A smart decision-making system and method for water conservancy and hydropower project construction management

Abstract

The present invention provides an intelligent decision-making system and method for water conservancy and hydropower project construction management, comprising a data layer, a platform layer, and an application layer. The application layer includes a safety management unit, a quality management unit, a progress management unit, and a decision-making recommendation unit. The present invention manages and comprehensively analyzes process data generated during the construction process, significantly improving the efficiency and targeted management of engineering projects. Furthermore, the system explores the relationships between process data such as safety, quality, progress, resource input, and images, and uses corresponding anomaly detection models to detect anomalies in safety data, quality data, and progress data. When an anomaly exists, the system constructs and uses corresponding analysis models to determine the cause of the anomaly and recommends decision-making recommendations for resolving the anomaly. This allows for timely optimization of resource allocation, improves resource utilization, and reduces project costs through quantitative control of resource input.

Description

Technical Field

[0001] The present invention belongs to the technical field of water conservancy and hydropower project construction management, and specifically relates to an intelligent decision-making system and method for water conservancy and hydropower project construction management. Background Art

[0002] Water conservancy and hydropower projects are massive, systematic projects characterized by long construction periods, extensive scope, high operational risks, numerous participating entities, and significant social impact. Currently, water conservancy and hydropower project construction management primarily utilizes database technology to manage various process data, including safety, quality, progress, and resource input data. This technology links each application unit to the project entity, enabling the digitization of the management process. Furthermore, BIM+GIS technology is used to display project entity and management process data in real time, enabling refined management of the project construction process.

[0003] Existing construction management systems use threshold comparison methods to detect anomalies in progress data within process data. When anomalies are detected, an early warning module alerts the responsible person to ensure that construction projects are completed on time and as planned. However, existing construction management systems primarily focus on process data management and presentation. Data analysis relies solely on simple comparisons for anomaly detection and early warning, without deeply exploring the relationships between process data. This results in monotonous and inaccurate anomaly detection. Furthermore, they lack guidance on how to resolve anomalies in process data, such as safety, quality, and progress, resulting in a poor user experience.

[0004] Therefore, there is an urgent need for an intelligent decision-making system for water conservancy and hydropower project construction management that can perform data analysis conveniently and accurately to improve user experience. Summary of the Invention

[0005] In view of the defects of the existing technology, the present invention provides an intelligent decision-making system for water conservancy and hydropower project construction management, which can effectively solve the above problems.

[0006] The technical solution adopted in the present invention is as follows:

[0007] The present invention provides a smart decision-making system for water conservancy and hydropower project construction management, including a data layer, a platform layer, and an application layer:

[0008] The data layer includes a model database and a process database; the model database is used to store multiple data anomaly detection models, multiple basic analysis models, and multiple model function components; wherein the data anomaly detection models at least include a safety data anomaly detection model, a quality data anomaly detection model, and a progress data anomaly detection model; the process database is used to store process data of each engineering unit, and the process data at least includes resource input data, safety data, progress data, quality data, and image data;

[0009] The application layer includes a security management unit, a quality management unit, a progress management unit and a decision suggestion unit;

[0010] The security management unit is used to call the security data and the security data anomaly detection model and output a security data anomaly label;

[0011] The quality management unit is used to call the quality data and the quality data anomaly detection model, and output a quality data anomaly label;

[0012] The progress management unit is used to call the progress data and the progress data anomaly detection model and output a progress data anomaly label;

[0013] The decision suggestion unit includes an anomaly coding module, a model building module, a decision generation module and a visualization module;

[0014] The anomaly encoding module is used to obtain the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label, and generate an anomaly label vector in the order of safety, quality, and progress;

[0015] The model building module is used to obtain the abnormal label vector and build a first analysis model according to the abnormal label vector, the basic analysis model and the model function component;

[0016] The decision generation module is used to call the first analysis model and the data to be analyzed, and generate a decision suggestion based on the first analysis model and the data to be analyzed;

[0017] The visualization module is used to output the decision suggestion.

[0018] Furthermore, constructing a first analysis model based on the abnormal label vector, the basic analysis model, and the model functional component specifically includes:

[0019] Obtaining a basic analysis model in the model database according to the abnormal label vector matching, wherein the basic analysis model includes at least one model function component;

[0020] Edit the basic analysis model through a user interface to obtain an initial analysis model;

[0021] A training data set is obtained from the model database; and the initial analysis model is trained using the training data set to obtain the first analysis model.

[0022] Furthermore, the basic analysis model is a template model pre-constructed according to the number of abnormal data types of safety data, quality data and progress data in the abnormal label vector;

[0023] The basic analysis model includes a first basic analysis model, a second basic analysis model and a third basic analysis model;

[0024] If only one type of data anomaly exists among the safety data, the quality data, and the progress data, then the first basic analysis model is matched; the first basic analysis model includes a first generation block, which includes a first encoding layer and a first decoding layer; the first encoding layer is used to receive the data to be analyzed corresponding to the abnormal type data, obtain a coding vector, and input it into the first decoding layer, and the first decoding layer analyzes the coding vector and outputs the cause of the anomaly and a decision recommendation;

[0025] If there are only two types of data anomalies among the safety data, the quality data, and the progress data, the second basic analysis model is matched; the second basic analysis model includes a 2-1 generation block, a 2-2 generation block, a 2-3 generation block, and a second fusion layer; the 2-1 generation block and the 2-2 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data to obtain a 2-1 encoding vector and a 2-2 encoding vector; the second fusion layer fuses and splices the 2-1 encoding vector and the 2-2 encoding vector to obtain a fused splicing vector, and inputs the fused splicing vector into the 2-3 generation block; the 2-3 generation block analyzes the fused splicing vector and outputs the cause of the anomaly and a decision recommendation;

[0026] If all three types of data among the safety data, the quality data and the progress data are abnormal, the third basic analysis model is matched; the third basic analysis model includes a 3-1 generation block, a 3-2 generation block, a 3-3 generation block, a 3-4 generation block and a 3rd fusion layer; the 3-1 generation block, the 3-2 generation block and the 3-3 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data, and obtain a 3-1 encoding vector, a 3-2 encoding vector and a 3-3 encoding vector; the 3rd fusion layer fuses and splices the 3-1 encoding vector, the 3-2 encoding vector and the 3-3 encoding vector to obtain a fused splicing vector, and inputs it into the 3-4 generation block; the 3-4 generation block analyzes the fused splicing vector and outputs the cause of the abnormality and decision recommendations.

[0027] Furthermore, the model functional components include a fusion layer, a sharing layer, and encoding layers, decoding layers, and generation blocks of different network structures;

[0028] The editing of the basic analysis model through the user interface to obtain the initial analysis model specifically includes: obtaining the process data type and quantity that need to be associated with the abnormal data type for analysis in the data to be analyzed based on the abnormal data type in the abnormal label vector, thereby determining the model function components that need to be recommended, and recommending the model function components to the user interface; based on the user interface, selecting and / or calling the recommended model function components; integrating the selected and / or called model function components into the basic analysis model to obtain the initial analysis model, so that the initial analysis model has a model framework for comprehensive analysis of the data to be analyzed.

[0029] Furthermore, the decision generation module calls the data to be analyzed, specifically including:

[0030] The decision generation module obtains the data to be analyzed from the engineering database by matching based on the abnormal label vector. When the abnormal label vector indicates that there is an abnormality in the safety data, the data to be analyzed includes safety data, resource input data and image data; when the abnormal label vector indicates that there is an abnormality in the quality data, the data to be analyzed includes quality data, resource input data and image data; when the abnormal label vector indicates that there is an abnormality in the progress data, the data to be analyzed includes progress data, resource input data and image data.

[0031] Furthermore, the security management unit includes a security data statistical analysis module, a security data anomaly detection module and a security data visualization module;

[0032] The safety data statistical analysis module is used to collect statistics and analyze the safety data of the engineering unit within a preset time period;

[0033] The security data anomaly detection module is used to call the security data anomaly detection model in the model database, perform anomaly detection on the security data at a preset time point or within a preset time period, and determine whether the security data has an anomaly; the input of the security data anomaly detection model is security data and image data;

[0034] The safety data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of safety data and safety abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display safety data, safety abnormality data, and statistical analysis results of safety data and / or safety abnormality data.

[0035] Furthermore, the quality management unit includes a quality data statistical analysis module, a quality data anomaly detection module and a quality data visualization module;

[0036] The quality data statistical analysis module is used to collect and analyze the quality data of the engineering unit within a preset time period;

[0037] The quality data anomaly detection module is used to call the quality data anomaly detection model in the model database, perform anomaly detection on the quality data at a preset time point or within a preset time period, and determine whether the quality data has an anomaly; the input of the quality data anomaly detection model is quality data and image data;

[0038] The quality data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of quality data and quality abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display quality data, quality abnormality data, and statistical analysis results of quality data and / or quality abnormality data.

[0039] Furthermore, the progress management unit includes a progress data statistical analysis module, a progress data anomaly detection module and a progress data visualization module;

[0040] The progress data statistical analysis module is used to collect statistics and analyze the progress data of the engineering unit within a preset time period;

[0041] The progress data anomaly detection module is used to call the progress data anomaly detection model in the model database, perform anomaly detection on the progress data at a preset time point or a preset time period, and determine whether the progress data has an anomaly; the input of the progress data anomaly detection model is the progress data and the image data;

[0042] The progress data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the progress data and progress exception data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display the progress data, progress exception data, and statistical analysis results of the progress data and / or progress exception data.

[0043] Furthermore, the application layer further includes a collection unit; the collection unit at least includes a first collection module, a second collection module and a third collection module;

[0044] The first acquisition module is used to collect engineering data of the engineering unit, and store the collected engineering data in a directory corresponding to the engineering unit database after adding an engineering data identifier. The engineering data at least includes the engineering location, engineering construction content, and engineering quantity.

[0045] The second acquisition module is used to collect process data of each engineering unit at different time points according to a preset period, and add a process data identifier to the collected process data and store it in the process database;

[0046] The third acquisition module is used to collect model data and store the model data in a model database, wherein the model data at least includes model function components, training data sets and model parameters;

[0047] The application layer also includes a project management unit;

[0048] The project management unit is used to edit and visualize the engineering project according to the engineering unit; the project management unit includes at least an editing module and a visualization module; the editing module is used to edit the engineering data corresponding to the engineering unit; the visualization module includes a two-dimensional visualization sub-module and a three-dimensional visualization sub-module, and the three-dimensional visualization sub-module uses the BIM model corresponding to the engineering unit to perform a three-dimensional visualization display of the engineering unit; the two-dimensional visualization sub-module uses two-dimensional charts and / or text to display the engineering information and statistical analysis results of the engineering unit.

[0049] The present invention also provides a method for applying the aforementioned intelligent decision-making system for water conservancy and hydropower project construction management, comprising the following steps:

[0050] Step S1, obtaining engineering data of each engineering unit in a water conservancy and hydropower engineering construction project, adding engineering data identifiers to the engineering data, and storing them in an engineering unit database;

[0051] Step S2: During the construction of the water conservancy and hydropower project, process data of each engineering unit at different time points is collected according to a preset period, and the collected process data is added with a process data identifier and stored in a process database; the process data includes at least resource input data, safety data, progress data, quality data, and image data;

[0052] Step S3, reading the safety data, the quality data, and the progress data stored in the most recent time period from the process database according to a preset analysis cycle;

[0053] Calling the safety data anomaly detection model to analyze the safety data and output a safety data anomaly label; calling the quality data anomaly detection model to analyze the quality data and output a quality data anomaly label; calling the progress data anomaly detection model to analyze the progress data and output a progress data anomaly label;

[0054] Step S4, sequentially concatenating the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label to generate an anomaly label vector;

[0055] Step S5, according to the abnormal label vector, matching and obtaining the data to be analyzed from the engineering unit database: when the abnormal label vector indicates that there is an abnormality in the safety data, the data to be analyzed includes the safety data, resource input data and image data of the corresponding time period; when the abnormal label vector indicates that there is an abnormality in the quality data, the data to be analyzed includes the quality data, resource input data and image data of the corresponding time period; when the abnormal label vector indicates that there is an abnormality in the progress data, the data to be analyzed includes the progress data, resource input data and image data of the corresponding time period.

[0056] Step S6: pre-establishing a model database; the model database stores a plurality of basic analysis models, a plurality of model function components, a training data set, and model parameters;

[0057] According to the number of abnormal labels in the abnormal label vector, the corresponding basic analysis model is matched from the model database; according to the type and number of process data that need to be analyzed in association with the abnormal data in the data to be analyzed, the model function components to be recommended are determined, and the model function components are recommended to the user interface;

[0058] Integrating the model functional components selected by the user into the basic analysis model to obtain an initial analysis model;

[0059] Obtaining a training data set from the model database; training the initial analysis model using the training data set, adjusting model parameters in the initial analysis model, and obtaining a trained first analysis model;

[0060] Step S7: Analyze the data to be analyzed using the first analysis model to generate decision recommendations.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] (1) The intelligent decision-making system for water conservancy and hydropower project construction management constructed by the present invention manages and comprehensively analyzes the process data generated during the project construction process, evaluates the project construction process from the three dimensions of safety, quality, and progress, and mines the correlation between process data such as safety, quality, progress, resource input and images based on the data of these three dimensions, providing process decision-making suggestions for project construction, greatly improving management efficiency and targeted management.

[0063] (2) The present invention uses anomaly vectors constructed using safety data, quality data, and progress data to construct or match corresponding analysis models. The causes of anomalies and decision-making suggestions for resolving anomalies are obtained through the analysis model, thereby improving the accuracy of decision-making, optimizing resource allocation in a timely manner, improving resource utilization, and reducing engineering costs through quantitative control of resource inputs.

[0064] (3) The present invention recommends a basic analysis model through anomaly vectors, and users can select corresponding functional components through the interface to modify the basic analysis model, thereby improving the scalability of the analysis model and further improving the accuracy of decision-making and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 A schematic diagram of an intelligent decision-making system for water conservancy and hydropower project construction management provided by an embodiment of the present invention;

[0066] Figure 2 A schematic diagram of a security management unit provided in an embodiment of the present invention;

[0067] Figure 3 A schematic diagram of a quality management unit provided in an embodiment of the present invention;

[0068] Figure 4 A schematic diagram of a progress management unit provided in an embodiment of the present invention;

[0069] Figure 5 A schematic diagram of a decision suggestion unit provided in an embodiment of the present invention;

[0070] Figure 6 A schematic diagram of a decision suggestion unit provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0071] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0072] See Figure 1 The present invention provides a water conservancy and hydropower project construction management intelligent decision-making system, including a data layer, a platform layer and an application layer:

[0073] (1) Data layer

[0074] The data layer includes a model database and a process database; the model database is used to store multiple data anomaly detection models, multiple basic analysis models and multiple model functional components; wherein, the data anomaly detection model includes at least a safety data anomaly detection model, a quality data anomaly detection model and a progress data anomaly detection model; the process database is used to store the process data of each engineering unit, and the process data includes at least resource input data, safety data, progress data, quality data and image data.

[0075] (2) Platform layer

[0076] The platform layer obtains the data to be displayed from the engineering unit database and / or process database, and sends the data to be displayed to the PC or mobile client for display; the platform layer obtains the data to be analyzed from the engineering unit database and / or process database, calls the data analysis model and / or builds the data analysis model, performs data analysis on the data to be analyzed to obtain data analysis results, and sends the data analysis results to the PC or mobile client for display.

[0077] (3) Application layer

[0078] The application layer includes a security management unit, a quality management unit, a progress management unit and a decision suggestion unit, as well as a collection unit and a project management unit.

[0079] (3.1) Acquisition unit

[0080] The acquisition unit at least includes a first acquisition module, a second acquisition module and a third acquisition module;

[0081] The first acquisition module is used to collect engineering data of the engineering unit, and store the collected engineering data in a directory corresponding to the engineering unit database after adding an engineering data identifier. The engineering data at least includes the engineering location, engineering construction content, and engineering quantity.

[0082] The second acquisition module is used to collect process data of each engineering unit at different time points according to a preset period, and add a process data identifier to the collected process data and store it in the process database;

[0083] The third acquisition module is used to acquire model data and store the model data in a model database. The model data at least includes model functional components, training data sets and model parameters.

[0084] (3.2) Security Management Unit

[0085] The security management unit is used to call the security data and the security data anomaly detection model and output a security data anomaly label;

[0086] The security management unit includes a security data statistical analysis module, a security data anomaly detection module and a security data visualization module;

[0087] The safety data statistical analysis module is used to collect statistics and analyze the safety data of the engineering unit within a preset time period;

[0088] The security data anomaly detection module is used to call the security data anomaly detection model in the model database, perform anomaly detection on the security data at a preset time point or within a preset time period, and determine whether the security data has an anomaly; the input of the security data anomaly detection model is security data and image data;

[0089] The safety data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of safety data and safety abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display safety data, safety abnormality data, and statistical analysis results of safety data and / or safety abnormality data.

[0090] (3.3) Quality Management Unit

[0091] The quality management unit is used to call the quality data and the quality data anomaly detection model, and output a quality data anomaly label;

[0092] The quality management unit includes a quality data statistical analysis module, a quality data anomaly detection module and a quality data visualization module;

[0093] The quality data statistical analysis module is used to collect and analyze the quality data of the engineering unit within a preset time period;

[0094] The quality data anomaly detection module is used to call the quality data anomaly detection model in the model database, perform anomaly detection on the quality data at a preset time point or within a preset time period, and determine whether the quality data has an anomaly; the input of the quality data anomaly detection model is quality data and image data;

[0095] The quality data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of quality data and quality abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display quality data, quality abnormality data, and statistical analysis results of quality data and / or quality abnormality data.

[0096] (3.4) Progress Management Unit

[0097] The progress management unit is used to call the progress data and the progress data anomaly detection model and output a progress data anomaly label;

[0098] The progress management unit includes a progress data statistical analysis module, a progress data anomaly detection module and a progress data visualization module;

[0099] The progress data statistical analysis module is used to collect statistics and analyze the progress data of the engineering unit within a preset time period;

[0100] The progress data anomaly detection module is used to call the progress data anomaly detection model in the model database, perform anomaly detection on the progress data at a preset time point or a preset time period, and determine whether the progress data has an anomaly; the input of the progress data anomaly detection model is the progress data and the image data;

[0101] The progress data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the progress data and progress exception data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display the progress data, progress exception data, and statistical analysis results of the progress data and / or progress exception data.

[0102] (3.5) Decision-making suggestion unit

[0103] The decision suggestion unit includes an anomaly coding module, a model building module, a decision generation module and a visualization module.

[0104] (3.5.1) Exception Coding Module

[0105] The anomaly encoding module is used to obtain the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label, and generate an anomaly label vector in the order of safety, quality, and progress;

[0106] (3.5.2) Model building module

[0107] The model building module is used to obtain the abnormal label vector and build a first analysis model according to the abnormal label vector, the basic analysis model and the model function component;

[0108] The model building module is specifically used to:

[0109] A basic analysis model in the model database is obtained according to the abnormal label vector matching, and the basic analysis model includes at least one model functional component; the basic analysis model is edited through a user interface to obtain an initial analysis model; a training data set in the model database is obtained; and the initial analysis model is trained using the training data set to obtain the first analysis model.

[0110] The specific process of obtaining the basic analysis model in the model database according to the abnormal label vector matching is as follows:

[0111] The basic analysis model is a template model pre-constructed according to the number of abnormal data types of safety data, quality data and progress data in the abnormal label vector;

[0112] The basic analysis model includes a first basic analysis model, a second basic analysis model and a third basic analysis model;

[0113] If only one type of data anomaly exists among the safety data, the quality data, and the progress data, then the first basic analysis model is matched; the first basic analysis model includes a first generation block, which includes a first encoding layer and a first decoding layer; the first encoding layer is used to receive the data to be analyzed corresponding to the abnormal type data, obtain a coding vector, and input it into the first decoding layer, and the first decoding layer analyzes the coding vector and outputs the cause of the anomaly and a decision recommendation;

[0114] If there are only two types of data anomalies among the safety data, the quality data, and the progress data, the second basic analysis model is matched; the second basic analysis model includes a 2-1 generation block, a 2-2 generation block, a 2-3 generation block, and a second fusion layer; the 2-1 generation block and the 2-2 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data to obtain a 2-1 encoding vector and a 2-2 encoding vector; the second fusion layer fuses and splices the 2-1 encoding vector and the 2-2 encoding vector to obtain a fused splicing vector, and inputs the fused splicing vector into the 2-3 generation block; the 2-3 generation block analyzes the fused splicing vector and outputs the cause of the anomaly and a decision recommendation;

[0115] If all three types of data among the safety data, the quality data and the progress data are abnormal, the third basic analysis model is matched; the third basic analysis model includes a 3-1 generation block, a 3-2 generation block, a 3-3 generation block, a 3-4 generation block and a 3rd fusion layer; the 3-1 generation block, the 3-2 generation block and the 3-3 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data, and obtain a 3-1 encoding vector, a 3-2 encoding vector and a 3-3 encoding vector; the 3rd fusion layer fuses and splices the 3-1 encoding vector, the 3-2 encoding vector and the 3-3 encoding vector to obtain a fused splicing vector, and inputs it into the 3-4 generation block; the 3-4 generation block analyzes the fused splicing vector and outputs the cause of the abnormality and decision recommendations.

[0116] Furthermore, the basic analysis model is edited through the user interface to obtain the initial analysis model:

[0117] The model functional components include a fusion layer, a sharing layer, and encoding layers, decoding layers, and generation blocks of different network structures; based on the abnormal data type in the abnormal label vector, the process data type and quantity that need to be associated with the abnormal data type for analysis in the data to be analyzed are obtained, thereby determining the model functional components that need to be recommended, and recommending the model functional components to the user interface; based on the user interface, the recommended model functional components are selected and / or called; the selected and / or called model functional components are integrated into the basic analysis model to obtain the initial analysis model, so that the initial analysis model has a model framework for comprehensive analysis of the data to be analyzed.

[0118] (3.5.3) Decision Generation Module

[0119] The decision generation module is used to call the first analysis model and the data to be analyzed, and generate a decision suggestion based on the first analysis model and the data to be analyzed.

[0120] Specifically, the decision generation module obtains the data to be analyzed from the engineering database according to the abnormal label vector. When the abnormal label vector indicates that there is an abnormality in the safety data, the data to be analyzed includes safety data, resource input data and image data; when the abnormal label vector indicates that there is an abnormality in the quality data, the data to be analyzed includes quality data, resource input data and image data; when the abnormal label vector indicates that there is an abnormality in the progress data, the data to be analyzed includes progress data, resource input data and image data.

[0121] (3.5.4) Visualization module

[0122] The visualization module is used to visually output the decision suggestion.

[0123] (3.6) Project Management Unit

[0124] The project management unit is used to edit and visualize the engineering project according to the engineering unit; the project management unit includes at least an editing module and a visualization module; the editing module is used to edit the engineering data corresponding to the engineering unit; the visualization module includes a two-dimensional visualization sub-module and a three-dimensional visualization sub-module, and the three-dimensional visualization sub-module uses the BIM model corresponding to the engineering unit to perform a three-dimensional visualization display of the engineering unit; the two-dimensional visualization sub-module uses two-dimensional charts and / or text to display the engineering information and statistical analysis results of the engineering unit.

[0125] The present invention also provides a method for applying the water conservancy and hydropower project construction management intelligent decision-making system, comprising the following steps:

[0126] Step S1, obtaining engineering data of each engineering unit in a water conservancy and hydropower engineering construction project, adding engineering data identifiers to the engineering data, and storing them in an engineering unit database;

[0127] Step S2: During the construction of the water conservancy and hydropower project, process data of each engineering unit at different time points is collected according to a preset period, and the collected process data is added with a process data identifier and stored in a process database; the process data includes at least resource input data, safety data, progress data, quality data, and image data;

[0128] Step S3, reading the safety data, the quality data, and the progress data stored in the most recent time period from the process database according to a preset analysis cycle;

[0129] Calling the safety data anomaly detection model to analyze the safety data and output a safety data anomaly label; calling the quality data anomaly detection model to analyze the quality data and output a quality data anomaly label; calling the progress data anomaly detection model to analyze the progress data and output a progress data anomaly label;

[0130] Step S4, sequentially concatenating the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label to generate an anomaly label vector;

[0131] Step S5, according to the abnormal label vector, matching and obtaining the data to be analyzed from the engineering unit database: when the abnormal label vector indicates that there is an abnormality in the safety data, the data to be analyzed includes the safety data, resource input data and image data of the corresponding time period; when the abnormal label vector indicates that there is an abnormality in the quality data, the data to be analyzed includes the quality data, resource input data and image data of the corresponding time period; when the abnormal label vector indicates that there is an abnormality in the progress data, the data to be analyzed includes the progress data, resource input data and image data of the corresponding time period.

[0132] Step S6: pre-establishing a model database; the model database stores a plurality of basic analysis models, a plurality of model function components, a training data set, and model parameters;

[0133] According to the number of abnormal labels in the abnormal label vector, the corresponding basic analysis model is matched from the model database; according to the type and number of process data that need to be analyzed in association with the abnormal data in the data to be analyzed, the model function components to be recommended are determined, and the model function components are recommended to the user interface;

[0134] Integrating the model functional components selected by the user into the basic analysis model to obtain an initial analysis model;

[0135] Obtaining a training data set from the model database; training the initial analysis model using the training data set, adjusting model parameters in the initial analysis model, and obtaining a trained first analysis model;

[0136] Step S7: Analyze the data to be analyzed using the first analysis model to generate decision recommendations.

[0137] An embodiment is described below:

[0138] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0139] Reference Figure 1The present invention provides an intelligent decision-making system for water conservancy and hydropower project construction management, including a data layer, a platform layer and an application layer. The data layer includes an engineering unit database DB1, a process database DB2 and a model database DB3. The engineering unit database DB1 is used to store engineering data such as the engineering location, engineering construction content, and engineering quantity of each engineering unit according to a hierarchical directory of unit projects, sub-projects and unit projects, and add engineering data identifiers to the engineering data such as the engineering location, engineering construction content, and engineering quantity. The engineering data identifiers are obtained by hierarchically splicing the unit engineering code value, sub-project code value and unit engineering code value. For example, the engineering data identifier of a certain engineering data is 000100010002, which indicates that the unit engineering code value, sub-project code value and unit engineering code value corresponding to the engineering data are 0001, 0001 and 0002; the engineering data identifier of a certain engineering data is 000100030002, which indicates that the unit engineering code value, sub-project code value and unit engineering code value corresponding to the engineering data are 0001, 0003 and 0002. Engineering data identification is mainly used for data association between engineering unit database and process database.

[0140] The process database DB2 is used to store process data for each engineering unit at different time points. Process data includes at least resource input data, safety data, progress data, quality data, and image data. Resource input data includes personnel input data, machinery and equipment input data, and material input data. Safety data represents the safety hazards present in the engineering unit at that time point, including their type, level, and location. Progress data includes the project unit's planned schedule, actual progress, and completion percentage at that time point. Quality data includes the project unit's quality target, quality grade, and yield rate at that time point. Image data includes video surveillance images, drone images, and / or on-site photos of the project unit at that time point. Process data is stored in the process database in a hierarchical directory structured by unit project, sub-project, and unit project. The process data identifier is generated using a project data identifier and a timestamp. For example, process data with a process data identifier of 00010001000220241122 represents process data with a unit project code value of 0001, a sub-project code value of 0001, a unit project code value of 0002, and a time of 20241122.

[0141] The model database DB3 is used to store the BIM models of each engineering unit, as well as multiple data anomaly detection models, multiple basic analysis models, multiple historical analysis models, and model functional components for constructing analysis models. The model functional components include fusion layers, sharing layers, encoding layers of different network structures, decoding layers, and generation blocks. The encoding layers include LSTM encoding layers, CNN encoding layers, BILSTM encoding layers, GRN encoding layers, GNN encoding layers, GCN encoding layers, Attention+LSTM encoding layers, Attention+CNN encoding layers, Attention+BILSTM encoding layers, Attention+GRN encoding layers, Attention+GNN encoding layers, and Attention+GCN encoding layers. The decoding layers include LSTM decoding layers, CNN decoding layers, BILSTM decoding layers, GRN decoding layers, Attention+LSTM decoding layers, Attention+CNN decoding layers, Attention+BILSTM decoding layers, and Attention+GRN decoding layers.

[0142] The platform layer consists of a PC platform and a mobile platform with a B / S architecture. The platform layer retrieves the data to be displayed from the engineering unit database and / or process database and sends it to the PC or mobile client for display. The platform layer also retrieves the data to be analyzed from the engineering unit database and / or process database, invokes and / or constructs data analysis models, analyzes the data to obtain analysis results, and sends the results to the PC or mobile client for display.

[0143] The application layer includes the acquisition unit, project management unit, security management unit, quality management unit, progress management unit and decision-making recommendation unit.

[0144] The acquisition unit includes at least a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used to collect engineering data of the engineering unit, and add an engineering data identifier to the collected engineering data and store it in the directory corresponding to the engineering unit database. The engineering data at least includes the project location, construction content, and engineering quantity; the second acquisition module is used to collect process data of each engineering unit at different time points according to a preset period, and add a process data identifier to the collected process data and store it in the process database; the third acquisition module is used to collect model data and store the model data in the model database. The model data at least includes the model functional components, training data sets and model parameters required to construct the data analysis model.

[0145] The project management unit is used to edit and visualize the engineering project according to the engineering unit. It includes at least an editing module and a visualization module. The editing module is used to edit the engineering data corresponding to the engineering unit; the visualization module is used to visualize the engineering data. The visualization module includes a two-dimensional visualization sub-module and a three-dimensional visualization sub-module. The three-dimensional visualization sub-module uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the engineering unit. The two-dimensional visualization sub-module uses two-dimensional charts to display the engineering information and statistical analysis results of the engineering unit.

[0146] See also Figure 2 The security management unit includes a security data visualization module, a security data statistical analysis module, a security data anomaly detection module and a security warning module; wherein the security data statistical analysis module is used to perform statistics and analysis on the safety data of the engineering unit within a preset time period; the security data anomaly detection module is used to call the security data anomaly detection model in the model database, perform anomaly detection on the security data within a preset time point or a preset time period, and obtain whether there is an anomaly in the security data. When there is an anomaly, it is marked with a label "1", and when there is no anomaly, it is marked with a label "0"; the security data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule, wherein the three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the safety data and safety anomaly data, and the two-dimensional visualization submodule uses two-dimensional charts and / or text to display the safety data, safety anomaly data, and the statistical analysis results of the safety data and / or safety anomaly data; the security warning module is used to issue an early warning to the user when an anomaly is detected in the security data.

[0147] For example, the security data anomaly detection model is a pre-built deep neural network model. The security data anomaly detection module adopts the security data anomaly detection model to perform anomaly detection on the security data at a preset time point or within a preset time period to determine whether there is an anomaly in the security data. Specifically, it includes: inputting the image data into the deep neural network model, identifying and obtaining the actual security data, comparing the security data with the actual security data, and judging whether there is an anomaly in the security data based on the comparison results. Among them, the deep neural network model used for security data anomaly detection can adopt a CNN model or a CNN model based on the Attention mechanism.

[0148] Reference Figure 3The quality management unit includes a quality data visualization module, a quality data statistical analysis module, a quality data anomaly detection module and a quality early warning module; wherein the quality data statistical analysis module is used to perform statistics and analysis on the quality data of the engineering unit within a preset time period; the quality data anomaly detection module is used to call the quality data anomaly detection model in the model database, perform anomaly detection on the quality data within a preset time point or a preset time period, and obtain whether there is an anomaly in the quality data. When there is an anomaly, it is marked with a label "1", and when there is no anomaly, it is indicated with a label "0"; the quality data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule, wherein the three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the quality data and quality anomaly data, and the two-dimensional visualization submodule uses two-dimensional charts and / or text to display the quality data, quality anomaly data, and the statistical analysis results of quality data and / or quality anomaly data; the quality early warning module is used to warn the user when an anomaly in the quality data is detected.

[0149] For example, the quality data anomaly detection model is a pre-built deep neural network model. The quality data anomaly detection module adopts the quality data anomaly detection model to perform anomaly detection on the quality data within a preset time point or a preset time period to obtain whether there is an anomaly in the quality data. Specifically, it includes: inputting the quality data and image data into the deep neural network model, and outputting whether there is an anomaly in the quality data through the deep neural network model. Among them, the deep neural network model used for quality data anomaly detection can adopt LSTM+CNN+LSTM model, BILSTM+CNN+BILSTM model, etc.

[0150] Reference Figure 4 The progress management unit includes a progress data visualization module, a progress data statistical analysis module, a progress data anomaly detection module and a progress warning module. The progress data statistical analysis module is used to perform statistics and analysis on the progress data of the engineering unit within a preset time period; the progress data anomaly detection module is used to call the progress data anomaly detection model in the model database, perform anomaly detection on the progress data at a preset time point or a preset time period, and determine whether there is an anomaly in the progress data. When an anomaly exists, it is marked with a label "1", and when no anomaly exists, it is marked with a label "0"; the progress data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule, wherein the three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the progress data and progress anomaly data, and the two-dimensional visualization submodule uses two-dimensional charts and / or text to display the progress data, progress anomaly data, and the statistical analysis results of the progress data and / or progress anomaly data. The progress warning module is used to warn the user when an anomaly is detected in the progress data.

[0151] For example, the progress data anomaly detection model includes a pre-built deep neural network model. The progress data anomaly detection module adopts the progress data anomaly detection model to perform anomaly detection on the progress data within a preset time point or a preset time period to determine whether there is an anomaly in the progress data. Specifically, the model includes: inputting image data and progress data into the deep neural network model, and outputting whether there is an anomaly in the progress data through the deep neural network model. Among them, the deep neural network model used for progress data anomaly detection can adopt an LSTM+CNN+LSTM model, a BILSTM+CNN+BILSTM model, etc.

[0152] Reference Figure 5 ,The decision suggestion unit includes an anomaly coding module, a model ,construction module, a decision generation module and a visualization module.

[0153] The anomaly encoding module is used to obtain anomaly detection results for safety, quality, and progress data, and generates an anomaly label vector in the order of safety, quality, and progress. In the anomaly label vector, "1" indicates that the corresponding data has an anomaly, and "0" indicates that the corresponding data does not have an anomaly. For example, if anomalies are detected in safety and quality data but not in progress data, the anomaly label vector can be represented as (1,1,0). If anomalies are detected in all three data types, the anomaly label vector can be represented as (1,1,1).

[0154] The model construction module is used to obtain anomaly label vectors and construct a first analysis model based on the anomaly label vectors. Specifically, this includes: obtaining a basic analysis model from a model database based on the anomaly label vectors, the basic analysis model including at least one model functional component; editing the basic analysis model through a user interface to obtain an initial analysis model, wherein the editing includes at least one of adding, deleting, and replacing model functional components; obtaining a training dataset from the model database; and training the initial analysis model using the training dataset to obtain a first analysis model.

[0155] The model database stores multiple basic analysis models and multiple model functional components. Each basic analysis model is assigned a label. After obtaining an anomaly label vector, the model construction module first matches the anomaly label vector to the model database to obtain the corresponding basic analysis model. The basic analysis model is a pre-built template model based on the type and number of anomalies present in the safety data, quality data, and progress data. It includes at least one generation block, each of which includes at least one encoding layer and at least one decoding layer. When at least two of the safety data, quality data, and progress data contain anomalies, the basic analysis model also includes at least one fusion layer.

[0156] The basic analysis model is edited through the user interface to obtain the initial analysis model, specifically including: recommending model functional components to the user interface based on the abnormal label vector; selecting and / or calling model functional components based on the user interface; and integrating the selected and / or called model functional components into the basic analysis model to obtain the initial analysis model.

[0157] For example, when there is only one "1" in the abnormal label vector, that is, only one type of data among the safety data, quality data and progress data has an anomaly, the basic analysis model consists of a generation block including an encoding layer and a decoding layer; when there are two "1"s in the abnormal label vector, that is, two types of data among the safety data, quality data and progress data have anomalies, the basic analysis model consists of three generation blocks and a fusion layer, wherein the first generation block and the second generation block are used to receive the data to be analyzed corresponding to the abnormal label, and the fusion layer is used to fuse and splice the outputs of the first generation block and the second generation block, and input them to the third generation block, and the third generation block is used to generate decision recommendations; when there are three "1"s in the abnormal label vector, that is, three types of data among the safety data, quality data and progress data have anomalies, the basic analysis model consists of four generation blocks and a fusion layer, wherein the first generation block, the second generation block and the third generation block are used to receive the data to be analyzed corresponding to the abnormal label, and the fusion layer is used to fuse and splice the outputs of the first generation block, the second generation block and the third generation block, and input them to the fourth generation block, and the fourth generation block is used to generate decision recommendations.

[0158] After matching and obtaining the basic analysis model, the user can add, delete and / or modify the model functional components through the user interface to edit the basic analysis model and obtain the initial analysis model. Through the user interface, the generation blocks, fusion layers, etc. in the basic analysis model can be edited to obtain an analysis model that better suits the application scenario. When editing the basic analysis model, you can choose to edit the generation blocks as the basic functional units, such as increasing the number of generation blocks to increase the dimension of the data to be analyzed; you can also choose to edit the model functional components as the basic functional units, such as adding coding layers, decoding layers, fusion layers, shared layers, and / or replacing the types of coding layers, decoding layers, fusion layers, and shared layers.

[0159] After obtaining the initial analysis model, the training data set in the model database is called to train the initial analysis model until the preset conditions are met to obtain the analysis model.

[0160] The decision generation module is used to call the first analysis model and the data to be analyzed, and generate decision suggestions based on the first analysis model and the data to be analyzed. The decision suggestions include the causes of the anomalies and decision suggestions for resolving the anomalies. The causes of the anomalies include whether the construction is carried out in accordance with the relevant standards for safety and quality management, whether the construction is carried out in accordance with the construction plan, whether there are major changes in the geological conditions during the construction process, whether there are major design changes during the construction process, whether there are major changes in the construction environment, whether the number of personnel and personnel structure invested meet the requirements, whether the types and quantities of mechanical equipment invested meet the requirements, whether the types and quantities of materials invested meet the requirements, etc.; the decision suggestions include adjustments to the construction team, adjustments to the construction plan, adjustments to the construction process, adjustments to the personnel structure and number of personnel, adjustments to the types and quantities of mechanical equipment, adjustments to the types and quantities of materials, etc.

[0161] When there is an anomaly in the security data, the data to be analyzed includes security data, resource input data and image data; when there is an anomaly in the quality data, the data to be analyzed includes quality data, resource input data and image data; when there is an anomaly in the progress data, the data to be analyzed includes progress data, resource input data and image data.

[0162] For example, when it is detected that there is no abnormality in the security data and quality data, but there is an abnormality in the progress data, the generated abnormal label vector is (0, 0, 1). The first basic analysis model is obtained by matching the abnormal label vector. The first basic analysis model includes a first generation block. The first generation block includes a first LSTM encoding layer and a first LSTM decoding layer. The first CNN encoding layer and the first fusion layer are added to the first generation block through the user interface to obtain the first initial analysis model. The first initial analysis model includes a first LSTM encoding layer, a first CNN encoding layer, a first fusion layer and a first LSTM decoding layer.

[0163] After obtaining the first initial analysis model, the training data set is called to train the first initial analysis model, thereby obtaining the first analysis model.

[0164] Obtain the data to be analyzed, which includes progress data, resource input data, and image data. Input the data to be analyzed into the first analysis model, and output the cause of the anomaly and decision-making recommendations. Specifically, input the progress data and resource input data into the first LSTM encoding layer for encoding to obtain a first encoding vector, and input the on-site photos into the first CNN encoding layer for encoding to obtain a second encoding vector; input the first encoding vector and the second encoding vector into the first fusion layer to obtain a first fusion vector, and input the first fusion vector into the first LSTM decoding layer to output the cause of the anomaly and decision-making recommendations.

[0165] For example, when anomalies are detected in security data and quality data, but no anomalies are detected in progress data, the generated anomaly label vector is (1, 1, 0). The second basic analysis model is obtained by matching the anomaly label vector. The second basic analysis model includes a first generation block, a second generation block, a third generation block and a third fusion layer, wherein the first generation block includes a first LSTM encoding layer and a first LSTM decoding layer, the second generation block includes a second LSTM encoding layer and a second LSTM decoding layer, and the third generation block includes a third LSTM decoding layer.

[0166] Edit the second basic analysis model through the user interface, add the first fusion layer to the first generation block of the second basic analysis model, add the second fusion layer to the second generation block, and add the third encoding layer, the first CNN encoding layer and the first shared layer at the same time, so as to obtain the second initial analysis model.

[0167] After the second initial analysis model is obtained, the training data set is called to train the second initial analysis model, thereby obtaining a second analysis model.

[0168] Obtain the data to be analyzed, including safety data, quality data, resource input data, and image data; input the data to be analyzed into the second analysis model, and output the cause of the anomaly and decision-making recommendations, specifically:

[0169] Input the security data into the first LSTM encoding layer to obtain the first encoding vector, and input the quality data into the second LSTM encoding layer to obtain the second encoding vector; input the resource input data into the third encoding layer to obtain the third encoding vector, and input the image data into the first CNN encoding layer to obtain the fourth encoding vector; input the third encoding vector and the fourth encoding vector into the first shared layer; the first fusion layer obtains the first encoding vector from the first LSTM encoding layer, obtains the third encoding vector and the fourth encoding vector from the first shared layer, fuses the first encoding vector, the third encoding vector, and the fourth encoding vector to obtain the first fused vector, and inputs the first fused vector into the first LSTM decoding layer to obtain the first decoded vector; the second fusion layer obtains the second encoding vector from the second LSTM decoding layer, obtains the third encoding vector and the fourth encoding vector from the first shared layer, fuses the second encoding vector, the third encoding vector, and the fourth encoding vector to obtain the second fused vector, and inputs the second fused vector into the second LSTM decoding layer to obtain the second decoded vector; the first decoded vector and the second decoded vector are input into the third fusion layer to obtain the third fused vector, and inputs the third fused vector into the third LSTM decoding layer, outputting the cause of the anomaly and decision recommendations.

[0170] For example, when anomalies are detected in security data, quality data, and progress data, the generated anomaly label vector is (1,1,1). The third basic analysis model is obtained by matching the anomaly label vectors. The third basic analysis model includes a first generation block, a second generation block, a third generation block, a fourth generation block, and a fourth fusion layer, wherein the first generation block includes a first LSTM encoding layer and a first LSTM decoding layer, the second generation block includes a second LSTM encoding layer and a second LSTM decoding layer, the third generation block includes a third LSTM encoding layer and a third LSTM decoding layer, and the fourth generation block includes a fourth LSTM decoding layer.

[0171] The user edits the third basic analysis model through the interface, adds the first fusion layer to the first generation block of the third basic analysis model, adds the second fusion layer to the second generation block, adds the third fusion layer to the third generation block, and adds the fourth LSTM encoding layer, the first CNN encoding layer and the first shared layer at the same time, and modifies the third LSTM decoding layer to the Attention+LSTM decoding layer, thereby obtaining the third initial analysis model.

[0172] After the third initial analysis model is obtained, the training data set is called to train the third initial analysis model, thereby obtaining the third analysis model.

[0173] Obtain the data to be analyzed, including safety data, quality data, progress data, resource input data, and image data; input the data to be analyzed into the third analysis model, and output the causes of the anomalies and decision-making recommendations, specifically:

[0174] Input the security data into the first LSTM encoding layer to obtain the first encoding vector; input the quality data into the second LSTM encoding layer to obtain the second encoding vector; input the progress data into the third LSTM encoding layer to obtain the third encoding vector; input the resource input data into the fourth encoding layer to obtain the fourth encoding vector, input the image data into the first CNN encoding layer to obtain the fifth encoding vector; input the fourth encoding vector and the fifth encoding vector into the first shared layer; the first fusion layer obtains the first encoding vector from the first LSTM encoding layer, obtains the fourth encoding vector and the fifth encoding vector from the first shared layer, fuses the first encoding vector, the fourth encoding vector and the fifth encoding vector to obtain the first fusion vector, inputs the first fusion vector into the first LSTM decoding layer to obtain the first decoding vector; the second fusion layer obtains the second encoding vector from the second LSTM decoding layer The fourth and fifth encoding vectors are obtained from the first shared layer, and the second, fourth, and fifth encoding vectors are fused to obtain a second fused vector. The second fused vector is input into the second LSTM decoding layer to obtain a second decoding vector. The third fusion layer obtains the third encoding vector from the third LSTM decoding layer, and the fourth and fifth encoding vectors are obtained from the first shared layer. The third, fourth, and fifth encoding vectors are fused to obtain a third fused vector. The third fused vector is input into the third LSTM decoding layer to obtain a third decoding vector. The first, second, and third decoding vectors are input into the fourth fusion layer to obtain a fourth fused vector. The fourth fused vector is input into the Attention+LSTM decoding layer, and the cause of the anomaly and decision recommendations are output.

[0175] The visualization module includes a three-dimensional visualization sub-module and a two-dimensional visualization sub-module. The three-dimensional visualization sub-module uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the causes of anomalies and decision-making suggestions, and the two-dimensional visualization sub-module uses two-dimensional charts and / or text to display the causes of anomalies and decision-making suggestions.

[0176] The present invention pre-sets the functional components and basic analysis model of the analysis model. Users can customize the analysis model that conforms to the actual application scenario through the interface. Through diversified analysis models, the accuracy of abnormal analysis and decision suggestion generation is improved, and the user experience is improved.

[0177] For example, refer to Figure 6 The model database of the present invention also stores multiple historical analysis models, and the decision recommendation unit also includes a model matching module. The model matching module is used to obtain the first historical analysis model according to the abnormal label vector after obtaining the abnormal label vector; the decision generation module calls the first historical analysis model and the data to be analyzed to generate a decision recommendation.

[0178] The historical analysis model is a trained analysis model. Each historical analysis model is set with an identifier, which includes at least the abnormal vector label, unit project code value, sub-project code value, unit project code value, timestamp and abnormal vector label. The 1st to 3rd digits of the identifier are the abnormal vector label value, the 4th to 7th digits are the unit project code value, the 8th to 11th digits are the sub-project code value, the 12th to 15th digits are the unit project code value, and the 16th to 23rd digits are the timestamp. For example, for an analysis model trained with a unit engineering code value of 0001, a sub-section engineering code value of 0001, a unit engineering code value of 0002, a time node of November 24, 2024, and an anomaly label of (1,1,1), 11100010001000220141124 is used to identify the analysis model and the model is stored in the model library as a historical analysis model. 111 is the anomaly vector label value, indicating that anomalies exist in the safety, quality, and progress data. 0001 is the unit engineering code value, 0001 is the sub-section engineering code value, 0002 is the unit engineering code value, and 20141124 is the timestamp.

[0179] The model matching module uses fuzzy matching to match and obtain the first historical analysis model. Fuzzy matching is to use at least one combination of an abnormal vector label and a unit project code value, a sub-project code value, a unit project code value, and a timestamp to match the identifier of the historical analysis model, thereby obtaining the first historical analysis model. For example, the abnormal vector label is used to match and obtain the first historical analysis model, or the abnormal vector label, the unit project code value, the sub-project code value, and the unit project code value are matched to obtain the first historical analysis model, or the abnormal vector label and the timestamp are matched to obtain the first historical analysis model, or the abnormal vector label, the unit project code value, the sub-project code value, the unit project code value, and the timestamp are matched to obtain the first historical analysis model.

[0180] In this embodiment, a plurality of trained historical analysis models are stored in the model database. Each historical analysis model has a model identifier. The first historical analysis model is obtained by fuzzy matching of the model identifiers, thereby improving the efficiency of data analysis.

[0181] Compared with the prior art, the present invention has the following beneficial effects:

[0182] (1) The intelligent decision-making system for water conservancy and hydropower project construction management constructed by the present invention manages and comprehensively analyzes the process data generated during the project construction process, evaluates the project construction process from the three dimensions of safety, quality, and progress, and mines the correlation between process data such as safety, quality, progress, resource input and images based on the data of these three dimensions, providing process decision-making suggestions for project construction, greatly improving management efficiency and targeted management.

[0183] (2) The present invention uses anomaly vectors constructed using safety data, quality data, and progress data to construct or match corresponding analysis models. The causes of anomalies and decision-making suggestions for resolving anomalies are obtained through the analysis model, thereby improving the accuracy of decision-making, optimizing resource allocation in a timely manner, improving resource utilization, and reducing engineering costs through quantitative control of resource inputs.

[0184] (3) The present invention recommends a basic analysis model through anomaly vectors, and users can select corresponding functional components through the interface to modify the basic analysis model, thereby improving the scalability of the analysis model and further improving the accuracy of decision-making and user experience.

[0185] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, the functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a high-density digital video disc (DVD)), or a semiconductor medium (eg, a solid state disc (SSD)).

[0186] The memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. The non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus RAM (DR RAM). It should be noted that the memories described herein are intended to include, but are not limited to, these and any other suitable types of memories.

[0187] It should be noted that the processor in the embodiments of the present application can be an integrated circuit chip with signal processing capabilities. During implementation, the system provided by the above embodiment can be completed by the hardware integrated logic circuit in the processor or by instructions in the form of software. The above processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The system disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The system disclosed in conjunction with the embodiments of the present application is directly embodied as a hardware decoding processor for execution, or is executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art such as random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and implements the above system in combination with its hardware.

[0188] The above description is merely a preferred embodiment of the present application and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of application involved in this application is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the aforementioned application concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions applied for in this application.

Claims

1. A water conservancy and hydropower project construction management intelligent decision-making system, characterized by: Including data layer, platform layer and application layer: The data layer includes a model database and a process database; the model database is used to store multiple data anomaly detection models, multiple basic analysis models, and multiple model function components; wherein the data anomaly detection models at least include a safety data anomaly detection model, a quality data anomaly detection model, and a progress data anomaly detection model; the process database is used to store process data of each engineering unit, and the process data at least includes resource input data, safety data, progress data, quality data, and image data; The application layer includes a security management unit, a quality management unit, a progress management unit and a decision suggestion unit; The security management unit is used to call the security data and the security data anomaly detection model and output a security data anomaly label; The quality management unit is used to call the quality data and the quality data anomaly detection model, and output a quality data anomaly label; The progress management unit is used to call the progress data and the progress data anomaly detection model and output a progress data anomaly label; The decision suggestion unit includes an anomaly coding module, a model building module, a decision generation module and a visualization module; The anomaly encoding module is used to obtain the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label, and generate an anomaly label vector in the order of safety, quality, and progress; The model building module is used to obtain the abnormal label vector and build a first analysis model according to the abnormal label vector, the basic analysis model and the model function component; Constructing a first analysis model according to the abnormal label vector, the basic analysis model, and the model function component specifically includes: Obtaining a basic analysis model in the model database according to the abnormal label vector matching, wherein the basic analysis model includes at least one model function component; Edit the basic analysis model through a user interface to obtain an initial analysis model; Acquire a training data set from the model database; train the initial analysis model using the training data set to obtain the first analysis model; The decision generation module is used to call the first analysis model and the data to be analyzed, and generate a decision suggestion based on the first analysis model and the data to be analyzed; The decision generation module calls the data to be analyzed, specifically including: The decision generation module obtains the data to be analyzed from the process database according to the abnormal label vector. When the abnormal label vector indicates that the safety data is abnormal, the data to be analyzed includes the safety data, resource input data and image data; when the abnormal label vector indicates that the quality data is abnormal, the data to be analyzed includes the quality data, resource input data and image data; when the abnormal label vector indicates that the progress data is abnormal, the data to be analyzed includes the progress data, resource input data and image data; The visualization module is used to output the decision suggestion; The basic analysis model is a template model pre-constructed according to the number of abnormal data types of safety data, quality data and progress data in the abnormal label vector; The basic analysis model includes a first basic analysis model, a second basic analysis model and a third basic analysis model; If only one type of data anomaly exists among the safety data, the quality data, and the progress data, then the first basic analysis model is matched; the first basic analysis model includes a first generation block, which includes a first encoding layer and a first decoding layer; the first encoding layer is used to receive the data to be analyzed corresponding to the abnormal type data, obtain a coding vector, and input it into the first decoding layer, and the first decoding layer analyzes the coding vector and outputs the cause of the anomaly and a decision recommendation; If there are only two types of data anomalies among the safety data, the quality data, and the progress data, the second basic analysis model is matched; the second basic analysis model includes a 2-1 generation block, a 2-2 generation block, a 2-3 generation block, and a second fusion layer; the 2-1 generation block and the 2-2 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data to obtain a 2-1 encoding vector and a 2-2 encoding vector; the second fusion layer fuses and splices the 2-1 encoding vector and the 2-2 encoding vector to obtain a fused splicing vector, and inputs the fused splicing vector into the 2-3 generation block; the 2-3 generation block analyzes the fused splicing vector and outputs the cause of the anomaly and a decision recommendation; If all three types of data among the safety data, the quality data and the progress data are abnormal, the third basic analysis model is matched; the third basic analysis model includes a 3-1 generation block, a 3-2 generation block, a 3-3 generation block, a 3-4 generation block and a 3rd fusion layer; the 3-1 generation block, the 3-2 generation block and the 3-3 generation block are respectively used to receive the data to be analyzed corresponding to each type of abnormal data, and obtain a 3-1 encoding vector, a 3-2 encoding vector and a 3-3 encoding vector; the 3rd fusion layer fuses and splices the 3-1 encoding vector, the 3-2 encoding vector and the 3-3 encoding vector to obtain a fused splicing vector, and inputs it into the 3-4 generation block; the 3-4 generation block analyzes the fused splicing vector and outputs the cause of the abnormality and decision recommendations.

2. The intelligent decision-making system for water conservancy and hydropower project construction management according to claim 1 is characterized in that: The model functional components include fusion layers, sharing layers, and encoding layers, decoding layers, and generation blocks of different network structures; The editing of the basic analysis model through the user interface to obtain the initial analysis model specifically includes: obtaining the process data type and quantity that need to be associated with the abnormal data type for analysis in the data to be analyzed based on the abnormal data type in the abnormal label vector, thereby determining the model function components that need to be recommended, and recommending the model function components to the user interface; based on the user interface, selecting and / or calling the recommended model function components; integrating the selected and / or called model function components into the basic analysis model to obtain the initial analysis model, so that the initial analysis model has a model framework for comprehensive analysis of the data to be analyzed.

3. The intelligent decision-making system for water conservancy and hydropower project construction management according to claim 1 is characterized in that: The security management unit includes a security data statistical analysis module, a security data anomaly detection module and a security data visualization module; The safety data statistical analysis module is used to collect statistics and analyze the safety data of the engineering unit within a preset time period; The security data anomaly detection module is used to call the security data anomaly detection model in the model database, perform anomaly detection on the security data within a preset time point or a preset time period, and determine whether there is an anomaly in the security data; The input of the security data anomaly detection model is security data and image data; The safety data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of safety data and safety abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display safety data, safety abnormality data, and statistical analysis results of safety data and / or safety abnormality data.

4. The intelligent decision-making system for water conservancy and hydropower project construction management according to claim 1 is characterized in that: The quality management unit includes a quality data statistical analysis module, a quality data anomaly detection module and a quality data visualization module; The quality data statistical analysis module is used to collect and analyze the quality data of the engineering unit within a preset time period; The quality data anomaly detection module is used to call the quality data anomaly detection model in the model database, perform anomaly detection on the quality data at a preset time point or within a preset time period, and determine whether the quality data has an anomaly; The input of the quality data anomaly detection model is quality data and image data; The quality data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of quality data and quality abnormality data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display quality data, quality abnormality data, and statistical analysis results of quality data and / or quality abnormality data.

5. The intelligent decision-making system for water conservancy and hydropower project construction management according to claim 1 is characterized in that: The progress management unit includes a progress data statistical analysis module, a progress data anomaly detection module and a progress data visualization module; The progress data statistical analysis module is used to collect statistics and analyze the progress data of the engineering unit within a preset time period; The progress data anomaly detection module is used to call the progress data anomaly detection model in the model database, perform anomaly detection on the progress data at a preset time point or a preset time period, and determine whether there is an anomaly in the progress data; The input of the progress data anomaly detection model is progress data and image data; The progress data visualization module includes a three-dimensional visualization submodule and a two-dimensional visualization submodule. The three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform three-dimensional visualization of the progress data and progress exception data. The two-dimensional visualization submodule uses two-dimensional charts and / or text to display the progress data, progress exception data, and statistical analysis results of the progress data and / or progress exception data.

6. The intelligent decision-making system for water conservancy and hydropower project construction management according to claim 1 is characterized in that: The application layer also includes a collection unit; the collection unit includes at least a first collection module, a second collection module and a third collection module; The first acquisition module is used to collect engineering data of the engineering unit, and store the collected engineering data in a directory corresponding to the engineering unit database after adding an engineering data identifier. The engineering data at least includes the engineering location, engineering construction content, and engineering quantity. The second acquisition module is used to collect process data of each engineering unit at different time points according to a preset period, and add a process data identifier to the collected process data and store it in the process database; The third acquisition module is used to collect model data and store the model data in a model database, wherein the model data at least includes model function components, training data sets and model parameters; The application layer also includes a project management unit; The project management unit is used to edit and visualize the engineering project according to the engineering unit; the project management unit includes at least an editing module and a visualization module; the editing module is used to edit the engineering data corresponding to the engineering unit; the visualization module includes a two-dimensional visualization submodule and a three-dimensional visualization submodule, and the three-dimensional visualization submodule uses the BIM model corresponding to the engineering unit to perform a three-dimensional visualization display of the engineering unit; The two-dimensional visualization submodule displays the engineering information and statistical analysis results of the engineering unit in the form of two-dimensional charts and / or text.

7. A method for applying the intelligent decision-making system for water conservancy and hydropower project construction management according to any one of claims 1 to 6, characterized in that: The following steps are involved: Step S1, obtaining engineering data of each engineering unit in a water conservancy and hydropower engineering construction project, adding engineering data identifiers to the engineering data, and storing them in an engineering unit database; Step S2: During the construction of the water conservancy and hydropower project, process data of each engineering unit at different time points is collected according to a preset period, and the collected process data is added with a process data identifier and stored in a process database; the process data includes at least resource input data, safety data, progress data, quality data, and image data; Step S3, reading the safety data, the quality data, and the progress data stored in the most recent time period from the process database according to a preset analysis cycle; Invoking a security data anomaly detection model to analyze the security data and output a security data anomaly label; Calling a quality data anomaly detection model to analyze the quality data and output a quality data anomaly label; Calling a progress data anomaly detection model to analyze the progress data and output a progress data anomaly label; Step S4, sequentially concatenating the safety data anomaly label, the quality data anomaly label, and the progress data anomaly label to generate an anomaly label vector; Step S5: According to the abnormal label vector, the data to be analyzed is matched from the process database to obtain data to be analyzed: when the abnormal label vector indicates that the safety data has an abnormality, the data to be analyzed includes the safety data, resource input data, and image data within the corresponding time period; when the abnormal label vector indicates that the quality data has an abnormality, the data to be analyzed includes the quality data, resource input data, and image data within the corresponding time period; when the abnormal label vector indicates that the progress data has an abnormality, the data to be analyzed includes the progress data, resource input data, and image data within the corresponding time period; Step S6: pre-establishing a model database; the model database stores a plurality of basic analysis models, a plurality of model function components, a training data set, and model parameters; According to the number of abnormal labels in the abnormal label vector, the corresponding basic analysis model is matched from the model database; according to the type and number of process data that need to be analyzed in association with the abnormal data in the data to be analyzed, the model function components to be recommended are determined, and the model function components are recommended to the user interface; Integrating the model functional components selected by the user into the basic analysis model to obtain an initial analysis model; Obtaining a training data set from the model database; Using the training data set to train the initial analysis model, adjusting model parameters in the initial analysis model to obtain a trained first analysis model; Step S7: Analyze the data to be analyzed using the first analysis model to generate decision recommendations.

Images