Intelligent measured quantity method based on bluetooth transmission and automatic input

Abstract

The present application relates to the field of engineering surveying, in particular to a kind of intelligent measured quantity method based on bluetooth transmission and automatic entry, comprising: based on the survey point layout rule of each building component, the survey point layout position of the component to be measured is obtained;With the survey point layout position in local space window as input, in combination with the component type of the component to be measured, the target measurement data of entity embedding is determined;Call the conflict area of the mark direction and state identification of target measurement data to determine the deviation distribution trend of conflict area;Conflict area is mapped to building component, and adjacent component grouping is carried out based on deviation distribution trend, and risk correlation matrix is established;Characteristic state division is carried out based on risk correlation matrix, and the early warning structure after comparison is determined for multiple characteristic states.It improves the disposal efficiency and accuracy of measurement early warning.

Description

Technical Field

[0001] This invention relates to the field of engineering measurement technology, specifically an intelligent measurement method based on Bluetooth transmission and automatic data entry. Background Technology

[0002] Traditional measurement methods rely heavily on manual operation, using basic tools such as levels and theodolites for data collection. However, these tools are susceptible to several factors during operation. First, the varying locations and densities of measurement points by different personnel can lead to data inaccuracies. Second, individual point data from on-site measurements are easily affected by factors such as human contact, environmental vibrations, and electromagnetic interference, rendering single-point measurements unreliable. Therefore, it is necessary to consider the overall data distribution to determine the commonalities in the current measurements.

[0003] For example, Chinese Patent Publication No. CN120890373A discloses an intelligent measurement tool and its measurement method for building electromechanical installation engineering. The system includes a measurement environment preparation unit, a multi-parameter acquisition unit, a spatial topology analysis unit, an error optimization unit, and a measurement verification unit. The measurement environment preparation unit acquires initial environmental parameters and generates an environmental parameter sequence; the multi-parameter acquisition unit simultaneously captures multiple types of data, calculates spatial adaptation values, and establishes an initial set of measurement features; the spatial topology analysis unit identifies the topological connection direction and conflict areas of equipment groups, generating a spatial topology deviation dataset; the error optimization unit calculates spatial positioning compensation values ​​and adjusts the equipment coordinate mapping relationship; the measurement verification unit compares the deviation amount, re-acquires data if necessary, and finally generates a measurement topology map.

[0004] For example, Chinese Patent Publication No. CN121346686A discloses a method, system, and device for measuring structural deformation of buildings, relating to the field of structural deformation measurement technology. The measurement method includes: acquiring the physical coordinates of laser points, the angle between the laser emission direction and the world coordinate system, and a building image; preprocessing the building image and extracting the pixel coordinates of the laser points; calculating the relative coordinates of the laser points on the object surface, solving the homography transformation matrix, performing perspective correction on the building image to obtain an orthographic projection image, and obtaining the conversion coefficient between pixels and actual distances; inputting the preprocessed building image into a pre-trained semantic segmentation model to obtain a beam segmentation mask and beam edge curves; and converting the beam's actual span using the conversion coefficients.

[0005] In existing technologies, measurement verification output is achieved through environmental adaptation, measurement of component topology relationships, and equipment coordinate repositioning; or coordinate measurement definition is achieved through coordinate projection correction. However, existing technologies tend to judge measurement data based on isolated points, failing to identify risk correlations between components. Consequently, they lack the ability to analyze the correlation and transmission of construction measurement anomalies and identify commonalities, resulting in reduced processing accuracy for overall project measurement control and affecting the feedback effect of real-time measurements. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an intelligent measurement method based on Bluetooth transmission and automatic input, including: S1, obtaining the measurement point layout position of the component to be measured based on the measurement point layout rules of each building component.

[0007] S2 takes the location of the measuring points within the local space window as input and, combined with the component type of the component to be measured, determines the target measurement data for the entity embedding.

[0008] S3 calls the marker direction and status indicator of the target measurement data to determine the conflict area and obtain the deviation distribution trend of the conflict area.

[0009] S4 maps the conflict area to the building components and groups adjacent components based on the deviation distribution trend to establish a risk correlation matrix.

[0010] S5. Based on the risk correlation matrix, feature states are divided to determine the early warning structure after comparison of multiple feature states.

[0011] The beneficial effects of this invention are as follows: First, this invention determines the effective measurement area boundary of the component by extracting a continuous set of coordinate points of the outer contour of the component to be measured, and automatically matches and generates the measurement point layout positions within the area by combining the measurement point layout rules of the corresponding detection items. This solves the problems of large differences in edge offset of irregularly shaped components and insufficient rationality of point layout at contour intersections, and improves the adaptability of measurement point layout to building components.

[0012] II. This invention uses the minimum number of measurement points and the maximum allowable interval corresponding to the component type as constraints, and forms several local spatial windows through neighborhood measurement point polling and spatial clustering. A spatial topology is constructed using measurement points within the window as nodes and spatial distance and component continuity as edges. Target measurement data with entity embedding is generated through graph convolution feature fusion and contrast enhancement. By dividing the local spatial windows and spatial clustering, isolated single-point data is transformed into structured data with neighborhood associations. Simultaneously, graph convolution feature fusion and contrast enhancement improve the reliability and feature dimensionality of the measurement data, providing data support for subsequent deviation trend analysis and conflict area identification.

[0013] Third, this invention calculates the deviation from the target measurement data, extracts the principal gradient direction of the deviation using principal component analysis as the marker direction, and sets status indicators based on the deviation value range and the proportion of out-of-tolerance points. Through rules for judging contiguous out-of-tolerance areas and clustered reverse deviations, it identifies local conflict areas and outputs the corresponding deviation distribution trend. This accurately captures contiguous out-of-tolerance areas and clustered reverse deviation areas masked by the overall pass rate, improving the defect detection rate. Furthermore, by using the principal gradient direction and deviation distribution trend, it clearly presents the spatial orientation of conflicting defects, providing a basis for subsequent defect cause analysis and targeted rectification.

[0014] IV. This invention maps conflict areas to corresponding building components to form a set of components to be analyzed; it calculates damage similarity based on the spatial gradient of the components and groups similar components; it integrates spatial adjacency factors, stress correlation factors, construction correlation factors, and damage similarity to obtain a weighted global risk correlation matrix. This enables the identification of quality problems with common origins, and the multi-factor risk correlation clearly presents the spatial transmission path of risks; thus improving the accuracy and efficiency of overall early warning.

[0015] V. This invention sorts risk values ​​based on a risk correlation matrix in ascending order and classifies the characteristic states of components using quartiles as a benchmark. It verifies the consistency of deviation distribution trends among adjacent components, identifying spatially connected and trend-consistent risk component sets as early warning structures and outputting tiered early warning results. It can automatically identify continuously connected clustered early warning structures, improving the control value and handling efficiency of quality early warnings. Attached Figure Description

[0016] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Figure 1 This is a flowchart illustrating an intelligent measurement method based on Bluetooth transmission and automatic data entry. Figure 2 This is a flowchart illustrating step S2 of an intelligent measurement method based on Bluetooth transmission and automatic data entry. Figure 3 This is a flowchart illustrating step S3 of an intelligent measurement method based on Bluetooth transmission and automatic data entry. Figure 4 This is a flowchart illustrating step S4 of an intelligent measurement method based on Bluetooth transmission and automatic data entry. Detailed Implementation

[0018] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0019] See Figure 1 A smart measurement method based on Bluetooth transmission and automatic data entry includes: S1, obtaining the measurement point layout position of the component to be measured based on the measurement point layout rules of each building component.

[0020] S2 takes the location of the measuring points within the local space window as input and, combined with the component type of the component to be measured, determines the target measurement data for the entity embedding.

[0021] S3 calls the marker direction and status indicator of the target measurement data to determine the conflict area and obtain the deviation distribution trend of the conflict area.

[0022] S4 maps the conflict area to the building components and groups adjacent components based on the deviation distribution trend to establish a risk correlation matrix.

[0023] S5. Based on the risk correlation matrix, feature states are divided to determine the early warning structure after comparison of multiple feature states.

[0024] In the current scenario, the emphasis is on the benchmark transfer and deviation control of data acquisition. This involves acquiring data measured by devices such as laser rangefinders, straightedges, floor slab thickness gauges, and digital angle gauges. After comparing this data with measurement standards, if the measured data shows deviations exceeding tolerances, the mobile terminal issues a prompt on-site and requires the inspection personnel to remeasure or record the rectification status. After all measurement points are completed, the system automatically summarizes and generates a measured report that includes pass rate statistics, deviation distribution, and a list of out-of-tolerance locations. The report is then simultaneously uploaded to the cloud management platform to achieve real-time generation of multi-location reports, improving the objectivity and timeliness of quality inspection.

[0025] In one embodiment of the present invention, one implementation of step S1 includes: S11, obtaining a continuous set of coordinate points of the outer contour of the component to be measured, and determining the region boundary of the component to be measured.

[0026] S12, based on the spatial range of the region boundary, delineate the detection items within the corresponding range, synchronize the detection items with the measurement point layout rules, and obtain the measurement point layout positions under the region boundary.

[0027] Specifically, discrete coordinate points of the outer contour of the component to be measured are automatically extracted from the BIM model or CAD vector construction drawings; for three-dimensional components such as walls and openings, they are uniformly mapped to the corresponding elevation / planar coordinate system to retain the three-dimensional spatial coordinate information; the obtained discrete coordinate points are sorted in clockwise and counterclockwise order to form a continuous and closed set of coordinate points, which serves as the basis for subsequent point layout.

[0028] The coordinate point set is then fitted into a boundary polygon to identify the area to be measured of the component. Next, the area to be measured is classified according to the detection items such as verticality, flatness, and horizontal length. Finally, the location of the measurement points at the boundary is determined one by one according to the measurement point layout rules corresponding to each type of detection item.

[0029] The measurement point layout rules are based on the type of component to be measured, such as concrete shear walls, door and window openings, cast-in-place floors, etc. The rules automatically retrieve the parameters of the corresponding test items from the measurement point layout rule library, including parameters such as standard measurement point spacing and minimum number of measurement points, to obtain a digital list containing measurement point locations and measurement standards.

[0030] In one embodiment of the present invention, the measurement area is divided into several continuous and non-overlapping local spatial windows according to the component type and the characteristics of the test items. Each local spatial window serves as an independent calculation unit. For example, the verticality test of a concrete wall can be divided into a 1m×1m window; the flatness test of the ground can be divided into a 2m×2m window; and the contour test of door and window openings can be divided into linear windows along the boundary at a length of 0.5m.

[0031] At the same time, for each local spatial window of measurement, the measurement interval, maximum allowable interval, and minimum number of measurement points are retrieved synchronously to determine whether the current layout of measurement points is compatible with the corresponding area, so as to avoid deviations in component measurement data due to irregular measurement areas.

[0032] Specifically, such as Figure 2 As shown, one implementation of step S2 is as follows: S21, based on the construction type of the component to be measured, obtain the minimum number of measurement points and the maximum allowable interval corresponding to the component to be measured; wherein, the minimum number of measurement points is the specified minimum number of measurement points for a single component, representing the minimum number of samples that a single component needs to achieve; the maximum allowable interval is the maximum spatial distance between adjacent measurement points, used to prevent missed detection.

[0033] S22, based on the minimum number of measuring points and the maximum allowable interval corresponding to the component to be measured, perform overall polling of spatially adjacent measuring points to determine the divided local spatial window.

[0034] When dividing the local spatial window, based on the spatial range of the measurement points, all spatially adjacent measurement points are successively polled to initially delineate the neighborhood relationships of the measurement points.

[0035] Using the spatial Euclidean distance between adjacent measurement points, the reciprocal of this distance is used as the clustering value to perform spatial clustering of the measurement points; when the number of measurement points in each cluster is greater than the minimum number of measurement points and the spatial Euclidean distance between any two measurement points is less than the maximum allowable interval, the measurement points of the corresponding cluster are regarded as a set of neighborhood points.

[0036] Furthermore, the clustering algorithm employs density clustering, with the neighborhood radius set based on the elbow rule, and the minimum number of points selected is 8-12.

[0037] The neighborhood point set is continuously spliced ​​according to its spatial distribution to form several continuous, non-overlapping local spatial windows, which serve as the smallest computational unit for subsequent corrections.

[0038] S23, each measuring point within each local window is regarded as a node, and edges are established based on spatial distance and the continuity of components within the window to construct a spatial topology; S24. For each measurement point within the spatial topology, aggregate the features of all adjacent measurement points to form a feature representation of entity embedding, and use the corresponding data as the output target measurement data.

[0039] Specifically, when forming the feature representation of entity embedding, for continuously input measurement data, when the measurement data fluctuates, an initial feature set for the corresponding measurement point is constructed. The initial feature set statistically analyzes the real-time variance, fluctuation values, range, etc., during measurement, and standardizes the corresponding data according to the measurement point number and component type to form the initial feature set.

[0040] For each measurement point within a local spatial window, feature fusion is performed on the initial feature set of the measurement point based on the spatial topology to generate graph convolution features for multi-point collaborative measurement.

[0041] By pooling the graph convolutional features, the features of each local spatial window are converted into an initial embedding vector; the initial embedding vector is then contrast-enhanced to obtain the target measurement data for entity embedding.

[0042] Furthermore, in this embodiment, the target measurement data embedded in the entity is processed using a local spatial window as a unit. The original measurement value and fluctuation statistics of a single measurement point are combined with the spatial adjacency relationship between measurement points and the continuity of the components. After graph convolution feature fusion and contrast enhancement processing, a low-dimensional structured feature vector is generated. This vector retains both the single-point deviation value and the spatial distribution correlation features of the local area. It is used to enhance the spatial continuity of the deviation and provide more reliable input data for subsequent deviation trend analysis and defect identification.

[0043] Furthermore, the graph convolution method is as follows: ; Wherein, it indicates that the measurement point i is in the feature layer. The output features represent the aggregated graph convolutional features; Indicates the activation function; This represents the node degree of measurement point i, and the number of its neighbors. Indicates the nodal degree of measurement point j; Let i represent the set of neighboring nodes of measurement point i; This indicates that measurement point j is in the feature layer. The output features, whose first-level values ​​represent the original features, i.e. the initial feature set initially entered; This represents the weight matrix. After image convolution, these features are subjected to global average pooling to obtain the features of each local spatial window after pooling. At the same time, these features are transformed linearly, and the 1×64 vector obtained by pooling is multiplied by a 64×128 weight matrix, plus the bias, to output a 1×128 initial embedding vector.

[0044] Next, the embedding vectors of adjacent walls on the same floor and of the same type that have been confirmed to be fully qualified after retesting and rectification are used as positive samples, and the embedding vectors of inferior walls with extremely large deviations and dispersion are used as negative samples. The current data is then compared and enhanced to obtain target measurement data that includes abstract feature representations and measured values.

[0045] It should be noted that contrast enhancement uses backpropagation based on the loss value of the currently input features. By adjusting the weight matrix, the output convolutional features are continuously adjusted, and the target measurement data is obtained when the loss is minimized. Typically, a triplet loss function is used for contrastive learning. The embedding vector of the current window is used as the anchor point, the embedding vector of qualified walls is used as the positive sample, and the embedding vector of inferior walls is used as the negative sample. The weight matrix is ​​iteratively updated through backpropagation, and the final entity embedding vector is output when the loss value converges.

[0046] In one embodiment of the present invention, the principal gradient direction angle of the target measurement data is extracted by performing principal component analysis or center direction normalization on the target measurement data; and the trend of change after aggregation of multi-measurement point data is determined according to the direction of the direction angle; if it points to the positive quadrant of the coordinate axis, it is recorded as a positive deviation trend, which means that the deviation of the corresponding detection item gradually increases along a certain spatial direction; if it points to the negative quadrant, it is recorded as a negative deviation trend, which means that the deviation of the corresponding detection item gradually decreases along a certain spatial direction; if the magnitude of the principal gradient direction angle is extremely short and the direction is random, it is marked as a noise / unreliable point, and then the measurement result under the aggregation of each measurement point is determined.

[0047] Specifically, such as Figure 3 As shown, one implementation of step S3 includes: S31, performing principal component analysis on the target measurement data based on the deviation between the target measurement data and the standard value, and using the principal gradient direction of the analysis as the label direction.

[0048] If the current data measurement does not fluctuate, and each measurement is consistent, the conflict area and its deviation distribution trend within the current local spatial window will be determined based on the status identifier settings.

[0049] S32, based on the range of deviation values ​​and the proportion of measurement points with deviations exceeding the allowable range, count them one by one according to the number of test items, and set the status indicator of the target measurement data.

[0050] Furthermore, the status indicators include intact, basically intact, slightly damaged, and dangerous; intact means that the difference between the measuring points is within the allowable range, such as ±5mm; basically intact means that the proportion of measuring points with deviations exceeding the allowable range is less than 20%, and the maximum deviation is less than 1.5 times the allowable value; the proportion corresponding to slightly damaged is between 20% and 50%, or the maximum deviation is between 1.5 and 2 times the allowable value; dangerous means that the corresponding proportion is greater than 50%, or the maximum deviation is greater than 2 times the allowable value, or there are cracks / exposed reinforcement.

[0051] S33. Based on the clustering area of ​​the marking direction and status indicator, determine the conflict area, and output the deviation distribution trend according to the combination of the marking direction and status indicator. When the overall deviation trend of the window contradicts the qualified status and deviation sign of the internal measuring points, it indicates that there is a sudden change in local deviation within the window, and the corresponding conflicting status needs to be marked.

[0052] Specifically, for each local spatial window, the clusters of positive deviation measurement points, negative deviation measurement points, and out-of-tolerance points are statistically analyzed; where out-of-tolerance points represent measurement points that exceed the allowable range, positive deviation measurement points represent measurement points with deviation values ​​greater than 0, and negative deviation measurement points represent measurement points with deviation values ​​less than 0.

[0053] Furthermore, for the identified out-of-range, positive-deviation, and negative-deviation measurement points, points with a distance less than 1.5 times the average distance between measurement points are considered adjacent. A cluster is defined as a group of three or more consecutively adjacent points, and the smallest circumcircle containing the clustered points is then considered the clustered region. If the current number of measurement points is too small, Kriging spatial interpolation is used to provide enough measurement points for the current spatial clustering to complete the relative spatial region clustering.

[0054] When the percentage of out-of-tolerance points is less than a preset threshold and there are at least three out-of-tolerance points in a contiguous area, the current clustered area is identified as a conflict area based on the location of the out-of-tolerance points. The preset threshold is set to 20% of the total number of measurement points. That is, when there are contiguous out-of-tolerance points in the current area and the overall pass rate reaches 80%, local areas affecting construction are identified and marked as local state conflicts. The preset threshold is determined according to project control requirements. Local spatial windows with out-of-tolerance points greater than or equal to the preset threshold are directly judged as production non-conformities without local conflict analysis, and an alarm is directly triggered for that area.

[0055] When the local space window is marked as a positive deviation trend, the current cluster area is determined as a conflict area based on the proportion and distribution of negative deviation measurement points. At this time, only when the negative deviation measurement points are greater than 30% of the total measurement points, the proportion of reverse deviation is high enough, and the continuous parts of the two points are combined to determine the local defect conflict part under the overall deviation distribution.

[0056] When a local spatial window is marked as having a negative deviation trend, the current cluster area is determined as a conflict area based on the proportion and distribution of positive deviation measurement points. At this time, it is only marked as an internal state conflict when the number of positive deviation measurement points is greater than 30% of the total number of measurement points.

[0057] In cases where there are many positive and negative deviations, it indicates that there are obvious wall protrusions and depressions in the corresponding areas. In such cases, an early warning can be issued directly instead of further analysis.

[0058] The above-mentioned judgment process based on the window principal gradient and direct differences is used to compensate for window coarse-grained defects. The conflict content of local defects is used as the analysis subject of subsequent damage description to determine the building components with overall deviation conflicts. As for other cases, they are more obvious than the current local defects and can be directly repaired on-site by weight or structure.

[0059] It should be noted that deviation represents the difference between the measured value of the measuring point and the design value or the allowable value of the specification, and is used to characterize the degree of deviation of the construction quality.

[0060] Furthermore, the output deviation distribution trend includes the data combination of the principal gradient direction angle, out-of-error points, positive deviation measurement points, negative deviation measurement points, and corresponding conflict regions as the current data evolves over time, in order to explain the dangerous state of each part.

[0061] In the current embodiment, within the local spatial window, the conflict area represents a clustered area where the overall trend of the deviation and the local measuring point status indicators contradict each other. This includes two types: first, areas where the overall pass rate meets the standard but there are contiguous areas with concentrated deviations; and second, areas where the overall deviation trend and the clustered local reverse deviations do not match. This area corresponds to localized hidden construction defects masked by the overall pass rate index. It is a refined defect unit distinct from overall unqualified components, used to support targeted modifications to the building.

[0062] In one embodiment of the present invention, for each conflict region, the spatial gradient of each measuring point in the conflict region is first quantified in the form of central difference, that is, by subtracting the deviation value of the previous point from the deviation value of the next point and dividing by the spatial distance between the two points; this method ignores the random error of the current point and emphasizes the degree of change of the overall deviation; then the obtained spatial gradient is used as the input value of the Pearson correlation coefficient to quantify the damage similarity of any two building components.

[0063] Subsequently, the mapping relationship between conflict areas and building components is introduced, including determining the spatial adjacency factor, stress correlation factor and construction correlation factor between different components. The building components with correlation are regarded as a set of analytical data to determine the similarity under similar component grouping.

[0064] Then, the weighted sum of damage similarity, spatial adjacency factor, stress correlation factor and construction correlation factor is regarded as the output risk value, and a global risk correlation matrix is ​​constructed.

[0065] Furthermore, the spatial adjacency factor represents the degree of closeness between components in space. For example, adjacent components with a direct common boundary are set to 1.0; vertically aligned load-bearing components on different floors are set to 0.8; close neighboring components with a distance of less than 1m are set to 0.6; non-adjacent components on the same floor are set to 0.2; and components with no spatial association are set to 0.

[0066] Furthermore, the force correlation factor is used to measure the force transmission relationship between components. For example, the core components of beams, columns, and walls at the same node are set to 0.9; the components of upper and lower layers on the vertical force transmission path are set to 0.7; the floor slabs and surrounding beams in the same section are set to 0.5; and the components without direct force transmission are set to 0.1.

[0067] Furthermore, the construction correlation factor is used to measure the common impact of construction by associating the construction scenario time with the current deviation. The factor selected here is only to emphasize the relative commonality of construction and is ignored when the sequential relationship of component construction is not considered. For example, the value is set to 0.8 for the same type of component constructed by the same team on the same day; 0.7 for the same batch of formwork and the same pouring batch; 0.3 for different teams on the same floor; and 0 for no construction commonality.

[0068] Specifically, such as Figure 4 As shown, one implementation of step S4 includes: S41, determining the set of components to be analyzed based on the mapping relationship between conflict areas and building components.

[0069] When determining the set of components to be analyzed, the pairing conditions of the building components are determined based on the topological relationship of the building components; building components that meet any pairing condition are entered into the component set.

[0070] The selected pairing conditions must meet at least the following criteria: 1. Multiple building components are located on the same vertical transmission chain, pairing walls at the same vertical position. These components bear the same vertical load, and their deformation is continuous. 2. Multiple building components are located on the same horizontal constraint chain, pairing adjacent horizontally arranged walls forming L-shaped or T-shaped corners. These components share a common constrained floor slab and exhibit common variations in their foundations. 3. The building components belong to the same type of component within the same construction section and unit type. Components meeting the above conditions exhibit relative commonality in their structural distribution. Abrupt deviations in their overall structure will reflect construction defects, thereby improving the accuracy of risk correlation handling.

[0071] It should be noted that the pairing conditions need to be based on the type of the component to be measured, and the measurement method and judgment rules of the corresponding component need to be obtained from the database to determine the commonalities of the current component under similar grouping.

[0072] S42, calculate damage similarity based on the spatial gradient of the deviation of building components in the component set, and group similar components based on the damage similarity to obtain multiple component groups.

[0073] Furthermore, similar components are grouped using a hierarchical clustering approach. By calculating the damage similarity matrix, the selected building components are used as initial clusters, and the initial clusters are iteratively merged to obtain the component grouping at the end of the iteration.

[0074] Specifically, based on the spatial gradient of the building components, after aligning them with the spatial gradients of other components, damage similarity is calculated. Each building component is regarded as an initial cluster, and an initial inter-cluster similarity matrix is ​​generated. The initial value of this matrix is ​​consistent with the damage similarity between components.

[0075] Next, the off-diagonal elements of the current inter-cluster similarity matrix are traversed to find the maximum similarity value, and the cluster numbers corresponding to this value are recorded. It is then determined whether the corresponding damage similarity is greater than or equal to 0.7. If this value is met, the similarity is considered sufficient for merging; otherwise, the clustering cycle is terminated directly. Here, 0.7 represents the threshold for iteration termination, which can be defined based on the average damage similarity between the same component and other components in historical data.

[0076] For the merged clusters, the initial inter-cluster similarity matrix is ​​updated based on the average damage similarity of all components between the clusters until the maximum similarity is less than 0.7, at which point the iteration stops. The remaining clusters are regarded as groups of similar components in the output to illustrate the relative commonalities under different damage conditions.

[0077] It should be noted that during spatial gradient alignment, the spatial gradient sequences of the two components are first normalized to the same scale along the length of the components, and then the gradient sequences are resampled to the same length through linear interpolation. Finally, the Pearson correlation coefficient of the two sets of gradient sequences is calculated to obtain the damage similarity.

[0078] S43. Statistically calculate the spatial adjacency factor, stress correlation factor, construction correlation factor, and damage similarity of each component group in the spatial distribution, and then sum them by weight to obtain the risk value of the risk correlation matrix.

[0079] In the weighted summation, since each component group contains multiple building components, the risk value corresponding to any two building components is obtained by weighted summation based on the spatial adjacency factor, stress correlation factor, construction correlation factor and damage similarity of the current component and other components. At this time, the weights are set to 0.25, 0.2, 0.15 and 0.4 respectively to emphasize the risk between different components.

[0080] In one embodiment of the present invention, an updatable feature state is set for each building component, and multiple feature states are obtained by dividing them according to the quartiles; at the same time, the output warning structure is determined based on the consistency of the feature states of adjacent building components.

[0081] Specifically, one implementation of step S5 includes: extracting the risk values ​​between all building components from the risk association matrix, and setting characteristic states for pairwise combinations of building components based on the quartiles of the risk values ​​sorted in ascending order.

[0082] For adjacent building components in any characteristic state, if the deviation distribution trend is consistent, the corresponding building component is considered as an output warning structure. Specifically, all components are sorted in ascending order of risk value to determine the quartiles at 25% and 75%. Based on the range of risk values ​​falling within these quartiles, characteristic states are categorized as low-risk or high-risk. Components in the same characteristic state are determined as independently output warning structures based on trend continuity, spatial correlation, and risk value range. These are then entered into an external control terminal. After multi-faceted verification of the measurements on the external control terminal, the processing method for the corresponding location is determined.

[0083] Furthermore, consistent deviation distribution trends indicate that the main gradient directions are consistent, and the way the marker direction and state identifier combination conflict is consistent, representing a description of cluster differences in continuous transmission; if the identified building components do not meet the trend consistency, an early warning is issued in the form of quartiles.

[0084] If the deviation distribution trends of adjacent building components are inconsistent, the risk values ​​between building components are used for screening to determine the output component order and obtain the output warning components. Specifically, if the risk value of a building component is greater than 0.6, adjacent building components with risk association will be output; for building components with a risk value less than this, warnings will be issued one by one according to the value of the maximum deviation from large to small to determine the warning structure after measurement.

[0085] In this embodiment, the early warning structure is a continuous set of building components that meet the requirements of spatial adjacency and connectivity, consistent deviation distribution trends, and risk level annotations. It is a clustered risk unit identified based on a risk correlation matrix, rather than an isolated single-point deviation warning point. It directly corresponds to batch and continuous quality problems caused by the same source of construction, and is used to support regional-level batch rectification and common process management to achieve batch early warning identification of multiple components.

[0086] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered by the protection scope of the present invention.

Claims

1. A smart measurement method based on Bluetooth transmission and automatic data entry, characterized in that, include: S1, based on the measurement point layout rules of each building component, obtain the measurement point layout position of the component to be measured; S2, taking the location of the measuring points within the local space window as input, and combining the component type of the component to be measured, determines the target measurement data for the entity embedding; S3, call the marker direction and status indicator of the target measurement data to determine the conflict area and obtain the deviation distribution trend of the conflict area; S4 maps the conflict area to the building components and groups adjacent components based on the deviation distribution trend to establish a risk correlation matrix; S5. Based on the risk correlation matrix, feature states are divided to determine the early warning structure after comparison of multiple feature states.

2. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 1, characterized in that, The implementation methods for setting up the measuring points in step S1 include: S11, Obtain the continuous coordinate point set of the outer contour of the component to be measured, and determine the region boundary of the component to be measured; S12, based on the spatial range of the region boundary, delineate the detection items within the corresponding range, synchronize the detection items with the measurement point layout rules, and obtain the measurement point layout positions under the region boundary.

3. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 1, characterized in that, The methods for implementing the target measurement data in step S2 include: S21, based on the component type of the component to be measured, obtain the minimum number of measuring points and the maximum allowable interval corresponding to the component to be measured; S22, based on the minimum number of measuring points and the maximum allowable interval corresponding to the component to be measured, perform overall polling of spatially adjacent measuring points to determine the divided local spatial window; S23, each measuring point within each local window is regarded as a node, and edges are established based on spatial distance and the continuity of components within the window to construct a spatial topology; S24. For each measurement point within the spatial topology, aggregate the features of all adjacent measurement points to form a feature representation of entity embedding, and use the corresponding data as the output target measurement data.

4. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 3, characterized in that, When determining the division of local spatial windows, the implementation methods include: Based on the spatial range of the measurement points, all spatially adjacent measurement points are successively polled to preliminarily determine the neighborhood relationships of the measurement points. Spatial clustering of measurement points is performed using the spatial Euclidean distance between adjacent measurement points. When the number of measurement points in each cluster is greater than the minimum number of measurement points and the spatial Euclidean distance between any two measurement points is less than the maximum allowable interval, the measurement points of the corresponding cluster are regarded as a set of neighborhood points. The neighborhood point set is continuously spliced ​​according to its spatial distribution to form several local spatial windows.

5. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 3, characterized in that, When forming the feature representation of entity embedding, the implementation methods include: For continuously input measurement data, when there are fluctuations in the measurement data, an initial feature set for the corresponding measurement point is constructed; For each measurement point within a local spatial window, feature fusion is performed on the initial feature set of the measurement point based on the spatial topology to generate graph convolution features for multi-point collaborative measurement. By pooling the graph convolutional features, the features of each local spatial window are converted into an initial embedding vector; the initial embedding vector is then contrast-enhanced to obtain the target measurement data for entity embedding.

6. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 1, characterized in that, The methods for achieving the deviation distribution trend in step S3 include: S31, Based on the deviation between the target measurement data and the standard value, principal component analysis is performed on the target measurement data, and the principal gradient direction of the analysis is used as the label direction; S32, set the status indicator of the target measurement data based on the range of deviation values ​​and the proportion of measurement points with deviations exceeding the allowable range; S33, determine the conflict area based on the clustering area of ​​the marker direction and the status indicator, and output the deviation distribution trend based on the combination of the marker direction and the status indicator.

7. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 6, characterized in that, The methods for determining conflict zones include: For each local spatial window, the clustering areas of positive deviation measurement points, negative deviation measurement points, and out-of-tolerance points are statistically analyzed. When the proportion of out-of-poor points is less than a preset threshold and there are at least three out-of-poor points in a contiguous area, the current cluster area is determined to be a conflict area based on the distribution location of the out-of-poor points. When a local spatial window is marked as having a positive deviation trend, the current clustering area is determined to be a conflict area based on the proportion and distribution of negative deviation measurement points. When a local spatial window is marked as having a negative deviation trend, the current clustering area is determined to be a conflict area based on the proportion and distribution of positive deviation measurement points.

8. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 1, characterized in that, Specifically, the implementation methods of the risk correlation matrix in step S4 include: S41, Based on the mapping relationship between conflict areas and building components, determine the set of components to be analyzed; S42, calculate damage similarity based on the spatial gradient of the deviation of building components in the component set, and group similar components based on damage similarity to obtain multiple component groups; S43. Statistically calculate the spatial adjacency factor, stress correlation factor, construction correlation factor, and damage similarity of each component group in the spatial distribution, and then sum them by weight to obtain the risk value of the risk correlation matrix.

9. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 8, characterized in that, When determining the set of components to be analyzed, the following methods can be used: Based on the topological relationships of building components, the pairing conditions of building components are determined, and building components that meet any pairing condition are entered into the component set.

10. The intelligent measurement method based on Bluetooth transmission and automatic data entry according to claim 1, characterized in that, The implementation methods of the early warning structure in step S5 include: Extract the risk values ​​between all building components from the risk correlation matrix, and set the characteristic states for the pairwise combinations of building components based on the quartiles of the risk values ​​sorted in ascending order. For adjacent building components with any characteristic state, if the deviation distribution trend is consistent, the corresponding building component is regarded as the output warning structure. If the deviation distribution trends of adjacent building components are inconsistent, the risk values ​​between building components are used for screening to determine the output component order and obtain the output warning components.

Images