Wind power project construction management method, device, equipment and medium

Abstract

The invention relates to a wind power project construction management method and device, equipment and a medium. The method comprises the steps of obtaining wind power project multi-source heterogeneous data, inputting the wind power project multi-source heterogeneous data into a mixed deep learning feature extraction model, and performing feature extraction on the wind power project multi-source heterogeneous data to obtain wind power project features; inputting the wind power project characteristics into a preset risk assessment model to obtain project risk assessment data; obtaining construction resource data, construction progress node information and construction process logic information, and dynamically optimizing the construction resource data, the construction progress node information and the construction process logic information based on the project risk assessment data to obtain dynamic construction scheme data; and generating a project management decision result based on the dynamic construction scheme data. By adopting the method, the overall adaptability, the resource utilization efficiency and the risk response capability of project management in a complex and changeable environment can be effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of new energy, and in particular relates to a method, device, equipment and medium for the construction and management of wind power projects. Background Technology

[0002] As a crucial component of the new energy sector, wind power projects hold irreplaceable value in driving energy structure transformation and achieving sustainable development. With the continuous growth of global demand for clean energy, the scale and complexity of wind power projects are also constantly increasing, and their construction and management directly impact the project's economic benefits and social impact. However, current wind power project management methods often fall short in dealing with complex environments and volatile factors, particularly in their inadequacy in integrating resources from multiple parties and dynamically adjusting strategies, making it difficult to meet the efficiency and adaptability requirements of modern projects.

[0003] A significant limitation of existing methods is their often lack of comprehensive processing capabilities for diverse information throughout the project lifecycle. Many management approaches tend to rely on experience-based judgment or fixed processes, neglecting the correlation and potential value of information from different sources within a project. This limitation makes it difficult for managers to quickly formulate effective response strategies when faced with unexpected situations or resource conflicts, thereby impacting the overall project schedule and quality.

[0004] Against this backdrop, the core technical challenges facing wind power project construction management have gradually emerged. The most pressing issue is extracting crucial decision-making information from a vast amount of diverse data, such as weather, equipment, and personnel factors. This information is often scattered and inconsistently formatted, making it difficult to directly guide practical work. The failure to effectively address this problem leads to another closely related technical challenge: a lack of dynamic adaptability in resource allocation and construction scheduling. For example, in a wind power project, the construction team might fail to keep abreast of weather changes and equipment status, resulting in construction delays, or even resource idleness or over-allocation. This disconnect between information and action is frequently observed in actual operations. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, device, equipment and medium for wind power project construction management to address the above-mentioned technical problems.

[0006] Firstly, this application provides a method for the construction and management of wind power projects, including:

[0007] S1. Obtain multi-source heterogeneous data of wind power projects and input the multi-source heterogeneous data of wind power projects into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects and obtain the features of wind power projects.

[0008] S2. Input the characteristics of the wind power project into the preset risk assessment model to obtain project risk assessment data; wherein, the project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event;

[0009] S3. Obtain construction resource data, construction progress node information, and construction process logic information, and based on project risk assessment data, dynamically optimize the construction resource data, construction progress nodes, and construction process logic information to obtain dynamic construction plan data.

[0010] S4. Generate project management decision results based on dynamic construction plan data; project management decision results include resource allocation methods, process adjustment content, and risk response measures.

[0011] In one embodiment, the multi-source heterogeneous data includes geospatial data, environmental time-series data, and material bill of materials structured engineering data;

[0012] Acquire multi-source heterogeneous data of wind power projects and input this data into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects, obtaining wind power project features, including:

[0013] S11. Acquire geospatial data, environmental time-series data, and material list structured engineering data;

[0014] S12. Input geospatial data, environmental time-series data, and material list structured engineering data into a hybrid deep learning feature extraction model to extract spatial features, time-series features, and resource features;

[0015] The hybrid deep learning feature extraction model includes a convolutional neural network branch, a long short-term memory network branch, a fully connected network branch, and a feature fusion module based on an attention mechanism. The convolutional neural network branch is used to extract spatial features of geospatial data, the long short-term memory network branch is used to extract temporal features of environmental time-series data, and the fully connected network branch is used to extract resource features of structured engineering data of bill of materials.

[0016] S13. Input the spatial features, temporal features and resource features into the feature fusion module to perform feature fusion and obtain the wind power project features.

[0017] In one embodiment, wind power project characteristics are input into a preset risk assessment model to obtain project risk assessment data, including:

[0018] S21. Input the characteristics of the wind power project into the preset risk assessment model to generate the probability of occurrence of each risk event type;

[0019] S22. Based on the preset weight coefficients for each risk event type, the preset impact coefficients for each risk event type, and the probability of occurrence of each risk event type, the project risk assessment data is calculated.

[0020] The expression for the project risk assessment data is as follows:

[0021]

[0022] In the formula, This represents project risk assessment data. Indicates the total number of risk event categories. Indicates the first Weighting coefficients for risk-like events Indicates the first The probability of occurrence of risk events. Indicates the first Impact coefficient of risk events.

[0023] In one embodiment, construction resource data, construction progress node information, and construction process logic information are acquired. Based on project risk assessment data, the construction resource data, construction progress nodes, and construction process logic information are dynamically optimized to obtain dynamic construction plan data, including:

[0024] S31. Construct a multi-objective optimization function with the goal of minimizing project delay costs and resource idle costs;

[0025] The expression for the multi-objective optimization function is as follows:

[0026]

[0027] In the formula, The overall optimization objective; The cost of project delay is determined by the amount of time delay and the unit time cost caused by high-probability delay risk events in the risk assessment data. The cost of idle resources is determined by the value of resources that are not fully utilized in resource allocation. and These are the weighting coefficients for construction delay costs and resource idle costs, respectively. ;

[0028] S32. Solve the multi-objective optimization function using an improved genetic algorithm to generate a Pareto optimal solution set;

[0029] S33. Based on preset decision preferences, select the optimal solution from the Pareto optimal solution set to obtain the optimized construction resource allocation, schedule node arrangement and process logic relationship;

[0030] S34. Integrate and optimize the allocation of construction resources, schedule nodes, and process logic to generate dynamic construction plan data.

[0031] In one embodiment, project management decision results are generated based on dynamic construction plan data, including:

[0032] S41. Based on the construction resource data, construction progress node information and construction process logic information in the dynamic construction scheme data, construct a simulated construction scenario with multiple discrete time steps.

[0033] S42. In the simulated construction scenario at each discrete time step, based on the project risk assessment data, the Monte Carlo method is used to simulate the construction process and the probability of risk events, and the simulation results of the construction process and the probability of risk events are obtained.

[0034] S43. Perform statistical analysis on the simulation results of the construction process and the simulation results of the probability of risk events to obtain the resource demand fluctuation value and risk accumulation value at each time step;

[0035] S44. Based on the fluctuation value of resource demand and the cumulative value of risk, generate the resource allocation method and process adjustment content to guide the next construction cycle;

[0036] S45. When the cumulative risk value exceeds the preset warning threshold, risk response measures are matched and generated from the preset strategy library based on the risk event category that triggered the threshold.

[0037] S46. Based on resource allocation methods, process adjustments, and risk response measures, generate project management decision results.

[0038] In one embodiment, based on the fluctuation value of resource demand and the cumulative value of risk, a resource allocation method and process adjustment content are generated to guide the next construction cycle, including:

[0039] S51. Establish a short-term decision optimization model with the goal of minimizing resource demand fluctuations and risk accumulation.

[0040] S52. Transform the short-term decision optimization model into a constrained linear programming problem, and solve it to obtain the optimal resource input vector and the optimal process state adjustment vector for future time steps.

[0041] S53. Decode the optimal resource input vector into specific resource types, quantities, and allocation instructions to form a resource allocation method;

[0042] S54. Decode the optimal process state adjustment vector into the process start time offset, parallelism adjustment and logical relationship change to form the process adjustment content.

[0043] Secondly, this application also provides a wind power project construction management device, comprising:

[0044] The data acquisition and feature extraction module is used to acquire multi-source heterogeneous data of wind power projects and input the multi-source heterogeneous data of wind power projects into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects and obtain the features of wind power projects.

[0045] The project risk assessment module is used to input the characteristics of wind power projects into a preset risk assessment model to obtain project risk assessment data; among which, the project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event;

[0046] The construction plan dynamic optimization module is used to acquire construction resource data, construction progress node information and construction process logic information, and based on project risk assessment data, dynamically optimize the construction resource data, construction progress nodes and construction process logic information to obtain dynamic construction plan data.

[0047] The project management decision module is used to generate project management decision results based on dynamic construction plan data; the project management decision results include resource allocation methods, process adjustment content, and risk response measures.

[0048] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0049] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0050] The aforementioned wind power project construction management method, device, equipment, and medium employ a pre-defined hybrid deep learning feature extraction model to extract deep features from multi-source heterogeneous data of wind power projects, obtaining high-quality project feature data. This data is then input into a risk assessment model to accurately quantify the probability of various risk events. Subsequently, based on this risk assessment data, construction resources, schedule nodes, and process logic are dynamically optimized collaboratively to generate dynamic construction plans that can respond to risk changes in real time. Finally, management decision results, including specific resource allocation, process adjustments, and risk response measures, are automatically generated directly based on this optimized plan. This achieves the integration and utilization of information throughout the entire project lifecycle, improves the accuracy of resource allocation in complex environments and the adaptability of dynamic adjustments to construction plans, and effectively enhances the overall risk resistance and decision-making efficiency of project management. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating a wind power project construction management method in one embodiment;

[0053] Figure 2 This is a schematic diagram of the structure of a wind power project construction management device in one embodiment;

[0054] Figure 3 This is a schematic diagram of a computer device in one embodiment. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] refer to Figure 1 The application presents a flowchart illustrating a wind power project construction management method, which includes the following steps:

[0057] S1. Obtain multi-source heterogeneous data of wind power projects and input the multi-source heterogeneous data of wind power projects into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects and obtain the features of wind power projects.

[0058] Specifically, firstly, the acquisition of multi-source heterogeneous data needs to cover all dimensions of project construction, including three core types of data. The first is structured data, acquired through real-time sensor data collection and database interface calls, covering operating parameters of wind turbine equipment, attendance and skill level data of construction personnel, and quantity and cost data of material procurement. The second is semi-structured data, which can be extracted using document parsing tools; construction progress reports, equipment maintenance records, and supervision opinion documents all fall into this category. After acquisition, it needs to be converted into key-value pair format for unified storage. The third is unstructured data, acquired through image acquisition equipment and natural language processing interfaces, such as drone aerial images of the construction site, disaster warning texts issued by meteorological departments, and voice recordings of equipment fault diagnosis. During data acquisition, a quality verification module needs to be set up, using outlier detection algorithms to remove invalid data caused by sensor malfunctions and data transmission errors. The isolated forest algorithm is suitable for this type of detection work. Simultaneously, timestamps and geographic location tags are used to align and correlate various types of data, laying the foundation for subsequent feature extraction.

[0059] The hybrid deep learning feature extraction model employs a branch extraction-feature fusion architecture, designing dedicated extraction branches for different types of data. For structured data, a fully connected neural network branch is used. The input layer maps standardized data to hidden layers, and the ReLU activation function enhances nonlinear expressive power, ultimately outputting a structured feature vector. The standardized data includes multi-dimensional information such as equipment parameters, personnel data, and material data. For unstructured image data, an improved CNN (Convolutional Neural Network) branch is used, which can be optimized based on the ResNet (Residual Network) series of networks. Through convolution operations and pooling layers, spatial features such as construction area boundaries and equipment installation status in the image are extracted. The fully connected layers of the original network are removed and replaced with global average pooling layers to output image feature vectors. For text-based data, including text transformed from semi-structured documents, the BERT (Bidirectional Encoder Representation from Transformers) series of pre-trained model branches are used. After the text is segmented into segments of fixed length, semantic features are extracted through the Transformer encoder. Information such as disaster types in weather warnings and problem descriptions in construction reports can be extracted in this way, and the final output is a text feature vector.

[0060] In the feature fusion stage, an attention mechanism is used to fuse the output features of the three branches. After constructing an attention weight matrix, the contribution of each feature dimension to project management is calculated using the Softmax function. The weight difference between equipment temperature features and personnel attendance features is reflected in this way, ultimately outputting a comprehensive feature vector for the wind power project. This vector contains core information such as equipment operating status, construction progress, and environmental risks, providing accurate input for subsequent risk assessment. During model training, the Adam optimizer is used, with feature reconstruction error as the optimization objective. The mean squared error loss function is used to calculate this error. The training dataset uses labeled data from historical wind power projects, and multiple rounds of iterative training ensure that the model's feature extraction accuracy meets project requirements.

[0061] S2. Input the characteristics of the wind power project into the preset risk assessment model to obtain project risk assessment data.

[0062] Among them, project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event.

[0063] Specifically, the risk assessment model is constructed based on the logic of feature mapping, risk classification, and probability prediction. First, it is necessary to clarify the core risk types of wind power project construction. Combining industry standards and historical data, these can be summarized into six key risk categories: meteorological disaster risk (typhoons, rainstorms, strong winds, etc.); equipment failure risk (e.g., damage to wind turbine gearboxes, generator failures); personnel safety risk (covering scenarios such as falls from heights and electric shocks); schedule delay risk (insufficient material supply, poor workflow coordination, etc.); cost overrun risk (equipment price increases, rework costs, etc. are the main causes); and technical risk (substandard construction techniques, geological conditions not matching the design, etc.). Each risk category corresponds to multiple specific risk events.

[0064] The pre-defined risk assessment model employs a combination of a classifier and a regressor. The classifier uses the LightGBM (Light Gradient Boosting Machine) gradient boosting tree algorithm. Inputting wind power project features, it constructs multiple decision trees to achieve multi-classification of risk types and outputs predicted labels for risk event types using a one-hot encoding format. The regressor uses a logistic regression model. For each type of risk event identified by the classifier, a probability prediction sub-model is constructed. Inputting local features related to the risk from the feature vector (for example, meteorological risks correspond to features such as temperature and wind speed), the output is mapped to a fixed interval using the sigmoid function to obtain the probability of occurrence for each risk event. The probability results must maintain a certain level of numerical accuracy.

[0065] During the model training phase, a dataset containing risk labels needs to be constructed. Historical risk events from projects are labeled according to the six categories mentioned above. Simultaneously, expert ratings are used to supplement sample labels. The Delphi method can be employed to organize industry experts to complete the labeling of risk occurrence probabilities. The dataset is proportionally divided into training, validation, and test sets. The F1 score and mean absolute error (MAO) are used as evaluation metrics, with the F1 score evaluating classification performance and the MAO evaluating regression performance. Model hyperparameters are optimized using a grid search method. The learning rate of the LightGBM algorithm and the regularization coefficient of logistic regression are determined in this way to ensure that the accuracy of risk type identification and the precision of probability prediction meet project requirements.

[0066] The project risk assessment data is output in structured tables, including fields such as risk event ID, risk type, detailed description, probability of occurrence, and degree of impact. The degree of impact is calculated based on probability and loss value, which can be determined with reference to industry average data. Risk events with a probability of occurrence reaching a certain threshold should be marked as high-risk and addressed as a priority in subsequent steps.

[0067] S3. Acquire construction resource data, construction progress node information, and construction process logic information, and based on project risk assessment data, dynamically optimize the construction resource data, construction progress nodes, and construction process logic information to obtain dynamic construction plan data.

[0068] Specifically, firstly, the acquisition of construction resource data, construction progress node information, and construction process logic information needs to achieve real-time updates and accurate quantification. Construction resource data specifically includes human resources, equipment resources, and material resources. Human resources cover the number of personnel, skill levels, and current work status of each type of work, which can be synchronized through personnel positioning systems and attendance software; equipment resources include the quantity, location, and operating status of wind turbine installation equipment, transport vehicles, and testing instruments, and relevant data are collected in real time through IoT modules; material resources involve the inventory quantity, storage location, and supply cycle of concrete, steel, and wind turbine components, which are obtained through the ERP (Enterprise Resource Planning) system interface. The ERP system is an information system that integrates various internal resource management within an enterprise, enabling centralized control of data in procurement, inventory, and other aspects. All resource data must be labeled with a unique identifier and an available time window.

[0069] Construction progress milestone information is decomposed based on the project's WBS (Work Breakdown Structure). WBS is a structured tool that decomposes a project into manageable task units, clearly defining the project scope and work content. Through this structure, multiple core milestones are identified, such as site leveling completion, foundation pouring completion, and wind turbine installation completion. Each milestone includes information such as planned completion time, actual completion progress, and dependencies on preceding milestones. Actual completion progress is represented by quantitative indicators and is synchronized in real time through a professional project management software interface. The construction process logic information needs to be combined with the wind power project construction specifications to clarify the sequence of each process. The installation of the wind turbine base can only be carried out after the foundation is poured, which is a typical process logic. At the same time, the process parameter requirements and quality inspection standards should be clarified. The information should be transformed into computable logical rules in the form of flowcharts. Petri nets can be used to describe the process flow relationship. Petri nets are a graph-based system modeling and analysis tool. Through elements such as places, transitions and arcs, they can intuitively and accurately express the dynamic behaviors such as concurrency, synchronization and conflict in discrete event systems. They are very suitable for characterizing the sequential dependence and flow logic of each process in wind power project construction.

[0070] The dynamic optimization process uses an improved genetic algorithm as the core optimization tool. The optimization objective is set as a multi-objective function that minimizes risk loss, maximizes resource utilization, and minimizes schedule deviation. The objective function expression is as follows:

[0071]

[0072] In the above formula, This represents the objective value of a multi-objective optimization function, and the optimization direction is to minimize it. , , These are all weighting coefficients, which can be dynamically adjusted according to the project stage. The weighting allocation in the early and middle stages of construction needs to be determined in conjunction with the actual management priorities. This represents the risk loss value, calculated based on the risk probability and individual risk loss. The risk loss from equipment failure can be determined by the sum of equipment maintenance costs and downtime losses. Representing resource utilization rate, it is a weighted average of the utilization rates of manpower, equipment, and materials. The weights of the three types of resources can be set according to their degree of impact on the project. The schedule deviation rate represents the percentage difference between the actual progress and the planned progress, and is used to quantify the degree of deviation between the progress execution and the plan.

[0073] The algorithm uses real-number encoding, with each chromosome corresponding to a construction plan, comprising three parts: a resource allocation vector, a schedule node adjustment vector, and a process parameter adjustment vector. The resource allocation vector specifies the number of personnel and equipment types allocated to each construction team. The schedule node adjustment vector represents the planned completion time offset for each node. The process parameter adjustment vector involves process-related parameters such as concrete pouring speed and wind speed thresholds for hoisting operations. In the genetic operation, the selection operator uses roulette wheel selection, the crossover operator uses arithmetic crossover, and the mutation operator uses Gaussian mutation. Through multiple generations of iterative optimization, the optimal solution is obtained, thus determining the optimal construction plan.

[0074] The optimization process requires setting constraints based on risk assessment data. For high-probability meteorological disaster risks, the allocation of resources should increase the proportion of emergency equipment, including rain shelters and emergency generators. A buffer period should be reserved during schedule adjustments to address potential delays. Protective measures for rainy weather construction should be added to the process logic; rainproof covering measures for foundation construction fall into this category. Regarding equipment failure risks, the availability of backup equipment should be prioritized in resource data, and the timing of equipment installation procedures should be adjusted appropriately during schedule adjustments to avoid overall construction stoppages due to single equipment failures.

[0075] The output of dynamic construction plan data includes an optimized resource allocation table, a schedule schedule, and process adjustment instructions. The resource allocation table clearly defines the usage time, location, and quantity of each resource; the schedule schedule updates the planned completion time and buffer time for each node; and the process adjustment instructions highlight the differences from the original process and implementation requirements. Simultaneously, visualization tools such as Matplotlib can be used to generate comparison charts of the plans, visually displaying changes in risk losses, resource utilization, and schedule deviations before and after optimization.

[0076] S4. Generate project management decision results based on dynamic construction plan data.

[0077] Specifically, the core of generating directly executable project management decisions based on dynamic construction plan data is to transform the optimized plan into specific and clear management instructions, ensuring the implementation and operability of the decisions. The generation of management decision results adopts a process of instruction decomposition, priority ranking, and measure refinement. First, the dynamic construction plan data is decomposed into three core instructions according to management dimensions: resource allocation, process adjustment, and risk response. Each type of instruction corresponds to a clear executing entity. Resource allocation instructions are usually executed by the materials management department, while process adjustment instructions correspond to the construction team.

[0078] The generation of resource allocation methods needs to be combined with the optimized resource allocation table, clearly defining the resource scheduling path and execution standards. Taking the allocation of wind turbine hoisting equipment as an example, the transportation route from the current storage location to the construction site needs to be marked. To improve reliability, multiple alternative routes need to be planned, and route selection should avoid areas with high meteorological risks. The transportation time requirements for the equipment and the personnel configuration for hoisting operations need to be clearly defined. The personnel configuration needs to clearly define the division of labor for each position, and the cost and efficiency indicators of resource allocation need to be calculated. Costs include transportation costs, labor costs, etc., and efficiency indicators can be reflected by equipment utilization rate. For manpower allocation, it is necessary to match personnel skill levels with risk event types. For high-risk processes involving high-altitude operations, priority should be given to allocating experienced personnel with good safety records. At the same time, the personnel's arrival time and safety training requirements need to be clearly defined. Risk prevention and control training is an important part of ensuring personnel safety.

[0079] Work process adjustments must be based on the schedule and process logic optimization results, and the adjustment details should be clearly defined using a work process network diagram. A double-symbol network diagram can be used for drawing such diagrams. When equipment failure risks cause delays in the wind turbine installation process, the work process adjustment instruction must explicitly initiate the subsequent cable laying process ahead of schedule, making full use of the downtime during equipment maintenance. The connection points between cable laying and equipment installation must be marked, and cable laying work must be stopped before equipment maintenance is completed, allowing sufficient connection time to ensure smooth process transitions. The scope of the project duration change after the work process adjustment must also be clearly defined. For work process adjustments involving multiple work teams, a cross-operation coordination plan must be generated, rationally allocating the work time for each team. The allocation of work time between the foundation construction team and the equipment installation team must avoid construction conflicts to ensure an orderly and efficient construction site.

[0080] The development of risk response measures requires the formulation of a comprehensive process encompassing prevention, response, and recovery for identified high-risk events. Taking typhoon risk as an example, preventive measures include the early reinforcement of wind turbine components and the dismantling of temporary structures at the construction site. Wind turbine components can be reinforced using steel cables. Response measures clearly define personnel evacuation routes and equipment power-off procedures during a typhoon. Personnel must be guided to safe emergency shelters, and equipment power-off must be carried out in a reasonable sequence, from the periphery to the center, to ensure operational safety. Recovery measures stipulate the on-site inspection process and resumption conditions after the typhoon. The inspection must first check the equipment damage and then assess the safety of the construction area. Resumption of work must meet relevant conditions such as wind speed and equipment condition.

[0081] The output of project management decisions adopts a combination of instruction documents and visual dashboards. The instruction documents are in PDF format and include fields such as decision instruction number, executing entity, execution time, execution standards, and acceptance indicators. They are pushed to relevant responsible persons through the project management system to ensure timely and accurate instruction delivery. The visual dashboards can be updated in real time via a web interface, displaying information such as resource allocation progress, process completion status, and risk event handling status, allowing managers to monitor the effectiveness of decision execution in real time.

[0082] Simultaneously, a decision feedback module is set up to acquire decision execution data through sensors and reports from on-site personnel. Resource availability rate and process completion quality are examples of this data. When the deviation between the execution data and the decision requirements reaches a certain level, the cyclical optimization process of steps S1 to S4 is automatically triggered to dynamically adjust management decisions, ensuring that management decisions can adapt to changes in the actual situation of the project and improve the flexibility and effectiveness of management.

[0083] In the aforementioned wind power project construction management method, a hybrid deep learning feature extraction model is used to extract features from the multi-source heterogeneous data of the wind power project, resulting in unified project feature data. This data is then input into a risk assessment model to quantify the probability of various risk events. Based on this probability, construction resources, schedule nodes, and process logic are dynamically optimized to generate dynamic construction plan data. Finally, based on this plan data, project management decision results are generated, including resource allocation methods, process adjustment content, and risk response measures. This enables comprehensive processing and dynamic adaptation of complex project information, effectively improving the efficiency and risk resistance of project management.

[0084] In an optional embodiment, the multi-source heterogeneous data includes geospatial data, environmental time-series data, and bill of materials structured engineering data;

[0085] Acquire multi-source heterogeneous data of wind power projects and input this data into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects, including the following steps:

[0086] S11. Obtain geospatial data, environmental time-series data, and structured engineering data of bill of materials.

[0087] The project acquires geospatial data, environmental time-series data, and structured engineering data from the bill of materials (BOM). These three types of data form a closed loop of project data from spatial, temporal, and resource dimensions. Geospatial data encompasses spatial attribute information such as the topography, geological structure, and distribution of surrounding infrastructure in the project area. It is acquired through a geographic information system platform, drone aerial photography, and geological survey reports, and spatially aligned using a coordinate system. Environmental time-series data, indexed by time, includes meteorological parameters and construction equipment operating parameters. It is collected in real time through on-site weather stations and equipment sensors to capture dynamic changes in the environment and equipment. Structured engineering data from the BIM (Building Information Modeling) system, enterprise resource planning (ERP) system, and design drawings. The BIM system is a three-dimensional model system that integrates data from the entire lifecycle of engineering design, construction, and operation and maintenance, providing accurate engineering structure and material parameter support for wind power projects. The BIM data is stored in a standardized data table format.

[0088] After data acquisition, quality control is only required for core issues. Geospatial data focuses on verifying spatial topological integrity and coordinate consistency, environmental time series data focuses on outlier removal and data completion for key periods, and bill of materials data ensures field validity and matching of engineering parameters, providing a reliable data foundation for subsequent feature extraction.

[0089] S12. Input geospatial data, environmental time-series data, and material list structured engineering data into a hybrid deep learning feature extraction model to extract spatial features, time-series features, and resource features.

[0090] The hybrid deep learning feature extraction model includes a convolutional neural network branch, a long short-term memory network branch, a fully connected network branch, and a feature fusion module based on an attention mechanism. The convolutional neural network branch is used to extract spatial features from geospatial data, the long short-term memory network branch is used to extract temporal features from environmental time-series data, and the fully connected network branch is used to extract resource features from structured engineering data in the bill of materials.

[0091] Specifically, the three types of data are input into the corresponding branches of the hybrid deep learning feature extraction model to achieve accurate feature extraction. The model consists of a convolutional neural network branch, a long short-term memory network branch, a fully connected network branch, and a feature fusion module based on an attention mechanism. Each branch is precisely adapted to the data type, forming a dedicated data-branch extraction path.

[0092] The convolutional neural network branch is used to process geospatial data. Leveraging its spatial feature extraction advantages, it captures spatial correlation information such as topography and geology through multi-scale convolutional operations, outputting spatial features that reflect the spatial attributes of the project. The long short-term memory network branch addresses the temporal dependence of environmental time-series data, employing a bidirectional structure to mine data change patterns from both positive and negative time dimensions. It combines a temporal attention layer to enhance features during critical construction periods, generating time-series features. The fully connected network branch adapts to the high-dimensional characteristics of structured engineering data such as bills of materials. Through encoding and nonlinear mapping, it extracts core information such as material supply and demand and resource allocation, obtaining resource features.

[0093] S13. Input the spatial features, temporal features and resource features into the feature fusion module to perform feature fusion and obtain the wind power project features.

[0094] Specifically, spatial, temporal, and resource features are input into a feature fusion module for integration, forming wind power project features. Before fusion, a fully connected layer maps the three types of features to the same dimensional space. Then, an attention mechanism is used to construct a weight matrix: a self-attention mechanism distinguishes the importance of each dimension within a single feature type, such as the weight priority of terrain carrying capacity in spatial features; a cross-attention mechanism explores the correlation between the three types of features, such as the correlation between traffic accessibility and material transportation efficiency. After weighted fusion and standardization, the integrated features of the wind power project, combining spatial, temporal, and resource information, are output, providing accurate input for subsequent risk assessment.

[0095] In one embodiment, the characteristics of the wind power project are input into a preset risk assessment model to obtain project risk assessment data, including the following steps:

[0096] S21. Input the characteristics of the wind power project into the preset risk assessment model to generate the probability of occurrence of each risk event type.

[0097] Specifically, the characteristics of the wind power project are input into a pre-defined risk assessment model to generate the probability of occurrence for each type of risk event. The risk assessment model employs a classifier + regressor architecture. The classifier identifies risk event types based on the comprehensive input features, accurately locating the core risk categories the project may face. Weather disasters, equipment failures, personnel safety, and schedule delays all fall into this category of core risks. The regressor then extracts strongly correlated local features from the feature vector for each identified risk event. Weather risks correspond to time-series features such as wind speed and precipitation, while equipment failure risks correspond to resource features related to equipment operating parameters. These features are mapped to values ​​within a specific interval using a sigmoid function; this value is the probability of occurrence of the first risk event. The probability of a risk event occurring is denoted as . .

[0098] S22. Based on the preset weight coefficients for each risk event type, the preset impact coefficients for each risk event type, and the probability of occurrence of each risk event type, the project risk assessment data is calculated.

[0099] The expression for the project risk assessment data is as follows:

[0100]

[0101] In the formula, This represents project risk assessment data. Indicates the total number of risk event categories. Indicates the first Weighting coefficients for risk-like events Indicates the first The probability of occurrence of risk events. Indicates the first Impact coefficient of risk events.

[0102] Specifically, project risk assessment data is calculated by combining the preset weighting coefficients, preset impact coefficients, and occurrence probabilities of each risk event type. Weighting coefficients With influence coefficient It is a core parameter for risk quantification. Weighting coefficients. This coefficient reflects the importance of various risks to project objectives, including schedule, cost, and safety. It is determined using the analytic hierarchy process (AHP) combined with industry expert scoring; risks with a significant impact on safety are typically assigned higher weights. Impact Coefficient To quantify the extent of loss after a risk occurs, refer to loss data from similar risks in historical projects. Data such as downtime and maintenance costs can be used as a reference and converted into calculable quantitative values ​​after normalization.

[0103] The project risk assessment data is calculated using the weighted summation formula described above. The meaning of each parameter in the formula is clearly defined. This represents the comprehensive risk assessment data of the project. Its value directly reflects the overall risk level of the project. The higher the value, the greater the comprehensive risk faced by the project. This represents the total number of risk event categories identified by the risk assessment model, covering all risk types that the model determines are relevant to the project. For the first The weighting coefficient of a risk event reflects its relative importance in the overall project risk; a higher weighting means a more critical impact on the project. For the first The probability of occurrence of risk-like events is generated by model calculation in step S21 and is the basic data for risk quantification. For the first The impact coefficient of a risk event is specifically used to quantify the potential loss caused by the occurrence of such a risk, providing support for risk assessment in terms of loss dimensions.

[0104] Project risk assessment data is output in a structured format, specifically a combination of risk event, probability, weight, impact coefficient, and comprehensive risk value. This output format includes detailed quantitative information on individual risks, facilitating targeted analysis, and also... The value intuitively presents the overall risk status of the project, facilitating macro-level control. For and High-risk events whose product reaches a preset threshold will be marked separately in the output results, providing a clear priority basis for subsequent construction resource optimization and risk response measures, ensuring that resources can be accurately allocated to high-risk links and improving risk management efficiency.

[0105] In an optional embodiment, construction resource data, construction progress node information, and construction process logic information are acquired, and based on project risk assessment data, the construction resource data, construction progress nodes, and construction process logic information are dynamically optimized to obtain dynamic construction plan data, including the following steps:

[0106] S31. Construct a multi-objective optimization function with the goal of minimizing project delay costs and resource idle costs;

[0107] The expression for the multi-objective optimization function is as follows:

[0108]

[0109] In the formula, The overall optimization objective; The cost of project delay is determined by the amount of time delay and the unit time cost caused by high-probability delay risk events in the risk assessment data. The cost of idle resources is determined by the value of resources that are not fully utilized in resource allocation. and These are the weighting coefficients for construction delay costs and resource idle costs, respectively. .

[0110] Specifically, a multi-objective optimization function is constructed with the goal of minimizing project delay costs and resource idle costs. This function directly relates to project risks and economic costs, achieving synergy between risk management and cost control. The optimization function is guided by minimizing total cost. In the above expression, This represents the overall optimization objective, which is the weighted sum of project delay costs and resource idle costs. The optimization direction is to minimize this value. The cost of project delay is calculated directly with risk assessment data, extracting the time delay corresponding to high-probability delay risk events, such as equipment failure and weather disasters. This is then combined with the project's pre-set unit time cost, which includes project-related expenses such as labor idle time and equipment rental fees. The product of the two is the cost of project delay. The cost of idle resources is determined by summing up the value per unit time of resources that have not reached the threshold of reasonable utilization in resource allocation. Idle hoisting equipment and construction teams waiting for work are examples of such underutilized resources.

[0111] and These are the weighting coefficients for construction delay costs and resource idle costs, respectively, and both satisfy... The constraints. The weight values ​​need to be dynamically adjusted based on the actual needs of the project. When the project is in a rush phase, Prioritize controlling delays by increasing the value; increase the value if project resources are tight. Weighting is crucial for minimizing idle resources and waste. Weighting coefficients can be determined by the project management team in conjunction with historical data and expert opinions, ensuring a precise match with the project's current management priorities.

[0112] S32. The multi-objective optimization function is solved by an improved genetic algorithm to generate a Pareto optimal solution set.

[0113] Specifically, the improved genetic algorithm addresses the multi-objective optimization characteristics by introducing a non-dominated sorting mechanism and crowding calculation on the basis of the traditional genetic algorithm, thus avoiding the optimization results from being biased towards a single objective. The algorithm uses construction resource allocation schemes and schedule node arrangements as encoding objects, and high-risk events in the risk assessment data as constraints, such as prohibiting outdoor hoisting operations during high-risk periods. Through iterative evolution using genetic operations such as selection, crossover, and mutation, it ultimately obtains multiple non-dominated Pareto optimal solutions, each corresponding to a feasible solution that balances the construction period and resource costs.

[0114] S33. Based on preset decision preferences, select the optimal solution from the Pareto optimal solution set to obtain the optimized construction resource allocation, schedule arrangement and process logic relationship.

[0115] Specifically, the optimal solution is selected from the Pareto optimal solution set based on pre-defined decision preferences. These preferences must be clearly defined in conjunction with the project's overall lifecycle goals and the priorities of the current stage. When a project signs a rigid schedule contract, the decision preference is set to prioritize controlling the cost of schedule delays, and the optimal solution is selected from the set of solutions. The minimum acceptable solution; given the project's stringent cost control requirements, the preference shifts to prioritizing controlling idle resource costs, and prioritizing [the following options]. The optimal solution; for balanced management requirements, a weighted scoring method can be used to evaluate each optimal solution. The values ​​are calculated, the scheme with the highest comprehensive score is selected, and the optimized construction resource allocation, schedule arrangement and process logic relationship are finally output.

[0116] S34. Integrate and optimize the allocation of construction resources, schedule nodes, and process logic to generate dynamic construction plan data.

[0117] Specifically, the optimized allocation of construction resources, schedule arrangements, and technological logic relationships are integrated to generate dynamic construction plan data. During the integration process, it is crucial to ensure logical consistency among all elements: resource allocation must clearly define the usage periods, quantities, and allocation paths of various resources, precisely matching the optimized schedule nodes; schedule node arrangements must indicate the adjusted times for each node, reserving buffer time for high-risk events; and technological logic relationships must incorporate specific technological requirements based on risk characteristics, such as requiring special protective measures for concrete pouring in rainy weather. Ultimately, this results in structured dynamic construction plan data, providing a direct basis for subsequent decision-making.

[0118] In one embodiment, project management decision results are generated based on dynamic construction plan data, including:

[0119] S41. Based on the construction resource data, construction progress node information and construction process logic information in the dynamic construction scheme data, construct a simulated construction scenario with multiple discrete time steps.

[0120] Specifically, based on construction resource data, progress node information, and process logic information in the dynamic construction plan, simulated construction scenarios with multiple discrete time steps are constructed. The division of time steps is determined by combining the construction process cycle and the high-risk event period. Key processes are divided into time steps according to time units matching their own cycles, while basic processes with longer cycles are divided into relatively longer time units. Each time step scenario must fully map the resource allocation, progress objectives, and process constraints for that period. Resource allocation includes core elements such as available equipment and personnel teams, progress objectives specify the required percentage of completion of nodes, and process constraints define the sequence of processes. A visual scenario model is built using engineering simulation software to ensure consistency with the actual construction process.

[0121] S42. In the simulated construction scenario at each discrete time step, based on the project risk assessment data, the Monte Carlo method is used to simulate the construction process and the probability of risk events, and the simulation results of the construction process and the probability of risk events are obtained.

[0122] Specifically, in the simulation scenario of each discrete time step, the Monte Carlo method is used to simulate the construction process and the probability of risk events, incorporating project risk assessment data. Before simulation, the risk types and probabilities of occurrence in the risk assessment data are converted into simulation parameters, and the probability of occurrence of a certain type of weather risk is set as the random event triggering condition for that time step. The construction process is simulated through a large number of random samples, with the number of samples reasonably set according to the complexity of the risk. The construction process simulation focuses on indicators such as resource consumption and the speed of process progress, while the risk event simulation captures the interference of risk triggering on construction. Finally, the simulation results of the construction process and the simulation results of the risk event probability at each time step are output. The former includes key indicators such as actual progress deviation and resource consumption value, while the latter covers core information such as the actual number of risk triggers and the duration of impact.

[0123] S43. Perform statistical analysis on the simulation results of the construction process and the simulation results of the probability of risk events to obtain the resource demand fluctuation value and risk accumulation value at each time step.

[0124] Specifically, statistical analysis is performed on the simulation results of the construction process and the probability of risk events to obtain the resource demand fluctuation value and risk accumulation value at each time step. The resource demand fluctuation value is calculated by the difference between the simulated resource consumption and the planned consumption at that time step. Positive values ​​indicate resource overconsumption, and negative values ​​indicate resource redundancy, intuitively reflecting the dynamic changes in resource demand. The risk accumulation value is calculated by superposition, converting the losses caused by risk events at each time step into quantitative values ​​according to preset standards and adding them up to form the total risk loss quantification result from project initiation to the current time step.

[0125] S44. Based on the fluctuation value of resource demand and the cumulative value of risk, generate resource allocation methods and process adjustment content to guide the next construction cycle.

[0126] Specifically, based on the fluctuation value of resource demand and the cumulative value of risk, resource allocation methods and process adjustments are generated to guide the next construction cycle. The resource allocation method is tailored to the fluctuation value: if the fluctuation value is positive, resources are prioritized from nearby backup resource pools or resource usage priorities are adjusted; if it is negative, redundant resources are allocated to parallel processes with high demand. Process adjustments are optimized based on the underlying causes of the fluctuation. For example, if resource fluctuations are caused by equipment failure risks, processes dependent on that equipment can be postponed, and processes with a high proportion of manual operations can be started earlier.

[0127] S45. When the cumulative risk value exceeds the preset warning threshold, risk response measures are matched and generated from the preset strategy library based on the risk event category that triggered the threshold.

[0128] Specifically, when the cumulative risk value exceeds the preset warning threshold, risk response measures are generated from the preset strategy library based on the risk event category that triggered the threshold. The warning threshold is set in conjunction with the project's total risk tolerance. The strategy library stores response plans categorized by risk type. For example, weather-related risks correspond to measures such as setting up rain shelters and suspending work processes, while equipment-related risks correspond to measures such as activating backup equipment and having maintenance teams on standby. The scope of risk impact is also taken into account during the matching process to ensure that the measures accurately cover the risk area.

[0129] S46. Based on resource allocation methods, process adjustments, and risk response measures, generate project management decision results.

[0130] Specifically, the integration of resource allocation methods, process adjustments, and risk response measures forms the project management decision. During integration, the execution time, responsible parties, and acceptance criteria for each decision must be clearly defined: resource allocation should specify resource type, quantity, and arrival time; process adjustments should clearly define the connection points between processes before and after the adjustment; and risk response measures should detail operational procedures and material support requirements. Ultimately, these should be output in the form of structured instructions to ensure the construction team can execute them directly.

[0131] In one embodiment, based on the fluctuation value of resource demand and the cumulative value of risk, a resource allocation method and process adjustment content are generated to guide the next construction cycle, including:

[0132] S51. Establish a short-term decision optimization model with the goal of minimizing resource demand fluctuations and risk accumulation.

[0133] Specifically, a short-term decision optimization model is established with the objective of minimizing resource demand fluctuations and cumulative risk. The core input parameters of the model include the resource demand fluctuation value at the current time step, denoted as ΔR; the cumulative risk value, denoted as S; the estimated basic resource demand for the next construction cycle; and the baseline plan for each work process. The objective function adopts a bi-objective optimization design, controlling resource demand fluctuations within an acceptable range while preventing the continuous growth of cumulative risk. Simultaneously, the model incorporates multiple constraints, including resource supply and demand balance constraints and work process logical association constraints. These constraints ensure that the model output highly matches the actual construction scenario and can directly guide on-site operations.

[0134] S52. Transform the short-term decision optimization model into a constrained linear programming problem, and solve it to obtain the optimal resource input vector and the optimal process state adjustment vector for future time steps.

[0135] Specifically, the short-term decision optimization model is transformed into a constrained linear programming problem. The core of this transformation is to clearly define the decision variables and constraint boundaries. The decision variables are set as the resource input and process state adjustment quantities for future time steps. The resource input corresponds to the planned investment scale of various construction resources, while the process state adjustment quantities are related to the key parameters of process execution. The constraints cover three core categories: the upper limit constraint on resource supply corresponds to the maximum supply capacity of available equipment, personnel teams, and other resources; the process time constraint includes time requirements such as the shortest construction cycle of the process and the completion time limit of the preceding process; and the risk-related constraint reasonably limits the adjustment range of the process during high-risk periods. The linear programming problem is solved using the simplex method or the interior point method, ultimately obtaining the optimal resource input vector and the optimal process state adjustment vector for future time steps. The former includes the target input quantities of various resources, while the latter covers the specific adjustment parameters of each process.

[0136] S53. Decode the optimal resource input vector into specific resource types, quantities, and allocation instructions to form a resource allocation method.

[0137] Specifically, the optimal resource input vector is decoded into a specific resource allocation method. The decoding operation relies on preset resource coding rules. These rules establish a correspondence between vector dimensions and actual resource types. For example, a specific dimension of the vector corresponds to hoisting equipment, while another dimension corresponds to concrete transport vehicles. The values ​​of each dimension of the vector are directly mapped to the required quantity of the corresponding resource. During the decoding process, the real-time status of the project resource library needs to be considered to clarify the resource allocation starting point, optimal transportation route, and arrival time. This information is then integrated with the resource type and required quantity to form a complete allocation instruction, which is ultimately transformed into a directly executable resource allocation method.

[0138] S54. Decode the optimal process state adjustment vector into the process start time offset, parallelism adjustment and logical relationship change to form the process adjustment content.

[0139] Specifically, the optimal process state adjustment vector is decoded into specific process adjustment content, with each dimension of the vector corresponding one-to-one with the adjustment parameters of different processes. The decoded information is summarized into three core adjustment categories: process start time offset, which clarifies the specific duration by which the process is advanced or delayed relative to the original plan; parallelism adjustment, which involves optimizing the process execution mode, such as adjusting two originally sequential processes to partially parallel execution; and logical relationship changes, which are only activated when the risk impact is severe, for temporarily adjusting the order of processes. These three types of adjustment information are integrated according to the process relationships to form a clear and organized set of process adjustment content, ensuring that the construction team can accurately grasp the direction of process optimization.

[0140] In the aforementioned wind power project construction management method, comprehensive data support for risk assessment is provided through the accurate collection and feature fusion of multi-source heterogeneous data; accurate risk identification and assessment are achieved by combining classification models and quantitative algorithms, clarifying risk priorities; a multi-objective optimization model is constructed based on risk data, and dynamic construction plans adapted to the risks are generated through improved algorithms; and resource allocation, process adjustment, and risk response instructions are output through simulation and real-time analysis. The entire process forms a closed-loop management system, improving risk prediction and response capabilities, optimizing resource allocation efficiency, reducing resource idleness and construction delays, providing scientific decision support for project construction, and ensuring the orderly and efficient progress of construction.

[0141] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0142] Based on the same inventive concept, this application also provides a wind power project construction management device for implementing the wind power project construction management method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the wind power project construction management device provided below can be found in the limitations of the wind power project construction management method described above, and will not be repeated here.

[0143] In one exemplary embodiment, such as Figure 2 As shown, a structural schematic diagram of a wind power project construction management device 20 is provided, including:

[0144] The data acquisition and feature extraction module 21 is used to acquire multi-source heterogeneous data of wind power projects, and input the multi-source heterogeneous data of wind power projects into the hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects and obtain the features of wind power projects.

[0145] The project risk assessment module 22 is used to input the characteristics of the wind power project into the preset risk assessment model to obtain project risk assessment data; wherein, the project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event;

[0146] The construction plan dynamic optimization module 23 is used to acquire construction resource data, construction progress node information and construction process logic information, and based on project risk assessment data, dynamically optimize the construction resource data, construction progress nodes and construction process logic information to obtain dynamic construction plan data.

[0147] Project management decision module 24 is used to generate project management decision results based on dynamic construction plan data; the project management decision results include resource allocation methods, process adjustment content and risk response measures.

[0148] Furthermore, the data acquisition and feature extraction module 21 is also used for:

[0149] S11. Acquire geospatial data, environmental time-series data, and material list structured engineering data;

[0150] S12. Input geospatial data, environmental time-series data, and material list structured engineering data into a hybrid deep learning feature extraction model to extract spatial features, time-series features, and resource features;

[0151] The hybrid deep learning feature extraction model includes a convolutional neural network branch, a long short-term memory network branch, a fully connected network branch, and a feature fusion module based on an attention mechanism. The convolutional neural network branch is used to extract spatial features of geospatial data, the long short-term memory network branch is used to extract temporal features of environmental time-series data, and the fully connected network branch is used to extract resource features of structured engineering data of bill of materials.

[0152] S13. Input the spatial features, temporal features and resource features into the feature fusion module to perform feature fusion and obtain the wind power project features.

[0153] Furthermore, the project risk assessment module 22 is also used for:

[0154] S21. Input the characteristics of the wind power project into the preset risk assessment model to generate the probability of occurrence of each risk event type;

[0155] S22. Based on the preset weight coefficients for each risk event type, the preset impact coefficients for each risk event type, and the probability of occurrence of each risk event type, the project risk assessment data is calculated.

[0156] The expression for the project risk assessment data is as follows:

[0157]

[0158] In the formula, This represents project risk assessment data. Indicates the total number of risk event categories. Indicates the first Weighting coefficients for risk-like events Indicates the first The probability of occurrence of risk events. Indicates the first Impact coefficient of risk events.

[0159] Furthermore, the dynamic optimization module 23 for construction plans is also used for:

[0160] S31. Construct a multi-objective optimization function with the goal of minimizing project delay costs and resource idle costs;

[0161] The expression for the multi-objective optimization function is as follows:

[0162]

[0163] In the formula, The overall optimization objective; The cost of project delay is determined by the amount of time delay and the unit time cost caused by high-probability delay risk events in the risk assessment data. The cost of idle resources is determined by the value of resources that are not fully utilized in resource allocation. and These are the weighting coefficients for construction delay costs and resource idle costs, respectively. ;

[0164] S32. Solve the multi-objective optimization function using an improved genetic algorithm to generate a Pareto optimal solution set;

[0165] S33. Based on preset decision preferences, select the optimal solution from the Pareto optimal solution set to obtain the optimized construction resource allocation, schedule node arrangement and process logic relationship;

[0166] S34. Integrate and optimize the allocation of construction resources, schedule nodes, and process logic to generate dynamic construction plan data.

[0167] Furthermore, the project management decision module 24 is also used for:

[0168] S41. Based on the construction resource data, construction progress node information and construction process logic information in the dynamic construction scheme data, construct a simulated construction scenario with multiple discrete time steps.

[0169] S42. In the simulated construction scenario at each discrete time step, based on the project risk assessment data, the Monte Carlo method is used to simulate the construction process and the probability of risk events, and the simulation results of the construction process and the probability of risk events are obtained.

[0170] S43. Perform statistical analysis on the simulation results of the construction process and the simulation results of the probability of risk events to obtain the resource demand fluctuation value and risk accumulation value at each time step;

[0171] S44. Based on the fluctuation value of resource demand and the cumulative value of risk, generate the resource allocation method and process adjustment content to guide the next construction cycle;

[0172] S45. When the cumulative risk value exceeds the preset warning threshold, risk response measures are matched and generated from the preset strategy library based on the risk event category that triggered the threshold.

[0173] S46. Based on resource allocation methods, process adjustments, and risk response measures, generate project management decision results.

[0174] Furthermore, the project management decision module 24 is also used for:

[0175] S51. Establish a short-term decision optimization model with the goal of minimizing resource demand fluctuations and risk accumulation.

[0176] S52. Transform the short-term decision optimization model into a constrained linear programming problem, and solve it to obtain the optimal resource input vector and the optimal process state adjustment vector for future time steps.

[0177] S53. Decode the optimal resource input vector into specific resource types, quantities, and allocation instructions to form a resource allocation method;

[0178] S54. Decode the optimal process state adjustment vector into the process start time offset, parallelism adjustment and logical relationship change to form the process adjustment content.

[0179] In one embodiment, such as Figure 3 A computer device 30 is provided, comprising:

[0180] At least one processor 31, and at least one memory 32 communicatively connected to said processor 31; said memory stores application code executable by said processor, said application code being executed by said processor to enable said processor to perform the steps of a wind power project construction management method as described above;

[0181] Computer equipment may also include: sensor 33;

[0182] Processor 31, memory 32, and sensor 33 can be connected via bus 34 or other means; the diagram shows an example using bus 34. Figure 3 The character is represented by a single thick line, but this does not mean that there is only one bus or a type of bus.

[0183] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0184] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0185] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for construction management of wind power projects, characterized in that, The method includes: S1. Obtain multi-source heterogeneous data of wind power projects, and input the multi-source heterogeneous data of wind power projects into a hybrid deep learning feature extraction model to extract features from the multi-source heterogeneous data of wind power projects and obtain wind power project features. S2. Input the characteristics of the wind power project into a preset risk assessment model to obtain project risk assessment data; wherein, the project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event; S3. Obtain construction resource data, construction progress node information and construction process logic information, and based on the project risk assessment data, dynamically optimize the construction resource data, the construction progress nodes and the construction process logic information to obtain dynamic construction scheme data. S4. Generate project management decision results based on the dynamic construction plan data; the project management decision results include resource allocation methods, process adjustment content, and risk response measures.

2. The method according to claim 1, characterized in that, The multi-source heterogeneous data includes geospatial data, environmental time-series data, and material bill of materials structured engineering data. The process involves acquiring multi-source heterogeneous data from wind power projects and inputting this data into a hybrid deep learning feature extraction model to extract features from the data, thereby obtaining wind power project features, including: S11. Obtain the geospatial data, the environmental time-series data, and the material list structured engineering data; S12. Input the geospatial data, the environmental time series data, and the material list structured engineering data into the hybrid deep learning feature extraction model to extract spatial features, time series features, and resource features; The hybrid deep learning feature extraction model includes a convolutional neural network branch, a long short-term memory network branch, a fully connected network branch, and a feature fusion module based on an attention mechanism. The convolutional neural network branch is used to extract the spatial features of the geospatial data, the long short-term memory network branch is used to extract the temporal features of the environmental time-series data, and the fully connected network branch is used to extract the resource features of the bill of materials structured engineering data. S13. Input the spatial features, temporal features and resource features into the feature fusion module to perform feature fusion and obtain the wind power project features.

3. The method according to claim 1, characterized in that, The step of inputting the characteristics of the wind power project into a preset risk assessment model to obtain project risk assessment data includes: S21. Input the characteristics of the wind power project into a preset risk assessment model to generate the probability of occurrence of each of the risk event types; S22. Based on the preset weight coefficients for each of the risk event types, the preset impact coefficients for each of the risk event types, and the probability of occurrence of each of the risk event types, the project risk assessment data is calculated. The expression for the project risk assessment data is as follows: In the formula, This represents project risk assessment data. Indicates the total number of risk event categories. Indicates the first Weighting coefficients for risk-like events Indicates the first The probability of occurrence of risk events. Indicates the first Impact coefficient of risk events.

4. The method according to claim 1, characterized in that, The process of acquiring construction resource data, construction progress node information, and construction process logic information, and then dynamically optimizing the construction resource data, construction progress nodes, and construction process logic information based on the project risk assessment data to obtain dynamic construction plan data, includes: S31. Construct a multi-objective optimization function with the goal of minimizing project delay costs and resource idle costs; The expression for the multi-objective optimization function is as follows: In the formula, The overall optimization objective; The cost of project delay is determined by the amount of time delay and the unit time cost caused by high-probability delay risk events in the risk assessment data. The cost of idle resources is determined by the value of resources that are not fully utilized in resource allocation. and These are the weighting coefficients for construction delay costs and resource idle costs, respectively. ; S32. Solve the multi-objective optimization function using an improved genetic algorithm to generate a Pareto optimal solution set; S33. Based on preset decision preferences, select the optimal solution from the Pareto optimal solution set to obtain the optimized construction resource allocation, schedule node arrangement and process logic relationship; S34. Integrate the optimized construction resource allocation, schedule arrangement and process logic relationship to generate the dynamic construction plan data.

5. The method according to claim 1, characterized in that, The generation of project management decision results based on the dynamic construction plan data includes: S41. Based on the construction resource data, construction progress node information and construction process logic information in the dynamic construction scheme data, construct a simulated construction scenario with multiple discrete time steps; S42. In the simulated construction scenario at each discrete time step, based on the project risk assessment data, the Monte Carlo method is used to simulate the construction process and the probability of risk events, and the simulation results of the construction process and the probability of risk events are obtained. S43. Perform statistical analysis on the simulation results of the construction process and the simulation results of the probability of risk events to obtain the resource demand fluctuation value and risk accumulation value at each time step; S44. Based on the resource demand fluctuation value and the risk accumulation value, generate resource allocation methods and process adjustment content to guide the next construction cycle; S45. When the cumulative risk value exceeds the preset warning threshold, risk response measures are matched and generated from the preset strategy library based on the risk event category that triggered the threshold. S46. Based on the resource allocation method, the process adjustment content, and the risk response measures, generate the project management decision result.

6. The method according to claim 5, characterized in that, The step of generating resource allocation methods and process adjustments to guide the next construction cycle based on the resource demand fluctuation value and the cumulative risk value includes: S51. Establish a short-term decision optimization model with the goal of minimizing resource demand fluctuations and risk accumulation. S52. Transform the short-term decision optimization model into a constrained linear programming problem, and solve it to obtain the optimal resource input vector and the optimal process state adjustment vector for future time steps. S53. Decode the optimal resource input vector into specific resource types, quantities, and allocation instructions to form the resource allocation method; S54. Decode the optimal process state adjustment vector into the process start time offset, parallelism adjustment and logical relationship change to form the process adjustment content.

7. A wind power project construction management device, used to implement the method described in any one of claims 1-6, characterized in that, The device includes: The data acquisition and feature extraction module is used to acquire multi-source heterogeneous data of wind power projects, input the multi-source heterogeneous data of wind power projects into a hybrid deep learning feature extraction model, and extract features from the multi-source heterogeneous data of wind power projects to obtain wind power project features. The project risk assessment module is used to input the characteristics of the wind power project into a preset risk assessment model to obtain project risk assessment data; wherein, the project risk assessment data is used to characterize various types of risk events and the probability of occurrence of each type of risk event; The construction plan dynamic optimization module is used to acquire construction resource data, construction progress node information and construction process logic information, and based on the project risk assessment data, dynamically optimize the construction resource data, the construction progress nodes and the construction process logic information to obtain dynamic construction plan data. The project management decision module is used to generate project management decision results based on the dynamic construction plan data; the project management decision results include resource allocation methods, process adjustment content, and risk response measures.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images