Road engineering contract performance control and risk identification system and method

Abstract

The invention discloses a road engineering contract performance control and risk identification system and method. The system comprises a contract data analysis module; a multi-source data acquisition module; an intelligent performance management and control module; a risk automatic identification module; an early warning pushing module; and a data storage and visualization module. According to the invention, through intelligent analysis and multi-source data fusion, real-time monitoring and dynamic evaluation of the performance state are realized, and manual intervention is reduced; according to the method, the performance risk is pre-judged and graded, the accuracy and timeliness of risk identification are improved through a deep learning model, and the pertinence and timeliness of risk disposal are ensured through a graded early warning mechanism; a sustainable optimization performance management system is formed, meanwhile, cyclic utilization of data is achieved through a management mechanism, continuous iteration of a performance evaluation index system and a risk identification model is promoted, and personalized requirements of different road engineering projects are met; therefore, digital and intelligent transformation of road engineering management is promoted.

Description

Technical Field

[0001] This invention relates to the field of road engineering management and intelligent information technology, specifically to a road engineering contract intelligent performance management and risk automatic identification system and method, applicable to contract performance monitoring, risk prediction and management scenarios throughout the entire life cycle of road engineering projects. Background Technology

[0002] Road engineering projects are characterized by large investment scale, long construction period, multiple participating entities, and complex construction environment. Contracts, as the core basis for binding the rights and obligations of all parties, directly determine the achievement of project progress, quality, cost, and safety goals. With the rapid development of the infrastructure construction industry, the complexity of road engineering contract management is constantly increasing. Traditional contract performance control models are no longer sufficient to meet the needs of modern project management, and related fields are gradually introducing information technology to improve management efficiency. Currently, some technical solutions related to engineering contract management have emerged, mainly focusing on contract text management, basic data recording, and simple risk alerts. For example, some engineering management systems have contract text archiving and query functions, enabling the electronic storage of basic contract information; some solutions use manual input of construction progress and quality inspection data to form simple performance ledgers, assisting managers in monitoring project progress; and some risk identification solutions use a preset rule base to match and warn known risk points, such as setting threshold alerts for common problems like delays and substandard quality. Meanwhile, the application of technologies such as the Internet of Things (IoT), big data, and artificial intelligence (AI) in the engineering field is gradually expanding. Some projects have begun to use IoT devices to collect data from the construction site, such as tracking the location of construction equipment through GPS locators and collecting structural parameters through sensors. Natural language processing technology is also beginning to be applied to information extraction from contract texts, attempting to improve the automation level of contract management. The initial application of these technologies has, to some extent, alleviated the cumbersome nature of traditional manual management models, laying the foundation for the intelligent upgrade of contract performance control.

[0003] Existing shortcomings: Although existing technologies have been explored in the field of road engineering contract management and performance control, there are still many shortcomings that need to be addressed in practical applications, specifically: (1) Existing technologies for processing road engineering contract texts are mostly limited to electronic archiving, lacking specialized intelligent parsing capabilities for contracts in the engineering field. Due to the complexity of road engineering contract terms and the density of professional terminology, existing natural language processing solutions have not been optimized in conjunction with domain corpora, making it difficult to accurately extract core elements such as project scope, schedule milestones, payment terms, and liability for breach of contract. As a result, contract elements and performance process data cannot be effectively linked, and performance management lacks clear and accurate contractual basis, still requiring a large amount of manual intervention for verification, which is inefficient and prone to errors.

[0004] (2) Most data entry relies on manual input, resulting in issues such as data lag, omissions, and distortion. Even when some solutions utilize IoT devices, they primarily collect single-type data, failing to achieve collaborative collection and fusion processing of multi-source data from IoT devices, business systems, and manual supplementation. Multiple dimensions of performance data, such as construction progress, quality inspection, cost consumption, personnel and equipment input, and weather conditions, are scattered across different systems or ledgers, making it impossible to form a comprehensive and unified performance data view. This hinders managers from gaining a real-time and comprehensive understanding of the project's true performance status.

[0005] (3) Existing risk identification technologies mostly rely on pre-set rule bases for matching, which can only identify known and clear risk points. They cannot adapt to the complex and ever-changing risk factors in the road engineering contract performance process, such as sudden risks such as extreme weather, policy adjustments, and supply chain fluctuations. At the same time, risk identification relies too much on the experience and judgment of management personnel, which is highly subjective and makes it difficult to predict potential risks in advance. Often, risks can only be dealt with after they occur, resulting in delayed risk handling and increased losses.

[0006] (4) Existing technologies have not established a collaborative mechanism for performance control and risk identification. The results of performance status monitoring cannot drive risk identification in a timely manner, and the results of risk identification are also difficult to guide the optimization of performance control strategies. There is a lack of a graded response mechanism after risk warning, the warning information is not targeted enough, and the risk handling process is not tracked and recorded, resulting in poor effectiveness of control measures and difficulty in achieving dynamic optimization of the performance process.

[0007] The technical problems caused by the defects of the existing technology are as follows: The defects of the existing technology lead to problems such as low efficiency, insufficient accuracy, delayed risk prediction and poor control in the performance management of road engineering contracts. This not only increases the project management cost, but also easily leads to a series of problems such as construction delays, quality risks, cost overruns and contract disputes, which seriously affect the smooth progress of road construction and the achievement of project goals. Summary of the Invention

[0008] Therefore, in view of the shortcomings of the existing similar technologies, the present invention provides a road engineering contract performance management and risk identification system and method, which is an implementation scheme that can realize intelligent analysis of contract elements, multi-source data fusion and collection, automatic and accurate risk identification, and performance management and risk disposal, in order to meet the relevant technical needs in the current field of road engineering contract management.

[0009] This invention is implemented by constructing an intelligent performance management and risk identification system for road engineering contracts, comprising: The contract data parsing module uses natural language processing algorithms to perform structured parsing of road engineering contract texts, extracting core contract elements, including the scope of the project, schedule milestones, quality standards, payment terms, liability for breach of contract, and the rights and obligations of both parties, generating a standardized contract element database. The multi-source data acquisition module collects multi-dimensional data in real time during the road engineering contract performance process through three methods: IoT devices, business system interfaces, and manual input. The multi-dimensional data includes construction progress data, quality inspection data, cost consumption data, personnel and equipment input data, weather and environmental data, and policy change data. The intelligent performance management module constructs a performance evaluation index system based on the standardized contract element database, compares and analyzes the performance data collected by the multi-source data acquisition module with the performance evaluation index system, realizes dynamic monitoring of schedule performance, quality performance, cost performance and safety performance, and generates a performance status assessment report. The automatic risk identification module constructs a road engineering contract performance risk identification model. The standardized contract element data and real-time performance data are input into the risk identification model. Through feature extraction and risk matching, the model automatically identifies the types and levels of potential risks in the performance process. The potential risk types include construction delay risk, substandard quality risk, cost overrun risk, safety accident risk, and contract dispute risk. The early warning push module generates corresponding early warning information based on the risk type and risk level output by the automatic risk identification module, and pushes the early warning information to the corresponding responsible parties using a hierarchical push mechanism, while also pushing risk response strategy suggestions. The data storage and visualization module stores the standardized contract element data, real-time performance data, performance status assessment reports, and risk identification results. It displays the performance progress curve, risk distribution heat map, and key indicator compliance status through a visualization interface, and supports data backtracking and multi-dimensional queries.

[0010] According to the system of the present invention, the natural language processing algorithm used in the contract data parsing module includes the BERT pre-trained model. By fine-tuning the contract corpus in the field of road engineering, the accuracy of extracting core contract elements is improved. The standardized contract element database is stored in a relational database, which supports fast retrieval and related queries of contract elements.

[0011] According to the system of the present invention, the Internet of Things devices in the multi-source data acquisition module include GPS locators, high-definition cameras, sensors and drones, which are used to collect construction equipment location data, construction site image data, engineering structure parameter data and large-scale construction progress image data, respectively. The business system interfaces include API interfaces of project management system, financial system, quality inspection system and human resource management system.

[0012] According to the system described in this invention, the performance evaluation index system constructed by the intelligent performance management module includes the following performance indicators: schedule performance indicators include key node completion rate and schedule deviation rate; quality performance indicators include inspection pass rate and quality problem rectification closure rate; cost performance indicators include cost deviation rate and budget execution rate; and safety performance indicators include safety hazard investigation rate and safety accident occurrence rate. Each indicator is assigned a weight coefficient, and a weighted summation method is used to calculate the comprehensive performance score.

[0013] According to the system of the present invention, the risk identification model in the automatic risk identification module adopts a deep learning model that integrates LSTM and CNN. LSTM is used to extract the temporal features of performance data, and CNN is used to extract the spatial correlation features of risk features. The weights of key risk features are strengthened through an attention mechanism, and the model parameters are optimized by cross-validation during the model training process.

[0014] According to the system described in this invention, the hierarchical push mechanism of the early warning push module includes four levels of early warning: blue warning (general risk), yellow warning (significant risk), orange warning (major risk), and red warning (particularly serious risk). Different levels of early warning correspond to different push channels and response time limits. Red warnings need to be pushed to project leaders and corporate management in real time, and the emergency response process needs to be initiated.

[0015] A method for contract performance management and risk identification in road engineering projects includes the following steps: S1. Structured parsing of contract text: Natural language processing algorithms are used to segment, tag, and identify entities in road engineering contract texts, extract core contract elements, and generate a standardized contract element database. S2. Multi-source performance data collection: Real-time collection of progress, quality, safety and environmental data at the construction site through IoT devices, synchronous acquisition of system data such as project management and financial payment through business system interfaces, and supplementation of unstructured data by manual input to achieve real-time aggregation of multi-dimensional performance data; S3. Dynamic Management of Performance Status: Based on a standardized contract element database, a performance evaluation index system is constructed that includes four dimensions: schedule, quality, cost, and safety. The collected real-time performance data is compared with the evaluation index to calculate the performance compliance rate and generate a dynamic performance status assessment report. S4. Automatic identification of performance risks: A risk identification model is built based on machine learning algorithms. The model is trained and optimized using historical performance data and risk cases. Standardized contract element data and real-time performance data are input into the model. Through feature extraction and risk matching, potential risk types and risk levels are automatically identified. S5. Risk Warning and Response: Generate graded warning information based on the identified risk type and risk level, push it to the corresponding responsible entity, and at the same time associate and match the preset risk response strategy to achieve timely handling and closed-loop management of risks.

[0016] According to the method described in this invention, after performing structured parsing on the contract text in step S1, the method further includes verifying the extracted core elements of the contract. By comparing with standard contract templates in the field of road engineering and through manual review, errors in element extraction are corrected to ensure the accuracy of the standardized contract element database.

[0017] According to the method described in this invention, the training process of the risk identification model in step S4 includes: collecting historical data on the performance of road engineering contracts and typical risk cases, constructing a risk sample dataset, labeling the sample data with features, dividing the labeled sample data into a training set, a validation set and a test set, training an initial model using the training set, optimizing the model parameters using the validation set, verifying the model's identification accuracy using the test set, and completing the model training when the accuracy reaches a preset threshold.

[0018] According to the method described in this invention, step S5 further includes tracking the risk handling process, recording risk handling measures, handling progress and handling effects, forming a closed-loop risk handling file, and feeding the risk handling data back to the risk identification model to achieve continuous iterative optimization of the model.

[0019] This invention has the following advantages: It targets the field of road engineering management and intelligent information technology, aiming to solve the following technical problems existing in the performance management of road engineering contracts: 1) Contract texts are unstructured data, and the extraction of core elements relies on manual labor, resulting in low efficiency, large errors, and difficulty in achieving accurate matching with the performance process; 2) Performance data sources are scattered, lacking a unified data collection and integration mechanism, leading to untimely and incomplete monitoring of performance status; 3) Risk identification relies on manual experience judgment, which is highly subjective, making it difficult to predict potential risks in advance, and resulting in delayed risk handling; 4) Performance management and risk identification are disconnected, lacking a closed-loop management mechanism, making it impossible to achieve dynamic optimization of the performance process.

[0020] The prominent substantive features obtained by this invention include: (1) This invention proposes an intelligent parsing scheme for core contract elements based on the BERT pre-trained model. By fine-tuning the contract corpus in the field of road engineering, it realizes the transformation of unstructured contract text into standardized element data, solves the problems of low efficiency and large error in the extraction of contract elements in the existing technology, and provides accurate contract basis for performance management.

[0021] (2) This invention constructs a multi-source data collection form of IoT device + business system interface + manual input, thereby realizing the real-time aggregation and fusion processing of multi-dimensional performance data such as construction progress, quality, cost, and safety, comprehensively covering the key links of road engineering performance, and solving the problem of single and lagging data sources in the existing technology.

[0022] (3) The present invention designs a deep learning risk identification model that integrates LSTM and CNN, and strengthens key risk features by combining attention mechanism. It does not rely on preset risk rules and can automatically learn the temporal and spatial correlation features in the performance data, thereby achieving accurate identification of complex and ever-changing risk factors and improving the flexibility and accuracy of risk identification.

[0023] (4) This invention establishes a management method of performance monitoring, risk identification, hierarchical early warning, response and disposal, and data feedback, thereby deeply integrating performance control and risk identification. By using risk disposal data to reverse optimize the performance evaluation index system and risk identification model, dynamic optimization of the performance process is achieved, solving the problem of disconnect between control and risk identification in the prior art.

[0024] The significant advancements brought about by this invention are also reflected in: 1. Significantly improving the efficiency and accuracy of contract performance management in road engineering projects. Through intelligent analysis and multi-source data fusion, real-time monitoring and dynamic evaluation of performance status are achieved, reducing manual intervention and lowering management costs; 2. Enabling early prediction and tiered handling of performance risks. Deep learning models improve the accuracy and timeliness of risk identification, and a tiered early warning mechanism ensures targeted and timely risk handling, effectively reducing risk losses; 3. Forming a sustainable and optimized performance management system. Through management mechanisms, data recycling is achieved, promoting the continuous iteration of performance evaluation indicator systems and risk identification models to adapt to the personalized needs of different road engineering projects; 4. Promoting the digital and intelligent transformation of road engineering management. Integrating various advanced technologies such as natural language processing, deep learning, and the Internet of Things, intelligent management of the entire contract management and performance control process is realized, thereby improving the overall level of road engineering project management. Attached Figure Description

[0025] Figure 1 This is a system structure diagram of this application; Figures 2-6 This is the flowchart of the application process. Detailed Implementation

[0026] The following will be combined with the appendix Figures 1-6 This invention will be described in detail, and the technical solutions in the embodiments of this invention will be clearly and completely described. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0027] Example 1: This invention provides an intelligent performance management and risk identification system for road engineering contracts. See [link to relevant documentation]. Figure 1 The system includes: The contract data parsing module uses natural language processing algorithms to perform structured parsing of road engineering contract texts, extracting core contract elements, including the scope of the project, schedule milestones, quality standards, payment terms, liability for breach of contract, and the rights and obligations of both parties, generating a standardized contract element database. The multi-source data acquisition module collects multi-dimensional data in real time during the road engineering contract performance process through three methods: IoT devices, business system interfaces, and manual input. The multi-dimensional data includes construction progress data, quality inspection data, cost consumption data, personnel and equipment input data, weather and environmental data, and policy change data. The intelligent performance management module constructs a performance evaluation index system based on the standardized contract element database, compares and analyzes the performance data collected by the multi-source data acquisition module with the performance evaluation index system, realizes dynamic monitoring of schedule performance, quality performance, cost performance and safety performance, and generates a performance status assessment report. The automatic risk identification module constructs a road engineering contract performance risk identification model. The standardized contract element data and real-time performance data are input into the risk identification model. Through feature extraction and risk matching, the model automatically identifies the types and levels of potential risks in the performance process. The potential risk types include construction delay risk, substandard quality risk, cost overrun risk, safety accident risk, and contract dispute risk. The early warning push module generates corresponding early warning information based on the risk type and risk level output by the automatic risk identification module, and pushes the early warning information to the corresponding responsible parties using a hierarchical push mechanism, while also pushing risk response strategy suggestions. The data storage and visualization module stores the standardized contract element data, real-time performance data, performance status assessment reports, and risk identification results. It displays the performance progress curve, risk distribution heat map, and key indicator compliance status through a visualization interface, and supports data backtracking and multi-dimensional queries.

[0028] Based on the system described in Embodiment 1, the natural language processing algorithm used in the contract data parsing module includes the BERT pre-trained model. By fine-tuning the contract corpus in the field of road engineering, the accuracy of extracting core contract elements is improved. The standardized contract element database is stored in a relational database, which supports fast retrieval and related queries of contract elements.

[0029] Based on the system described in Embodiment 1, the IoT devices in the multi-source data acquisition module include GPS locators, high-definition cameras, sensors, and drones, which are used to collect construction equipment location data, construction site image data, engineering structure parameter data, and large-scale construction progress image data, respectively. The business system interfaces include API interfaces of project management system, financial system, quality inspection system, and human resource management system.

[0030] Based on the system described in Example 1, the performance evaluation index system constructed by the intelligent performance management module includes the following performance indicators: schedule performance indicators include key node completion rate and schedule deviation rate; quality performance indicators include inspection pass rate and quality problem rectification closure rate; cost performance indicators include cost deviation rate and budget execution rate; and safety performance indicators include safety hazard investigation rate and safety accident occurrence rate. Each indicator is assigned a weight coefficient, and a weighted summation method is used to calculate the comprehensive performance score.

[0031] Based on the system described in Example 1, the risk identification model in the automatic risk identification module adopts a deep learning model that integrates LSTM and CNN. LSTM is used to extract the temporal features of performance data, and CNN is used to extract the spatial correlation features of risk features. The weights of key risk features are strengthened through an attention mechanism, and the model parameters are optimized by cross-validation during the model training process.

[0032] Based on the system described in Embodiment 1, the hierarchical push mechanism of the early warning push module includes four levels of early warning: blue warning (general risk), yellow warning (significant risk), orange warning (major risk) and red warning (particularly serious risk). Different levels of early warning correspond to different push channels and response time limits. Red warnings need to be pushed to the project leader and enterprise management in real time and the emergency response process needs to be initiated.

[0033] Example 2: A method for intelligent performance management and automatic risk identification of road engineering contracts, such as... Figures 2-6 This includes the following steps: S1. Structured parsing of contract text: Natural language processing algorithms are used to segment, tag, and identify entities in road engineering contract texts, extract core contract elements, and generate a standardized contract element database. S2. Multi-source performance data collection: Real-time collection of progress, quality, safety and environmental data at the construction site through IoT devices, synchronous acquisition of system data such as project management and financial payment through business system interfaces, and supplementation of unstructured data by manual input to achieve real-time aggregation of multi-dimensional performance data; S3. Dynamic Management of Performance Status: Based on a standardized contract element database, a performance evaluation index system is constructed that includes four dimensions: schedule, quality, cost, and safety. The collected real-time performance data is compared with the evaluation index to calculate the performance compliance rate and generate a dynamic performance status assessment report. S4. Automatic identification of performance risks: A risk identification model is built based on machine learning algorithms. The model is trained and optimized using historical performance data and risk cases. Standardized contract element data and real-time performance data are input into the model. Through feature extraction and risk matching, potential risk types and risk levels are automatically identified. S5. Risk Warning and Response: Generate graded warning information based on the identified risk type and risk level, push it to the corresponding responsible entity, and at the same time associate and match the preset risk response strategy to achieve timely handling and closed-loop management of risks.

[0034] In step S1 of the method described in Example 2, after the contract text is parsed in a structured manner, the core elements of the extracted contract are verified. By comparing with standard contract templates in the field of road engineering and through manual review, errors in element extraction are corrected to ensure the accuracy of the standardized contract element database.

[0035] In step S4 of the method described in Example 2, the training process of the risk identification model includes: collecting historical data on the performance of road engineering contracts and typical risk cases, constructing a risk sample dataset, labeling the sample data with features, dividing the labeled sample data into a training set, a validation set and a test set, training an initial model using the training set, optimizing the model parameters using the validation set, verifying the model's recognition accuracy using the test set, and completing the model training when the accuracy reaches a preset threshold.

[0036] In step S5 of the method described in Example 2, the risk disposal process is further tracked, risk disposal measures, disposal progress and disposal effect are recorded to form a closed-loop risk disposal file, and the risk disposal data is fed back to the risk identification model to realize the continuous iterative optimization of the model.

[0037] The following is an implementation example of an intelligent performance management and automatic risk identification system and method for road engineering contracts. I. Overview of Implementation Examples This embodiment uses a provincial expressway expansion project (K0+000-K25+600 section) as the application scenario. The project is 25.6 kilometers long, with a total contract value of 1.28 billion yuan and a construction period of 24 months, involving multiple sub-projects such as roadbed engineering, pavement engineering, and bridge engineering. Based on the technical solution defined in claims 1-10, this embodiment builds an intelligent performance management and automatic risk identification system, fully executes the corresponding management methods, and verifies the feasibility and effectiveness of the solution.

[0038] This embodiment aims to solve problems such as "low efficiency in extracting contract elements, scattered performance data, and delayed risk prediction" in the contract performance process of this project. By integrating technologies such as natural language processing, deep learning, and the Internet of Things, it achieves intelligent management and control of the entire contract performance process, ensuring that the project progresses in accordance with contract requirements.

[0039] II. System Implementation Examples 2.1 System Overall Architecture The intelligent performance management and automatic risk identification system in this embodiment adopts a three-tier "cloud-edge-device" architecture: the terminal layer deploys IoT devices and data acquisition terminals, the edge layer implements data preprocessing and local caching, and the cloud layer deploys core algorithm modules, data storage modules, and a visualization platform. The system hardware configuration includes: an edge computing gateway (CPU: Intel Core i7-12700H, 32GB memory, 1TB storage), a cloud server cluster (8 high-performance servers, deployed with load balancing), and IoT terminal devices (68 devices in total, including GPS locators and high-definition cameras). The software environment includes: an operating system of Ubuntu 22.04 LTS, a programming language of Python 3.9, a deep learning framework of TensorFlow 2.10, and databases of MySQL 8.0 (relational database) and Redis 6.2 (caching database).

[0040] 2.2 Specific Implementation of Each Module 2.2.1 Contract Data Parsing Module This module employs a natural language processing algorithm based on the BERT pre-trained model, as implemented below: Model Selection and Fine-tuning: A BERT-Base-Chinese pre-trained model was selected and fine-tuned based on a corpus from the road engineering domain. The domain corpus contains 1,000 similar highway engineering contract texts (approximately 8.6 million characters in total), with 32,000 annotated samples covering core elements such as project scope, construction schedule, and quality standards. Fine-tuning parameters: learning rate 2e-5, 10 iterations, batch size 32, and the cross-entropy loss function was used to optimize the model parameters.

[0041] Core element extraction: After preprocessing the project contract text (PDF format, 186 pages in total) (format conversion, noise reduction), word segmentation (using jieba word segmentation tool, custom road engineering domain dictionary, including 2300 professional terms such as "subgrade compaction" and "asphalt paving thickness"), part-of-speech tagging (based on NLTK tool) and entity recognition were performed to extract core elements including: project scope (clearly defining the subgrade, pavement and other engineering contents of section K0+000-K25+600), construction period (start date 202X, key nodes: subgrade completion date 202X, pavement completion date 202Y, etc., 8 key nodes), quality standards (subgrade compaction ≥96%, pavement smoothness σ≤1.2mm, etc.), payment terms (progress payments are made monthly, with a payment ratio of 80% of the completed work), breach of contract clauses (a penalty of 0.05% of the total contract amount will be paid for each day of delay in the construction period), and the rights and obligations of both parties.

[0042] Standardized Database Construction: A standardized contract element database is built using MySQL 8.0, comprising six tables including a basic contract information table, a schedule milestone table, and a quality standard table. These tables are linked by contract number. The database supports fast retrieval by dimensions such as "project division," "time milestone," and "element type," with a retrieval response time of ≤0.5 seconds.

[0043] Element verification: By comparing with the "Construction Project Construction Contract (Model Text)" (GF-2017-0201) and combining manual review by 3 road engineering experts, 2 errors in element extraction were corrected (such as "asphalt pavement paving temperature ≥150℃" was mistakenly extracted as "≥140℃"), ensuring that the database accuracy rate is ≥99.5% (corresponding to claim 8).

[0044] 2.2.2 Multi-source data acquisition module This module adopts a multi-source data acquisition architecture of "IoT devices + business system interfaces + manual data entry", and the specific implementation is as follows: IoT device deployment: ① GPS locators (model: GT06N): installed on 32 construction machines (excavators, pavers, etc.) to collect equipment location data (positioning accuracy ±1m), data update frequency once every 30 seconds; ② High-definition cameras (model: Hikvision DS-2CD3T46WD-I3): 15 monitoring points deployed at the construction site to collect construction site image data (resolution 1080P, frame rate 25 frames / second); ③ Sensors: 16 sensors including concrete strength sensor (model: JMZX-212) and roadbed compaction sensor (model: HC-CY20) are deployed to collect engineering structural parameter data (data accuracy 0.1%FS); ④ Drones (model: DJI Phantom 4) RTK: Flights are conducted twice daily, at 10:00 AM and 4:00 PM, to collect large-scale construction progress image data (aerial photography accuracy 1cm / pixel); ⑤ Data transmission: Using the E860-DTU(4040-4G)-V2 4G wireless transmission module, data is transmitted to the edge computing gateway via the MQTT protocol, with a transmission delay of ≤5 seconds.

[0045] Business System Interface Integration: Four business systems are integrated via RESTful API interfaces: ① Project Management System (Glodon BIM5D): Synchronously acquires data such as construction progress plans and completion status of sub-projects, with a synchronization frequency of once per hour; ② Financial System (Yonyou U9 Cloud): Acquires data such as project payment records and cost consumption details, with a synchronization frequency of once per day; ③ Quality Inspection System (Ministry of Transport Highway Engineering Quality Inspection Management System): Acquires data such as raw material inspection reports and sub-project quality assessments, with a synchronization frequency of once per 24 hours; ④ Human Resources Management System: Acquires data such as construction worker attendance and special operation personnel qualifications, with a synchronization frequency of once per day. The interfaces use OAuth2.0 authentication to ensure secure data transmission.

[0046] Manual data entry: Standardized data entry forms are designed for on-site management personnel to supplement and enter unstructured data (such as construction logs, on-site meeting minutes, emergency situation reports, etc.). The forms support multiple formats such as text, images, and audio, and the entered data is automatically associated with timestamps and geographical location information.

[0047] 2.2.3 Intelligent Performance Management Module This module constructs a performance evaluation index system based on a standardized contract element database, as specifically implemented below: Evaluation index system construction: Referring to the "Implementation Rules for Performance Evaluation of Construction Contractors in Longgang District", 4 primary indicators, 8 secondary indicators, and weight coefficients were determined (weights for primary indicators: construction period 0.3, quality 0.3, cost 0.2, safety 0.2). The specific indicators are as follows: Dynamic monitoring and evaluation: A weighted summation method is used to calculate the comprehensive performance score (comprehensive score = Σ (secondary indicator score × corresponding weight)). The secondary indicator scores are divided into 5 levels according to the performance rate (100%, 80%, 60%, 30%, 0%). The system compares the collected performance data with the indicator standards in real time. For example, if the completion rate of key nodes of a roadbed project in a certain month is 92% (score 92 points × 0.18 = 16.56 points) and the schedule deviation rate is -2% (ahead of schedule, score 100 points × 0.12 = 12 points), the final comprehensive performance score for that month is 89.6 points, generating a "Monthly Performance Status Assessment Report" that identifies weaknesses such as "cost deviation rate exceeding 5%" and "safety hazard rectification closure rate of 82%".

[0048] 2.2.4 Automatic Risk Identification Module This module uses a deep learning model that combines LSTM and CNN, and the specific implementation is as follows: Model Construction: The model consists of an input layer, an LSTM temporal feature extraction layer, a CNN spatial feature extraction layer, an attention mechanism layer, and an output layer. The LSTM layer has two layers with 128 hidden units, used to extract temporal features of the performance data (such as cost trends over three consecutive months). The CNN layer has two convolutional layers (3×3 kernels, 64 kernels) and one pooling layer (2×2 kernels), used to extract spatial correlation features of risk characteristics (such as the correlation between quality problems and construction worker qualifications). The attention mechanism layer uses the Bahdanau attention mechanism to strengthen the weights of key risk features such as "sudden increase in schedule deviation rate" and "quality inspection failure".

[0049] Model Training: Historical performance data (120,000 records) from 50 similar highway projects and 86 typical risk cases (such as "delayed construction during the rainy season" and "cost overruns due to rising asphalt material prices") were collected to construct a risk sample dataset. The sample data was feature-labeled (labeling risk type, risk level, etc.) and divided into training, validation, and test sets in a 7:2:1 ratio. Cross-validation (50% fold) was used to optimize model parameters. After training, the model achieved a test set accuracy of 94.2%, meeting the requirements for practical applications.

[0050] Risk identification: Standardized contract element data (such as project schedule and cost budget) and real-time performance data (such as current project schedule deviation rate of 3% and cost deviation rate of 6%) are input into the trained model. Through feature extraction and risk matching, potential risks are automatically identified as "cost overrun risk (significant risk)" and "project delay potential risk (general risk)". The risk level is determined based on the risk score (0-30 points are general risk, 31-60 points are significant risk, 61-85 points are major risk, and 86-100 points are particularly major risk).

[0051] 2.2.5 Early Warning Push Module This module adopts a four-level hierarchical push mechanism, as implemented below: Warning levels and judgment thresholds: ① Blue warning (general risk): risk score 0-30 points, such as "project schedule deviation rate ≤3%"; ② Yellow warning (significant risk): risk score 31-60 points, such as "cost deviation rate 5%-10%"; ③ Orange warning (major risk): risk score 61-85 points, such as "quality inspection pass rate ≤85%"; ④ Red warning (particularly serious risk): risk score 86-100 points, such as "occurrence of a general safety accident".

[0052] Push channels and response time limits: ① Blue alert: Pushed to the construction team leader via system message, response time limit 48 hours; ② Yellow alert: Pushed to the project manager via system message + SMS, response time limit 24 hours; ③ Orange alert: Pushed to the project leader and the company's engineering management department via system message + SMS + email, response time limit 12 hours; ④ Red alert: Pushed in real time (≤5 minutes) to the project leader, company management and supervision unit via system message + SMS + telephone voice, and the emergency response process is activated.

[0053] Risk response strategy push: The system has a preset risk response strategy library. For the identified "cost overrun risk (major risk)", the system will push the following strategies: "Optimize the construction plan and reduce material waste; recalculate the material purchase price and negotiate with the supplier to reduce the price; apply for adjustment of the budget approval process".

[0054] 2.2.6 Data Storage and Visualization Module This module adopts an architecture of "MySQL + Redis + visualization platform", and its specific implementation is as follows: Data storage: The MySQL database stores standardized contract element data, real-time performance data, performance assessment reports, and risk identification results (storage period is 5 years after project completion); the Redis cache database stores frequently accessed data (such as real-time performance indicators and current risk status) to improve data query efficiency.

[0055] Visualization: The web-based visualization interface is built using ECharts visualization components, displaying the following content: ① Progress curve (comparing the changing trends of actual and planned progress); ② Risk distribution heatmap (displaying risk level distribution by project segment, with red areas indicating high-risk zones); ③ Key indicator compliance dashboard (intuitively displaying the scores and compliance status of 8 secondary indicators). It supports multi-dimensional queries by "time dimension (day / week / month)," "project segment," and "risk type," with data backtracking response time ≤ 1 second.

[0056] III. Method Examples The intelligent performance management and automatic risk identification method in this embodiment is executed according to the following steps: 3.1 Step S1: Structured Parsing of Contract Text S11: Obtain the contract text (PDF format) for the highway expansion project, convert it to TXT format using a format conversion tool, and remove redundant information such as headers, footers, and watermarks; S12: Employs a natural language processing algorithm based on the BERT fine-tuning model to perform word segmentation (using a custom domain dictionary), part-of-speech tagging (tags "noun", "verb", "technical term", etc.) and entity recognition on the contract text, extracting 6 core elements such as project scope and project schedule. S13: The extracted core elements are compared with the "Construction Project Construction Contract (Model Text)" and manually reviewed by 3 road engineering experts to correct any extraction errors; S14: Enter the verified core elements into the MySQL database according to the preset format to generate a standardized contract element database.

[0057] 3.2 Step S2: Multi-source performance data collection S21: Real-time data collection via IoT devices: GPS locators collect construction machinery location data, sensors collect engineering structural parameters such as roadbed compaction and concrete strength, high-definition cameras collect construction site images, and drone aerial photography collects construction progress images. The data is transmitted to the edge computing gateway via a 4G module. S22: Synchronize business system data via API interface: Obtain construction schedule from project management system, cost consumption data from financial system, and inspection report from quality inspection system. Data synchronization frequency is set according to system requirements. S23: On-site management personnel manually enter unstructured data such as construction logs and meeting minutes using standardized forms; S24: The edge computing gateway preprocesses multi-source data (cleaning, deduplication, format conversion) and uploads the processed data to the cloud server.

[0058] 3.3 Step S3: Dynamic Management of Contract Performance Status S31: Based on a standardized contract element database, construct a performance evaluation index system that includes four dimensions: schedule, quality, cost, and safety, and determine the weight and calculation method of each index; S32: Compare the preprocessed real-time performance data with the evaluation indicator standards, and calculate the performance rate and score of each secondary indicator; S33: Calculate the overall performance score using the weighted summation method, and generate a dynamic performance status assessment report that includes indicator scores, weak links, and improvement suggestions; S34: Push the assessment report to the relevant responsible parties to support performance decisions.

[0059] 3.4 Step S4: Automatic identification of performance risks S41: Collect historical performance data and typical risk cases of similar projects, construct a risk sample dataset, and perform feature annotation on the sample data; S42: Construct a deep learning risk identification model that integrates LSTM and CNN. Divide the sample dataset into training set, validation set and test set. Train the model using the training set, optimize the parameters using the validation set, and verify the model accuracy (reaching the threshold ≥90%) using the test set. S43: Input standardized contract element data and real-time performance data into the trained model. The model extracts temporal features through LSTM, extracts spatial correlation features through CNN, and strengthens key features by combining an attention mechanism. S44: Automatically identify potential risk types (such as cost overrun risk) and risk levels (such as major risk) through risk matching algorithms.

[0060] 3.5 Step S5: Risk Warning and Response S51: Generate graded early warning information based on risk type and risk level, specifying the warning content, risk level, and scope of impact; S52: The warning information will be pushed to the responsible entities through the corresponding channels according to the four-level push mechanism, and the emergency response will be activated when a red warning is issued; S53: Related push notifications include pre-defined risk response strategies; the responsible party formulates and implements handling plans based on these strategies. S54: The system tracks the risk management process, records the management measures, progress and effects, and forms a closed-loop risk management archive; S55: Feed the risk management data back to the risk identification model, optimize the model parameters, and achieve continuous model iteration (corresponding to claim 10).

[0061] The social benefits and use value obtained by implementing this invention will be described below; Social benefits: The implementation of this invention will generate significant social benefits from multiple levels, including the development of the road engineering industry, the protection of public interests, and the optimization of resource allocation. Specifically, it is reflected in the following aspects: (1) Through the intelligent performance management module, the dynamic monitoring of quality and safety indicators, and the automatic risk identification module, the early prediction of risks such as substandard quality and safety accidents, can significantly reduce the occurrence of quality hazards and safety accidents during the construction of road engineering projects. The completed road engineering projects will be more secure in terms of structural stability and traffic safety, directly benefiting a wide range of traffic participants, reducing public safety incidents such as traffic congestion and accidents caused by engineering quality problems, and safeguarding the public interest. (2) This invention integrates advanced technologies such as natural language processing, deep learning, and the Internet of Things to break the backward model of traditional road engineering contract management that is dominated by manual labor and driven by experience, and builds a full-process intelligent performance management system. Its implementation will provide the industry with a replicable and scalable digital transformation solution, guide industry enterprises to improve their technology application capabilities, promote the upgrading of the overall management level of the industry, and help the engineering construction field achieve the goal of high-quality development under the background of "new infrastructure". (3) By using multi-source data fusion analysis to achieve precise control over cost performance, it is possible to effectively reduce problems such as material waste and idle funds during the construction process. At the same time, dynamic monitoring and optimization of the construction period can shorten the project construction cycle and reduce the long-term interference of construction on the surrounding environment (such as noise and dust pollution). In addition, the early handling of risks avoids the consumption of resources such as rework and reconstruction due to the expansion of risks, which is in line with the concept of green development and reflects the construction of a conservation-oriented society. (4) This patent achieves precise analysis of the core elements of the contract through the BERT model, which clarifies the key contents such as the rights and obligations of both parties and the performance standards, and provides a clear contractual basis for performance control. At the same time, the closed-loop management mechanism ensures that the performance process is traceable throughout and the risk handling process is fully documented, which can effectively reduce disputes between the parties due to misunderstandings of the contract and inadequate performance control. This helps to regulate the market transaction order of the road engineering industry, maintain a fair competitive environment in the industry, and reduce social conflicts caused by engineering disputes.

[0062] Use value: The technology of this invention addresses the core needs of various stakeholders in road construction (construction unit, construction unit, supervision unit, and industry regulatory department), and its use value is significant: (1) For construction units, it enables precise control and risk prevention of project performance, reducing project management costs. Construction units can monitor the performance status of core indicators such as schedule, quality, cost, and safety in real time through a visual interface, without relying on manual summarization and analysis; the automatic risk identification and graded early warning function can predict potential risks such as cost overruns and schedule delays in advance, and combined with the related push response strategies, it helps them make quick decisions and timely handling, effectively avoiding investment losses caused by risk out of control, and ensuring that the project is promoted efficiently according to the contract requirements. (2) For construction units, it improves performance efficiency, strengthens process control capabilities, and enhances market competitiveness. Construction units can use multi-source data acquisition modules to achieve real-time aggregation of construction site data, reducing the workload of manual data entry; the performance evaluation index system provides clear performance standard guidance, helping them to accurately locate weak links in performance and rectify them in a timely manner; the feedback optimization mechanism of risk disposal data can continuously improve their risk response capabilities, enhance corporate reputation through standardized performance, and strengthen their competitive advantage in the market. (3) For supervision units, it can improve the accuracy and efficiency of supervision work and reduce the workload of supervision. Supervision units can directly obtain real-time data such as construction progress and quality inspection through the system without repeated on-site verification; the risk identification results provide clear guidance for key supervision work, enabling them to focus on high-risk links to carry out supervision work, improve supervision efficiency and quality, and better fulfill their supervision responsibilities. (4) For industry regulatory departments, it can realize dynamic supervision and macro-control of road engineering projects and improve the efficiency of industry governance. Regulatory authorities can access project performance data and risk information through the system to gain a comprehensive understanding of road construction within their jurisdiction, promptly identify common industry problems and prominent risks, and conduct precise supervision and law enforcement inspections based on data backtracking and multi-dimensional query functions. At the same time, it can provide real and comprehensive data support for industry policy formulation, thereby improving the scientific nature and pertinence of industry governance.

[0063] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A road engineering contract performance management and risk identification system, characterized in that, The system includes: The contract data parsing module uses natural language processing algorithms to perform structured parsing of road engineering contract texts, extract core contract elements, and generate a standardized contract element database. The multi-source data acquisition module collects multi-dimensional data in real time during the road engineering contract performance process through three methods: IoT devices, business system interfaces, and manual input. The intelligent performance management module constructs a performance evaluation index system based on the standardized contract element database. It is used to compare and analyze the performance data collected by the multi-source data acquisition module with the performance evaluation index system, realize dynamic monitoring of schedule performance, quality performance, cost performance and safety performance, and generate a performance status assessment report. The automatic risk identification module constructs a road engineering contract performance risk identification model. The standardized contract element data and real-time performance data are input into the risk identification model. Through feature extraction and risk matching, the potential risk types and risk levels in the performance process are automatically identified. The early warning push module is configured to generate corresponding early warning information based on the risk type and risk level output by the risk automatic identification module, and push the early warning information to the corresponding responsible parties using a hierarchical push mechanism, while also pushing risk response strategy suggestions. The data storage and visualization module stores the standardized contract element data, real-time performance data, performance status assessment reports, and risk identification results. It displays the performance progress curve, risk distribution heat map, and key indicator compliance status through a visualization interface, and supports data backtracking and multi-dimensional queries.

2. The road engineering contract performance management and risk identification system according to claim 1, characterized in that, The contract data parsing module employs a natural language processing algorithm including the BERT pre-trained model. By fine-tuning the contract corpus in the road engineering field, it improves the accuracy of extracting core contract elements. The standardized contract element database is stored using a relational database, supporting rapid retrieval and related queries of contract elements.

3. The road engineering contract performance management and risk identification system according to claim 1, characterized in that, The IoT devices in the multi-source data acquisition module include GPS locators, high-definition cameras, sensors, and drones, which are used to collect construction equipment location data, construction site image data, engineering structure parameter data, and large-scale construction progress image data, respectively. The business system interfaces include API interfaces for project management system, financial system, quality inspection system, and human resource management system.

4. The road engineering contract performance management and risk identification system according to claim 1, characterized in that, The performance evaluation index system constructed by the intelligent performance management module includes the following performance indicators: schedule performance indicators include the completion rate of key nodes and schedule deviation rate; quality performance indicators include the inspection pass rate and the closed-loop rectification rate of quality problems; cost performance indicators include the cost deviation rate and budget execution rate; and safety performance indicators include the safety hazard investigation rate and the safety accident occurrence rate. Each indicator is assigned a weight coefficient, and the comprehensive performance score is calculated using a weighted summation method.

5. The road engineering contract performance management and risk identification system according to claim 1, characterized in that, The risk identification model in the automatic risk identification module adopts a deep learning model that integrates LSTM and CNN. LSTM is used to extract the temporal features of performance data, and CNN is used to extract the spatial correlation features of risk features. The weights of key risk features are strengthened through an attention mechanism, and the model parameters are optimized by cross-validation during the model training process.

6. The road engineering contract performance management and risk identification system according to claim 1, characterized in that, The tiered push mechanism of the early warning push module includes four levels of early warning: blue warning (general risk), yellow warning (significant risk), orange warning (major risk), and red warning (extremely serious risk). Different levels of early warning correspond to different push channels and response time limits. Red warnings need to be pushed to project leaders and corporate management in real time and the emergency response process needs to be initiated.

7. A method for contract performance management and risk identification in road engineering projects, characterized in that, Includes the following steps: S1. Structured parsing of contract text: Natural language processing algorithms are used to segment, tag, and identify entities in road engineering contract texts, extract core contract elements, and generate a standardized contract element database. S2. Multi-source performance data collection: Real-time collection of progress, quality, safety and environmental data at the construction site through IoT devices, synchronous acquisition of system data such as project management and financial payment through business system interfaces, and supplementation of unstructured data by manual input to achieve real-time aggregation of multi-dimensional performance data; S3. Dynamic Management of Performance Status: Based on a standardized contract element database, a performance evaluation index system is constructed that includes four dimensions: schedule, quality, cost, and safety. The collected real-time performance data is compared with the evaluation index to calculate the performance compliance rate and generate a dynamic performance status assessment report. S4. Automatic identification of performance risks: A risk identification model is built based on machine learning algorithms. The model is trained and optimized using historical performance data and risk cases. Standardized contract element data and real-time performance data are input into the model. Through feature extraction and risk matching, potential risk types and risk levels are automatically identified. S5. Risk Warning and Response: Generate graded warning information based on the identified risk type and risk level, push it to the corresponding responsible entity, and at the same time associate and match the preset risk response strategy to achieve timely handling and closed-loop management of risks.

8. The method for performance management and risk identification of road engineering contracts according to claim 7, characterized in that, After performing structured parsing of the contract text in step S1, the core elements of the extracted contract are also verified. Errors in element extraction are corrected by comparing them with standard contract templates in the field of road engineering and by manual review, so as to ensure the accuracy of the standardized contract element database.

9. The method for performance management and risk identification of road engineering contracts according to claim 7, characterized in that, The training process of the risk identification model in step S4 includes: collecting historical data on the performance of road engineering contracts and typical risk cases, constructing a risk sample dataset, labeling the sample data with features, dividing the labeled sample data into a training set, a validation set and a test set, training the initial model using the training set, optimizing the model parameters using the validation set, verifying the model's recognition accuracy using the test set, and completing the model training when the accuracy reaches a preset threshold.

10. The method for performance management and risk identification of road engineering contracts according to claim 7, characterized in that, Step S5 also includes tracking the risk management process, recording risk management measures, progress and effects, forming a closed-loop risk management file, and feeding the risk management data back to the risk identification model to achieve continuous iterative optimization of the model.

Images