Intelligent standardization method and system for hydropower engineering quantity list

Abstract

According to the water-power engineering quantity list intelligent standardization method and system, the water-power engineering quantity list intelligent standardization system is constructed, a first party makes a standard requirement table on an upper computer and applies for the engineering quantity list in a special format, and for a second party with qualification, the upper computer installed with water-power engineering quantity list standardization application software is used for authorization; a list table-oriented authorization function in the hydropower engineering quantity list standardized application software is obtained in a second-party host, a standard requirement table is called by means of the authorization function, filling is performed by means of cost software and list software, and a five-step first-party and second-party interaction bid invitation mode is performed when qualification does not exist. And high efficiency and intelligent automation of engineering project application table expression standardization are realized.

Description

Technical Field

[0001] The present invention relates to intelligent standardization of hydropower project quantities lists, and in particular to automatic keyword retrieval and filling of natural language documents based on image recognition, as well as an intelligent expert review scheme, belonging to the field of intelligent review. Background Art

[0002] Existing engineering review systems generally require the creation of a standard checklist format, which is then subject to expert review. A significant drawback is that after the design firm completes the checklist using cost and checklist software, it must be standardized and re-formatted using standardization software before it can be submitted to the ECP system and submitted to the contractor. The contractor then has to convert the checklist back to the standard format, create the application form, and then, after a second round of standardization, upload it to the ECP again, repeatedly switching between file formats. This process not only ignores whether the document's substantive content meets the contractor's requirements, but also significantly reduces overall project reporting efficiency due to the frequent and unnecessary format conversions, which increase local system load.

[0003] Therefore, the problem that needs to be solved is very clear. How to express the substantive content of the Party A's form clearly and concisely, fill it out in a special format list at one time, and send it back to Party A is the core issue of the overall efficient and standardized solution for the engineering project. Summary of the Invention

[0004] Based on the above problems of the prior art, the present invention aims to provide a technical solution to solve the technical problems by adopting the following technical measures: First, highlight the "conditional off-site authorization" of the inventory function of the standardized application software for hydropower project quantities, complete the standardized inventory locally at the qualified Party B, upload it at one time, and reduce the burden on the system.

[0005] This key "conditional remote authorization" solution eliminates the traditional step of Party A creating specialized forms locally for project requirements, handing the form-filling task directly to a qualified Party B. This eliminates the need for format conversions or the two-step process of document upload and download. Instead, the focus is on completing the required substantive content. The focus is on results.

[0006] Therefore, "one of the keys" to solving this problem lies in the confirmation of qualifications, so that conditional authorization can be achieved, and "the second key" is how to clearly express Party A's list filling requirements.

[0007] Second, consider intelligently filling out standardized checklists. This involves automatically extracting and filling in keywords from natural language documents using image recognition. The key to achieving this intelligent filling lies in quickly developing data that demonstrates Party B's competence in accepting a project, based on Party A's requirements.

[0008] Third, regarding natural language models, there is the problem of unique word spatial distribution for filling in the form, so how to use this spatial distribution to quickly obtain keywords becomes the key to the algorithm; at the same time, considering the appropriate language model, taking into account the standardization of vocabulary expressions in special tables.

[0009] Fourth, the present invention solves the technical problem through hardware system technology, from conditional remote authorization, and image-based intelligent filling, two scientific and technological means.

[0010] Specifically, in order to implement the above solution, the present invention provides an intelligent standardization method for hydropower project quantity list, which specifically includes the following steps: The first step is to build an intelligent standardization system for hydropower project quantity lists. The system includes a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A and uses a locally preset model to input the qualification parameters provided by Party B and provide a classification result of whether the qualification is obtained. If you have, execute steps 2 to 4: In the second step, Party A prepares the standard requirement form and the special format for applying for the bill of quantities on the host computer and saves them on the server; In the third step, the qualified Party B installs the upper computer authorization of the hydropower bill of quantities standardization application software, obtains the authorization function for the bill form in the hydropower bill of quantities standardization application software on the Party B host, and uses this authorization function to call the standard requirement form and fill it out using the cost software and bill software; In the fourth step, Party B uses the authorization function in Party B's host computer to retrieve the manually completed standard requirement form through the natural language document automatic keyword extraction based on image recognition, complete the retrieved special format to apply for the bill of quantities, and submit it to the ECP system; If not, perform the following steps: S1: The design unit uses cost software, bill of quantities software, etc. to prepare the tender bill of quantities in standard Excel format; S2: The design unit uses the standardized application software for hydropower bill of quantities to verify the standardization of the form. Once the verification is qualified, the standard Excel format tender bill of quantities is converted into a special format tender bill of quantities and uploaded as an attachment to the ECP. After the technical specification is generated, other work such as review of the bidding documents for submission is carried out; S3: After the tender is issued, the applicant downloads the tender document package containing the tender bill of quantities and uses the hydropower bill of quantities standardization application software to convert the tender bill of quantities in the special format into a standard Excel format bill of quantities; S4: The applicant shall use the standard Excel format for bidding bill of quantities and use cost estimation software, bill of quantities software, etc. to prepare the standard Excel format for application bill of quantities; S5: The applicant uses the standardized application software for hydropower bill of quantities to verify the form and identify any discrepancies. Once the verification is qualified, the applicant converts the standard Excel format bill of quantities application into a special format bill of quantities application and submits it to the ECP; Step 5. This step follows Step 4 and S5 respectively, depending on whether the qualification classification is available. The bid evaluation expert uses the bidding software to receive special bidding documents and bid opening list information from the ECP, generates a bid evaluation environment, and uses the bid evaluation software to receive a special format application bill of quantities from the ECP for review. The comparison results, difference prompts, and mutually exclusive relationship reviews are obtained, and the expert is assisted in the bid evaluation, and the rejection / scoring situation is returned to the ECP.

[0011] In the above overall plan, it should be understood that when the preset model predicts that Party B is not qualified, the small probability that it is competent should not be immediately ignored. In order not to dampen its enthusiasm, a non-conditional authorization list filling mechanism can be adopted, namely steps S1-S5.

[0012] Optionally, the preset model includes a multidimensional support vector machine, and its training method includes: Party A collects the actual qualification parameters of Party B, and statistically analyzes the actual qualification parameters to produce training parameters and verification parameters, and sets labels for the training parameters and verification parameters as having and not having qualifications, constructs a multidimensional support vector machine, trains using the training parameters, and verifies using the verification parameters, optimizes the hyperplane, and stops training until the classification accuracy is maximized.

[0013] Optionally, the qualification parameters include the number of projects accepted over the years, project investment amount, project completion evaluation level, project personnel's education level, length of service, and annual turnover rate.

[0014] Optionally, the statistical analysis adopts normal distribution function fitting, and takes the qualification parameter distributed between positive and negative standard deviations σ.

[0015] The method for automatically extracting keywords from a natural language document based on image recognition and completing the application for a bill of quantities in a special format includes the following steps: Step I: Take a screenshot of the manually filled standard requirement form and establish a mapping relationship between the filled-in items in the standard requirement form and the corresponding items in the special format application bill of quantities; Step II: Use the preset natural language model to identify keywords in the screenshot and retrieve them; Step III: Call the special format to apply for the bill of quantities, and use the mapping to map each keyword to the special format to complete the bill of quantities application.

[0016] The natural language model is an artificial intelligence model for recognizing documents. However, in form-filling tasks, its essence is to retrieve information in fixed spatial locations in the created form, which is different from the word segmentation problem of generating image characters in general articles.

[0017] Therefore, preferably, the natural language model construction method includes: Q1 Establish a two-dimensional spatial coordinate system in the standard requirement form and calibrate the spatial coordinates according to the position of the filled-in items; Q2 collects screenshots of multiple standard requirement forms filled out by Party B, extracts the filled-in item sub-graphs according to the calibrated spatial coordinates, and builds a vocabulary library for each sub-graph; Q3 builds a feedforward neural network language model to identify the documents in the subgraph and convert them into the word prediction results with the highest probability in the corresponding vocabulary library to form keywords with standard expressions.

[0018] The feedforward neural network language model is considered suitable for form filling because the advantages of this model are: There is no need to expand and update the vocabulary library. Instead, it is only necessary to predict the most likely expression of the document to be tested based on the existing vocabulary. Therefore, as long as a vocabulary library of standard vocabulary is set up, expressions that may not be professional and unified can be converted into standard expressions that the vocabulary library considers to be the most likely. For example, when filling in the education level as a postgraduate, it generally refers to a master's degree. Through model training, it will be given that the education level parameter is actually most likely to express a master's degree. That is, the model output recognition is a master's degree, thus solving the problem of standardization of special form expressions and the expression of substantive content. This achieves the aforementioned "second key" to clearly express the requirements for Party A to fill in the list. The key to achieving this standardized expression is that the model uses low-dimensional dense vectors, which is the basis for words to find similar standard expressions in the vocabulary library.

[0019] Another object of the present invention is to provide an intelligent standardization system for hydropower engineering quantity lists, including a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A, the host computer is connected to the server, and the host computer of Party B communicates with the server. The host computer is installed with standardized application software for hydropower engineering quantity lists. The application software includes an interface presentation layer, a business logic layer, a basic business layer, a system component layer, and data storage. The above method is implemented through layer-by-layer calling, processing and analysis of data.

[0020] According to the present invention, the substantial content of the form filled in by Party A can be expressed clearly and concisely, and filled in a special format list at one time and sent back to Party A, effectively providing an overall efficient and standardized solution for the engineering project.

[0021] According to the present invention, the standardized application software for hydropower project quantity lists involves "conditional off-site authorization" of the list function, and the standardized list is completed locally by a qualified Party B and uploaded at one time to reduce the burden on the system.

[0022] According to the present invention, the technical problem can be effectively solved through hardware system technology, from conditional remote authorization, and image-based intelligent filling, two scientific and technological means. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flowchart of the intelligent standardization method for hydropower project quantity bills in embodiment 1 of the present invention, Figure 2 Logic diagram of the hydropower project quantity list standardization application software architecture, Figure 3 The preset model includes a multi-dimensional support vector machine training method flow chart, Figure 4 The distribution diagram of data points in the process of constructing a multidimensional support vector machine in the six-dimensional qualification parameter space, Figure 5 Screenshots and mapping example diagrams of the method for automatically extracting keywords and filling in special format for applying for bill of quantities in natural language documents based on image recognition, Figure 6 The method of automatically extracting keywords from natural language documents based on image recognition and completing the application of a special format for a bill of quantities is shown in the figure below: Figure 7 Configuration diagram of the intelligent standardization system for hydropower project quantity bills according to embodiment 2 of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be described in detail below in combination with specific embodiments with reference to the accompanying drawings. The description is for illustrative purposes only, and the present invention is not limited to the specific embodiments. Example 1

[0025] This embodiment will illustrate the intelligent standardization method of hydropower project quantity list. Figure 1 The specific steps shown are as follows: The first step is to build an intelligent standardization system for hydropower project quantity list. The system includes a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A. Through a locally preset model, the qualification parameters provided by Party B are input and a classification result is given as to whether the qualification is met. If the qualification is met, the second to fourth steps are executed: In the second step, Party A prepares the standard requirement form and the special format for applying for the bill of quantities on the host computer and saves them on the server; In the third step, the qualified Party B installs the upper computer authorization of the hydropower bill of quantities standardization application software, obtains the authorization function for the bill form in the hydropower bill of quantities standardization application software on the Party B host, and uses this authorization function to call the standard requirement form and fill it out using the cost software and bill software; Among them Figure 2 The logic diagram of the hydropower project quantity list standardization application software architecture shows the layered structure of the specific software modules, among which, The interface presentation layer includes Winform forms and DSkin controls; the business logic layer includes project management module, bill of quantities module, normative inspection module, list inspection, and clarification management module; the basic business module includes project information management module, bid package information management module, normative inspection configuration, list inspection configuration module, clarification information management module, quotation inspection configuration module, report inspection configuration module, and report preview module; the system software layer includes log components, XML components, EXCEL components, dictionary components, encryption and decryption components; data storage includes XML, XMD, and local file system.

[0026] The authorization is completed by the collaboration of the engineering management module and the engineering information management module. The former collects qualification parameters through the engineering management module, and the latter inputs the qualification parameters into the preset model to give a classification result of whether the qualification is obtained.

[0027] In the fourth step, Party B uses the authorization function in Party B's host computer to retrieve the manually completed standard requirement form through the natural language document automatic keyword extraction based on image recognition, complete the retrieved special format to apply for the bill of quantities, and submit it to the ECP system; If not, perform the following steps: S1: The design unit uses cost software, bill of quantities software, etc. to prepare the tender bill of quantities in standard Excel format; S2: The design unit uses the standardized application software for hydropower bill of quantities to verify the standardization of the form. Once the verification is qualified, the standard Excel format tender bill of quantities is converted into a special format tender bill of quantities and uploaded as an attachment to the ECP. After the technical specification is generated, other work such as review of the bidding documents for submission is carried out; S3: After the tender is issued, the applicant downloads the tender document package containing the tender bill of quantities and uses the hydropower bill of quantities standardization application software to convert the tender bill of quantities in the special format into a standard Excel format bill of quantities; S4: The applicant shall use the standard Excel format for bidding bill of quantities and use cost estimation software, bill of quantities software, etc. to prepare the standard Excel format for application bill of quantities; S5: The applicant uses the standardized application software for hydropower bill of quantities to verify the form and identify any discrepancies. Once the verification is qualified, the applicant converts the standard Excel format bill of quantities application into a special format bill of quantities application and submits it to the ECP; Step 5. This step follows Step 4 and S5 respectively, depending on whether the qualification classification is available. The bid evaluation expert uses the bidding software to receive special bidding documents and bid opening list information from the ECP, generates a bid evaluation environment, and uses the bid evaluation software to receive a special format application bill of quantities from the ECP for review. The comparison results, difference prompts, and mutually exclusive relationship reviews are obtained, and the expert is assisted in the bid evaluation, and the rejection / scoring situation is returned to the ECP.

[0028] The preset model includes a multidimensional support vector machine, and its training method includes the following Figure 3 As shown: Party A collects the actual qualification parameters of Party B in society, and fits the normal distribution function, takes the qualification parameters distributed between positive and negative standard deviations σ to make training parameters and verification parameters, and sets the labels of qualified and unqualified for the training parameters and verification parameters, constructs a multidimensional support vector machine, trains with the training parameters, and verifies with the verification parameters, optimizes the hyperplane, and stops training until the classification accuracy is maximized.

[0029] The qualification parameters include the number of projects accepted over the years, project investment amount, project completion evaluation level, project personnel's education level, length of service, and annual turnover rate.

[0030] Figure 4 As shown, during the specific training, the multidimensional support vector machine is divided into a first multidimensional support vector machine SVM1 and a second multidimensional support vector machine SVM2, which are respectively responsible for training two groups of qualification parameters: the number of projects accepted over the years, the amount of project investment, the project completion evaluation level, and the education level, length of service, and annual turnover rate of project personnel.

[0031] The method for automatically retrieving keywords from a natural language document based on image recognition and completing the application for a bill of quantities in a special format includes the following steps: Step I: If Figure 5 As shown, take a screenshot of the manually filled standard requirement form, and establish a mapping relationship between the filled-in items of the standard requirement form (such as the four matrix boxes in the figure) and the corresponding items (project name, unit, quantity, unit price) in the special format application bill of quantities; like Figure 6 As shown, step II: using a preset feedforward neural network language model to identify keywords in the screenshot and retrieve them; Step III: Call the special format to apply for the bill of quantities, and use the mapping to map each keyword to the special format to complete the bill of quantities application.

[0032] The natural language model construction method includes: Q1 Establish a two-dimensional space coordinate system in the standard requirement table (such as Figure 6 In the coordinate system O), the spatial coordinates are calibrated according to the position of the filled-in items; for example, the coordinate of the number of projects accepted over the years is 2B, and the project investment amount is 2C.

[0033] Q2 collects screenshots of multiple standard requirement forms filled in by Party B, and extracts the filled-in item sub-graphs according to the calibrated spatial coordinates ( Figure 6 ), build a vocabulary library for each sub-graph; the box filled with the number of projects accepted in previous years is sub-graph a, and the box filled with the project investment amount is sub-graph b.

[0034] Q3 builds a feedforward neural network language model to identify the documents in the subgraph and convert them into the word prediction results with the highest probability in the corresponding vocabulary library to form keywords with standard expressions.

[0035] Therefore, in step II, a preset natural language model is used to identify the keywords in the screenshot. The specific process is also to take a screenshot of the standard requirement form, extract the fill-in item subgraph according to the calibrated spatial coordinates, and use the trained feedforward neural network language model to identify the documents in the subgraph and convert them into the word prediction results with the highest probability in the corresponding vocabulary library to form keywords with standard expressions.

[0036] Example 2 like Figure 7 As shown, the intelligent standardization system for the bill of quantities of hydropower projects includes a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A and connected to the server. The host computer of Party B communicates with the server. The host computer is installed with the standardized application software for the bill of quantities of hydropower projects.

[0037] like Figure 2 As shown, the application software includes an interface presentation layer, a business logic layer, a basic business layer, a system component layer, and data storage, and implements the method of Example 1 through layer-by-layer calling, processing and analysis of data.

[0038] In summary, the present invention has been described in detail with reference to specific embodiments. Those skilled in the art will understand that various modifications and changes may be made thereon. As long as they do not depart from the purpose and spirit of the present invention, these modifications and changes should fall within the scope of protection of the present invention, and the scope of protection of the present invention is defined by the appended claims.

Claims

1. Intelligent standardization method for hydropower project quantity list, characterized by: The specific steps include: The first step is to build an intelligent standardization system for hydropower project quantity list. The system includes a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A. Through a locally preset model, the qualification parameters provided by Party B are input and a classification result is given as to whether the qualification is met. If the qualification is met, the second to fourth steps are executed: The second step is for Party A to prepare the standard requirement form and the special format for applying for the bill of quantities on the host computer and save it on the server. In the third step, the qualified Party B installs the upper computer authorization of the hydropower bill of quantities standardization application software, obtains the authorization function for the bill form in the hydropower bill of quantities standardization application software on the Party B host, and uses this authorization function to call the standard requirement form and fill it out using the cost software and bill software; In the fourth step, Party B uses the authorization function in Party B's host computer to retrieve the manually completed standard requirement form through the natural language document automatic keyword extraction based on image recognition, complete the retrieved special format to apply for the bill of quantities, and submit it to the ECP system; If not, perform the following steps: S1: The design unit uses cost software, bill of quantities software, etc. to prepare the tender bill of quantities in standard Excel format; S2: The design unit uses the standardized application software for hydropower bill of quantities to verify the standardization of the form. Once the verification is qualified, the standard Excel format tender bill of quantities is converted into a special format tender bill of quantities and uploaded as an attachment to the ECP. After the technical specification is generated, other work such as review of the bidding documents for submission is carried out; S3: After the tender is issued, the applicant downloads the tender document package containing the tender bill of quantities and uses the hydropower bill of quantities standardization application software to convert the tender bill of quantities in the special format into a standard Excel format bill of quantities; S4: The applicant shall use the standard Excel format for bidding bill of quantities and use cost estimation software, bill of quantities software, etc. to prepare the standard Excel format for application bill of quantities; S5: The applicant uses the standardized application software for hydropower bill of quantities to verify the form and identify any discrepancies. Once the verification is qualified, the applicant converts the standard Excel format bill of quantities application into a special format bill of quantities application and submits it to the ECP; Step 5. This step follows Step 4 and S5 respectively, depending on whether the qualification classification is available. The bid evaluation expert uses the bidding software to receive special bidding documents and bid opening list information from the ECP, generates a bid evaluation environment, and uses the bid evaluation software to receive a special format application bill of quantities from the ECP for review. The comparison results, difference prompts, and mutually exclusive relationship reviews are obtained, and the expert is assisted in the bid evaluation, and the rejection / scoring situation is returned to the ECP.

2. The method according to claim 1, characterized in that The preset model includes a multidimensional support vector machine, and its training method includes: Party A collects the actual qualification parameters of Party B, and statistically analyzes the actual qualification parameters to produce training parameters and verification parameters, and sets labels for the training parameters and verification parameters as having and not having qualifications, constructs a multidimensional support vector machine, trains using the training parameters, and verifies using the verification parameters, optimizes the hyperplane, and stops training until the classification accuracy is maximized.

3. The method according to claim 1, characterized in that The qualification parameters include the number of projects accepted over the years, project investment amount, project completion evaluation level, project personnel's education level, length of service, and annual turnover rate.

4. The method according to claim 3, characterized in that The statistical analysis adopts normal distribution function fitting, and takes the qualification parameter distributed between positive and negative standard deviations σ.

5. The method according to any one of claims 1 to 4, characterized in that The method for automatically extracting keywords from a natural language document based on image recognition and completing the application for a bill of quantities in a special format includes the following steps: Step I: Take a screenshot of the manually filled standard requirement form and establish a mapping relationship between the filled-in items in the standard requirement form and the corresponding items in the special format application bill of quantities; Step II: Use the preset natural language model to identify keywords in the screenshot and retrieve them; Step III: Call the special format to apply for the bill of quantities, and use the mapping to map each keyword to the special format to complete the bill of quantities application.

6. The method according to claim 5, characterized in that In step II, a preset natural language model is used to identify keywords in the screenshot. The specific process is also a screenshot of the standard requirement form. The entry subgraph is extracted according to the calibrated spatial coordinates. The trained feedforward neural network language model is used to identify the documents in the subgraph and convert them into the word prediction results with the highest probability in the corresponding vocabulary library to form keywords with standard expressions.

7. The method according to claim 6, characterized in that Natural language model construction methods include: Q1 Establish a two-dimensional spatial coordinate system in the standard requirement form and calibrate the spatial coordinates according to the position of the filled-in items; Q2 collects screenshots of multiple standard requirement forms filled out by Party B, extracts the filled-in item sub-graphs according to the calibrated spatial coordinates, and builds a vocabulary library for each sub-graph; Q3 builds a feedforward neural network language model to identify the documents in the subgraph and convert them into the word prediction results with the highest probability in the corresponding vocabulary library to form keywords with standard expressions.

8. The intelligent standardization system for hydropower project quantity list is characterized by: It includes a server, a host computer, and a host computer of Party B. The host computer is located locally at Party A and connected to the server. The host computer of Party B communicates with the server. The host computer is installed with standardized application software for the bill of quantities of hydropower engineering projects. The application software includes an interface presentation layer, a business logic layer, a basic business layer, a system component layer, and data storage. Through layer-by-layer calling, processing and analysis of data, the method as claimed in any one of claims 1 to 7 is implemented.

Images