Online engineering bidding management system and method

Abstract

This invention provides an online engineering bidding and tendering management system, comprising: a data acquisition module configured to collect internal and external data related to bidding and tendering projects; the internal data including tender documents, bid documents, and historical project records; and the external data including publicly available financial data, legal litigation data, public opinion data, supply chain data, and IoT device data of bidding companies; and a blockchain module configured to store key node data in the bidding and tendering process, including the publication of tender announcements, submission of bid documents, bid opening records, bid evaluation results, and contract signing, on the blockchain in the form of transactions. This invention introduces a Dynamic Performance Capability Index (DPCI) to incorporate the historical performance, real-time status, and future potential of bidding companies into the evaluation system, making the bid evaluation results more scientific and objective, effectively screening out truly capable contractors, and significantly reducing performance risks such as project delays, cost overruns, and substandard quality from the source.

Description

Technical Field

[0001] This invention relates to the fields of information technology and engineering management technology, and in particular to an online engineering bidding management system and method. Background Technology

[0002] Engineering bidding is the starting point for engineering project construction. Its fairness, scientific nature and efficiency are directly related to the success or failure of the project and the investment benefits. The traditional offline bidding model has problems such as cumbersome process, long cycle, low transparency and easy to breed corruption. With the development of information technology, a large number of online engineering bidding management systems have emerged. These systems typically digitize processes such as the publication of bidding announcements, online bidding, electronic bid opening, and expert evaluation, improving efficiency to some extent. However, existing technologies still have the following significant drawbacks:

[0003] The evaluation dimensions are static and singular: existing systems mainly rely on the bid documents themselves, i.e., "one-time" quotations and technical solutions. Within a limited timeframe, evaluation experts find it difficult to comprehensively and thoroughly examine the true and dynamic performance capabilities of bidding companies. This leads to a "bad money drives out good" phenomenon, where companies with low bids but poor actual capabilities may win bids, posing significant performance risks later in the project.

[0004] Information silos and asymmetry: Bidding parties have difficulty obtaining comprehensive and real-time information on the operating status of bidding companies, such as the latest financial health, potential legal disputes, and supply chain stability. Bidding parties also have difficulty fully understanding the potential hidden risks of the project. This information asymmetry is the main reason for decision-making errors and project failures. Bid rigging and collusion are difficult to identify: Although digitalization has improved some transparency, the ability to identify bid rigging and collusion is still weak. Some people can coordinate bids in a more covert way, and the existing system lacks effective data analysis and behavioral pattern recognition tools to proactively detect these anomalies. Insufficient trust in the process: Although the bidding process is electronic, core data and operation records may still be tampered with by the administrator of the centralized platform. The lack of a record mechanism that is trusted by all participants and cannot be tampered with leads to frequent questioning and disputes. Lack of monitoring during the contract execution phase: Most systems complete their mission after determining the successful bidder, lacking the ability to continuously monitor and warn of dynamic risks during the project execution phase, and failing to form a closed-loop management of "pre-assessment - in-process monitoring - post-evaluation". To address this, an online engineering bidding management system and method are proposed. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide an online engineering bidding management system and method to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0006] The technical solution of this invention is implemented as follows: an online engineering bidding management system, comprising: The data acquisition module is configured to collect internal and external data related to bidding projects. The internal data includes bidding documents, tender documents, and historical project records. The external data includes publicly available financial data, judicial litigation data, public opinion data, supply chain data, and IoT device data of bidding companies. The blockchain module is configured to store key data from the bidding process, including the publication of bidding announcements, submission of bid documents, bid opening records, bid evaluation results, and contract signing, on the blockchain in the form of transactions, ensuring the integrity and immutability of the data. An artificial intelligence analysis engine, connected to the data acquisition module, is configured for: Based on the collected internal and external data, the dynamic performance capability index (DPCI) of each bidding company is calculated through a preset machine learning model. The DPCI is a quantitative score that is updated over time and reflects the current and future performance capability of the bidding company. Based on the project characteristics in the bidding documents, the project risk prediction submodule analyzes and outputs the potential risk level and risk points of the project. The intelligent evaluation and recommendation module, connected to the artificial intelligence analysis engine and the blockchain module, is configured for: Receive tender documents and extract tender prices and technical scores; Combining the bid price, technical score, and the DPCI of the corresponding bidding company calculated by the artificial intelligence analysis engine, a weighted algorithm is used to calculate the comprehensive value score of each bid. Bids are ranked according to their overall value score, and a recommended list of winning candidates is generated. The visualization monitoring module is connected to the data acquisition module, the blockchain module, the artificial intelligence analysis engine, and the intelligent evaluation and recommendation module. It is configured to provide a visualization interface for the entire project process to the bidding party and the regulatory party, and to display DPCI changes, project risk warnings, blockchain evidence storage status, and bid evaluation and recommendation results in real time.

[0007] In some embodiments, when the artificial intelligence analysis engine performs calculations through the Dynamic Performance Capability Index (DPCI) calculation submodule, the data dimensions it relies on include at least historical project performance records, corporate financial health status, legal risk index, supply chain stability, key technical personnel qualifications, and real-time equipment resource availability.

[0008] In some embodiments, the calculation model of the Dynamic Performance Capability Index (DPCI) is a weighted scoring model or a deep neural network model, wherein the weights of different data dimensions are dynamically adjusted according to the project type and industry characteristics.

[0009] In some embodiments, the artificial intelligence analysis engine further includes a bid-rigging and collusion identification submodule. This submodule is configured to analyze data such as bid prices, technical solution similarity, and historical bid correlation of all bidding companies, identify potential bid-rigging and collusion behaviors through anomaly detection algorithms, and issue warnings to regulators.

[0010] In some embodiments, the blockchain module adopts a consortium blockchain architecture, with nodes jointly maintained by the tendering party, bidders, regulatory agencies, financial institutions, and third-party auditing agencies to ensure access control and transparency in data sharing.

[0011] In some embodiments, the blockchain module is also deployed with smart contracts, which automatically trigger progress payments to the contractor when preset conditions are met, thereby improving the efficiency of capital flow and the rigidity of contract execution. Preset conditions include the successful acceptance of key project milestones.

[0012] An online engineering bidding management method includes the following steps: S1. Data Acquisition and Blockchain Initialization: The data acquisition module collects internal and external data of the bidding project and stores key information such as the bidding announcement on the blockchain through the blockchain module. S2. Construction of dynamic profile of bidders: During the bidding period, the artificial intelligence analysis engine continuously collects and analyzes multi-source data of each bidding company, and calculates and updates its dynamic performance capability index (DPCI) in real time. S3. Tender Receipt and Intelligent Analysis: Receives tender documents submitted by each bidder, records the submissions on the blockchain for evidence storage, and uses an artificial intelligence analysis engine to parse the tender documents and conduct risk assessment in conjunction with a project risk prediction model. S4. Comprehensive Value Assessment: The intelligent assessment and recommendation module obtains the bid price, technical score and real-time DPCI of each bidder, and calculates the comprehensive value score. S5. Generation of Recommendations and Public Announcement of Results: A list of recommended candidates for winning bids is generated based on the comprehensive value score, and the evaluation process and results are stored on the blockchain for public announcement. S6. Contract Performance and Intelligent Monitoring: After winning the bid, key contract terms will be written into the smart contract. During the project execution process, the system will continuously monitor the changes in the bidder's DPCI and project risks, and provide real-time early warnings to all parties through the visualization monitoring module.

[0013] In some embodiments, in S2, the calculation of the Dynamic Performance Capability Index (DPCI) specifically includes quantifying historical project performance records, financial data, judicial data, supply chain data, and IoT data, inputting them into a pre-trained machine learning model, and outputting a score between 0 and 100, with a higher score representing stronger performance capability.

[0014] In some embodiments, prior to S4, a bid-rigging identification step is also included. The artificial intelligence analysis engine uses cluster analysis and isolated forest algorithms to analyze the feature vectors of all bidding schemes. If highly similar bid clusters or abnormally deviating bidding behaviors are found, they are marked as suspected bid-rigging and submitted for manual review.

[0015] In some embodiments, in S6, the execution process of the smart contract is as follows: the project supervisor or the bidding party confirms on the blockchain that a certain milestone node has been completed, the smart contract automatically verifies the validity of the confirmation, and after the verification is successful, it automatically releases the corresponding funds from the escrow account to the winning bidder's account.

[0016] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions: I. This invention introduces the Dynamic Performance Capability Index (DPCI), which incorporates the bidding company's historical performance, real-time status, and future potential into the evaluation system. This makes the evaluation results more scientific and objective, effectively selecting truly capable contractors and significantly reducing performance risks such as project delays, cost overruns, and substandard quality from the source.

[0017] Second, this invention provides the bidding party with a 360-degree panoramic view of the risks of bidding companies and projects through multi-source data fusion, assisting them in making more informed decisions. The AI-driven risk warning and recommendation functions free experts from tedious information screening and allow them to focus on judging core values.

[0018] Third, this invention ensures that all key operations and data records cannot be unilaterally altered once they are recorded on the chain through the application of blockchain technology. All authorized participants can verify them, which greatly enhances the transparency and credibility of the bidding process and effectively curbs opaque operations and corruption.

[0019] Fourth, this invention, through its built-in sub-module for identifying bid rigging and collusion, can proactively discover suspicious bidding patterns through big data analysis, providing precise clues for regulatory authorities, which helps to combat illegal and irregular activities and maintain a fair and competitive market order.

[0020] Fifth, this invention not only covers the bidding stage but also extends management to the contract performance stage. It achieves automatic payment through smart contracts and continuously monitors DPCI and project risks, thus realizing closed-loop management of "prevention before the event, control during the event, and evaluation after the event" and improving the overall level of engineering project management.

[0021] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a system diagram of the present invention. Detailed Implementation

[0024] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0025] It is important to note that terms such as "first," "second," "symmetric," "array," "set in," and "set with" are used only to distinguish between descriptive and positional descriptions and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features specified with terms such as "first" or "symmetric" may explicitly or implicitly include one or more of that feature; similarly, when the quantity of certain features is not limited by words such as "two" or "three," it should be noted that such features also explicitly or implicitly include one or more features.

[0026] In this invention, unless otherwise explicitly specified and limited, terms such as "installation," "connection," and "fixation" should be interpreted broadly; for example, they can refer to a fixed connection, a detachable connection, or an integral molding; they can refer to a mechanical connection, a direct connection, a welding connection, or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the accompanying drawings and specific circumstances.

[0027] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0028] like Figure 1 As shown, this embodiment of the invention provides an online engineering bidding and tendering management system, including: The data acquisition module is configured to collect internal and external data related to bidding projects. Internal data includes bidding documents, tender documents, and historical project records, while external data includes publicly available financial data, legal litigation data, public opinion data, supply chain data, and IoT device data of bidding companies. The blockchain module is configured to store key data from the bidding process, including the publication of bidding announcements, submission of bid documents, bid opening records, bid evaluation results, and contract signing, on the blockchain in the form of transactions, ensuring the integrity and immutability of the data. An artificial intelligence analysis engine, connected to the data acquisition module, is configured for: Based on the collected internal and external data, the dynamic performance capability index of each bidding company is calculated through a pre-set machine learning model. DPCI is a quantitative score that is updated over time and reflects the current and future performance capability of the bidding company. Based on the project characteristics in the bidding documents, the project risk prediction submodule analyzes and outputs the potential risk level and risk points of the project. The intelligent evaluation and recommendation module, connected to the artificial intelligence analysis engine and blockchain module, is configured for: Receive tender documents and extract tender prices and technical scores; By combining the bid price, technical score, and the corresponding bidder's DPCI calculated by the artificial intelligence analysis engine, a weighted algorithm is used to calculate the comprehensive value score of each bid. Bids are ranked according to their overall value score, and a recommended list of winning candidates is generated. The visualization monitoring module is connected to the data acquisition module, blockchain module, artificial intelligence analysis engine, and intelligent evaluation and recommendation module. It is configured to provide a visualization interface for the entire project process to the bidding party and regulators, displaying DPCI changes, project risk warnings, blockchain evidence storage status, and bid evaluation and recommendation results in real time.

[0029] In this embodiment, specifically, when the artificial intelligence analysis engine performs calculations through the Dynamic Performance Capability Index (DPCI) calculation submodule, the data dimensions it relies on include at least historical project performance records, corporate financial health status, legal risk index, supply chain stability, key technical personnel qualifications, and real-time equipment resource availability.

[0030] In this embodiment, the Dynamic Performance Capability Index (DPCI) is specifically calculated using a weighted scoring model or a deep neural network model, where the weights of different data dimensions are dynamically adjusted according to the project type and industry characteristics.

[0031] In this embodiment, the artificial intelligence analysis engine specifically includes a bid-rigging and collusion identification submodule. This submodule is configured to analyze data such as the bid prices, technical solution similarity, and historical bid correlation of all bidding companies, identify potential bid-rigging and collusion behaviors through anomaly detection algorithms, and issue warnings to regulators.

[0032] In this embodiment, the blockchain module adopts a consortium blockchain architecture, and the nodes are jointly maintained by the bidding party, the tendering party, the regulatory agency, the financial institution and the third-party auditing agency to ensure access control and transparency of data sharing.

[0033] In this embodiment, the blockchain module is also equipped with smart contracts. When the smart contracts meet preset conditions, they automatically trigger the payment of progress payments to the contractor, thereby improving the efficiency of capital flow and the rigidity of contract execution. The preset conditions include the successful acceptance of key project nodes.

[0034] In this embodiment, specifically, The data acquisition module is the system's data entry point, and it collects data from different sources through various means such as API interfaces, web crawlers, and data exchange platforms. Internal data source: The system's own database, which stores historical bidding project information, a database of bidding companies, an expert database, historical bid evaluation records, etc. External data source: Business / judicial data APIs: such as Tianyancha and Qichacha, to obtain enterprise registration information, shareholder structure, legal proceedings, and records of dishonesty; Financial Data API: Obtain financial reports of listed companies, corporate credit ratings, etc. Public opinion monitoring system: Captures news reports and social media comments about bidding companies online and conducts sentiment analysis; Supply chain platform: Connect with major building material suppliers or logistics platforms to understand the stability of their supply chains; IoT Platform: For bidding companies with large equipment, they can access the IoT data of their equipment management system to obtain the real-time location, working status, and availability of key equipment.

[0035] In this embodiment, specifically, The blockchain module adopts a consortium blockchain architecture, built using Hyperledger Fabric or FISCO BCOS. The nodes are composed of the bidding unit (owner), several core bidding units, industry regulatory agencies, banks (funds custodians), notary offices, etc., and each node stores a complete copy of the ledger. Data on the blockchain: When key events such as the release of bidding announcements, submission of bid documents by bidders (the hash value of the documents is on the blockchain), bid opening, determination of bid evaluation results, and signing of contracts occur, the system will package the core information of the event (such as timestamps, operators, and key data summaries) into a "transaction" and broadcast it to all nodes for consensus verification. After verification, it will be permanently recorded on the blockchain. Smart contracts: During the contract signing phase, payment terms in the contract (such as "20% payment upon completion of foundation pit excavation, 30% payment upon completion of main structure capping") are written into smart contract code and deployed on the blockchain. Project funds are first deposited into an escrow account controlled by the smart contract.

[0036] In this embodiment, specifically, The artificial intelligence analysis engine consists of a Dynamic Performance Capability Index (DPCI) calculation submodule, a project risk prediction submodule, and a bid rigging and collusion identification submodule; Dynamic Performance Capability Index (DPCI) Calculation Submodule: Data preprocessing: The multi-source heterogeneous data obtained from the data acquisition module will be cleaned, standardized and vectorized. For example, "no major safety accidents in the past three years" will be quantified as +5 points and "flow ratio is less than 1" will be quantified as -10 points. Model building: Employ weighted scoring models or more complex machine learning / deep learning models such as Gradient Boosting Decision Tree (GBDT) and Long Short-Term Memory Network (LSTM); Calculation example (weighted model): DPCI = w1 (Historical Performance Score) + w2 (Financial Health Score) + w3 (Legal Risk Score) + w4 (Supply Chain Score) + w5 (Real-time Resource Score) Among them, the weights w1, w2, w3 and w4 are dynamically adjusted according to the project type (such as building construction, bridges, software). For technology-intensive projects, the weight of w5 (real-time resource score, such as the number of senior engineers) will be higher. Dynamic updates: The model is not calculated once, but is recalculated daily or weekly based on the inflow of external data (such as new litigation news, financial statement releases) to ensure the timeliness of DPCI; The project risk prediction submodule uses natural language processing (NLP) technology to parse the tender documents and extract features such as project scale, technical complexity, schedule requirements, and geographical location. These features are then input into a risk prediction model trained based on historical project data to output the main risks that the project may face (such as "Delay risk: High" and "Cost overrun risk: Medium"). The bid rigging and collusion identification submodule extracts the feature vectors of all bidding proposals (such as price breakdowns, technical solution keywords, project management team backgrounds, etc.) and uses the K-Means clustering algorithm. If the feature vectors of a few bidding proposals are found to be highly clustered, while other proposals are far away from the cluster, then the cluster may be a bid rigging group. At the same time, the isolated forest algorithm is used to detect "dummy bidding" behavior with abnormal pricing. In this embodiment, specifically, The process of parsing the tender document using Natural Language Processing (NLP) technology is shown below. Step 1: Text Preprocessing This is the foundation of all NLP tasks, the purpose of which is to clean up raw, messy text into a format that is easier for computers to process; Text cleaning and formatting: Remove irrelevant information such as headers, footers, table of contents, and page numbers; Handle special characters and garbled text, and unify the encoding format (such as UTF-8); Convert files of different formats, such as PDF and DOCX, into plain text. Clause segmentation and word segmentation: Sentence segmentation: Divide the entire document into independent sentences; For example, the sentence "The total investment of the project is approximately 500 million yuan. The construction period is required to be 730 calendar days" can be split into two sentences.

[0037] Word segmentation: This involves breaking down a sentence into its smallest semantic units—words. For example, "the total investment of the project is about 500 million yuan" can be segmented into ["project", "total", "investment", "about", "5", "100 million yuan"]. This is a crucial step in Chinese. Part-of-speech tagging: Each word is labeled with its part of speech, such as noun (n), verb (v), numeral (m), quantifier (q), etc. For example, ["project(n)", "total(d)", "investment(v)", "about(d)", "5(m)", "hundred million yuan(q)"]. This helps to extract information more accurately in the future. Named entity recognition: Identifying entities with specific meanings in text, such as names of people, places, organizations, times, and dates, is a key technology for extracting geographic locations. Example: In the sentence "The project construction site is located in Zhongguancun Software Park, Haidian District, Beijing", the NER model can identify Zhongguancun Software Park, Haidian District, Beijing as a "Location" entity; Step 2: Feature Extraction and Quantization After preprocessing, the system will use different strategies to accurately extract different target features; Extracting project scale: Keyword matching: The system will predefine a keyword dictionary related to scale, such as ["investment amount", "total investment", "contract amount", "construction cost", "building area", "total length", "processing capacity"]; Regular expressions: Combine keywords with regular expressions to match and extract numbers and units that immediately follow them; Example: For the text "The total investment of the project is approximately 500 million RMB", the regular expression r"total investment.*?(\d{1,3}(,\d{3})*\s*million RMB)" can accurately extract 500 million RMB. For the text "Total building area is approximately 125,000 square meters", the regular expression r"building area.*?(\d+\.?\d*)\s*ten thousand?square meters" can extract 125,000 square meters; Quantification: The extracted text (such as "50,000,000 yuan") is converted into a pure numerical value of 500,000,000 (unit: yuan) to facilitate model calculation; Extraction time requirements: Keyword matching: Keywords include ["construction period", "calendar days", "construction cycle", "start date", "completion date", "delivery date"]; Regular expressions and date parsing: Directly extract the project duration: r"Project duration is (\d+)\s*calendar days", which extracts the number 730; Indirect calculation of construction period: If the document only provides the start and completion dates, such as "planned to start on January 1, 2026 and be completed on January 1, 2028", the system needs to identify these two date entities and calculate the difference in days between them; Quantification: The final result is a value in "days"; Extracting geographic location: Relying on the NER model: As described in step one, use the trained NER model to directly identify all "location" entities; Geocoding: The extracted place names (such as "Zhongguancun Software Park, Haidian District, Beijing") are converted into precise latitude and longitude coordinates (such as [116.3039, 39.9834]) by calling geocoding service APIs (such as Gaode Maps, Baidu Maps API). Latitude and longitude are standardized numerical features, which are very convenient for model analysis. Assess the technical complexity: This is the most challenging aspect because it involves semantic understanding, not just simple numerical extraction. Technical keyword dictionary method: Build a dictionary containing highly complex technical terms, such as ["BIM", "prefabricated", "deep foundation pit", "long span", "irregular structure", "artificial intelligence", "Internet of Things", "blockchain", "smart construction site"]. By statistically analyzing the frequency and weight of these keywords in the bidding documents, a preliminary complexity score can be given. Text classification / embedding models (more advanced): Use pre-trained language models (such as BERT, RoBERTa) to transform the entire tender document or its technical requirements section into a high-dimensional vector (text embedding). This vector can be viewed as a "semantic fingerprint" of the document's technical content; A classifier can be trained using a small amount of labeled (high, medium, and low complexity) historical project data to automatically classify the complexity of new tender documents; Step 3: Structured Output Finally, all extracted and quantified features are integrated into a structured data object, such as JSON format, for direct use by the "Project Risk Prediction Model".

[0038] The JSON format is shown below. { "project_id": "ZH-2025-001", "scale": { "total_investment": 500000000, "investment_unit": "CNY" "building_area": ​​125000 }, "timeline": { "duration_days": 730, "start_date": "2026-01-01", "end_date": "2028-01-01" }, "location": { "address": "Zhongguancun Software Park, Haidian District, Beijing" "coordinates": [116.3039, 39.9834] }, "technical_complexity": { "score": 0.85, "level": "High", "keywords_found": ["BIM", "prefabricated construction", "smart construction site"] }} In this embodiment, specifically, The detailed process of the K-Means clustering algorithm in identifying bid rigging and collusion is shown below. Step 1: Feature Engineering—Vectorizing the Bidding Proposal This is the most crucial step; each bidding proposal must be converted into a mathematical vector so that the algorithm can process it. Quotation feature vectorization: Don't just look at the total price. Bidding rigged groups usually differentiate between different total prices, but maintain a high degree of consistency in the proportion of each internal bid. Method: Extract the itemized price list (such as earthwork, concrete, steel reinforcement, labor costs, etc.) from all bid documents. Assuming there are N items, the price of each bid can be represented as an N-dimensional vector V_price=[price_1, price_2, ..., price_N]. Normalization: In order to eliminate the influence of the total price, the vector is usually normalized. For example, the price of each item is divided by the total price to obtain a cost composition ratio vector V_ratio=[ratio_1, ratio_2, ..., ratio_N]. Bids from the same bidder will have very similar V_ratios. Feature vectorization of technical solutions: The technical solution is text, which needs to be converted into vectors; Method 1: TF-IDF: TF-IDF is a statistical technique that uses the term frequency to inverse document frequency of keywords to generate a sparse vector. The vector dimension is the size of the dictionary, and each value represents the importance of the word in the document. Method 2: Text embedding: Using models such as BERT, the entire technical solution text is transformed into a dense, low-dimensional semantic vector (e.g., 768-dimensional). This vector can better capture semantic similarity, that is, different words but similar meanings. Combined feature vectors: The price quote vector and the technical solution vector are concatenated to form a comprehensive feature vector V_combined=[V_ratio, V_tech]; Before splicing, it may be necessary to scale the different sub-vectors to prevent the numerical range of a certain part from being too large and dominating the clustering results; Step 2: Apply the K-Means algorithm Determine the K value (number of clusters): K-Means requires pre-specifying how many clusters to divide into. In the scenario of identifying bid rigging, we do not know how many groups there are. Methods: The "elbow rule" or "profile coefficient" can be used to help determine a relatively reasonable K value, or a large K value can be set, and then focus on those clusters with few members but very compact internal structures; Algorithm execution flow: Initialization: Randomly select K points in the data space as the initial cluster centers (centroids); Assignment: Calculate the distance (usually Euclidean distance) from each bid vector V_combined to the K centroids, and assign it to the cluster containing the nearest centroid; Update: Recalculate the average of all vectors in each cluster and use this average as the new centroid; Iteration: Repeat the "assign" and "update" steps until the position of the centroid no longer changes significantly, or the maximum number of iterations is reached; Step 3: Results Analysis and Early Warning Identify suspicious clusters: After the algorithm converged, we obtained K clusters; Suspicious pattern: The patent explicitly states that "the feature vectors of a few bidding schemes are highly clustered, while the feature vectors of other schemes are far from the cluster"; Quantitative indicators: Calculate the sum of squares (WCSS) or average distance within each cluster. A cluster with few members (e.g., 2-3) but extremely low WCSS is a highly suspicious bid-rigging group. Generate an alert: The system automatically marks these suspicious clusters and generates early warning reports; The report may include: members within the cluster (list of bidding companies), their similarity scores, and their differences from other bidding proposals; Final decision: This warning will be submitted to the regulator for manual review, and a human will be responsible for determining whether it constitutes bid rigging or collusion.

[0039] In this embodiment, specifically, The detailed process of the Isolation Forest algorithm in identifying abnormal bidding behavior is shown below. Step 1: Understanding the core concept The principles of Isolation Forest are completely different from K-Means. It does not focus on the "group" but on the degree of abnormality of the "individual". Basic assumption: Outliers are “few and distinct”, and therefore they are easier to “isolate” than normal points; Implementation method: By constructing a series of random "decision trees" (iTrees), data points are split in these trees. If a point is quickly split to a leaf node (i.e., the path is very short), it is considered abnormal. Step 2: Algorithm Execution Flow Building a forest: Sampling: Randomly select a portion of the sample from all the bidding data (without repeating the sampling); Build an iTree: In the current sample data, randomly select a feature dimension (e.g., "total quote"); Randomly select a split point between the maximum and minimum values ​​of this feature; Based on this dividing point, the data is divided into two parts; The above process is recursively repeated for the segmented sub-data until there is only one sample point in the sub-data (successfully isolated); Or all sample points have the same feature values; Or the tree has reached the preset maximum height; Repeat: Repeat the above steps of "sampling" and "building iTree" multiple times (e.g., 100 times) to build an "isolated forest"; Calculate the outlier score: For each bidding proposal (data point), let it "pass through" every iTree in the forest; Record the path length of the point on each tree (the number of edges traversed from the root node to the leaf node); Calculate the average path length E(h(x)) of this point across all trees; Core logic: Outliers are different from other points and can usually be separated after several random partitions, so their average path length E(h(x)) will be very short. The algorithm calculates an anomaly score based on the average path length and the total number of samples. The score is usually between 0 and 1. The closer the score is to 1, the more likely it is to be an anomaly. Step 3: Results Analysis and Early Warning Set threshold: Set a threshold for outlier scores, for example, 0.7; Identifying Abnormal Bids: Iterate through all bid proposals and calculate their anomaly scores; Any bid with a score exceeding 0.7 will be marked as "abnormal"; Example scenario: Among 10 bids, 9 bids are concentrated between 100 million and 110 million, while one bid is 150 million. This 150 million bid is very easy to isolate in terms of the "bid" feature. Its path length will be very short and its anomaly score will be very high, thus being identified as a "dummy bid". Generate an alert: The system will add bids with high abnormal scores to the warning list and point out the abnormalities (such as "the bid is much higher than the average level"). Similarly, this warning will be submitted for manual review as another clue to detect bid-rigging and collusion.

[0040] In this embodiment, specifically, The intelligent evaluation and recommendation module starts working after the bid opening. It receives the price (P) of each bid and the technical score (T) given by the evaluation experts. Obtain the latest DPCI from each bidder using an AI analytics engine; Calculate the Comprehensive Value Score (CVS): CVS=a*(1 / P_normalized)+b*T_normalized+c*DPCI_normalized Among them, P_normalized, T_normalized, and DPCI_normalized are normalized values, while a, b, and c are weights set by the bidding party, reflecting its emphasis on price, technology, and performance capabilities. Based on CVS ranking from highest to lowest, generate a recommendation list and provide a detailed explanation of the ranking criteria; In this embodiment, specifically, the visual monitoring module is a web-based dashboard; The bidding party's view shows the project progress, the historical curves and comparisons of each bidder's DPCI, the project risk radar chart, the evaluation recommendation list, the blockchain evidence storage status, etc. Regulatory perspective: It provides a macro-level view of all regulated projects, allows for focused review of system-issued warnings of bid rigging and collusion, and enables tracing of the complete on-chain records of any project; Bidder View: You can see your company's DPCI score and composition, and understand your own strengths and weaknesses.

[0041] An online engineering bidding management method includes the following steps: S1. Data Acquisition and Blockchain Initialization: The data acquisition module collects internal and external data of the bidding project and stores key information such as the bidding announcement on the blockchain through the blockchain module. S2. Construction of dynamic profile of bidders: During the bidding period, the artificial intelligence analysis engine continuously collects and analyzes multi-source data of each bidding company, and calculates and updates its dynamic performance capability index (DPCI) in real time. S3. Tender Receipt and Intelligent Analysis: Receives tender documents submitted by each bidder, records the submissions on the blockchain for evidence storage, and uses an artificial intelligence analysis engine to parse the tender documents and conduct risk assessment in conjunction with a project risk prediction model. S4. Comprehensive Value Assessment: The intelligent assessment and recommendation module obtains the bid price, technical score and real-time DPCI of each bidder, and calculates the comprehensive value score. S5. Generation of Recommendations and Public Announcement of Results: A list of recommended candidates for winning bids is generated based on the comprehensive value score, and the evaluation process and results are stored on the blockchain for public announcement. S6. Contract Performance and Intelligent Monitoring: After winning the bid, key contract terms will be written into the smart contract. During the project execution process, the system will continuously monitor the changes in the bidder's DPCI and project risks, and provide real-time early warnings to all parties through the visualization monitoring module.

[0042] In this embodiment, specifically in S2, the calculation of the Dynamic Performance Capability Index (DPCI) includes quantifying historical project performance records, financial data, judicial data, supply chain data, and IoT data, inputting them into a pre-trained machine learning model, and outputting a score between 0 and 100, with a higher score representing stronger performance capability.

[0043] In this embodiment, specifically before S4, there is a bid-rigging and collusion identification step. The artificial intelligence analysis engine uses cluster analysis and isolated forest algorithm to analyze the feature vectors of all bidding schemes. If highly similar bidding clusters or abnormally deviating bidding behaviors are found, they are marked as suspected bid-rigging and collusion and submitted for manual review.

[0044] In this embodiment, specifically in S6, the execution process of the smart contract is as follows: the project supervisor or the bidding party confirms on the blockchain that a certain milestone node has been completed, the smart contract automatically verifies the validity of the confirmation, and after the verification is successful, it automatically releases the corresponding funds from the escrow account to the winning bidder's account.

[0045] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in the present invention, and these should all be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An online engineering bidding management system, characterized in that, include: The data acquisition module is configured to collect internal and external data related to bidding projects. The internal data includes bidding documents, tender documents, and historical project records. The external data includes publicly available financial data, judicial litigation data, public opinion data, supply chain data, and IoT device data of bidding companies. The blockchain module is configured to store key data from the bidding process, including the publication of bidding announcements, submission of bid documents, bid opening records, bid evaluation results, and contract signing, on the blockchain in the form of transactions, ensuring the integrity and immutability of the data. An artificial intelligence analysis engine, connected to the data acquisition module, is configured for: Based on the collected internal and external data, the dynamic performance capability index (DPCI) of each bidding company is calculated through a preset machine learning model. The DPCI is a quantitative score that is updated over time and reflects the current and future performance capability of the bidding company. Based on the project characteristics in the bidding documents, the project risk prediction submodule analyzes and outputs the potential risk level and risk points of the project. The intelligent evaluation and recommendation module, connected to the artificial intelligence analysis engine and the blockchain module, is configured for: Receive tender documents and extract tender prices and technical scores; Combining the bid price, technical score, and the DPCI of the corresponding bidding company calculated by the artificial intelligence analysis engine, a weighted algorithm is used to calculate the comprehensive value score of each bid. Bids are ranked according to their overall value score, and a recommended list of winning candidates is generated. The visualization monitoring module is connected to the data acquisition module, the blockchain module, the artificial intelligence analysis engine, and the intelligent evaluation and recommendation module. It is configured to provide a visualization interface for the entire project process to the bidding party and the regulatory party, and to display DPCI changes, project risk warnings, blockchain evidence storage status, and bid evaluation and recommendation results in real time.

2. The online engineering bidding management system according to claim 1, characterized in that: When the AI ​​analysis engine performs calculations through the Dynamic Performance Capability Index (DPCI) calculation submodule, the data dimensions it relies on include at least historical project performance records, corporate financial health status, legal risk index, supply chain stability, key technical personnel qualifications, and real-time equipment resource availability.

3. The online engineering bidding management system according to claim 2, characterized in that: The Dynamic Performance Capability Index (DPCI) is calculated using a weighted scoring model or a deep neural network model, where the weights of different data dimensions are dynamically adjusted based on project type and industry characteristics.

4. The online engineering bidding management system according to claim 1, characterized in that: The artificial intelligence analysis engine also includes a bid-rigging and collusion identification submodule. This submodule is configured to analyze data such as the bid prices, technical solution similarity, and historical bid correlation of all bidding companies, identify potential bid-rigging and collusion behaviors through anomaly detection algorithms, and issue warnings to regulators.

5. The online engineering bidding management system according to claim 1, characterized in that: The blockchain module adopts a consortium blockchain architecture, with nodes jointly maintained by the tendering party, bidders, regulatory agencies, financial institutions, and third-party auditing institutions to ensure access control and transparency in data sharing.

6. The online engineering bidding management system according to claim 1, characterized in that: The blockchain module is also equipped with smart contracts. When preset conditions are met, the smart contracts automatically trigger progress payments to the contractor, thereby improving the efficiency of capital flow and the rigidity of contract execution. Preset conditions include the successful acceptance of key project milestones.

7. An online engineering bidding management method, characterized in that, Includes the following steps: S1. Data Acquisition and Blockchain Initialization: The data acquisition module collects internal and external data of the bidding project and stores key information such as the bidding announcement on the blockchain through the blockchain module. S2. Construction of dynamic profile of bidders: During the bidding period, the artificial intelligence analysis engine continuously collects and analyzes multi-source data of each bidding company, and calculates and updates its dynamic performance capability index (DPCI) in real time. S3. Tender Receipt and Intelligent Analysis: Receives tender documents submitted by each bidder, records the submissions on the blockchain for evidence storage, and uses an artificial intelligence analysis engine to parse the tender documents and conduct risk assessment in conjunction with a project risk prediction model. S4. Comprehensive Value Assessment: The intelligent assessment and recommendation module obtains the bid price, technical score and real-time DPCI of each bidder, and calculates the comprehensive value score. S5. Generation of Recommendations and Public Announcement of Results: A list of recommended candidates for winning bids is generated based on the comprehensive value score, and the evaluation process and results are stored on the blockchain for public announcement. S6. Contract Performance and Intelligent Monitoring: After winning the bid, key contract terms will be written into the smart contract. During the project execution process, the system will continuously monitor the changes in the bidder's DPCI and project risks, and provide real-time early warnings to all parties through the visualization monitoring module.

8. The online engineering bidding management method according to claim 7, characterized in that: In S2, the calculation of the Dynamic Performance Capability Index (DPCI) specifically includes quantifying historical project performance records, financial data, judicial data, supply chain data, and IoT data, inputting them into a pre-trained machine learning model, and outputting a score between 0 and 100, with a higher score representing stronger performance capability.

9. The online engineering bidding management method according to claim 7, characterized in that: Prior to S4, there is also a bid-rigging and collusion identification step. The artificial intelligence analysis engine uses cluster analysis and isolated forest algorithms to analyze the feature vectors of all bidding schemes. If highly similar bid clusters or abnormally deviating bidding behaviors are found, they are marked as suspected bid-rigging and collusion and submitted for manual review.

10. The online engineering bidding management method according to claim 7, characterized in that: In S6, the execution process of the smart contract is as follows: the project supervisor or the bidding party confirms on the blockchain that a certain milestone node has been completed, the smart contract automatically verifies the validity of the confirmation, and after the verification is successful, it automatically releases the corresponding funds from the escrow account to the winning bidder's account.

Images