generates a material consumption cost anomaly detection result, and transmits the material consumption cost anomaly detection result to the material spatial risk linkage identification module.

[0009] The material spatial risk linkage identification module obtains the abnormal judgment results of the material consumption cost and the material consumption cost of the same construction section, and uses the K-means clustering algorithm to perform spatial clustering analysis to output material spatial linkage risk information.

[0010] As a further aspect of the present invention, the weighted and corrected material consumption amount specifically refers to the corrected consumption amount and the correction basis parameters; the material dynamic consumption cost benchmark range includes the lower limit of cost, the upper limit of cost, and the range of fluctuation; the material consumption cost anomaly judgment result specifically refers to the anomaly state, the anomaly type, and the associated risk; and the material spatial linkage risk information includes the degree of spatial aggregation, the risk distribution type, and the scope of influence.

[0011] As a further aspect of the present invention, the material consumption data acquisition module includes:

[0012] The on-site data acquisition submodule collects the actual material consumption, planned consumption, progress percentage, and equipment runtime of on-site equipment. It calculates the consumption difference based on the actual material consumption and the planned consumption, integrates the progress percentage and the equipment runtime, and associates and classifies the data of different time periods to establish the basic data volume of on-site material consumption.

[0013] The consumption weight determination submodule calls the on-site material consumption basic data, extracts the equipment running time of multiple cycles according to the time proximity principle in the weighted moving average model, and uses the running time as the core factor for weight allocation, assigns differentiated weight values ​​to the material consumption data of multiple cycles, normalizes all weight values, and generates moving average weight coefficients.

[0014] The weighted moving average correction submodule calls the planned material consumption within the on-site material consumption basic data, and applies the weighted moving average model according to the moving average weight coefficient. It multiplies the planned consumption of multiple periods by the corresponding weight coefficient, and sums the calculated products of all periods to obtain the weighted corrected material consumption.

[0015] As a further aspect of the present invention, the material dynamic consumption benchmark generation module includes:

[0016] The regression variable screening submodule obtains the weighted adjusted material consumption, meteorological parameters and material unit price. Based on the multiple linear regression model, the weighted adjusted material consumption is used as the dependent variable, and the meteorological parameters and material unit price are used as independent variables. By calculating the correlation coefficient between the independent variables and the dependent variable, the independent variables with a correlation degree exceeding the preset correlation degree threshold are screened to obtain the core influencing factors of the multiple regression model.

[0017] The dynamic consumption correction submodule calls the core influencing factors of the multiple regression model, constructs a regression equation based on the multiple linear regression model, multiplies the current values ​​of multiple core influencing factors with the corresponding regression coefficients in the regression equation, sums them, and adds them to the constant term to correct the weighted material consumption and generate the environmental factor corrected material consumption.

[0018] The cost benchmark interval calculation submodule calculates the cost center value by multiplying the material consumption and material unit price after the environmental factors are corrected, and calculates the confidence interval of the cost center value at a specified confidence level based on the prediction standard error of the multiple linear regression model. The upper and lower limits of the interval are used as fluctuation boundaries to establish the dynamic material consumption cost benchmark interval.

[0019] As a further aspect of the present invention, the material consumption cost anomaly detection module includes:

[0020] The actual cost accounting submodule obtains the actual consumption of the material and the unit price of the material, multiplies the consumption value and the unit price value, and sums up all the product results to calculate the total cost within the period and establish the material cost incurred in the current period.

[0021] The cost range discrimination submodule calls the current material cost and the material dynamic consumption cost benchmark range, compares the value of the current material cost with the upper and lower limits of the material dynamic consumption cost benchmark range. If the cost value exceeds the upper and lower limits of the range, it will be marked as an abnormal state; if it is within the range, it will be marked as a normal state, and an abnormal material consumption cost discrimination result will be generated.

[0022] As a further aspect of the present invention, the material space risk linkage identification module includes:

[0023] The cost data integration submodule obtains the material consumption cost anomaly judgment result and the material consumption cost of the same construction section work point. It integrates the two types of data with the construction work point as the basic unit, performs numerical processing on the anomaly judgment result to form a risk identifier, and forms a feature vector with the corresponding work point material consumption cost data. It merges the feature vectors of all work points to establish a work point material consumption feature matrix.

[0024] The spatial clustering analysis submodule calls the material consumption feature matrix of the work site, uses the K-means clustering algorithm to initialize the cluster centers according to the preset number of clusters K, and iteratively calculates the Euclidean distance between multiple work site data and multiple cluster centers. The work site is assigned to the nearest cluster center and the center position is updated until the position of the cluster center no longer changes, thus obtaining the spatial clustering assignment of material consumption.

[0025] The risk linkage identification submodule, based on the spatial clustering of material consumption, counts the number of work points with risk identifiers within each cluster and calculates their proportion within the cluster. It then compares the proportion with a set risk identification benchmark value, filters out clusters whose proportions exceed the risk identification benchmark value, and generates material spatial linkage risk information.

[0026] As a further aspect of the present invention, the consumption weight determination submodule is specifically implemented using the following formula:

[0027] ;

[0028] Calculate the moving average weighting coefficient;

[0029] in, Indicates the first The moving average weighting coefficient for each period, Indicates the first The equipment runtime per cycle, This represents the total number of periods used to calculate the weighted moving average. Indicates all The sum of the operating time of the equipment in each cycle;

[0030] The dynamic consumption correction submodule specifically uses the following formula:

[0031] ;

[0032] Calculate the material consumption after adjusting for the aforementioned environmental factors;

[0033] in, This indicates the material consumption after adjusting for the aforementioned environmental factors. This represents the constant term in the regression equation. Indicates the first The regression coefficients corresponding to the core influencing factors Indicates the first The current values ​​of the core impact factors, This represents the total number of core influencing factors in the multiple regression model. This represents the random error term of the model.

[0034] As a further aspect of the present invention, the meteorological parameters obtained by the regression variable screening submodule include the daily average temperature, daily cumulative precipitation, daily average wind speed and daily average air humidity of the construction area;

[0035] The anomaly types in the material consumption cost anomaly identification results include cost overrun anomalies and cost savings anomalies.

[0036] Among them, the cost overrun anomaly is analyzed by tracing the source of the deviation between the actual value and the planned value of the material unit price and the deviation between the actual material consumption and the material consumption after the environmental factors are corrected, and iteratively classified into unit price fluctuation anomaly, usage overconsumption anomaly and comprehensive impact anomaly.

[0037] The risk identification benchmark value used by the risk linkage identification submodule is obtained by acquiring historical completed project data that has a similarity to the current project in terms of engineering geology, construction scale, and climate zoning exceeding a preset similarity threshold, statistically analyzing the proportion distribution of abnormal work points within risk clusters in historical projects, and generating the risk identification benchmark value based on the 90th percentile of the distribution.

[0038] As a further aspect of the present invention, the spatial clustering degree in the material spatial linkage risk information is quantified by calculating the average profile coefficient of each cluster, and the average profile coefficient is used as a measure of the spatial clustering degree, wherein the average profile coefficient is calculated using the formula:

[0039] calculate;

[0040] in, Represents cluster The average profile coefficient, Represents cluster Total number of work sites within the area Represents cluster A single work site within, Representative work site The average Euclidean distance between the work site and all other work sites within the same cluster. Representative work site The minimum average Euclidean distance from the point in the node to all other clusters. represent and The larger value in;

[0041] The risk distribution type is determined by calculating the minimum boundary rectangle of the spatial coordinates of all work points within each selected cluster and solving the aspect ratio of the rectangle. If the aspect ratio is greater than a preset shape ratio threshold, the risk distribution type is determined as a corridor-type risk; otherwise, it is determined as a regional-type risk.

[0042] The scope of influence is defined by constructing the minimum convex hull polygon of the spatial coordinates of all work sites within the selected cluster, and the set of geographic coordinate vertices of the minimum convex hull polygon is used as the boundary data of the scope of influence. At the same time, the area of ​​the polygon is calculated as a quantitative indicator of the scope of influence.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0044] In this invention, material consumption data is collected in real time by on-site equipment. Combined with multi-dimensional factors such as progress and equipment operation, and using weighted moving average and multiple linear regression models to continuously and dynamically adjust the consumption benchmark, the planned material consumption can be accurately corrected. Furthermore, by judging the interval between actual consumption cost and dynamic benchmark, abnormal situations can be detected in a timely manner. With the help of spatial cluster analysis, material consumption risks in the same area can be effectively identified. This realizes the upgrade from single-point data monitoring to multi-dimensional dynamic judgment and spatial risk linkage identification, effectively improving the scientificity and timeliness of cost control. It can proactively discover and warn of potential anomalies, reduce the impact of human statistical errors and data lag, help rationally allocate resources and refine cost management, and improve the overall capital utilization efficiency and risk prevention and control capabilities. Attached Figure Description

[0045] Figure 1 This is a flowchart of the system structure of the present invention;

[0046] Figure 2 This is a flowchart of the material consumption data acquisition module of the present invention;

[0047] Figure 3 This is a flowchart of the material dynamic consumption benchmark generation module of the present invention;

[0048] Figure 4 This is a flowchart of the material consumption cost anomaly detection module of the present invention;

[0049] Figure 5 This is a flowchart of the material space risk linkage identification module of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0051] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0052] Please see Figure 1 and Figure 2 The present invention provides a technical solution: a road construction material cost control system comprising:

[0053] The material consumption data acquisition module is used to obtain the actual material consumption, planned consumption, progress percentage and equipment runtime through field equipment, correct the planned material consumption using a weighted moving average model, generate the weighted corrected material consumption and transmit it to the material dynamic consumption benchmark generation module.

[0054] The material consumption data acquisition module includes:

[0055] The on-site data acquisition submodule collects the actual material consumption, planned consumption, progress percentage, and equipment runtime of on-site equipment. It calculates the consumption difference based on the actual and planned material consumption, integrates the progress percentage and equipment runtime, and associates and classifies the data of different time periods to establish the basic data volume of on-site material consumption.

[0056] In a specific implementation scenario, focusing on the consumption management of asphalt concrete (AC-25) in the third section (chainage K10+000 to K15+000) of a highway project, the execution unit of the field data acquisition submodule interfaces with the central control system of the asphalt mixture mixing plant in this section. During a designated data acquisition period, specifically the first week of July 2025, this submodule obtains various data by parsing the production log files output by the control system.

[0057] This submodule first retrieves the weekly construction plan pre-entered into the system by the construction management department, which specifies the planned material consumption. This week's plan is to pave 5 kilometers of road surface, and based on the design drawings, the total planned consumption of AC-25 is precisely calculated to be 10,000 tons. Simultaneously, the submodule continuously reads the actual material consumption in real time through the sensor data interface of the mixing plant's control system. As of the midpoint of the cycle (Wednesday at 5 PM), the system recorded a cumulative total of 4,850 tons of AC-25 produced and shipped. The project progress data entered into the system by the project supervision engineer shows that road paving is 45% complete; this is the progress percentage. In addition, the submodule extracts the cumulative operating time of the mixing plant's programmable logic controller (PLC) since the start of construction from the PLC's operating records, recording it as 25 hours.

[0058] Next, the consumption variance is calculated. This calculation process is as follows: First, based on the percentage of progress completed, the theoretically planned consumption for the current node is calculated, which is 10,000 tons multiplied by 45%, resulting in 4,500 tons. Then, the actual material consumption (4,850 tons) is compared with this theoretical planned consumption, calculated using the formula (actual material consumption / theoretical planned consumption) - 1. Substituting the values, the calculation is (4,850 / 4,500) - 1, resulting in 0.078. This variance is a dimensionless positive value, and its magnitude directly reflects the degree to which consumption exceeds the plan.

[0059] Finally, this submodule integrates all the collected and calculated data items to form a structured periodic data record, specifically: {Period: First week of July 2025, Material: AC-25, Planned consumption: 10,000 tons, Actual consumption: 4,850 tons, Progress percentage: 45%, Equipment runtime: 25 hours, Consumption variance: 0.078}. The system uses a four-week period as a complete monthly assessment cycle, associating and categorizing the data records generated each week to establish a basic data volume of on-site material consumption containing four weeks of data, providing data support for subsequent weight determination.

[0060] The consumption weight determination submodule calls the on-site material consumption basic data, extracts the equipment running time of multiple cycles according to the time proximity principle in the weighted moving average model, and uses the running time as the core factor for weight allocation, assigns differentiated weight values ​​to the material consumption data of multiple cycles, normalizes all weight values, and generates moving average weight coefficients.

[0061] The consumption weight determination submodule uses the following formula:

[0062] ;

[0063] Calculate the moving average weighting coefficient;

[0064] in, Indicates the first The moving average weighting coefficient for the nth period represents the weighting coefficient of the nth period. The equipment runtime per cycle, This represents the total number of periods used to calculate the weighted moving average. Indicates all The sum of the operating time of the equipment in each cycle;

[0065] This submodule calls upon the on-site material consumption baseline data established in the preceding steps, which includes the equipment runtime for four consecutive cycles. The equipment runtime for these four cycles are as follows: Week 1 ( ) = 25 hours, second week ( ) = 35 hours, third week ( ) = 30 hours, fourth week ( =40 hours. These data constitute the core factors for weight allocation. This submodule calculates the moving average weight coefficient for each period based on the following formula:

[0066] ;

[0067] In the above formula, Representing the The moving average weighting coefficient for the nth period, which is a unitless scale value used to quantify the nth period. The importance of data from each period in subsequent weighted correction calculations. (Subscript) Used to uniquely identify a specific period, its value ranges from 1 to... . Representing the The equipment runtime per cycle is a parameter obtained through direct data collection, measured in hours, and reflects the absolute intensity of production activities within that cycle. This represents the total number of periods used to calculate the weighted moving average. In this embodiment, the rolling assessment period set by the project management department is four weeks. The value of is 4. The representative will be from the first cycle to the second cycle. The runtime of all devices in each cycle ( The total device runtime over a specified time span is calculated by summing the device runtime of a single cycle. Divide the total runtime of all devices in a given period by the total runtime of that period to obtain the percentage of runtime of that period in the total runtime. This percentage is defined as the moving average weighting coefficient for that period.

[0068] Based on the above settings, the total number of cycles The device runtime for each cycle is 4. The values ​​have been determined, and are respectively Hour, Hour, Hour, Hours. The first step is to calculate the sum of the runtimes of all cycle devices. : Hours. The second step is to calculate the moving average weighting coefficient for each period. The moving average weighting coefficient for the first period The moving average weighting coefficient for the second period The moving average weighting coefficient for the third period The moving average weighting coefficient for the 4th period After the calculation is completed, a normalization check is performed, and all weight coefficients are summed: The sum of the calculated results equals 1, which meets the normalization requirement. The final moving average weight coefficients are organized into an array: [0.192, 0.269, 0.231, 0.308].

[0069] The weighted correction submodule calls the planned material consumption within the basic data of on-site material consumption, and applies the weighted moving average model according to the moving average weight coefficient. It multiplies the planned consumption of multiple periods by the corresponding weight coefficient, and sums the calculated products of all periods to obtain the weighted corrected material consumption.

[0070] The weighted adjusted material consumption is specifically the adjusted consumption and the adjustment basis parameters.

[0071] This submodule calls the planned material consumption for each period from the on-site material consumption baseline data and applies the moving average weight coefficient array generated in the previous step. The planned material consumption for the four periods are as follows: Week 1 ( = 9,500 tons, second week ( ) = 11,000 tons, third week ( ) = 10,500 tons, fourth week ( = 12,000 tons.

[0072] This submodule applies a weighted moving average model to calculate the planned consumption for each period ( ) and its corresponding moving average weighting coefficient ( Perform multiplication, then sum the products of calculations from all periods. The formula for this process is: Weighted adjusted material consumption = Substituting specific values: Weighted adjusted material consumption = = = ton.

[0073] The calculated result, 10904.5 tons, is the weighted adjusted material consumption. This value is defined as the adjusted consumption. The array of planned consumption for each period [9500, 11000, 10500, 12000] and the array of moving average weighting coefficients [0.192, 0.269, 0.231, 0.308] used for the calculation together constitute the adjustment basis parameters and are stored together for verification.

[0074] Please see Figure 1 and Figure 3 The material dynamic consumption benchmark generation module is used to obtain the weighted and corrected material consumption, meteorological parameters, and material unit price, and uses a multiple linear regression model to correct the material consumption, generate the material dynamic consumption cost benchmark range, and transmit it to the material consumption cost anomaly detection module.

[0075] The material dynamic consumption baseline generation module includes:

[0076] The regression variable screening submodule obtains the weighted adjusted material consumption, meteorological parameters and material unit price. Based on the multiple linear regression model, the weighted adjusted material consumption is used as the dependent variable, and the meteorological parameters and material unit price are used as independent variables. By calculating the correlation coefficient between the independent variables and the dependent variable, the independent variables with a correlation degree exceeding the preset correlation degree threshold are screened to obtain the core influencing factors of the multiple regression model.

[0077] The meteorological parameters obtained by the regression variable screening submodule include the daily average temperature, daily cumulative precipitation, daily average wind speed, and daily average air humidity of the construction area.

[0078] After obtaining the weighted adjusted material consumption of 10,904.5 tons, this module begins execution. The regression variable screening submodule extracts consumption data of the same material (AC-25), meteorological parameters, and market price data for the past 12 consecutive months from the project's historical database to identify factors that have a decisive impact on material consumption.

[0079] Table 1 Historical Data Table

[0080] ;

[0081] Table 1 shows the weighted adjusted material consumption (as dependent variable Y) for the past 12 data periods, along with four meteorological parameters and one market parameter (as independent variable X). The submodule calculates the Pearson correlation coefficient between each independent and dependent variable. This calculation is performed by taking the covariance of the two variables and dividing it by the product of their standard deviations. The calculated correlation coefficients are: daily average temperature -0.68; material unit price -0.55; daily cumulative precipitation -0.45; daily average wind speed -0.15; and daily average air humidity -0.28.

[0082] Next, a preset correlation threshold is set. To ensure objectivity and applicability, the system selected five completed projects. These projects, along with the current project, shared similarities in three core dimensions: engineering geological conditions, construction technology, and management level. All similarities were assessed using expert scoring and calculated to exceed 0.85. The system retrieved complete construction data from these five projects and repeated the correlation analysis process for each project, obtaining a set of correlation coefficients between all independent and dependent variables. By statistically analyzing the distribution of the absolute values ​​of all correlation coefficients in this set, it was found that the absolute values ​​of the correlation coefficients of variables ultimately included in the regression model by successful projects were all at the top of the distribution. All these absolute values ​​of correlation coefficients were sorted in descending order, and the value at the 70th percentile was determined to be 0.5. Therefore, the preset correlation threshold for this project was set to 0.5.

[0083] The absolute values ​​of the calculated correlation coefficients for each variable are compared with a threshold of 0.5:

[0084] The absolute value of the correlation coefficient of the daily average temperature is |0.68|, which is considered to be greater than 0.5, so it is selected and retained.

[0085] The absolute value of the correlation coefficient of the unit price of materials is |0.55|, which is considered greater than 0.5 and is therefore selected for retention.

[0086] The absolute value of the correlation coefficient of the daily cumulative precipitation is |-0.45|, which is 0.45. It is determined to be less than 0.5 and is therefore discarded.

[0087] The absolute value of the correlation coefficient of the daily average wind speed is |-0.15|, which is 0.15. It is judged to be less than 0.5 and is therefore rejected.

[0088] The absolute value of the correlation coefficient for daily average air humidity is |-0.28|, which is 0.28, and is therefore considered less than 0.5 and is thus discarded. Through this screening process, the core influencing factors of the multiple regression model are finally determined to be: daily average temperature and material unit price.

[0089] The dynamic consumption correction submodule calls the core influencing factors of the multiple regression model, constructs a regression equation based on the multiple linear regression model, multiplies the current values ​​of multiple core influencing factors with the corresponding regression coefficients in the regression equation, sums them, and adds them to the constant term to correct the weighted material consumption and generate the environmental factor corrected material consumption.

[0090] The dynamic consumption correction submodule uses the following formula:

[0091] ;

[0092] Calculate material consumption after environmental factor correction;

[0093] in, This indicates the amount of materials consumed after adjusting for environmental factors. This represents the constant term in the regression equation. Indicates the first The regression coefficients corresponding to the core influencing factors Indicates the first The current values ​​of the core impact factors, This represents the total number of core influencing factors in a multiple regression model. Represents the random error term of the model;

[0094] This submodule calls the selected core influencing factors (daily average temperature, material unit price) and uses historical data from Table 1 to construct a multiple linear regression equation using the least squares method. This construction process involves finding a set of coefficients that minimizes the sum of squares of the differences between the actual values ​​and the model predictions for all historical data points. The resulting regression equation is: Among them, the constant term tons, average daily temperature ( regression coefficients tons / ℃, material unit price ( regression coefficients Tons (RMB / ton).

[0095] This submodule calculates the material consumption after environmental factor correction based on the following formula:

[0096] ;

[0097] In the above formula, This is a forecast of material consumption after taking into account environmental and market factors, expressed in tons. It is the constant term of the regression equation, with a value of 4515.8 tons, representing the material consumption under the baseline condition that all core influencing factors are zero. This represents the linear summation of the effects of all core influencing factors, where This represents the total number of core impact factors, which is 2 in this example. It is the first The regression coefficients corresponding to the core influencing factors quantify the impact of each unit change in that factor on... The average degree of impact. It is the first The current values ​​of the core influencing factors are the instantaneous parameters input when making predictions. This is the model's random error term, representing the combined effect of all other random factors that the model fails to explain. Its expected value is 0 when performing point prediction calculations, therefore it is not involved in specific numerical calculations. The overall formula's operational logic is based on a base consumption ( Starting from ), plus the current values ​​of each core influencing factor ( ) and their corresponding influence coefficients ( The calculated correction amount ultimately yields a dynamic correction consumption amount that integrates multiple key factors. In this calculation, the regression coefficients... The use of units ensures that the units of all product results are consistent.

[0098] Entering the current forecast period (the fifth week of July 2025), the system obtains the forecast daily average temperature of 31.5℃ from the integrated meteorological data service and the expected unit price of materials for this period of 460 yuan / ton from the materials department's procurement plan. Therefore, the current values ​​of the core influencing factors are as follows: ℃, Yuan / ton. Total number of core influencing factors. Substitute these values ​​into the regression equation to calculate: Tons. The calculated result of 13697.7 tons represents the material consumption after adjusting for environmental factors.

[0099] The cost benchmark interval calculation submodule calculates the cost center value by multiplying the material consumption and material unit price after environmental factor correction, and calculates the confidence interval of the cost center value at a specified confidence level based on the prediction standard error of the multiple linear regression model. The upper and lower limits of the interval are used as fluctuation boundaries to establish the dynamic consumption cost benchmark interval of materials.

[0100] The benchmark range for dynamic material consumption costs includes the lower limit of cost, the upper limit of cost, and the range of fluctuation.

[0101] This submodule first multiplies the calculated material consumption (13697.7 tons) adjusted for environmental factors by the current material unit price (460 yuan / ton) to obtain the cost center value: Cost center value = When performing multiple linear regression analysis, the model also outputs a prediction standard error, which quantifies the uncertainty of the model's predicted values. This value, obtained from the regression analysis report, is 250 tons.

[0102] Next, a confidence level is set. Following standard practice in cost control within the construction industry, a 95% confidence level is chosen. Consulting the standard normal distribution table, the Z-value for a 95% confidence level is 1.96. This value is used to define the width of the interval. Then, the confidence interval for the cost center value is calculated. This process first calculates the confidence interval for the consumption itself: Interval half-width = Lower limit of the confidence interval for consumption = Tons. Upper limit of the confidence interval for consumption = Tons. Multiply the upper and lower limits of the consumption range by the material unit price to obtain the cost confidence interval: lower limit of cost = Cost ceiling = Range of fluctuation = Ultimately, the established benchmark range for dynamic material consumption costs is [6,075,542 yuan, 6,526,342 yuan]. This range includes three specific data points: the lower limit of cost, the upper limit of cost, and the range of fluctuation, which together define the reasonable fluctuation range of material costs under the current conditions.

[0103] The cost range discrimination submodule calls the current period material cost and the material dynamic consumption cost benchmark range, compares the current period material cost with the upper and lower limits of the material dynamic consumption cost benchmark range. If the cost exceeds the upper and lower limits of the range, it will be marked as an abnormal state; if it is within the range, it will be marked as a normal state, and an abnormal material consumption cost discrimination result will be generated.

[0104] The specific results of the material consumption cost anomaly identification are the anomaly status, anomaly type, and associated risks.

[0105] The anomaly types in the material consumption cost anomaly identification results include cost overrun anomalies and cost savings anomalies;

[0106] Among them, cost overrun anomalies are analyzed by tracing the source of the deviation between the actual value and the planned value of the material unit price and the deviation between the actual material consumption and the material consumption after environmental factor correction, and iteratively classified into unit price fluctuation anomalies, usage overconsumption anomalies and comprehensive impact anomalies.

[0107] After generating the dynamic cost baseline range, this module begins execution. The actual cost accounting submodule retrieves all purchase and outbound records related to AC-25 materials from the project's materials management department's database for a specified period (the fifth week of July 2025). Three purchase batches occurred during this period, with the following quantities and unit prices: the first batch of 10,000 tons at 458 yuan / ton; the second batch of 3,000 tons at 461 yuan / ton; and the third batch of 1,500 tons at 465 yuan / ton. Simultaneously, the actual material consumption records from the mixing plant show that the actual material consumption for this period was 14,300 tons.

[0108] This submodule first calculates the weighted average purchase price for the period to accurately reflect the actual unit cost: Total purchase cost = = Total purchase quantity = Weighted average unit price = Then, multiply the actual material consumption for that period (14,300 tons) by the weighted average unit price (459.34 yuan / ton) to calculate the total cost for the period. Current period material costs = This figure of 6,568,562 yuan was established as the material cost incurred in the current period.

[0109] Please see Figure 1 and Figure 4 The material consumption cost anomaly detection module calculates the actual material consumption cost based on the actual material consumption and the unit price of the material, calls the dynamic material consumption cost benchmark range for range judgment, generates the material consumption cost anomaly detection result, and transmits the material consumption cost anomaly detection result to the material spatial risk linkage identification module.

[0110] The material consumption cost anomaly detection module includes:

[0111] The actual cost accounting submodule obtains the actual consumption of materials and the unit price of materials, multiplies the consumption value and the unit price value, and sums up all the product results to calculate the total cost within the period and establish the material costs incurred in the current period.

[0112] This submodule retrieves the current period's material costs calculated in the previous step (6,568,562 yuan) and obtains the dynamic material consumption cost benchmark range from the previous module, which is [6,075,542 yuan, 6,526,342 yuan]. This submodule performs a numerical comparison operation, comparing the current period's material costs with the upper and lower limits of the range. The comparison process is as follows: the current period's material costs of 6,568,562 yuan are greater than the upper limit of 6,526,342 yuan. Because the cost value exceeds the upper limit of the range, this submodule marks this state as an abnormal state.

[0113] Next, a source analysis is performed on this anomaly to determine its specific anomaly type. This process is a further processing of the judgment results. First, because the actual cost is higher than the upper limit of the benchmark range, the anomaly type is initially determined to be a cost overrun anomaly. Then, deviation decomposition is performed to identify its driving factors: 1. Unit price deviation analysis: Compare the actual value and planned value of the material unit price. The planned value here is the expected unit price of 460 yuan / ton used in the benchmark calculation in the previous module. The actual value is the weighted average unit price of 459.34 yuan / ton calculated in this period. Unit price deviation = The negative result indicates that the actual purchase price was lower than expected, and the unit price factor itself did not lead to cost overruns.

[0114] 2. Usage Deviation Analysis: Compare the actual and predicted material consumption. Actual consumption was 14,300 tons. Predicted consumption was 13,697.7 tons, adjusted for environmental factors, calculated in the previous module. Usage Deviation = This result is positive, indicating that the actual usage exceeded the predicted usage after adjustments for environmental factors. The degree of over-consumption is defined as follows: a deviation rate of less than 5% is considered slight over-consumption, 5%-10% is moderate over-consumption, and above 10% is severe over-consumption. The deviation rate in this case is... This is considered a slight overconsumption.

[0115] 3. Comprehensive Judgment: Based on the above two analyses, the driving factor for the cost overrun is entirely due to the excessive consumption of materials. Therefore, among the three types of anomalies—unit price fluctuation anomalies, excessive consumption anomalies, and comprehensive impact anomalies—this anomaly iteration is classified as an excessive consumption anomaly.

[0116] Finally, the generated material consumption cost anomaly identification result is updated into a structured data containing complete information: {Identification result: anomaly status, anomaly type: cost overrun anomaly (excessive consumption type), associated risks: construction waste, measurement error or design change}.

[0117] Please see Figure 1 and Figure 5 The material spatial risk linkage identification module obtains the abnormal judgment results of material consumption costs and the material consumption costs of the same construction section and work points, and uses the K-means clustering algorithm to perform spatial clustering analysis to output material spatial linkage risk information.

[0118] The material space risk linkage identification module includes:

[0119] The cost data integration submodule obtains the material consumption cost anomaly identification results and the material consumption cost of the same construction section. It integrates the two types of data with the construction site as the basic unit, performs numerical processing on the anomaly identification results to form risk labels, and forms feature vectors with the corresponding construction site material consumption cost data. It merges the feature vectors of all construction sites to establish a construction site material consumption feature matrix.

[0120] After receiving the cost anomaly assessment results for a specific work site, this module begins execution. This module integrates the analysis results using a single construction work site as the basic unit. The entire highway project is divided into 10 consecutive construction work sites, starting from chainage K0+000, with one work site defined every 2 kilometers. The core area of ​​the third section (chainage K10+000 to K12+000), where the "excessive consumption anomaly" appeared in the aforementioned analysis, is defined as work site 6. This submodule obtains the material consumption cost anomaly assessment results for all 10 work sites.

[0121] Next, the anomaly detection results are numerically processed to form risk labels. The processing rules are defined as follows: normal status is assigned a value of 0, cost surplus anomaly is assigned a value of -1, and cost overrun anomaly is assigned a value of 1. According to this rule, the risk label for work point 6 is assigned a value of 1. Simultaneously, the cost deviation rate is calculated for each work point, calculated as (current period material costs / cost center value) - 1. For work point 6, its cost deviation rate is (6,568,562 yuan / 6,300,942 yuan) - 1. 0.042.

[0122] This submodule constructs a feature vector for each work point, which consists of the work point's spatial coordinates (simplified to the center station number in this linear project), risk identifier, and cost deviation rate. The feature vector for work point 6 is {station number: 11.0, risk identifier: 1, cost deviation rate: 0.042}. The feature vectors of all 10 work points are merged to establish the following material consumption feature matrix for each work point.

[0123] Table 2. Material Consumption Characteristics Matrix of Work Sites

[0124] ;

[0125] As shown in Table 2, the matrix integrates the spatial, risk, and cost data of 10 work sites, among which work sites 3, 5, 6, and 7 are marked as work sites with the risk of cost overrun.

[0126] The spatial clustering analysis submodule calls the material consumption feature matrix of the work site, uses the K-means clustering algorithm to initialize the cluster centers according to the preset number of clusters K, and iteratively calculates the Euclidean distance between multiple work site data and multiple cluster centers. The work site is assigned to the nearest cluster center and the center position is updated until the cluster center position no longer changes, thus obtaining the spatial clustering assignment of material consumption.

[0127] This submodule calls the aforementioned material consumption feature matrix for each work site and employs the K-means clustering algorithm. In this embodiment, cluster analysis is performed using data from two dimensions: risk identification and cost deviation rate. The preset cluster size K is set to 3, a value based on prior knowledge from project management experience that classifies work sites into three categories: "normal," "low-risk," and "high-risk."

[0128] 1. Initialization: From the data of 10 work sites, 3 sites are randomly selected as initial cluster centers. Work site 2 {risk indicator: 0, cost deviation rate: 0.008}, work site 5 {risk indicator: 1, cost deviation rate: 0.062}, and work site 9 {risk indicator: 0, cost deviation rate: -0.011} are selected as cluster centers C1, C2, and C3 respectively.

[0129] 2. Initial Assignment: Calculate the Euclidean distance from the data vectors of the remaining 7 work points to the 3 cluster centers. Calculate the distance from work point 6 {1, 0.042} to the three centers: d(work point 6, C1) = d(work point 6, C2) = d(work point 6, C3) = Since the distance from work point 6 to C2 is the smallest (0.02), work point 6 is assigned to cluster 2. This operation is performed on all work points to obtain the initial clustering results.

[0130] 3. Update centroid: Based on the initial clustering results, recalculate the centroid of each cluster, which is the average value of the feature vectors of all nodes within the cluster.

[0131] 4. Iterative Process: Repeat the assignment and update steps. After 3 iterations, the positions of the cluster centers no longer change, and the algorithm converges. The final material consumption space cluster assignments are: Cluster 1 (C1'): {Work Points 1, 2, 4, 8, 9, 10} Cluster 2 (C2'): {Work Points 3, 5, 6, 7} Cluster 3 (C3'): No work point affiliation; this center was eliminated during the iteration due to lack of members. Ultimately, all work points are divided into two clusters.

[0132] The risk linkage identification submodule determines the number of work points with risk labels in each cluster based on the spatial clustering of material consumption, calculates the proportion of each cluster, compares the proportion with the set risk identification benchmark value, filters out clusters with a proportion exceeding the risk identification benchmark value, and generates material spatial linkage risk information.

[0133] The risk identification benchmark value used by the risk linkage identification submodule is obtained by acquiring historical completed project data that has a similarity to the current project in terms of engineering geology, construction scale, and climate zoning that exceeds a preset similarity threshold. The distribution of the proportion of abnormal work points within the risk clusters in the historical projects is statistically analyzed, and the risk identification benchmark value is generated based on the 90th percentile of the distribution.

[0134] Material spatial linkage risk information includes spatial clustering degree, risk distribution type, and impact range;

[0135] The spatial clustering degree in the material spatial linkage risk information is quantified by calculating the average profile coefficient of each cluster, and the average profile coefficient is used as a measure of the spatial clustering degree. The average profile coefficient is calculated using the following formula:

[0136] calculate;

[0137] in, Represents cluster The average profile coefficient, Represents cluster Total number of work sites within the area Represents cluster A single work site within, Representative work site The average Euclidean distance between the work site and all other work sites within the same cluster. Representative work site The minimum average Euclidean distance from the point in the node to all other clusters. represent and The larger value in;

[0138] Risk distribution type is determined by calculating the minimum boundary rectangle of the spatial coordinates of all work points within each selected cluster and solving for the aspect ratio of the rectangle. If the aspect ratio is greater than the preset shape ratio threshold, the risk distribution type is determined as corridor risk; otherwise, it is determined as area risk.

[0139] The scope of influence is defined by constructing the minimum convex hull polygon of the spatial coordinates of all work sites within the selected clusters, and the set of geographic coordinate vertices of the minimum convex hull polygon is used as the boundary data of the scope of influence. At the same time, the area of ​​the polygon is calculated as a quantitative indicator of the scope of influence.

[0140] This submodule, based on the aforementioned clustering results, counts the number of work points with a risk label (risk label is 1) within each cluster and their proportion within the cluster. - Cluster 1: Contains 6 work points, of which 0 are risky. Risky work point proportion = 0 / 6 = 0%. - Cluster 2: Contains 4 work points {3, 5, 6, 7}, of which 4 are risky. Risky work point proportion = 4 / 4 = 100%.

[0141] Next, the risk proportion of each cluster is compared with the established risk identification benchmark. The process for setting the risk identification benchmark is as follows: The system acquires data from 10 historical completed projects that have a similarity exceeding 0.9 with the current project in terms of engineering geology, construction scale, and climate zoning. The same anomaly detection and spatial clustering analysis is performed on the data of these 10 projects to identify the main risk clusters in each project and calculate the proportion of anomalous work points within each cluster. This results in an array containing 10 proportion values: [0.65, 0.80, 0.75, 0.90, 0.70, 0.82, 0.78, 0.91, 0.85, 0.88]. This array is then sorted in ascending order as: [0.65, 0.70, 0.75, 0.78, 0.80, 0.82, 0.85, 0.88, 0.90, 0.91]. The 90th percentile of this distribution is taken as the benchmark value, i.e., the 9th value after sorting. Therefore, the risk identification benchmark is set at 0.90.

[0142] Comparative analysis: Cluster 1 has a risk percentage of 0%, less than 0.90. Cluster 2 has a risk percentage of 100%, greater than 0.90. Therefore, Cluster 2 is selected as the cluster with spatial linkage risk. Based on this, material spatial linkage risk information is generated, which includes the following specific details:

[0143] Spatial clustering degree: quantified by calculating the average silhouette coefficient of cluster 2. This coefficient is expressed by the formula... Calculation. In this formula, It is the average silhouette coefficient of cluster C (cluster 2 in this example); The total number of work points within the cluster is 4; Represents a single work point within a cluster; It is a work site The average Euclidean distance based on feature vectors between the work points and all other work points within the same cluster; It is a work site The minimum average Euclidean distance to all work points in another cluster. Calculate the profile coefficient for each work point {3,5,6,7} in cluster 2, and then take its arithmetic mean. The calculated average profile coefficient is... The value is 0.82. The criteria for the silhouette coefficient are defined as follows: [0.7, 1.0] indicates a strong clustering structure, and [0.5, 0.7) indicates a reasonable clustering structure. A value of 0.82 falls within the strong clustering structure range, indicating that the work points within cluster 2 are highly similar in risk behavior and significantly different from normal work points.

[0144] Risk distribution type: Obtain the spatial coordinates - station (5.0, 9.0, 11.0, 13.0) of all work points (3, 5, 6, 7) in cluster 2. Calculate the minimum boundary rectangle for this point set. The length of this rectangle along the road direction is... The rectangle's width is 0.05 km, which is the width of the road's right-of-way. Calculate the aspect ratio of this rectangle as follows: A shape ratio threshold was set. This threshold was set based on the general characteristics of linear engineering projects, namely, when the length-to-width ratio exceeds a certain value, the geometric shape becomes corridor-like. Several similar linear engineering projects were selected, and the length-to-width ratio distribution of their risk clusters was statistically analyzed. The 25th percentile of this distribution was taken as the threshold, which was calculated to be 5.0. Since the calculated length-to-width ratio of 160 is much greater than the threshold of 5.0, the risk distribution type was determined to be corridor-type risk.

[0145] Impact Range: Construct the minimum convex hull polygon for the spatial coordinates of all work points (3, 5, 6, 7) in the selected cluster 2. Since these work points are distributed along a straight line (highway), their minimum convex hull is represented as a line segment on the two-dimensional plane, extending from station K5+000 to K13+000. The geographic coordinates (latitude and longitude) of the two endpoints of this line segment are extracted and stored as boundary data of the impact range. Simultaneously, the quantitative indicator of this impact range, i.e., the area of ​​the polygon, is calculated. Area = Length of the affected road segment. Road width = Square meters. This data clarifies the geographical scope and scale of the risk linkage.

[0146] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the scope of protection defined by the claims of the present invention.

Claims

1. A road construction material cost control system, characterized in that, The system includes: The material consumption data acquisition module is used to obtain the actual material consumption, planned consumption, progress percentage and equipment runtime through field equipment, correct the planned material consumption using a weighted moving average model, generate the weighted corrected material consumption and transmit it to the material dynamic consumption benchmark generation module. The material consumption data acquisition module includes: The on-site data acquisition submodule collects the actual material consumption, planned consumption, progress percentage, and equipment runtime of on-site equipment. It calculates the consumption difference based on the actual material consumption and the planned consumption, integrates the progress percentage and the equipment runtime, and associates and classifies the data of different time periods to establish the basic data volume of on-site material consumption. The consumption weight determination submodule calls the on-site material consumption basic data, extracts the equipment running time of multiple cycles according to the time proximity principle in the weighted moving average model, and uses the running time as the core factor for weight allocation, assigns differentiated weight values ​​to the material consumption data of multiple cycles, normalizes all weight values, and generates moving average weight coefficients. The weighted moving average correction submodule calls the planned material consumption within the on-site material consumption basic data, and applies the weighted moving average model according to the moving average weight coefficient. It multiplies the planned consumption of multiple periods by the corresponding weight coefficient, and sums the calculated products of all periods to obtain the weighted corrected material consumption. The material dynamic consumption benchmark generation module is used to obtain the weighted and corrected material consumption, meteorological parameters, and material unit price, correct the material consumption using a multiple linear regression model, generate a material dynamic consumption cost benchmark range, and transmit it to the material consumption cost anomaly detection module. The material consumption cost anomaly detection module calculates the actual material consumption cost based on the actual material consumption and the material unit price, calls the material dynamic consumption cost benchmark range for range judgment, generates a material consumption cost anomaly detection result, and transmits the material consumption cost anomaly detection result to the material spatial risk linkage identification module. The material spatial risk linkage identification module obtains the abnormal judgment results of material consumption cost and the material consumption cost of the same construction section, and uses the K-means clustering algorithm to perform spatial clustering analysis to output material spatial linkage risk information. The material spatial linkage risk information includes the degree of spatial clustering, risk distribution type, and scope of impact. The spatial clustering degree in the material spatial linkage risk information is quantified by calculating the average profile coefficient of each cluster, and the average profile coefficient is used as a measure of the spatial clustering degree. The average profile coefficient is calculated using the following formula: calculate; in, Represents cluster The average profile coefficient, Represents cluster Total number of work sites within the area Represents cluster A single work site within, Representative work site The average Euclidean distance between the work site and all other work sites within the same cluster. Representative work site The minimum average Euclidean distance from the point in the node to all other clusters. represent and The larger value in; The risk distribution type is determined by calculating the minimum boundary rectangle of the spatial coordinates of all work points within each selected cluster and solving the aspect ratio of the rectangle. If the aspect ratio is greater than a preset shape ratio threshold, the risk distribution type is determined as a corridor-type risk; otherwise, it is determined as a regional-type risk. The scope of influence is defined by constructing the minimum convex hull polygon of the spatial coordinates of all work sites within the selected cluster, and the set of geographic coordinate vertices of the minimum convex hull polygon is used as the boundary data of the scope of influence. At the same time, the area of ​​the polygon is calculated as a quantitative indicator of the scope of influence.

2. The road construction material cost control system according to claim 1, characterized in that, The weighted adjusted material consumption specifically refers to the adjusted consumption and the adjustment basis parameters. The material dynamic consumption cost benchmark range includes the lower limit of cost, the upper limit of cost, and the range of fluctuation. The material consumption cost anomaly judgment result specifically refers to the abnormal state, the abnormal type, and the associated risk.

3. The road construction material cost control system according to claim 1, characterized in that, The material dynamic consumption benchmark generation module includes: The regression variable screening submodule obtains the weighted adjusted material consumption, meteorological parameters and material unit price. Based on the multiple linear regression model, the weighted adjusted material consumption is used as the dependent variable, and the meteorological parameters and material unit price are used as independent variables. By calculating the correlation coefficient between the independent variables and the dependent variable, the independent variables with a correlation degree exceeding the preset correlation degree threshold are screened to obtain the core influencing factors of the multiple regression model. The dynamic consumption correction submodule calls the core influencing factors of the multiple regression model, constructs a regression equation based on the multiple linear regression model, multiplies the current values ​​of multiple core influencing factors with the corresponding regression coefficients in the regression equation, sums them, and adds them to the constant term to correct the weighted material consumption and generate the environmental factor corrected material consumption. The cost benchmark interval calculation submodule calculates the cost center value by multiplying the material consumption and material unit price after the environmental factors are corrected, and calculates the confidence interval of the cost center value at a specified confidence level based on the prediction standard error of the multiple linear regression model. The upper and lower limits of the interval are used as fluctuation boundaries to establish the dynamic material consumption cost benchmark interval.

4. The road construction material cost control system according to claim 3, characterized in that, The material consumption cost anomaly detection module includes: The actual cost accounting submodule obtains the actual consumption of the material and the unit price of the material, multiplies the consumption value and the unit price value, and sums up all the product results to calculate the total cost within the period and establish the material cost incurred in the current period. The cost range discrimination submodule calls the current material cost and the material dynamic consumption cost benchmark range, compares the value of the current material cost with the upper and lower limits of the material dynamic consumption cost benchmark range. If the cost value exceeds the upper and lower limits of the range, it will be marked as an abnormal state; if it is within the range, it will be marked as a normal state, and an abnormal material consumption cost discrimination result will be generated.

5. The road construction material cost control system according to claim 4, characterized in that, The material space risk linkage identification module includes: The cost data integration submodule obtains the material consumption cost anomaly judgment result and the material consumption cost of the same construction section work point. It integrates the two types of data with the construction work point as the basic unit, performs numerical processing on the anomaly judgment result to form a risk identifier, and forms a feature vector with the corresponding work point material consumption cost data. It merges the feature vectors of all work points to establish a work point material consumption feature matrix. The spatial clustering analysis submodule calls the material consumption feature matrix of the work site, uses the K-means clustering algorithm to initialize the cluster centers according to the preset number of clusters K, and iteratively calculates the Euclidean distance between multiple work site data and multiple cluster centers. The work site is assigned to the nearest cluster center and the center position is updated until the position of the cluster center no longer changes, thus obtaining the spatial clustering assignment of material consumption. The risk linkage identification submodule, based on the spatial clustering of material consumption, counts the number of work points with risk identifiers within each cluster and calculates their proportion within the cluster. It then compares the proportion with a set risk identification benchmark value, filters out clusters whose proportions exceed the risk identification benchmark value, and generates material spatial linkage risk information.

6. The road construction material cost control system according to claim 5, characterized in that, The consumption weight determination submodule specifically uses the following formula: ； Calculate the moving average weighting coefficient; in, Indicates the first The moving average weighting coefficient for each period, Indicates the first The equipment runtime per cycle, This represents the total number of periods used to calculate the weighted moving average. Indicates all The sum of the operating time of the equipment in each cycle; The dynamic consumption correction submodule specifically uses the following formula: ； Calculate the material consumption after adjusting for the aforementioned environmental factors; in, This indicates the material consumption after adjusting for the aforementioned environmental factors. This represents the constant term in the regression equation. Indicates the first The regression coefficients corresponding to the core influencing factors Indicates the first The current values ​​of the core impact factors, This represents the total number of core influencing factors in the multiple regression model. This represents the random error term of the model.

7. The road construction material cost control system according to claim 6, characterized in that, The meteorological parameters obtained by the regression variable screening submodule include the daily average temperature, daily cumulative precipitation, daily average wind speed, and daily average air humidity of the construction area. The anomaly types in the material consumption cost anomaly identification results include cost overrun anomalies and cost surplus anomalies. Among them, the cost overrun anomaly is analyzed by tracing the source of the deviation between the actual value and the planned value of the material unit price and the deviation between the actual material consumption and the material consumption after the environmental factors are corrected, and iteratively classified into unit price fluctuation anomaly, usage overconsumption anomaly and comprehensive impact anomaly. The risk identification benchmark value used by the risk linkage identification submodule is obtained by acquiring historical completed project data that has a similarity to the current project in terms of engineering geology, construction scale, and climate zoning exceeding a preset similarity threshold, statistically analyzing the proportion distribution of abnormal work points within risk clusters in historical projects, and generating the risk identification benchmark value based on the 90th percentile of the distribution.

Images