A method and apparatus for generating an engineering monitoring report

Abstract

This disclosure provides a method and apparatus for generating engineering monitoring reports, relating to the field of data processing technology. The method includes: determining structured monitoring data containing time-series observation data based on the sensitivity weights of each monitoring point; rendering the time-series observation data according to the marker line density and visual difference feature points to obtain various types of charts and images; filling different groups of time-series observation data corresponding to each monitoring point into a preset result table and determining derived columns; determining the adaptation height corresponding to the process line graph based on the row and column structure of the result table, and configuring the process line graph in the result table according to the adaptation height to obtain integrated text and graphics result table data; and writing the corresponding text, various types of charts and images containing feature value graphs, and result table data into the corresponding template placeholders according to the priority order of each component in the preset report template to obtain the target engineering monitoring report. This disclosure can improve the quality of generated engineering monitoring reports.

Description

Technical Field

[0001] This disclosure relates to the field of data technology, and in particular to a method and apparatus for generating engineering monitoring reports. Background Technology

[0002] With the continuous development of engineering safety monitoring technology, monitoring systems have been collecting and accumulating massive amounts of multi-source heterogeneous monitoring data. In order to intuitively reflect the actual operating status of the project, it is usually necessary to process the above monitoring data and transform it into a monitoring report document that includes assessment and analysis text and visualization charts.

[0003] To overcome the inefficiency of frequent manual screenshotting and mechanical form filling, existing technologies attempt to introduce an automatic formatting engine in the report generation stage. This engine directly binds the underlying structured time-series database with the upper-level office document components, attempting to automate form filling and batch image insertion through specific row and column mapping relationships in a fixed template.

[0004] However, this fixed mapping approach fails to effectively clean up abnormal signals in the data stream when facing real-world engineering environments with multiple measurement points and varying amplitudes, directly leading to inaccurate data sources for reports. Simultaneously, when the automatic layout engine writes charts into the document, the lack of adaptive adjustment capabilities in the generated graphic resources makes text overlap and blurring highly likely during secondary scaling within the page. Furthermore, when large-scale projects undergo complex evolutions such as the deployment of new measurement points, discontinuation of old measurement points, or changes in monitoring frequency, the fixed table column structure easily introduces data alignment errors and page breaks when filling in data. During the report generation stage, issues such as content overlap, space encroachment, and misalignment between components can easily occur, ultimately affecting the quality of the automated report generation. Summary of the Invention

[0005] To overcome the problems existing in related technologies, this disclosure provides a method and apparatus for generating engineering monitoring reports, which can realize adaptive generation of engineering monitoring reports from structured data to documents throughout the entire process, ensuring the quality of report generation.

[0006] According to a first aspect of the present disclosure, a method for generating an engineering monitoring report is provided, the method comprising: A random binary search tree is constructed based on the sensitivity weight of each monitoring point. The monitoring data obtained is then processed according to the weighted path length and preset confidence interval of the monitoring data of each monitoring point in the random binary search tree to obtain structured monitoring data containing time-series observation data. The sensitivity weight is used to characterize the sensitivity of the spatial location of each monitoring point to the response of the engineering structure damage evolution process. The marker line density and visual difference feature points corresponding to the time series observation data are determined, and the time series observation data are rendered according to the marker line density and the visual difference feature points to obtain multiple types of chart images, including process line charts and feature value charts. The monitoring points are grouped according to their time-series monitoring status. The time-series observation data of different groups are filled into a preset result table and the derived columns are determined. The adaptation height of the process line graph is determined according to the row and column structure of the result table. The process line graph is configured in the result table according to the adaptation height to obtain the result table data with integrated graphics and text. Obtain the priority order of each component in the preset report template, and write the corresponding text, the multi-type chart images containing the feature value map, and the result table data into the corresponding template placeholders according to the priority order to obtain the target project monitoring report.

[0007] In an exemplary embodiment of this disclosure, the step of constructing a random binary search tree based on the sensitivity weights of each monitoring point, and processing the monitoring data obtained from the monitoring based on the weighted path length and preset confidence interval corresponding to the monitoring data of each monitoring point in the random binary search tree, to obtain structured monitoring data containing time-series observation data, includes: The sensitivity weights of each monitoring point are normalized to obtain the cutting bias factor of the corresponding feature space, and the random binary search tree is constructed based on the cutting bias factor. The monitoring data of each monitoring point is traversed layer by layer along the random binary search tree. The weighted path length corresponding to each monitoring data is determined based on the depth traversed from the root node to the leaf node and the cutting bias factor corresponding to each node. Determine the dynamic pruning threshold and preset confidence interval for each monitoring data point; Monitoring data whose weighted path length corresponding to each monitoring point is less than the dynamic pruning threshold and exceeds the corresponding preset information interval are identified as abnormal data and deleted to obtain processed monitoring data. The time-series observation data is determined based on the processed monitoring data, and the acquired inspection data and the time-series observation data are combined to obtain the structured monitoring data.

[0008] In one exemplary embodiment of this disclosure, after obtaining structured monitoring data containing time-series observation data, the method further includes: The curvature coefficient of time-degradation is determined based on the engineering location and physical quantity type corresponding to each monitoring point. Based on the time-degradation curvature coefficient and the access frequency of each structured monitoring data, a comprehensive popularity value for each structured monitoring data is determined, and the structured monitoring data is stored in the corresponding storage level according to the comprehensive popularity value. The storage levels include a hot data storage level, a warm data storage level, and a cold data storage level. The comprehensive popularity value of the structured monitoring data stored in the hot data storage level is greater than that of the structured monitoring data stored in the warm data storage level, and the comprehensive popularity value of the structured monitoring data stored in the warm data storage level is greater than that of the structured monitoring data stored in the cold data storage level.

[0009] In an exemplary embodiment of this disclosure, determining the marker line density and visual difference feature points corresponding to the time-series observation data includes: The time-series observation data is divided into multiple time windows according to time sequence; The time series observation data of each monitoring point within each time window are processed by first-order difference to obtain the corresponding difference sequence. The median of the absolute value of each difference sequence is determined as the fluctuation intensity index of each monitoring point within the corresponding time window. The corresponding comprehensive volatility index is determined based on the volatility intensity index corresponding to each time window, and each comprehensive volatility index is multiplied by the preset basic marker line density value to obtain the marker line density corresponding to each time window. For each time window, if the comprehensive volatility index of that time window is greater than a preset multiple of the average of the comprehensive volatility indices of all time windows, then the marker line density of that time window is adjusted to a preset upper limit value; if the comprehensive volatility index of that time window is less than the average of the comprehensive volatility indices of all time windows, then the marker line density of that time window is adjusted to a preset lower limit value. Differential scanning analysis is performed on the time-series observation data of each monitoring point to determine the visual difference feature points of each monitoring point. The visual difference feature points include at least one or a combination of trend turning points, early warning approach points, and rate change points.

[0010] In one exemplary embodiment of this disclosure, the rendering process of the time-series observation data based on the marker line density and the visual difference feature points to obtain multiple types of chart images includes: Obtain the page size and layout constraints of the preset charts in the preset report template; Based on the ratio between the layout size of the preset chart and the preset chart design resolution, the corresponding rendering configuration parameters are determined, and the coordinate axis tick labels, data series lines, marker symbols and legend items in the process line chart and the feature value chart are adjusted according to the rendering configuration parameters to generate the corresponding chart instance. The time-series observation data is divided into multiple time windows in chronological order. Based on the marker line density corresponding to each time window, the vertical dashed line position corresponding to each time window in the preset coordinate axis area of ​​the chart instance is determined, and a vertical dashed line marker line is generated in each time window based on the vertical dashed line position. Based on the visual difference feature points corresponding to each monitoring point, the feature marker position of each monitoring point is determined, and corresponding abnormal element markers are configured in the process line graph and the feature value graph in the chart instance according to the feature marker position; Asynchronous rendering is performed on each of the chart instances containing the vertical dashed marker line and the abnormal element marker to obtain the multi-type chart images.

[0011] In one exemplary embodiment of this disclosure, the grouping includes a stable monitoring state group and a dynamic monitoring state group. The step of filling the time-series observation data of different groups into a preset result table and determining derived columns includes: For the time-series observation data of the stable monitoring state group, the corresponding time-series observation data are sequentially filled into the results table according to the preset arrangement order of monitoring points and the mapping relationship between cells; For the time-series observation data of the dynamic monitoring status group, based on the valid monitoring points in the monitoring status during the valid duration, an empty template structure containing only a header format is generated and the monitoring point number column is associated and filled. The time-series observation data of the dynamic monitoring status group is then filled into the empty template structure in time sequence to obtain the result table. Based on the monitoring data value range characteristics in the filled-in results table, the corresponding derivative column type is determined, and a derivative column is added to the corresponding table according to the derivative column type. The corresponding derivative column data is written into the cell corresponding to the derivative column to obtain the results table data.

[0012] In one exemplary embodiment of this disclosure, the step of determining the adaptation height corresponding to the process line diagram based on the row and column structure of the result table, and configuring the process line diagram in the result table according to the adaptation height to obtain integrated graphic and text result table data includes: Get the column width distribution, row height distribution, and merged cell range of the current results table; The corresponding adaptation height is calculated based on the row span range of the monitoring points associated with each process line diagram in the results table; The total height of the corresponding target row region is determined based on the row height distribution. If the sum of the adaptation heights of all the process line diagrams to be written into the result table is less than or equal to the total height of the target row area, then each process line diagram is configured in the row span range corresponding to the result table according to the adaptation height. If the sum of the adaptation heights of all the process line graphs to be written into the results table is greater than the total height of the target row area, then the corresponding process line graph is embedded in the corresponding cell in the form of a thumbnail, and a link mapping relationship between the thumbnail and the original high-resolution graphic is generated to obtain the integrated graphic results table data.

[0013] In one exemplary embodiment of this disclosure, the structured monitoring data further includes inspection data, and the method further includes: Obtain the inspection anomaly description text from the inspection data, and use a preset inspection thesaurus to perform word segmentation and keyword matching on the inspection anomaly description text to determine the defect type label and severity level corresponding to each inspection anomaly description text. Based on the inspection form template in the preset report template, determine the row and column writing areas for each inspection item; Based on the defect type labels, the inspection anomaly description texts are grouped into similar categories to obtain corresponding defect merging groups; The inspection anomaly description texts within the defect merging group are sorted in descending order according to the severity level to obtain the target display order. The corresponding inspection data are then sequentially filled into the corresponding row and column writing areas according to the target display order to obtain a result table containing the inspection data.

[0014] According to a second aspect of the present disclosure, an apparatus for generating an engineering monitoring report is provided, comprising: The processing module is used to construct a random binary search tree based on the sensitivity weight of each monitoring point, and to process the monitoring data obtained by monitoring based on the weighted path length and preset confidence interval corresponding to the monitoring data of each monitoring point in the random binary search tree, so as to obtain structured monitoring data containing time series observation data. The chart module is used to determine the marker line density and visual difference feature points corresponding to the time series observation data, and to render the time series observation data according to the marker line density and the visual difference feature points to obtain multiple types of chart images. The multiple types of chart images include process line graphs and feature value graphs. The sensitivity weight is used to characterize the sensitivity of the spatial location of each monitoring point to the response of the engineering structure damage evolution process. The table module is used to group the time-series monitoring data of each monitoring point according to the time-series monitoring status, fill the time-series observation data of different groups into a preset result table and determine the derived columns; determine the adaptation height corresponding to the process line graph according to the row and column structure of the result table, and configure the process line graph in the result table according to the adaptation height to obtain the result table data with integrated graphics and text. The generation module is used to obtain the priority order of each component in the preset report template, and write the corresponding text, the multi-type chart images containing the feature value map and the result table data into the corresponding template placeholders according to the priority order, so as to obtain the target project monitoring report.

[0015] According to a third aspect of the present disclosure, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0016] This disclosure provides a method and apparatus for generating engineering monitoring reports. By constructing a random binary search tree based on the sensitivity weights of each monitoring point, and processing the monitoring data using a weighted path length and a pre-set confidence interval, structured monitoring data containing time-series observation data is obtained. Compared to traditional data processing methods, this method can assign differentiated anomaly detection biases to monitoring points in different spatial locations, significantly reducing the interference of anomalous samples on the monitoring report and ensuring the accuracy of multi-source heterogeneous monitoring data at the source.

[0017] Furthermore, by determining the marker line density and visual difference feature points corresponding to the time-series observation data, and rendering the time-series observation data based on the marker line density and visual difference feature points, various types of chart images, including process line plots and feature value plots, are obtained. This method can directly control the rendering and generation of graphic resources based on the characteristics of the data itself, and the headless browser can directly render and output high-resolution static images in memory that match the pixel specifications required by the target report placeholders, avoiding problems such as text overlap and blurring that are very likely to occur when chart images are scaled up twice within the page.

[0018] Furthermore, by grouping data according to the time-series monitoring status of each monitoring point, the time-series observation data of different groups are filled into a pre-defined results table, and derived columns are determined. Simultaneously, the adaptation height corresponding to the process line graph is determined based on the row and column structure of the results table, and the process line graph is configured in the results table according to the adaptation height, resulting in integrated text and graphics results table data. Therefore, compared to the defects of fixed table column structures, which are prone to data alignment errors and page breaks during data filling, this method allows table rows and columns to be reconstructed and filled based on the time-series monitoring status of the monitoring points. Through the geometric adaptation of the table row and column structure with the height of the process line graph, problems such as page breaks and overlapping table writing areas caused by graph insertion are eliminated.

[0019] Finally, by obtaining the priority order of each component in the preset report template, and writing the corresponding text, various charts and images including feature value graphs, and result table data into the corresponding template placeholders according to the priority order, the target project monitoring report is obtained. Therefore, compared to the shortcomings of traditional layout methods where components are prone to conflict, this method uses the priority order between components to apply writing control constraints, solving the technical problems of content overlap, space squeezing, and order misalignment that easily occur when multiple resources are finally written into the template placeholders. This ensures the automated production of high-quality monitoring documents while improving the quality of the automated report output.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this disclosure, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0022] Figure 1 This is a flowchart illustrating a method for generating an engineering monitoring report according to an exemplary embodiment of the present disclosure.

[0023] Figure 2 This is a schematic diagram of the structure of an apparatus for generating an engineering monitoring report according to an exemplary embodiment of the present disclosure.

[0024] Figure 3 This disclosure is a hardware structure diagram of a computer device according to an exemplary embodiment. Detailed Implementation

[0025] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0026] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms first, second, third, etc., may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are used only to distinguish information of the same type from one another. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0028] The embodiments of this disclosure will now be described in detail.

[0029] like Figure 1 As shown, Figure 1 This disclosure is a flowchart illustrating a method for generating an engineering monitoring report according to an exemplary embodiment, comprising the following steps: Step 101: Construct a random binary search tree based on the sensitivity weight of each monitoring point, and process the monitoring data obtained from the monitoring based on the weighted path length and preset confidence interval of the monitoring data of each monitoring point in the random binary search tree to obtain structured monitoring data containing time series observation data. Step 102: Determine the marker line density and visual difference feature points corresponding to the time series observation data, and render the time series observation data according to the marker line density and visual difference feature points to obtain multiple types of chart images, including process line graphs and feature value graphs. Step 103: Group the data according to the time-series monitoring status of each monitoring point, fill the time-series observation data of different groups into the preset result table and determine the derived columns; determine the adaptation height corresponding to the process line diagram according to the row and column structure of the result table, and configure the process line diagram in the result table according to the adaptation height to obtain the integrated graphic and text result table data; Step 104: Obtain the priority order of each component in the preset report template, and write the corresponding text, multiple types of charts and images containing feature value graphs, and result table data into the corresponding template placeholders according to the priority order to obtain the target project monitoring report.

[0030] The method for generating engineering monitoring reports provided in the exemplary embodiments of this disclosure constructs a random binary search tree based on the sensitivity weights of each monitoring point, and processes the monitoring data obtained by combining the weighted path length and a preset confidence interval to obtain structured monitoring data containing time-series observation data. Compared with traditional data processing methods that rely on fixed threshold cleaning, this method can give differentiated processing to measuring points in different spatial locations, and combine the processing depth of the underlying time-series data with the sensitivity weights of each monitoring point, thereby improving the data processing accuracy of the input time-series observation data and reducing the negative interference of abnormal samples on subsequent analysis and monitoring report generation.

[0031] Furthermore, by determining the marker line density and visual difference feature points corresponding to the time-series observation data, and rendering the time-series observation data based on the marker line density and visual difference feature points, various types of chart images are obtained. Thus, compared with the traditional graphics rendering method that does not consider the marker line density and visual difference feature points, this method realizes the corresponding control and processing of the graphic resources on the marker line density and visual difference feature points, so that the generated various types of chart images, including process line graphs and feature value graphs, carry visual difference feature points and marker line density, improving the problem of loss of chart visual features due to the lack of relevant rendering features.

[0032] Furthermore, by grouping data according to the time-series monitoring status of each monitoring point, the time-series observation data of different groups are filled into a preset result table and the derived columns are determined. At the same time, the adaptation height of the process line graph is determined according to the row and column structure of the result table, and the process line graph is configured in the result table according to the adaptation height, resulting in integrated text and graphics result table data. Thus, compared with the traditional table filling method that uses a fixed table column structure for hard data filling and lacks height adaptation when inserting graphs, this method can group and fill data based on the time-series monitoring status of each monitoring point and automatically determine the derived columns. Moreover, it can directly determine and dynamically adapt the height of the configured process line graph through the row and column structure of the result table, eliminating table row and column alignment conflicts or spatial adaptation problems caused by simple graph insertion, and ensuring the integrity of the integrated text and graphics of the result table data.

[0033] Finally, by obtaining the priority order of each component in the preset report template, and writing the corresponding text, various charts and images containing feature value graphs, and result table data into the corresponding template placeholders according to the priority order, the target project monitoring report is obtained. Thus, the entire data processing based on the priority order of components in the preset report template can be achieved without relying on manual typesetting. This solves the layout conflicts such as overlapping content, space occupation, and misalignment of components that are very easy to occur when writing multiple components in traditional automated typesetting. While ensuring the automated output of high-quality monitoring reports, it also improves the generation quality of engineering monitoring reports.

[0034] The following will provide a detailed description of a method for generating an engineering monitoring report in this example embodiment.

[0035] In step 101, a random binary search tree is constructed based on the sensitivity weight of each monitoring point. The monitoring data obtained from the monitoring are then processed according to the weighted path length and preset confidence interval of the monitoring data of each monitoring point in the random binary search tree to obtain structured monitoring data containing time-series observation data.

[0036] In the exemplary implementation of this disclosure, multi-source heterogeneous monitoring data can be continuously collected from various monitoring points in an engineering project. This data includes not only instrumental observation values ​​such as displacement, seepage pressure, temperature, and stress-strain resulting from large-scale engineering projects (such as hydropower dams) spanning the construction and operation periods, but also unstructured inspection data text reported daily by inspection technicians via mobile terminals and entered into the system. Based on the spatial location of the monitoring points, the spatial distance to the preset key bearing sections can be determined, along with quantitative parameters such as the pre-calibrated structural importance coefficient in the engineering design model. This allows for the determination of the sensitivity of each monitoring point to the structural damage evolution process, assigning different sensitivity weights. Furthermore, based on this, anomaly cleaning and removal are performed on the collected monitoring data using a constructed random binary search tree. This reduces the need for manual subjective screening, filtering and removing interfering data from heterogeneous monitoring data, ensuring the accuracy and objectivity of subsequent report preparation.

[0037] In some example implementations, when processing the monitoring data obtained from the monitoring, the sensitivity weight of each monitoring point can be determined first based on the positional relationship between each monitoring point and the preset key bearing section.

[0038] For example, this sensitivity weight is used to characterize the sensitivity of each monitoring point to the damage evolution process of the engineering structure. Specifically, the sensitivity weight can be determined jointly based on the spatial distance from each monitoring point to the preset key bearing section and the pre-calibrated structural importance coefficient in the engineering design model. For instance, in specific calculations, the sensitivity weight can be set as a mathematical function that is inversely proportional to the spatial distance and directly proportional to the structural importance coefficient. The specific values ​​of the aforementioned structural importance coefficient can be determined based on the criticality of each measuring point in the overall reliability or load-bearing instability evolution of the project. Specifically, for measuring points located at the geometric center of the dam top and the transverse joints of the arch crown, which directly bear and characterize the overall stability and key progressive instability features of the dam (such as seepage pressure monitoring points on the main axis of the dam foundation), a high sensitivity weight can be assigned because these spatial locations are within the preset key bearing sections or their core stress concentration areas, the corresponding spatial distances are small, and the structural importance coefficients are high. Failure of these points would trigger a huge engineering disaster. On the other hand, for measuring items that are far from the main bearing sections, such as auxiliary environmental temperature measuring points and conventional rainfall measuring points, which have a smaller impact on the overall progressive instability of the structure, a lower sensitivity weight is assigned.

[0039] After determining the sensitivity weight of each monitoring point, the cutting bias factor of the corresponding feature space can be determined based on the sensitivity weight of each monitoring point, and a random binary search tree can be constructed based on the cutting bias factor.

[0040] For example, the sensitivity weights of each monitoring point can be normalized, and the normalized value can be determined as the cutting bias factor of the corresponding feature space. A random binary search tree can then be constructed based on the cutting bias factor. Here, the feature space refers to a multi-dimensional data representation space composed of various monitoring points in the project. Each feature dimension in the feature space corresponds to a monitoring point in the project, and the data points in the feature space are used to represent the set of monitoring data collected by each monitoring point at the same sampling time node.

[0041] When splitting specific tree nodes, instead of using a uniform probability distribution to select feature axes, a cutting bias factor is used as a selection probability bias. This adjusts the selection probability when monitoring data corresponding to each monitoring point is extracted as the basis for splitting, increasing the probability that monitoring data collected from monitoring points with high sensitivity weights will be selected as cutting features. This biased splitting mechanism allows early, weak data distributed among high-sensitivity monitoring points that have not yet exceeded the conventional statistical threshold but have already shown trend drift characteristics to be split earlier or more frequently in the feature space when constructing a random binary search tree. This shortens the path length of their leaf nodes in the topological structure and improves the sensitivity of identifying weak and abnormal data.

[0042] Next, during the process of traversing each monitoring data point from the root node to the leaf node layer by layer, the sequence of path nodes traversed by the monitoring data to reach the corresponding leaf node is recorded, and the cutting bias factor corresponding to each node in the path node sequence is extracted. Based on this, the cutting bias factor of each node along the way is used as a weight adjustment coefficient to perform a weighted calculation on the depth of downward traversal at each level (for example, multiplying the depth corresponding to each level by the cutting bias factor corresponding to the node at that level), and the calculation results of each node in the sequence are accumulated, thereby obtaining the weighted path length corresponding to the monitoring data through specific numerical iterative calculation steps. Further, monitoring data whose weighted path length corresponding to each monitoring point is less than the dynamic pruning threshold and whose value exceeds the corresponding preset confidence interval are identified as abnormal data and deleted to obtain the processed monitoring data. In some example embodiments, when constructing the random binary search tree, a tree set composed of multiple random binary search trees can be generated according to a preset number of constructions; the preset number of constructions is a constant pre-set based on the computing resource processing capacity and algorithm requirements. Accordingly, when determining the weighted path length of a certain monitoring data, it specifically means obtaining the path length of the monitoring data in each random binary search tree in the tree set, calculating the arithmetic mean of all path lengths, and using the obtained arithmetic mean as the final weighted path length of the monitoring data for comparison and judgment with the dynamically calculated dynamic pruning threshold.

[0043] For example, a dynamic pruning threshold is first dynamically calculated based on the input monitoring data. This can be achieved by calculating the expected average path length of a random binary search tree at that sample size using the total number of samples for each monitoring data point. A preset function model containing the natural logarithm and a constant term is used to estimate the total number of samples to correct for deviations in the binary tree construction process, thus obtaining the theoretical expected average path length. A preset pruning ratio coefficient is then obtained, and the expected average path length is multiplied by this coefficient to determine the dynamic pruning threshold suitable for the current sample size. This pruning ratio coefficient is a parameter preset based on engineering experience, typically ranging from 0 to 1. Simultaneously, for each input monitoring data point, its corresponding sample expected value and sample standard deviation are statistically analyzed. Chebyshev's inequality is used for joint discrimination. Using the calculated sample expected value as the reference center, the sample standard deviation is multiplied by a preset probability deviation factor, and the resulting product is used as the fluctuation range of the monitoring data. Next, based on the baseline center, the fluctuation amplitude is added to and subtracted from each other to construct a dynamic upper limit (i.e., Chebyshev upper bound) and a lower limit (preset confidence interval). The preset probability deviation factor refers to the standard deviation product factor controlling the confidence interval width in the Chebyshev inequality; its value is greater than 1. This factor is set based on a preset normal data tolerance probability. The larger the factor, the wider the constructed confidence interval, and the higher the tolerance for normal fluctuations in engineering monitoring data. The preset normal data tolerance probability represents the lower limit of the proportion of monitoring data belonging to normal physical fluctuations expected to be included in the engineering safety monitoring scenario. When the weighted path length of a monitoring data point is less than the dynamic pruning threshold, and the monitoring data simultaneously exceeds the corresponding preset confidence interval, the monitoring data is judged as abnormal data and deleted. The remaining data are retained as normal observation data, thus finally outputting the processed monitoring data. Specifically, the monitoring data exceeding the corresponding preset confidence interval includes the monitoring data being greater than the upper limit of the corresponding preset confidence interval, or the monitoring data being less than the lower limit of the corresponding preset confidence interval. This method effectively overcomes the shortcomings of the traditional fixed standard deviation threshold method in responding slowly to early progressive structural anomalies, and reduces the probability of missing detection of small-amplitude but highly hazardous samples in highly sensitive monitoring sequences.

[0044] Finally, time-series observation data is determined based on the processed monitoring data, and the acquired inspection data and time-series observation data are combined to obtain structured monitoring data. For example, the processed monitoring data retained after the above weighted isolated forest and confidence interval filtering is determined as a valid sequence, and time-series alignment and normalization reconstruction are performed in memory according to the timestamp and monitoring point identifier, thereby determining it as time-series observation data that can be called by the system later.

[0045] Meanwhile, the inspection data reported that day was retrieved via a mobile field terminal. This data included irregular natural language text describing anomalies in various parts of the dam by the inspection personnel. The multi-source heterogeneous monitoring data integration module, by calling the data assembly interface, combined and merged the acquired inspection data and the cleaned and aligned time-series observation data into multi-modal data, ultimately outputting and packaging it into structured monitoring data containing complete time-series physical quantities and text logs.

[0046] In the exemplary implementation of this disclosure, after obtaining the structured monitoring data, the time-degradation curvature coefficient can be determined based on the engineering part and physical quantity type corresponding to each monitoring point.

[0047] For example, based on the structural safety criticality level of each monitoring point's engineering component and the failure sensitivity of the monitored physical quantities, differentiated time-degradation curvature coefficients are pre-configured for different time-series observation and inspection data. For instance, for data located in critical engineering components characterizing progressive structural instability, such as deformation within the dam body and the main axis of the dam foundation, and for key physical quantities like horizontal displacement, vertical displacement, and uplift pressure of the dam, a lower time-degradation curvature coefficient is configured to support long-term trend-based safety analysis and retrospective assessment, thus slowing down the rate of thermal decay. Conversely, for environmental measurements such as ambient temperature and rainfall, which only play a supporting role in the overall structural reliability analysis, a higher time-degradation curvature coefficient is configured, allowing for both computational needs and freeing up limited storage resources.

[0048] Secondly, based on the curvature coefficient of time-degradation and the access frequency of each structured monitoring data, the comprehensive heat value of each structured monitoring data is determined, and the structured monitoring data is stored in the corresponding storage level according to the comprehensive heat value.

[0049] For example, after the data is persistently written to the system, the external access frequency of each or each batch of structured monitoring data is continuously counted. By constructing a comprehensive popularity evaluation function, the weighted value of the external access frequency of each data tuple is combined with the time decay weight value calculated based on the time decay curvature coefficient to dynamically determine the current comprehensive popularity value of each structured monitoring data.

[0050] After obtaining the overall heat value, the migration of structured monitoring data between different storage physical media layers is dynamically determined and triggered based on the currently calculated value. Specifically, the storage layers include hot data storage layer, warm data storage layer, and cold data storage layer. Among them, the overall heat value of structured monitoring data stored in the hot data storage layer (such as in-memory databases or high-performance solid-state drive arrays) is greater than that of structured monitoring data stored in the warm data storage layer (such as conventional hard drive arrays); the overall heat value of structured monitoring data stored in the warm data storage layer is greater than that of structured monitoring data stored in the cold data storage layer (such as large-capacity mechanical hard drives or cloud archive storage services).

[0051] For example, a first switching threshold and a second switching threshold can be set (with the first switching threshold being greater than the second switching threshold). When a highly sensitive monitoring data (such as displacement data of the core part of a dam) has an extremely low time-degradation curvature coefficient, and its comprehensive heat value remains above the first switching threshold for a long time, it is locked into the high-performance hot data storage layer to ensure immediate read and write availability. As time goes by or the access frequency decreases, when its comprehensive heat value falls below the first switching threshold but is still higher than the second switching threshold, the storage migration mechanism is automatically triggered to transfer it to the warm data storage layer. For auxiliary environmental monitoring data, due to its extremely high time-degradation curvature coefficient, its comprehensive heat value will quickly fall below the second switching threshold after the statistics are completed. Therefore, it can be directly deposited as cold data in the cold data storage layer, thereby smoothly releasing high-value storage resources.

[0052] This step determines the curvature coefficient of time-degradation by identifying the corresponding engineering parts and physical quantity types of each monitoring point, and then accurately quantifies the comprehensive heat value by combining the access frequency. This enables dynamic hierarchical storage of multi-source heterogeneous engineering safety monitoring data assets and achieves a dynamic balance between storage resource consumption and structural safety decision-making efficiency under high-load data environments.

[0053] In step 102, the marker line density and visual difference feature points corresponding to the time series observation data are determined, and the time series observation data are rendered based on the marker line density and visual difference feature points to obtain multiple types of chart images, including process line graphs and feature value graphs.

[0054] In this exemplary embodiment, all sampled time-series observation data within the target monitoring period can be divided into several time windows according to their chronological order. For the time-series observation data corresponding to each monitoring point within each time window, a fluctuation intensity index is calculated. The fluctuation intensity index is the median of the absolute values ​​of the data sequence within the current window after first-order differencing. The target monitoring period refers to the time span of the observation data intercepted or defined when generating the engineering monitoring report.

[0055] Subsequently, the fluctuation intensity index calculated for each window is normalized and multiplied by the preset base marker line density value to obtain the marker line density corresponding to that time window.

[0056] For each time window, after obtaining the marker line density corresponding to that time window, if the comprehensive volatility index of that time window is greater than a preset multiple of the average comprehensive volatility index of all time windows, then that period is determined to belong to an abnormal attention segment of structural response. Subsequently, the marker line density within that time window is increased to a preset upper density value (e.g., generating one marker line per day) to finely analyze and display the gradual evolution of data during abnormal periods. Conversely, for stable segments where the comprehensive volatility index is less than the average comprehensive volatility index of all time windows, the marker line density is reduced to a preset lower density value (e.g., generating one marker line every 5 days) to minimize the interference of redundant visual elements on long-term trend interpretation. Finally, based on the calculated marker line density of each time window, the specific date sequence of marker line distribution is determined. This preset multiple can be determined according to actual circumstances, and this disclosure does not impose any special limitations.

[0057] Secondly, differential scanning analysis is performed on the time-series observation data to be rendered to determine the visual difference feature points of each monitoring point. These visual difference feature points may include at least one or a combination of several of the following: trend inflection points, early warning approaching points, and rate abrupt change points.

[0058] For example, trend inflection points are used to represent sampling points in the first-order difference sequence of data where the absolute value exceeds a preset inflection amplitude threshold and the signs of the preceding and following points reverse. These points indicate a sudden change or reversal in the direction of structural deformation. Warning approach points are used to represent critical sampling points in the data sequence where the physical value has exceeded a preset warning threshold by a certain proportion (such as reaching a preset safety alarm line but not yet formally triggering a platform alarm). These points indicate a critical precursor to a shift from a safe to a dangerous state in the structure. Rate abrupt change points are used to represent sampling points in the second-order difference sequence of data where the absolute value exceeds a preset rate increment threshold. These points indicate an instantaneous intensification or anomaly in the rate of structural deformation. The preset inflection amplitude threshold and the preset rate increment threshold can be determined according to actual conditions, and this disclosure does not impose any special limitations.

[0059] The time-series observation data are rendered based on the marked line density and visual difference feature points to obtain various types of charts and images, including process line graphs and feature value graphs.

[0060] First, the layout size and constraints of the preset charts in the preset report template can be obtained. For example, before the chart is rendered, a layout parser is invoked to scan the currently selected preset report template, obtaining the actual layout size declared by the library resource component in the template's target area (e.g., a width of 595px corresponds to an A4 portrait layout, and a width of 842px corresponds to an A4 landscape layout). Simultaneously, layout constraint parameters such as its relative position to surrounding text paragraphs, adjacent table components, and margin limits are extracted. This layout size and constraints will be used as core boundary conditions and passed to the underlying graphics rendering engine.

[0061] Secondly, based on the ratio between the preset chart size and the preset chart design resolution, the corresponding rendering configuration parameters are determined, and the coordinate axis tick labels, data series lines, marker symbols, and legend items in the process line chart and feature value chart are adjusted according to the rendering configuration parameters to generate the corresponding chart instance.

[0062] For example, after receiving the actual layout size, the rendering engine adjusts key configuration items in the chart instance, such as the font of the axis tick labels, the thickness of the data series lines, the radius of the marker symbols, and the spacing of the legend items, based on the ratio between the target layout size and the chart design resolution. This ensures that the chart is rendered with visual specifications that perfectly match the report layout, avoiding text overlap, blurring, and line distortion that are easily caused by secondary scaling after the chart is written to the document. This ratio includes horizontal and vertical scaling factors. The horizontal and vertical scaling factors can be calculated by obtaining the actual display width and height of the preset chart in the preset report template as the target layout size, and comparing the actual display width and height with the baseline pixel width and height at the preset chart design resolution. Based on these rendering configuration parameters, fine-tuning is performed on the prepared process line graphs and feature value graphs, including distribution maps, correlation maps, and contour maps. This includes adjusting the font size and rotation angle of the axis tick labels to prevent label overlap due to insufficient space; dynamically scaling the pixel thickness of data series lines to ensure the smoothness and continuity of line curves; synchronously reshaping the pixel radius of various feature marker symbols; and adaptively adjusting the spacing and font size of legend items. This ensures that the final rendered charts exist in a physical format that perfectly matches the report document layout, avoiding text overlap, blurring, jagged lines, or non-linear deformation problems that can easily occur when the graphics are directly written into the document due to mechanical secondary scaling.

[0063] Next, the time series observation data is divided into multiple time windows according to the time sequence. Based on the marker line density corresponding to each time window, the vertical dashed line position corresponding to each time window in the preset coordinate axis area of ​​the chart instance is determined, and vertical dashed line markers are generated in each time window according to the vertical dashed line position.

[0064] For example, regarding the adjustment of markers along the time axis, all time-series observation data within the monitoring period are divided into several time windows of equal initial length according to the chronological order of sampling time. Based on the geometric coordinate span occupied by the current time window on the X-axis (time axis) of the chart instance, the corresponding marker line density is converted into specific pixel geometric coordinates, thereby accurately determining the vertical dashed line position corresponding to each time window within the preset coordinate axis area of ​​the chart instance. Finally, based on the calculated vertical dashed line position, a vertical dashed marker line perpendicular to the X-axis is automatically drawn and generated at each corresponding date coordinate. For time windows in the structural response anomaly focus area, visual features different from normal areas (such as darker line color or thicker line width) are added to their corresponding marker lines to help report readers quickly focus on key periods.

[0065] Then, based on the visual difference feature points corresponding to each monitoring point, the feature marker positions of each monitoring point are determined, and corresponding abnormal element markers are configured in the process line graph and feature value graph in the chart instance according to the feature marker positions.

[0066] For example, the visual difference feature points identified in the data sequence scans for each monitoring point in the aforementioned steps are retrieved (including trend reversal points indicating a reversal of deformation direction, warning approach points indicating impending critical safety, and rate abrupt change points indicating a sudden increase in deformation rate). By establishing a mapping equation between the physical values ​​of the data and the two-dimensional coordinate system of the chart, the physical sampled values ​​of these feature points are converted into pixel coordinates within the chart grid area, thereby accurately determining the feature marker positions of each monitoring point. Subsequently, based on the feature marker positions, corresponding abnormal element markers are dynamically configured in the process line graph and feature value graph of the chart instance. For example, a circular marker with a directional arrow is configured for the feature marker position of the trend reversal point; a warning color marker with a fill color different from the normal sequence and a specific physical value label is attached for the feature marker position of the warning approach point; and a semi-transparent highlighted bar perpendicular to the X-axis is configured for the feature marker position of the rate abrupt change point to identify the abnormal time period.

[0067] Finally, each chart instance containing vertical dashed marker lines and abnormal element markers is rendered asynchronously to obtain various types of chart images.

[0068] For example, after the chart instance completes the layout adaptation attribute adjustment, vertical dashed line embedding, and abnormal element mark injection in memory, its complete rendering configuration parameters are solidified. Then, a headless browser in the background is launched to execute differentiated asynchronous rendering instructions. Specifically, the headless browser performs parallel asynchronous graphics rendering on each chart instance in the background, simulating the constraints of the target report template page. The final output is a high-resolution static image (such as PNG or JPEG), which is persistently stored in an object storage service, thus obtaining multiple types of chart images. This method achieves a complete layout closed loop of layout awareness, feature extraction, and mark solidification, enabling the generated static images to carry key abnormal information for structural safety monitoring.

[0069] In step 103, the time-series monitoring data of each monitoring point are grouped according to their time-series monitoring status. The time-series observation data of different groups are filled into the preset result table and the derived columns are determined. The adaptation height of the process line graph is determined according to the row and column structure of the result table. The process line graph is configured in the result table according to the adaptation height to obtain the result table data with integrated graphics and text.

[0070] In the example implementation of this disclosure, the time-series monitoring status of each monitoring point can be grouped, and the time-series observation data of different groups can be filled into a preset result table and the derived columns can be determined.

[0071] For example, the grouping may specifically include a stable monitoring status group and a dynamic monitoring status group.

[0072] For the time-series observation data of the stable monitoring status group, identify the effective time-series observation data that is continuously generated during the target monitoring period, and fill the corresponding time-series observation data into the preset result table in sequence according to the preset arrangement order of monitoring points and the physical mapping relationship of table cells.

[0073] For the time-series observation data of the dynamic monitoring status group, a set of non-stationary monitoring points was identified within the target monitoring period, including newly established monitoring points, discontinued old monitoring points, or monitoring frequencies that had changed. Valid monitoring points that were actually in monitoring status and generating data during the effective duration were then selected. Next, an empty template structure containing only a header format was constructed, and the selected valid monitoring points were associated and filled into the corresponding monitoring point number column. Subsequently, the time-series observation data of the dynamic monitoring status group was filled into the empty template structure horizontally or vertically according to the timestamp order, thus combining them to obtain the result table. During this process, missing or blank rows due to frequency changes in dynamic monitoring point segments were hidden to ensure the compactness of the table structure. The effective duration period characterizes the time interval within which monitoring points can normally collect and generate valid monitoring data during the target monitoring period.

[0074] After the data is populated, the monitoring data range features in the populated results table are extracted and semantically parsed. These range features include the magnitude range of the physical quantities, their positive and negative distribution characteristics, and whether they include instrument over-range flags. Based on the monitoring data range features, the corresponding derived column types are matched and determined. Derivative columns are then automatically added to the corresponding tables according to their types. Finally, in-situ aggregation calculations accurately write the corresponding derived column data into the corresponding cells of the derived columns to obtain the results table data.

[0075] For example, if the current displacement physical quantity is detected to have a value span covering millimeters to centimeters and including both positive and negative directions within the target monitoring period, the corresponding derived column type is determined to be a direction identifier, and a derived column for the direction identifier is then added to the table; if the analysis detects that the value of the seepage pressure physical quantity has obvious periodic hydrological fluctuation characteristics and the target monitoring period includes a complete hydrological cycle, a derived column for the same period comparison difference and the year-on-year change rate is added; if the instrument calibration coefficient has been changed, a calibration status identifier column is added and its derived calibration label is backfilled in the corresponding measuring point row.

[0076] Secondly, determine the appropriate height for the process line diagram based on the row and column structure of the results table.

[0077] For example, when preparing to configure the corresponding process line graph in the observation data result table, the complete row and column topology of the result table to be written is first globally analyzed to determine the column width distribution, row height distribution, and technical parameters of the merged cell area of ​​the current result table.

[0078] Next, the physical row span range actually occupied by the specific monitoring point associated with the process line graph in the results table is identified and obtained in real time (i.e., the start and end row numbers of the data row displayed by the measurement point on the time axis). By accumulating the pixel values ​​of all row heights within the row span range in the row height distribution, an optimal graphic height that coincides with the physical height of the current measurement point's data row is calculated. This optimal graphic height is equal to the sum of the heights of all rows within the aforementioned data row span range, ensuring that after the process line graph is inserted into the cell, it is vertically aligned with the associated data row without overflowing the cell boundary or leaving gaps. This optimal graphic height is then confirmed as the adaptive height of the process line graph. This dynamic calculation of the adaptive height ensures that the subsequent graphic presentation ratio is geometrically bound to the table rows, avoiding the destruction of layout readability due to graphics being too tall or too short.

[0079] Finally, the process line diagram is configured in the results table according to the adaptation height to obtain the integrated graphic and text results table data.

[0080] For example, after obtaining the adaptation height of the process line chart, when multiple process line charts need to be embedded simultaneously in the same result table according to business analysis requirements, the vertical space conflict of multiple graphics within the current result table cell grid area is detected. If it is detected that the physical space corresponding to the current measurement point row area is sufficient to fully accommodate all the graphic resources to be inserted, the process line chart is directly configured in the specified cell of the corresponding row span range of the result table according to the adaptation height, in the format of an embedded object. If it is detected that multiple graphics coexist and the current row area is insufficient to accommodate all the graphic resources to be inserted, in order to prevent the large result table from causing page breaks due to graphics crowding, the image scaling operator is dynamically called to convert the original high-resolution process line chart into a low-position thumbnail form proportionally and embed it in the corresponding cell, and add a clickable enlargement visual label in the cell; at the same time, a link mapping relationship is established and generated between the thumbnail in the cell and the original high-resolution graphics persisted in the object storage service. Through this execution method, while ensuring the overall compactness of the table layout, the accessibility of high-resolution graphics is not lost, and the result table data with high visual regularity and integrated text and graphics is finally output.

[0081] In the example implementation of this disclosure, the structured monitoring data also includes inspection data.

[0082] After obtaining the inspection data, the inspection anomaly description text in the inspection data can be obtained. The inspection anomaly description text can be segmented and matched with keywords using a preset inspection thesaurus to determine the defect type label and severity level corresponding to each inspection anomaly description text.

[0083] For example, the preset inspection terminology is a professional terminology in the field of engineering safety monitoring. It contains pre-set and encapsulated keywords for high-frequency defect types (such as cracks, seepage, settlement, displacement, corrosion, detachment, blockage, etc.) and corresponding severity modifiers (such as minor, obvious, severe, penetrating, etc.).

[0084] The input inspection anomaly description text is processed using a domain-specific thesaurus, and keyword matching is performed on the segmentation results. By extracting core defect words and degree modifiers from the text, the defect type label (such as categorized as crack, seepage, or blockage) and physical severity level (such as quantitatively divided into minor, obvious, and severe levels) are automatically determined for each inspection anomaly description text according to mapping rules.

[0085] Secondly, based on the inspection form template in the preset report template, determine the row and column writing areas for each inspection item.

[0086] For example, before assembling the results table, the inspection table template referenced in the preset report template is obtained, and the grid topology of the template is parsed. The core parameters configured for the project are read, including the inspection date column, the column where the inspection item is located, and the starting row of the data. Based on this, the row and column writing areas to which each specific inspection item, such as the dam surface, flood discharge facilities, drainage system, and monitoring equipment, belongs in the spreadsheet are determined through coordinate offset calculation, providing accurate physical spatial grid boundaries for subsequent data distribution and writing.

[0087] Next, the inspection anomaly description texts are grouped into similar categories based on the defect type labels to obtain the corresponding defect groupings.

[0088] For example, based on the defect type label, all similar items in the inspection anomaly description text within the current monitoring period are merged and summarized, and inspection records with the same defect type label are divided into a unified set, thereby aggregating the corresponding defect merging groups and achieving preliminary structured classification of unstructured text.

[0089] Finally, the inspection anomaly description texts within the defect merging group are sorted in descending order according to each severity level to obtain the target display order. The corresponding inspection data are then sequentially filled into the corresponding row and column writing areas according to the target display order to obtain the result table containing the inspection data.

[0090] For example, for each defect grouped together, the severity level corresponding to each inspection anomaly description text is obtained and mapped to a comparable numerical risk weight (e.g., severe / pervasive is mapped to the highest risk, and minor is mapped to the lowest risk). Based on the relationship between severity levels, the inspection anomaly description texts are sorted in descending order from high risk to low risk within each group, thereby calculating the target display order that best meets the timeliness requirements of engineering safety management.

[0091] Subsequently, in accordance with the order of target display, the corresponding time-series inspection data (including inspection date, abnormal defect classification, severity and specific description text, etc.) are sequentially filled back into the rows and columns of the corresponding inspection items in the aforementioned area cells.

[0092] By performing the aforementioned defect semantic classification and high-risk priority ranking, a result table containing inspection data with strong logical guidance is ultimately generated. This method utilizes the dual semantic drive of defect type and severity to adaptively reconstruct the table, ensuring that the most critical and serious structural safety hazard information in the final automated report is prioritized and presented within the core visual area of ​​the report reader. This solves the engineering challenge of traditional technologies where critical defect hazard information is easily overwhelmed by massive amounts of redundant, stable-period inspection text.

[0093] In step 104, the priority order of each component in the preset report template is obtained, and the corresponding text, multiple types of charts and images containing feature value maps, and result table data are written into the corresponding template placeholders according to the priority order to obtain the target project monitoring report.

[0094] In the example implementation of this disclosure, the priority order of each component in the preset report template can be obtained first.

[0095] For example, upon responding to a report generation instruction, a preset report template is first retrieved, and a resource tree topology is reconstructed, including various report components such as text content, library resources (e.g., process line graphs, feature maps including distribution maps, correlation maps, and contour maps), inspection checklists, and result table data. Based on this, priority parameters pre-maintained and bound to each report component are read (e.g., pre-setting process line graph components to take precedence over distribution map components, core result table data components to take precedence over regular text components, etc.), thereby establishing the global component priority order for this report assembly.

[0096] Secondly, according to priority, the corresponding text, various charts and images containing feature value graphs, and result table data are written into the corresponding template placeholders to obtain the target project monitoring report.

[0097] For example, in order to eliminate space contention and content overlap conflicts caused by writing visual resources, the writing of each component and the adaptive page layout calculation can be triggered sequentially from high to low according to the component priority order obtained above.

[0098] Specifically, when writing text content and basic report information, the system scans text paragraphs in the preset report template and directly replaces the preset lightweight text placeholders (such as ${report date} representing the compilation date, ${compiler} representing the personnel name, etc.) with the currently extracted actual business text through string matching, thereby realizing the replacement and dynamic updating of basic conclusions and key texts.

[0099] When writing multiple types of chart images containing feature value maps, for each independent static image resource, the original pixel dimensions of the image are first read (e.g., width 1200px, height 800px), and the paper page width declared by the current preset report template (e.g., the effective width of an A4 portrait layout) is obtained. Based on the principle that the image width completely and proportionally fills the page width or adapts to the page width according to a preset ratio, the corresponding pre-selected writing height is calculated horizontally using a formula.

[0100] Next, the pre-selected write height is compared with the total available height of the report template page (e.g., the height of an A4 portrait layout). If the calculated pre-selected write height is less than or equal to the total available height of the page, the image is written into the corresponding placeholder using the page width and the pre-selected height. Conversely, if the calculated pre-selected write height exceeds the maximum available height of a single page, to prevent layout distortion and out-of-bounds formatting, the pre-selected write height of the image is limited to the current total available height of the page. Then, the most suitable write width is calculated based on the original scale of the corresponding chart image, scaled, and written into the corresponding image placeholder. In addition, for layouts with multiple images per row, automatic segmentation and physical line breaks are performed on the images in the gallery according to the preset constraints on the number of images displayed per row, ensuring that charts do not overlap or intrude into adjacent text or table display areas.

[0101] When writing integrated text and graphics result tables, the large number of columns and embedded process line charts in these tables can easily lead to data alignment errors and page breaks. Therefore, during the formatted writing of table data to the result table placeholders, the physical properties of the current page are captured and analyzed in real time. If the column width of the current result table exceeds the maximum physical boundary of the vertical page, the paper orientation of the current page is reset to landscape (e.g., the layout space of an A4 landscape page). This prioritizes the row and column layout and visual regularity of large-volume tables with many columns, ensuring that all data columns are fully displayed within the physical boundaries of the current page, avoiding data overlap, truncation, or unexpected page breaks due to excessive column width compression. After the result table data is fully written and the data stream crosses the page boundary, the paper orientation of subsequent pages is adaptively restored to portrait, allowing subsequent components to continue writing.

[0102] By leveraging a priority mechanism and applying temporal and spatial control constraints during multi-component collaborative layout, high-priority components are ensured to be prioritized for filling unoccupied space on the current page, with sufficient physical dimensions to accommodate their entire text and graphic content without requiring page-wide splitting. Low-priority components are written within the remaining available space using either compression or page-adaptive fine-tuning modes. After all component resources are written to their corresponding placeholders and the layout is integrated, a neatly formatted and highly standardized target project monitoring report can be output in a closed loop without relying on manual secondary proofreading or layout fine-tuning, thus improving the overall execution efficiency of the monitoring system's automated report generation.

[0103] Corresponding to the embodiments of the foregoing methods, this disclosure also provides embodiments of the apparatus and the terminal to which it is applied.

[0104] like Figure 2 As shown, Figure 2This is a schematic diagram of an engineering monitoring report generation apparatus according to an exemplary embodiment of the present disclosure. The apparatus includes: a processing module 210, a chart module 220, a table module 230, and a generation module 240. The processing module 210 is used to construct a random binary search tree according to the sensitivity weight of each monitoring point, and process the monitoring data obtained by monitoring according to the weighted path length and preset information interval corresponding to the monitoring data of each monitoring point in the random binary search tree, so as to obtain structured monitoring data containing time series observation data. The chart module 220 is used to determine the marker line density and visual difference feature points corresponding to the time series observation data, and to render the time series observation data according to the marker line density and visual difference feature points to obtain multiple types of chart images, including process line charts and feature value charts. Table module 230 is used to group the time-series monitoring data of each monitoring point according to the time-series monitoring status, fill the time-series observation data of different groups into the preset result table and determine the derived columns; determine the adaptation height of the process line diagram according to the row and column structure of the result table, and configure the process line diagram in the result table according to the adaptation height to obtain the result table data with integrated graphics and text. The generation module 240 is used to obtain the priority order of each component in the preset report template, and write the corresponding text, multiple types of charts and images containing feature value maps and result table data into the corresponding template placeholders according to the priority order to obtain the target project monitoring report.

[0105] It should be noted that the engineering monitoring report generation device in this embodiment is used to implement the corresponding engineering monitoring report generation method in the foregoing method embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0106] The embodiments of the engineering monitoring report generation device disclosed herein can be applied to computer equipment, such as servers or terminal devices. The device embodiments can be implemented through software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor reading the corresponding computer program instructions from non-volatile memory into memory and executing them. From a hardware perspective, such as... Figure 3 The diagram shown is a hardware structure diagram of a computer device containing the engineering monitoring report generation apparatus according to an embodiment of this disclosure, except... Figure 3 In addition to the processor 310, memory 330, network interface 320, and non-volatile memory 340 shown, the server or electronic device where the engineering monitoring report generation device is located in the embodiment may also include other hardware depending on the actual function of the computer device, which will not be described in detail here.

[0107] Accordingly, this disclosure also provides an apparatus for generating an engineering monitoring report, the apparatus including a processor; a memory for storing processor-executable instructions; wherein the processor is configured for the above-described method for generating an engineering monitoring report.

[0108] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0109] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0110] The foregoing has described specific embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous. Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention applied herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not claimed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0111] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

[0112] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method for generating an engineering monitoring report, characterized in that, include: The sensitivity weights of each monitoring point are normalized to obtain the cutting bias factor of the corresponding feature space. A random binary search tree is constructed based on the cutting bias factor. The monitoring data obtained by monitoring are processed according to the weighted path length and preset confidence interval of the monitoring data of each monitoring point in the random binary search tree to obtain structured monitoring data containing time series observation data. The sensitivity weights are used to characterize the sensitivity of the spatial location of each monitoring point to the response of the engineering structure damage evolution process. The marker line density and visual difference feature points corresponding to the time series observation data are determined, and the time series observation data are rendered according to the marker line density and the visual difference feature points to obtain multiple types of chart images, including process line charts and feature value charts. The monitoring points are grouped according to their time-series monitoring status. The time-series observation data of different groups are filled into a preset result table and the derived columns are determined. The adaptation height of the process line graph is determined according to the row and column structure of the result table. The process line graph is configured in the result table according to the adaptation height to obtain the result table data with integrated graphics and text. Obtain the priority order of each component in the preset report template, and write the corresponding text, the multi-type chart images containing the feature value map, and the result table data into the corresponding template placeholders according to the priority order to obtain the target project monitoring report; The process of processing the monitoring data obtained from each monitoring point in the random binary search tree based on the weighted path length and preset confidence interval to obtain structured monitoring data containing time-series observation data includes: The monitoring data of each monitoring point is traversed layer by layer along the random binary search tree. Based on the depth traversed from the root node to the leaf node and the cutting bias factor corresponding to each node, the weighted path length corresponding to each monitoring data is determined. The dynamic pruning threshold and preset confidence interval corresponding to each monitoring data are determined respectively. The monitoring data whose weighted path length corresponding to each monitoring point is less than the dynamic pruning threshold and exceeds the corresponding preset confidence interval are identified as abnormal data and deleted to obtain the processed monitoring data. Among them, the total number of samples of each monitoring data is estimated using a preset function model containing natural logarithm and constant term to obtain the theoretical expected value of the average path length. The theoretical expected value is multiplied by the preset pruning ratio coefficient to obtain the dynamic pruning threshold. The time-series observation data is rendered based on the marked line density and the visual difference feature points to obtain various types of chart images, including: The process involves: obtaining the layout size and constraints of preset charts in the preset report template; determining the corresponding rendering configuration parameters based on the ratio between the layout size and the preset chart design resolution; adjusting the coordinate axis tick labels, data series lines, marker symbols, and legend items in the process line graph and the feature value graph according to the rendering configuration parameters to generate corresponding chart instances; dividing the time-series observation data into multiple time windows according to time sequence; determining the vertical dashed line position of each time window within the preset coordinate axis area of ​​the chart instance based on the marker line density corresponding to each time window; generating vertical dashed line markers within each time window based on the vertical dashed line position; determining the feature marker position of each monitoring point based on the visual difference feature points corresponding to each monitoring point; configuring corresponding abnormal element markers in the process line graph and the feature value graph of the chart instance based on the feature marker positions; and asynchronously rendering each chart instance containing the vertical dashed line markers and the abnormal element markers to obtain the multiple types of chart images.

2. The method according to claim 1, characterized in that, After obtaining the structured monitoring data containing time-series observation data, the method further includes: The curvature coefficient of time-degradation is determined based on the engineering location and physical quantity type corresponding to each monitoring point. Based on the time-degradation curvature coefficient and the access frequency of each structured monitoring data, a comprehensive popularity value for each structured monitoring data is determined, and the structured monitoring data is stored in the corresponding storage level according to the comprehensive popularity value. The storage levels include a hot data storage level, a warm data storage level, and a cold data storage level. The comprehensive popularity value of the structured monitoring data stored in the hot data storage level is greater than that of the structured monitoring data stored in the warm data storage level, and the comprehensive popularity value of the structured monitoring data stored in the warm data storage level is greater than that of the structured monitoring data stored in the cold data storage level.

3. The method according to claim 1, characterized in that, The step of determining the marked line density and visual difference feature points corresponding to the time-series observation data includes: The time-series observation data is divided into multiple time windows according to time sequence; The time series observation data of each monitoring point within each time window are processed by first-order difference to obtain the corresponding difference sequence. The median of the absolute value of each difference sequence is determined as the fluctuation intensity index of each monitoring point within the corresponding time window. The corresponding comprehensive volatility index is determined based on the volatility intensity index corresponding to each time window, and each comprehensive volatility index is multiplied by the preset basic marker line density value to obtain the marker line density corresponding to each time window. For each time window, if the comprehensive volatility index of that time window is greater than a preset multiple of the average of the comprehensive volatility indices of all time windows, then the marker line density of that time window is adjusted to a preset upper limit value; if the comprehensive volatility index of that time window is less than the average of the comprehensive volatility indices of all time windows, then the marker line density of that time window is adjusted to a preset lower limit value. Differential scanning analysis is performed on the time-series observation data of each monitoring point to determine the visual difference feature points of each monitoring point. The visual difference feature points include at least one or a combination of trend turning points, early warning approach points, and rate change points.

4. The method according to claim 1, characterized in that, The grouping includes a stable monitoring status group and a dynamic monitoring status group. The step of filling the time-series observation data of different groups into a pre-set results table and determining derived columns includes: For the time-series observation data of the stable monitoring state group, the corresponding time-series observation data are sequentially filled into the results table according to the preset arrangement order of monitoring points and the mapping relationship between cells; For the time-series observation data of the dynamic monitoring status group, based on the valid monitoring points in the monitoring status during the valid duration, an empty template structure containing only a header format is generated and the monitoring point number column is associated and filled. The time-series observation data of the dynamic monitoring status group is then filled into the empty template structure in time sequence to obtain the result table. Based on the monitoring data value range characteristics in the filled-in results table, the corresponding derivative column type is determined, and a derivative column is added to the corresponding table according to the derivative column type. The corresponding derivative column data is written into the cell corresponding to the derivative column to obtain the results table data.

5. The method according to claim 1, characterized in that, The step of determining the adaptation height corresponding to the process line diagram based on the row and column structure of the result table, and configuring the process line diagram in the result table according to the adaptation height to obtain integrated graphic and text result table data includes: Get the column width distribution, row height distribution, and merged cell range of the current results table; The corresponding adaptation height is calculated based on the row span range of the monitoring points associated with each process line diagram in the results table; The total height of the corresponding target row region is determined based on the row height distribution. If the sum of the adaptation heights of all the process line diagrams to be written into the result table is less than or equal to the total height of the target row area, then each process line diagram is configured in the row span range corresponding to the result table according to the adaptation height. If the sum of the adaptation heights of all the process line graphs to be written into the results table is greater than the total height of the target row area, then the corresponding process line graph is embedded in the corresponding cell in the form of a thumbnail, and a link mapping relationship between the thumbnail and the original high-resolution graphic is generated to obtain the integrated graphic results table data.

6. The method according to claim 1, characterized in that, The structured monitoring data also includes inspection data, and the method further includes: Obtain the inspection anomaly description text from the inspection data, and use a preset inspection thesaurus to perform word segmentation and keyword matching on the inspection anomaly description text to determine the defect type label and severity level corresponding to each inspection anomaly description text. Based on the inspection form template in the preset report template, determine the row and column writing areas for each inspection item; Based on the defect type labels, the inspection anomaly description texts are grouped into similar categories to obtain corresponding defect merging groups; The inspection anomaly description texts within the defect merging group are sorted in descending order according to the severity level to obtain the target display order. The corresponding inspection data are then sequentially filled into the corresponding row and column writing areas according to the target display order to obtain a result table containing the inspection data.

7. A device for generating engineering monitoring reports, characterized in that, include: The processing module is used to normalize the sensitivity weights of each monitoring point to obtain the cutting bias factor of the corresponding feature space, and construct a random binary search tree based on the cutting bias factor. The monitoring data obtained by monitoring is processed according to the weighted path length and preset confidence interval of the monitoring data of each monitoring point in the random binary search tree to obtain structured monitoring data containing time series observation data. The sensitivity weights are used to characterize the sensitivity of the spatial location of each monitoring point to the response of the engineering structure damage evolution process. The chart module is used to determine the marker line density and visual difference feature points corresponding to the time series observation data, and to render the time series observation data according to the marker line density and the visual difference feature points to obtain multiple types of chart images, including process line charts and feature value charts. The table module is used to group the time-series monitoring data of each monitoring point according to the time-series monitoring status, fill the time-series observation data of different groups into a preset result table and determine the derived columns; determine the adaptation height corresponding to the process line graph according to the row and column structure of the result table, and configure the process line graph in the result table according to the adaptation height to obtain the result table data with integrated graphics and text. The generation module is used to obtain the priority order of each component in the preset report template, and write the corresponding text, the multi-type chart images containing the feature value map and the result table data into the corresponding template placeholders according to the priority order to obtain the target project monitoring report. The processing module is specifically configured to: traverse the monitoring data of each monitoring point layer by layer along the random binary search tree; determine the weighted path length corresponding to each monitoring data based on the depth traversed from the root node to the leaf node and the cutting bias factor corresponding to each node; determine the dynamic pruning threshold and preset confidence interval corresponding to each monitoring data; identify monitoring data whose weighted path length corresponding to each monitoring point is less than the dynamic pruning threshold and exceeds the corresponding preset confidence interval as abnormal data and delete the abnormal data to obtain processed monitoring data; wherein, the total number of samples of each monitoring data is estimated using a preset function model containing natural logarithm and constant term to obtain the theoretical expected value of the average path length, and the theoretical expected value is multiplied by a preset pruning ratio coefficient to obtain the dynamic pruning threshold; The chart module is specifically used for: obtaining the layout size and constraints of the preset chart in the preset report template; determining the corresponding rendering configuration parameters according to the ratio between the layout size and the preset chart design resolution, and adjusting the coordinate axis tick labels, data series lines, marker symbols, and legend items in the process line graph and the feature value graph according to the rendering configuration parameters to generate corresponding chart instances; dividing the time series observation data into multiple time windows according to time order, determining the vertical dashed line position of each time window in the preset coordinate axis area of ​​the chart instance according to the marker line density corresponding to each time window, and generating vertical dashed line markers in each time window according to the vertical dashed line position; determining the feature marker position of each monitoring point according to the visual difference feature points corresponding to each monitoring point, and configuring corresponding abnormal element markers in the process line graph and the feature value graph of the chart instance according to the feature marker positions; asynchronously rendering each chart instance containing the vertical dashed line markers and the abnormal element markers to obtain the multiple types of chart images.

8. A computer device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1-6.

Images