Scientific and technological project data management method, system and medium for construction method revision process

Abstract

The present invention discloses a scientific and technological project data management method, system and medium for the construction method revision process; it relates to the technical field of data management; according to the same-period rule change data and revision process data, this solution trains a construction method revision prediction model to realize the construction method revision approval prediction based on scientific and technological project data, so as to guide the revision process of the target construction method; this solution organically combines scientific and technological project data management with the construction method revision and approval processes, improving the quality and traceability of construction method revision; on the other hand, this solution also inputs the same-period rule change data and the preprocessed revision process data into a machine learning network to train a construction method revision prediction model, so as to analyze historical construction method revision data, predict the success rate of the current construction method revision and possible problems, provide intelligent suggestions for the revisers, monitor the revision process in real time, and timely remind the revisers of possible omissions.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and particularly to a data management method, system and medium for scientific and technological projects in the process of construction method revision. Background Art

[0002] In the traditional process of scientific and technological project management, the data management of scientific and technological projects and the management of scientific research achievements (such as construction method revision and construction method approval) are usually two independent links. The data management of scientific and technological projects mainly involves stages such as project establishment, implementation, and acceptance, covering a large amount of information such as technical data, experimental data, design drawings, and construction records; while the management of scientific research achievements focuses on the declaration, revision, and approval of achievements such as construction methods, patents, and papers generated by the project. This separated management mode brings many problems in actual operation. Especially in the process of construction method revision or construction method approval, it is often necessary to reorganize and extract relevant data in scientific and technological projects, resulting in low work efficiency, increased repetitive labor, and even possible impact on the revision quality and approval efficiency of construction methods due to data omission or error.

[0003] Firstly, the data of scientific and technological projects are usually scattered in different management systems or documents, lacking unified integration and standardized management. For example, technical data such as construction techniques, material parameters, and equipment configurations may be stored in different folders, databases, or paper documents, and key information needs to be extracted from these scattered data for construction method revision or approval. This data extraction process is not only time-consuming and laborious but also prone to incomplete or inaccurate data due to human negligence, thus affecting the revision quality and approval progress of construction methods.

[0004] Secondly, there is often information asymmetry between the data of scientific and technological projects and the requirements of construction method revision or approval. Construction method revision or approval has high requirements for the standardization, integrity, and timeliness of data, while the data management of scientific and technological projects may not be optimized for these requirements. For example, construction method revision requires detailed technical scheme descriptions, process improvement explanations, and experimental data support, while the data records of scientific and technological projects may lack systematic arrangement of these contents, resulting in repeated review and modification during revision, increasing the workload and time cost.

[0005] In addition, the traditional management mode may also lead to delays in construction method revision and approval. Since the data management of scientific and technological projects is disconnected from the construction method revision and approval processes, additional time is often required for data sorting and revision preparation after the project is completed, which not only extends the update cycle of construction methods but may also miss some important approval windows. For example, during the construction method approval process, the approval department may require the provision of the latest technical data or experimental verification results, and if these data are not sorted and updated in a timely manner, it may lead to approval delays or even rejection.

[0006] More importantly, the revision process of the construction method requires repeated review and modification of relevant technical data, which is extremely inefficient under the traditional management model. For example, the revision staff may need to review different versions of technical documents or experimental records multiple times to ensure the accuracy and consistency of the revised content. This repetitive work not only increases the workload, but may also lead to confusion in data versions, further affecting the quality and traceability of the construction method.

[0007] In summary, the traditional management mode of independent management of scientific and technological project data management and revision and approval of construction methods not only increases the workload, but also may lead to data quality problems and low approval efficiency. Therefore, there is an urgent need for a management method or system that can organically combine scientific and technological project data management with revision and approval processes of construction methods, so as to improve data utilization efficiency, reduce duplication of work, and ensure the timeliness and accuracy of revision and approval of construction methods. Summary of the invention

[0008] The technical problem to be solved by the present invention is that the traditional management mode of scientific and technological project data management and construction method revision and approval is independent of each other, which not only increases the workload, but also may lead to data quality problems and low approval efficiency; the purpose of the present invention is to provide a scientific and technological project data management method, system and medium for the construction method revision process, organically combine scientific and technological project data management with construction method revision and approval process, and improve the quality and traceability of construction method revision; on the other hand, this scheme also inputs the rule change data of the same period and the pre-processed revision process data into the machine learning network to train a construction method revision prediction model, so as to analyze the historical construction method revision data, predict the success rate of the current construction method revision and the problems that may be encountered, and provide intelligent suggestions for revision personnel, monitor the revision process in real time, and promptly remind revision personnel of possible omissions.

[0009] The present invention is achieved through the following technical solutions:

[0010] This solution provides a scientific and technological project data management method for the revision process of the construction method, including:

[0011] Collect the revision process data of historical construction methods and the data of rule changes during the same period, and pre-process the revision process data; the revision process data includes: revision data of construction methods and revision verification results;

[0012] Input the concurrent rule change data and the pre-processed revision process data into the machine learning network to train a method revision prediction model;

[0013] Obtain the revision data of the target construction method and the rule change data of the same period from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain the revision and approval prediction results;

[0014] The revision process of the target method is guided based on current science and technology project data and revised and approved forecast results.

[0015] A further optimization solution is that the preprocessing of the revised process data includes the following methods:

[0016] Extract a basic feature set related to the construction method approval result from the revised process data;

[0017] Screen out multiple feature sets from the basic feature set based on different screening models;

[0018] Combine multiple feature sets to obtain a total feature set;

[0019] Delete the duplicate features in the total feature set, and then calculate the correlation coefficient between each feature in the total feature set and the passing of the construction method revision review. If the correlation coefficient calculated for the current feature does not reach the evaluation threshold, delete the current feature.

[0020] A further optimization solution is that the calculation method of the correlation coefficient between each feature in the total feature set and the passing of the construction method revision review includes:

[0021] The correlation coefficient r between each feature and the passing of the construction method revision review is calculated according to the following formula:

[0022] ;

[0023] where n represents the total number of times of construction method revision; X k represents the k-th revised construction method; Y k represents the review result of the k-th revised construction method; represents the expectation of the k-th revised construction method; represents the expectation of the review result of the k-th revised construction method.

[0024] A further optimization solution is that inputting the preprocessed revised process data into a machine learning network to train a construction method revision prediction model; includes the following methods:

[0025] Conduct quality inspection on the total feature set based on the synchronous rule change data, and construct inspection features;

[0026] After integrating the inspection features and the total feature set, divide them into a test set and a training set;

[0027] Input the test set and the training set into the XGboost model to train a construction method revision prediction model, and update the parameters of the XGboost model based on the optimization algorithm during the training process.

[0028] A further optimization solution is that conducting quality inspection on the total feature set based on the synchronous rule change situation, and constructing inspection features; includes the following methods:

[0029] Extract all rule conditions from the synchronous rule change situation;

[0030] Search for the matching features of each rule condition in the feature set to generate the matching feature sets of each rule condition;

[0031] Determine the quality of the total feature set according to the matching feature sets of each rule condition: Determine whether the following formula holds:

[0032] ;

[0033] If it holds, determine that the quality of the total feature set is excellent; otherwise, determine that the quality of the total feature set is poor; where m represents the total number of features in the feature set; n represents the total number of rule conditions; S i represents the number of matching features in the matching feature set of the i-th rule condition; c represents the number of matching feature sets with the number of matching features less than c0; c0 represents the first matching feature number threshold; d represents the number of matching feature sets with the number of matching features less than d0; d0 represents the second matching feature number threshold; c0 > d0;

[0034] Construct test features based on the total feature set with poor quality.

[0035] A further optimization scheme is that the constructing test feature set according to the total feature set with poor quality includes the method:

[0036] Obtain all the rule conditions corresponding to the total feature set with poor quality to form a difference rule condition set, and a supplementary feature set; the supplementary feature set is the difference set between the basic feature set and the total feature set;

[0037] Perform a goodness-of-fit test on the supplementary features in the supplementary feature set and each difference rule condition:

[0038] The goodness-of-fit test value X of supplementary feature i and structured rule j ij is calculated according to the following formula:

[0039] ;

[0040] where N represents the goodness-of-fit coefficient; V ij represents the string matching degree between supplementary feature i and structured rule j; Z ij represents the cosine similarity between supplementary feature i and structured rule j;

[0041] If the goodness-of-fit test value of basic feature i and difference rule condition j exceeds the preset goodness-of-fit threshold Xe, then include basic feature i in the test feature set.

[0042] A further optimization scheme is that the updating the parameters of the XGboost model based on the optimization algorithm includes the method:

[0043] Taking the parameters of the XGboost model as the optimization object, initialize the improved whale optimization algorithm to calculate the fitness value of each whale, update the parameters of the XGboost model with the fitness values of each whale, train the XGboost model through the training set for construction method revision prediction, and then use the mean absolute percentage error of the construction method revision prediction result as the fitness value of each whale for iteration to determine the optimal solution; the parameters include tree depth, learning rate, sample weight, and subsample rate.

[0044] The further optimization plan is that the iteration method of the improved whale optimization algorithm includes:

[0045] Taking L as the surrounding step size:

[0046] ;

[0047] Among them, represents the convergence factor; b represents a random number between [0, 1]; represents the position vector of the prey; represents the position vector of the whale in the t-th iteration;

[0048] Among them, the convergence factor converges in a non-linear decreasing manner, and the decreasing form is:

[0049] a = 2 · [1 - cos(t / T max ) 2 · (π / 2);

[0050] a2 = (-1) · [1 - cos(t / T max ) 2 · (π / 2);

[0051] Among them, t represents the current iteration number; T max represents the maximum iteration number;

[0052] If , then update the position of the whale according to Equation , otherwise update the position of the whale according to Equation ; among them, represents the position vector of the whale in the (t + 1)-th iteration; represents the randomly selected position vector of the whale.

[0053] This solution also provides a scientific and technological project data management system for the construction method revision process, which is used to implement the above-mentioned scientific and technological project data management method for the construction method revision process; the system includes:

[0054] A preprocessing module, configured to collect the revision process data of historical construction methods and the contemporaneous rule change data, and preprocess the revision process data; the revision process data includes: construction method revision data and revision approval results;

[0055] A model training module, configured to input the contemporaneous rule change data and the preprocessed revision process data into a machine learning network to train a construction method revision prediction model;

[0056] A prediction module, configured to obtain the construction method revision data and the contemporaneous rule change data of a target construction method from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain a revision approval prediction result;

[0057] A guidance module, configured to guide the revision process of the target construction method based on the current scientific and technological project data and the revision approval prediction result.

[0058] A computer-readable medium of this solution, on which a computer program is stored, and when the computer program is executed by a processor, it can implement the scientific and technological project data management method for the construction method revision process as described above.

[0059] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0060] 1. The scientific and technological project data management method, system and medium for the construction method revision process provided by the present invention; this solution organically combines the scientific and technological project data management with the construction method revision and approval processes, improving the quality and traceability of the construction method revision; on the other hand, this solution also inputs the contemporaneous rule change data and the preprocessed revision process data into a machine learning network to train a construction method revision prediction model to analyze the historical construction method revision data, predict the success rate of the current construction method revision and possible problems, and provide intelligent suggestions for the revision personnel, monitor the revision process in real time, and timely remind the revision personnel of possible omissions;

[0061] 2. The scientific and technological project data management method, system and medium for the construction method revision process provided by the present invention; this solution effectively combines the advantages of different screening models based on a hybrid feature selection method, and relies on the powerful generalization ability of machine learning algorithms to achieve more accurate classification; at the same time, based on an optimization algorithm, the parameters of the XGboost model are updated to quickly and effectively find the optimal parameters and improve the performance of the XGboost model. Description of the Drawings

[0062] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the embodiments. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings. In the drawings:

[0063] Figure 1 It is a schematic flow chart of a scientific and technological project data management method for the construction method revision process;

[0064] Figure 2 It is a schematic diagram of a scientific and technological project data management system for the construction method revision process. Specific embodiments

[0065] To make the purpose, technical solutions and advantages of the present invention more clear, the following will further elaborate on the present invention in combination with the embodiments and the accompanying drawings. The illustrative embodiments of the present invention and their descriptions are only used to explain the present invention and do not limit the present invention.

[0066] In the traditional management mode where the data management of scientific and technological projects is independent of the construction method revision and approval, not only does it increase the workload, but it may also lead to data quality problems and low approval efficiency. In view of this, the following embodiments are provided in this solution to solve the above technical problems.

[0067] Embodiment 1: This embodiment provides a scientific and technological project data management method for the construction method revision process, as Figure 1 shown, including:

[0068] Step 1, collect the revision process data of the historical construction method and the synchronous rule change data, and preprocess the revision process data; the revision process data includes: construction method revision data and revision approval results;

[0069] In this step, the construction method revision data mainly includes: construction method revision technical parameters, revision times, interval time for each revision, approval process, departments involved in approval, revision approval opinions, revision records, and approval results for each revision, etc.; the synchronous rule change data mainly includes contemporaneous regulations revision documents, industry standard revision documents, and relevant standard revision documents formulated based on big data summary, etc.

[0070] The preprocessing of the revision process data includes the following methods:

[0071] S11, extract the basic feature set related to the construction method approval result from the revision process data;

[0072] S12. Select multiple feature sets from the basic feature set based on different screening models. The screening models include common ones such as random forest, mutual information method, L1 regularization method, and sequential backward selection method. Generally, three screening models are selected to construct three feature sets.

[0073] S13. Combine the multiple feature sets to obtain the total feature set.

[0074] S14. Delete the duplicate features in the total feature set, and then calculate the correlation coefficient between each feature in the total feature set and the passing of the construction method revision review. If the correlation coefficient calculated for the current feature does not reach the evaluation threshold, delete the current feature.

[0075] The calculation method of the correlation coefficient between each feature in the total feature set and the passing of the construction method revision review includes:

[0076] The correlation coefficient r between each feature and the passing of the construction method revision review is calculated according to the following formula:

[0077] ;

[0078] Among them, n represents the total number of construction method revisions; X k represents the k-th revised construction method; Y k represents the review result of the k-th revised construction method; represents the expectation of the k-th revised construction method; represents the expectation of the review result of the k-th revised construction method.

[0079] Step 2. Input the synchronous rule change data and the preprocessed revision process data into the machine learning network to train the construction method revision prediction model. This step specifically includes the following methods:

[0080] S21. Conduct quality inspection on the total feature set based on the synchronous rule change data and construct inspection features. This step specifically includes the following methods:

[0081] S211. Extract all rule conditions from the synchronous rule change situation.

[0082] S212. Search for the matching features of each rule condition in the feature set to generate the matching feature set of each rule condition.

[0083] S213. Determine the quality of the total feature set according to the matching feature set of each rule condition: Determine whether the following formula holds:

[0084] ;

[0085] If it holds, determine that the quality of the total feature set is excellent; otherwise, determine that the quality of the total feature set is poor. Among them, m represents the total number of features in the feature set; n represents the total number of rule conditions; Si represents the number of matching features in the matching feature set of the i-th rule condition; c represents the number of matching feature sets whose number of matching features is less than c0; c0 represents the first matching feature number threshold; d represents the number of matching feature sets whose number of matching features is less than d0; d0 represents the second matching feature number threshold; c0>d0; in this embodiment, c0 is 3 and d0 is 0;

[0086] S214, constructing a test feature set based on the total feature set with poor quality. This step specifically includes the following method:

[0087] S2141, obtaining all rule conditions corresponding to the total feature set with poor quality to form a poor rule condition set and a supplementary feature set; the supplementary feature set is a difference set between the basic feature set and the total feature set;

[0088] S2142, perform a goodness of fit test on the supplementary features in the supplementary feature set and each difference rule condition:

[0089] The goodness-of-fit test value X between supplementary feature i and structured rule j ij Calculated according to the following formula:

[0090] ;

[0091] Where N represents the goodness of fit coefficient; V ij represents the string matching degree between supplementary feature i and structured rule j; Z ij represents the cosine similarity between supplementary feature i and structured rule j;

[0092] S2143, if the goodness-of-fit test value between the basic feature i and the difference rule condition j exceeds the preset goodness threshold value Xe, the basic feature i is included in the test feature set;

[0093] S22, integrating the test feature set with the total feature set to divide them into a test set and a training set; combining the requirements of the rule change data of the same period to improve the accuracy of the features, so that the subsequent XGboost model can efficiently search for the best values ​​of these parameters, thereby improving the performance and accuracy of the model;

[0094] S23, input the test set and the training set into the XGboost model to train a method revision prediction model, and update the parameters of the XGboost model based on the optimization algorithm during the training process.

[0095] The method of updating the parameters of the XGboost model based on the optimization algorithm includes:

[0096] Taking the parameters of the XGboost model as the optimization object, initialize the improved whale optimization algorithm to calculate the fitness value of each whale, update the parameters of the XGboost model with the fitness values of each whale, train the XGboost model through the training set for construction method revision prediction, and then use the mean absolute percentage error of the construction method revision prediction result as the fitness value of each whale for iteration to determine the optimal solution; the parameters include tree depth, learning rate, sample weight, and subsample rate.

[0097] The iterative method of the improved whale optimization algorithm includes:

[0098] Taking L as the surrounding step size:

[0099] ;

[0100] Among them, represents the convergence factor; b represents a random number between [0, 1]; represents the position vector of the prey; represents the position vector of the whale in the t-th iteration;

[0101] Among them, the convergence factor converges in a non-linear decreasing manner, and the decreasing form is:

[0102] a = 2 · [1 - cos(t / T max ) 2 · (π / 2);

[0103] a2 = (-1) · [1 - cos(t / T max ) 2 · (π / 2);

[0104] Among them, t represents the current iteration number; T max represents the maximum iteration number;

[0105] If , then update the position of the whale according to Equation , otherwise update the position of the whale according to Equation ; among them, represents the position vector of the whale in the (t + 1)-th iteration; represents the randomly selected position vector of the whale.

[0106] The advantages of the XGboost model are its efficient computing power, powerful feature processing function, anti-overfitting performance, and excellent parallel design. However, its large number of parameters and complex initial settings are its major disadvantages; parameters such as tree depth, learning rate, sample weight, and sub-sample rate in the model will significantly affect the final performance. Although XGBoost contains regularization terms, overfitting may still occur if the hyperparameters are not set properly, especially on small data sets. Manually adjusting parameters is not only time-consuming, but also difficult to find the best combination. Therefore, the use of automated optimization algorithms helps to find the optimal parameters quickly and effectively, making parameter optimization a key factor affecting the performance of the algorithm. In order to ensure the effectiveness of the XGboost model, this solution uses an optimization algorithm to optimize the initial parameters.

[0107] This scheme is based on the improved whale optimization algorithm to optimize the initial parameters. The improved whale optimization algorithm improves the convergence factor for the prediction of the revision and approval results of the construction method, so that the convergence factor converges in a nonlinear decreasing manner, and can quickly and accurately predict the revision and approval results of the construction method.

[0108] Step 3: Obtain the revision data of the target construction method and the rule change data of the same period from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain the revision and approval prediction results;

[0109] Step 4: Guide the revision process of the target method based on current science and technology project data and revised approval forecast results.

[0110] This embodiment organically combines the management of scientific and technological project data with the revision and approval process of construction methods to improve the quality and traceability of construction method revisions; on the other hand, this solution also inputs the rule change data of the same period and the pre-processed revision process data into the machine learning network to train a construction method revision prediction model to analyze the historical construction method revision data, predict the success rate of the current construction method revision and possible problems, and provide intelligent suggestions for revision personnel, monitor the revision process in real time, and promptly remind revision personnel of possible omissions.

[0111] Embodiment 2: This embodiment provides a technology project data management system for the process of revision of the construction method, which is used to implement the technology project data management method for the process of revision of the construction method described in Embodiment 1; Figure 2 As shown, the system comprises:

[0112] A preprocessing module is used to collect revision process data of historical construction methods and concurrent rule change data, and to preprocess the revision process data; the revision process data includes: revision data of construction methods and revision verification results;

[0113] A model training module, which is used to input the synchronous rule change data and the preprocessed revision process data into a machine learning network to train a construction method revision prediction model;

[0114] A prediction module, which is used to obtain the construction method revision data and the synchronous rule change data of the target construction method from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain a revision approval prediction result;

[0115] A guidance module, which is used to guide the revision process of the target construction method based on the current scientific and technological project data and the revision approval prediction result.

[0116] Embodiment 3: This embodiment provides a computer-readable medium, on which a computer program is stored. When the computer program is executed by a processor, it can implement the scientific and technological project data management method for the construction method revision process as described in Embodiment 1, and specifically execute the following steps:

[0117] Step 1: Collect the revision process data and the synchronous rule change data of the historical construction method, and preprocess the revision process data; the revision process data includes: construction method revision data and revision approval results;

[0118] Step 2: Input the synchronous rule change data and the preprocessed revision process data into a machine learning network to train a construction method revision prediction model;

[0119] Step 3: Obtain the construction method revision data and the synchronous rule change data of the target construction method from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain a revision approval prediction result;

[0120] Step 4: Guide the revision process of the target construction method based on the current scientific and technological project data and the revision approval prediction result.

[0121] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A scientific and technological project data management method for the revision process of a construction method, characterized in that: include: Collect the revision process data of historical construction methods and the rule change data of the same period, and pre-process the revision process data; The revision process data includes: revision data of the construction method and revision verification results; the preprocessing of the revision process data includes: extracting a basic feature set related to the construction method verification results from the revision process data; filtering out multiple feature sets from the basic feature set based on different screening models; merging multiple feature sets to obtain a total feature set; deleting duplicate features in the total feature set, and then calculating the correlation coefficient between each feature in the total feature set and the revision verification of the construction method, and if the correlation coefficient calculated for the current feature does not reach the evaluation threshold, deleting the current feature; Inputting the concurrent rule change data and the pre-processed revision process data into the machine learning network to train a construction method revision prediction model; including the following method: performing quality inspection on the total feature set based on the concurrent rule change data and constructing inspection features; integrating the inspection features with the total feature set to divide them into a test set and a training set; inputting the test set and the training set into the XGboost model to train a construction method revision prediction model, and updating the parameters of the XGboost model based on the optimization algorithm during the training process; The method of performing quality inspection on the total feature set based on the concurrent rule change status and constructing inspection features includes: extracting all rule conditions from the concurrent rule change status; searching for matching features of each rule condition in the feature set to generate a matching feature set of each rule condition; judging the quality of the total feature set according to the matching feature set of each rule condition: judging whether the following formula is true: ; If it is established, the quality of the total feature set is judged to be excellent; otherwise, the quality of the total feature set is judged to be poor; where m represents the total number of features in the feature set; n represents the total number of rule conditions; S i represents the number of matching features in the matching feature set of the i-th rule condition; c represents the number of matching feature sets whose number of matching features is less than c0; c0 represents the first matching feature number threshold; d represents the number of matching feature sets whose number of matching features is less than d0; d0 represents the second matching feature number threshold; c0>d0; construct the inspection feature based on the total feature set with poor quality; Obtain the revision data of the target construction method and the rule change data of the same period from the current scientific and technological project data, and input them into the construction method revision prediction model to obtain the revision and approval prediction results; The revision process of the target method is guided based on current science and technology project data and revised and approved forecast results.

2. The scientific and technological project data management method for the process of revision of the construction method according to claim 1 is characterized in that: The calculation method of the correlation coefficient between each feature in the total feature set and the revision and approval of the working method includes: The correlation coefficient r between each characteristic and the revision and approval of the construction method is calculated according to the following formula: ； Where n represents the total number of revisions to the construction method; X k represents the kth revision of the construction method; Y k Indicates the review result of the kth revision of the construction method; represents the expectation of the kth revision of the construction method; Represents the expectation of the results of the k-th revision of the construction method review.

3. The scientific and technological project data management method for the process of revision of the construction method according to claim 1 is characterized in that: The method of constructing a test feature set based on a total feature set with poor quality includes: Acquire all rule conditions corresponding to the total feature set with poor quality to form a poor rule condition set and a supplementary feature set; the supplementary feature set is a difference set between the basic feature set and the total feature set; Convert all rule conditions into structured rules, and perform goodness of fit test between the supplementary features in the supplementary feature set and each structured rule: The goodness-of-fit test value X between supplementary feature i and structured rule j ij Calculated according to the following formula: ； Where N represents the goodness of fit coefficient; V ij represents the string matching degree between supplementary feature i and structured rule j; Z ij represents the cosine similarity between supplementary feature i and structured rule j; If the goodness-of-fit test value between the basic feature i and the difference rule condition j exceeds the preset goodness threshold Xe, the basic feature i is included in the test feature set.

4. The scientific and technological project data management method for the process of revision of the construction method according to claim 1 is characterized in that: The parameters of the XGboost model are updated based on the optimization algorithm; Included methods: The parameters of the XGboost model are taken as the optimization object, and the improved whale optimization algorithm is initialized to calculate the fitness value of each whale. The parameters of the XGboost model are updated with the fitness value of each whale. The XGboost model is trained with the training set to predict the revision of the working method. The mean absolute percentage error of the prediction result of the revision of the working method is used as the fitness value of each whale for iteration to determine the optimal solution. The parameters include tree depth, learning rate, sample weight and sub-sample rate.

5. The scientific and technological project data management method for the process of revision of the construction method according to claim 4 is characterized in that: The iterative methods of the improved whale optimization algorithm include: Take L as the bracketing step length: ;in, represents the convergence factor; b represents a random number between [0, 1]; represents the position vector of the prey; represents the position vector of the whale in the tth iteration; Among them, the convergence factor converges in a nonlinear decreasing manner, and the decreasing form is: a=2· [1-cos(t / T max ) 2 ] ·(π / 2); a2 =(-1)· [1-cos(t / T max ) 2 ] ·(π / 2); Where t represents the current iteration number; T max Indicates the maximum number of iterations; like , then according to the formula Update the whale's position, otherwise press Update the whale's position; where, represents the position vector of the whale in the t+1th iteration; represents a randomly selected whale position vector.

6. A scientific and technological project data management system used in the revision process of construction methods, characterized in that: A method for managing scientific and technological project data for a process of revising a construction method according to any one of claims 1 to 5; the system comprises: A preprocessing module is used to collect revision process data of historical construction methods and concurrent rule change data, and to preprocess the revision process data; the revision process data includes: revision data of construction methods and revision verification results; The model training module is used to input the concurrent rule change data and the pre-processed revision process data into the machine learning network to train the method revision prediction model; The prediction module is used to obtain the revision data of the target construction method and the rule change data of the same period from the current scientific and technological project data, and input it into the construction method revision prediction model to obtain the revision and approval prediction results; The guidance module is used to guide the revision process of the target method based on the current science and technology project data and the revision and approval forecast results.

7. A computer readable medium having a computer program stored thereon, characterized in that: The computer program is executed by a processor to implement the scientific and technological project data management method for the process of revising a construction method as described in any one of claims 1 to 5.

Images