Construction cost visual dynamic analysis system based on multi-source heterogeneous data identification

Abstract

The present application relates to the technical field of construction cost analysis, and particularly relates to a construction cost visual dynamic analysis system based on multi-source heterogeneous data recognition, which is provided with a UAV module, a data recognition module, a data analysis module, a region recognition and regulation module, and a database, acquires point cloud data and surface images of a building to be recognized through the UAV module, determines whether the building to be recognized has recognition risks through the data recognition module, marks a fuzzy tendency region based on a pointing sub-vector of each recognition sub-region through the data analysis module, determines the feature category of the fuzzy tendency region through the region recognition and regulation module, selects an optimized mode of UAV surveying and mapping, and constructs a twin model through the database and stores it in a cost accounting library. The present application realizes quick determination of whether the building to be recognized has recognition risks, adaptively adjusts the optimized mode of UAV surveying and mapping, and improves the efficiency and reliability of construction cost visual dynamic analysis.

Description

Technical Field

[0001] This invention relates to the field of construction cost analysis technology, and in particular to a construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification. Background Technology

[0002] In the context of increasingly complex construction projects and intensifying market competition, traditional construction cost management models are no longer adequate to meet the core needs of refined control, risk warning, and dynamic decision-making. They commonly suffer from prominent problems such as information silos, weak process control, and reliance on experience for decision-making. Manual inspections struggle to cover high-altitude and concealed areas and are prone to judgment errors, while information transmission is often delayed. In recent years, cutting-edge technologies such as Building Information Modeling (BIM), UAV surveying, and digital twins have been widely applied in construction projects. Using UAVs equipped with multiple sensors for aerial surveying, high-precision 3D models of real-world scenes are rapidly generated and combined with digital twin platforms, significantly improving data acquisition efficiency and the level of intelligent construction scheduling. However, despite the automation of data acquisition achieved by UAVs and other equipment, the complex environment of construction sites remains a challenge. The data collected by drones is often subject to various variations, such as reflections from glass curtain walls and obstructions from irregularly shaped structures. These variations often result in fixed flight paths, failing to dynamically adjust flight parameters based on regional characteristics. This leads to insufficient data quality in complex structures or high-risk areas, and the collected point cloud and image data are prone to omissions or errors. Existing systems lack the ability to diagnose data quality in real time during flight data collection, and problems are often only discovered during post-processing. Re-collection at this point is costly, undermining the foundation for subsequent modeling. Defective raw data directly distorts the reconstructed digital twin model, making the model-based automated component identification and quantity extraction results unreliable. This impacts the reliability of the construction cost visualization and dynamic analysis system. Therefore, improving the efficiency and reliability of construction cost visualization and dynamic analysis is an urgent technical problem to be solved.

[0003] For example, Chinese patent application publication number CN117522160A discloses a construction project cost calculation model and method, including cost management software. This software includes a quota, material, and equipment library editor, data conversion, intelligent pricing, and data comparison analysis functions. Users can add, compile, and maintain quotas through the quota, material, and equipment library editor to form enterprise pricing standards, which can be repeatedly applied to different projects. The data conversion function transforms standard-compiled cost documents into cost documents priced based on enterprise quotas. These cost documents, priced based on enterprise quotas, are cost files. The intelligent pricing function includes AI guidance, batch cloning, and pricing scheme functions. The data comparison analysis function includes comparison of labor, materials, and equipment costs, as well as comparison of individual item amounts. This solution saves enterprises manpower and time costs and establishes an enterprise quota library and pricing scheme, providing data support for subsequent cost calculation work.

[0004] The following problems still exist in the existing technology:

[0005] Existing technologies do not consider that UAVs fly with fixed preset parameters, making them unable to cope with the complex and ever-changing optical interference and geometric occlusion at construction sites. This affects the reliability of modeling data acquisition, and consequently the reliability of construction cost analysis. Existing technologies cannot quickly determine whether there is an identification risk for the building to be identified, nor can they adaptively adjust the optimization method of UAV mapping, thus affecting the efficiency and reliability of construction cost visualization and dynamic analysis. Summary of the Invention

[0006] To address this, the present invention provides a construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification, which overcomes the problems of existing technologies being unable to quickly determine whether a building to be identified has identification risks, and being unable to adaptively adjust the optimization method of UAV mapping, thus affecting the efficiency and reliability of construction cost visualization and dynamic analysis.

[0007] To achieve the above objectives, this invention provides a construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification, comprising:

[0008] The drone module is used to control the drone to move along a preset flight path and acquire point cloud data and surface images of the building to be identified.

[0009] The data recognition module, which is connected to the UAV module, is used to divide the surface of the building to be identified into several recognition areas, acquire surface images of the same recognition area at different times within a preset recognition period, and determine the risk tendency parameter of the recognition area based on the comparison between the surface images, so as to determine whether the building to be identified has a recognition risk.

[0010] The data analysis module is connected to the UAV module and the data recognition module respectively. It is used to divide each recognition area into several recognition sub-regions, determine nonlinear characterization parameters based on the pointing sub-vectors of each recognition sub-region, and mark the fuzzy tendency region according to the comparison between the nonlinear characterization parameters.

[0011] The region identification and control module is connected to the UAV module and the data analysis module respectively. It is used to determine the feature category of the fuzzy tendency region based on the risk tendency parameter of each fuzzy tendency region, select the feature pointing sub-vector group determined by the nonlinear characterization parameter according to the feature category to mark the virtual mutation trajectory, and determine the movement direction of the UAV according to the virtual mutation trajectory.

[0012] The reduction in flight speed of the UAV is determined based on several nonlinear characterization parameters of the fuzzy tendency region and the risk tendency parameter.

[0013] A database, connected to the drone module, is used to construct a twin model of the building to be identified based on the data collected by the drone, and to store the twin model in a cost accounting database.

[0014] The pointing sub-vector is determined based on the point cloud data of the identified sub-region.

[0015] Furthermore, the data recognition module is used to determine the risk propensity parameter of the recognition area, wherein,

[0016] The data recognition module determines several grayscale mean values ​​based on the surface images of the recognition area at different times, and determines the variance of the grayscale mean values ​​as the risk tendency parameter of the recognition area.

[0017] Furthermore, the data recognition module is used to determine whether the building to be identified poses a risk of identification.

[0018] The data identification module determines that the building to be identified has an identification risk based on the judgment condition that the risk tendency parameter of the identification area meets the identification risk condition.

[0019] The risk identification condition is the existence of an identification region where the risk propensity parameter exceeds a preset risk propensity parameter threshold.

[0020] Furthermore, the data analysis module is used to determine nonlinear characterization parameters based on the pointing sub-vectors of each identified sub-region, wherein,

[0021] The data analysis module determines several normal vectors based on the point cloud data of each identification sub-region, and determines the normal vectors as the pointing sub-vectors of the identification sub-region;

[0022] This is used to determine the average angle between the pointing sub-vector of any identification sub-region and the other pointing sub-vectors as the nonlinear characterization parameter of the identification sub-region.

[0023] Furthermore, the data analysis module is used to mark regions with a tendency towards ambiguity, wherein,

[0024] If the fuzziness tendency condition is met, the data analysis module will mark the identification area as a fuzziness tendency area;

[0025] The fuzzy tendency condition is that the mean of the nonlinear characterization parameter exceeds a preset mean threshold.

[0026] Furthermore, the region identification and control module is used to determine the feature category of the region with a tendency to be blurry, wherein,

[0027] The region identification and control module determines the feature category of the ambiguous tendency region as the first feature category based on the determination result that the risk tendency parameter of the ambiguous tendency region does not exceed the preset feature threshold.

[0028] Based on the determination result that the risk tendency parameter of the fuzzy tendency region exceeds a preset feature threshold, the feature category of the fuzzy tendency region is determined to be the second feature category.

[0029] Furthermore, the region identification and control module is used to select the optimization method for the UAV based on the feature category, wherein,

[0030] The region identification and control module determines the feature pointing sub-vector group to mark the virtual mutation trajectory based on the determination result that the feature category is the first feature category, according to the nonlinear characterization parameter, and determines the movement direction of the UAV according to the virtual mutation trajectory;

[0031] The region identification and control module determines the reduction in the flight speed of the UAV based on the determination result that the feature category is the second feature category, according to several nonlinear characterization parameters of the fuzzy tendency region and the risk tendency parameter.

[0032] Furthermore, the region identification and control module is used to determine the feature pointing sub-vector group based on the nonlinear representation parameters, wherein,

[0033] The region identification and control module acquires several nonlinear characterization parameters of the ambiguous tendency region;

[0034] Determine the identification sub-region corresponding to the maximum value of the nonlinear characterization parameter;

[0035] The vector angles between the pointing sub-vectors of the identification sub-region and the other pointing sub-vectors are determined respectively. The two pointing sub-vectors corresponding to the largest vector angle are determined to obtain the feature pointing sub-vector group.

[0036] Furthermore, the region identification and control module is used to mark virtual mutation trajectories, wherein,

[0037] The region identification and control module obtains the identification sub-region corresponding to the feature pointing sub-vector group, and determines the virtual mutation trajectory by connecting the midpoints of the identification sub-regions.

[0038] Furthermore, the area identification and control module is used to determine the direction of movement of the drone and the extent of the decrease in the drone's flight speed, wherein,

[0039] The direction of movement of the drone is perpendicular to the virtual mutation trajectory;

[0040] The decrease in flight speed is positively correlated with the mean of the nonlinear characterization parameter and with the risk tendency parameter. The mean of the nonlinear characterization parameter is the mean of several nonlinear characterization parameters within the fuzzy tendency region.

[0041] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention includes a drone module, a data recognition module, a data analysis module, a region recognition and control module, and a database. The drone module controls the drone to move along a preset flight path, acquiring point cloud data and surface images of the building to be identified. The data recognition module divides the surface of the building into several recognition regions, acquiring surface images of the same recognition region at different times within a preset recognition period. Based on the comparison between surface images, risk tendency parameters for the recognition regions are determined to determine whether the building has identification risk. The data analysis module divides each recognition region into several recognition sub-regions, determining nonlinear characterization parameters based on the directional vectors of each sub-region. Ambiguous tendency regions are marked based on the comparison between nonlinear characterization parameters. The region recognition and control module determines the feature category of each ambiguous tendency region based on its risk tendency parameters, selects an optimization method for drone mapping based on the feature category, and constructs a twin model of the building to be identified based on the data collected by the drone using the database. The twin model is stored in a cost accounting database. Therefore, this invention enables rapid determination of whether a building has identification risk, adaptive adjustment of the drone mapping optimization method, and improved efficiency and reliability of construction cost visualization and dynamic analysis.

[0042] In particular, this invention uses a data recognition module to determine whether a building faces identification risks based on comparisons between surface images. This means that a risk scan of the entire building surface is completed simultaneously during the drone's flight data collection, shifting the time of quality problem detection from after the operation to during the operation itself. This enables immediate risk warnings, avoiding rework due to invalid data collection, saving time and costs. It transforms subjective and vague image evaluations into quantifiable values, making data quality measurable, comparable, and traceable. This eliminates inconsistencies in standards caused by subjective judgments from different projects or operators, improving the scientific and standardized level of project management and quality control. Accurate dynamic analysis of construction costs highly depends on the quality of input data. Risk identification lays the foundation for subsequent analysis and control, ensuring that the original image data used to build the digital twin model has undergone stability verification. This guarantees the accuracy and reliability of subsequent model reconstruction, automatic extraction of quantities, and even cost accounting results. Ultimately, this enables rapid determination of whether a building faces identification risks, improving the efficiency and reliability of dynamic analysis of construction costs.

[0043] In particular, this invention uses a data analysis module to determine nonlinear characterization parameters based on the pointing sub-vectors of each identified sub-region to mark areas prone to ambiguity. It is understood that in traditional construction scenarios, the judgment of structurally complex areas relies on on-site observation or drawing analysis by technicians, lacking quantitative standards and prone to omissions due to experience-based judgments, such as misjudging hidden irregular components or flat surfaces as complex areas. Through vectorized analysis, the abstract structural complexity is transformed into calculable and comparable numerical indicators, achieving standardization and objectification of complex area identification. Before modeling, high-risk geometric areas that may cause modeling voids, distortions, or noise are pre-identified, triggering optimized data acquisition. The point cloud data input to the modeling engine ensures sufficient density and quality even in complex areas, improving the geometric fidelity and integrity of the entire digital twin model from the source. This lays a solid geometric data foundation for accurate engineering quantity extraction and cost accounting based on the model. By using quantitative indicators to accurately mark areas with complex structures that are prone to data collection distortion, the blindness of uniform collection across the entire area can be avoided. Subsequently, UAV flight parameters can be optimized in a targeted manner, optimizing collection only in areas with a tendency to be ambiguous, while maintaining the efficiency of regular collection in other areas. This achieves precise targeting, reduces data storage and processing costs, and ultimately realizes the marking of areas with a tendency to be ambiguous, improving the efficiency and reliability of dynamic analysis of construction costs.

[0044] In particular, this invention, through a region identification and control module, determines a set of feature pointing sub-vectors based on nonlinear representation parameters to mark virtual mutation trajectories when the feature category in a region with a tendency towards ambiguity is the first feature category. The movement direction of the UAV is then determined based on the virtual mutation trajectory. It is understood that traditional fixed flight paths, such as grid lines, are prone to blindness when facing complex geometry; the flight path direction may be parallel to the direction of key features, failing to fully capture information from certain angles and creating modeling blind spots. By marking virtual mutation trajectories and optimizing the path based on key geometric features on site, the effectiveness of data is improved from the source for areas difficult to model accurately, such as irregular curved surfaces, dense decorative components, and complex steel structure nodes. Movement perpendicular to the virtual mutation trajectory provides optimal input data for the 3D reconstruction algorithm, enabling the UAV to traverse major geometric features. The mutated images possess the greatest parallax and rich cross-view texture information, significantly improving the success rate and accuracy of feature matching. This results in improved fidelity of geometric details, edge sharpness, and continuity of complex surfaces in the reconstructed digital twin model, laying a geometric foundation for subsequent automatic extraction of accurate engineering quantities. Optimized acquisition efficiency eliminates the need for uniform high-density scanning of the entire building facade; instead, high-precision directional enhancement acquisition modes are activated only in these areas. For geometrically simple, flat areas, a more efficient conventional mode can be used. Based on feature-based differentiated resource allocation, flight time and battery consumption are saved while ensuring overall modeling accuracy, thus improving overall operational efficiency. Consequently, adaptive adjustments to the optimization methods of UAV mapping are achieved, enhancing the efficiency and reliability of dynamic analysis of construction cost visualization.

[0045] In particular, this invention, through the region identification and control module, determines the reduction in UAV flight speed based on several nonlinear characterization parameters and risk tendency parameters of the fuzzy tendency region when the feature category of the fuzzy tendency region is the second feature category. It can be understood that the second feature category, i.e. the fuzzy tendency region, faces a dual challenge: its own complex structure and relatively unstable light reflection. It requires giving the imaging system a sufficient time window to cope with changes in light and the fine identification of complex geometric structures. The UAV is instructed to reduce its flight speed according to a positive correlation with this index, which has profound engineering implications. In aerial surveying, flight speed determines spatial sampling density and effective observation time at a single point. Reducing speed increases the number of photos taken per unit distance, resulting in denser point cloud scan lines. This provides sufficient data samples for the detailed reconstruction of complex structures. Simultaneously, reducing speed is equivalent to extending the exposure or integration time of the sensor for the same ground feature unit, allowing the camera to attempt multiple exposures to synthesize high dynamic range images and overcome optical instability. The mean value of the nonlinear characterization parameter can characterize the overall structural complexity of the blur-prone area. The higher the mean value of the nonlinear characterization parameter, the more irregular the structure, such as multiple turns and dense irregular nodes. The UAV needs to collect point cloud data and images for a longer time to ensure that the point cloud density meets the standard and the image is distortion-free. The risk tendency parameter can characterize the degree of surface change. The larger the risk tendency parameter, the more necessary it is to reduce the speed to capture denser data. Thus, the optimization method of UAV surveying is adaptively adjusted, improving the efficiency and reliability of construction cost visualization and dynamic analysis. Attached Figure Description

[0046] Figure 1 This is a functional block diagram of the construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification, as described in an embodiment of the present invention.

[0047] Figure 2 This is a flowchart illustrating the logic of the data recognition module in this embodiment of the invention for determining whether a building to be identified has a risk of being identified.

[0048] Figure 3 This is a flowchart illustrating the logic of marking ambiguous regions in the data analysis module of this invention.

[0049] Figure 4 The flowchart shows the logic of selecting the optimal method for the UAV in the region identification and control module of this embodiment of the invention. Detailed Implementation

[0050] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0051] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0052] It should be noted that in the description of this invention, the terms "upper," "lower," "inner," "outer," etc., which indicate the direction or positional relationship, are based on the direction or positional relationship shown in the drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0053] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation" and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0054] Please see Figure 1 The diagram shown is a functional block diagram of the construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification according to an embodiment of the present invention. The construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification according to the present invention includes:

[0055] The drone module is used to control the drone to move along a preset flight path and acquire point cloud data and surface images of the building to be identified.

[0056] Specifically, the embodiments of the present invention do not impose specific limitations on the structure of the UAV module. Preferably, it can be a UAV equipped with a LiDAR scanner, i.e., an airborne LiDAR, used to obtain point cloud data and surface images of the building to be identified. Of course, other forms can also be used to control the UAV to move along a preset flight path, which is widely used in the field of surveying and mapping, etc., and will not be elaborated here.

[0057] The data recognition module, which is connected to the UAV module, is used to divide the surface of the building to be identified into several recognition areas, acquire surface images of the same recognition area at different times within a preset recognition period, and determine the risk tendency parameter of the recognition area based on the comparison between the surface images, so as to determine whether the building to be identified has a recognition risk.

[0058] Specifically, the embodiments of the present invention do not impose specific limitations on the structure of the data identification module. Preferably, it can be a microprocessor to determine whether the building to be identified has identification risks. Of course, other forms can also be used, which will not be elaborated here.

[0059] Specifically, the identification area can be uniformly divided into grids. The area of ​​the identification area is the product of the surface area of ​​the building to be identified and the division factor. The division factor and the preset identification period can be set by those skilled in the art according to the accuracy requirements of the construction cost visualization dynamic analysis. The higher the accuracy requirement, the smaller the value should be. The value range of the division factor can be [0.005, 0.015], and the value range of the preset identification period can be [20, 40], with the unit of interval being seconds. Preferably, the division factor can be 0.01 and the preset identification period can be 30 seconds.

[0060] The data analysis module is connected to the UAV module and the data recognition module respectively. It is used to divide each recognition area into several recognition sub-regions, determine nonlinear characterization parameters based on the pointing sub-vectors of each recognition sub-region, and mark the fuzzy tendency region according to the comparison between the nonlinear characterization parameters.

[0061] Specifically, the embodiments of the present invention do not impose specific limitations on the structure of the data analysis module. Preferably, it can be a microprocessor used to mark the ambiguous tendencies region. Of course, other forms can also be used, which will not be elaborated here.

[0062] Specifically, the identified sub-regions can be uniformly divided in a grid pattern. The area of ​​the identified sub-region is the product of the area of ​​the identified region and the sub-region factor. The sub-region factor can be set by those skilled in the art based on the accuracy requirements of the construction cost visualization dynamic analysis. The higher the accuracy requirement, the smaller the value should be. The value range can be [0.05, 0.1], and preferably, it can be 0.08.

[0063] The region identification and control module is connected to the UAV module and the data analysis module respectively. It is used to determine the feature category of the fuzzy tendency region based on the risk tendency parameter of each fuzzy tendency region, select the feature pointing sub-vector group determined by the nonlinear characterization parameter according to the feature category to mark the virtual mutation trajectory, and determine the movement direction of the UAV according to the virtual mutation trajectory.

[0064] The reduction in flight speed of the UAV is determined based on several nonlinear characterization parameters of the fuzzy tendency region and the risk tendency parameter.

[0065] Specifically, the embodiments of the present invention do not specifically limit the structure of the region identification and control module. Preferably, it can be a processor used in a computer to determine the feature category of the fuzzy tendency region and select the UAV optimization method. Of course, other forms can also be used, which will not be elaborated here.

[0066] A database, connected to the drone module, is used to construct a twin model of the building to be identified based on the data collected by the drone, and to store the twin model in a cost accounting database.

[0067] The pointing sub-vector is determined based on the point cloud data of the identified sub-region.

[0068] Specifically, the embodiments of the present invention do not impose specific limitations on the structure of the database. Preferably, it can be a microprocessor to build a twin model and store it in the cost accounting database. Of course, other forms can also be used, which will not be elaborated here.

[0069] Specifically, after receiving the optimization and adjustment instructions, the UAV module performs local replanning through the built-in flight path planning unit to obtain modeling data of the building to be identified.

[0070] Specifically, the data identification module is used to determine the risk propensity parameter of the identified area, wherein,

[0071] The data recognition module determines several grayscale mean values ​​based on the surface images of the recognition area at different times, and determines the variance of the grayscale mean values ​​as the risk tendency parameter of the recognition area.

[0072] Please see Figure 2 The diagram shown is a flowchart illustrating the logic of the data identification module in an embodiment of the present invention for determining whether a building to be identified poses an identification risk. The data identification module is used to determine whether the building to be identified poses an identification risk.

[0073] The data identification module determines that the building to be identified has an identification risk based on the judgment condition that the risk tendency parameter of the identification area meets the identification risk condition.

[0074] Based on the fact that the risk tendency parameters of the identified area do not meet the criteria for identifying risk, it is determined that the building to be identified does not pose any identification risk.

[0075] The risk identification condition is the existence of an identification region where the risk propensity parameter exceeds a preset risk propensity parameter threshold.

[0076] Specifically, the preset risk propensity parameter threshold is the product of the risk propensity parameter reference value and the risk factor. The risk propensity parameter reference value is the average value of the risk propensity parameter under the same working conditions in historical data. The same working conditions may include building type, climate conditions, drone model, etc. The risk factor can be set by those skilled in the art according to the accuracy requirements of the construction cost visualization dynamic analysis. The higher the accuracy requirement, the smaller the value should be. The value range can be [1.05, 1.15], preferably 1.1.

[0077] Specifically, this invention uses a data recognition module to determine whether a building faces identification risks based on comparisons between surface images. This means that a risk scan of the entire building surface is completed simultaneously during the drone's flight data collection, shifting the time of quality problem detection from after the operation to during the operation itself. This enables immediate risk warnings, avoids rework due to invalid data collection, and saves time and costs. Subjective and vague image evaluations are transformed into quantifiable values, making data quality measurable, comparable, and traceable. This eliminates inconsistencies in standards caused by subjective judgments from different projects or operators, improving the scientific and standardized level of project management and quality control. Accurate dynamic analysis of construction costs highly depends on the quality of input data. Risk identification lays the foundation for subsequent analysis and control, ensuring that the original image data used to build the digital twin model has undergone stability verification. This guarantees the accuracy and reliability of subsequent model reconstruction, automatic extraction of quantities, and even cost accounting results. Ultimately, this enables rapid determination of whether a building faces identification risks, improving the efficiency and reliability of dynamic analysis of construction costs.

[0078] Specifically, it can be understood that the surface of the building to be identified is divided into several identification areas by gridding. For each identification area, instead of analyzing a single image, multiple consecutive surface images of the same area at different times within a preset time period are acquired for dynamic observation. The overall grayscale mean of each image is extracted. This value reflects the overall brightness and darkness of the image at the moment of shooting under the combined influence of lighting, material reflection, and occlusion. The variance of the grayscale mean is determined by the risk tendency parameter, which can characterize the stability of the optical performance of the building surface area within the observation period. The larger the risk tendency parameter, the more drastic and unexpected fluctuations in the brightness and darkness of the image have occurred. This may be due to strong specular reflections such as glass or metal, or dynamically changing shadows such as cloud movement, as well as interference from temporary obstructions. Current UAV mapping may have identification risks, requiring further analysis. Thus, it is possible to quickly determine whether the building to be identified has identification risks, improving the efficiency and reliability of construction cost visualization and dynamic analysis.

[0079] Specifically, the data analysis module is used to determine nonlinear characterization parameters based on the pointing sub-vectors of each identified sub-region, wherein,

[0080] The data analysis module determines several normal vectors based on the point cloud data of each identification sub-region, and determines the normal vectors as the pointing sub-vectors of the identification sub-region;

[0081] This is used to determine the average angle between the pointing sub-vector of any identification sub-region and the other pointing sub-vectors as the nonlinear characterization parameter of the identification sub-region.

[0082] There are no restrictions on the method for constructing the pointing sub-vectors of the identification sub-regions. For example, the normal vector of the plane equation can be obtained by fitting the local plane using algorithms such as the least squares method. This will not be elaborated further.

[0083] Please see Figure 3 The diagram shown is a logical flowchart of the data analysis module marking ambiguous regions according to an embodiment of the present invention. The data analysis module is used to mark ambiguous regions, wherein...

[0084] If the fuzziness tendency condition is met, the data analysis module will mark the identification area as a fuzziness tendency area;

[0085] If the fuzzy tendency condition is not met, the data analysis module will not mark the identification area;

[0086] The fuzzy tendency condition is that the mean of the nonlinear characterization parameter exceeds a preset mean threshold.

[0087] Specifically, the preset mean threshold is the product of the mean reference value and the fuzzy factor. The mean reference value is the average value of the mean of the nonlinear characterization parameter under the same working conditions in historical data. The fuzzy factor can be set by those skilled in the art according to the accuracy requirements of the construction cost visualization dynamic analysis. The higher the accuracy requirement, the smaller the value should be. The value range can be [1.05, 1.2], preferably 1.1.

[0088] Specifically, this invention uses a data analysis module to determine nonlinear characterization parameters based on the pointing sub-vectors of each identified sub-region to mark areas prone to ambiguity. It is understood that in traditional construction scenarios, the judgment of structurally complex areas relies on on-site observation or drawing analysis by technicians, lacking quantitative standards and prone to omissions due to experience-based judgments, such as misjudging hidden irregular components or flat surfaces as complex areas. Through vectorization analysis, abstract structural complexity is transformed into calculable and comparable numerical indicators, achieving standardization and objectification of complex area identification. Before modeling, high-risk geometric areas that may cause modeling voids, distortions, or noise are pre-identified, triggering optimized data acquisition. This collection ensures that the point cloud data input to the modeling engine has sufficient density and quality even in complex areas, improving the geometric fidelity and integrity of the entire digital twin model from the source. This lays a solid geometric data foundation for accurate engineering quantity extraction and cost accounting based on the model. By using quantitative indicators to accurately mark areas where complex structures are prone to data collection distortion, the blindness of uniform collection across the entire area is avoided. Subsequently, UAV flight parameters can be optimized in a targeted manner, optimizing collection only in areas with a tendency to be ambiguous, while maintaining the efficiency of regular collection in other areas. This achieves precise targeting, reduces data storage and processing costs, and thus realizes the marking of areas with a tendency to be ambiguous, improving the efficiency and reliability of dynamic analysis of construction costs.

[0089] Specifically, it can be understood that the structural complexity of building surfaces, such as corners, protrusions, depressions, and irregularly shaped components, is mostly manifested in local detailed features. The division into sub-regions can refine the granularity of structural feature analysis, providing a prerequisite for accurately quantifying structural complexity. Point cloud data is a collection of discrete points in three-dimensional space, and its pointing vectors can represent the core geometric features of surface orientation. The surface of any sub-region can be approximated as a plane, and the pointing vector, i.e., the normal vector, can represent the spatial orientation of that sub-region. Differences in the pointing vectors of different sub-regions can correspond to structural undulations, turns, or abrupt changes in shape. For example, the pointing vectors of adjacent sub-regions of a right-angled component are perpendicular, while those of adjacent sub-regions of a curved component are perpendicular. The domain pointing sub-vectors exhibit a gradually changing angle. The smaller the nonlinear characterization parameter, the closer the surface orientation of the identified sub-regions, and the more obvious the linear features. Conversely, the larger the nonlinear characterization parameter, the greater the difference in surface orientation of the identified sub-regions, and the more complex the structure, such as convex and concave corners and irregular nodes, and the more prominent the nonlinear features. The degree of concentration of the distribution of the nonlinear characterization parameter reflects the structural complexity level of the entire identified area. The more significant the difference in surface orientation of the identified sub-regions within the identified area, the more complex and irregular the overall structure. When UAVs collect data, they are more prone to data acquisition distortion such as image blurring and uneven point cloud density due to insufficient angle adaptation. Thus, the marking of blurry tendency areas is realized, improving the efficiency and reliability of construction cost visualization and dynamic analysis.

[0090] Specifically, the region identification and control module is used to determine the feature category of regions with a tendency to be blurry, wherein,

[0091] The region identification and control module determines the feature category of the ambiguous tendency region as the first feature category based on the determination result that the risk tendency parameter of the ambiguous tendency region does not exceed the preset feature threshold.

[0092] Based on the determination result that the risk tendency parameter of the fuzzy tendency region exceeds a preset feature threshold, the feature category of the fuzzy tendency region is determined to be the second feature category.

[0093] Specifically, the preset feature threshold is the product of the risk tendency parameter reference value and the feature factor. The feature factor can be set by those skilled in the art based on the accuracy requirements of the construction cost visualization dynamic analysis. The higher the accuracy requirement, the smaller the value should be. The value range can be [1.16, 1.25], and preferably, it can be 1.2.

[0094] Please see Figure 4 The diagram shown is a logical flowchart of the region identification and control module selecting the optimization method for a UAV according to an embodiment of the present invention. The region identification and control module is used to select the optimization method for the UAV based on the feature category.

[0095] The region identification and control module determines the feature pointing sub-vector group to mark the virtual mutation trajectory based on the determination result that the feature category is the first feature category, according to the nonlinear characterization parameter, and determines the movement direction of the UAV according to the virtual mutation trajectory;

[0096] The region identification and control module determines the reduction in the flight speed of the UAV based on the determination result that the feature category is the second feature category, according to several nonlinear characterization parameters of the fuzzy tendency region and the risk tendency parameter.

[0097] Specifically, the region identification and control module is used to determine the feature pointing sub-vector group based on the nonlinear representation parameters, wherein,

[0098] The region identification and control module acquires several nonlinear characterization parameters of the ambiguous tendency region;

[0099] Determine the identification sub-region corresponding to the maximum value of the nonlinear characterization parameter;

[0100] The vector angles between the pointing sub-vectors of the identification sub-region and the other pointing sub-vectors are determined respectively. The two pointing sub-vectors corresponding to the largest vector angle are determined to obtain the feature pointing sub-vector group.

[0101] Specifically, the region identification and control module is used to mark virtual mutation trajectories, wherein,

[0102] The region identification and control module obtains the identification sub-region corresponding to the feature pointing sub-vector group, and determines the virtual mutation trajectory by connecting the midpoints of the identification sub-regions.

[0103] Specifically, in this embodiment of the invention, when the feature category of a region with a tendency for ambiguity is the first feature category, the region identification and control module determines the feature pointing sub-vector group based on nonlinear representation parameters to mark the virtual mutation trajectory. The movement direction of the UAV is then determined based on the virtual mutation trajectory. It is understood that traditional fixed flight paths, such as grid lines, are prone to blindness when facing complex geometry; the flight path direction may be parallel to the direction of key features, failing to fully capture information from certain angles and creating modeling blind spots. By marking the virtual mutation trajectory and optimizing the path based on key geometric features on site, the effectiveness of data is improved from the source for areas that are difficult to model accurately, such as irregular curved surfaces, dense decorative components, and complex steel structure nodes. Movement perpendicular to the virtual mutation trajectory provides optimal input data for the 3D reconstruction algorithm, enabling it to cross major... Images with geometrical mutations possess the greatest parallax and rich cross-view texture information, significantly improving the success rate and accuracy of feature matching. This results in improved fidelity of geometric details, edge sharpness, and continuity of complex surfaces in the reconstructed digital twin model, laying a geometric foundation for subsequent accurate automatic extraction of engineering quantities. Optimized acquisition efficiency eliminates the need for uniform high-density scanning of the entire building facade; instead, high-precision directional enhancement acquisition modes are activated only in these areas. For geometrically simple, flat areas, a more efficient conventional mode can be used. Based on feature-based differentiated resource allocation, flight time and battery consumption are saved while ensuring overall modeling accuracy, thus improving overall operational efficiency. Furthermore, this enables adaptive adjustments to the optimization of UAV mapping methods, enhancing the efficiency and reliability of dynamic analysis of construction cost visualization.

[0104] Specifically, it can be understood that the fuzzy tendency region of the first feature category, i.e., the surface illumination reflection state is relatively stable, but the structure is more complex, such as irregular components, multi-faceted turning nodes, etc. When collecting data via traditional fixed flight paths, it is easy to cause partial occlusion and poor imaging angles due to the flight direction being parallel to the direction of structural change, such as the inability to capture turning details from the side. From this fuzzy tendency region, the feature pointing sub-vector group is found, which is the position where the geometric orientation changes most drastically in this region. Connecting the centers of these two key sub-regions forms a virtual change trajectory, which can represent the direction of the most important geometric discontinuity feature in this region, i.e., the direction of the most drastic change. When the observation direction, i.e., the flight direction, is perpendicular to the feature change direction, the image acquired by the exposure can simultaneously and clearly capture relatively complete surface information on both sides of the change trajectory. In subsequent motion recovery structure or dense matching algorithms, the corresponding image points crossing this change can be identified and matched to the greatest extent possible, thereby reconstructing the sharp or complex geometric feature in the digital model, avoiding smoothing distortion or breakage. In this way, the optimization method of UAV mapping is adaptively adjusted, improving the efficiency and reliability of construction cost visualization dynamic analysis.

[0105] Specifically, the area identification and control module is used to determine the direction of movement of the drone and the extent of the decrease in the drone's flight speed.

[0106] The direction of movement of the drone is perpendicular to the virtual mutation trajectory;

[0107] The decrease in flight speed is positively correlated with the mean of the nonlinear characterization parameter and with the risk tendency parameter. The mean of the nonlinear characterization parameter is the mean of several nonlinear characterization parameters within the fuzzy tendency region.

[0108] Specifically, the reduction in flight speed is calculated as: First Influence Factor × Mean of Nonlinear Characterization Parameter / Reference Value of Mean of Nonlinear Characterization Parameter + Second Influence Factor × Risk Propensity Parameter / Reference Value of Risk Propensity Parameter. The reference value of the mean of nonlinear characterization parameter is the average of the mean values ​​of nonlinear characterization parameters under the same operating conditions in historical data, and the reference value of the risk propensity parameter is the average of the risk propensity parameter under the same operating conditions in historical data. The first influence factor and the second influence factor can be calculated based on data from several experiments. The value range of the first influence factor can be [0.1, 0.3], and the value range of the second influence factor can be [0.15, 0.3], to avoid the reduction being too large or too small. Preferably, the first influence factor can be 0.2, and the second influence factor can be 0.2.

[0109] Specifically, in this embodiment of the invention, when the feature category of the fuzzy tendency region is the second feature category, the region identification and control module determines the reduction in the UAV's flight speed based on several nonlinear characterization parameters and risk tendency parameters of the fuzzy tendency region. It can be understood that the second feature category, i.e. the fuzzy tendency region, faces a dual challenge: its own complex structure and relatively unstable light reflection. It is necessary to give the imaging system a sufficient time window to cope with changes in light and the fine identification of complex geometric structures. The UAV is instructed to reduce its flight speed according to a positive correlation with this index, which has profound engineering implications. In aerial surveying, flight speed determines spatial sampling density and effective observation time at a single point. Reducing speed increases the number of photos taken per unit distance, resulting in denser point cloud scan lines. This provides sufficient data samples for the detailed reconstruction of complex structures. Simultaneously, reducing speed is equivalent to extending the exposure or integration time of the sensor for the same ground feature unit, allowing the camera to attempt multiple exposures to synthesize high dynamic range images and overcome optical instability. The mean value of the nonlinear characterization parameter can characterize the overall structural complexity of the blur-prone area. The higher the mean value of the nonlinear characterization parameter, the more irregular the structure, such as multiple turns and dense irregular nodes. The UAV needs to collect point cloud data and images for a longer time to ensure that the point cloud density meets the standard and the image is distortion-free. The risk tendency parameter can characterize the degree of surface change. The larger the risk tendency parameter, the more necessary it is to reduce the speed to capture denser data. Thus, the optimization method of UAV surveying is adaptively adjusted, improving the efficiency and reliability of construction cost visualization and dynamic analysis.

[0110] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0111] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification, characterized in that, include: The drone module is used to control the drone to move along a preset flight path and acquire point cloud data and surface images of the building to be identified. The data recognition module, which is connected to the UAV module, is used to divide the surface of the building to be identified into several recognition areas, acquire surface images of the same recognition area at different times within a preset recognition period, and determine the risk tendency parameter of the recognition area based on the comparison between the surface images, so as to determine whether the building to be identified has a recognition risk. The data analysis module is connected to the UAV module and the data recognition module respectively, and is used to divide each recognition area into several recognition sub-regions; Based on the point cloud data of each identification sub-region, several normal vectors are determined. The normal vector of the identification sub-region is determined as the pointing sub-vector of the identification sub-region. The average of the vector angles between the pointing sub-vector of any identification sub-region and the other pointing sub-vectors is determined as the nonlinear characterization parameter of the identification sub-region. If the mean of the nonlinear characterization parameter exceeds a preset mean threshold, the data analysis module will mark the identification region as a region with a tendency to be ambiguous. The region identification and control module is connected to the UAV module and the data analysis module respectively, and is used to determine the feature category of the ambiguous tendency region as the first feature category based on the determination result that the risk tendency parameter of the ambiguous tendency region does not exceed the preset feature threshold. Based on the determination result that the risk tendency parameter of the fuzzy tendency region exceeds a preset feature threshold, the feature category of the fuzzy tendency region is determined to be the second feature category; Based on the determination result that the feature category is the first feature category, several nonlinear characterization parameters of the fuzzy tendency region are obtained, the recognition sub-region corresponding to the maximum value of the nonlinear characterization parameter is determined, the vector angle between the pointing sub-vector of the recognition sub-region corresponding to the maximum value of the nonlinear characterization parameter and the other pointing sub-vectors is determined, the two pointing sub-vectors corresponding to the maximum vector angle are determined, a feature pointing sub-vector group is obtained, the recognition sub-region corresponding to the feature pointing sub-vector group is obtained, the line connecting the midpoints of the recognition sub-regions is determined as the virtual mutation trajectory, the movement direction of the UAV is determined according to the virtual mutation trajectory, and the movement direction of the UAV is perpendicular to the virtual mutation trajectory; Based on the determination result that the feature category is the second feature category, the flight speed reduction of the UAV is determined according to several nonlinear characterization parameters of the fuzzy tendency region and the risk tendency parameter. The flight speed reduction is positively correlated with the mean of several nonlinear characterization parameters in the fuzzy tendency region and positively correlated with the risk tendency parameter. A database, connected to the drone module, is used to construct a twin model of the building to be identified based on the data collected by the drone, and to store the twin model in a cost accounting database.

2. The construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification according to claim 1, characterized in that, The data recognition module is used to determine the risk propensity parameter of the recognition area, wherein, The data recognition module determines several grayscale mean values ​​based on the surface images of the recognition area at different times, and determines the variance of the grayscale mean values ​​as the risk tendency parameter of the recognition area.

3. The construction cost visualization and dynamic analysis system based on multi-source heterogeneous data identification according to claim 2, characterized in that, The data recognition module is used to determine whether the building to be identified poses a risk of identification. The data identification module determines that the building to be identified has an identification risk based on the judgment condition that the risk tendency parameter of the identification area meets the identification risk condition. The risk identification condition is the existence of an identification region where the risk propensity parameter exceeds a preset risk propensity parameter threshold.

Images