Power transmission project cost risk whole-process dynamic early warning method and system

Abstract

The application relates to a power transmission and transformation project cost risk whole-process dynamic early warning method and system. The method comprises the following steps: acquiring cost data and benchmark data of each stage of a power transmission and transformation project, and constructing a standardized cost database; based on preset risk rules, a risk index system containing static risk indexes, dynamic risk indexes and stage correlation indexes is constructed; a preset deep learning model is used to compare and trend forecast the cost data and the benchmark data in real time, and accurately identify cost risk points; the risk level is determined according to the deviation degree of the risk points, the graded early warning information is generated and pushed to the corresponding terminal; the processing result is recorded and the database is updated, the whole-process dynamic monitoring and risk early warning are realized, the real-time performance, accuracy and decision efficiency of the cost risk management are effectively improved, and the problems of early warning lag and incomplete risk coverage in the traditional method are solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of engineering cost risk management, and particularly relates to a power transmission and transformation engineering cost risk whole-process dynamic early warning method and system. BACKGROUND

[0002] With the development of the power transmission and transformation engineering field, especially under the impetus of large-scale construction and intelligent transformation of power infrastructure, traditional cost risk management technology based on static indicators and historical data has emerged. The characteristics of this technology are to collect cost data at each stage manually, combine historical similar engineering benchmarks, and identify risks through simple threshold comparison. Its core advantage lies in intuitive operation, easy implementation, and providing a basic framework for engineering cost control. In the traditional cost risk early warning method, data is mainly collected in stages: in terms of data processing, estimated cost data in the feasibility study stage, budget cost data in the preliminary design stage, and budget cost data in the construction drawing stage are obtained respectively, and a preliminary database is constructed through standardized processing; in terms of risk identification, based on preset static risk indicators (such as fixed deviation threshold), the actual data is compared with the historical benchmark manually to identify the cost deviation of a single stage. In terms of early warning mechanism, the post-analysis method is adopted, and after the generation of early warning information, the rule engine is used to push the information to the decision terminal, but there is a lack of dynamic prediction and cross-stage correlation analysis.

[0003] However, the current traditional method has the following problems: early warning lag: risk identification relies on static data comparison, and cannot respond to dynamic changes in real time (such as fluctuations in the prices of equipment and materials or adjustments in the progress of the project). For example, dynamic risk indicators (such as the cumulative value of sub-item cost based on the progress node of the project) are not effectively integrated, resulting in delayed early warning and affecting the timeliness of risk disposal. Risk coverage is not comprehensive: the traditional method focuses on static risk indicators (such as the deviation of building engineering cost), ignores dynamic trends and stage-related risks, and leads to fragmented risk point identification, which cannot capture cross-stage transmission effects and increases the overall risk of cost overrun. Decision-making efficiency is low, the early warning information push path is rigid, lacks intelligent grading, and is not associated with historical disposal references, resulting in insufficient decision support and resource waste. SUMMARY

[0004] Therefore, it is necessary to provide a power transmission and transformation engineering cost risk whole-process dynamic early warning method and system to solve the above problems.

[0005] In a first aspect, the application provides a power transmission and transformation engineering cost risk whole-process dynamic early warning method, comprising:

[0006] obtaining cost data and benchmark data of each stage of the power transmission and transformation project, and constructing a standardized cost database;

[0007] Based on the preset risk rules, a risk index system is constructed, and the risk index system includes static risk indexes, dynamic risk indexes and stage correlation indexes;

[0008] Based on the risk index system, the preset deep learning model is used to compare the cost data in the cost database with the benchmark data in real time and predict the trend, and identify cost risk points;

[0009] According to the deviation degree of the cost risk points, the risk level is determined and the graded early warning information is generated and pushed to the corresponding terminal;

[0010] The processing result of the graded early warning information is recorded, and the standardized cost database is updated according to the processing result.

[0011] In one of the embodiments, the cost data and benchmark data of each stage of the power transmission and transformation project are obtained, and a standardized cost database is constructed, including:

[0012] The estimated cost data of the feasibility study stage, the budget cost data of the preliminary design stage and the budget cost data of the construction drawing stage are obtained respectively, and the cost data covers building engineering cost, equipment purchase cost, installation engineering cost and other costs;

[0013] The cost data of the historical similar projects, the reference prices of the equipment and materials in the project area and the industry standard rates are obtained as the benchmark data;

[0014] The cost data and benchmark data of each stage are standardized, and a standardized cost database is constructed.

[0015] In one of the embodiments, the risk index system is constructed, including:

[0016] Based on the benchmark data, the corresponding static risk indexes are set for building engineering cost, equipment purchase cost, installation engineering cost and other costs;

[0017] According to the historical price fluctuation law, the dynamic risk indexes of equipment and material prices are set, and combined with the engineering progress, the stage dynamic risk indexes based on the engineering progress nodes are set, and the dynamic deviation rate is calculated according to the cumulative value of the sub-item cost;

[0018] For the paths from the feasibility study stage to the preliminary design stage and from the preliminary design stage to the construction drawing stage, the stage correlation threshold is set as the stage correlation index.

[0019] In one of the embodiments, the preset deep learning model is used to compare the cost data in the cost database with the benchmark data in real time and predict the trend, and identify cost risk points, including:

[0020] The static risk indicators, dynamic risk indicators and stage correlation indicators are input into the deep learning model, and feature comparison and cross-stage trend deduction of the cost data and the benchmark data are synchronously performed by the deep learning model to generate a risk decision vector containing static deviation, dynamic trend and connection mutation;

[0021] When it is detected that the static deviation exceeds the static risk indicator, or the dynamic trend exceeds the dynamic risk indicator, or the connection mutation coefficient breaks through the stage correlation indicator, a corresponding type of cost risk point is identified, and the section identification position of the cost risk point in the cost list, the cost impact range and the current cost deviation are determined and output.

[0022] In one embodiment, according to the deviation degree of the cost risk point, the risk level is determined and the graded warning information is generated and pushed to the corresponding terminal, including:

[0023] Based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, a composite deviation rate is calculated, and the risk level is determined according to a preset mapping rule; the risk level includes slight, general or serious;

[0024] Based on the risk level, the warning information is generated; the warning information includes the section identification position of the risk point in the cost list corresponding to the risk level; a quantitative matrix reflecting the current cost deviation value and the predicted deviation value of the risk point; the cost impact range of the risk point;

[0025] The warning information is matched and pushed to the path according to the risk level through the rule engine; when the risk level is serious, it is pushed to the first type of terminal and triggers the preset emergency response process on the first type of terminal; when the risk level is slight or general, it is pushed to the second type of terminal and associated with the preset historical risk disposal reference information.

[0026] In one embodiment, based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, a composite deviation rate is calculated, which is realized by the following formula:

[0027]

[0028] Wherein is the composite deviation rate, is the static deviation amount, represents the actual value of the current stage sub-cost, represents the benchmark cost data, represents the deviation threshold corresponding to the static risk indicator set for the construction engineering cost, the equipment purchase cost, the installation engineering cost and other costs; is the dynamic trend probability, which represents the probability value that the future N periods cost exceeding the dynamic risk indicator corresponding to the dynamic trend output by the deep learning model, and the value is ; is a connection mutation coefficient, represents the actual total cost value of the current engineering stage, represents the actual total cost value of the previous engineering stage; , and is a dynamic weight factor, satisfying .

[0029] In one of the embodiments, the method further comprises inputting the hierarchical early warning information into the visualization engine, and performing the following steps:

[0030] Based on the chapter identification position in the early warning information, a heat map is generated, and the color scale depth of the heat map maps the value of the composite deviation rate CDR;

[0031] Based on the current cost deviation value and the predicted deviation value in the early warning information, a deviation fluctuation atlas is generated, wherein the actual deviation value is associated with the static deviation amount , and the prediction interval is associated with the dynamic trend probability ;

[0032] Based on the cost influence range and the connection mutation coefficient in the early warning information, a cross-stage risk transmission tree diagram is generated, and the influence weight between nodes in the tree diagram is quantified by value.

[0033] In a second aspect, the application also provides a power transmission and transformation project cost risk whole-process dynamic early warning system, comprising:

[0034] A data acquisition and processing module is configured to acquire cost data and benchmark data of each stage of the power transmission and transformation project, and construct a standardized cost database;

[0035] A risk index management module is configured to construct a risk index system based on a preset risk rule, and the risk index system comprises static risk indexes, dynamic risk indexes and stage correlation indexes;

[0036] An intelligent risk analysis module is configured to compare the cost data and the benchmark data in the cost database in real time and predict trends based on the risk index system and a preset deep learning model, and identify cost risk points;

[0037] An early warning generation and response module is configured to determine a risk level and generate hierarchical early warning information based on the deviation degree of the cost risk points, and push the hierarchical early warning information to a corresponding terminal;

[0038] A feedback optimization module is configured to record the processing results of the hierarchical early warning information, and update the standardized cost database according to the processing results.

[0039] In a third aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the power transmission and transformation project cost risk whole-process dynamic early warning method when executing the computer program.

[0040] In a fourth aspect, the present application also provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the power transmission and transformation project cost risk whole-process dynamic early warning method when executed by a processor.

[0041] The power transmission and transformation project cost risk whole-process dynamic early warning method, system, computer device and storage medium, by acquiring cost data and benchmark data of each stage of the power transmission and transformation project to construct a standardized cost database, and based on a preset risk rule to construct a risk index system covering static risk indexes, dynamic risk indexes and stage correlation indexes, using a preset deep learning model to compare and predict the trend of the cost data and the benchmark data in real time, accurately identifying cost risk points; according to the deviation degree of the risk points, determining the risk level, generating graded early warning information and pushing it to the corresponding terminal, and recording the processing results and updating the database. The problem of early warning lag is solved: the real-time comparison and trend prediction mechanism of the deep learning model can dynamically respond to the fluctuation of equipment and material prices or the adjustment of the project progress, avoiding the delay caused by relying on static data; the introduction of dynamic risk indexes and stage correlation indexes in the risk index system solves the problem of incomplete risk coverage, captures the price fluctuation rule and cross-stage correlation effect, and avoids the risk of project cost overrun caused by fragmented risk identification; the risk level determination and graded pushing mechanism based on the deviation degree optimizes the decision-making efficiency, realizes the intelligent distribution of early warning information, ensures the triggering of the emergency response process when serious risks occur, and improves the real-time, accuracy and overall decision-making efficiency of risk management when minor or general risks occur. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiment or related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0043] Figure 1 The flowchart of the power transmission and transformation project cost risk whole-process dynamic early warning method of the present application;

[0044] Figure 2 The structural diagram of the power transmission and transformation project cost risk whole-process dynamic early warning system of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0046] In one embodiment, as shown in Figure 1 A power transmission and transformation project cost risk whole-process dynamic early warning method is provided. In this embodiment, the method is applied to an engineering site cost data acquisition terminal. The terminal is provided with a built-in sensor and a data interface, and can obtain cost data (including construction cost, equipment purchase cost, etc.) of the feasibility study stage, preliminary design stage and construction drawing stage of the power transmission and transformation project in real time. It can be understood that the method can also be deployed on a cloud analysis server to synchronously process multiple engineering data streams through a deep learning model; or applied to a terminal-server collaborative system: the terminal is responsible for collecting real-time cost data and receiving early warning instructions, the server runs a standardized cost database and a risk analysis algorithm, and interacts through a 4G / 5G private network. When the power transmission and transformation project faces dynamic price fluctuation risks (such as sudden changes in the price of equipment and materials during the construction stage) or abnormal cross-stage cost connection, the terminal uploads the current sub-item cost data to the server. The server calls a risk index system (including static deviation threshold, dynamic trend probability, etc.), generates a risk decision vector through a deep learning model. If the composite deviation rate is detected to be out of limit, the server immediately triggers a hierarchical push: through a rule engine, early warning information containing a heat map (mapping the deviation position) is sent to the project manager mobile terminal (first type of terminal), and the pre-installed emergency response process of the terminal is activated synchronously; if it is a general risk, it is pushed to the cost engineer terminal (second type of terminal) and associated with a historical disposal scheme library. After the terminal feeds back the disposal result, the server updates the database to close the risk control process. In this embodiment, the method comprises the following steps:

[0047] S01, obtaining cost data and reference data of each stage of the power transmission and transformation project, and constructing a standardized cost database.

[0048] The cost data of each stage (estimated cost data of the feasibility study stage, budget cost data of the preliminary design stage, and budget cost data of the construction drawing stage, which can include construction cost, equipment purchase cost, installation cost and other key sub-item costs) and reference data (cost data of similar historical projects, reference prices of equipment and materials in the project area, and industry standard rates, which are used to provide comparison references); the above data can be obtained in real time or asynchronously through a terminal with built-in sensors or a cloud interface, and standardized processing is performed, including data cleaning, format unification and numerical standardization, to eliminate inconsistencies; the processed data is integrated by a database management system to construct a structured and stored standardized cost database, which supports subsequent real-time access and updating.

[0049] S02, based on the preset risk rules, a risk index system is constructed, and the risk index system includes static risk indexes, dynamic risk indexes, and stage correlation indexes.

[0050] The static risk indexes (for sub-item costs such as construction engineering costs, equipment purchase costs, installation engineering costs, and other costs, based on a benchmark data such as a historical similar engineering data and a deviation threshold set by an industry standard) are used to quantify the static deviation of the actual value from the benchmark value; the dynamic risk indexes (combining the historical price fluctuation rule and the dynamic deviation rate index of the engineering progress node, including the risk probability model of the equipment and material price and the index for calculating the dynamic trend based on the cumulative value of the sub-item cost, to capture the dynamic risk of the price and progress change); and the stage correlation indexes (for the transition path of the engineering stage, such as the connection mutation threshold set for the transition from the feasibility study stage to the preliminary design stage and from the preliminary design stage to the construction drawing stage, to identify the abnormal mutation of the total cost value between stages). When implemented, the preset risk rules can be applied to index construction: the rule engine is used to analyze the benchmark data to set the static risk indexes for each sub-item cost, such as formulaic processing based on the maximum deviation threshold; the historical data is used to train the model to generate the dynamic risk indexes of the equipment and material price, and the dynamic deviation rate is calculated in combination with the progress node; and the stage correlation threshold is defined for the stage path to quantify the connection mutation coefficient.

[0051] S03, based on the risk index system, a preset deep learning model is used to compare and predict the trends of the cost data and the benchmark data in the cost database in real time, to identify the cost risk points.

[0052] The preset deep learning model (a neural network architecture formed by training the time series data of the cost data and the benchmark data in a large number of different engineering cases, used to simultaneously perform feature comparison and trend deduction); the cost risk points (abnormal states generated when the actual data deviates from the benchmark, specifically quantified by a risk decision vector); when implemented, the risk index system is input into the deep learning model, the model extracts multi-scale features of the cost data and the benchmark data through a convolution layer, performs cross-stage time series deduction through a recurrent neural network layer, and generates a risk decision vector including a static deviation amount, a dynamic trend probability, and a connection mutation coefficient; when the static deviation exceeds the static risk index threshold, or the dynamic trend probability exceeds the dynamic risk index range, or the connection mutation coefficient breaks through the stage correlation threshold, the model locates the risk point at the chapter identification position in the cost list, and simultaneously outputs the cost impact range and the current deviation value.

[0053] S04, according to the deviation degree of the cost risk point, the risk level is determined and the graded warning information is generated and pushed to the corresponding terminal.

[0054] Among them, the risk level (three levels of slight, general or serious based on the composite deviation rate (CDR) mapping, used to quantify the urgency of risk points); hierarchical warning information (contains the section identification position of the risk point in the cost list, used to locate the risk source, reflects the quantitative matrix of the current cost deviation value and the predicted deviation value, shows the deviation trend through the numerical model and the cost impact range of the risk point, describes the structured data of the spread of the risk to the engineering sub-item cost); when pushed to the corresponding terminal (the warning information is intelligently distributed to the specified terminal through the rule engine, which specifically includes pushing to the first type of terminal (such as the project manager mobile terminal) and triggering the preset emergency response process when the risk level is serious, and pushing to the second type of terminal (such as the cost engineer terminal) and associating to provide historical risk disposal reference information when the risk level is slight or general), based on the static deviation amount, dynamic trend probability and connection mutation coefficient of the risk decision vector in the cost risk point, the composite deviation rate CDR value is calculated; according to the preset mapping rule (such as CDR threshold interval: slight corresponds to CDR<0.3, general corresponds to 0.3≤CDR<0.6, and serious corresponds to CDR≥0.6), the risk level is determined; hierarchical warning information is generated, and the location identifier, quantitative matrix and impact range are integrated; through the rule engine, the dynamic matching of the push path is realized, and the information distribution is realized.

[0055] S05, record the processing result of the hierarchical warning information, and update the standardized cost database according to the processing result.

[0056] Among them, the processing result of the hierarchical warning information (the disposal measures taken by the terminal for the received warning information and the effect data generated thereby, such as the revised cost data or risk resolution record after manual intervention); update the standardized cost database (based on the processing result, dynamically optimize the benchmark data in the database, such as historical similar engineering data, equipment material reference price and industry standard rate, to improve the accuracy of subsequent risk analysis). When implemented, the warning processing result feedback from the terminal can be received in real time, the effectiveness data of the disposal measures is analyzed, and the corresponding benchmark field in the standardized cost database is dynamically modified (such as when the processing result shows that the price of equipment and materials has undergone structural changes, the reference price library is updated; or when the stage connection mutation is manually corrected, the calculation model of the stage correlation threshold is adjusted); through the formation of a closed-loop feedback mechanism, the disposal experience is converted into the continuous optimization of the benchmark data, ensuring the dynamic adaptability of the risk index system, and realizing the self-iterative upgrade of the whole-process warning system.

[0057] The whole-process dynamic early warning method for the construction cost risk of the power transmission and transformation project, by obtaining the cost data and benchmark data of each stage of the power transmission and transformation project, and constructing a standardized cost database; based on the preset risk rules, a risk index system is constructed, covering static risk indexes, dynamic risk indexes (such as equipment and material price fluctuation trend and dynamic deviation rate based on engineering progress nodes), and stage correlation indexes (such as the connection mutation threshold from the feasibility study to the preliminary design and from the preliminary design to the construction drawing stage), to ensure the comprehensiveness of the risk identification framework; the preset deep learning model (trained by historical time series data) is used to compare and predict the cost data and benchmark data in real time, and the feature comparison and cross-stage deduction are simultaneously performed, to generate a risk decision vector containing static deviation, dynamic trend and connection mutation, and identify the cost risk points; the risk level (slight, general or serious) is determined according to the deviation degree of the risk points, and the graded early warning information is generated, which is pushed to the corresponding terminal through the rule engine; the processing results are recorded and the database is updated, forming a closed-loop feedback mechanism. In view of the early warning lag problem, the deep learning model integrates dynamic risk indexes (such as price fluctuation trend), which can respond to equipment and material price changes or engineering progress adjustment in real time, avoid the delay caused by relying on static data, and improve the timeliness of early warning; in view of the problem of incomplete risk coverage, the stage correlation indexes capture the connection mutation of the feasibility study to the preliminary design and other paths, and the static and dynamic indexes are combined to avoid fragmented risk identification and prevent the engineering cost overrun risk caused by cross-stage transmission effect; in view of the low decision efficiency problem, the risk level determination (based on the composite deviation rate) and the graded pushing mechanism optimize the early warning path, ensure that the emergency process is triggered when the risk is serious, and the historical reference is associated when the risk is slight, and improve the real-time, accuracy and decision efficiency of risk management.

[0058] In one of the embodiments, the cost data and benchmark data of each stage of the power transmission and transformation project are obtained, and a standardized cost database is constructed, including:

[0059] S11, respectively obtaining the estimated cost data of the feasibility study stage, the budget cost data of the preliminary design stage and the budget cost data of the construction drawing stage, the cost data covering building engineering cost, equipment purchase cost, installation engineering cost and other costs;

[0060] S12, obtaining the cost data of the historical similar projects, the reference price of the equipment and materials in the project area and the standard rate of the industry as the benchmark data;

[0061] S13, standardizing the cost data and benchmark data of each stage to construct a standardized cost database.

[0062] Specifically, the estimation cost data in the feasibility study stage (the engineering feasibility study stage, wherein the estimation cost data represents the estimated construction cost, equipment purchase and installation cost based on the preliminary design, specifically covering the construction engineering cost (such as the civil engineering and structural engineering cost), equipment purchase cost (such as the transformer and switch equipment procurement cost), installation engineering cost (such as the equipment installation and commissioning cost) and other costs (such as the land use cost, project management and service cost)) of the building; the preliminary design stage (the preliminary design stage, wherein the budget cost data is the budget value prepared after the detailed design); and the construction drawing stage (the construction drawing design stage, wherein the budget cost data is the detailed construction drawing budget value). These data can be collected in real time or asynchronously through the terminal or cloud data interface, ensuring coverage of the entire life cycle of the project. The benchmark data (including the cost data of similar historical projects, such as the past cost records of similar power transmission and transformation projects, the equipment and material reference prices in the region where the project is located, such as the regional market prices of cables or insulation materials, and the industry standard rates, such as the standard ratio of installation labor or management cost) can be obtained from the engineering database or external market information platform, providing a benchmark for risk analysis. When implemented, the estimation data in the feasibility study stage, the budget data in the preliminary design stage and the budget data in the construction drawing stage are extracted respectively; the benchmark data can be retrieved from the historical database and the regional price database by using the query engine; the above-mentioned data is standardized, including data cleaning (removing outliers or duplicates), format unification (such as converting to standard JSON or CSV format) and numerical standardization (such as normalizing the numerical range by the z-score method), to construct a standardized cost database stored in a structured manner, supporting subsequent real-time access.

[0063] In one of the embodiments, a risk index system is constructed, including:

[0064] S21, based on the benchmark data, setting corresponding static risk indexes for the construction engineering cost, equipment purchase cost, installation engineering cost and other costs respectively;

[0065] S23, setting the dynamic risk index of the equipment and material price according to the historical price fluctuation rule, and setting the stage dynamic risk index based on the engineering progress node, the dynamic deviation rate calculated by the sub-item cost cumulative value according to the engineering progress;

[0066] S24, setting the stage correlation threshold as the stage correlation index for the path from the feasibility study stage to the preliminary design stage and the path from the preliminary design stage to the construction drawing stage.

[0067] Exemplarily, the static risk indicator is a deviation threshold set for key sub-item costs of power transmission and transformation projects, such as construction cost, equipment purchase cost, installation cost and other costs. The construction cost covers the cost of civil structure, the equipment purchase cost includes the procurement cost of transformers and the like, the installation cost involves the labor cost for equipment commissioning, and the other costs cover the costs generated by land acquisition, project management, pre-construction consultation, production preparation and the like, which can be generated based on benchmark data such as historical similar project cost and industry standard rate through rule engine analysis. When implemented, a preset risk rule (such as a maximum deviation threshold formula) can be used to set a static threshold for each sub-item cost, such as calculating the allowed deviation range based on the benchmark value obtained by database query, to ensure the static basis of risk identification; the dynamic risk indicator includes a fluctuation risk probability model of equipment and material prices and a dynamic deviation rate based on engineering progress nodes. The risk indicator of the equipment and material prices (such as cables or insulation materials) is generated by a probability model trained based on historical price data to capture periodic fluctuations, and the dynamic deviation rate is calculated in real time based on the cumulative value of sub-item cost according to engineering progress nodes such as foundation completion or equipment installation stage, such as the ratio of cumulative cost to benchmark. When implemented, machine learning algorithms can be used to analyze historical price patterns and integrate progress sensor data to dynamically update the deviation rate model to address market changes and engineering delay risks; the stage association indicator is a transition path of engineering stages, such as the threshold of connection mutation between the researchable stage and the preliminary design stage and the preliminary design stage and the construction drawing stage. The stage path represents the evolution process of the project from feasibility study to preliminary design to construction drawing, and the connection mutation coefficient is quantified by the absolute value ratio of the total cost value change between the previous stage and the current stage. When implemented, a rule engine can be used to define the path boundary by setting a stage association threshold such as the maximum mutation rate, and trigger an early warning when the mutation coefficient exceeds the limit, dynamically optimize the indicator through a closed-loop feedback mechanism, and realize dynamic monitoring throughout the process.

[0068] In one of the embodiments, a preset deep learning model is used to compare and predict the cost data in the cost database with the benchmark data in real time, identify cost risk points, including:

[0069] S31, input the static risk indicator, the dynamic risk indicator and the stage association indicator into the deep learning model, and perform feature comparison and cross-stage trend deduction of the cost data and the benchmark data through the deep learning model to generate a risk decision vector containing static deviation, dynamic trend and connection mutation;

[0070] S32, when detecting that the static deviation exceeds the static risk indicator, or the dynamic trend exceeds the dynamic risk indicator, or the connection mutation coefficient breaks through the stage association indicator, identify the corresponding type of cost risk point and determine the section identification position of the cost risk point in the cost list, the cost impact range and the current cost deviation and output.

[0071] Specifically, the preset deep learning model is a neural network architecture trained by historical power transmission and transformation project time series data, which is used to synchronously perform feature comparison (extracting the difference features of the current cost data and the benchmark data such as the cost of the similar historical project) and cross-stage trend deduction (deducing the future risk trend based on the evolution of the project stage such as from feasibility study to preliminary design); the risk decision vector is a data structure output by the model, which includes static deviation (the deviation amount of the actual sub-item cost such as construction engineering cost from the benchmark value), dynamic trend (the future probability of equipment and material price fluctuation) and connection mutation (the coefficient of the total value change of the stage connection). In implementation, the static risk indicators (such as the construction engineering cost deviation threshold) in the risk indicator system, the dynamic risk indicators (such as the equipment and material price fluctuation probability) and the stage correlation indicators (such as the stage connection mutation threshold) are input into the model, the model extracts multi-scale features (such as the numerical distribution of the sub-item cost) through the convolution layer and performs time series analysis through the recurrent neural network layer to generate the risk decision vector; the vector is detected in real time, when the static deviation exceeds the static risk indicator (such as the construction engineering cost deviation amount is greater than the threshold), or the dynamic trend exceeds the dynamic risk indicator (such as the price fluctuation probability is out of limit), or the connection mutation coefficient breaks through the stage correlation indicator (such as the mutation rate from feasibility study to preliminary design exceeds the threshold), the model accurately identifies the corresponding risk point type (static type, dynamic type or connection type), determines its chapter identification position in the cost list (such as the specific cost chapter code), the cost impact range (such as the range of the affected equipment purchase cost) and the current cost deviation value (such as the difference between the actual and the benchmark), and outputs the warning. Through the analysis of the risk core by artificial intelligence technology, the real-time and accuracy of the warning are improved, and the problems of lagging warning and incomplete risk coverage are solved.

[0072] In one of the embodiments, according to the deviation degree of the cost risk point, the risk level is determined and the graded warning information is generated and pushed to the corresponding terminal, including:

[0073] S41, based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, the composite deviation rate is calculated, and the risk level is determined according to the preset mapping rule; the risk level includes slight, general or serious;

[0074] S42, based on the risk level, the warning information is generated; the warning information includes the chapter identification position of the risk point corresponding to the risk level in the cost list; the quantization matrix reflecting the current cost deviation value and the predicted deviation value of the risk point; the cost impact range of the risk point;

[0075] S43, the warning information is matched and pushed to the path according to the risk level through the rule engine; when the risk level is serious, it is pushed to the first type of terminal and triggers the preset emergency response process on the first type of terminal; when the risk level is slight or general, it is pushed to the second type of terminal and associated with the preset historical risk disposal reference information.

[0076] Exemplarily, the composite deviation rate (CDR) is an index of the deviation degree of the comprehensive quantitative risk point, which is calculated by a static deviation amount, a dynamic trend probability and a connection mutation coefficient; the risk level includes three levels of slight, general or serious, which can be determined based on a preset mapping rule (such as CDR threshold interval: slight corresponds to CDR < 0.3, general corresponds to 0.3 ≤ CDR < 0.6, and serious corresponds to CDR ≥ 0.6); the early warning information includes the section identification position of the risk point in the cost list (such as the specific cost section code), the quantitative matrix reflecting the current cost deviation value and the predicted deviation value (the deviation trend is displayed through the numerical model), and the cost influence range of the risk point (the degree of affecting the engineering sub-item cost is described); the first type of terminal refers to the terminal receiving a serious risk warning (such as a project manager mobile terminal), triggering a preset emergency response process, and the second type of terminal refers to the terminal receiving a slight or general risk warning (such as a cost engineer terminal) and being associated with the preset historical risk disposal reference information. When implemented, the CDR value can be calculated based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector (from the previous step), and the risk level can be determined by applying the preset mapping rule through the rule engine; the early warning information can be generated according to the risk level, and the section identification position, the quantitative matrix (actual deviation value association, prediction interval association) and the influence range can be integrated; through the rule engine matching and pushing path, the serious risk is pushed to the first type of terminal and triggers the emergency response process (such as automatic alarm and intervention instruction) of the first type of terminal, and the slight or general risk is pushed to the second type of terminal and is associated with the historical disposal reference information library (such as the risk solution case of similar projects), so as to realize the closed loop from risk identification to early warning distribution.

[0077] In one of the embodiments, S51, based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, the composite deviation rate is calculated, which is realized by the following formula:

[0078]

[0079] wherein is the composite deviation rate, is the static deviation amount, represents the actual value of the sub-item cost of the current stage, represents the benchmark cost data, represents the deviation threshold value corresponding to the static risk index set for the construction engineering cost, the equipment purchase cost, the installation engineering cost and other costs; is the dynamic trend probability, which represents the probability value of the cost exceeding the dynamic risk index in the future N periods corresponding to the dynamic trend output by the deep learning model, and the value is ; is the connection mutation coefficient, represents the actual total cost value of the current engineering stage, actual total cost value of the previous engineering stage; , and is a dynamic weight factor, satisfying .

[0080] Specifically, the composite deviation rate (CDR) can be calculated by the formula , is the absolute deviation ratio of the actual value of the construction cost and the benchmark cost data such as historical similar engineering data , and is the deviation threshold value (such as the maximum allowed deviation value) corresponding to the static risk index set for the construction cost, equipment purchase cost, installation engineering cost and other costs, is a probability value ranging from 0 to 1, representing the probability of the cost exceeding the dynamic risk index (such as the equipment and material price fluctuation threshold) in the future N periods (such as the next 3 months) corresponding to the dynamic trend output by the deep learning model, and the linkage mutation coefficient is the change ratio of the actual total cost value of the current engineering stage (such as the construction drawing stage) and the actual total cost value of the previous engineering stage (such as the preliminary design stage) , quantifying the degree of cost mutation between stages, , and are dynamic weight factors, satisfying constraints, used to dynamically adjust the weights of each deviation component according to the engineering context (such as focusing on static deviation, focusing on dynamic trend stability). When implemented, the values of , and can be calculated by the rule engine from the risk decision vector (such as the data structure generated in the previous step), where and are calculated based on the real-time sub-cost data (such as the actual value of equipment purchase cost) and the stage total value (such as the transition total value from the feasibility study stage to the preliminary design stage) in the standardized cost database, predicted and output by the deep learning model, and the weight factor is dynamically optimized by engineering historical data, ensuring that the composite deviation rate accurately reflects the multi-dimensional deviation characteristics of the risk point, solving the problems of early warning lag and incomplete risk coverage.

[0081] In one of the embodiments, the method further comprises inputting the hierarchical early warning information into a visualization engine and performing the following steps:

[0082] S61, based on the chapter identification position in the early warning information, generating a heat map, and the color depth of the heat map maps the value of the composite deviation rate CDR;

[0083] S62, generating a deviation fluctuation atlas based on the current cost deviation value and the predicted deviation value in the early warning information, wherein the actual deviation value is associated with a static deviation amount , and the prediction interval is associated with a dynamic trend probability ;

[0084] S63, generating a cross-stage risk transmission tree diagram based on the cost influence range and the connection mutation coefficient in the early warning information , and the influence weight between nodes in the tree diagram is quantified by the value.

[0085] Specifically, the visualization engine is a software module specially used for converting structured early warning information into visual charts, and the graphic interface assists in risk decision-making; the heat map refers to a color-coded chart generated based on the chapter identification position in the early warning information (such as the specific chapter code of the construction engineering cost or the equipment purchase cost in the cost list), and the depth of the color scale is mapped by an algorithm to the value of the composite deviation rate CDR (for example, the color scale from light green to dark red represents the CDR from low to high, which intuitively displays the spatial distribution and severity of the risk point); the deviation fluctuation atlas is a chart showing the time sequence changes of the current cost deviation value and the predicted deviation value, wherein the actual deviation value is associated with a static deviation amount, representing the instantaneous deviation of the actual cost from the benchmark, and the prediction interval is associated with a dynamic trend probability (output by a deep learning model, representing the probability range of the cost exceeding the dynamic risk indicator in the next N periods such as 3 months); the cross-stage risk transmission tree diagram describes the propagation path of the risk point between the engineering stages (such as the research stage to the construction drawing stage), and the node represents the sub-item cost or the engineering stage, and the edge represents the transmission relationship, and the influence weight is quantified by the connection mutation amount, which is used to represent the influence intensity of the mutation degree on the adjacent node. When implemented, the chapter identification position in the early warning information can be parsed, the rendering algorithm of the visualization engine is called to generate the heat map, for example, the CDR value is mapped to the HSV color scale model using the OpenCV library, and a color gradient image is output to highlight the high-risk chapter; based on the deviation data (including the current actual deviation value and the predicted deviation value) in the early warning information, a time series analysis library (such as Matplotlib) is used to generate a deviation fluctuation atlas, the actual deviation line is plotted based on the value, and the prediction interval is calculated based on the confidence interval; a tree diagram is constructed by a graph theory algorithm (such as NetworkX), taking the cost influence range (such as the equipment purchase cost fluctuation affecting the installation engineering cost) in the early warning information as the node, and the value as the edge weight (such as the edge thickness increasing when the value is greater than 0.5), and the cross-stage transmission path of the risk is visualized. The heat map provides risk positioning, the deviation fluctuation atlas enhances the explainability of trend prediction, and the tree diagram reveals systematic transmission risks, which improves the accuracy of the whole process monitoring by intuitively displaying and feeding back to the closed-loop system.

[0086] The above power transmission and transformation project cost risk whole-process dynamic early warning method constructs a standardized cost database by obtaining cost data and benchmark data of each stage of the power transmission and transformation project, and constructs a risk index system covering static risk indexes (such as deviation threshold values of construction engineering cost, equipment purchase cost, installation engineering cost and other costs), dynamic risk indexes (such as equipment material price fluctuation trend and dynamic deviation rate based on engineering progress nodes), and stage correlation indexes (such as connection mutation threshold values from the feasibility study stage to the preliminary design stage and from the preliminary design stage to the construction drawing stage), uses a preset deep learning model to compare and trend forecast the cost data and the benchmark data in real time, synchronously executes feature comparison and cross-stage deduction to generate a risk decision vector (including static deviation amount, dynamic trend probability and connection mutation coefficient), identifies cost risk points and their section identification positions in the cost list, cost influence range and current deviation value; according to the deviation degree of the risk points, the composite deviation rate is calculated to determine the slight, general or serious risk level, and the rule engine generates graded early warning information (including section identification position, quantitative matrix and influence range) to push to the corresponding terminal (serious risk pushes to the first type of terminal to trigger the emergency response process, slight or general risk pushes to the second type of terminal to associate historical disposal reference information), and records the processing results to update the database to form a closed-loop feedback mechanism. In view of the early warning lag problem: the deep learning model integrates dynamic risk indexes (such as price fluctuation trend), responds to equipment material price changes or engineering progress adjustments in real time, and avoids delay caused by relying on static data; the risk coverage is not comprehensive: the stage correlation index captures the cross-stage transmission effect, and the cooperative analysis of static and dynamic indexes avoids the risk of engineering cost overrun caused by fragmented risk identification; in view of the low decision efficiency: the risk level determination and intelligent grading push mechanism based on the composite deviation rate optimize the early warning path, ensure that the serious risk triggers the emergency process quickly, and the slight risk associates the historical reference to improve the decision support efficiency, significantly improving the real-time, accuracy and overall decision efficiency of cost risk management.

[0087] It should be understood that, although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the direction of the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, there is no strict order limitation for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0088] Based on the same inventive concept, the embodiments of the present application also provide a power transmission and transformation project cost risk whole-process dynamic early warning system for implementing the power transmission and transformation project cost risk whole-process dynamic early warning method described above. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method, so the specific limitations in one or more power transmission and transformation project cost risk whole-process dynamic early warning system embodiments provided below can be referred to the limitations of the power transmission and transformation project cost risk whole-process dynamic early warning method described above, which will not be repeated here.

[0089] In one exemplary embodiment, as shown in Figure 2 a power transmission and transformation project cost risk whole-process dynamic early warning system is provided, comprising:

[0090] The data acquisition and processing module 101 is configured to acquire cost data and benchmark data of each stage of the power transmission and transformation project, and construct a standardized cost database.

[0091] The risk index management module 102 is configured to construct a risk index system based on preset risk rules, wherein the risk index system comprises static risk indexes, dynamic risk indexes, and stage correlation indexes.

[0092] The intelligent risk analysis module 103 is configured to compare the cost data and the benchmark data in the cost database in real time and predict trends based on the risk index system and using a preset deep learning model, and identify cost risk points.

[0093] The early warning generation and response module 104 is configured to determine a risk level and generate graded early warning information based on the deviation degree of the cost risk points, and push the graded early warning information to a corresponding terminal.

[0094] The feedback optimization module 105 is configured to record the processing results of the graded early warning information, and update the standardized cost database according to the processing results.

[0095] In one embodiment, the data acquisition and processing module 101 is further configured to:

[0096] acquire estimated cost data of the feasibility study stage, budget cost data of the preliminary design stage, and budget cost data of the construction drawing stage, wherein the cost data covers building engineering cost, equipment purchase cost, installation engineering cost, and other costs;

[0097] acquire cost data of similar historical projects, reference prices of equipment and materials in the region where the project is located, and industry standard rates as benchmark data;

[0098] standardize the cost data of each stage and the benchmark data, and construct a standardized cost database.

[0099] In one embodiment, the risk index management module 102 is further configured to:

[0100] Based on the benchmark data, corresponding static risk indicators are set for construction engineering cost, equipment purchase cost, installation engineering cost and other costs;

[0101] According to the historical price fluctuation rule, the dynamic risk indicators of equipment and material prices are set, and combined with the engineering progress, the stage dynamic risk indicators based on the engineering progress nodes are set, and the stage dynamic risk indicators are calculated according to the cumulative value of the sub-item cost;

[0102] For the path from the feasibility study stage to the preliminary design stage and the path from the preliminary design stage to the construction drawing stage, the stage correlation threshold is set as the stage correlation indicator.

[0103] In one of the embodiments, the intelligent risk analysis module 103 is further used for:

[0104] The static risk indicators, dynamic risk indicators and stage correlation indicators are input into the deep learning model, and the feature comparison of the cost data and the benchmark data and the cross-stage trend deduction are simultaneously performed through the deep learning model to generate a risk decision vector containing static deviation, dynamic trend and connection mutation;

[0105] When it is detected that the static deviation exceeds the static risk indicator, or the dynamic trend exceeds the dynamic risk indicator, or the connection mutation coefficient breaks through the stage correlation indicator, the corresponding type of cost risk point is identified, and the chapter identification position of the cost risk point in the cost list, the cost impact range and the current cost deviation are determined and output.

[0106] In one of the embodiments, the early warning generation and response module 104 is further used for:

[0107] Based on the static deviation amount, dynamic trend probability and connection mutation coefficient in the risk decision vector, the composite deviation rate is calculated, and the risk level is determined according to the preset mapping rule; the risk level includes slight, general or serious;

[0108] Based on the risk level, the early warning information is generated; the early warning information includes the chapter identification position of the risk point in the cost list corresponding to the risk level; the quantization matrix reflecting the current cost deviation value and the predicted deviation value of the risk point; the cost impact range of the risk point;

[0109] The early warning information is matched and pushed to the path according to the risk level through the rule engine; when the risk level is serious, it is pushed to the first type of terminal and triggers the preset emergency response process on the first type of terminal; when the risk level is slight or general, it is pushed to the second type of terminal and associated with the preset historical risk disposal reference information.

[0110] In one of the embodiments, the early warning generation and response module 104 is further used for calculating the composite deviation rate based on the static deviation amount, dynamic trend probability and connection mutation coefficient in the risk decision vector through the following formula:

[0111]

[0112] wherein is a composite deviation rate, is a static deviation amount, represents an actual value of a current stage sub-construction cost, represents a reference construction cost data, represents a deviation threshold corresponding to a static risk index set for construction engineering cost, equipment purchase cost, installation engineering cost and other costs; is a dynamic trend probability, representing a probability value of a future N period cost exceeding the dynamic risk index corresponding to the dynamic trend output by the deep learning model, and taking a value of ; is a connection mutation coefficient, represents an actual total cost value of a current engineering stage, represents an actual total cost value of a previous engineering stage; , and is a dynamic weight factor, satisfying .

[0113] In one of the embodiments, a visualization module is further included, configured to:

[0114] generate a heat map based on the chapter identification position in the early warning information, and the color depth of the heat map is mapped to the value of the composite deviation rate CDR;

[0115] generate a deviation fluctuation atlas based on the current cost deviation value and the predicted deviation value in the early warning information, wherein the actual deviation value is associated with the static deviation amount , and the prediction interval is associated with the dynamic trend probability ;

[0116] generate a cross-stage risk transmission tree diagram based on the cost influence range and the connection mutation coefficient in the early warning information, and the influence weight between nodes in the tree diagram is quantified by the value.

[0117] In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements the steps of the power transmission and transformation engineering cost risk whole-process dynamic early warning method as described above when executing the computer program.

[0118] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0119] For the device embodiment, since it basically corresponds to the method embodiment, the relevant part can be seen from the part of the method embodiment. The device embodiment described above is only schematic, wherein the components shown as separate components can or can not be physically separate, and the components shown as a unit can or can not be a physical unit, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present disclosure. Those skilled in the art can understand and implement it without creative labor.

[0120] The above-described embodiments only express several implementation manners of the present application, which are described in detail, but cannot be understood as a limitation on the patent scope of the application. It should be pointed out that, for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are all within the protection scope of the present application.

Claims

1. A power transmission project cost risk whole-process dynamic early warning method, characterized in that, The method comprises: Obtaining cost data and benchmark data of each stage of power transmission and transformation projects, and constructing a standardized cost database; Based on the preset risk rules, a risk index system is constructed, which includes static risk indicators, dynamic risk indicators and stage correlation indicators; Based on the risk index system, the cost data and benchmark data in the cost database are compared and trend forecasted in real time by using a preset deep learning model to identify cost risk points, wherein the use of the preset deep learning model to compare and trend forecast the cost data and benchmark data in the cost database to identify cost risk points comprises: Input the static risk indicators, dynamic risk indicators and stage correlation indicators into the deep learning model, and perform feature comparison and cross-stage trend deduction of the cost data and benchmark data by the deep learning model to generate a risk decision vector containing static deviation, dynamic trend probability and connection mutation coefficient; When the static deviation exceeds the static risk indicator, or the dynamic trend exceeds the dynamic risk indicator, or the connection mutation coefficient breaks through the stage correlation indicator, identify the corresponding type of cost risk point and determine the section identification position, cost impact range and current cost deviation of the cost risk point in the cost list and output; According to the deviation degree of the cost risk point, determine the risk level and generate a graded warning information and push it to the corresponding terminal, wherein the determination of the risk level and the generation of the graded warning information and the pushing to the corresponding terminal according to the deviation degree of the cost risk point comprises: Based on the static deviation, dynamic trend probability and connection mutation coefficient in the risk decision vector, calculate the compound deviation rate, and determine the risk level according to the preset mapping rule; the risk level includes slight, general or serious, wherein the compound deviation rate is calculated based on the static deviation, dynamic trend probability and connection mutation coefficient in the risk decision vector, which is realized by the following formula: wherein is a composite deviation rate, is a static deviation amount, represents the actual value of the current stage sub-construction cost, represents the reference construction cost data, represents the deviation threshold corresponding to the static risk index set for the construction cost, equipment purchase cost, installation engineering cost and other costs; is a dynamic trend probability, representing the probability value of the future N period construction cost exceeding the dynamic risk index corresponding to the dynamic trend output by the deep learning model, taking values ; is a connection mutation coefficient, represents the actual total construction cost of the current engineering stage, represents the actual total construction cost of the previous engineering stage; , and is a dynamic weight factor, satisfying ; Based on the risk level, generate warning information; the warning information includes the section identification position of the risk point in the cost list corresponding to the risk level; the quantization matrix reflecting the current cost deviation value and the predicted deviation value of the risk point; the cost impact range of the risk point; Through the rule engine, the warning information is matched and pushed to the path according to the risk level, the risk level is serious, and the first type of terminal is pushed to trigger the preset emergency response process on the first type of terminal, and the risk level is slight or general, and the second type of terminal is pushed and associated to provide preset historical risk disposal reference information; Record the processing result of the graded warning information, and update the standardized cost database according to the processing result.

2. The method of claim 1, wherein, The acquisition of cost data and benchmark data of each stage of power transmission and transformation projects, and the construction of a standardized cost database, comprises: Respectively acquiring estimated cost data of the feasibility study stage, budget cost data of the preliminary design stage and budget cost data of the construction drawing stage, the cost data covers building engineering cost, equipment purchase cost, installation engineering cost and other costs; Obtaining cost data of similar projects in history, reference prices of equipment and materials in the region where the project is located, and industry standard rates as the benchmark data; Standardizing the cost data of each stage and the benchmark data to construct the standardized cost database.

3. The method of claim 2, wherein, The construction of the risk index system includes: Based on the benchmark data, the construction cost, equipment purchase cost, installation cost and other costs are set with corresponding static risk indexes; According to the historical price fluctuation rule, the dynamic risk index of equipment and material price is set, and the stage dynamic risk index based on the engineering progress node is set according to the engineering progress and the cumulative value of sub-item cost; For the path from the feasibility study stage to the preliminary design stage and the path from the preliminary design stage to the construction drawing stage, set the stage correlation threshold as the stage correlation index.

4. The method of claim 1, wherein, The method further includes inputting the hierarchical early warning information into a visualization engine and performing the following steps: Based on the chapter identification position in the early warning information, a heat map is generated, and the color scale depth of the heat map maps the value of the composite deviation rate CDR; generate a deviation fluctuation atlas based on a current cost deviation value and a predicted deviation value in the early warning information, wherein the current cost deviation value is associated with the static deviation amount , and the predicted deviation value is associated with the dynamic trend probability ; Based on the cost impact range in the early warning information and the connection mutation coefficient , a cross-stage risk transmission tree diagram is generated, and the influence weight between nodes in the tree diagram is quantified by value.

5. A power transmission project cost risk whole-process dynamic early warning system, characterized in that, The system includes: A data acquisition and processing module is configured to obtain cost data of each stage of a power transmission and transformation project and benchmark data, and construct a standardized cost database; A risk index management module is configured to construct a risk index system based on preset risk rules, which includes static risk indexes, dynamic risk indexes, and stage correlation indexes; An intelligent risk analysis module is configured to compare and predict the trend of cost data and benchmark data in the cost database in real time based on the risk index system and a preset deep learning model, and identify cost risk points, wherein the comparison and prediction of the trend of cost data and benchmark data in the cost database based on the deep learning model includes: Inputting the static risk indexes, dynamic risk indexes and stage correlation indexes into the deep learning model, and synchronously executing feature comparison and cross-stage trend deduction of cost data and benchmark data through the deep learning model to generate a risk decision vector containing static deviation amount, dynamic trend probability and connection mutation coefficient; When it is detected that the static deviation amount exceeds the static risk index, or the dynamic trend exceeds the dynamic risk index, or the connection mutation coefficient breaks through the stage correlation index, the corresponding type of cost risk point is identified, the chapter identification position, cost impact range and current cost deviation of the cost risk point in the cost list are determined, and the cost risk point is outputted; An early warning generation and response module is configured to determine a risk level and generate hierarchical early warning information based on the deviation degree of the cost risk point, and push the hierarchical early warning information to a corresponding terminal, wherein the determination of the risk level and the generation of the hierarchical early warning information based on the deviation degree of the cost risk point and the pushing of the hierarchical early warning information to the corresponding terminal include: Based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, a composite deviation rate is calculated, and a risk level is determined according to a preset mapping rule; the risk level includes slight, general or serious, wherein the composite deviation rate is calculated based on the static deviation amount, the dynamic trend probability and the connection mutation coefficient in the risk decision vector, and is realized by the following formula: wherein is a composite deviation rate, is a static deviation amount, represents the actual value of the current stage sub-cost, represents the reference cost data, represents the deviation threshold corresponding to the static risk index set for the construction cost, equipment purchase cost, installation cost and other costs; is a dynamic trend probability, representing the probability value of the cost exceeding the dynamic risk index in the next N periods corresponding to the dynamic trend output by the deep learning model, taking values ; is a connection mutation coefficient, represents the actual total cost of the current project stage, represents the actual total cost of the previous project stage; , and is a dynamic weight factor, satisfying ; Based on the risk level, early warning information is generated; the early warning information includes a section identification position of a risk point corresponding to the risk level in the cost list; a quantization matrix reflecting a current cost deviation value and a predicted deviation value of the risk point; a cost influence range of the risk point; The early warning information is matched and pushed to a path according to the risk level by a rule engine; when the risk level is serious, the early warning information is pushed to a first type of terminal and triggers a preset emergency response process on the first type of terminal; when the risk level is slight or general, the early warning information is pushed to a second type of terminal and associated preset historical risk disposal reference information is provided; A feedback optimization module is configured to record a processing result of the graded early warning information, and update the standardized cost database according to the processing result. 6.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-4 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the method in any one of claims 1 to 4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method in any one of claims 1 to 4.

Images