Artificial intelligence-based engineering cost method and device

Abstract

The embodiment of the application provides a kind of based on artificial intelligence engineering cost method and device, by real-time total engineering project market material price, design drawing and engineering log is synchronized to federal learning network, simultaneously obtain building information model parameter, and based on above-mentioned data constructs dynamic updated engineering digital twin, according to federal learning network constructs including design institute node, construction party node and material node cost prediction model, respectively extracts engineering drawing feature, log time series data and market material price, to predict total engineering cost, compress cost prediction model into lightweight model, receive the real-time data of engineering digital twin and feedback to blockchain notarization, according to current engineering quantity and material demand optimization construction mode, and the optimized construction mode is updated to engineering digital twin.The method effectively solves the accuracy and real-time of engineering cost, significantly improves the accuracy, real-time and efficiency of engineering cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering cost, in particular to an engineering cost method and device based on artificial intelligence. BACKGROUND

[0002] Engineering cost refers to all the funds invested in a certain engineering construction, covering all the costs of one-time expenditure for the realization of planned reproduction of fixed assets, formation of corresponding intangible assets and replenishment of working capital. In engineering cost, it mainly includes but is not limited to equipment and tool purchase cost, construction engineering and installation engineering cost and other engineering construction costs.

[0003] In the prior art, the main way for the management of engineering cost is to build an engineering cost model, analyze the construction data such as building area, structure type, decoration standard, etc., and the construction period data, predict the overall cost of the project and the specific cost of each stage and each sub-item of the project, so as to realize the fine management of engineering cost. However, in the early stage of the project, a large amount of resources need to be invested to collect and analyze various data affecting the engineering cost, and the analyzed data need to be input into the model, which results in that the accuracy and real-time performance of the data collection and analysis link of the engineering cost have a significant impact on the cost result, and the difficulty of cost management increases with the complexity of the project structure. SUMMARY

[0004] In view of the problems in the prior art, the present application provides an engineering cost method and device based on artificial intelligence, which can effectively solve the deficiencies of traditional technology in cost accuracy, real-time performance and cost difficulty increasing with the complexity of the project structure, and significantly improve the accuracy, real-time performance and cost difficulty increasing with the complexity of the project structure of the engineering cost.

[0005] In order to solve at least one of the above problems, the present application provides the following technical solutions:

[0006] In a first aspect, the present application provides an engineering cost method based on artificial intelligence, comprising:

[0007] The market material price, design drawing and engineering log of the total engineering project are synchronized to the federated learning network in real time through the cross-chain protocol of the blockchain, the construction site image collected by the image collection device is received, the boundary information and the attribute parameters corresponding to the boundary information in the construction site image are extracted, the building information model parameters are obtained, and the dynamic updated engineering digital twin is constructed based on the collected market material price and building information model parameters;

[0008] Based on the federated learning network, a cost prediction model including a design institute node, a construction party node and a material node is constructed, wherein the design institute node obtains engineering drawing features based on design drawings, the construction party node obtains log time series data based on engineering logs, and the material node obtains a predicted total price corresponding to the total engineering project based on market material prices;

[0009] The cost prediction model is compressed into a lightweight model, and the cost prediction model receives real-time data in the engineering digital twin to dynamically determine the engineering quantity and material demand required for the total engineering project, and feeds back the determined current engineering quantity and material demand to the blockchain for notarization;

[0010] Based on the current engineering quantity and material demand, the construction risk is evaluated in real time, the geological information and weather information corresponding to the total engineering project are obtained, the construction method of the current total engineering project is optimized based on the construction risk, the geological information and the weather information, and the optimized construction method is updated to the engineering digital twin.

[0011] Further, it also includes: identifying boundary information in the construction site image through a lightweight convolutional neural network model, matching standard component codes in the building information library, and labeling the matching result as attribute parameters corresponding to the boundary information matched successfully based on the standard component codes, the attribute parameters including material type, construction progress and quality defect mark;

[0012] Map the construction site image to the three-dimensional point cloud space to compare the construction site image and the design drawing to obtain deviation data, and adjust the attribute parameters corresponding in the design drawing based on the deviation data;

[0013] Based on the boundary information, the attribute parameters and the design data included in the design drawing, the building information model parameters are obtained.

[0014] Further, it also includes: analyzing the topological relationship in the design drawing through a preset graph neural network, and extracting the geometric attributes and non-geometric attributes in the design drawing, determining the engineering drawing features based on the topological relationship, the geometric attributes and the non-geometric attributes;

[0015] Convert the preset key events in the engineering log into log time series data through a preset natural language model, obtain the health index of the construction equipment, process the log time series data and the health index through a time series convolution network, and obtain the predicted failure probability of each construction equipment;

[0016] Based on the market material price and the required material type and material quantity in the total engineering project, the cost of each material type is estimated by a weighted balance method to determine the predicted total price.

[0017] Further, it also includes: inputting the design drawing into a preset teacher model to make the preset teacher model extract the drawing features of the design drawing;

[0018] The drawing features are input to the preset student model, the preset student model is made to reproduce the attention weight distribution of each channel in the preset teacher model through a lightweight attention mechanism, and redundant filters in the preset student model are removed according to a preset pruning stage.

[0019] Further, the method further comprises: obtaining geological information and weather information corresponding to the total engineering project, and establishing a four-target optimization function based on the construction risk, the geological information and the weather information, wherein the four targets corresponding to the four-target optimization function include a cost target, a construction period target, a quality defect target and a carbon emission target;

[0020] The non-dominated sorting genetic algorithm is used to process the high-dimensional target space in the four-target optimization function to generate a Pareto frontier solution set;

[0021] Based on the Pareto frontier solution set, an optimized construction method is obtained;

[0022] The execution effect of the optimized construction method is simulated by the engineering digital twin, and mechanical conflicts in the process of executing the construction method are detected.

[0023] Further, the method further comprises: predicting the engineering state at the next preset time through the building information model parameter, and determining the corresponding consumption score based on the engineering state at the next preset time, wherein when the consumption score is greater than a preset consumption threshold, the engineering project corresponding to the consumption score is determined as abnormal data;

[0024] The abnormal reason corresponding to the abnormal data is analyzed through the explanation model, and the engineering digital twin is adjusted and updated based on the abnormal data and the abnormal reason.

[0025] Further, the method further comprises: obtaining a carbon footprint coefficient of a specified material, and accessing controller local area network bus data of each construction equipment to determine real-time carbon emissions in the process of executing the total engineering project, wherein the specified material includes cement and steel;

[0026] A carbon cost node is added to the cost prediction model, so that the carbon cost node determines the carbon cost based on the carbon footprint coefficient and the material consumption corresponding to the total engineering project.

[0027] The carbon emissions and the carbon cost are converted into a monetary cost, and the monetary cost is added to the predicted total price.

[0028] In a second aspect, the application provides an artificial intelligence engineering cost device, comprising:

[0029] The acquisition module is configured to synchronize market material prices, design drawings and engineering logs of the total engineering project to the federated learning network in real time through a cross-chain protocol of a block chain, receive construction site images collected by an image collection device, extract boundary information in the construction site images and attribute parameters corresponding to the boundary information, obtain building information model parameters, and construct a dynamically updated engineering digital twin based on the collected market material prices and the building information model parameters.

[0030] The first processing module is configured to construct a cost prediction model including a design institute node, a construction party node and a material node based on the federated learning network, wherein the design institute node obtains engineering drawing features based on the design drawings, the construction party node obtains log time series data based on the engineering logs, and the material node obtains a predicted total price corresponding to the total engineering project based on the market material prices.

[0031] The second processing module is configured to compress the cost prediction model into a lightweight model, and enable the cost prediction model to receive real-time data in the engineering digital twin to dynamically determine engineering quantities and material requirements required by the total engineering project, and feed back the determined current engineering quantities and material requirements to the block chain for notarization.

[0032] The third processing module is configured to evaluate construction risks in real time based on the current engineering quantities and material requirements, acquire geological information and weather information corresponding to the total engineering project, optimize the construction method of the current total engineering project based on the construction risks, the geological information and the weather information, and update the optimized construction method to the engineering digital twin.

[0033] In a third aspect, the present application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to realize the steps of the method.

[0034] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to realize the steps of the artificial intelligence-based engineering cost method.

[0035] In a fifth aspect, the present application provides a computer program product comprising computer programs / instructions, wherein the computer programs / instructions are executed by a processor to realize the steps of the artificial intelligence-based engineering cost method.

[0036] According to the technical solution, the application provides an engineering cost method and device based on artificial intelligence. The market material price, design drawing and engineering log of the total engineering project are synchronized to the federated learning network in real time through the innovative blockchain cross-chain protocol, which can ensure the independence of each group of data and enable the data to be combined with each other. Meanwhile, the construction site image is received, the boundary information and attribute parameters are extracted, the building information model parameters are obtained, and the dynamic updated engineering digital twin is constructed based on the above data. According to the federated learning network, the cost prediction model including the design institute node, the construction party node and the material node is constructed, the engineering drawing features, the log time series data and the market material price are extracted respectively, the total engineering cost is predicted, the cost prediction model is compressed into a lightweight model, the real-time data of the engineering digital twin is received, the engineering quantity and the material demand are dynamically determined, and the data is fed back to the blockchain for storage, thereby ensuring the real-time and accuracy of various data. The construction risk is evaluated in real time according to the current engineering quantity and material demand, the construction method is optimized in combination with the geological information and weather information of the construction site, the adaptability of the engineering cost is improved, the engineering cost is more accurate, the optimized construction method is updated to the engineering digital twin, the engineering digital twin cooperates through the federated learning network, and the overall cooperation efficiency is improved. The method effectively solves the problem that the accuracy and real-time of data collection and analysis link of the engineering cost significantly affect the cost result, and the difficulty of cost management increases with the complexity of the engineering project structure, significantly improves the accuracy, real-time and efficiency of the engineering cost of the complex engineering project, and eliminates the problem of data silos. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 The flowchart of the engineering cost method based on artificial intelligence in the embodiments of the application is shown.

[0039] Figure 2 The structural diagram of the engineering cost device based on artificial intelligence in the embodiments of the application is shown.

[0040] Figure 3 The structural diagram of the electronic device in the embodiments of the application is shown.

[0041] Reference signs:

[0042] The electronic device 9600, the central processing unit 9100, the memory 9140, the communication module 9110, the input unit 9120, the audio processor 9130, the display 9160, the power supply 9170, the buffer memory 9141, the application / function storage section 9142, the data storage section 9143, the driver program storage section 9144, the antenna 9111, the speaker 9131, and the microphone 9132. DETAILED DESCRIPTION

[0043] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0044] The acquisition, storage, use, processing, etc. of data in the technical solutions of the present application comply with relevant provisions of national laws and regulations.

[0045] In view of the problems in the prior art, the present application provides an artificial intelligence-based engineering cost method and device. The market material price, design drawings and engineering log of the total engineering project are synchronized to the federated learning network in real time through an innovative blockchain cross-chain protocol, which can ensure the independence of each group of data and enable the data to be combined with each other. Meanwhile, construction site images are received, boundary information and attribute parameters are extracted, building information model parameters are obtained, and a dynamically updated engineering digital twin is constructed based on the above data. According to the federated learning network, a cost prediction model including a design institute node, a construction party node and a material node is constructed, engineering drawing features, log time series data and market material prices are extracted respectively to predict the total cost of the project, the cost prediction model is compressed into a lightweight model, real-time data of the engineering digital twin is received, the engineering quantity and material demand are dynamically determined, and feedback is fed back to the blockchain for notarization, ensuring the real-time and accuracy of various data. The construction risk is evaluated in real time according to the current engineering quantity and material demand, the construction method is optimized in combination with the geological information and weather information of the construction site, the degree of adaptation to local conditions of the engineering cost is improved, the engineering cost is more accurate, and the optimized construction method is updated to the engineering digital twin, so that the engineering digital twin cooperates through the federated learning network, the overall cooperation efficiency is improved, and the problem of data silos is eliminated.

[0046] In order to effectively solve the influence of the accuracy and real-time of the data collection and analysis link of the traditional technology on the cost result, the difficulty of cost management increases with the complexity of the structure of the engineering project and the shortage of data island, and significantly improves the accuracy, real-time and efficiency of the engineering cost of the complex engineering project. An embodiment of an artificial intelligence engineering cost method is provided, as shown in Figure 1 , the artificial intelligence engineering cost method specifically includes the following contents:

[0047] Step S101: synchronizing the market material price, design drawing and engineering log of the total engineering project to the federated learning network in real time through the cross-chain protocol of the blockchain, receiving the construction site image collected by the image collection device, extracting the boundary information in the construction site image and the attribute parameters corresponding to the boundary information, obtaining the building information model parameters, and constructing a dynamically updated engineering digital twin based on the collected market material price and building information model parameters.

[0048] Optionally, the embodiment connects the building material trading platform in real time through the cross-chain protocol of the blockchain to obtain the spot and futures prices of steel, cement and other main materials. The market material price can also be stored in the alliance chain according to the time stamp to ensure that the market material price cannot be tampered with, while supporting historical backtracking of the material market price. The cross-chain protocol can be a cross-chain message passing scheme (Cross-Chain Message Passing, Polkadot XCMP). The design drawing and the construction log can be uploaded to the blockchain through distributed storage, and a unique content identifier is generated respectively. The unique content identifier and the corresponding design drawing and construction log are written into the blockchain, and the market material price, design drawing and engineering log are synchronized to the federated learning network.

[0049] Receiving the construction site image collected by the image collection device, wherein the image collection device can be a drone. After receiving the construction site image, the boundary information in the construction site image and the attribute parameters corresponding to the boundary information are extracted, wherein the boundary information of the construction site image can be extracted by a joint model composed of an object detection and image segmentation model YOLOv8 and a visual segmentation model SAM (Segment Anything Model), which identifies the component boundary and segments the semantic region. The attribute parameters can be obtained by querying the preset knowledge graph, matching the boundary information with the design parameters, such as boundary ID 2031 and concrete strength C30. The boundary information in the construction site image and the attribute parameters corresponding to the boundary information are determined as the building information model parameters.

[0050] According to the market material price and the building information model parameter, a dynamic updated engineering digital twin is constructed, and the market material price associated in the engineering digital twin can determine the cost of the sub-part engineering in real time, and the construction log and the construction site picture progress data are fused, so that the accuracy and real-time performance of the total engineering project progress are improved.

[0051] The embodiment realizes the non-tamperability and transparency of the market material price, the design drawing and the engineering log through the blockchain cross-chain protocol, and expands the sharing range of the above data, synchronizes the above data to the federated learning network, so that multiple participants can perform model training without sharing the original data, the data privacy is protected, the information asymmetry is further reduced through the synchronous update of the engineering digital twin, and the cooperation efficiency is improved.

[0052] Step S102: based on the federated learning network, a cost prediction model including a design institute node, a construction party node and a material node is constructed.

[0053] The design institute node obtains engineering drawing features based on the design drawing, the construction party node obtains log time sequence data based on the engineering log, and the material node obtains a predicted total price corresponding to the total engineering project based on the market material price.

[0054] Optionally, the embodiment constructs a cost prediction model including multiple nodes based on the federated learning network, so as to realize accurate prediction of the total engineering project cost. The cost prediction model includes a design institute node, a construction party node and a material node.

[0055] The design institute node can extract engineering drawing features of the design drawing by analyzing the input design drawing, and the engineering drawing features include but are not limited to three-dimensional geometric features and component topological relationship, which reflect the design requirements and complexity of the engineering. The construction party node extracts time sequence data in the engineering log according to the engineering log, and the time sequence data represents the engineering progress, various events in the construction process and possible problems, and provides dynamic information in the actual construction process for cost prediction. The material node is used for processing market material price, and the predicted total price corresponding to the total engineering project is predicted by analyzing market material price data.

[0056] Further, after local training of the design institute node, the construction party node and the material node, the encrypted gradient can be uploaded to enhance the privacy of the data.

[0057] The embodiment integrates the data of the design institute node, the construction party node and the material node through the federated learning network, so that the cost prediction model can comprehensively consider multiple factors such as design, construction and material cost, provide more accurate predicted total price, and improve the generalization ability of the cost prediction model.

[0058] Step S103: compress the cost prediction model into a lightweight model, and make the cost prediction model receive real-time data in the engineering digital twin to dynamically determine the engineering quantity and material demand required for the total engineering project, and feed back the determined current engineering quantity and material demand to the blockchain for notarization.

[0059] Optionally, in order to improve the efficiency and adaptability of the cost prediction model, the cost prediction model is compressed into a lightweight model, so that the lightweight model can process various data more quickly while maintaining a high-precision predicted total price, and adapt to real-time updating requirements.

[0060] The lightweight model can receive real-time data in the engineering digital twin, so as to dynamically determine the engineering quantity and material demand required for the total engineering project, and dynamically update the mechanism to enable the predicted total price output by the cost prediction model to timely reflect the actual progress of the project and market changes.

[0061] Further, the determined current engineering quantity and material demand are fed back to the blockchain for notarization, so as to ensure the authenticity and reliability of the current engineering quantity and material demand through the tamper-proof feature of the blockchain.

[0062] The embodiment realizes a lightweight cost prediction model, so that the cost prediction model can efficiently operate in a resource-limited environment, improves the response speed and real-time performance of the cost prediction model, and enables the cost prediction model to receive real-time data in the engineering digital twin in real time, thereby improving the real-time performance.

[0063] Step S104: real-time evaluation of construction risk based on current engineering quantity and material demand, obtaining geological information and weather information corresponding to the total engineering project, optimizing the construction method of the current total engineering project based on the construction risk, geological information and weather information, and updating the optimized construction method to the engineering digital twin.

[0064] Optionally, the embodiment can real-time evaluate the construction risk according to the current determined engineering quantity and material demand, and at the same time, obtain the geological information and weather information corresponding to the total engineering project, i.e., the geological information and weather information corresponding to the construction site. The geological information can be obtained through optical character recognition (OCR) according to the geological exploration report of the construction site. The weather information can also include meteorological and hydrological information.

[0065] For example, according to the geological information, the leakage probability is determined by Monte Carlo simulation of the penetration path, and the reinforcement cost is determined. According to the weather information, the future precipitation is predicted to determine whether to trigger the emergency plan, etc., so as to avoid the influence of geological conditions and weather changes on the construction progress and safety.

[0066] In combination with the construction risk, the geological information and the weather information, the construction mode of the current total engineering project is optimized, and the optimized construction mode is updated to the engineering digital twin, so as to ensure that the digital twin reflects the latest construction strategy and plan in real time.

[0067] The dynamic optimization mechanism based on real-time data in this embodiment can effectively reduce the construction risk, improve the construction efficiency and quality, and ensure that the engineering project can be completed on time and with quality.

[0068] Through the above technical innovations, the embodiment realizes efficient management and sharing of various data involved in the engineering cost, can accurately and timely obtain the predicted total price through the cost prediction model, and can update and display the construction progress in real time through the dynamic engineering digital twin and evaluate and optimize the construction risk, thereby improving the management efficiency and quality of the engineering project and improving the accuracy of the predicted total price.

[0069] In some embodiments, boundary information in the construction site image and attribute parameters corresponding to the boundary information are extracted to obtain building information model parameters, including:

[0070] The boundary information in the construction site image is identified through the lightweight convolutional neural network model, the standard component code in the building information library is matched, and the matching result is labeled as the attribute parameters corresponding to the matching successful boundary information based on the standard component code. The attribute parameters include material type, construction progress and quality defect mark;

[0071] The construction site image is mapped to a three-dimensional point cloud space to compare the construction site image and the design drawing to obtain deviation data, and the attribute parameters corresponding to the design drawing are adjusted based on the deviation data;

[0072] Based on the boundary information, the attribute parameters and the design data included in the design drawing, building information model parameters are obtained.

[0073] Optionally, the embodiment analyzes the construction site image through the lightweight convolutional neural network model to identify the boundary information therein. The analysis of the construction site image can be to extract key feature points in the construction site image to identify the boundary information. The lightweight convolutional neural network model can be a CNN model (Convolutional Neural Network). Through the lightweight convolutional neural network model, the consumption of computing resources can be reduced while maintaining a high recognition accuracy.

[0074] In addition, the identified boundary information is matched with a standard building code in a building information database. For boundary information that matches successfully, the corresponding attribute parameters are labeled according to the standard building code. The building information database stores standard codes of various building components and corresponding attribute parameters thereof. The attribute parameters include, but are not limited to, material types, construction progress, and quality defect markers, to ensure that each boundary information in the construction site image can be accurately labeled.

[0075] In addition, the construction site image is mapped to a three-dimensional point cloud space. The mapping process can be achieved by a dense point cloud technology, such as an ORB-SLAM3 algorithm (a visual sparse algorithm supporting multiple sensors), which can generate a dense three-dimensional point cloud from a two-dimensional image. Based on the construction site image, detailed three-dimensional spatial information is provided for the engineering digital twin.

[0076] By comparing the point cloud data generated from the construction site image and the design drawing, deviation data between the two is calculated. According to the deviation data, the corresponding attribute parameters in the design drawing are adjusted, so that the design institute node can be updated in real time according to the changes in the construction site image.

[0077] In addition, according to the extracted boundary information, the labeled attribute parameters, and the design data in the design drawing, a complete building information model parameter is constructed. The above-mentioned parameters include geometric information, and also cover material types, construction progress, and quality defects. The above-mentioned parameters are integrated into the building information model to realize real-time monitoring and dynamic updating of the construction site, and to provide accurate data support for the whole life cycle management of the general engineering project.

[0078] Through the above technical innovation, the application of the lightweight convolutional neural network model makes the extraction of boundary information more efficient. At the same time, through the matching with the standard component code, the attribute parameters can be quickly and accurately labeled, reducing the error of manual operation, improving the accuracy and reliability of the data, and providing accurate data support for the whole life cycle management of the general engineering project through the processing of the construction site image, which can also improve the overall efficiency.

[0079] In some embodiments, the construction cost prediction model is constructed by:

[0080] The topology relationship in the design drawing is parsed by a preset graph neural network, and the geometric attributes and non-geometric attributes in the design drawing are extracted. Based on the topology relationship, the geometric attributes, and the non-geometric attributes, the engineering drawing features are determined.

[0081] The preset key events in the engineering log are converted into log time series data by a preset natural language model, the health degree index of the construction equipment is obtained, the log time series data and the health degree index are processed by a time series convolution network, and the predicted failure probability of each construction equipment is obtained.

[0082] Based on the market material price and the required material types and material quantities included in the total engineering project, the cost of each material type is estimated by a weighted balance method to determine the predicted total price.

[0083] Optionally, the embodiment analyzes the design drawing by a preset graph neural network, extracts the topological relationship in the design drawing, and extracts the geometric properties and non-geometric properties in the design drawing, and comprehensively determines the construction drawing features of the construction drawing according to the topological relationship, the geometric properties and the non-geometric properties. It can be a weighted fusion of topological relationship, geometric properties and non-geometric properties through attention mechanism to obtain engineering drawing features.

[0084] Among them, the topological relationship represents the connection relationship between elements in the drawing, including but not limited to components and nodes, and the graph neural network can update the representation of the node through the message passing mechanism, thereby capturing the complex dependency between nodes. Geometric properties include size, shape, position and other information of components, such as length, area, volume, spatial coordinates and curvature, etc. Non-geometric properties include material type, component purpose, construction requirements and other information, such as material specification (such as concrete grade, steel strength), process requirements (such as corrosion coating thickness), cost coding, etc.

[0085] In addition, the preset key events in the engineering log are converted into log time series data by a preset natural language processing (NLP) model, wherein the preset natural language model can be an NLP (Neuro-Linguistic Programming) model, which can understand the text information in the engineering log and convert it into structured time series data.

[0086] Among them, the preset natural language model can classify engineering events by event types in the engineering log, such as mechanical failure, personnel change, weather influence, etc. After classification, the key entities are extracted and converted into structured time series labels to obtain log time series data.

[0087] The health degree index of the construction equipment is a parameter for evaluating the running state of the equipment, and the log time series data and the health degree index are processed by a time series convolution network, wherein the health degree index can be obtained by sensor data or maintenance records of the construction equipment in the construction site, and the time series convolution network can be a TCN (Temporal Convolutional Networks). TCN can effectively capture the local dependency in time series and extract features through convolution operation.

[0088] The TCN can include a causal dilated convolution and a residual connection to prevent gradient disappearance, and can process log time series data and health indicators through a time series convolution network to obtain a predicted failure probability of each construction equipment. Further, if the predicted failure probability exceeds a preset failure probability, a prompt message can be sent so that relevant personnel of the overall engineering project can timely understand the failure situation.

[0089] In addition, the market material price and the required material types and material quantities included in the overall engineering project are received, weights are assigned to each material type, and the weights are adjusted in stages according to each stage of the overall engineering project. For example, the weight of the main material can be set to 0.6, and the weight of the auxiliary material can be set to 0.4. The main material can include but is not limited to steel and cement, and the auxiliary material can include but is not limited to a water stop and an anchor rod.

[0090] Further, a risk coefficient can be added to the predicted total price. For example, the risk coefficient of geological risk can be set in the interval of 1.05 to 1.15, and the risk coefficient of supply chain risk can be set in the interval of 1.0 to 1.2. This reduces the fluctuations in the predicted total price caused by geological reasons and supply chain reasons at the construction site, and can more accurately reflect the actual cost of the overall engineering project.

[0091] Through the above technical innovations, the topological relationship, geometric properties and non-geometric properties of the design drawing are extracted by the graph neural network, which can more comprehensively understand the drawing information contained in the design drawing, improve the accuracy of the predicted total price, and dynamically adjust the predicted total price by real-time processing of the construction log and the health indicators of the construction equipment. The dynamic adaptability of the cost prediction model is enhanced, the cost overrun caused by construction changes is reduced, and the predicted total price is determined by a weighted balancing method, which can reasonably allocate resources according to market conditions and project needs.

[0092] In some embodiments, the cost prediction model is compressed into a lightweight model, including:

[0093] The design drawing is input into a preset teacher model to enable the preset teacher model to extract drawing features of the design drawing;

[0094] The drawing features are input into a preset student model, the preset student model reproduces the attention weight distribution of each channel in the preset teacher model through a lightweight attention mechanism, and redundant filters in the preset student model are removed according to a preset pruning stage.

[0095] Optionally, the embodiment inputs the design drawing into a preset teacher model, wherein the preset teacher model can adopt a 50-layer deep residual network ResNet50 as a basic network, and the last convolutional output feature map. In the feature extraction stage, the preset teacher model processes the key areas in the design drawing, such as the steel bar dense area, through a spatial attention module, and integrates low-level details, such as cracks, and high-level semantics, such as structure types, through a feature pyramid network.

[0096] Further, before inputting the design drawing into the preset teacher model, the design drawing can be parsed and converted into a standardized tensor, and the training samples can be expanded through data augmentation to improve the robustness of the preset teacher model.

[0097] In addition, the drawing features are input into a preset student model to construct a preset teacher model-preset student model framework, and an attention transfer loss (Attention Transfer) is introduced into the preset teacher model to force the preset student model to reproduce the attention weight distribution of the convolutional layer of the preset teacher model, wherein the backbone network of the preset student model can be a small version of the lightweight network MobileNetV3-Small, and the last convolutional layer thereof can be replaced with a depth separable convolution, and an adaptive layer can be added to align the feature dimension of the preset teacher model.

[0098] In addition, the preset student model is processed through a lightweight attention mechanism, i.e., a simplified attention channel is generated through convolution, and the above-mentioned convolutional block attention module (CBAM) can be replaced with coordinate attention to reduce the computational amount of the preset student model, so that it becomes a lightweight model. The preset student model processed through the lightweight attention mechanism is pruned according to a preset pruning stage, wherein the preset pruning stage can be a three-stage pruning, the first pruning stage removes low-activation filters, the second pruning stage prunes redundant channels, and the third pruning stage quantizes the remaining filters to a preset shaping number, which can improve the distillation loss weight and increase the learning rate of the preset student model.

[0099] Further, after each stage pruning, the test is performed on the validation set, and if the accuracy loss exceeds a preset accuracy threshold, the pruning process of the preset student model is rolled back to the last pruning stage and the pruning rate is reduced.

[0100] The above technical innovations compress the cost prediction model into a lightweight model, reduce the demand for storage resources and computing resources of the cost prediction model, and retain high performance.

[0101] In some embodiments, geological information and weather information corresponding to the total engineering project are obtained, and the construction mode of the current total engineering project is optimized based on the construction risk, the geological information and the weather information, including:

[0102] Geological information and weather information corresponding to the total engineering project are obtained, and a four-objective optimization function is established based on the construction risk, the geological information and the weather information, the four objectives corresponding to the four-objective optimization function including a cost objective, a duration objective, a quality defect objective and a carbon emission objective;

[0103] The high-dimensional objective space in the four-objective optimization function is processed by a non-dominated sorting genetic algorithm to generate a Pareto frontier solution set;

[0104] Based on the Pareto frontier solution set, an optimized construction mode is obtained;

[0105] The execution effect of the optimized construction mode is simulated by an engineering digital twin, and mechanical conflicts in the process of executing the construction mode are detected.

[0106] Optionally, the geological information and weather information corresponding to the total engineering project are obtained, wherein the geological information can be obtained through a geological cloud platform, which includes scientific data of geological environment and geological disaster investigation to obtain detailed geological information of the area where the total engineering project is located, such as soil type, groundwater level, seismic activity, etc. The weather information is real-time weather data of the location of the total engineering project, and historical weather data and weather forecast data can also be obtained, including but not limited to temperature, precipitation probability, wind speed and weather warning, etc., which are used to evaluate the impact of weather on construction progress and safety.

[0107] The four-objective optimization function is established according to the construction risk, the geological information and the weather information, wherein the four-objective optimization function includes the cost objective, the duration objective, the quality defect objective and the carbon emission objective. The cost objective is used to minimize the total cost of the project, including material cost and labor cost, etc. The duration objective is used to shorten the total duration of the project to ensure that the total engineering project is completed on time. The quality defect objective is used to reduce quality defects in the construction process to improve engineering quality. The carbon emission objective is used to reduce carbon emissions in the construction process to meet environmental protection requirements.

[0108] In order to effectively process the high-dimensional objective space in the four-objective optimization function, a non-dominated sorting genetic algorithm can be used for processing and generating a Pareto frontier solution set, wherein the non-dominated sorting genetic algorithm (NSAG-II) is a high-efficiency multi-objective optimization algorithm that simulates the natural selection process to gradually optimize the solution set and find the optimal trade-off between multiple objectives. In the process of processing the four-objective optimization function, the following steps can be implemented:

[0109] Initialization of population: generate a set of random solutions as the initial population;

[0110] Non-dominated sorting: sort the population according to the non-dominated relationship of solutions, and divide the solutions into different fronts;

[0111] Crowding degree calculation: calculate the crowding degree of solutions within each front to maintain the diversity of the population;

[0112] Selection operation: select the next generation population according to the non-dominated sorting and crowding degree;

[0113] Crossover and mutation: perform crossover and mutation operations on the selected population to generate new solutions;

[0114] Iterative optimization: repeat the above steps until a preset number of iterations is reached or a convergence condition is met, and obtain the Pareto front solution set.

[0115] In addition, from the Pareto front solution set, according to the specific needs and priorities of the project, one or more optimized construction methods are selected, and according to the optimized construction method, simulation and verification are performed through the engineering digital twin, wherein the mechanical scheduling table and process time in the optimized construction method can be input into the engineering digital twin to obtain conflict detection and progress detection, wherein the conflict detection is, for example, the interference warning of tower crane and concrete pump truck arm, and the progress detection can be, for example, the Gantt chart of weather delay.

[0116] Further, the actual progress can be detected by taking pictures of the construction site by a drone, and the simulation deviation between the actual progress and the engineering digital twin is compared to trigger NSGA-II re-optimization to update the construction method, and the simulation parameters, optimization decisions and execution results can also be recorded through the cross-chain protocol of the blockchain.

[0117] Through the above technical innovations, the embodiment obtains geological information and weather information, combines with construction risk assessment, establishes a four-objective optimization function, and uses non-dominated sorting genetic algorithm to process high-dimensional target space to obtain an optimized construction method. The execution effect of the optimized construction method is simulated through the engineering digital twin, and mechanical conflicts are detected to ensure the feasibility and efficiency of the construction method, thereby improving the scientificity and accuracy of engineering management.

[0118] In some embodiments, the optimized construction method is updated to the engineering digital twin, including:

[0119] The next preset time of the engineering state is predicted through the building information model parameters, and the corresponding consumption score is determined based on the next preset time of the engineering state. When the consumption score is greater than the preset consumption threshold, the engineering project corresponding to the consumption score is determined as abnormal data;

[0120] The abnormal reason corresponding to the abnormal data is analyzed through the explanation model, and the engineering digital twin is adjusted and updated based on the abnormal data and the abnormal reason.

[0121] Optionally, the embodiment can predict the engineering state at the next preset time through the building information model parameter, and the engineering state at the next time can be predicted through a machine learning algorithm or a simulation model. The building information model includes engineering information such as geometric properties of components, material types, construction progress, etc. The prediction model can be Kalman filtering, which predicts the engineering state at the next time through a corresponding state equation and an observation equation.

[0122] In addition, after predicting the engineering state at the next time, the resource consumption of the engineering needs to be evaluated to determine the consumption score, wherein the consumption score is used to evaluate the resource consumption of the current engineering at the next time, and the resource consumption includes but is not limited to material consumption, energy consumption and manpower consumption. Through a preset scoring rule or model, the consumption score is determined according to the predicted engineering state, for example, different weights are set for different consumption types, and the consumption scores of different types of resource consumption are weighted and summed to obtain the consumption score. For example, the engineering state includes material cost deviation, construction delay and quality defect, the weight of the material cost deviation is set to 0.4, the weight of the construction delay is set to 0.3, and the weight of the quality defect is set to 0.3. The consumption score is obtained by weighted summation.

[0123] In addition, a preset consumption threshold is obtained, for example, the preset consumption threshold can be divided into three levels, the first level is a normal consumption range, the value is less than or equal to 0.1, the second level is a mild abnormal consumption range, the value is greater than 0.1 and less than or equal to 0.3, and the third level is a severe abnormal consumption range, the value is greater than 0.3. When the consumption score is greater than the preset consumption threshold, the engineering project corresponding to the consumption score is determined as abnormal data, that is, when the consumption score is greater than 0.3, the corresponding engineering project is marked as abnormal data.

[0124] In addition, the abnormal reason corresponding to the abnormal data is analyzed according to the explanation model, which can be input into the knowledge graph for reasoning and determination according to the abnormal data and the abnormal reason. The engineering digital twin is adjusted and updated, wherein the abnormal reason includes but is not limited to material waste, low construction efficiency or construction equipment failure, etc.

[0125] According to the abnormal reason, a corresponding adjustment strategy is developed, for example, the abnormal reason is material waste, which can be solved by optimizing the construction process or adjusting the material supply to solve the abnormal reason. The adjusted information is updated to the engineering digital twin to ensure that the engineering digital twin can reflect the real-time state of the total engineering project in real time.

[0126] The embodiment can update the optimized construction mode to the engineering digital twin, thereby monitoring the engineering state in real time, discovering and processing abnormal conditions in time, and ensuring the progress of the total engineering project.

[0127] In some embodiments, further comprising:

[0128] The carbon footprint coefficient of the specified material is obtained, and the controller area network bus data of each construction equipment is accessed to determine the real-time carbon emission in the execution of the total engineering project, and the specified material includes cement and steel;

[0129] A carbon cost node is added to the cost prediction model, so that the carbon cost is determined based on the carbon footprint coefficient and the material consumption corresponding to the total engineering project;

[0130] The carbon emission and the carbon cost are converted into a monetary cost, and the monetary cost is added to the predicted total price.

[0131] Optionally, the embodiment obtains the carbon footprint coefficient of the specified material, and accesses the controller area network bus (CAN bus) data of each construction equipment, thereby determining the real-time carbon emission in the execution of the total engineering project, wherein the specified material includes cement and steel, and the CAN bus can collect operation data of construction equipment including but not limited to excavators, tower cranes and concrete pump trucks to determine the real-time carbon emission, and the operation data includes but is not limited to energy consumption, operation time and other information of the construction equipment.

[0132] In addition, a carbon cost node is added to the cost prediction model to determine the carbon cost according to the carbon footprint coefficient and the material consumption corresponding to the total engineering project, so as to consider the economic impact of carbon emission in the consideration range of the engineering cost.

[0133] Further, the carbon emission can be involved in the Pareto frontier solution as the fourth target to generate a low-carbon construction mode.

[0134] In addition, the carbon emission and the carbon cost are converted into a monetary cost, the monetary cost of the carbon emission can be determined according to the carbon emission and the price of the current carbon trading market, and the monetary cost is added to the predicted total price, for example, the carbon emission is 100 tons of carbon dioxide equivalent (CO2e), and the carbon trading price is 50 yuan per ton, so the corresponding monetary cost is 5000 yuan.

[0135] The embodiment obtains the carbon footprint coefficient of the specified material, accesses the CAN bus data of the construction equipment, determines the implementation carbon emission, and adds the monetary cost corresponding to the carbon cost to the predicted total price, which can more comprehensively evaluate the carbon emission and economic impact of the total engineering project in the execution process, support the cost and evaluation of green construction, and improve the sustainability of the engineering project.

[0136] In order to effectively solve the influence of the accuracy and real-time of the data collection and analysis link of the traditional technology on the cost result, and the difficulty of cost management increases with the complexity of the structure of the engineering project and the shortage of data island, the present application provides an embodiment of a construction cost device based on artificial intelligence for realizing all or part of the content of the construction cost method based on artificial intelligence, see Figure 2 The construction cost device based on artificial intelligence specifically includes the following contents:

[0137] The acquisition module 10 is used to synchronize the market material price, design drawing and engineering log of the total engineering project to the federated learning network in real time through the cross-chain protocol of the block chain, receive the construction site image collected by the image collection device, extract the boundary information in the construction site image and the attribute parameters corresponding to the boundary information, obtain the building information model parameters, and construct a dynamically updated engineering digital twin based on the collected market material price and building information model parameters;

[0138] The first processing module 20 is used to construct a cost prediction model including a design institute node, a construction party node and a material node based on the federated learning network, wherein the design institute node obtains the engineering drawing features based on the design drawing, the construction party node obtains the log time series data based on the engineering log, and the material node obtains the predicted total price corresponding to the total engineering project based on the market material price;

[0139] The second processing module 30 is used to compress the cost prediction model into a lightweight model, and make the cost prediction model receive real-time data in the engineering digital twin to dynamically determine the engineering quantity and material demand required by the total engineering project, and feed back the determined current engineering quantity and material demand to the block chain for notarization;

[0140] The third processing module 40 is used to evaluate the construction risk in real time based on the current engineering quantity and material demand, obtain the geological information and weather information corresponding to the total engineering project, optimize the construction method of the current total engineering project based on the construction risk, geological information and weather information, and update the optimized construction method to the engineering digital twin.

[0141] As described above, the AI-based engineering cost estimation device provided in this application can synchronize the market material prices, design drawings, and engineering logs of the entire project to the federated learning network in real time through an innovative blockchain cross-chain protocol. This ensures the independence of each data set while enabling data integration. Simultaneously, it receives construction site images, extracts boundary information and attribute parameters to obtain building information model parameters, and constructs a dynamically updated digital twin of the project based on the aforementioned data. According to the federated learning network, it constructs a cost prediction model including design institute nodes, construction party nodes, and material nodes, respectively extracting features from engineering drawings and log data. This method uses sequenced data and market material prices to predict the total project cost. The cost prediction model is compressed into a lightweight model, receiving real-time data from the project's digital twin to dynamically determine the project quantity and material requirements. This data is then fed back to the blockchain for storage, ensuring the real-time nature and accuracy of all data. Construction risks are assessed in real-time based on the current project quantity and material requirements. Construction methods are optimized by combining local geological and weather information, improving the localized nature of project cost estimates and making them more accurate. The optimized construction methods are then updated to the project's digital twin, enabling collaboration through a federated learning network, thus improving overall efficiency. This method effectively addresses the shortcomings of data collection and analysis in project cost estimation, where the accuracy and real-time nature of data collection significantly impact cost results. Furthermore, the complexity of cost management increases with the complexity of project structures. This significantly improves the accuracy, real-time nature, and efficiency of project cost estimation for complex projects, while also eliminating data silos.

[0142] From a hardware perspective, in order to effectively address the shortcomings in the accuracy and real-time performance of data collection and analysis in engineering cost estimation, which significantly impact cost results, and the increasing complexity of cost management as engineering projects become more complex, this application provides an embodiment of an electronic device for implementing all or part of an AI-based engineering cost estimation method. The electronic device specifically includes the following components:

[0143] The system comprises a processor, memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to realize information transmission between the engineering cost estimation device based on acceptable risk and core business systems, user terminals, and related databases and other related devices; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the artificial intelligence-based engineering cost estimation method and the artificial intelligence-based engineering cost estimation device, the content of which is incorporated herein, and repeated details will not be described again.

[0144] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0145] In practical applications, the AI-based engineering cost estimation method can be partially executed on the electronic device side as described above, or all operations can be completed on the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations on this. If all operations are completed on the client device, the client device may further include a processor.

[0146] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0147] Figure 3 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 3 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 3 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0148] In one embodiment, the AI-based engineering cost estimation method function can be integrated into the central processing unit 9100. The central processing unit 9100 can be configured to perform the following control:

[0149] Step S101: Synchronize the market material prices, design drawings and engineering logs of the overall project to the federated learning network in real time through the blockchain cross-chain protocol, receive construction site images collected by the image acquisition device, extract the boundary information and the corresponding attribute parameters in the construction site images, obtain the building information model parameters, and construct a dynamically updated digital twin of the project based on the collected market material prices and building information model parameters.

[0150] Step S102: Based on the federated learning network, construct a cost prediction model including design institute nodes, construction party nodes and material nodes. The design institute node obtains engineering drawing features based on the design drawings, the construction party node obtains log time series data based on the engineering log, and the material node obtains the predicted total price corresponding to the total project based on the market material price.

[0151] Step S103: Compress the cost prediction model into a lightweight model, and enable the cost prediction model to receive real-time data from the engineering digital twin to dynamically determine the total engineering quantity and material requirements required for the project, and feed back the determined current engineering quantity and material requirements to the blockchain for storage.

[0152] Step S104: Based on the current workload and material requirements, assess the construction risks in real time, obtain the geological and weather information corresponding to the overall project, optimize the construction method of the current overall project based on the construction risks, geological information and weather information, and update the optimized construction method to the project digital twin.

[0153] As described above, the electronic device provided in this application embodiment synchronizes the market material prices, design drawings, and project logs of the entire project to the federated learning network in real time through an innovative blockchain cross-chain protocol. This ensures the independence of each data set while enabling data to be combined. Simultaneously, it receives construction site images, extracts boundary information and attribute parameters to obtain building information model parameters, and constructs a dynamically updated digital twin of the project based on the aforementioned data. According to the federated learning network, it constructs a cost prediction model including design institute nodes, construction party nodes, and material nodes, respectively extracting features from engineering drawings, log time-series data, and market material prices. This method uses on-site material prices to predict the total project cost. The cost prediction model is compressed into a lightweight model, receiving real-time data from the project's digital twin. It dynamically determines the project quantity and material requirements, feeding this data back to the blockchain for storage, ensuring the real-time nature and accuracy of all data. Based on the current project quantity and material requirements, construction risks are assessed in real-time. Construction methods are optimized by combining local geological and weather information, improving the localized nature of project cost estimates and making them more accurate. The optimized construction methods are then updated in the project's digital twin, enabling collaboration through a federated learning network, thus improving overall efficiency. This method effectively addresses the shortcomings of data collection and analysis in project cost estimation, where the accuracy and real-time nature of data significantly impact cost results. Furthermore, the complexity of cost management increases with the complexity of project structures. It significantly improves the accuracy, real-time nature, and efficiency of cost estimation for complex projects, while eliminating data silos.

[0154] In another embodiment, the AI-based engineering cost estimation device can be configured separately from the central processing unit 9100. For example, the AI-based engineering cost estimation device can be configured as a chip connected to the central processing unit 9100, and the AI-based engineering cost estimation method function can be realized through the control of the central processing unit.

[0155] like Figure 3 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 3 All components shown; in addition, the electronic device 9600 may also include Figure 3 For components not shown, please refer to existing technologies.

[0156] like Figure 3 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0157] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0158] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0159] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0160] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device for communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0161] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0162] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored audio via the speaker 9131.

[0163] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the AI-based engineering cost estimation method with a server or client as the execution subject in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the AI-based engineering cost estimation method with a server or client as the execution subject in the above embodiments. For example, when the processor executes the computer program, it implements the following steps:

[0164] Step S101: Synchronize the market material prices, design drawings and engineering logs of the overall project to the federated learning network in real time through the blockchain cross-chain protocol, receive construction site images collected by the image acquisition device, extract the boundary information and the corresponding attribute parameters in the construction site images, obtain the building information model parameters, and construct a dynamically updated digital twin of the project based on the collected market material prices and building information model parameters.

[0165] Step S102: Based on the federated learning network, construct a cost prediction model including design institute nodes, construction party nodes and material nodes. The design institute node obtains engineering drawing features based on the design drawings, the construction party node obtains log time series data based on the engineering log, and the material node obtains the predicted total price corresponding to the total project based on the market material price.

[0166] Step S103: Compress the cost prediction model into a lightweight model, and enable the cost prediction model to receive real-time data from the engineering digital twin to dynamically determine the total engineering quantity and material requirements required for the project, and feed back the determined current engineering quantity and material requirements to the blockchain for storage.

[0167] Step S104: Based on the current workload and material requirements, assess the construction risks in real time, obtain the geological and weather information corresponding to the overall project, optimize the construction method of the current overall project based on the construction risks, geological information and weather information, and update the optimized construction method to the project digital twin.

[0168] As described above, the computer-readable storage medium provided in this application embodiment synchronizes the market material prices, design drawings, and project logs of the entire project to the federated learning network in real time through an innovative blockchain cross-chain protocol. This ensures the independence of each data set while enabling data to be combined. Simultaneously, it receives construction site images, extracts boundary information and attribute parameters to obtain building information model parameters, and constructs a dynamically updated digital twin of the project based on the aforementioned data. According to the federated learning network, it constructs a cost prediction model including design institute nodes, construction party nodes, and material nodes, extracting features from the project drawings and time-series data from the logs. Based on market material prices, the total project cost is predicted. The cost prediction model is compressed into a lightweight model, receiving real-time data from the project's digital twin to dynamically determine the project quantity and material requirements. This data is then fed back to the blockchain for storage, ensuring the real-time nature and accuracy of all types of data. Construction risks are assessed in real-time based on the current project quantity and material requirements. Construction methods are optimized by combining local geological and weather information, improving the localized nature of project cost estimates and making them more accurate. The optimized construction methods are then updated to the project's digital twin, enabling collaboration through a federated learning network, thus improving overall efficiency. This method effectively addresses the shortcomings of data collection and analysis in project cost estimation, where the accuracy and real-time nature of data collection significantly impact cost results. Furthermore, the complexity of cost management increases with the complexity of project structures. This method significantly improves the accuracy, real-time nature, and efficiency of project cost estimation for complex projects, while also eliminating data silos.

[0169] Embodiments of this application also provide a computer program product capable of implementing all steps of the AI-based engineering cost estimation method described above, where the execution subject is a server or client. When executed by a processor, this computer program / instruction implements the steps of the AI-based engineering cost estimation method. For example, the computer program / instruction implements the following steps:

[0170] Step S101: Synchronize the market material prices, design drawings and engineering logs of the overall project to the federated learning network in real time through the blockchain cross-chain protocol, receive construction site images collected by the image acquisition device, extract the boundary information and the corresponding attribute parameters in the construction site images, obtain the building information model parameters, and construct a dynamically updated digital twin of the project based on the collected market material prices and building information model parameters.

[0171] Step S102: Based on the federated learning network, construct a cost prediction model including design institute nodes, construction party nodes and material nodes. The design institute node obtains engineering drawing features based on the design drawings, the construction party node obtains log time series data based on the engineering log, and the material node obtains the predicted total price corresponding to the total project based on the market material price.

[0172] Step S103: Compress the cost prediction model into a lightweight model, and enable the cost prediction model to receive real-time data from the engineering digital twin to dynamically determine the total engineering quantity and material requirements required for the project, and feed back the determined current engineering quantity and material requirements to the blockchain for storage.

[0173] Step S104: Based on the current workload and material requirements, assess the construction risks in real time, obtain the geological and weather information corresponding to the overall project, optimize the construction method of the current overall project based on the construction risks, geological information and weather information, and update the optimized construction method to the project digital twin.

[0174] As described above, the computer program product provided in this application embodiment synchronizes the market material prices, design drawings, and project logs of the entire project to the federated learning network in real time through an innovative blockchain cross-chain protocol. This ensures the independence of each data set while enabling data to be combined. Simultaneously, it receives construction site images, extracts boundary information and attribute parameters to obtain building information model parameters, and constructs a dynamically updated digital twin of the project based on the aforementioned data. According to the federated learning network, it constructs a cost prediction model including design institute nodes, construction party nodes, and material nodes, extracting features from the project drawings and time-series data from the logs. This method uses market material prices to predict the total project cost, compressing the cost prediction model into a lightweight model. It receives real-time data from the project's digital twin, dynamically determines the project quantity and material requirements, and feeds this data back to the blockchain for storage, ensuring the real-time nature and accuracy of all data. Based on the current project quantity and material requirements, it assesses construction risks in real time and optimizes construction methods by combining local geological and weather information, improving the localized nature of project cost estimates and making them more accurate. The optimized construction methods are then updated in the project's digital twin, enabling collaboration through a federated learning network, thus improving overall efficiency. This method effectively addresses the shortcomings of data collection and analysis in project cost estimation, where the accuracy and real-time nature of data collection significantly impact cost results, and the difficulty of cost management increases with the complexity of project structures. It significantly improves the accuracy, real-time nature, and efficiency of project cost estimation for complex projects, while eliminating data silos.

[0175] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0176] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0177] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0178] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0179] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An artificial intelligence-based engineering cost estimation method, characterized in that, The method includes: The project's market material prices, design drawings, and project logs are synchronized in real-time to the federated learning network via a blockchain cross-chain protocol. Construction site images captured by image acquisition devices are received. A lightweight convolutional neural network model identifies boundary information in the construction site images, matches it with standard component codes in the building information database, and labels the attribute parameters corresponding to the successfully matched boundary information based on the standard component codes. The construction site images are mapped to a 3D point cloud space to compare them with the design drawings, obtaining deviation data. The corresponding attribute parameters in the design drawings are adjusted based on the deviation data. Based on the boundary information, the attribute parameters, and the design data included in the design drawings, building information model parameters are obtained. A dynamically updated digital twin of the project is constructed based on the collected market material prices and the building information model parameters. The attribute parameters include material type, construction progress, and quality defect markers. Based on the federated learning network, a cost prediction model is constructed, including design institute nodes, construction party nodes, and material nodes. The design institute node obtains engineering drawing features based on the design drawings, the construction party node obtains log time-series data based on the engineering log, and the material node obtains the predicted total price corresponding to the total project based on the market material prices. The cost prediction model is compressed into a lightweight model, and the cost prediction model receives real-time data from the digital twin of the project to dynamically determine the engineering quantity and material requirements required for the total project, and the determined current engineering quantity and material requirements are fed back to the blockchain for storage. Based on the current workload and material requirements, the construction risks are assessed in real time. Geological and weather information corresponding to the overall project is obtained. A four-objective optimization function is established based on the construction risks, geological information, and weather information. The construction method of the current overall project is optimized based on the four-objective optimization function, and the optimized construction method is updated to the digital twin of the project. The four objectives corresponding to the four-objective optimization function include cost objective, schedule objective, quality defect objective, and carbon emission objective.

2. The method according to claim 1, characterized in that, The construction process of the cost prediction model includes: The topological relationships in the design drawings are analyzed by a preset graph neural network, and the geometric and non-geometric attributes in the design drawings are extracted. Based on the topological relationships, the geometric attributes, and the non-geometric attributes, the features of the engineering drawings are determined. The preset key events in the project log are converted into log time series data through a preset natural language model to obtain the health index of the construction equipment. The log time series data and the health index are processed through a temporal convolutional network to obtain the predicted failure probability of each construction equipment. Based on the market material prices and the required material types and quantities included in the total project, the cost of each material type is estimated using a weighted balancing method to determine the predicted total price.

3. The method according to claim 1, characterized in that, The step of compressing the cost prediction model into a lightweight model includes: The design drawings are input into a preset teacher model so that the preset teacher model can extract the drawing features of the design drawings; The drawing features are input into a preset student model. A lightweight attention mechanism is used to make the preset student model reproduce the attention weight distribution of each channel in the preset teacher model. Redundant filters in the preset student model are removed according to a preset pruning stage.

4. The method according to claim 1, characterized in that, The process of obtaining geological and weather information corresponding to the overall project, and optimizing the construction method of the current overall project based on the construction risks, the geological information, and the weather information, includes: Obtain the geological and weather information corresponding to the overall project, and establish a four-objective optimization function based on the construction risk, the geological information, and the weather information. The four objectives of the four-objective optimization function include cost objective, schedule objective, quality defect objective, and carbon emission objective. The high-dimensional objective space in the four-objective optimization function is processed by a non-dominated sorting genetic algorithm to generate a Pareto front solution set; Based on the Pareto front solution set, the optimized construction method is obtained; The optimized construction method is simulated using the engineering digital twin, and mechanical conflicts during the execution of the construction method are detected.

5. The method according to claim 1, characterized in that, The step of updating the optimized construction method to the project digital twin includes: The project status at the next preset time is predicted by the building information model parameters, and the corresponding consumption score is determined based on the project status at the next preset time. When the consumption score is greater than the preset consumption threshold, the project corresponding to the consumption score is identified as abnormal data. The abnormal data is analyzed by interpreting the model to determine the cause of the abnormality, and the engineering digital twin is adjusted and updated based on the abnormal data and the cause of the abnormality.

6. The method according to claim 1, characterized in that, Also includes: The carbon footprint coefficient of the specified materials is obtained, and the controller local area network bus data of each construction equipment is accessed to determine the real-time carbon emissions during the execution of the total project. The specified materials include cement and steel. A carbon cost node is added to the cost prediction model so that the carbon cost node determines the carbon cost based on the carbon footprint coefficient and the material consumption corresponding to the total project. The carbon emissions and carbon costs are converted into monetary costs, and the monetary costs are added to the predicted total price.

7. An engineering cost estimation device based on artificial intelligence, characterized in that, The device includes: The acquisition module is used to synchronize market material prices, design drawings, and project logs of the entire project to the federated learning network in real time via a blockchain cross-chain protocol. It receives construction site images captured by image acquisition devices, identifies boundary information in the construction site images using a lightweight convolutional neural network model, matches them with standard component codes in the building information database, and labels the attribute parameters corresponding to the successfully matched boundary information based on the standard component codes. The module maps the construction site images to a 3D point cloud space to compare the construction site images with the design drawings, obtains deviation data, and adjusts the corresponding attribute parameters in the design drawings based on the deviation data. Based on the boundary information, the attribute parameters, and the design data included in the design drawings, it obtains building information model parameters. Based on the acquired market material prices and the building information model parameters, it constructs a dynamically updated digital twin of the project, where the attribute parameters include material type, construction progress, and quality defect markers. The first processing module is used to construct a cost prediction model based on the federated learning network, including design institute nodes, construction party nodes, and material nodes. The design institute nodes obtain engineering drawing features based on the design drawings, the construction party nodes obtain log time-series data based on the engineering logs, and the material nodes obtain the predicted total price corresponding to the total project based on the market material prices. The second processing module is used to compress the cost prediction model into a lightweight model, and enable the cost prediction model to receive real-time data from the engineering digital twin to dynamically determine the engineering quantity and material requirements required for the total engineering project, and to feed back the determined current engineering quantity and material requirements to the blockchain for storage. The third processing module is used to assess construction risks in real time based on the current workload and material requirements, obtain geological and weather information corresponding to the overall project, establish a four-objective optimization function based on the construction risks, geological information, and weather information, optimize the construction method of the current overall project based on the four-objective optimization function, and update the optimized construction method to the project digital twin. The four objectives corresponding to the four-objective optimization function include cost target, schedule target, quality defect target, and carbon emission target.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the artificial intelligence-based engineering cost estimation method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the artificial intelligence-based engineering cost estimation method as described in any one of claims 1 to 6.

Images