Photovoltaic project construction information management system based on industrial internet

Abstract

This invention relates to the field of industrial internet management and control technology, specifically a photovoltaic project construction information management system based on the industrial internet. The system includes: collecting 3D real-scene scanned point cloud data and environmental meteorological time-series data of the photovoltaic construction site; employing an improved spatial semantic segmentation algorithm, dynamically adjusting the local convolution kernel based on the spatial variability of point cloud density and normal vectors, parsing the point cloud data, dividing the site into various regions, and building a structured construction site model with spatial semantic annotations; and utilizing a construction progress simulation engine to couple meteorological conditions and construction techniques to generate a time-series virtual construction event chain, driving the dynamic operation of the 3D model based on the event chain, and outputting a sequence of visualized construction simulation animations. This system improves the digital analysis capabilities of photovoltaic construction sites, realizing a 3D dynamic visualization of the entire construction process.

Description

Technical Field

[0001] This invention relates to the field of industrial internet management and control technology, and in particular to a photovoltaic project construction information management system based on industrial internet. Background Technology

[0002] Photovoltaic projects are mostly deployed in complex outdoor environments. Currently, photovoltaic construction information management generally adopts a management method that combines manual surveying with two-dimensional ledger recording. For the collection and analysis of three-dimensional spatial information of the site, conventional spatial semantic segmentation algorithms are mostly used. These algorithms use local convolutional kernels of fixed specifications to complete the point cloud data operation and processing throughout the process. Photovoltaic construction sites have large terrain undulations and complex surface structures. The point cloud data obtained by real-scene scanning has the characteristics of uneven density distribution and large range of differences in normal vector space. Convolutional kernels with fixed parameters cannot adapt to the operation requirements of point clouds in different areas, which can easily cause confusion in the identification of site elements. It is difficult to accurately distinguish various physical facilities, shading and obstacle areas and undeveloped construction areas within the site. The generated site model lacks semantic annotation content, and the spatial structure information is fragmented.

[0003] Outdoor photovoltaic (PV) construction is highly susceptible to interference from weather factors such as temperature, sunlight, and wind and rain. Existing construction progress simulation methods mostly rely on static calculations based on pre-set fixed procedures, with meteorological monitoring data only recorded and stored independently, failing to integrate with PV-specific construction processes. The temporal arrangement of construction stages cannot dynamically change with environmental conditions, and 3D display modes can only present static site images, unable to simulate the dynamic changes in resource allocation and construction stage progress under different weather conditions. Therefore, the dynamic simulation and display of the entire construction process has shortcomings. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a photovoltaic project construction information management system based on the Industrial Internet.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a photovoltaic project construction information management system based on the Industrial Internet, comprising:

[0006] The data acquisition module acquires 3D real-scene scanning point cloud data and environmental meteorological time series data of the photovoltaic construction site;

[0007] The site analysis module uses an improved spatial semantic segmentation algorithm to analyze the 3D real-scene scanned point cloud data. The improved spatial semantic segmentation algorithm dynamically adjusts the local convolution kernel based on the spatial variability of point cloud density and normal vector. Through the improved spatial semantic segmentation algorithm, construction geographic entities, obstacles and unconstructed areas are identified and separated from the 3D real-scene scanned point cloud data, forming a structured construction site model with spatial semantic annotation.

[0008] The progress simulation module inputs the environmental meteorological time series data into the construction progress simulation engine. The construction progress simulation engine couples meteorological conditions with construction technology to generate a virtual construction event chain sorted by time axis.

[0009] The visualization simulation module, based on the instructions of the virtual construction event chain, drives the dynamic three-dimensional models of the corresponding construction geographic entities and virtual construction resources in the structured construction site model to generate a sequence of visualized construction simulation animations.

[0010] As a further aspect of the present invention, the improved spatial semantic segmentation algorithm dynamically adjusts the local convolution kernel based on the spatial variability of point cloud density and normal vector, including:

[0011] An octree spatial index is constructed on the input 3D real-scene scan point cloud data to quickly query a subset of point clouds within any spatial neighborhood;

[0012] For each point to be processed, the number of points in the neighborhood of its preset radius is counted to calculate the local density of the point cloud, and the variance of the angle between the normal vectors of all points in the neighborhood is calculated to obtain the normal vector variability.

[0013] Construct a dynamic convolution kernel generation function. The input of the dynamic convolution kernel generation function is the local density and normal vector variability of the point cloud, and the output is the shape parameters and weight distribution of the local three-dimensional convolution kernel corresponding to the point to be processed.

[0014] As the local density of the point cloud increases, the dynamic convolution kernel generation function outputs a convolution kernel with a larger spatial receptive field; as the variability of the normal vector increases, the dynamic convolution kernel generation function outputs a convolution kernel with weights more concentrated at the center point.

[0015] The dynamic convolution kernel generation function generates a unique local 3D convolution kernel for each point to be processed, and the local 3D convolution kernel is used to perform weighted aggregation of the features of the neighboring point cloud to complete the context-aware extraction of point cloud features.

[0016] As a further aspect of the present invention, the step of identifying and separating construction geographic entities, obstacles, and unconstructed areas from three-dimensional real-scene scanned point cloud data to form a structured construction site model with spatial semantic annotation includes:

[0017] Using the point cloud features extracted by the improved spatial semantic segmentation algorithm, semantic category prediction is performed on each point in the 3D real-scene scan point cloud. The semantic categories include installed photovoltaic panels, photovoltaic support foundations, construction roads, temporary buildings, natural obstacles, and open spaces awaiting construction.

[0018] Point clouds predicted to have the same semantic category are spatially clustered to form multiple independent 3D point cloud clusters;

[0019] For each 3D point cloud cluster, perform minimum bounding cube fitting to calculate its 3D spatial coordinate range, volume and center point position;

[0020] The semantic category, spatial coordinate range, volume, and center point location information of each 3D point cloud cluster are associated and stored with the original point cloud data, together forming the structured construction site model with clearly defined element attributes in the three-dimensional space.

[0021] As a further aspect of the present invention, the environmental meteorological time series data is input into a construction progress simulation engine. The construction progress simulation engine couples meteorological conditions with construction technology to generate a virtual construction event chain ordered along a time axis, including:

[0022] Analyze environmental meteorological time series data to extract hourly data on light intensity, wind speed, precipitation, and temperature for the future simulation period;

[0023] Establish a construction process knowledge base, which defines the resource types and standard working hours required for each photovoltaic construction procedure, as well as the sensitivity thresholds and influence coefficients of each procedure to various meteorological factors.

[0024] Starting from the current moment, based on the logical relationship of the construction process defined in the construction process knowledge base, the start of each process is simulated sequentially;

[0025] When simulating the start of each process, the meteorological data of the planned execution time of the process is queried. If the meteorological data exceeds the sensitivity threshold defined in the construction process knowledge base, the standard working hours of the process are extended according to the impact coefficient, and the completion time of the process is recalculated.

[0026] The simulated start time, resource requirements, execution location, and simulated completion time after weather adjustment for all processes are arranged in chronological order to form the virtual construction event chain.

[0027] As a further aspect of the present invention, in the structured construction site model, based on the instructions of the virtual construction event chain, the dynamic three-dimensional models of the corresponding construction geographic entities and virtual construction resources are driven to generate a sequence of visualized construction simulation animations, including:

[0028] For each process in the virtual construction event chain, a corresponding virtual construction resource 3D model is configured, which includes crane, transport vehicle, and installation worker models.

[0029] Locate the three-dimensional spatial coordinate range corresponding to the position of the process execution in the structured construction site model;

[0030] In a 3D visualization environment, the corresponding virtual construction resource 3D model is controlled to move to the 3D spatial coordinate range corresponding to the execution position of the process;

[0031] Drive the virtual construction resource 3D model to perform standardized construction operations, and synchronously update the status and appearance of the corresponding construction geographic entities in the structured construction site model to reflect the construction progress;

[0032] According to the time sequence of the virtual construction event chain, the state changes of the three-dimensional model corresponding to all procedures are recorded, and a continuous sequence of the visualized construction simulation animation is generated.

[0033] As a further aspect of the present invention, the system further includes:

[0034] The status synchronization module synchronously collects the actual construction status data reported by the on-site IoT sensor devices. The actual construction status data includes material consumption records, equipment working hour records, and process check-in records.

[0035] The scheduling correction module compares the actual construction status data with the virtual construction event chain data at the same time point to detect whether there is a status deviation pattern. If a status deviation pattern is detected, the adaptive task scheduler is triggered. The adaptive task scheduler, combined with the latest structured construction site model, recalculates the start time, resource requirements and execution location of subsequent virtual construction events.

[0036] The closed-loop update module updates the virtual construction event chain in real time based on the calculation results of the adaptive task scheduler and pushes the update command to the field terminal. Based on the updated virtual construction event chain and the real-time collected actual construction status data, it cyclically executes the closed-loop management process of construction simulation, status comparison and scheduling update.

[0037] The actual construction status data is compared with the virtual construction event chain data at the same time point to detect whether there is a status deviation pattern, including:

[0038] Extract actual process completion time, actual material consumption, and actual equipment working time from actual construction status data;

[0039] Find the planned completion time, planned material consumption, and planned equipment man-hours of the work processes within the same time period from the virtual construction event chain.

[0040] Calculate the difference between the actual completion time and the simulated completion time, calculate the ratio of actual material consumption to planned material consumption, and calculate the ratio of actual equipment working time to planned equipment working time.

[0041] If the difference exceeds the preset time tolerance, or the ratio exceeds the preset quantity tolerance range, then it is determined that the state deviation mode exists in the current time period.

[0042] Record the process type, deviation direction, and deviation amount associated with the status deviation pattern.

[0043] As a further aspect of the present invention, if a state deviation pattern is detected, an adaptive task scheduler is triggered. This adaptive task scheduler, in conjunction with the latest structured construction site model, recalculates the start time, resource requirements, and execution location of subsequent virtual construction events, including:

[0044] The moment when the state deviation pattern is detected is taken as the new scheduling benchmark point, and the latest structured construction site model is obtained. The structured construction site model reflects the real-time physical state of the construction site.

[0045] Based on the deviation direction and deviation amount recorded by the state deviation pattern, the subsequent standard working hours of the affected process in the construction process knowledge base are dynamically corrected.

[0046] A resource-constrained project scheduling algorithm is adopted, with the modified standard working hours of the process, the logical relationship of the process, and the resource availability as constraints, and the goal function is to minimize the total project duration. The start time of all unstarted processes is replanned starting from the scheduling baseline.

[0047] Based on the replanned process start time and combined with the constructability status of each area in the structured construction site model, a suitable execution location is assigned to each process.

[0048] The output includes an updated scheduling scheme with recalculated start time, resource requirements, and execution location.

[0049] As a further aspect of the present invention, the step of updating the virtual construction event chain in real time based on the calculation results of the adaptive task scheduler includes:

[0050] Replace the simulated start time of the corresponding process in the original virtual construction event chain with the recalculated start time of each process in the updated scheduling scheme;

[0051] Replace the resource requirement description of the corresponding process in the original virtual construction event chain with the resource requirements of each process in the updated scheduling scheme;

[0052] Replace the position information of the corresponding process in the original virtual construction event chain with the execution position of each process in the updated scheduling scheme;

[0053] All updated process information is reordered according to the new start time to form an updated virtual construction event chain.

[0054] As a further aspect of the present invention, the step of pushing the update command to the field terminal includes:

[0055] In the updated virtual construction event chain, the information of several upcoming procedures is encapsulated into a structured task instruction package. The structured task instruction package includes task code, three-dimensional coordinates of execution location, list of required resources, and planned start and end times.

[0056] The structured task instruction package is sent to the field terminal held by the corresponding construction team through the industrial internet network;

[0057] The on-site terminal parses and displays the contents of the structured task instruction package to provide guidance for on-site construction.

[0058] As a further aspect of the present invention, the closed-loop management process of cyclically executing construction simulation, status comparison, and scheduling updates based on the updated virtual construction event chain and real-time collected actual construction status data includes:

[0059] While driving a new round of visualized construction simulation animation sequences with the updated virtual construction event chain, newly generated actual construction status data are continuously collected;

[0060] The newly collected actual construction status data will be compared with the updated virtual construction event chain on which the new round of simulations are based.

[0061] Based on the pattern comparison results, a decision is made as to whether to trigger the adaptive task scheduler again to start a new round of scheduling updates;

[0062] Repeat the steps from construction simulation to status comparison and then to scheduling update until all construction activities of the photovoltaic project are completed.

[0063] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0064] By leveraging the spatial variability of point cloud density and normal vectors, local convolution kernels are dynamically adjusted to adapt to the differentiated spatial characteristics of point cloud data in different areas of photovoltaic construction sites. The point cloud data segmentation operation logic is optimized to mitigate the adaptation bias caused by uniform fixed convolution kernel operations, achieving accurate segmentation of construction geographic entities, obstacles, and unconstructed areas. Simultaneously, spatial semantic information binding and annotation are completed, improving the structured attributes of the 3D site model, enriching the refined expression of spatial data, enhancing the analytical depth of real-scene scanned point cloud data, and improving the accuracy of site spatial element identification and differentiation, resulting in a more complete and standardized digital representation of site spatial information.

[0065] Realize the fusion operation of environmental meteorological time-series data and construction technology, break down the data barriers between environmental monitoring data and construction process simulation, and generate a continuous and ordered virtual construction event chain according to the time dimension. The time-sequential event chain drives the linked changes of the three-dimensional models of site geographic entities and virtual construction resources, and continuously generates coherent dynamic picture content. It conforms to the environmental change law of outdoor construction operations, expands the expression form of digital simulation of construction processes, strengthens the dynamic display capability of spatial scenes, restores the construction operation state under the influence of environmental factors, and improves the digital dynamic display form of the entire construction process. Description of Drawings

[0066] Figure 1 is a timing diagram of the construction information management system for photovoltaic projects based on industrial Internet according to the present invention;

[0067] Figure 2 is a flow chart of dynamically adjusting local convolution kernels by the improved spatial semantic segmentation algorithm;

[0068] Figure 3 is a flow chart for forming a structured construction site model. Detailed Description of the Embodiments

[0069] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0070] In the description of the present invention, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and other indicated orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings, which are only for facilitating the description of the present invention and simplifying the description, rather than indicating or implying that the indicated device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation to the present invention. In addition, in the description of the present invention, "a plurality" means two or more unless otherwise explicitly specified and limited.

[0071] Referring to Figure 1 , the present invention provides a photovoltaic project construction information management system based on industrial Internet, and the specific system includes:

[0072] The system comprises a data acquisition module, a site analysis module, a progress simulation module, and a visualization and simulation module. The data acquisition module acquires 3D real-scene scanned point cloud data and environmental meteorological time-series data of the photovoltaic construction site. The site analysis module receives the 3D real-scene scanned point cloud data and analyzes it using an improved spatial semantic segmentation algorithm. This improved algorithm dynamically adjusts the local convolution kernel based on the spatial variability of point cloud density and normal vectors. Through this algorithm, construction geographic entities, obstacles, and unconstructed areas are identified and separated from the point cloud data, thus forming a structured construction site model with spatial semantic annotations. The progress simulation module receives the environmental meteorological time-series data and inputs it into the construction progress simulation engine. This engine couples meteorological conditions with construction technology, generating a virtual construction event chain ordered along a timeline. The visualization and simulation module, within the constructed structured construction site model, drives the dynamic 3D models of corresponding construction geographic entities and virtual construction resources according to the instructions of the virtual construction event chain, thereby generating a visualized construction simulation animation sequence.

[0073] In one embodiment of the present invention, see [reference] Figure 2 An octree spatial index is constructed on the input 3D real-scene scanned point cloud data to enable fast querying of point cloud subsets within any spatial neighborhood. For each point to be processed in the point cloud, the number of points in its neighborhood with a preset radius is counted to calculate the local density of the point cloud. Simultaneously, the variance of the angle between the normal vectors of all points in this neighborhood is calculated to obtain the normal vector variability. A dynamic convolution kernel generation function is constructed. The input of this function is the calculated local density and normal vector variability of the point cloud, and the output is the shape parameters and weight distribution of the local 3D convolution kernel corresponding to the point to be processed. The logic of the dynamic convolution kernel generation function is that, based on the increase of the local density of the point cloud, the output convolution kernel has a larger spatial receptive field; based on the increase of the normal vector variability, the output convolution kernel has weights more concentrated at the center point. The dynamic convolution kernel generation function is used to generate a unique local 3D convolution kernel for each point to be processed, and this local 3D convolution kernel is used to perform weighted aggregation of the features of the neighboring point cloud, thereby completing the context-aware extraction of point cloud features.

[0074] In practical implementation, an octree spatial index is constructed for the input 3D real-scene scanned point cloud data. The octree spatial index recursively divides the 3D space into eight sub-cubes until the number of point clouds contained in each cube is below a preset threshold or the maximum depth is reached. This data structure enables rapid querying and retrieval of a subset of point clouds within a preset radius neighborhood of any spatial point. Specifically, for each point to be processed in the 3D real-scene scanned point cloud data, a query operation is performed on a spherical neighborhood centered on that point with a fixed radius R. The number of points within this neighborhood is counted and divided by the volume of the spherical neighborhood to calculate the local density D of the point cloud. When calculating the normal vector variability, the normal vector of each point within this neighborhood is first obtained, then the cosine of the angle between the normal vectors of any two points is calculated, and finally the variance of all cosine values ​​of the angle is obtained. This variance is denoted as the normal vector variability V.

[0075] In practical implementation, constructing a dynamic convolution kernel generation function is the core step. This function is a mapping relationship; its inputs are the calculated local density D of the point cloud and the normal vector variability V, and its output is a set of shape parameters K for a 3D convolution kernel and a corresponding weight distribution matrix W. The local density D of the point cloud is negatively correlated with the size of the convolution kernel's spatial receptive field. When the local density D is small, it indicates a sparse point cloud distribution. The dynamic convolution kernel generation function will output a convolution kernel shape parameter with a larger spatial coverage to aggregate contextual information from a wider area. The normal vector variability V is positively correlated with the concentration of the convolution kernel weights. When the normal vector variability V is large, it indicates drastic changes in local surface curvature or the presence of edges. The dynamic convolution kernel generation function will output a weight distribution matrix W with weights more concentrated at the center point to enhance the features of the center point and suppress interference from neighboring points that may come from different objects. The specific form of the dynamic convolution kernel generation function can be defined by a parameterized formula, for example:

[0076]

[0077] in: and It is a pre-defined function. and K is the set of learnable or fixed parameters of the function. The shape parameter set K can contain the size of the convolution kernel in each spatial dimension. The weight distribution matrix W defines the weight of each position in the convolution kernel to the feature of the center point.

[0078] In practice, a dynamic convolution kernel generation function is used to generate a unique local 3D convolution kernel for each point to be processed. For each point, based on its local point cloud density D and normal vector variability V, a set of unique shape parameters K and a weight distribution matrix W are calculated using the dynamic convolution kernel generation function. This set of parameters defines an irregular 3D convolution window centered on the point to be processed. The generated local 3D convolution kernel is used to perform weighted aggregation of the features of the neighboring point cloud. The operation involves determining the range of neighboring points to be calculated based on the shape parameter set K of the convolution kernel, and then performing a weighted summation of the feature vectors of each neighboring point within the range based on the weight distribution matrix W. The result of the weighted summation serves as a new feature representation of the point to be processed, rich in contextual information, thereby completing the context-aware extraction process of point cloud features.

[0079] In one embodiment of the present invention, see [reference] Figure 3 This study utilizes an improved spatial semantic segmentation algorithm to extract point cloud features, predicting the semantic category of each point in a 3D real-scene scanned point cloud. The predicted semantic categories include installed photovoltaic panels, photovoltaic support foundations, construction roads, temporary buildings, natural obstacles, and open spaces awaiting construction. Point clouds predicted to have the same semantic category are spatially clustered to form multiple independent 3D point cloud clusters. Each 3D point cloud cluster is fitted with a minimum bounding cube to calculate its 3D spatial coordinate range, volume, and center point location. The semantic category, spatial coordinate range, volume, and center point location information of each 3D point cloud cluster are associated and stored with the original point cloud data, collectively forming a structured construction site model with clearly defined element attributes within the 3D space.

[0080] In specific implementation, point cloud features extracted using an improved spatial semantic segmentation algorithm are used to predict the semantic category of each point in the 3D real-scene scan point cloud. The predicted semantic categories include installed photovoltaic panels, photovoltaic support foundations, construction roads, temporary buildings, natural obstacles, and open spaces awaiting construction. The prediction process involves inputting the context-aware feature vector extracted from each point and its neighborhood using the improved spatial semantic segmentation algorithm into a pre-trained classifier. The output of the classifier is the probability distribution of the point belonging to each predefined semantic category, and the category with the highest probability is assigned to the point as its semantic label. In some embodiments, the classifier can be a multilayer perceptron network, with the last layer activated using a softmax function to obtain the probability distribution. In practice, point clouds predicted to have the same semantic category are spatially clustered to form multiple independent 3D point cloud clusters. The clustering algorithm is based on the 3D Euclidean distance between points and the consistency of the semantic labels of the points. For example, an Euclidean clustering algorithm based on a distance threshold is used to group points in 3D space that are less than the set threshold and have the same semantic label into the same 3D point cloud cluster. Each 3D point cloud cluster represents an independent physical object or continuous region, such as a complete photovoltaic panel, a section of construction road, or a mound of earth.

[0081] In practice, a minimum bounding cube is fitted to each 3D point cloud cluster obtained through clustering. The minimum bounding cube is a cube aligned with the coordinate axes. It is calculated by finding the maximum and minimum values ​​of all points in the 3D point cloud cluster along the X, Y, and Z coordinate axes, thus determining the cube's boundaries. The 3D spatial coordinate range is defined by the coordinates of the eight vertices of the minimum bounding cube, the volume is calculated by the product of the three side lengths, and the center point is determined by the average of the minimum and maximum values ​​along the three coordinate axes. In some embodiments, the minimum bounding cube may not fit precisely to irregularly shaped objects, but it provides a standardized description of the object's spatial occupancy. Optionally, after calculating the minimum bounding cube, other geometric properties of the 3D point cloud cluster, such as the principal orientation, can be further calculated. In practical implementation, the semantic category, spatial coordinate range, volume, and center point location information of each 3D point cloud cluster are associated and stored with the original point cloud data. This association can be achieved by storing the unique identifier of the 3D point cloud cluster, its semantic category label, calculated spatial coordinate range, volume, and center point location in a single record in the database. A link is then established between this record and the indexes of all original point cloud data belonging to this cluster, collectively forming a structured construction site model with clearly defined feature attributes within the 3D space. This structured construction site model allows for programmatic querying and access to the semantic, geometric, and location information of each geographic entity.

[0082] In practice, spatial clustering can use a specific criterion to determine whether two points belong to the same cluster. An optional criterion formula can be expressed as:

[0083]

[0084] in: It is a Boolean value indicating whether point i and point j can be classified into the same cluster; and Let i and j represent the predicted semantic category labels, respectively. and Let i and j represent the three-dimensional coordinate vectors of point i and point j, respectively. This indicates the calculation of the Euclidean distance between two points; It is a preset distance threshold. Only when two points have the same semantic category label and the spatial distance is less than the threshold will the distance be limited. Only when two points are considered members of the same candidate cluster are they considered to be in that cluster. It can be understood that by iteratively applying this criterion, all points satisfying the connection condition are grouped together. It can also be understood that the construction of the structured construction site model provides a precise, semantically understood 3D environmental foundation for subsequent progress simulation and visualization.

[0085] In one embodiment of the invention, environmental meteorological time series data is analyzed to extract hourly data on solar irradiance, wind speed, precipitation, and temperature for the future simulation period. A construction process knowledge base is established, defining the resource types, standard working hours, sensitivity thresholds, and influence coefficients of each photovoltaic construction procedure to various meteorological factors. Starting from the current moment, the start-up of each procedure is simulated sequentially according to the logical relationships defined in the construction process knowledge base. When simulating the start-up of each procedure, the meteorological data for the planned execution time of that procedure is queried. If the meteorological data exceeds the sensitivity threshold defined in the construction process knowledge base, the standard working hours of the procedure are extended according to the influence coefficient, and the completion time of the procedure is recalculated. The simulated start-up time, resource requirements, execution location, and meteorologically adjusted simulated completion time of all procedures are arranged in chronological order to form a virtual construction event chain. For each procedure in the virtual construction event chain, a corresponding virtual construction resource 3D model is configured, including models of cranes, transport vehicles, and installation workers. The 3D spatial coordinate range corresponding to the execution location of the procedure is located in the structured construction site model. In a 3D visualization environment, the corresponding virtual construction resource 3D model is controlled to move to the 3D spatial coordinate range corresponding to the work process execution position. The virtual construction resource 3D model is driven to perform standardized construction operations, and the state and appearance of the corresponding construction geographic entities in the structured construction site model are updated synchronously to reflect the construction progress. Following the chronological order of the virtual construction event chain, the state changes of the 3D model corresponding to all work processes are recorded, and a continuous sequence of visualized construction simulation animations is rendered.

[0086] In practical implementation, the construction progress simulation engine begins by parsing the input environmental meteorological time series data, extracting hourly data on solar irradiance, wind speed, precipitation, and temperature for the future simulation period. A construction process knowledge base is then established, defining in a structured manner the resource types and standard working hours required for each photovoltaic construction procedure, as well as the sensitivity thresholds and influence coefficients of each procedure to different meteorological factors. Partial definitions of the construction process knowledge base are shown in Table 1.

[0087] Table 1: Construction Technology Knowledge Base Table

[0088]

[0089] In practice, starting from the current moment, the startup of each process is simulated according to the logical relationships defined in the construction technology knowledge base. These logical relationships include sequential and parallel relationships, and the simulation process follows the critical path method or a similar project scheduling algorithm. When simulating the startup of each process, the sensitivity thresholds of that process to various meteorological factors are retrieved from the construction technology knowledge base, and hourly meteorological data corresponding to the planned execution time of that process is read. The meteorological data is compared with the sensitivity thresholds. If any meteorological data exceeds the corresponding sensitivity threshold defined in the construction technology knowledge base, the standard working hours of the process are extended according to the corresponding impact coefficient defined in the knowledge base. The extended working hours of the process are calculated using a formula:

[0090]

[0091] in: This indicates the actual estimated working hours after weather adjustments. This refers to the standard working hours defined in the construction technology knowledge base. It is a binary indicator function, when the first... A meteorological factor (such as wind speed and precipitation) takes a value of 1 when its sensitivity threshold is exceeded, and 0 otherwise. Indicates the first One meteorological factor, This refers to the meteorological factors defined in the construction technology knowledge base. Influence coefficient, This represents the total number of meteorological factors considered. The completion time of the process is recalculated; it equals the simulated start time of the process plus the adjusted actual estimated working hours. The simulated start time, resource requirements, execution location, and simulated completion time after weather adjustment for all processes are arranged in chronological order to form a virtual construction event chain.

[0092] In practical implementation, a corresponding 3D virtual construction resource model is configured for each process in the virtual construction event chain. This model includes crane models, transport vehicle models, and installation worker models, each with a predefined geometric appearance and motion skeleton. The 3D spatial coordinate range corresponding to the process execution position is located within the structured construction site model. This location operation involves querying and matching the execution position description of the process in the virtual construction event chain within the structured construction site model database to obtain the minimum bounding cube spatial coordinates of that position. In the 3D visualization environment, the corresponding virtual construction resource 3D model is controlled to move from its previous or initial position to the center point or designated work point of the 3D spatial coordinate range corresponding to the process execution position. The system drives standardized construction operations on the virtual construction resource 3D model. These operations are pre-made keyframe animations or procedural animations for each type of resource model, such as the lifting, rotating, and lowering sequence of a crane. Simultaneously, it updates the state and appearance of corresponding construction geographic entities in the structured construction site model to reflect construction progress. For example, when the animation for the "photovoltaic panel installation" process plays, the semantic state of the target location "vacant land awaiting construction" is updated to "photovoltaic panels installed," and its 3D rendering appearance is changed. Following the chronological order of the virtual construction event chain, the system records the state changes of all 3D models corresponding to each process, including the model's position, posture, action frames, and the state attribute changes of geographic entities in the scene. The rendering engine generates a continuous sequence of visualized construction simulation animations based on these continuously recorded state frames.

[0093] In one embodiment of the present invention, actual construction status data reported by on-site IoT sensors is collected synchronously. This data includes material consumption records, equipment time records, and process check-in records. The actual process completion time, actual material consumption, and actual equipment working time are extracted from the actual construction status data. Processes planned for completion within the same time period are searched from the virtual construction event chain, and their simulated completion time (adjusted for weather conditions), planned material consumption, and planned equipment time are obtained. The difference between the actual process completion time and the simulated completion time is calculated, as are the ratios of actual material consumption and planned material consumption, and the ratios of actual equipment working time and planned equipment time. If the difference exceeds a preset time tolerance, or the ratio exceeds a preset quantity tolerance range, a status deviation pattern is determined to exist in the current time period. The process type, deviation direction, and deviation amount associated with the status deviation pattern are recorded. If a status deviation pattern is detected, an adaptive task scheduler is triggered. The adaptive task scheduler uses the moment the status deviation pattern is detected as the new scheduling reference point and obtains the latest structured construction site model, which reflects the real-time physical state of the construction site. Based on the deviation direction and magnitude recorded in the state deviation pattern, the subsequent standard working hours of affected processes in the construction technology knowledge base are dynamically corrected. A resource-constrained project scheduling algorithm is adopted, using the corrected standard working hours of the processes, the logical relationships between processes, and resource availability as constraints, and minimizing the total project duration as the objective function. Starting from the scheduling baseline, the start times of all unstarted processes are replanned. Based on the replanned process start times and combined with the constructability status of each area in the structured construction site model, a suitable execution location is assigned to each process. The adaptive task scheduler outputs an updated scheduling scheme containing the recalculated start times, resource requirements, and execution locations.

[0094] In practical implementation, actual construction status data reported by on-site IoT sensors is collected synchronously. This data includes material consumption records, equipment working hours records, and process check-in records. The actual process completion time, actual material consumption, and actual equipment working hours are extracted from this data. The planned processes within the same time period are searched from the virtual construction event chain, and the simulated completion time, planned material consumption, and planned equipment working hours (adjusted for weather conditions) are obtained. It is understood that status comparison requires data alignment based on unified process codes and time periods. In practical implementation, the difference between the actual process completion time and the simulated completion time is calculated, as are the ratios of actual material consumption and planned material consumption, and the ratios of actual equipment working hours and planned equipment working hours. Preset time tolerance thresholds and quantity tolerance ranges are set. If the time difference exceeds the preset time tolerance, or the material or equipment ratios exceed the preset quantity tolerance range, a status deviation pattern is determined for the current time period. The process type, deviation direction, and deviation amount associated with the status deviation pattern are recorded. The deviation direction indicates whether the schedule is delayed or ahead of schedule, or whether material or equipment usage is overspending or cost-saving. Refer to Table 2 for an illustration of data comparison:

[0095] Table 2: State Deviation Pattern Detection Table

[0096]

[0097] In practice, if a state deviation pattern is detected, the adaptive task scheduler is triggered. The adaptive task scheduler uses the moment the state deviation pattern is detected as the new scheduling reference point, obtains the latest structured construction site model, and continuously scans and analyzes the structured construction site model to reflect the real-time physical state of the construction site. Based on the deviation direction and magnitude recorded by the state deviation pattern, the subsequent standard working hours of the affected processes in the construction technology knowledge base are dynamically corrected. The correction process can follow a predefined rule function, for example:

[0098]

[0099] in: This indicates the adjusted standard working hours for subsequent processes affected by the deviation in operating conditions. This refers to the original standard working hours in the construction technology knowledge base. This indicates the amount of deviation (such as time delay or material overage percentage) in the recorded status deviation pattern. This is a pre-set correction coefficient based on the process type and deviation nature, used to propagate the impact of the current deviation to subsequent related processes. A resource-constrained project scheduling algorithm is employed, constrained by the corrected standard process time, process logical relationships, and resource availability, with the objective function of minimizing the total project duration. Starting from the scheduling baseline, the start times of all unstarted processes are replanned. In some embodiments, the resource-constrained project scheduling algorithm can be a heuristic algorithm based on priority rules or a mathematical programming method. In specific implementations, based on the replanned process start times and combined with the workability status of each area in the structured construction site model, a suitable execution location is assigned to each process. The workability status is determined by the semantic category of each area in the structured construction site model; for example, "vacant land awaiting construction" is a workable area, while "already installed photovoltaic panels" areas cannot be reassigned with installation processes. The output includes an updated scheduling scheme containing recalculated start times, resource requirements, and execution locations. The output format of the updated scheduling scheme is compatible with the virtual construction event chain.

[0100] In one embodiment of the present invention, the virtual construction event chain is updated in real time based on the calculation results of the adaptive task scheduler. The simulated start time of the corresponding process in the original virtual construction event chain is replaced with the recalculated start time of each process in the updated scheduling scheme. The resource requirement description of the corresponding process in the original virtual construction event chain is replaced with the resource requirement of each process in the updated scheduling scheme. The position information of the corresponding process in the original virtual construction event chain is replaced with the execution position of each process in the updated scheduling scheme. All updated process information is reordered according to the new start time to form an updated virtual construction event chain. Information on several upcoming processes in the updated virtual construction event chain is encapsulated into a structured task instruction package, which includes a task code, three-dimensional coordinates of the execution position, a list of required resources, and planned start and end times. The structured task instruction package is distributed to the field terminal held by the corresponding construction team via the industrial internet network. The field terminal parses and displays the content of the structured task instruction package, providing guidance for on-site construction. Based on the updated virtual construction event chain and real-time collected actual construction status data, a closed-loop management process of construction simulation, status comparison, and scheduling update is executed cyclically. While driving a new round of visualized construction simulation animation sequences with the updated virtual construction event chain, newly generated actual construction status data is continuously collected. This newly collected actual construction status data is then compared with the updated virtual construction event chain upon which the new round of simulations is based. Based on the comparison results, a decision is made as to whether to trigger the adaptive task scheduler again to initiate a new round of scheduling updates. This process of construction simulation, status comparison, and scheduling updates is repeated until all construction activities of the photovoltaic project are completed.

[0101] In practical implementation, the virtual construction event chain is updated in real time based on the calculation results of the adaptive task scheduler. The specific update operations involve replacing the simulated start time of the corresponding process in the original virtual construction event chain with the recalculated start time of each process in the updated scheduling scheme; replacing the resource requirement description of the corresponding process in the original virtual construction event chain with the updated resource requirement of each process in the updated scheduling scheme; and replacing the position information of the corresponding process in the original virtual construction event chain with the execution position of each process in the updated scheduling scheme. It can be understood that the replacement operation is based on matching the unique identifier of the process, overwriting the old data entry with the same identifier in the virtual construction event chain with the new data entry of a process in the updated scheduling scheme. All the updated process information is then reordered according to the new start time to form the updated virtual construction event chain. This reordering ensures the logical correctness of the processes on the timeline. In some embodiments, the update of the process start time can be formally represented by an assignment operation:

[0102]

[0103] in: This indicates the first event in the updated virtual construction event chain. The object of each process event Indicates the corresponding number in the updated scheduling scheme Each process item It is a process start time attribute. It is a recalculated start-up time attribute, and this operation is performed sequentially on all processes in the chain that need to be updated.

[0104] In practice, pushing update instructions to the field terminal involves encapsulating the upcoming process information from the updated virtual construction event chain into a structured task instruction package. This package includes a task code, three-dimensional coordinates of the execution location, a list of required resources, and planned start and end times. Optionally, the structured task instruction package may also include a brief description of the process or safety precautions. The encapsulated structured task instruction package is then distributed to the field terminal held by the corresponding construction team via the industrial internet network. The field terminal is a mobile device or vehicle-mounted terminal with a dedicated client installed. The field terminal receives and parses the content of the structured task instruction package, displaying the task code, three-dimensional coordinates of the execution location, a list of required resources, and planned start and end times on its graphical interface, providing clear guidance for on-site construction. In some embodiments, the three-dimensional coordinates of the execution location can be further converted into an area code or geographic marker that is more easily identifiable by on-site personnel.

[0105] In practical implementation, based on the updated virtual construction event chain and real-time collected actual construction status data, a closed-loop management process of construction simulation, status comparison, and scheduling update is executed cyclically. The driving logic of this cycle is that while the updated virtual construction event chain drives a new round of visualized construction simulation animation sequence, newly generated actual construction status data is continuously collected automatically through on-site IoT sensing devices. The newly collected actual construction status data is compared with the updated virtual construction event chain on which the new round of simulation is based. The comparison method includes calculating the time difference, material consumption ratio, and equipment time ratio, and comparing them with preset tolerance thresholds. Based on the comparison results, it is decided whether to trigger the adaptive task scheduler again to start a new round of scheduling update. The condition for this decision is whether a new state deviation pattern is detected in the comparison. The steps from construction simulation to status comparison to scheduling update are repeated, with each cycle based on the latest virtual construction event chain and actual construction status data, until all construction activities of the photovoltaic project are completed. The completion sign is that the status of all procedures in the virtual construction event chain is marked as completed and there are no new state deviations. Understandably, this closed-loop process enables construction management to dynamically respond to changes on site, and each iteration of the closed-loop management process corrects the deviation between the plan and the actual situation.

[0106] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A photovoltaic project construction information management system based on the Industrial Internet, characterized in that, include: The data acquisition module acquires 3D real-scene scanning point cloud data and environmental meteorological time series data of the photovoltaic construction site; The site analysis module uses an improved spatial semantic segmentation algorithm to analyze the 3D real-scene scanned point cloud data. The improved spatial semantic segmentation algorithm dynamically adjusts the local convolution kernel based on the spatial variability of point cloud density and normal vector. Through the improved spatial semantic segmentation algorithm, construction geographic entities, obstacles and unconstructed areas are identified and separated from the 3D real-scene scanned point cloud data, forming a structured construction site model with spatial semantic annotation. The progress simulation module inputs the environmental meteorological time series data into the construction progress simulation engine. The construction progress simulation engine couples meteorological conditions with construction technology to generate a virtual construction event chain sorted by time axis. The visualization simulation module, based on the instructions of the virtual construction event chain, drives the dynamic three-dimensional models of the corresponding construction geographic entities and virtual construction resources in the structured construction site model to generate a sequence of visualized construction simulation animations.

2. The photovoltaic project construction information management system based on the Industrial Internet as described in claim 1, characterized in that, The improved spatial semantic segmentation algorithm dynamically adjusts the local convolution kernel based on the spatial variability of point cloud density and normal vector, including: An octree spatial index is constructed on the input 3D real-scene scan point cloud data to quickly query a subset of point clouds within any spatial neighborhood; For each point to be processed, the number of points in the neighborhood of its preset radius is counted to calculate the local density of the point cloud, and the variance of the angle between the normal vectors of all points in the neighborhood is calculated to obtain the normal vector variability. Construct a dynamic convolution kernel generation function. The input of the dynamic convolution kernel generation function is the local density and normal vector variability of the point cloud, and the output is the shape parameters and weight distribution of the local three-dimensional convolution kernel corresponding to the point to be processed. As the local density of the point cloud increases, the dynamic convolution kernel generation function outputs a convolution kernel with a larger spatial receptive field; as the variability of the normal vector increases, the dynamic convolution kernel generation function outputs a convolution kernel with weights more concentrated at the center point. The dynamic convolution kernel generation function generates a unique local 3D convolution kernel for each point to be processed, and the local 3D convolution kernel is used to perform weighted aggregation of the features of the neighboring point cloud to complete the context-aware extraction of point cloud features.

3. The photovoltaic project construction information management system based on the Industrial Internet as described in claim 2, characterized in that, The process of identifying and separating construction geographic entities, obstacles, and unconstructed areas from 3D real-scene scanned point cloud data to form a structured construction site model with spatial semantic annotations includes: Using the point cloud features extracted by the improved spatial semantic segmentation algorithm, semantic category prediction is performed on each point in the 3D real-scene scan point cloud. The semantic categories include installed photovoltaic panels, photovoltaic support foundations, construction roads, temporary buildings, natural obstacles, and open spaces awaiting construction. Point clouds predicted to have the same semantic category are spatially clustered to form multiple independent 3D point cloud clusters; For each 3D point cloud cluster, perform minimum bounded cube fitting to calculate its 3D spatial coordinate range, volume and center point position; The semantic category, spatial coordinate range, volume, and center point location information of each 3D point cloud cluster are associated and stored with the original point cloud data, together forming the structured construction site model with clearly defined element attributes in the three-dimensional space.

4. The photovoltaic project construction information management system based on the Industrial Internet as described in claim 1, characterized in that, The environmental meteorological time series data is input into the construction progress simulation engine, which couples meteorological conditions with construction technology to generate a virtual construction event chain ordered along a time axis, including: Analyze environmental meteorological time series data to extract hourly data on light intensity, wind speed, precipitation, and temperature for the future simulation period; Establish a construction process knowledge base, which defines the resource types and standard working hours required for each photovoltaic construction procedure, as well as the sensitivity thresholds and influence coefficients of each procedure to various meteorological factors. Starting from the current moment, based on the logical relationship of the construction process defined in the construction process knowledge base, the start of each process is simulated sequentially; When simulating the start of each process, the meteorological data of the planned execution time of the process is queried. If the meteorological data exceeds the sensitivity threshold defined in the construction process knowledge base, the standard working hours of the process are extended according to the impact coefficient, and the completion time of the process is recalculated. The simulated start time, resource requirements, execution location, and simulated completion time after weather adjustment for all processes are arranged in chronological order to form the virtual construction event chain.

5. The photovoltaic project construction information management system based on the Industrial Internet according to claim 4, characterized in that, In the structured construction site model, based on the instructions of the virtual construction event chain, the dynamic 3D models of the corresponding construction geographic entities and virtual construction resources are driven to generate a sequence of visualized construction simulation animations, including: For each process in the virtual construction event chain, a corresponding virtual construction resource 3D model is configured, which includes crane, transport vehicle, and installation worker models. Locate the three-dimensional spatial coordinate range corresponding to the position of the process execution in the structured construction site model; In a 3D visualization environment, the corresponding virtual construction resource 3D model is controlled to move to the 3D spatial coordinate range corresponding to the execution position of the process; Drive the virtual construction resource 3D model to perform standardized construction operations, and synchronously update the status and appearance of the corresponding construction geographic entities in the structured construction site model to reflect the construction progress; According to the time sequence of the virtual construction event chain, the state changes of the three-dimensional model corresponding to all procedures are recorded, and a continuous sequence of the visualized construction simulation animation is generated.

6. The photovoltaic project construction information management system based on the Industrial Internet according to claim 1, characterized in that, The system also includes: The status synchronization module synchronously collects the actual construction status data reported by the on-site IoT sensor devices. The actual construction status data includes material consumption records, equipment working hour records, and process check-in records. The scheduling correction module compares the actual construction status data with the virtual construction event chain data at the same time point to detect whether there is a status deviation pattern. If a status deviation pattern is detected, the adaptive task scheduler is triggered. The adaptive task scheduler, combined with the latest structured construction site model, recalculates the start time, resource requirements and execution location of subsequent virtual construction events. The closed-loop update module updates the virtual construction event chain in real time based on the calculation results of the adaptive task scheduler and pushes the update command to the field terminal. Based on the updated virtual construction event chain and the real-time collected actual construction status data, it cyclically executes the closed-loop management process of construction simulation, status comparison and scheduling update. The actual construction status data is compared with the virtual construction event chain data at the same time point to detect whether there is a status deviation pattern, including: Extract actual process completion time, actual material consumption, and actual equipment working time from actual construction status data; Find the planned completion time, planned material consumption, and planned equipment man-hours of the work processes within the same time period from the virtual construction event chain. Calculate the difference between the actual completion time and the simulated completion time, calculate the ratio of actual material consumption to planned material consumption, and calculate the ratio of actual equipment working time to planned equipment working time. If the difference exceeds the preset time tolerance, or the ratio exceeds the preset quantity tolerance range, then it is determined that the state deviation mode exists in the current time period. Record the process type, deviation direction, and deviation amount associated with the status deviation pattern.

7. The photovoltaic project construction information management system based on the Industrial Internet according to claim 6, characterized in that, If a state deviation pattern is detected, an adaptive task scheduler is triggered. This scheduler, combined with the latest structured construction site model, recalculates the start time, resource requirements, and execution location of subsequent virtual construction events, including: The moment when the state deviation pattern is detected is taken as the new scheduling benchmark point, and the latest structured construction site model is obtained. The structured construction site model reflects the real-time physical state of the construction site. Based on the deviation direction and deviation amount recorded by the state deviation pattern, the subsequent standard working hours of the affected process in the construction process knowledge base are dynamically corrected. A resource-constrained project scheduling algorithm is adopted, with the modified standard working hours of the process, the logical relationship of the process, and the resource availability as constraints, and the goal function is to minimize the total project duration. The start time of all unstarted processes is replanned starting from the scheduling baseline. Based on the replanned process start time and combined with the constructability status of each area in the structured construction site model, a suitable execution location is assigned to each process. The output includes an updated scheduling scheme with recalculated start time, resource requirements, and execution location.

8. The photovoltaic project construction information management system based on the Industrial Internet according to claim 7, characterized in that, The step of updating the virtual construction event chain in real time based on the calculation results of the adaptive task scheduler includes: Replace the simulated start time of the corresponding process in the original virtual construction event chain with the recalculated start time of each process in the updated scheduling scheme; Replace the resource requirement description of the corresponding process in the original virtual construction event chain with the resource requirements of each process in the updated scheduling scheme; Replace the position information of the corresponding process in the original virtual construction event chain with the execution position of each process in the updated scheduling scheme; All updated process information is reordered according to the new start time to form an updated virtual construction event chain.

9. The photovoltaic project construction information management system based on the Industrial Internet as described in claim 8, characterized in that, The step of pushing the update command to the field terminal includes: In the updated virtual construction event chain, the information of several upcoming procedures is encapsulated into a structured task instruction package. The structured task instruction package includes task code, three-dimensional coordinates of execution location, list of required resources, and planned start and end times. The structured task instruction package is sent to the field terminal held by the corresponding construction team through the industrial internet network; The on-site terminal parses and displays the contents of the structured task instruction package to provide guidance for on-site construction.

10. The photovoltaic project construction information management system based on the Industrial Internet according to claim 9, characterized in that, The closed-loop management process, which involves cyclically executing construction simulation, status comparison, and scheduling updates based on the updated virtual construction event chain and real-time collected actual construction status data, includes: While driving a new round of visualized construction simulation animation sequences with the updated virtual construction event chain, newly generated actual construction status data are continuously collected; The newly collected actual construction status data will be compared with the updated virtual construction event chain on which the new round of simulations are based. Based on the pattern comparison results, a decision is made as to whether to trigger the adaptive task scheduler again to start a new round of scheduling updates; Repeat the steps from construction simulation to status comparison and then to scheduling update until all construction activities of the photovoltaic project are completed.

Images