Fusion visualization method and device for multi-source engineering monitoring data

Abstract

The invention relates to a fusion visualization method and device for multi-source engineering monitoring data. The method comprises the following steps: carrying out preprocessing of verification, cleaning and restoration on collected monitoring data; constructing a data matrix based on the preprocessed multi-source heterogeneous monitoring data; calculating a comprehensive attribute value sequence based on the influence weights of the attributes of different dimensions; and converting the comprehensive attribute value sequence into visual data coordinates based on the determined data proportion. According to the invention, the monitoring data are verified, cleaned and supplemented, so that errors and abnormal data in the monitoring data are reduced; converting the spatial position information and the attribute information of the monitoring data into intermediate data in a visual format; interpolating in a three-dimensional space according to the intermediate data and forming a three-dimensional visual curved surface to realize the visualization of the monitoring data; and through monitoring and early warning judgment, the attribute abnormity level and early warning information are represented in the three-dimensional visual curved surface through colors.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering monitoring, and in particular to a method and device for fusion visualization of multi-source engineering monitoring data. Background Art

[0002] In modern engineering construction, with the expansion of project scale and the increase of complexity, engineering monitoring has become an important means to ensure construction safety, control quality and reduce environmental impact. Engineering monitoring not only helps to grasp the structural changes, safety performance, construction progress and other conditions of buildings or structures in real time, but also can timely discover potential safety hazards and quality problems, take corresponding preventive measures to ensure the safety and quality of the project. In addition, engineering monitoring can also help optimize construction plans, improve engineering efficiency, reduce resource waste and reduce construction costs. Therefore, engineering monitoring is not only an important means to ensure construction safety, but also a key link in realizing refined management of engineering projects.

[0003] However, the monitoring process often generates complex data that is multi-source, heterogeneous, massively growing, and dynamically changing. These data come from a wide range of sources, including sensors, cameras, drones, satellite remote sensing, and other equipment, involving a variety of physical quantities such as temperature, humidity, pressure, displacement, and stress. The data formats are diverse, including both structured data (such as numerical data) and unstructured data (such as images and videos). Data from different sources differ in time, space, and semantics, which increases the difficulty of data integration. At the same time, with the popularization of monitoring equipment and the advancement of technology, the generation speed and storage volume of engineering monitoring data are growing exponentially, posing huge challenges to data processing and analysis.

[0004] Faced with such complex monitoring data, traditional data processing and analysis methods are difficult to meet the needs of rapid decision-making. Simply relying on raw monitoring data for analysis is not only time-consuming and laborious, but also difficult to intuitively display the trends and laws behind the data, and cannot provide effective decision-making support for managers. Therefore, how to convert complex monitoring data into a visual form that is easy to understand and use has become an urgent problem to be solved in the field of engineering management. As an important tool to connect data processing with practical applications, data visualization can present abstract monitoring data in the form of graphics, charts, models or dynamic displays, helping managers to quickly identify problems, evaluate status and optimize decisions. Through data visualization, complex monitoring data can become more intuitive and easy to understand, allowing managers to obtain key information in a short time and make scientific and reasonable decisions. In addition, data visualization can also promote communication and collaboration among team members and improve the overall management level of the project.

[0005] In recent years, with the rapid development of emerging technologies such as big data, cloud computing, and artificial intelligence, data processing and visualization technologies in the field of engineering monitoring have also made significant progress. By combining advanced data processing technology and visualization methods, not only can the accuracy and efficiency of monitoring data be improved, but also early warning and dynamic analysis can be achieved, providing strong support for intelligent and digital management in the engineering field. For example, by verifying, cleaning, and supplementing monitoring data, errors and abnormal data can be effectively reduced, and the quality and reliability of data can be improved. Using big data platforms and distributed computing technologies, monitoring data from different sources can be integrated to form a unified data set. This multi-source data fusion not only improves the integrity and consistency of the data, but also provides a solid foundation for subsequent analysis and visualization. Through three-dimensional modeling technology, the spatial location information and attribute information of monitoring data can be converted into intermediate data in a visual format, and interpolated in three-dimensional space to form a three-dimensional visualization surface. This three-dimensional visualization method can intuitively display the actual situation of the engineering site and help managers better understand the progress of the project, structural changes, and safety conditions. Based on machine learning and deep learning algorithms, monitoring data can be analyzed in real time, potential risks and abnormal situations can be identified, and early warning information can be intuitively displayed on the visualization interface through color coding, icon annotation, and other methods. This dynamic early warning system can not only detect safety hazards in advance, but also provide managers with timely decision-making support to avoid accidents.

[0006] Although engineering monitoring plays an irreplaceable role in modern engineering construction, the existing monitoring data processing and analysis methods still have some obvious defects, which limit their effectiveness in practical applications. In engineering monitoring, especially for large projects, the amount of data generated is very large. How to effectively display this data in a limited visualization interface is a challenge. If the visualization interface is not designed properly, it may lead to information overload and users will find it difficult to extract useful information from it. For example, too many charts, graphs, and data points may make the interface look cluttered, making it difficult for users to quickly find the data they care about. In addition, some unnecessary information may take up valuable screen space, distract users' attention, and reduce decision-making efficiency.

[0007] Although some visualization tools already have basic early warning functions, most of these warnings are based on preset thresholds or simple rules and lack the ability of intelligent analysis and dynamic adjustment. When monitoring data is abnormal, the system may not be able to issue accurate early warning signals in time, or the false alarm rate is high, causing managers to ignore potential risks. In addition, existing early warning systems usually only provide single alarm information, lack in-depth analysis of the root cause of the problem, and it is difficult to help managers formulate effective response measures.

[0008] For example, CN119168494A discloses a quality control platform and method for asphalt pavement engineering. By performing quality control planning and multi-source monitoring, multi-source monitoring data is obtained; video monitoring is performed on multiple key construction areas; quality control and registration records of materials entering and leaving the site are performed; equipment performance monitoring, status monitoring and work monitoring are performed; asphalt mixing monitoring and asphalt test monitoring are performed; intelligent compaction monitoring and completion acceptance monitoring are performed. This technical solution can generate control planning data, perform multi-source monitoring, display real-time monitoring videos, visual management data of materials entering and leaving the site, visual data of equipment monitoring, visual data of asphalt monitoring, and visual data of compaction test acceptance monitoring, and can achieve automatic and precise tracking and strict management of asphalt pavement engineering projects, and the engineering quality control is comprehensive and meticulous, and can achieve remote and real-time engineering supervision. However, this technical solution does not repair some damaged data, resulting in a reduction in the accuracy of the monitoring information of the visual interface, and at the same time, the accuracy of the early warning is also reduced.

[0009] For another example, CN117056867A discloses a multi-source heterogeneous data fusion method and system that can be used for digital twins, extracts format information and protocol information from multiple data sources, obtains multi-source data through a server parser, standardizes and maps the multi-source data to obtain different standardized data sources, extracts and transforms multiple actual feature information from different standardized data sources, determines whether the multiple actual feature information meets the preset feature requirement range, obtains a multi-source heterogeneous data set, semantically maps the multi-source heterogeneous data set to construct a decision tree model, obtains multi-source heterogeneous fusion data based on the decision tree model, establishes a multi-source heterogeneous fusion database based on the multi-source heterogeneous fusion data, obtains real-time multi-source data, and visualizes the real-time multi-source data to obtain a digital twin model of real-time multi-source data. However, the application of this technical solution in engineering monitoring visualization has many limitations. Specifically, this technical solution has the problems of lack of specific needs for engineering monitoring, limitations of decision tree models, insufficiency of visual expression, lack of real-time early warning mechanism, lack of support for data interpretability, and complexity of technical implementation, which limits its applicability in the field of engineering monitoring.

[0010] In summary, the existing engineering monitoring data visualization methods have many deficiencies in terms of information overload, early warning mechanism, etc., which limit their effectiveness in practical applications. In order to overcome these problems, the present invention provides a fusion visualization method and device for multi-source engineering monitoring data, hoping to solve the current problems.

[0011] In addition, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making the present invention, but due to space limitations, not all details and contents are listed in detail. However, this does not mean that the present invention does not have the characteristics of these prior arts. On the contrary, the present invention already has all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art to the background technology. Summary of the invention

[0012] Faced with complex monitoring data, traditional data processing and analysis methods are difficult to meet the needs of rapid decision-making. Relying solely on raw monitoring data for analysis is not only time-consuming and laborious, but also difficult to intuitively display the trends and laws behind the data, and cannot provide effective decision-making support for managers. Therefore, how to convert complex monitoring data into a visual form that is easy to understand and use has become an urgent problem to be solved in the field of engineering management.

[0013] In view of the shortcomings of the prior art, the present invention provides a fusion visualization method for multi-source engineering monitoring data from a first aspect, the method comprising: preprocessing the collected monitoring data for verification, cleaning and repair; constructing a data matrix based on the preprocessed multi-source heterogeneous monitoring data; calculating a comprehensive attribute value sequence based on the influence weights of attributes of different dimensions; and transforming the comprehensive attribute value sequence into visualized data coordinates based on a determined data ratio.

[0014] The present invention effectively removes noise and outliers, repairs missing data, and ensures the quality and reliability of data through preprocessing steps. This provides a solid foundation for subsequent analysis and visualization. The present invention can extract the most representative information from multidimensional data by constructing a data matrix, reducing the dimension of the data while retaining the most important features. This method not only improves the interpretability of the data, but also simplifies the visualization process. The present invention introduces the concept of data ratio, which can be flexibly adjusted according to the specific conditions of the engineering area, so that the distribution of data points on the visualization interface is more reasonable, avoiding information overload or too sparse situations. The present invention can also achieve real-time monitoring and dynamic display through the processing and visualization of time series data, helping users to discover potential risks and problems in a timely manner.

[0015] According to a preferred embodiment, the preprocessing step includes: verifying the location information and / or time information in the monitoring data and deleting duplicate data; judging and identifying outliers in the monitoring data based on the improved standard score, and deleting the outliers; performing time repair and space repair on the monitoring data based on the calculated time repair data and space repair data, thereby obtaining repaired multi-source heterogeneous monitoring data.

[0016] Traditional methods often ignore the integrity and consistency of data when processing raw monitoring data, resulting in unreliable subsequent analysis results. The present invention ensures the accuracy and integrity of data through strict preprocessing steps, eliminates redundant and abnormal data, and improves data quality. The present invention simplifies the subsequent data analysis process by deleting duplicate data and abnormal values, reduces unnecessary calculations, and improves processing efficiency; it also fills in the gaps in the data through time and space repair, ensures the continuity and real-time nature of the data, and makes the monitoring results more accurate and reliable, and can promptly reflect changes in the project status.

[0017] According to a preferred embodiment, the step of constructing a data matrix includes: acquiring displacement monitoring data and stress monitoring data of monitoring points; arranging the displacement monitoring data and stress monitoring data in time sequence, thereby constructing a data matrix of a single monitoring point.

[0018] Traditional methods lack effective organization and sorting methods when processing multidimensional data, making the data difficult to manage and analyze. The present invention constructs a data matrix to arrange different types of monitoring data in chronological order, forming a structured data set, which is convenient for subsequent trend analysis and prediction. Through time series arrangement, the present invention can clearly track the historical changes of each monitoring point, help users understand the changing trends of the data, and discover potential problems. The construction of the data matrix provides a unified format for subsequent feature extraction and visualization, simplifies the data processing process, and improves overall efficiency.

[0019] According to a preferred embodiment, the step of calculating the comprehensive attribute value sequence includes: calculating the covariance matrix of the standardized data; performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors corresponding thereto; sorting the eigenvalues ​​and selecting the eigenvector corresponding to the largest eigenvalue; and calculating the comprehensive attribute value sequence based on the selected eigenvectors and the original monitoring data.

[0020] Traditional methods are prone to "dimensionality disaster" when processing high-dimensional data, resulting in inaccurate analysis results. The present invention effectively reduces the dimension of the data through feature decomposition and principal component analysis, extracts the most important change pattern in the data, avoids information overload, and improves the interpretability of the data. The present invention ensures that the extracted information is the most representative and stable by sorting the eigenvalues ​​and selecting the eigenvector corresponding to the maximum eigenvalue, avoiding misjudgment caused by interference from minor factors. The comprehensive attribute value sequence is a dimensionality reduction representation of multidimensional data, which can more clearly reflect the main characteristics and trends of the data, facilitate subsequent visualization, and help users quickly understand the meaning behind the data.

[0021] According to a preferred embodiment, the step of determining the data ratio includes: calculating the length and width of the plane rectangle of the engineering area; calculating the distance between the monitoring points based on the spatial average point spacing of the engineering area; and determining the data ratio based on the length, width and distance between the monitoring points of the plane rectangle of the engineering area.

[0022] Traditional methods lack reasonable proportion adjustment when performing data visualization, resulting in data points being too densely or sparsely distributed on the interface, affecting the user's perception and understanding. The present invention introduces the concept of data proportion and flexibly adjusts it according to the specific conditions of the engineering area, so that the distribution of data points on the visualization interface is more reasonable, avoiding information overload or being too sparse. Through reasonable proportion adjustment, the present invention ensures that the distribution of data points on the interface is neither too crowded nor too scattered, improves the accuracy and aesthetics of visualization, and enables users to observe the changing trend of data more intuitively. According to the size of different engineering areas and the distribution of monitoring points, the present invention can dynamically adjust the data proportion, so that the method can adapt to engineering projects of various sizes and complexities, and has a wide range of applicability.

[0023] According to a preferred embodiment, the step of transforming the comprehensive attribute value sequence based on the determined data ratio includes: visualization data point coordinates = comprehensive attribute value sequence × data ratio.

[0024] Traditional methods often lack an effective coordinate transformation mechanism when performing data visualization, resulting in unreasonable distribution of data points and difficulty in intuitively displaying the changing trend of data. The present invention uses a coordinate transformation formula to map abstract comprehensive attribute values ​​to a specific two-dimensional or three-dimensional coordinate system, so that data points can be intuitively displayed on the visualization interface, helping users to quickly understand the meaning behind the data. Through coordinate transformation, the present invention realizes real-time updating and dynamic display of data, and users can view the latest monitoring data at any time and discover potential risks and problems in a timely manner. At the same time, the visualization interface is highly interactive, and users can flexibly adjust the view according to their needs and conduct multi-dimensional analysis. Through intuitive visualization, users can more clearly observe the changing trends and distribution patterns of data, helping them make more scientific decisions and improve the efficiency and quality of engineering management.

[0025] The present invention provides a fusion visualization device for multi-source engineering monitoring data from a second aspect, the device includes a processor, and the processor includes a preprocessing module, a matrix construction module and a calculation module. The preprocessing module is used for preprocessing the collected monitoring data for verification, cleaning and repair; the matrix construction module is used for constructing a data matrix based on the preprocessed multi-source heterogeneous monitoring data; the calculation module is used for calculating a comprehensive attribute value sequence based on the influence weights of attributes of different dimensions, and transforming the comprehensive attribute value sequence into a visualized data coordinate based on a determined data ratio.

[0026] The device verifies, cleans and repairs the collected monitoring data through the preprocessing module, ensuring the accuracy and consistency of the data. In particular, the verification of location information and time information, as well as the identification and deletion of outliers, eliminates noise and errors in the data and improves the reliability of the data. By repairing the missing time and space data, the present invention fills the gaps in the data, ensures the integrity and continuity of the data, and avoids analysis errors caused by incomplete data. The various modules of the device (preprocessing module, matrix construction module and calculation module) can work in parallel, making full use of the advantages of multi-core processors, further improving the efficiency of data processing, and shortening the cycle from data collection to visualization.

[0027] According to a preferred embodiment, the preprocessing module is configured to: verify the location information and / or time information in the monitoring data and delete duplicate data; judge and identify outliers in the monitoring data based on the improved standard score, and delete the outliers; perform time repair and space repair on the monitoring data based on the calculated time repair data and space repair data, thereby obtaining repaired multi-source heterogeneous monitoring data.

[0028] The preprocessing module can efficiently complete the verification, cleaning and repair of data, reduce the repeated processing and redundant calculations that may occur in subsequent analysis, and improve the speed of overall data processing. The verification step of the preprocessing module ensures that the time and space coordinates of the data are accurate, avoiding analysis deviations caused by position or time errors. Based on the improved standard score method, the distance between the data point and the average value can be quantified to ensure that outliers will not affect the subsequent analysis results. By calculating the time repair data and space repair data, the preprocessing module can reasonably interpolate and complete the missing data, ensuring the continuity and integrity of the data and avoiding analysis errors caused by missing data.

[0029] According to a preferred embodiment, the matrix construction module is configured to: obtain displacement monitoring data and stress monitoring data of a monitoring point; arrange the displacement monitoring data and stress monitoring data in time sequence, thereby constructing a data matrix of a single monitoring point.

[0030] This structured data organization of the matrix building module ensures the temporal order and logical consistency of the data, which is convenient for subsequent trend analysis and feature extraction. Through time series arrangement, the matrix building module can eliminate duplicate data records, ensure that the data at each time point is unique, and avoid analysis errors caused by data redundancy.

[0031] According to a preferred embodiment, the calculation module is configured to: calculate the covariance matrix of the standardized data; perform eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors corresponding thereto; sort the eigenvalues ​​and select the eigenvector corresponding to the largest eigenvalue; calculate the comprehensive attribute value sequence based on the selected eigenvector and the original monitoring data; visualize the data point coordinates = comprehensive attribute value sequence × data ratio.

[0032] The calculation module can extract the most important change patterns in the data, reduce the interference of irrelevant information, and ensure the accuracy of feature extraction. By sorting the eigenvalues ​​and selecting the eigenvector corresponding to the maximum eigenvalue, the calculation module can retain the most critical information in the data and avoid analysis errors caused by minor factors. Based on the selected eigenvectors and the original monitoring data, the calculation module can accurately calculate the comprehensive attribute value sequence, ensuring that the dimensionality reduction representation of the data is both concise and comprehensive, avoiding information loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a simplified module connection relationship diagram of a fusion visualization device for multi-source engineering monitoring data provided by the present invention;

[0034] Figure 2 It is a flow chart of a fusion visualization method of multi-source engineering monitoring data provided by the present invention;

[0035] Figure 3 It is a principle logic schematic diagram of a fusion visualization device for multi-source engineering monitoring data provided by the present invention.

[0036] Reference numerals list

[0037] 100: processor; 110: preprocessing module; 120: matrix building module; 130: computing module; 200: terminal; 210: display interface; 220: interactive component; 230: terminal processor. DETAILED DESCRIPTION

[0038] The following is a detailed description with reference to the accompanying drawings.

[0039] When dealing with complex monitoring data, traditional analytical methods often fail to meet the requirements of rapid decision-making. Relying solely on raw monitoring data for interpretation is not only inefficient, but also difficult to clearly present the potential patterns and trends in the data, thus limiting managers' ability to gain effective insights. Therefore, how to convert these complex monitoring data into intuitive and easy-to-operate visualizations has become a technical problem that urgently needs to be solved in engineering management.

[0040] In view of the deficiencies of the prior art, the present invention provides a method and device for fusion visualization of multi-source engineering monitoring data. The present invention may also provide a processor 100, which can run the coding program of the fusion visualization method of multi-source engineering monitoring data of the present invention. The hardware of the processor 100 includes a dedicated integrated chip, a server, a cloud server and a combination thereof. Preferably, the fusion visualization device of multi-source engineering monitoring data may also include a terminal 200 connected to the processor 100. Preferably, the terminal 200 includes a display interface 210, an interactive component 220 and a terminal processor 230. The display interface 210 is used to display the data point coordinates and data curves of the monitoring point in a visual manner. The interactive component 220 is used for the user to interact with the terminal 200. The interactive component 220 includes a touch screen, a mouse, a keyboard, etc. The terminal processor 230 is a processor or a microprocessor arranged in the terminal 200, which is used for data reception and visualization control of the terminal 200.

[0041] The present invention may also provide a storage medium for storing the coding program of the fusion visualization method of multi-source engineering monitoring data of the present invention. The storage medium may be a random access memory (RAM), a read-only memory (ROM), a magnetic surface memory, an optical memory, a solid state drive (SSD), a register, etc.

[0042] The fusion visualization device of multi-source engineering monitoring data of the present invention is as follows Figure 1 As shown, the device includes a processor 100. Figure 1 As shown, the processor 100 includes a pre-processing module 110, a matrix construction module 120 and a calculation module 130 which are sequentially data-connected. Preferably, the matrix construction module 120 and the calculation module 130 can also be integrated into a calculation function module.

[0043] Preferably, when the processor 100 is an independent processor 100 , the preprocessing module 110 , the matrix construction module 120 and the calculation module 130 are functional modules inside the processor 100 .

[0044] Preferably, the preprocessing module 110, the matrix building module 120 and the computing module 130 may also be independent physical hardware, such as processors. The preprocessing module 110, the matrix building module 120 and the computing module 130 may be connected to each other via a data line to achieve data transmission.

[0045] First, the preprocessing module 110 is used to verify, clean and repair the collected monitoring data to ensure the quality and consistency of the data. The output of the preprocessing module 110 is processed high-quality data, which is passed as input to the matrix construction module 120. Next, the matrix construction module 120 uses the preprocessed data to construct a data matrix and integrates the multi-source heterogeneous monitoring data into a unified format for subsequent calculations and analysis. The constructed data matrix will be passed as input to the calculation module 130. Finally, the calculation module 130 calculates the influence weights of attributes of different dimensions based on the data matrix and generates a sequence of comprehensive attribute values. At the same time, the module also converts these comprehensive attribute values ​​into visual data coordinates for further data analysis and display.

[0046] The device verifies, cleans and repairs the collected monitoring data through the preprocessing module 110, ensuring the accuracy and consistency of the data. In particular, the verification of location information and time information, as well as the identification and deletion of outliers, eliminates noise and errors in the data and improves the reliability of the data. By repairing the missing time and space data, the present invention fills the gaps in the data, ensures the integrity and continuity of the data, and avoids analysis errors caused by incomplete data. The various modules of the device (preprocessing module 110, matrix construction module 120 and calculation module 130) can work in parallel, making full use of the advantages of the multi-core processor 100, further improving the efficiency of data processing, and shortening the cycle from data collection to visualization.

[0047] The receiving port of the processor 100 of the device of the present invention receives the location coordinates, data attribute values ​​and monitoring time of the monitoring data, and stores the data in a database form.

[0048] The engineering monitoring data of the present invention generally includes geological monitoring, structural monitoring and environmental monitoring. Specifically, geological monitoring is used to monitor the changes in terrain and geology, and generally uses instruments such as total stations, radars, sensors, etc. for monitoring, and the result data is generally displacement or positioning data; structural monitoring is used to monitor the changes in stress and strain, and generally uses instruments such as strain gauges and strain meters for monitoring, and the result data is stress and strain data; environmental monitoring is used to monitor the impact of environmental factors, including noise, air, precipitation and other information, and the result data is attribute data of various environmental factors.

[0049] In this embodiment, the engineering monitoring data acquired by data collection are displacement data for geological monitoring and stress data for structural monitoring.

[0050] Each set of collected engineering monitoring data should contain location information, attribute information and time information. Specifically, the location information refers to the actual location of the monitoring instrument that collected the monitoring data, which needs to include the longitude coordinate X, latitude coordinate Y and elevation coordinate H. The attribute information is the attribute information monitored by the instrument itself, such as the attribute information of stress monitoring data, which is the stress value at the location of the instrument. The time information is the time when the instrument monitors the current value, which needs to be distinguished from the time when the data is uploaded.

[0051] In this embodiment, the location information and time information structure of the displacement monitoring data and the stress monitoring data are the same. For attribute information, the displacement monitoring data contains displacements in three directions, namely, longitude displacement DX, latitude displacement DY, and elevation displacement DH. For stress information, there is only one dimension of data, namely stress data. This reflects the multi-source heterogeneity of monitoring data.

[0052] Preferably, the preprocessing module 110 performs step S1, such as Figure 2 shown.

[0053] S1: Preprocessing of collected monitoring data for verification, cleaning and repair.

[0054] The acquired multi-source heterogeneous monitoring data needs to go through a series of preprocessing before visualization. Specifically, the original monitoring data cannot guarantee that all of them are highly accurate, complete and comprehensive data. For example, engineering monitoring data may generate noise, erroneous values, missing values ​​and duplicate values ​​due to sensor failure, communication interruption or environmental interference. These erroneous data may be included in the analysis, leading to wrong conclusions or risk warnings. Data integrity is reflected in the fact that due to the uniqueness of the monitoring equipment and the regularity of the monitoring time, there is a certain two-dimensional correlation between the monitoring equipment and the monitoring time, that is, for each type of monitoring, each monitoring device represents a set of monitoring data in each monitoring time period. Therefore, for the acquired monitoring data, the preprocessing process of verification, cleaning and repair should be carried out first.

[0055] In this embodiment, each time a batch of monitoring data is received, in an ideal situation without abnormalities or errors, it is the displacement monitoring data and stress monitoring data of different monitoring positions in the same monitoring time period. Different monitoring positions generally correspond to monitoring points set in the engineering monitoring area. That is, in this embodiment, the distribution of all monitoring data in space is in a discrete form with monitoring points as aggregation points.

[0056] S11: Verify the location information and / or time information in the monitoring data and delete duplicate data.

[0057] The purpose of monitoring data verification is to determine whether a new set of monitoring data is a duplicate value. In this embodiment, monitoring data verification is the verification of the location information and time information in each set of collected monitoring data. In actual engineering monitoring, monitoring instruments of the same type will not monitor at the same spatial position or very close spatial positions at the same monitoring time; based on the nature of time, there should not be two or more monitoring data of the same type at the same time.

[0058] Specifically, when new monitoring data is obtained, its monitoring time is compared with the same type of monitoring data that has been obtained. If monitoring data with exactly the same monitoring time is found, the location information of these data is compared. If the location information is exactly the same or close, it is judged as a duplicate value. The old data will be deleted.

[0059] The reason for deleting the old data is that the source of the old monitoring data may be repeated data collected for technical reasons, or it may be an estimated value for monitoring data repair. Replacing the old monitoring data with new original monitoring data is more beneficial to the final visualization and can better reflect the actual situation. In this embodiment, the proximity of monitoring data location information means that the spatial distance d between two sets of monitoring data of the same type is less than 1 meter. Assume that the longitude coordinates of the two sets of monitoring data are x 1 and x 2 , and the latitude coordinates are y 1 and 2 , the elevation coordinates are h 1 and h 2 , then the spatial distance between the two sets of monitoring data is This formula can also be applied to the calculation of all the following spatial distances.

[0060] S12: Clean the monitoring data.

[0061] The purpose of monitoring data cleaning is to determine whether a new set of monitoring data is noise or error values, that is, outliers, and delete such outliers.

[0062] The cleaning step focuses on the attribute information of the monitoring data for judgment, i.e., stress value, displacement value, etc. In this embodiment, the monitoring data is judged and identified as an outlier by means of the improved standard score. The improved standard score Z is modified on the basis of the standard score calculation formula to make it more consistent with the characteristics of the monitoring data changing over time and the characteristics of the spatial points, and to make the outliers more significant. By comparing the value of the improved standard score to judge whether the monitoring data is an outlier, the data identified as an outlier will be deleted.

[0063] S121: When a new set of monitoring data is obtained, its attribute information is read and a set of reference data sets is obtained from the same type of monitoring data obtained. Taking stress monitoring data as an example, the target data read is stress data, and the reference data set is 10 sets of data in the acquired stress monitoring data that are close to the target monitoring data and relatively new in time. Its role is to serve as a standard for judging the deviation between the target data and the overall data.

[0064] S122: Calculate the mean μ and standard deviation σ of the reference data set.

[0065] The value of the improved standard score Z of the target data X can be calculated from the calculated mean μ and standard deviation σ of the reference data set.

[0066]

[0067] In this formula, sign(X-μ) is a sign function that preserves the sign of the difference between the target data and the reference data set mean.

[0068] S123: Determine the calculated improvement standard score Z of the target data. When |Z|≥1, the current stress monitoring data is considered to be an abnormal value and should be deleted.

[0069] Preferably, the condition for determining that the locations are close is that the spatial distance d corresponding to the two sets of stress monitoring data is less than 5 meters.

[0070] S13: Perform time repair and space repair on the monitoring data based on the calculated time repair data and space repair data, so as to obtain repaired multi-source heterogeneous monitoring data.

[0071] The function of monitoring data repair is to fill missing values ​​and deleted abnormal values ​​so that the two-dimensional data matrix of spatial monitoring points and time period series remains intact. The monitoring data repair step is different from the verification and cleaning steps. The latter two only need to be triggered when new monitoring data is acquired, while the repair step not only needs to be triggered when new monitoring data is acquired to supplement the missing monitoring data during acquisition, but also needs to be triggered after the verification and cleaning steps are completed to supplement the deleted monitoring data. Due to the spatiotemporal two-dimensionality of monitoring data, this method combines time series interpolation and spatial point interpolation. Interpolation is to estimate data points at unknown locations through known discrete data points. Due to different types of monitoring data, the influence relationship between time and space is different. The present invention dynamically adjusts the influence coefficient of time series interpolation and spatial point interpolation with the help of weight factors, so that the final repaired monitoring data is closer to the actual situation.

[0072] Taking the process of obtaining new monitoring data as an example, first determine whether there is missing data in the new monitoring data. The specific judgment method is: in the monitoring data of the same category, whether there is a monitoring point that has no corresponding monitoring data. If there is missing data, the data is repaired through the following steps.

[0073] S131: Calculate time repair data.

[0074] Find the data time series of the monitoring points and monitoring types corresponding to the missing data, and obtain the two sets of monitoring data of the same type that are closest in time to the target repair data.

[0075] For example, if the missing data is the stress monitoring data X at time t, the two most recent sets of stress monitoring data at the corresponding monitoring point are x 2 and x 1 , the corresponding monitoring time is t 2 and t 1 , where x 2 It is the monitoring data closest in time to the missing stress data.

[0076] From this, the target time repair data can be calculated.

[0077]

[0078] Among them, α is the weight factor.

[0079] Preferably, in order to emphasize the importance of time in the stress data series, that is, to perform data repair based on recent data, the weight factor α in the formula is set to 0.7.

[0080] S132: Calculate spatial repair data.

[0081] The data of the same category as the target repair data are distinguished from the monitoring data of the same monitoring time batch, and 8 groups of data adjacent to the target repair data in space are obtained as the spatial reference data set.

[0082] Preferably, the method for finding adjacent monitoring data is: calculating the spatial distance d between the target repair data and the monitoring data of the same category at the same time, arranging and selecting 8 groups of data with the smallest spatial distance d as the spatial reference data set.

[0083] Calculate the spatial weight w of each of the 8 groups of data in the spatial reference data set respectively. The calculation of spatial weight is: Here, d represents the spatial distance, and p represents the distance attenuation parameter.

[0084] Since the distance has a great influence on the monitored attribute value, and the monitored attribute value has a certain local variation property, the distance attenuation parameter p should take a larger value. In this embodiment, the value of p is 3.

[0085] Assuming that the attribute value of each data in the spatial data reference set is y, the spatial repair data of the target can be estimated:

[0086] Among them, w i represents the spatial weight of the i-th monitoring point; y i represents the attribute value of the i-th monitoring point; N represents the number of groups in the spatial reference data set.

[0087] S133: Calculate the final restoration data of time and space fusion.

[0088] Final repaired data value: S=αX+(1-α)Y.

[0089] Since different types of monitoring data have different sensitivities to spatial and temporal changes, the displacement monitoring data that needs to be repaired and the stress monitoring data that needs to be repaired have different proportions of time repair data factors and space repair data factors. Specifically, displacement monitoring data is more sensitive to spatial position, while stress monitoring data is more inclined to a certain degree of temporal continuity. In this embodiment, if the monitoring data that needs to be repaired and supplemented is displacement monitoring data, the weight factor α=0.4; if the monitoring data that needs to be repaired and supplemented is stress monitoring data, the weight factor α=0.6.

[0090] One possible scenario is that the missing data is displacement monitoring data. In this case, since there are three sets of displacements corresponding to each set of displacement monitoring data, the longitude displacement, latitude displacement, and elevation displacement are missing, and the repair process should calculate and process each unidirectional displacement separately.

[0091] After the above series of preprocessing steps, the monitoring data has a certain degree of accuracy and completeness. The following is a visualization of the monitoring data.

[0092] The matrix construction module 120 is configured to perform step S2, such as Figure 2 shown.

[0093] S2: Construct a data matrix based on the preprocessed multi-source heterogeneous monitoring data.

[0094] Multi-source data fusion aims to fuse the pre-processed multi-source heterogeneous monitoring data and reduce the dimensions of multiple attribute dimensions of the engineering monitoring area. The overall situation of the engineering monitoring area is reflected through single-dimensional data, that is, the monitoring data is reduced in dimension. Specifically, the longitude displacement, latitude displacement, elevation displacement and stress value of the displacement monitoring data are fused into the comprehensive attribute value of engineering monitoring. Multi-source data is fused through the following steps:

[0095] Obtain the displacement monitoring data and stress monitoring data of the monitoring point. Arrange the displacement monitoring data and stress monitoring data in time sequence to construct a data matrix of a single monitoring point.

[0096] Specifically, multi-source data fusion is performed based on the time series of monitoring data of each monitoring point. For a certain monitoring point, the displacement monitoring data includes three groups of sub-data, namely, the longitude displacement D 1 , Latitude displacement D 2 And the elevation displacement D 3 Each group of sub-data contains n monitoring data in the monitoring time sequence. The stress monitoring data S contains n monitoring data in the time sequence.

[0097] From this, the data matrix of a single monitoring point can be constructed:

[0098] Each column of the data matrix represents displacement monitoring data or stress monitoring data.

[0099] According to a preferred embodiment, the calculation module 130 is configured to perform steps S3 and S4, such as Figure 2 shown.

[0100] S3: Calculate the comprehensive attribute value sequence based on the influence weights of attributes in different dimensions.

[0101] S31: Adjust the influence weight of each attribute.

[0102] The data matrix is ​​introduced into the sequence of influencing weight factors to scale the influence of various attributes.

[0103] Generally speaking, the impact of each group of data on the final attribute is different, so the data matrix is ​​introduced into the sequence of influencing weight factors to scale the impact of each type of attribute.

[0104] The influence weight factor sequence refers to a set of influence weight values ​​corresponding to the features, which is used to adjust the importance of each feature in the model.

[0105] The scaled data matrix is: X′=X original ·diag(β 1 , β 2 , β 3 , β4 ).

[0106] Specifically in the current embodiment, the longitude displacement and the latitude displacement have little influence on the final attribute, and the elevation displacement has a greater influence weight because it represents the settlement displacement to the greatest extent. 1 =1,β 2 =1,β 3 =3,β 4 =2.

[0107] S32: Standardize the displacement monitoring data and the stress monitoring data.

[0108] Since the displacement monitoring data and stress monitoring data have different dimensions, they should be standardized. The specific method is to calculate the mean μ and standard deviation σ of each column of monitoring data in the data matrix, and apply the transformation formula to the data matrix X to obtain the standardized data matrix

[0109] Preferably, for the jth column (i.e., the jth feature), the mean μ j It can be calculated by the following formula:

[0110]

[0111] Here, X ij represents the value in row i and column j. That is, add up all the values ​​in the column and divide by the total amount of data n.

[0112] Standard deviation j It measures the discreteness of the eigenvalue and can be calculated by the following formula:

[0113]

[0114] X ij -μ j It represents the deviation between each monitoring data and the mean, and then sums them after squaring, and finally takes the square root of the average. Note that the denominator uses n-1 instead of n, because unbiased estimation is usually used to calculate the standard deviation.

[0115] S33: Calculate the comprehensive attribute value sequence of engineering monitoring.

[0116] S331: Calculate the covariance matrix of the standardized data,

[0117] S332: Perform eigendecomposition on the covariance matrix to obtain eigenvalues ​​and corresponding eigenvectors.

[0118] S333: Sort the eigenvalues, select the eigenvector corresponding to the largest eigenvalue, and calculate the comprehensive attribute value sequence based on the selected eigenvector and the original monitoring data.

[0119] Specifically, calculate the covariance matrix of the standardized data:

[0120] The covariance matrix C is a 4×4 symmetric matrix. X′ T is the transposed matrix of the data matrix.

[0121] The covariance matrix C is decomposed by the expression Cv=λv to obtain the eigenvalue λ and the corresponding eigenvector v. Sort the eigenvalues ​​from large to small and select the eigenvector v corresponding to the largest eigenvalue. 1 , project the data matrix X′ onto the feature vector v 1 The fused single-dimensional engineering monitoring comprehensive attribute value sequence of this monitoring point is obtained: Z = X′v 1 .

[0122] S4: Based on the determined data ratio, the comprehensive attribute value sequence is transformed into visual data coordinates.

[0123] The data scaling step aims to scale the fused comprehensive attribute data to a range suitable for displaying the fluctuation of the project area attributes. The scaling is related to the size of the project area plane and the distance between the monitoring points.

[0124] S41: Calculate the length and width of the plane rectangle of the project area.

[0125] The influencing factor of the plan size of the project area is described by the length of the project monitoring boundary.

[0126] Specifically, obtain and sort the location information of the monitoring data, and obtain the maximum longitude value X of all monitoring data. max , minimum longitude value X min , maximum latitude value Y max and minimum latitude value Y min , from which the length and width of the plane rectangle of the project area can be calculated.

[0127] The length calculation formula is: ΔX = X max -X min .

[0128] The calculation formula of width is: ΔY=Y max -Y min .

[0129] S42: Calculate the distance between monitoring points based on the spatial average point spacing of the project area.

[0130] Obtain any type of monitoring data at any monitoring time. These data can describe the distribution of monitoring points in the project area with the least amount of data, that is, the distribution of monitoring points. Use these data as the average distance calculation data set. Assuming that for any two data, the spatial distance between them is d, then the distance between each pair of points can be calculated to form a set of distance data sets, and the average value of the data set is d. avg .

[0131] S43: Determine the data ratio based on the length and width of the plane rectangle of the project area and the distance between the monitoring points.

[0132] The data ratio is:

[0133] Visualized data point coordinates = comprehensive attribute value sequence × data ratio.

[0134] From this, the z coordinate of the monitored data after proportional transformation can be calculated:

[0135]

[0136] The transformed coordinates of the data points used for visualization are (x, y, z′). By drawing a surface based on the data points, the fusion visualization of multi-source engineering monitoring data can be realized.

[0137] The visualization method based on multi-source engineering monitoring data of the present invention integrates and visualizes various types of monitoring data collected during engineering construction, providing monitoring personnel with a clear and intuitive engineering status display interface. At the same time, through the visualization method of monitoring and early warning, the engineering status and early warning information generated based on the monitoring data are integrated into the visualization display interface, enriching the information and connotation of the engineering status display interface.

[0138] Specifically, if Figure 3 As shown, various monitoring data in engineering construction are obtained, and the monitoring data includes spatial location information, attribute information and time information; the monitoring data is verified, cleaned and supplemented to reduce the errors and abnormal data therein; the spatial location information and attribute information of the monitoring data are converted into intermediate data in a visual format; in three-dimensional space, the intermediate data is interpolated and a three-dimensional visualization surface is formed to realize the visualization of the monitoring data; through monitoring and early warning judgment, the attribute abnormality level and early warning information are represented by color in the three-dimensional visualization surface. Preferably, in the three-dimensional visualization surface, the attribute abnormality level and early warning information can be represented by color coding.

[0139] It should be noted that the above-mentioned specific embodiments are exemplary, and those skilled in the art can come up with various solutions inspired by the disclosure of the present invention, and these solutions also belong to the disclosure scope of the present invention and fall within the protection scope of the present invention. Those skilled in the art should understand that the present invention specification and its drawings are illustrative and do not constitute a limitation on the claims. The scope of protection of the present invention is defined by the claims and their equivalents. The present invention specification contains multiple inventive concepts, such as "preferably" and "according to a preferred embodiment", which means that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to file a divisional application based on each inventive concept.

Claims

1. A fusion visualization method for multi-source engineering monitoring data, characterized in that: The method comprises: Pre-process the collected monitoring data for verification, cleaning and repair; Construct a data matrix based on preprocessed multi-source heterogeneous monitoring data; Calculate the comprehensive attribute value sequence based on the influence weights of attributes in different dimensions; The comprehensive attribute value sequence is transformed into visual data coordinates based on the determined data ratio.

2. The method according to claim 1, characterized in that The pre-processing steps include: Verifying the location information and / or time information in the monitoring data and deleting duplicate data; Judging and identifying outliers in the monitoring data based on the improved standard score, and deleting the outliers; The monitoring data is time-repaired and space-repaired based on the calculated time-repaired data and space-repaired data, thereby obtaining repaired multi-source heterogeneous monitoring data.

3. The method according to claim 1 or 2, characterized in that: The steps of constructing the data matrix include: Obtain the displacement monitoring data and stress monitoring data of the monitoring points; The displacement monitoring data and stress monitoring data are arranged in time sequence to construct a data matrix of a single monitoring point.

4. The method according to any one of claims 1 to 3, characterized in that: The step of calculating the comprehensive attribute value sequence comprises: Compute the covariance matrix of the standardized data; Performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors corresponding thereto; sorting the eigenvalues ​​and selecting the eigenvector corresponding to the largest eigenvalue; The comprehensive attribute value sequence is calculated based on the selected feature vector and the original monitoring data.

5. The method according to any one of claims 1 to 4, characterized in that: The step of determining the data ratio includes: Calculate the length and width of the plane rectangle of the project area; Calculate the distance between monitoring points based on the spatial average point spacing of the project area; The data ratio is determined based on the length and width of the plane rectangle of the engineering area and the distance between the monitoring points.

6. The method according to any one of claims 1 to 5, characterized in that: The step of transforming the comprehensive attribute value sequence based on the determined data ratio comprises: Visualized data point coordinates = comprehensive attribute value sequence × data ratio.

7. A fusion visualization device for multi-source engineering monitoring data, characterized in that: comprising a processor (100), The processor (100) comprises: A preprocessing module (110) performs preprocessing of verification, cleaning and repair on the collected monitoring data; A matrix construction module (120) constructs a data matrix based on the pre-processed multi-source heterogeneous monitoring data; A calculation module (130) calculates a comprehensive attribute value sequence based on the influence weights of attributes of different dimensions, and transforms the comprehensive attribute value sequence into visual data coordinates based on a determined data ratio.

8. The device according to claim 7, characterized in that The pre-processing module (110) is configured to: Verifying the location information and / or time information in the monitoring data and deleting duplicate data; Judging and identifying outliers in the monitoring data based on the improved standard score, and deleting the outliers; The monitoring data is time-repaired and space-repaired based on the calculated time-repaired data and space-repaired data, thereby obtaining repaired multi-source heterogeneous monitoring data.

9. The device according to claim 7 or 8, characterized in that The matrix building module (120) is configured to: Obtain the displacement monitoring data and stress monitoring data of the monitoring points; The displacement monitoring data and stress monitoring data are arranged in time sequence to construct a data matrix of a single monitoring point.

10. The device according to any one of claims 7 to 9, characterized in that: The calculation module (130) is configured to: Compute the covariance matrix of the standardized data; Performing eigendecomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors corresponding thereto; sorting the eigenvalues ​​and selecting the eigenvector corresponding to the largest eigenvalue; Calculating the comprehensive attribute value sequence based on the selected feature vector and the original monitoring data; The coordinates of the visualization data points = comprehensive attribute value sequence × data ratio.

Images