Project site cost risk management method and system

Abstract

The invention discloses a project site cost risk management method and system, and the method comprises the steps: obtaining a frame mode layer based on the site cost management scene historical data and site cost management business process historical data of a project object sample; processing the framework mode layer and an on-site cost standard database obtained through a selected on-site cost management file and a historical engineering settlement review manuscript to obtain a data layer; determining a field cost management knowledge graph according to the framework mode layer and the data layer; determining a risk management list of the target project object according to a matching result of the target project object and the field cost management knowledge graph; and screening the field cost file data of the target project object according to the field cost standard database, and generating a risk management scheme corresponding to a screening result based on the field cost management knowledge graph. According to the method, online indexing of project site cost knowledge, automatic identification and positioning of site cost risks and automatic generation of a risk management scheme are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of project management, and in particular, to a project site cost risk management method and system. Background Art

[0002] Project site cost management is an important link in the whole-process cost control. There are many design changes, on-site visas and concealed works in the project site cost process, and the data and materials are complicated, which are all managed manually offline, resulting in great difficulty in process control; and project site cost management is often post-control, and only exposed in the sub-settlement and final settlement stages; there is also a large degree of uncertainty in project site cost management, which seriously affects the efficiency of sub-settlement and final settlement. The existing cost management methods only focus on the engineering design stage and the settlement stage, involve less management of the project site cost stage, and cannot realize automatic identification and early warning of project cost risks, and cannot provide scientific decision-making for on-site cost management.

[0003] Therefore, how to effectively manage the project site cost and automatically identify risks has become an urgent technical problem for those skilled in the art to solve. Summary of the Invention

[0004] The present invention provides a project site cost risk management method and system to solve the technical problem of how to effectively manage the project site cost and automatically identify risks, and achieve the effects of intelligent management of project site cost, automatic identification of on-site cost risks, and automatic generation of risk early warning and risk management plans.

[0005] In a first aspect, the present invention provides a project site cost risk management method, which is applied to a project object composed of at least multiple data subjects, and the method includes:

[0006] Based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, determine the on-site cost data of the project object samples, and use a knowledge-driven algorithm to process the on-site cost data to obtain a framework pattern layer reflecting the relationships between the data subjects;

[0007] Perform character recognition on the selected on-site cost management documents and historical engineering settlement review drafts, construct an on-site cost standard database according to the character recognition results, and use a data-driven algorithm to process the framework pattern layer and the on-site cost standard database to obtain a data layer reflecting the associated data between the data subjects;

[0008] Determine an on-site cost management knowledge graph according to the framework pattern layer and the data layer;

[0009] In the process of project site cost risk management, the inference engine algorithm is used to match the target project object with the on-site cost management knowledge graph, and the risk management list of the target project object is determined according to the matching result;

[0010] The on-site cost file data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph.

[0011] Preferably, the knowledge-driven algorithm is used to process the on-site cost data to obtain a framework pattern layer reflecting the relationships between the data subjects, including:

[0012] Extract the first text content in the on-site cost data;

[0013] Perform content recognition on each of the first text contents to obtain a recognition result, where the recognition result includes each data subject and the association relationship between each data subject;

[0014] Define the attribute types of each data subject to obtain the attributes of each data subject;

[0015] Perform structured processing on each data subject, the association relationship between each data subject, and the attributes of each data subject to obtain the framework pattern layer.

[0016] Preferably, perform text recognition on the selected on-site cost management documents and historical project settlement review drafts, and construct an on-site cost standard database according to the text recognition results, including:

[0017] Extract the third text content of the selected on-site cost management documents, and perform content recognition on each of the third text contents to obtain a second recognition result;

[0018] Construct a cost management document database and a cost management rule database according to the second recognition result;

[0019] Extract the fourth text content of the selected historical project settlement review drafts, and perform content recognition on each of the fourth text contents to obtain a third recognition result;

[0020] Construct a cost problem database and a cost impact factor database according to the third recognition result;

[0021] Extract keywords from the text of the cost management document database and the cost management rule database, and perform recognition of the data subject and attribute value on the extracted keywords;

[0022] Structurally represent the recognized data subjects and attribute values to obtain structured features;

[0023] Input the structured features into a random forest model to obtain a quantitative determination condition;

[0024] Based on the historical data of the on-site cost management business process, divide the on-site cost management process into nodes to obtain a number of business nodes;

[0025] Determine the business node to which the quantitative determination condition belongs, use the quantitative determination condition as the warning rule for the node, and generate a warning rule database based on each warning rule.

[0026] Preferably, using a data-driven algorithm to process the pattern layer and the on-site cost standard database to obtain a data layer that reflects the associated data between each data subject, including:

[0027] Based on the pattern layer, set data subject extraction rules and data subject relationship extraction rules;

[0028] Extract the second text content of the selected on-site cost management documents and historical project settlement review drafts, and extract each data subject and the corresponding attribute value from the second text content according to the data subject extraction rules;

[0029] According to the data subject relationship extraction rules, extract the associated data between each data subject from the second text content;

[0030] In a manner corresponding one by one to the data subjects in the pattern layer, add the attribute values and the associated data to the corresponding data subjects to obtain a data layer.

[0031] Preferably, screening the on-site cost document data of the target project object according to the on-site cost standard database, and generating a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph, including:

[0032] Extract the fourth text content of the on-site cost document of the target project object to determine the business node where the target project object is located;

[0033] Based on the business node, screen the warning rules of the business node from the warning rule database;

[0034] Based on the screened warning rules, conduct risk control and risk warning on the target project object, and trace the risk warning according to the on-site cost management knowledge graph to generate a corresponding risk management plan.

[0035] In a second aspect, the present invention further provides a project site cost risk management system for implementing the project site cost risk management method described above. The system includes: a framework mode layer construction unit, a data layer construction unit, a site cost management knowledge graph determination unit, a risk management list generation unit, and a risk management plan generation unit;

[0036] The framework mode layer construction unit is used to determine the on-site cost data of the project object samples based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, and process the on-site cost data using a knowledge-driven algorithm to obtain a framework mode layer reflecting the relationships between the data subjects;

[0037] The data layer construction unit is used to perform character recognition on the selected on-site cost management documents and historical engineering settlement review drafts, construct an on-site cost standard database based on the character recognition results, and process the framework mode layer and the on-site cost standard database using a data-driven algorithm to obtain a data layer reflecting the associated data between the data subjects;

[0038] The on-site cost management knowledge graph determination unit is used to determine the on-site cost management knowledge graph according to the framework mode layer and the data layer;

[0039] The risk management list generation unit is used to match the target project object with the on-site cost management knowledge graph using an inference engine algorithm during the project site cost risk management process, and determine the risk management list of the target project object according to the matching result;

[0040] The risk management plan generation unit is used to screen the on-site cost document data of the target project object according to the on-site cost standard database, and generate a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph.

[0041] Preferably, the process of using a knowledge-driven algorithm to process the on-site cost data to obtain a framework mode layer reflecting the relationships between the data subjects includes:

[0042] Extract the first text content in the on-site cost data;

[0043] Perform content recognition on each of the first text contents to obtain a recognition result, where the recognition result includes each data subject and the association relationship between each data subject;

[0044] Define the attribute types of each data subject to obtain the attributes of each data subject;

[0045] Structurally process each of the said data subjects, the association relationships between each of the said data subjects, and the attributes of each of the said data subjects to obtain a framework pattern layer. Preferably, the method for performing character recognition on the selected on-site cost management documents and historical project settlement review drafts, and constructing an on-site cost standard database according to the character recognition results includes:

[0046] Extract the third text content of the selected on-site cost management documents, and perform content recognition on each of the said third text contents to obtain a second recognition result;

[0047] Construct a cost management document database and a cost management rule database according to the second recognition result;

[0048] Extract the fourth text content of the selected historical project settlement review drafts, and perform content recognition on each of the said fourth text contents to obtain a third recognition result;

[0049] Construct a cost problem database and a cost influencing factor database according to the third recognition result;

[0050] Extract keywords from the texts of the cost management document database and the cost management rule database, and perform recognition of the said data subjects and attribute values on the extracted keywords;

[0051] Structurally represent the recognized data subjects and attribute values to obtain structural features;

[0052] Input the structural features into a random forest model to obtain quantitative determination conditions;

[0053] Based on the historical data of the on-site cost management business process, divide the on-site cost management process into nodes to obtain a number of business nodes;

[0054] Determine the business node to which the quantitative determination condition belongs, use the quantitative determination condition as the warning rule for the node, and generate a warning rule database based on each of the warning rules.

[0055] Preferably, the method for processing the pattern layer and the on-site cost standard database using a data-driven algorithm to obtain a data layer reflecting the association data between each of the said data subjects includes:

[0056] Based on the pattern layer, set data subject extraction rules and data subject relationship extraction rules;

[0057] Extract the second text content of the selected on-site cost management documents and historical project settlement review drafts, and extract each of the said data subjects and the attribute values corresponding to the attributes from the second text content according to the data subject extraction rules;

[0058] Extract the associated data among the data subjects from the second text content according to the rules for extracting relationships among data subjects.

[0059] Add the attribute values and the associated data to the corresponding data subjects in a one-to-one correspondence with the data subjects in the schema layer to obtain the data layer.

[0060] Preferably, screening the on-site cost file data of the target project object according to the on-site cost standard database, and generating a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph, including:

[0061] Extract the fourth text content of the on-site cost file of the target project object to determine the business node where the target project object is located.

[0062] Based on the business node, screen the warning rules of the business node from the warning rule database.

[0063] Based on the screened warning rules, perform risk control and risk warning on the target project object, and trace the risk warning according to the on-site cost management knowledge graph to generate a corresponding risk management plan.

[0064] The present invention provides a method and system for risk management of on-site project costs. Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0065] (1) Construct an on-site cost management knowledge graph that integrates on-site cost data subjects and historical cost problems, realizes online indexing of on-site project cost knowledge, automatic identification and positioning of on-site cost risks, traceability analysis of management documents related to risk problems to automatically generate a risk management plan, and guide on-site cost personnel to perform timely, scientific and targeted control.

[0066] (2) Improve the lag of on-site project cost management and improve the efficiency and intelligent level of on-site project cost management. Description of the Drawings

[0067] Figure 1 is a schematic diagram of the steps of a method for risk management of on-site project costs provided by a preferred embodiment of the present invention;

[0068] Figure 2 is a schematic diagram of the structure of a system for risk management of on-site project costs provided by a preferred embodiment of the present invention. Detailed Embodiments

[0069] The embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings. The given embodiments are only for illustrative purposes and should not be construed as limiting the present invention. The included drawings are for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention. In the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0070] In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, the terms "installed", "connected", "coupled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are only for illustrative purposes and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0071] In the description of the present invention, it should be noted that unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0072] Please refer to Figure 1 , in an embodiment of the present invention, a method for project site cost risk management is provided, and the method includes:

[0073] S1. Based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, determine the on-site cost data of the project object samples, and use knowledge-driven algorithms to process the on-site cost data to obtain a framework model layer reflecting the relationships between various data subjects; among them, on-site cost risk management for project sites runs through the entire process from project start to completion and commissioning, and plays a crucial role in the entire process cost management work chain. Taking all types of project objects as samples and analyzing each project object sample, it is found that the on-site cost management scenarios for transmission and transformation projects at least include: design change management, on-site visa management, payment of migrant workers' wages, progress payment, sectional settlement, concealed project management, construction progress tracking, correction of construction quantities, quality management during construction, management of major equipment / materials / fittings supplied by the employer, and management of project equipment / materials / fittings supplied by the contractor. Among them, for each project object sample, although the on-site cost management business processes are basically the same and all involve the executing entity, execution content, execution requirements, etc., for different project object samples, there are also differences in their on-site cost management business processes, such as the setting of task nodes and the time of task execution.

[0074] This application constructs an on-site cost management knowledge graph for on-site cost management knowledge retrieval, rapid identification of target project object risks, risk warning, and tracing of solutions to warning problems for project engineering on-site costs. The on-site cost management knowledge graph includes a framework model layer and a data layer. Among them, the framework model layer is also called the ontology layer, which is a description framework of data subjects and the relationships between data subjects. The data layer is also called the data subject layer, which is used to store the data subjects corresponding to the ontology layer and the associated data between data subjects. Both the framework model layer and the data layer are stored in the form of "data subject - relationship - data subject".

[0075] The traditional methods for constructing a knowledge graph mainly include two methods: the top-down method driven by knowledge and the bottom-up method driven by data. For the construction of the framework model layer, the top-down method driven by knowledge is adopted. The on-site cost management business processes of each on-site cost management scenario are extracted in text to determine the on-site cost data of the project object samples. The first text content related to on-site cost risk management is extracted from the on-site cost data, and content recognition is performed on each piece of the first text content to obtain the recognition results. The recognition results include various data subjects and the association relationships between various data subjects. In the preferred embodiment of the present invention, the obtained data subjects include engineering projects, construction management units, design units, construction units, supervision units, cost problems, cost influencing factors, management rules, and management documents. In this application, the attribute types of each data subject are defined to obtain the attributes of each data subject. The attributes of a data subject are the specific characteristics and properties of the data subject, such as the issuing unit, issuing date, and document status of a management document, such as the voltage level, project type, and project characteristics of an engineering project, such as the keywords of a cost problem. Each data subject has corresponding attributes. The various data subjects, the association relationships between the various data subjects, and the attributes of the various data subjects are structured to construct a structured model layer of "data subject - relationship - data subject" triples, thereby obtaining the framework model layer.

[0076] S2. Perform text recognition on the selected on-site cost management documents and historical engineering settlement review drafts, construct an on-site cost standard database according to the text recognition results, and use a data-driven algorithm to process the framework model layer and the on-site cost standard database to obtain a data layer reflecting the association data between the various data subjects; wherein, the on-site cost standard database includes: a cost management document database, a cost management rule database, a cost problem database, a cost influencing factor database, a warning rule database, an equipment material database, and a material price database.

[0077] The construction of the on-site cost standard database mainly involves performing text recognition and rapid extraction on on-site cost management documents and historical engineering settlement review drafts, and storing them as structured and standardized texts to form a cost management document database, a cost management rule database, a cost problem database, a cost influencing factor database, a warning rule database, an equipment price database, and a material price database, providing a data basis for the construction of the data layer of the on-site cost management knowledge graph.

[0078] Specifically, extract the third text content of the selected on-site cost management documents, and perform content recognition on each third text content to obtain the second recognition result; construct a cost management document database and a cost management rule database according to the second recognition result; for example, the cost management document database includes file name, document number, issuing unit, issuing date, status, etc.; the cost management rule database is a management rule entry related to on-site project cost, including rule name, rule text, keyword, source document, etc.

[0079] Extract the fourth text content of the selected historical project settlement review drafts, and perform content recognition on each fourth text content to obtain the third recognition result; construct a cost problem database and a cost influencing factor database according to the third recognition result; for example, for the cost problem database, by mining project cost problems from historical project settlement review drafts, extract key problem information, and classify and organize the cost problems according to different on-site cost management stages and scenarios corresponding to cost problem keywords, to obtain major problem categories or minor problem categories such as "low quality of bidding document preparation", "irregular change and visa management", "inaccurate settlement basis", etc.; for the cost influencing factor database, by mining the main factors causing different categories of cost problems, classify and integrate them, and form an association with the cost problem database to form a cost influencing factor database.

[0080] Extract keywords from the texts of the cost management document database and the cost management rule database, and perform data subject recognition and attribute value recognition on the extracted keywords. The attribute value is a specific value or description associated with a certain attribute, indicating the specific manifestation or measurement of the attribute on a specific thing or object. The attribute value can be a number, text, symbol, etc.; represent the recognized data subject and attribute value in a structured manner to obtain a structured feature; input the structured feature into a random forest model to obtain a quantitative determination condition; based on the historical data of the on-site cost management business process, divide the on-site cost management process into nodes to obtain several business nodes; determine the business node to which the quantitative determination condition belongs, use the quantitative determination condition as the warning rule for the node, and generate a warning rule database based on each warning rule. The warning rule database is used to identify potential risks in on-site project cost risk management and generate risk management plans. Specifically, take the management regulation text "Within 15 days after the completion of the acceptance inspection of 220 kV and above transmission and transformation projects, and within 10 days after the completion of the acceptance inspection of 110 kV and below transmission and transformation projects, the construction, design, supervision and other units and relevant departments such as materials and development shall prepare the corresponding project settlement documents and submit them to the owner's project department" as an example to illustrate the process of automatically constructing a warning rule database by extracting information from the cost management document database and the cost management rule database:

[0081] Data preprocessing, which includes text recognition, word segmentation, stop word removal, and keyword extraction. The result of word segmentation is: Within 15 days after the completion acceptance of power transmission and transformation projects of 220 kV and above, and within 10 days after the completion acceptance of power transmission and transformation projects of 110 kV and below, construction, design, supervision units, and relevant departments such as materials and development shall compile the corresponding project settlement documents and submit them to the owner's project department. After removing common and meaningless stop words such as "of", "is", "in", etc., the result of removing stop words from the sample text is: Within 15 days after the completion acceptance of power transmission and transformation projects of 220 kV and above, within 10 days after the completion acceptance of power transmission and transformation projects of 110 kV and below, construction, design, supervision units, materials, development relevant departments compile the project settlement documents and submit them to the owner's project department. Extract keywords related to project site cost risk management. The extraction result of the sample text is as follows: Power transmission and transformation projects of 220 kV and above, completion acceptance, within 15 days, power transmission and transformation projects of 110 kV and below, within 10 days, construction, design, supervision units, materials, development relevant departments, compile, project settlement documents, submit, owner's project department.

[0082] Feature extraction. For the text after data preprocessing, natural language processing technology is used to identify the data subject category and extract the data subject features. The identification result of the sample data subject category is:

[0083] Technical specification data subject: Identify data subjects related to voltage levels, such as "220 kV and above" and "110 kV and below".

[0084] Time data subject: Identify data subjects related to time limits, such as "within 15 days" and "within 10 days".

[0085] Organization data subject: Identify the units and departments involved, such as "construction unit", "design unit", "supervision unit", "materials department", and "development department".

[0086] Document data subject: Identify document data subjects related to project settlement, such as "project settlement documents".

[0087] Process data subject: Identify data subjects related to project processes, such as "completion acceptance".

[0088] Action data subject: Identify data subjects related to the submission action, such as "submit to the owner's project department".

[0089] The feature extraction result of the sample text:

[0090] Voltage level: Extract "220 kV and above" and "110 kV and below" as the features of the voltage level.

[0091] Time limit: Extract "15 days" and "10 days" as the characteristics of the time limit.

[0092] Responsible parties: Extract "construction, design, and supervision units" and "materials and development-related departments" as the characteristics of the responsible parties.

[0093] Document preparation: Extract "complete the corresponding project settlement documents" as the characteristic of document preparation.

[0094] Submission action: Extract "submit to the owner's project department" as the characteristic of the submission action.

[0095] Structured feature representation: Structurally represent the identified data subject categories and data subject characteristics to obtain structured features. The structured features of the example text are:

[0096] {"voltage level": ["220 kV and above", "110 kV and below"],

[0097] "time limit": ["within 15 days", "within 10 days"],

[0098] "responsible parties": ["construction unit", "design unit", "supervision unit", "materials department", "development department"],

[0099] "document preparation": "complete the corresponding project settlement documents",

[0100] "submission action": "submit to the owner's project department"}.

[0101] Rule extraction: Input the structured features into the random forest model to obtain the quantitative judgment conditions in the on-site cost management regulations. Convert the extracted structured features into a format that can be recognized by the machine learning model. Common coding techniques include one-hot encoding or label encoding. Build a random forest model, input the encoded structured features into the random forest model, form multiple decision tree results, and obtain the quantitative judgment conditions in the on-site cost management regulations. The example quantitative judgment conditions are:

[0102] If voltage level = "220 kV and above" AND time limit = "15 days":

[0103] Then submission action = "submit to the owner's project department".

[0104] If voltage level = "110 kV and below" AND time limit = "10 days":

[0105] Then submission action == "submit to the owner's project department".

[0106] Rule construction: Build an early warning rule database based on quantitative judgment conditions. An example of an early warning rule is as follows:

[0107] For transmission and transformation projects of 220 kV and above, within 15 days after the completion of the acceptance check, it is necessary to complete the preparation of the project settlement documents and submit them to the owner's project department.

[0108] For transmission and transformation projects of 110 kV and below, within 10 days after the completion of the acceptance check, it is necessary to complete the preparation of the project settlement documents and submit them to the owner's project department.

[0109] Furthermore, based on the on-site cost management business process, divide the on-site cost management process into several nodes; determine the node to which the early warning rule belongs. For example, if it is the project settlement node, then this early warning rule is classified as [Settlement Progress Early Warning] in the [Project Settlement] node.

[0110] The equipment price database is constructed by identifying price documents, contract documents, and market price information related to equipment. The material price database is constructed by identifying price documents, contract documents, and market price information related to materials.

[0111] As shown in Table 1 is an example of the on-site cost standard database.

[0112] Table 1

[0113]

[0114] In the on-site cost management knowledge graph, the data layer is also represented by the triple of "data subject - relationship - data subject". Relying on the constructed schema layer framework, set the data subject extraction rules and the relationship extraction rules between data subjects. For each data subject, it consists of a data subject name and multiple attributes of the data subject. The data subject extraction rule is expressed as:

[0115] Entity = Name{P1, P2, …}

[0116] In the formula, Entity represents the data subject, Name represents the data subject name, and P1, P2 represent multiple different attributes of the data subject, which are used to distinguish different types of data subjects.

[0117] In this application, the second text content of the selected on-site cost management documents and historical engineering settlement review drafts is extracted, and the attribute values corresponding to each data subject and attribute are extracted from the second text content according to the data subject extraction rules. Taking the on-site cost management of power transmission and transformation projects as an example, the data subjects related to on-site project costs include engineering projects, construction management units, design units, construction units, supervision units, cost issues, cost influencing factors, management rules, and management documents. For different document types, in this application, two text recognition and extraction methods based on preset rules and deep learning algorithms are respectively adopted. For on-site cost management documents that are standardized documents, preset rules are used to extract the data subjects and the attribute values of the data subjects. In the preferred embodiment of this application, the preset rule is to extract and identify the text at specified positions. The name and release unit of the on-site cost management document are extracted at the beginning and end of the document respectively. Each paragraph in the body content part of the on-site cost management document is an independent rule entry. The historical engineering settlement review draft is a standardized template document, which is extracted starting from a fixed position in a specific "review item" fixed table, and focuses on the review issues of "design changes and on-site visas", "compliance of settlement basis", "verification of settlement quantities", and "verification of settlement prices".

[0118] For the content that cannot be extracted using the preset rules, an identification model is constructed through a deep learning model for text recognition and extraction. For the training of the deep learning model, first, a corpus of data subjects for on-site cost risk management of projects is constructed. According to the corpus of data subjects for on-site cost risk management of projects, the content in the on-site cost management document samples and historical engineering settlement review draft samples that cannot be recognized and extracted using the preset rules is manually marked, and the marked text is input into the deep learning model for training to obtain a trained identification model. The on-site cost management documents and historical engineering settlement review drafts that need to be recognized are input into the identification model, and the recognition results are output. Table 2 shows the recognition methods for the data subjects and the attribute values of the data subjects for on-site cost risk management of projects.

[0119] Table 2

[0120]

[0121]

[0122] During the process of data subject extraction, due to differences in word usage and writing habits, there may be data subject names with similar meanings that are easily confused. Therefore, it is necessary to identify different names or relationship representations of the same data subject in multi-source heterogeneous data, and perform fusion and unification to avoid data redundancy. Knowledge fusion mainly unifies different word expressions with the same meaning. For example, "construction management unit" and "construction and management unit" should be the same noun; "design change and on-site visa", "change visa", and "design change visa" should be the same noun. In this application, cosine similarity is used to calculate the similarity of different data subject names. Specifically, the text is tokenized, then the two word lists are merged and duplicates are removed, the eigenvalues are calculated respectively, the calculated eigenvalues are vectorized, and finally the cosine value of the text similarity is calculated. By comparing whether the cosine value exceeds the threshold, it is determined whether two independent data subjects are the same data subject. If so, they are merged. Among them, the cosine similarity calculation formula is:

[0123]

[0124] Among them, and are the vector representations of two compared texts respectively, and are the vector values in two compared texts respectively.

[0125] Furthermore, according to the rule for extracting relationships between data subjects, the associated data between each data subject is extracted from the second text content; among them, the rule for extracting relationships between data subjects is expressed as:

[0126] Relationship=Name{Entity start →[Relationship]→Entity end}

[0127] In the formula, Entitystart and Entityend respectively represent the head data subject and the tail data subject, the arrow indicates the relationship direction between the two data subjects, and Relationship represents the relationship name.

[0128] Before extracting the associated data between data subjects, it is necessary to construct an association relationship library based on the framework pattern layer. For example, the relationship between the construction and management unit and the project belongs to the construction management relationship, the relationship between the design unit and the project belongs to the design relationship, the relationship between the project and the cost problem belongs to the cost problem relationship, the relationship between the rule entry and the management document belongs to the source document relationship, etc.

[0129] Further, based on the framework mode layer, the structured processing is performed on each data subject, the attribute values of each data subject, and the associated data between each data subject. Specifically, in a manner corresponding one-to-one with the data subjects in the framework mode layer, the attribute values and the associated data are added to the corresponding data subjects to obtain the data layer. There is a one-to-one correspondence between the data subjects in the framework mode layer and the data subjects in the data layer.

[0130] S3. Determine the on-site cost management knowledge graph according to the framework mode layer and the data layer; for the completed on-site cost management knowledge graph, use Neo4j for visual display to construct an on-site cost management knowledge retrieval platform, and the on-site cost management knowledge retrieval platform quickly searches and matches according to the input keywords.

[0131] Due to the continuous acceleration of informatization construction, the relevant on-site cost management documents and engineering settlement review drafts are also constantly updated. Therefore, the on-site cost management knowledge graph also needs to be continuously updated and improved to ensure the timeliness and accuracy of the data. When there are new project on-site cost risk management documents and batch engineering settlement review drafts, it is necessary to timely identify, import, and maintain the new on-site cost management documents and engineering settlement review drafts. During the update process, if there are data subjects that are inconsistent with the data subjects in the framework mode layer, it indicates that new data subjects have emerged. Update the new data subjects to the data subjects in the framework mode layer, and define the attribute types of the newly added data subjects in the framework mode layer and construct the relationships between the data subjects to update the framework mode layer. Based on the updated framework mode layer, update the data layer to achieve the update of the on-site cost management knowledge graph.

[0132] S4. In the process of project site cost risk management, the inference engine algorithm is used to match the target project object with the on-site cost management knowledge graph, and the risk management list of the target project object is determined according to the matching result. When a new project starts, the basic feature data set of the target project object is extracted, mainly including the core fields of the target project object such as project type, project nature, construction scale, voltage level, construction location, etc. The pre-constructed multi-layer perceptron is used to extract key features from the basic feature data set to obtain the key feature set. The key feature set is input into the on-site cost management knowledge retrieval platform and matched with the on-site cost management knowledge graph, and the cost problem database and the cost influencing factor database are associated according to the matching result. Specifically, the key feature set is vectorized based on semantics to obtain the project feature value vector, and the similarity between the project feature value vector and the attribute value vector of the data subject of the on-site cost management knowledge graph is calculated for matching, and the attribute value and the affiliated data subject with the highest similarity to the project feature value vector are obtained, and thus associated with the cost problem database and the cost influencing factor library, so as to automatically identify risks before the project starts and make the on-site cost risk management of the project more targeted.

[0133] S5. Screen the on-site cost file data of the target project object according to the on-site cost standard database, and generate a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph; during the project process, extract the fourth text content of the on-site cost file generated at the project site, and input the fourth text content into the on-site cost management knowledge retrieval platform to determine the business node where the target project object is currently located. Based on the determined business node, screen the warning rules of the business node from the warning rule database, and based on the screened warning rules, conduct risk control and risk warning on the target project object, and trace the source of the risk warning according to the on-site cost management knowledge graph to generate a corresponding risk management plan. For example, a reminder is given before the settlement document is submitted to take preventive control measures in advance to avoid the recurrence of the same problem. If the settlement document is not submitted after the specified time, the warning rule is triggered to generate a risk warning. Further, trace the source of the risk warning according to the on-site cost management knowledge graph, associate it with the cost management rule related to the risk warning, and generate a risk management plan to provide a standardized reference solution.

[0134] In a preferred embodiment of the present invention, based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, the on-site cost data of the project object samples is determined. A knowledge-driven algorithm is used to process the on-site cost data to obtain a framework pattern layer reflecting the relationships between various data subjects; the selected on-site cost management documents and historical engineering settlement review drafts are subjected to character recognition, and an on-site cost standard database is constructed according to the character recognition results. A data-driven algorithm is used to process the framework pattern layer and the on-site cost standard database to obtain a data layer reflecting the associated data between various data subjects; according to the framework pattern layer and the data layer, an on-site cost management knowledge graph is determined; in the process of on-site cost risk management of a project, an inference engine algorithm is used to match the target project object with the on-site cost management knowledge graph, and a risk management list of the target project object is determined according to the matching result; the on-site cost document data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph. The project on-site cost risk management method provided by this application constructs an on-site cost management knowledge graph that integrates on-site cost data subjects and historical cost problems, realizes online indexing of on-site cost knowledge in the process of project on-site cost risk management, automatically identifies and locates on-site cost risks, traces and analyzes relevant management documents of risk problems to automatically generate a risk management plan, and guides on-site cost personnel to conduct timely, scientific and targeted control.

[0135] Correspondingly, as Figure 2 shown, based on a project on-site cost risk management method, an embodiment of the present invention further provides a project on-site cost risk management system, which is applied to a project object composed of at least multiple data subjects and realizes the project on-site cost risk management method disclosed in the embodiment of the present invention, including: a framework pattern layer construction unit 1, a data layer construction unit 2, an on-site cost management knowledge graph determination unit 3, a risk management list generation unit 4, and a risk management plan generation unit 5;

[0136] The framework pattern layer construction unit 1 is configured to determine the on-site cost data of the project object samples based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, and use a knowledge-driven algorithm to process the on-site cost data to obtain a framework pattern layer reflecting the relationships between various data subjects.

[0137] The data layer construction unit 2 is configured to perform character recognition on the selected on-site cost management documents and historical engineering settlement review drafts, construct an on-site cost standard database according to the character recognition results, and use a data-driven algorithm to process the framework pattern layer and the on-site cost standard database to obtain a data layer reflecting the associated data between various data subjects.

[0138] The on-site cost management knowledge graph determination unit 3 is used to determine the on-site cost management knowledge graph according to the framework pattern layer and the data layer.

[0139] The risk management list generation unit 4 is used to match the target project object with the on-site cost management knowledge graph by using the inference engine algorithm during the risk management process of the on-site cost of the project, and determine the risk management list of the target project object according to the matching result.

[0140] The risk management plan generation unit 5 is used to screen the on-site cost file data of the target project object according to the on-site cost standard database, and generate a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph.

[0141] Among them, the use of the knowledge-driven algorithm to process the on-site cost data to obtain the framework pattern layer reflecting the relationships between the data subjects includes:

[0142] Extract the first text content in the on-site cost data;

[0143] Perform content recognition on each of the first text contents to obtain recognition results, where the recognition results include each data subject and the association relationships between each of the data subjects;

[0144] Define the attribute types of each of the data subjects to obtain the attributes of each of the data subjects;

[0145] Perform structured processing on each of the data subjects, the association relationships between each of the data subjects, and the attributes of each of the data subjects to obtain the framework pattern layer.

[0146] Among them, the use of the data-driven algorithm to process the pattern layer and the on-site cost standard database to obtain the data layer reflecting the associated data between the data subjects includes:

[0147] Based on the pattern layer, set the data subject extraction rule and the data subject relationship extraction rule;

[0148] Extract the second text content of the selected on-site cost management documents and historical project settlement review drafts, and extract the attribute values corresponding to each of the data subjects and the attributes from the second text content according to the data subject extraction rule;

[0149] Extract the associated data between each of the data subjects from the second text content according to the data subject relationship extraction rule;

[0150] The attribute values and the associated data are added to the corresponding data bodies in a one-to-one correspondence with the data bodies of the pattern layer, resulting in a data layer.

[0151] Among them, the character recognition of the selected on-site cost management documents and historical project settlement review drafts, and the construction of an on-site cost standard database according to the character recognition results, include:

[0152] Extract the third text content of the selected on-site cost management documents, and perform content recognition on each of the third text contents to obtain a second recognition result;

[0153] Construct a cost management document database and a cost management rule database according to the second recognition result;

[0154] Extract the fourth text content of the selected historical project settlement review drafts, and perform content recognition on each of the fourth text contents to obtain a third recognition result;

[0155] Construct a cost problem database and a cost influencing factor database according to the third recognition result;

[0156] Extract keywords from the text of the cost management document database and the cost management rule database, and perform data body recognition and attribute value recognition on the extracted keywords;

[0157] Structurally represent the recognized data bodies and attribute values to obtain structural features;

[0158] Input the structural features into a random forest model to obtain quantitative judgment conditions;

[0159] Based on the historical data of the on-site cost management business process, divide the on-site cost management process into nodes to obtain a number of business nodes;

[0160] Judge the business node to which the quantitative judgment condition belongs, use the quantitative judgment condition as the warning rule for the node, and generate a warning rule database based on each warning rule.

[0161] Among them, the screening of the on-site cost document data of the target project object according to the on-site cost standard database, and the generation of a risk management plan corresponding to the screening result based on the on-site cost management knowledge graph, include:

[0162] Extract the fourth text content of the on-site cost document of the target project object to determine the business node where the target project object is located;

[0163] Based on the business node, screen the warning rules of the business node from the warning rule database;

[0164] Based on the screened early warning rules, risk control and risk early warning are carried out on the target project object, and the risk early warning is traced back according to the on-site cost management knowledge graph, and a corresponding risk management plan is generated.

[0165] For the specific limitations of a project on-site cost risk management system, reference can be made to the above limitations on a project on-site cost risk management method, which will not be elaborated here. Those of ordinary skill in the art can realize that, in combination with the various modules and steps described in the embodiments disclosed in the present invention, they can be implemented by hardware, software, or a combination of both. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0166] A project on-site cost risk management method and system provided in this embodiment are used to solve the technical problem of how to effectively manage the on-site cost of a project and automatically identify risks. Based on the historical data of on-site cost management scenarios and the historical data of on-site cost management business processes including all types of project object samples, the on-site cost data of the project object samples are determined, and knowledge-driven algorithms are used to process the on-site cost data to obtain a framework pattern layer reflecting the relationships between various data subjects; character recognition is performed on the selected on-site cost management documents and historical engineering settlement review drafts, and an on-site cost standard database is constructed according to the character recognition results. Data-driven algorithms are used to process the framework pattern layer and the on-site cost standard database to obtain a data layer reflecting the associated data between various data subjects; according to the framework pattern layer and the data layer, an on-site cost management knowledge graph is determined; in the process of project on-site cost risk management, the inference engine algorithm is used to match the target project object with the on-site cost management knowledge graph, and the risk management list of the target project object is determined according to the matching result; the on-site cost document data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph. The project on-site cost risk management method provided in this application constructs an on-site cost management knowledge graph that integrates on-site cost data subjects and historical cost problems, realizes online indexing of on-site cost knowledge, automatic identification and positioning of on-site cost risks, and traceability analysis of management documents related to risk problems to automatically generate a risk management plan during the process of project on-site cost risk management, guiding on-site cost personnel to conduct timely, scientific, and targeted control.

[0167] Each embodiment in this specification is described in a progressive manner. For the parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, reference can be made to the description in the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0168] The above embodiments only represent several preferred embodiments of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those of ordinary skill in the art in this technical field, without departing from the technical principle of the present invention, several improvements and substitutions can still be made, and these improvements and substitutions should also be regarded as the protection scope of the present invention. Therefore, the protection scope of the invention patent should be subject to the protection scope of the claims.

Claims

1. A project site cost risk management method, characterized in that: Applied to a project object consisting of at least a plurality of data subjects, the method comprises: Based on the on-site cost management scenario historical data and on-site cost management business process historical data of all types of project object samples, the on-site cost data of the project object samples are determined, and the on-site cost data are processed using a knowledge-driven algorithm to obtain a framework model layer reflecting the relationship between each of the data subjects; Performing text recognition on the selected on-site cost management documents and historical engineering settlement review manuscripts, constructing an on-site cost standard database based on the text recognition results, and processing the framework model layer and the on-site cost standard database using a data-driven algorithm to obtain a data layer reflecting the associated data between each of the data subjects; Determine a knowledge graph of on-site cost management according to the framework model layer and the data layer; In the process of project site cost risk management, the target project object is matched with the site cost management knowledge graph using the inference engine algorithm, and the risk management list of the target project object is determined according to the matching result; The on-site cost document data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph.

2. The project site cost risk management method according to claim 1, characterized in that: The knowledge-driven algorithm is used to process the on-site cost data to obtain a framework model layer reflecting the relationship between each of the data subjects, including: Extracting the first text content in the on-site cost data; Performing content recognition on each of the first text contents to obtain a recognition result, wherein the recognition result includes each data subject and an association relationship between each of the data subjects; Defining the attribute type of each of the data subjects to obtain the attributes of each of the data subjects; Each of the data subjects, the associations between the data subjects, and the attributes of each of the data subjects are structured to obtain a framework model layer.

3. The project site cost risk management method according to claim 2, characterized in that: The text recognition of the selected on-site cost management documents and historical engineering settlement review manuscripts is performed, and a site cost standard database is constructed according to the text recognition results, including: Extracting the third text content of the selected on-site cost management file, and performing content recognition on each of the third text contents to obtain a second recognition result; Constructing a cost management document database and a cost management rule database according to the second recognition result; Extracting the fourth text content of the selected historical project settlement review manuscript, and performing content recognition on each of the fourth text contents to obtain a third recognition result; Constructing a cost problem database and a cost influencing factor database according to the third recognition result; Extracting keywords from the texts of the cost management document database and the cost management rule database, and identifying the data subject and attribute values ​​of the extracted keywords; The identified data subject and attribute value are structured to obtain structured features; Inputting the structured features into a random forest model to obtain quantitative judgment conditions; Based on the historical data of the on-site cost management business process, the on-site cost management process is divided into nodes to obtain a plurality of business nodes; The service node to which the quantitative determination condition belongs is determined, so as to use the quantitative determination condition as an early warning rule of the node, and to generate an early warning rule database based on each of the early warning rules.

4. The project site cost risk management method according to claim 3, characterized in that: The data-driven algorithm is used to process the model layer and the on-site cost standard database to obtain a data layer reflecting the associated data between the various data subjects, including: Based on the pattern layer, setting data subject extraction rules and relationship extraction rules between data subjects; Extracting the second text content of the selected on-site cost management documents and historical engineering settlement review manuscripts, and extracting attribute values ​​corresponding to each of the data subjects and the attributes from the second text content according to the data subject extraction rule; Extracting the associated data between each of the data subjects from the second text content according to the relationship extraction rule between the data subjects; The attribute value and the associated data are added to the corresponding data subject in a one-to-one correspondence manner with the data subject of the pattern layer to obtain a data layer.

5. The project site cost risk management method according to claim 4, characterized in that: The on-site cost document data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph, including: Extracting the fourth text content of the on-site cost document of the target project object to determine the business node where the target project object is located; Based on the service node, screening the warning rules of the service node from the warning rule database; Based on the screened warning rules, risk control and risk warning are carried out on the target project object, and the risk warning is traced according to the on-site cost management knowledge graph to generate a corresponding risk management plan.

6. A project site cost risk management system, applied to a project object consisting of at least a plurality of data subjects, to implement the project site cost risk management method according to any one of claims 1 to 5, characterized in that: The system comprises: a framework model layer construction unit, a data layer construction unit, a site cost management knowledge graph determination unit, a risk management list generation unit and a risk management plan generation unit; The framework model layer construction unit is used to determine the on-site cost data of the project object sample based on the on-site cost management scenario history data and on-site cost management business process history data of all types of project object samples, and process the on-site cost data using a knowledge-driven algorithm to obtain a framework model layer that reflects the relationship between each of the data subjects; The data layer construction unit is used to perform text recognition on the selected on-site cost management documents and historical engineering settlement review manuscripts, construct an on-site cost standard database based on the text recognition results, and use a data-driven algorithm to process the framework model layer and the on-site cost standard database to obtain a data layer reflecting the associated data between each of the data subjects; The on-site cost management knowledge graph determining unit is used to determine the on-site cost management knowledge graph according to the framework model layer and the data layer; The risk management list generating unit is used to match the target project object with the on-site cost management knowledge graph using an inference engine algorithm during the project on-site cost risk management process, and determine the risk management list of the target project object according to the matching result; The risk management solution generating unit is used to screen the on-site cost document data of the target project object according to the on-site cost standard database, and generate a risk management solution corresponding to the screening result based on the on-site cost management knowledge graph.

7. The project site cost risk management system according to claim 6, characterized in that: The knowledge-driven algorithm is used to process the on-site cost data to obtain a framework model layer reflecting the relationship between each of the data subjects, including: Extracting the first text content in the on-site cost data; Performing content recognition on each of the first text contents to obtain a recognition result, wherein the recognition result includes each data subject and an association relationship between each of the data subjects; Defining the attribute type of each of the data subjects to obtain the attributes of each of the data subjects; Each of the data subjects, the associations between the data subjects, and the attributes of each of the data subjects are structured to obtain a framework model layer.

8. The project site cost risk management system according to claim 7, characterized in that: The text recognition of the selected on-site cost management documents and historical engineering settlement review manuscripts is performed, and a site cost standard database is constructed according to the text recognition results, including: Extracting the third text content of the selected on-site cost management file, and performing content recognition on each of the third text contents to obtain a second recognition result; Constructing a cost management document database and a cost management rule database according to the second recognition result; Extracting the fourth text content of the selected historical project settlement review manuscript, and performing content recognition on each of the fourth text contents to obtain a third recognition result; Constructing a cost problem database and a cost influencing factor database according to the third recognition result; Extracting keywords from the texts of the cost management document database and the cost management rule database, and identifying the data subject and attribute values ​​of the extracted keywords; The identified data subject and attribute value are structured to obtain structured features; Inputting the structured features into a random forest model to obtain quantitative judgment conditions; Based on the historical data of the on-site cost management business process, the on-site cost management process is divided into nodes to obtain a plurality of business nodes; The service node to which the quantitative determination condition belongs is determined, so as to use the quantitative determination condition as an early warning rule of the node, and to generate an early warning rule database based on each of the early warning rules.

9. The project site cost risk management system according to claim 8, characterized in that: The data-driven algorithm is used to process the model layer and the on-site cost standard database to obtain a data layer reflecting the associated data between the various data subjects, including: Based on the pattern layer, setting data subject extraction rules and relationship extraction rules between data subjects; Extracting the second text content of the selected on-site cost management documents and historical engineering settlement review manuscripts, and extracting attribute values ​​corresponding to each of the data subjects and the attributes from the second text content according to the data subject extraction rule; Extracting the associated data between each of the data subjects from the second text content according to the relationship extraction rule between the data subjects; The attribute value and the associated data are added to the corresponding data subject in a one-to-one correspondence manner with the data subject of the pattern layer to obtain a data layer.

10. The project site cost risk management system according to claim 9, characterized in that: The on-site cost document data of the target project object is screened according to the on-site cost standard database, and a risk management plan corresponding to the screening result is generated based on the on-site cost management knowledge graph, including: Extracting the fourth text content of the on-site cost document of the target project object to determine the business node where the target project object is located; Based on the service node, screening the warning rules of the service node from the warning rule database; Based on the screened warning rules, risk control and risk warning are carried out on the target project object, and the risk warning is traced according to the on-site cost management knowledge graph to generate a corresponding risk management plan.

Images