Concrete bubble real-time monitoring method and system applied to water conservancy projects

Abstract

The application provides a concrete bubble real-time monitoring method and system applied to water conservancy projects, through synchronous acquisition of ultrasonic-laser co-path perception signals, slurry shear rate data and viscosity data of water conservancy concrete slurry, a multi-source slurry monitoring data set containing time sequence synchronization marks is obtained, the ultrasonic-laser co-path perception signals in the multi-source slurry monitoring data set are subjected to interference component stripping in combination with the slurry shear rate data and viscosity data, pure bubble perception signals after interference stripping are obtained, signal analysis is carried out on the pure bubble perception signals, bubble particle size distribution, instantaneous position and stress state information are obtained, in combination with the dynamic change of the viscosity data, bubble migration trajectories, coalescence critical conditions and escape time consumption information are derived, and corresponding relationship between the three and the impermeability and frost resistance of concrete is associated and characterized, so that bubble evolution state monitoring results are obtained. The application effectively improves the support value of bubble monitoring on the construction of water conservancy concrete.

Description

Technical Field

[0001] This invention relates to the field of data processing, and more specifically, to a method and system for real-time monitoring of air bubbles in concrete applied to water conservancy projects. Background Technology

[0002] With increasing demands for the durability of concrete structures in water conservancy projects, real-time monitoring of concrete air bubbles has become crucial for ensuring construction quality. Capturing and analyzing the state of air bubbles within the grout provides a basis for assessing the concrete's impermeability and freeze-thaw resistance. Current technologies typically collect bubble sensing signals and grout rheological parameters separately, using conventional signal processing methods to remove interference. After analyzing the bubble state, post-construction specimen testing indirectly correlates the bubble state with concrete performance. Tracking changes in bubble state are largely limited to independent recording. However, existing technologies suffer from several drawbacks. The lack of coordination between signal and rheological parameter acquisition, the inability of interference processing to match dynamic changes in the grout, and the resulting bias in bubble state analysis, coupled with the reliance on post-construction testing for correlation between bubbles and performance leading to insufficient timeliness, and the failure to consider the environmental impact of the grout during bubble behavior tracking, result in discrepancies between the assessment results and actual conditions, making it difficult to support precise control of construction quality. Summary of the Invention

[0003] This invention provides a method and system for real-time monitoring of air bubbles in concrete applied to water conservancy projects.

[0004] In a first aspect, embodiments of the present invention provide a method for real-time monitoring of air bubbles in concrete applied to hydraulic engineering. The method includes: synchronously acquiring ultrasonic-laser copath sensing signals, slurry shear rate data, and viscosity data of hydraulic concrete slurry to obtain a multi-source slurry monitoring dataset containing time-series synchronization markers, wherein the ultrasonic-laser copath sensing signals, slurry shear rate data, and viscosity data in the multi-source slurry monitoring dataset correspond to the same acquisition time node; and interfering with the ultrasonic-laser copath sensing signals in the multi-source slurry monitoring dataset by combining the slurry shear rate data and viscosity data in the multi-source slurry monitoring dataset. The components were stripped to obtain the pure bubble sensing signal after interference removal. The pure bubble sensing signal after interference removal was analyzed to obtain information on bubble size distribution, instantaneous position and stress state. Combined with the dynamic changes of viscosity data in the multi-source slurry monitoring dataset, the evolution law of bubble size distribution, instantaneous position and stress state information was derived to obtain information on bubble migration trajectory, coalescence critical condition and escape time. The correlation between bubble migration trajectory, coalescence critical condition and escape time information and concrete impermeability and frost resistance was characterized to obtain the bubble evolution state monitoring results related to concrete performance.

[0005] Secondly, embodiments of the present invention provide a computer system, including: a memory storing a computer program; and a processor for loading the computer program to implement the above-described method for real-time monitoring of concrete bubbles applied to water conservancy projects.

[0006] This invention obtains a time-synchronized multi-source slurry monitoring dataset by simultaneously acquiring ultrasonic-laser co-path sensing signals, slurry shear rate data, and viscosity data of hydraulic concrete slurry. Based on this dataset, interference components are removed from the ultrasonic-laser co-path sensing signals by combining the slurry shear rate and viscosity data. This overcomes the limitations of conventional methods that rely solely on the signal's intrinsic content to identify interference. By leveraging changes in the slurry's rheological state, it accurately locates and removes interference signals unrelated to bubbles, effectively improving the purity of the bubble sensing signal. Analysis of the interference-free pure bubble sensing signal yields information on bubble size distribution, instantaneous position, and stress state, covering the core state of bubbles in the slurry, resulting in more comprehensive and accurate bubble state capture. By combining the dynamic changes in viscosity data from the multi-source slurry monitoring dataset, the evolutionary laws of bubble size distribution, instantaneous position, and stress state information are derived, directly linking bubble state changes to the dynamic influence of the slurry environment. This accurately tracks the bubble migration process, coalescence trigger nodes, and escape process duration, making the derivation of bubble evolution laws more closely reflect the real environment of hydraulic engineering concrete slurry. Finally, the correlation between bubble migration trajectory, coalescence critical conditions, and escape time information and the concrete impermeability and frost resistance properties was characterized, breaking the disconnect between bubble monitoring and concrete performance analysis. This directly establishes a correspondence between the core information of bubble evolution and the core performance requirements of the project, enabling the monitoring results to directly serve the judgment of concrete performance and effectively enhancing the supporting value of bubble monitoring for concrete construction in water conservancy projects. Attached Figure Description

[0007] Figure 1 This is a flowchart of a method for real-time monitoring of air bubbles in concrete applied to water conservancy projects, provided by an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of the composition of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0009] Please see Figure 1 This is a flowchart of a real-time monitoring method for concrete air bubbles in water conservancy projects, provided by an embodiment of the present invention. Applied to a computer system, it includes the following steps:

[0010] Step S100: Synchronously collect ultrasonic-laser co-path sensing signals, slurry shear rate data, and viscosity data of hydraulic concrete slurry to obtain a multi-source slurry monitoring dataset containing time-series synchronization markers. The ultrasonic-laser co-path sensing signals, slurry shear rate data, and viscosity data in the multi-source slurry monitoring dataset correspond to the same acquisition time node.

[0011] The ultrasonic-laser co-path sensing signal is obtained by sensing hydraulic concrete slurry through ultrasound and laser along the same path. Ultrasound has penetrating power; as it propagates through the slurry, it produces reflections and scattering due to differences in the slurry's internal structure and density. The signals corresponding to these phenomena reflect the internal structural information of the slurry. Laser, on the other hand, can sense particles within the slurry; the characteristics of its scattered light reflect the particle size and distribution. Slurry shear rate data describes the relative velocity between the internal layers of the slurry when subjected to shear force; different shear rates result in different flow states. Viscosity data is a physical quantity used to measure the viscosity of the slurry, reflecting its resistance to flow; highly viscous slurries are relatively difficult to flow under external forces. Time synchronization markers are used to identify the data acquisition time, ensuring a one-to-one correspondence between the ultrasonic-laser co-path sensing signal, slurry shear rate data, and viscosity data in the multi-source slurry monitoring dataset, thereby guaranteeing the correlation and accuracy of the data in subsequent analyses.

[0012] In practical data acquisition, an ultrasonic-laser co-path sensing sensor can be used to acquire ultrasonic-laser co-path sensing signals. This sensor consists of an ultrasonic transmitting and receiving module and a laser transmitting and receiving module, simultaneously emitting ultrasonic waves and lasers, and recording the signals after reflection or scattering by the slurry. For acquiring slurry shear rate data, a rotary rheometer can be used. The rotary rheometer uses a rotating rotor within the slurry to measure the torque and rotational speed experienced by the rotor. Based on rheological principles, the slurry shear rate can be calculated using known rotor geometric parameters and measured torque, rotational speed, and other data. The rheometer measures the torque experienced by the rotor in real time and calculates the current slurry shear rate according to a preset algorithm, while simultaneously recording the acquisition time.

[0013] A rotational rheometer can also be used to collect viscosity data. The rheometer calculates the viscosity of the slurry based on the resistance experienced by the rotor as it rotates in the slurry, combined with information such as the rotor's geometric parameters and rotational speed. Specifically, the rotational rheometer continuously collects data and records the calculated viscosity values ​​in relation to the collection time.

[0014] Step S200: Combining the slurry shear rate data and viscosity data in the multi-source slurry monitoring dataset, the interference components of the ultrasonic-laser co-path sensing signal in the multi-source slurry monitoring dataset are removed to obtain the pure bubble sensing signal after interference removal.

[0015] In one implementation, step S200 may specifically include the following steps S210-S260:

[0016] Step S210: Read the time-domain sampling content of the ultrasonic-laser co-path sensing signal in the multi-source slurry monitoring dataset, bind the continuous acquisition content of the slurry shear rate data in the multi-source slurry monitoring dataset with the time-domain sampling content point by point according to the acquisition time node, generate shear rate-sensing signal time-domain binding data, so that each sensing signal sampling point corresponds to a unique shear rate acquisition content.

[0017] The time-domain sampling content of the ultrasonic-laser copath sensing signal consists of a series of data points obtained by discretely sampling the ultrasonic-laser copath sensing signal in the time domain. These data points record information such as the signal amplitude at different times. The continuous acquisition content of the slurry shear rate data consists of the slurry shear rate values ​​continuously acquired over a period of time. Point-by-point binding according to the acquisition time node associates each sampling point of the ultrasonic-laser copath sensing signal with the slurry shear rate value acquired at the same time node.

[0018] Specifically, the temporal sampling content of the ultrasonic-laser copath sensing signal can be read from the multi-source slurry monitoring dataset, and this content can be stored in the form of a digital sequence. Then, the continuously acquired content of the slurry shear rate data is read. Since time synchronization markers have been added to these data in step S100, the two can be bound point by point according to the timestamp. For example, assuming that the sampling point of the ultrasonic-laser copath sensing signal acquired in the first second is S1, and the corresponding slurry shear rate acquired in the first second is V1, then S1 and V1 are bound together.

[0019] Step S220: Read the continuous acquisition content of viscosity data in the multi-source slurry monitoring dataset, bind the shear rate-sensing signal time-domain binding data with the continuous acquisition content of the viscosity data point by point according to the acquisition time node, generate triple time-domain associated sensing data, so that each data node is simultaneously associated with the sensing signal, shear rate and viscosity acquisition content.

[0020] Specifically, the continuous collection of viscosity data is first read from the multi-source slurry monitoring dataset. Since time-series synchronization markers have already been added to all data, the shear rate-sensing signal time-domain bound data and viscosity data can be bound point-by-point according to the timestamp. For example, for the case at the 2nd second, the shear rate-sensing signal time-domain bound data has already been associated with the sensing signal sampling point and shear rate value at that moment, and now the viscosity value collected at that moment is bound to it.

[0021] Step S230: Perform time-domain differential processing on the continuously acquired slurry shear rate data and viscosity data in the triple time-domain correlated sensing data, mark the time-domain intervals where the differential results of the acquired content conform to the preset content, and generate a set of time-domain intervals for state change.

[0022] In one implementation, step S230 may specifically include the following steps S231-S236:

[0023] Step S231: Read the continuous acquisition content of slurry shear rate data in the triple time-domain correlated sensing data, divide it into multiple groups of continuous subsequences of different durations according to the preset time-domain resolution division strategy, and obtain a multi-resolution shear rate subsequence set, each subsequence containing a fixed number of continuous acquisition contents.

[0024] Specifically, assume the slurry shear rate data in the triple time-domain correlated sensing data is a time series of length N. According to a preset time-domain resolution partitioning strategy, the time series is divided into different durations (e.g., groups of 10 data points, groups of 20 data points, etc.). For each duration, starting from the beginning of the time series, a fixed number of data points are sequentially extracted to form subsequences. For example, when partitioning into groups of 10 data points, the first subsequence contains data points 1-10, the second subsequence contains data points 11-20, and so on. In this way, a multi-resolution shear rate subsequence set is obtained, providing a data foundation of different time scales for subsequent differential processing.

[0025] Step S232: Perform temporal difference processing on each subsequence in the multi-resolution shear rate subsequence set, calculate the difference between adjacent acquired contents, generate the shear rate difference sequence of the corresponding subsequence, and integrate the difference sequences of all subsequences to obtain the shear rate multi-resolution difference sequence set.

[0026] For example, for each subsequence in the multi-resolution shear rate subsequence set, assuming the subsequence is [S1, S2, S3, ..., Sn], its difference sequence is [ΔS1=S2-S1, ΔS2=S3-S2, ..., ΔSn-1=Sn-Sn-1]. This process is performed on each subsequence, and then the difference sequences of all subsequences are integrated according to their resolution and temporal order to obtain the multi-resolution shear rate difference sequence set. For example, for subsequences grouped by 10 data points and subsequences grouped by 20 data points, their difference sequences are calculated separately, and then these difference sequences are arranged together according to their resolution and temporal order.

[0027] Step S233: Read the continuously collected viscosity data from the triple time-domain correlated sensing data, and divide it into multiple groups of continuous subsequences with different durations using the same time-domain resolution partitioning strategy to obtain a multi-resolution viscosity subsequence set.

[0028] Similarly, for the viscosity data in the triple time-domain correlated sensing data, the same time-domain resolution partitioning strategy as used for processing the slurry shear rate data is adopted, dividing it into multiple sets of continuous subsequences of different durations to obtain a multi-resolution viscosity subsequence set. This ensures consistency and comparability in the processing of shear rate and viscosity data in subsequent analyses.

[0029] Specifically, the continuous collection of viscosity data from the triple temporal correlation sensing data is read, and the viscosity data is divided into subsequences of different durations according to a preset temporal resolution partitioning strategy (such as grouping by different numbers of data points as mentioned above). For example, the viscosity data is partitioned into groups of 10 data points and groups of 20 data points respectively to obtain subsequences at different resolutions, forming a multi-resolution viscosity subsequence set.

[0030] Step S234: Perform temporal difference processing on each subsequence in the multi-resolution viscosity subsequence set, calculate the difference between adjacent acquired contents, generate the viscosity difference sequence of the corresponding subsequence, and integrate the difference sequences of all subsequences to obtain the viscosity multi-resolution difference sequence set.

[0031] Each subsequence in the multi-resolution viscosity subsequence set is subjected to time-domain differencing, similar to the method used for the shear rate subsequence. This involves calculating the difference between two adjacent data points in the subsequence to obtain the viscosity difference sequence for each subsequence. Then, the difference sequences of all subsequences are integrated to obtain a multi-resolution viscosity difference sequence set, enabling analysis of viscosity variations across multiple time scales.

[0032] Step S235: Bind the shear rate multi-resolution difference sequence set and the viscosity multi-resolution difference sequence set point by point according to the corresponding resolution, perform cross-validation on the bound content, mark the time domain intervals where both difference sequences conform to the preset content, and obtain the cross-resolution abrupt change interval set.

[0033] For shear rate and viscosity difference sequences at the same resolution, their corresponding data points are bound together. For example, at a resolution of 10 data points per group, the first data point of the shear rate difference sequence is bound to the first data point of the viscosity difference sequence, the second data point to the second data point, and so on. Then, the bound data points are checked according to preset criteria (such as a set difference threshold). If, at a certain time point, the values ​​of both the shear rate and viscosity difference sequences exceed the preset threshold, the time domain interval containing that time point is marked. All time domain intervals that meet the criteria are summarized to obtain a set of cross-resolution abrupt change intervals.

[0034] Step S236: Merge overlapping time-domain intervals in the cross-resolution mutation interval set, perform duration verification on the merged intervals, remove intervals whose durations do not conform to the preset content, and obtain the state mutation time-domain interval set.

[0035] Specifically, overlapping time-domain intervals in the set of cross-resolution mutation intervals can be merged. For example, if there are two time-domain intervals [1-5] and [3-7] with overlapping parts, they can be merged into [1-7]. Then, the merged interval is validated for duration. Assuming the preset duration requirement is that the interval length is not less than 3 seconds, if the length of a merged interval is less than 3 seconds, it is removed from the set.

[0036] Step S240: Map the set of time-domain intervals of state change to the ultrasonic-laser co-path sensing signal part in the triple time-domain associated sensing data, extract the sensing signal content in the corresponding time-domain interval, and generate a set of suspected interference sensing signal content.

[0037] Mapping the set of state change time-domain intervals to the ultrasound-laser co-path sensing signal portion involves associating the state change time-domain intervals with corresponding time periods in the ultrasound-laser co-path sensing signal based on timestamps. Extracting the sensing signal content within the corresponding time-domain intervals involves extracting the ultrasound-laser co-path sensing signals within these potentially interfering time periods, forming a set of suspected interfering sensing signal content. Specifically, based on the start and end times of each interval in the state change time-domain interval set, the corresponding time period is found in the ultrasound-laser co-path sensing signal portion of the triple-time-domain associated sensing data. For example, if there is an interval of [2-5] seconds in the state change time-domain interval set, then the signal content within the 2-5 second time period is extracted from the ultrasound-laser co-path sensing signal.

[0038] Step S250: Compare the frequency domain content of each signal segment in the suspected interference sensing signal content set with the sensing signal segments in the adjacent non-abrupt time domain interval, mark the signal segments whose frequency domain content differences match the preset content, and generate a determined interference sensing signal content set.

[0039] In one implementation, step S250 may specifically include the following steps S251-S256:

[0040] Step S251: Read a single suspected interference signal segment from the suspected interference sensing signal content set, perform time-frequency joint transformation on the time-domain sampling content of the suspected interference signal segment, and generate a time-frequency feature map containing time, frequency and amplitude information, wherein each grid corresponds to the amplitude content of a preset time window and frequency range.

[0041] Specifically, a single suspected interference signal segment is selected from the set of suspected interference sensing signals. Assume the time-domain sampling content of this signal segment is a time series of length M. A time-frequency joint transformation method (such as short-time Fourier transform) is used to process this signal segment. First, the signal segment is divided into multiple overlapping time windows. A Fourier transform is performed on the signal within each time window to obtain the frequency distribution of the signal within that time window. Then, the frequency distributions of all time windows are combined to form a time-frequency feature map. For example, the signal segment is divided into 10-millisecond time windows, and a Fourier transform is performed within each window to obtain the frequency distribution within each window, ultimately forming a time-frequency feature map with time as the horizontal axis, frequency as the vertical axis, and amplitude represented by color intensity.

[0042] Step S252: Read the sensing signal segment in the non-abrupt time domain interval adjacent to the suspected interference signal segment in the triple time domain correlated sensing data, process it using the same time-frequency joint transformation strategy, and generate a reference time-frequency feature map containing the same time and frequency dimensions.

[0043] Specifically, based on the temporal location of the suspected interference signal segment, the adjacent non-abrupt time domain interval of the sensing signal segment is found in the triple time-domain correlated sensing data. Then, the same time-frequency joint transformation method (such as the short-time Fourier transform mentioned above, also divided into 10-millisecond time windows) is used to process the signal segment to generate a reference time-frequency feature map.

[0044] Step S253: Compare the time-frequency feature map with the reference time-frequency feature map grid by grid, record the difference in amplitude content of each grid, and generate a grid difference set containing descriptions of all grid differences, with each description corresponding to the difference of one grid.

[0045] The time-frequency grid-by-grid comparison involves comparing each grid cell at the same location in the time-frequency feature map with the reference time-frequency feature map, calculating the amplitude differences for each cell. The differences across all cells are recorded to form a grid difference set. Each description in this set corresponds to the difference of one cell, reflecting the amplitude differences between the suspected interfering signal segment and the reference signal segment within that time window and frequency range.

[0046] Specifically, for each corresponding grid in the time-frequency feature map and the reference time-frequency feature map, their amplitude difference is calculated. For example, if the amplitude of a certain grid in the time-frequency feature map is A, and the amplitude of the corresponding grid in the reference time-frequency feature map is B, then the difference of that grid is |AB|. The difference values ​​of all grids are recorded to form a grid difference set.

[0047] Step S254: Perform continuity analysis on the content in the grid difference set, and statistically analyze the length of grids that continuously conform to the preset difference content in the time dimension and the coverage of the frequency dimension, and generate the difference continuity and concentration description content.

[0048] Continuity analysis examines the content of the grid difference set to determine which grid differences occur consecutively. Statistical analysis of the length of grids that continuously match preset difference content in the time dimension and the coverage in the frequency dimension is used to analyze the distribution of differences in time and frequency. The description of difference continuity and concentration is a description of these statistical results, reflecting the degree of concentration and continuity of interference signals in time and frequency.

[0049] Specifically, the content in the set of grid differences is checked based on preset difference criteria (such as a set difference threshold). If the difference value of a grid exceeds the preset threshold, the grid is considered to meet the difference criteria. Then, the length of the grids that continuously meet the difference criteria in the time dimension (i.e., the number of consecutive time windows that meet the criteria) and the coverage in the frequency dimension (i.e., the frequency range that continuously meets the criteria) are statistically analyzed. For example, if it is found that within a consecutive time window (from the 5th to the 10th time window), the differences of grids in a certain frequency range (e.g., 100-200Hz) all exceed the preset threshold, then the length of this time dimension is recorded as 5 time windows, and the frequency coverage range is recorded as 100-200Hz. All such statistical results are summarized to generate a description of the continuity and concentration of differences.

[0050] Step S255: Compare the description of the difference continuity and concentration with the preset interference feature content, and mark the suspected interference signal segments whose description content completely matches the preset content.

[0051] The preset interference characteristics are descriptions of the time and frequency features of interference signals pre-defined based on experience and experiments. The descriptions of the continuity and concentration of differences are compared with the preset interference characteristics. If the descriptions completely match the preset characteristics, the suspected interference signal segment is considered highly likely to be an interference signal and is marked.

[0052] Specifically, assume that the preset interference characteristics are a time dimension with a length of no less than 3 time windows and a frequency dimension coverage of no less than 50Hz. The description of difference continuity and concentration is compared with this preset content. If the description of difference continuity and concentration of a suspected interference signal segment satisfies a time dimension length of 4 time windows and a frequency dimension coverage of 60Hz, then the suspected interference signal segment is marked.

[0053] Step S256: Summarize all marked suspected interference signal segments, check and remove duplicate signal segments, and integrate the remaining signal segments to obtain a set of confirmed interference sensing signal contents.

[0054] Specifically, all marked suspected interference signal segments are collected. Then, it is checked whether there are duplicate signal segments. For example, if there are two signal segments [1-5] and [1-5], which are duplicates, one of them is removed. The remaining signal segments are integrated to finally obtain the set of confirmed interference sensing signal content.

[0055] Step S260: Remove the set of identified interfering sensing signals from the ultrasonic-laser co-path sensing signals in the multi-source slurry monitoring dataset, and perform time-domain completion and splicing on the remaining sensing signals to obtain the pure bubble sensing signals after interference removal.

[0056] Removing the set of identified interfering sensing signals from the ultrasound-laser copath sensing signal involves removing the portions identified as interfering signals from the original signal. Temporal completion and stitching process the remaining sensing signal after interference removal, filling in the time gaps created by the removal and stitching the remaining signal segments together to form a continuous signal. The resulting interference-free pure bubble sensing signal contains only signals generated by the bubble, providing a clean signal foundation for subsequent analysis of bubble-related information.

[0057] Specifically, based on the determined temporal position of each signal segment in the set of interference sensing signals, the corresponding portion is located in the ultrasonic-laser co-path sensing signal of the multi-source slurry monitoring dataset and removed. After removal, some time gaps may appear. For these gaps, interpolation methods can be used for temporal completion. For example, using linear interpolation, the signal value within the gap is calculated based on the signal values ​​before and after the gap. Then, the remaining signal segments are spliced ​​together in chronological order to finally obtain the interference-free pure bubble sensing signal.

[0058] Step S300: Perform signal analysis on the pure bubble sensing signal after interference removal to obtain information on bubble size distribution, instantaneous position and stress state.

[0059] In one implementation, step S300 may specifically include the following steps S310-S360:

[0060] Step S310: Read the time-domain sampling content of the pure bubble sensing signal after interference removal, perform time-frequency joint transformation processing on the time-domain sampling content, and generate a time-frequency transformation spectrum containing the correspondence between time, frequency and amplitude. Each data point in the time-frequency transformation spectrum corresponds to the amplitude acquisition content of the preset time window and frequency range, so as to cover the time and frequency dimensions of all time-domain sampling content.

[0061] Assuming the time-domain sampling content of the interference-free pure bubble sensing signal is a time series of length N, a time-frequency joint transformation method (such as short-time Fourier transform) is used to process this signal. First, the signal is divided into multiple overlapping time windows, and a Fourier transform is performed on the signal within each time window to obtain the frequency distribution of the signal within that time window. Then, the frequency distributions of all time windows are combined to form a time-frequency transformation map. For example, the signal is divided into 20-millisecond time windows, and a Fourier transform is performed within each window to obtain the frequency distribution within each window. Finally, a time-frequency transformation map is formed with time as the horizontal axis, frequency as the vertical axis, and amplitude represented by color intensity. This map covers the time and frequency dimensions of the entire time-domain sampling content of the signal.

[0062] Step S320: Divide the time-frequency conversion spectrum into windows according to the preset multi-resolution rules to generate a set of multiple continuous time-frequency windows of different durations. Each window contains time-frequency conversion spectrum content with a fixed time range and the full frequency range, so that there is a preset proportion of time overlap between the windows, covering the time dimension of the entire time-frequency conversion spectrum.

[0063] In one implementation, step S320 may specifically include the following steps S321-S326:

[0064] Step S321: Read the time dimension range of the time-frequency conversion graph, divide the time dimension range into multiple time intervals of different lengths according to the preset multi-resolution ratio, and obtain a set of multi-resolution time intervals. The duration of each interval increases sequentially according to the preset ratio, covering the entire time dimension range.

[0065] The preset multi-resolution ratio is a pre-defined set of ratio values ​​used to divide the time dimension of the time-frequency conversion spectrum into time intervals of different lengths. The multi-resolution time interval set is a collection of these time intervals of different lengths, with the duration of each interval increasing sequentially according to a preset ratio. These intervals can cover the entire time dimension, thus allowing for the analysis of the time-frequency conversion spectrum at different time scales.

[0066] Specifically, the time dimension of the time-frequency conversion graph is read, assuming the time range is from 0 to 100 seconds. Based on a preset multi-resolution ratio (e.g., 1:2:4), this time dimension is divided into multiple time intervals of different lengths. For example, the first interval is 10 seconds long, the second is 20 seconds long, and the third is 40 seconds long. Starting from the beginning of the time dimension, these intervals are sequentially divided to obtain a set of multi-resolution time intervals, such as [(0-10), (10-20), (20-30), (30-50), (50-90), (90-100)], etc.

[0067] Step S322: For each time interval in the multi-resolution time interval set, extract all frequency dimension content of the corresponding time range from the time-frequency conversion graph to generate the initial content of the single-resolution time-frequency window. Each initial content corresponds to the time-frequency conversion graph content of a time interval.

[0068] Extracting all frequency dimensions of a corresponding time range from the time-frequency conversion graph involves finding the corresponding time range in the time-frequency conversion graph for each time interval in the multi-resolution time interval set, and then extracting the signal content of all frequency dimensions within that time range. The initial content of the generated single-resolution time-frequency window is a time-frequency window containing signal content of a preset time range and the full frequency range. Each initial content corresponds to the time-frequency conversion graph content of one time interval.

[0069] Step S323: Adjust the boundaries of the initial content of the single-resolution time-frequency window to ensure that the window content of adjacent time intervals has a preset proportion of time overlap. The start and end time points of the adjusted window are moved according to the overlap ratio to cover the entire time dimension range.

[0070] In one implementation, step S323 may include the following steps S3231-S3236:

[0071] Step S3231: Read the start and end time points of the time interval of the initial content of the single-resolution time-frequency window, generate the time boundary content of the single window, and each content corresponds to the start and end time markers of a window.

[0072] For the initial content of each single-resolution time-frequency window, read the start and end times of its time interval. For example, if the time interval of the initial content of a certain single-resolution time-frequency window is (10-20) seconds, then record the start time of the window as 10 seconds and the end time as 20 seconds, forming the single-window time boundary content. Perform this processing on the initial content of all single-resolution time-frequency windows to obtain the single-window time boundary content for each window.

[0073] Step S3232: Calculate the duration of the initial content of the single-resolution time-frequency window, calculate the overlap time length of adjacent windows according to the preset overlap ratio, and generate window overlap duration content, which is the preset proportion of the total window duration.

[0074] The duration of the initial content of a single-resolution time-frequency window is calculated by subtracting the start time from the end time of the window. The preset overlap ratio is a pre-defined proportion of time overlap between adjacent windows. Based on the duration and the preset overlap ratio, the overlap time length between adjacent windows is calculated, generating the window overlap duration content. This duration content is a preset proportion of the total window duration and is used for subsequent adjustments to the window's time boundaries. Specifically, for the initial content of a single-resolution time-frequency window, assuming its start time is t1 and end time is t2, the window duration is t2-t1. Assuming the preset overlap ratio is 0.2, if the window duration is 10 seconds, the overlap time length between adjacent windows is 10*0.2=2 seconds, generating a window overlap duration content of 2 seconds. This calculation is performed on the initial content of all single-resolution time-frequency windows to obtain the window overlap duration content for each window.

[0075] Step S3233: Subtract the window overlap duration from the end time of the current window to obtain the start time of the next window. The end time of the next window is the start time plus the total window duration, generating the adjusted window time boundary content.

[0076] Assume the current window ends at time t2, and the window overlap duration is Δt. Then the next window starts at time t2 - Δt. Assume the total window duration is T, then the next window ends at time (t2 - Δt) + T. For example, if the current window ends at 20 seconds, the window overlap duration is 2 seconds, and the total window duration is 10 seconds, then the next window starts at time 20 - 2 = 18 seconds and ends at time 18 + 10 = 28 seconds. Record these calculated start and end times to generate the adjusted window time boundary content.

[0077] Step S3234: Map the adjusted window time boundary content to the time-frequency conversion graph, extract all frequency dimension content of the corresponding time range, and generate the adjusted single-resolution time-frequency window content, with each content corresponding to an adjusted window.

[0078] Mapping the adjusted window time boundary content to the time-frequency conversion graph involves finding the corresponding time range in the graph based on the start and end times of the adjusted window. Then, all frequency dimension content within that time range is extracted to generate the adjusted single-resolution time-frequency window content. Each piece of content corresponds to an adjusted window, and these windows have a preset proportion of time overlap. Specifically, the corresponding time range is found in the time-frequency conversion graph based on the start and end times of the adjusted window time boundary content. For example, if the adjusted window starts at 18 seconds and ends at 28 seconds, all frequency dimension content within the time range of 18-28 seconds is extracted from the time-frequency conversion graph to generate the adjusted single-resolution time-frequency window content. This process is repeated for all adjusted windows to obtain the content for each adjusted window.

[0079] Step S3235: Perform overlap verification between the adjusted single-resolution time-frequency window content and the content of the previous window, check whether the frequency dimension content within the overlap time range is completely consistent, and generate window overlap verification results.

[0080] Specifically, for the content of the adjusted single-resolution time-frequency window and the content of the previous window, their overlapping time range is determined. For example, if the adjusted window time range is (18-28) seconds and the previous window time range is (10-20) seconds, the overlapping time range is (18-20) seconds. In the time-frequency conversion graph, it is checked whether the frequency dimension content of these two windows within the overlapping time range of (18-20) seconds is completely consistent. If they are consistent, it is recorded as a successful verification; if they are inconsistent, it is recorded as a failed verification. The verification results of the overlapping parts of all windows are summarized to generate the window overlap verification result.

[0081] Step S3236: Based on the window overlap verification results, fine-tune the start and end time points of the window to make the content of the overlapping part completely consistent and without time dimension gaps, covering the entire time dimension range, and generating a single-resolution time-frequency window set with boundary adjustment.

[0082] Based on the window overlap verification results, if there are inconsistencies in the overlapping content or time dimension gaps, the start and end time points of the windows are fine-tuned. The purpose of fine-tuning is to ensure that the content of the overlapping parts is completely consistent and to eliminate time dimension gaps, ultimately enabling all windows to cover the entire time dimension range. The generated set of single-resolution time-frequency windows with adjusted boundaries is the set of windows that meets the requirements after fine-tuning.

[0083] Specifically, if the window overlap verification result shows that the overlapping content of a certain window is inconsistent, such as a difference in amplitude at a certain frequency point within the overlapping time range, then the start or end time point of the window is fine-tuned. An iterative method can be used, fine-tuning a small time step (such as 0.1 seconds) each time, re-capturing the window content and performing overlap verification, until the overlapping content is completely consistent and there are no time dimension gaps.

[0084] Step S324: Classify the initial content of the adjusted single-resolution time-frequency window according to the resolution, and generate multiple sets of time-frequency window sequences with the same resolution. Each set of sequences contains all consecutive time-frequency windows under the same resolution, and the windows in each sequence are arranged in chronological order.

[0085] Classifying by resolution involves grouping the initial content of the adjusted single-resolution time-frequency window according to its resolution. The generated multiple sets of time-frequency window sequences with the same resolution are sequences composed of all consecutive time-frequency windows at the same resolution. The windows in each sequence are arranged in chronological order, which facilitates the unified processing and analysis of windows at the same resolution in subsequent steps.

[0086] Specifically, the windows are categorized based on the resolution of their initial content (i.e., the duration of the window) after adjustment. For example, windows with a duration of 10 seconds are grouped together, and windows with a duration of 20 seconds are grouped into another group. For each group of windows, they are arranged in chronological order to form a sequence of time-frequency windows with the same resolution.

[0087] Step S325: Perform content verification on each set of time-frequency window sequences with the same resolution, check whether the frequency dimension content of each window completely covers the entire frequency range, remove windows with incomplete frequency dimension content, and generate a set of multi-resolution time-frequency window verification sequences.

[0088] Content verification checks the frequency dimension content of each window in each set of time-frequency window sequences at the same resolution to determine whether it completely covers the entire frequency range. If the frequency dimension content of a window is incomplete, it may affect the subsequent analysis results, so these windows are removed from the sequence. The generated multi-resolution time-frequency window verification sequence set is the sequence set after content verification, with windows containing incomplete frequency dimension content removed.

[0089] Specifically, for each set of time-frequency window sequences at the same resolution, the frequency dimension content of each window is checked. Assuming the frequency range of the time-frequency transformation spectrum is 0-1000Hz, the frequency dimension content within each window is checked to ensure it covers the entire frequency range. If a window only covers the 0-800Hz frequency range, its frequency dimension content is considered incomplete, and it is removed from the sequence. This content verification is performed on all sets of time-frequency window sequences at the same resolution, ultimately generating a multi-resolution time-frequency window verification sequence set.

[0090] Step S326: Integrate all windows in the multi-resolution time-frequency window verification sequence set, arrange all windows of different resolutions in chronological order, and generate multiple continuous time-frequency window sets of different durations.

[0091] Specifically, windows of different resolutions in the multi-resolution time-frequency window verification sequence set are merged. Then, all windows are sorted according to their time start points. For example, there are windows A (time range 0-10 seconds, resolution 10 seconds), windows B (time range 5-15 seconds, resolution 20 seconds), etc., which are arranged in chronological order as [window A, window B]. All windows of different resolutions are arranged in this way to generate multiple sets of continuous time-frequency windows of different durations.

[0092] Step S330: Perform phase calibration processing on each window in the multi-resolution time-frequency window set, adjust the phase acquisition content of all data points in the window to a unified reference range, generate a calibrated time-frequency window set, and ensure that the phase acquisition content of different frequency ranges in the same window maintains a consistent reference correspondence.

[0093] Phase calibration is a process that adjusts the phase acquisition content of data points within each window of a multi-resolution time-frequency window set. During signal acquisition, various factors can cause phase differences between data points in different frequency ranges. Phase calibration adjusts the phase acquisition content of all data points within a window to a unified reference range, ensuring a consistent reference correspondence between phase acquisition content in different frequency ranges within the same window. This eliminates the impact of phase differences on subsequent analysis and improves the accuracy of the analysis.

[0094] Specifically, for each window in the multi-resolution time-frequency window set, a unified reference phase range (e.g., 0-2π) can be determined. Then, the phase acquisition content of each data point within the window is adjusted. A phase compensation method can be used: based on the phase value of a data point at a certain reference frequency within the window, the phase difference between other frequency data points and the reference frequency data point is calculated, and then the phase of the other frequency data points is compensated to adjust their phase values ​​to the unified reference range. For example, assuming the phase of the reference frequency data point is π / 2, and the phase of another frequency data point is 3π / 2, with a phase difference of π, then the phase of the other frequency data point is subtracted by π to adjust it to the unified reference range.

[0095] Step S340: Extract features from each window in the calibrated time-frequency window set, analyze the energy distribution, phase coherence and scattering spectrum features of the preset frequency band within the window, match the extracted features with the preset bubble acoustic and optical scattering feature library, mark the windows whose matching degree meets the preset threshold, and generate a candidate time-frequency window set containing bubble response content. Each window corresponds to a possible bubble sensing response time interval.

[0096] Feature extraction involves processing each window in the calibrated time-frequency window set to extract useful features. Preset frequency bands are frequency ranges predetermined based on the acoustic and optical properties of the bubble. Energy distribution, phase coherence, and scattering spectrum characteristics are analyzed within these bands. Energy distribution describes the magnitude and distribution of signal energy within the preset frequency band; phase coherence reflects the phase relationship between different frequency components; and scattering spectrum characteristics are related to the bubble's scattering characteristics of ultrasound and laser. A pre-built database of bubble acoustic and optical scattering features is established, containing various bubble acoustic and optical scattering characteristics. The extracted features are matched against this database to determine if the features within a window match the bubble's features. Windows with a matching degree meeting a preset threshold are marked; these windows may contain signal responses generated by the bubble. These windows are then aggregated to form a candidate time-frequency window set containing bubble response content, with each window corresponding to a possible bubble sensing response time interval.

[0097] Specifically, for each window in the calibrated time-frequency window set, a preset frequency band (e.g., 100-200Hz) can be determined. Then, the energy distribution within the preset frequency band is calculated, for example, by summing the squares of the signal amplitudes within that band. Phase coherence can be analyzed using methods such as cross-correlation analysis to calculate the phase correlation between different frequency components. Scattering spectrum characteristics can be obtained through scattering spectrum analysis of the signal. These extracted features are then matched with features in a preset bubble acoustic and optical scattering feature library.

[0098] Step S350: Match the time-frequency content of each window in the candidate time-frequency window set with the preset bubble particle size corresponding content to generate bubble particle size description content corresponding to each window, and integrate all bubble particle size description content in the order of collection time to obtain bubble particle size distribution.

[0099] The preset bubble size correspondence is a pre-established database containing the relationship between different time-frequency features and bubble sizes. The time-frequency content of each window in the candidate time-frequency window set is matched against this database, and bubble size descriptions are generated for each window based on the matching results. All bubble size descriptions are integrated in chronological order of acquisition time, and the number or proportion of bubbles of different sizes is statistically analyzed to obtain the final bubble size distribution, which reflects the distribution of bubble sizes at different time points.

[0100] Specifically, for each window in the candidate time-frequency window set, its time-frequency content (such as energy distribution, phase coherence, and other characteristics of a preset frequency band) is read. Then, the bubble size that best matches the time-frequency content is searched in a preset bubble size corresponding content database. For example, the database records that a certain preset energy distribution and phase coherence characteristics correspond to a bubble with a size of 50 micrometers. If the time-frequency content of a certain window matches the characteristics, the bubble size description content corresponding to that window is generated as 50 micrometers. The bubble size description content of all windows is arranged in chronological order of acquisition time, and the frequency or proportion of bubbles of different sizes is counted to obtain the bubble size distribution.

[0101] Step S360: Match the time-frequency content of each window in the candidate time-frequency window set with the preset content corresponding to the instantaneous position of the bubble and the content corresponding to the force state of the bubble, and integrate all the matched instantaneous position descriptions and force state descriptions of the bubble to obtain the instantaneous position and force state information of the bubble.

[0102] The preset instantaneous bubble position information is a pre-established database containing the correspondence between different time-frequency features and the instantaneous bubble position. The preset bubble force state information is another pre-established database containing the correspondence between different time-frequency features and the bubble force state. The time-frequency content of each window in the candidate time-frequency window set is matched against these two databases. Based on the matching results, a description of the bubble's instantaneous position and force state is generated for each window. All the matched descriptions are integrated to finally obtain the bubble's instantaneous position and force state information, which reflects the bubble's instantaneous position and force state at different time points.

[0103] Specifically, for each window in the candidate time-frequency window set, its time-frequency content is read. First, the time-frequency content is matched with a preset database of instantaneous bubble positions, and a description of the instantaneous bubble position (such as coordinates) is generated for that window based on the matching result. For example, if the database records a preset time-frequency feature corresponding to the bubble's coordinate position in the slurry (10, 20, 30), and the time-frequency content of a window matches this feature, then the instantaneous bubble position description for that window is generated as (10, 20, 30). Then, the time-frequency content is matched with a preset database of bubble force states, and a description of the bubble's force state (such as buoyancy and resistance) is generated for that window based on the matching result.

[0104] Step S400: Combining the dynamic changes of viscosity data in the multi-source slurry monitoring dataset, the evolution law of bubble particle size distribution, instantaneous position and stress state information is derived to obtain bubble migration trajectory, coalescence critical condition and escape time information.

[0105] In one implementation, step S400 may specifically include the following steps S410-S460:

[0106] Step S410: Read the continuous collection content of viscosity data in the multi-source slurry monitoring dataset, and bind the continuous collection content with the bubble particle size distribution, instantaneous position and stress state information point by point according to the collection time node to generate viscosity-bubble information time domain binding data, so that each data node is simultaneously associated with viscosity collection content and bubble information.

[0107] Continuous viscosity data acquisition consists of slurry viscosity values ​​collected over a period of time. Point-by-point binding according to acquisition time nodes establishes a one-to-one correspondence between the continuous viscosity data acquisition and information on bubble size distribution, instantaneous location, and stress state over time. The generated viscosity-bubble information time-domain bound data is a dataset containing viscosity data and bubble information. Each data node is associated with both the viscosity value at the same time point and information such as bubble size, location, and stress state. This facilitates subsequent analysis of bubble evolution patterns in conjunction with viscosity changes. Specifically, the continuous viscosity data acquisition content from the multi-source slurry monitoring dataset is read. This content is typically stored in time series format, with each data point corresponding to an acquisition time and a viscosity value. Simultaneously, information on bubble size distribution, instantaneous location, and stress state is read, which is also associated with the acquisition time. Based on the acquisition time node, each data point of the viscosity data is bound to the bubble size, location, and stress state information at the same time point.

[0108] Step S420: Continuously track the instantaneous position of bubbles in the viscosity-bubble information time-domain binding data, record the position change of each bubble at adjacent acquisition time nodes, and generate a set of continuous bubble position change trajectories.

[0109] In one implementation, step S420 may specifically include the following steps S421-S426:

[0110] Step S421: Read the instantaneous bubble position content and the corresponding time-frequency characteristics of the pure bubble sensing signal after interference removal from the viscosity-bubble information time-domain binding data. Bind the bubble position description content of each acquisition time node with the corresponding time-frequency characteristics to generate bubble time-frequency characteristic-position association data, so that each bubble position corresponds to a unique time-frequency characteristic description, covering the bubble position content of all acquisition time nodes.

[0111] The instantaneous bubble position information is the position information of each bubble at different acquisition time points recorded in the viscosity-bubble information time-domain binding data. The time-frequency characteristics of the pure bubble sensing signal after interference removal are time-frequency information reflecting bubble characteristics obtained after the previous steps. Binding the bubble position description at each acquisition time point with the corresponding time-frequency characteristics is to establish the correspondence between bubble position and its time-frequency characteristics. The generated bubble time-frequency characteristic-position association data is a dataset containing the correspondence between bubble position and time-frequency characteristics, where each bubble position corresponds to a unique time-frequency characteristic description, and this dataset covers the bubble position content at all acquisition time points. This allows for more accurate identification and location of bubbles when tracking bubble position changes in subsequent steps, by combining their unique time-frequency characteristics.

[0112] Specifically, the instantaneous position information of the bubble at each acquisition time point is read from the viscosity-bubble information time-domain bound data. Assuming the bubble position is represented by three-dimensional coordinates (x, y, z), the time-frequency characteristics of the interference-free pure bubble sensing signal at the corresponding acquisition time point are also read. These time-frequency characteristics may include information such as energy distribution and phase coherence within a preset frequency band. For each bubble at each acquisition time, its position coordinates are bound to the corresponding time-frequency characteristics one by one.

[0113] Step S422: Perform consistency verification on the bubble time-frequency features of adjacent acquisition time nodes in the bubble time-frequency feature-location association data, mark bubble pairs whose features match the preset correspondence, and generate a cross-node bubble matching mark set. Each mark contains the unique bubble identifier of the preceding and following nodes and a feature consistency description, covering the bubble comparison content of all adjacent nodes.

[0114] Consistency verification compares the time-frequency features of bubbles at adjacent acquisition time points to determine whether they are similar or conform to a pre-defined correspondence. The pre-defined correspondence is a set of judgment conditions predetermined based on the bubble's motion characteristics and the changing patterns of its time-frequency features. It is used to determine whether two time-frequency features belong to the same bubble at adjacent time points. Marking bubble pairs whose features conform to the pre-defined correspondence is to identify cases where they belong to the same bubble at adjacent acquisition time points. The generated cross-node bubble matching tag set is a collection recording information on all bubble pairs that meet the conditions. Each tag contains a unique bubble identifier from the preceding and following nodes (used to distinguish different bubbles) and a feature consistency description (such as the specific similarity value). This set covers the bubble comparison content of all adjacent nodes, providing a matching basis for subsequently constructing the bubble's motion trajectory.

[0115] Specifically, for bubble time-frequency feature-location associated data, the bubble time-frequency features of adjacent acquisition time nodes are compared sequentially according to the acquisition time order. Assuming a similarity calculation method (such as cosine similarity) is used to measure the similarity between two time-frequency features, the default correspondence is a similarity greater than 0.8. The similarity is calculated for the time-frequency feature of a bubble at the nth acquisition time node and the time-frequency features of all bubbles at the (n+1)th acquisition time node. If a similarity is greater than 0.8, the two bubbles are considered to be the same bubble in adjacent time states, the pair of bubbles is marked, and their unique identifier (e.g., bubble number) and similarity value are recorded.

[0116] Step S423: For each bubble pair in the cross-node bubble matching tag set, extract the position description of the preceding and following nodes, and combine it with the change trend of the corresponding viscosity data to construct a multi-scale motion feature sequence of the bubble. The sequence includes short-term position changes and long-term motion trends, covering all cross-node matching data for each bubble.

[0117] Extracting the positional descriptions of preceding and following nodes involves obtaining the positional information of each pair of matched bubbles at two consecutive acquisition time points from the cross-node bubble matching tag set. The corresponding viscosity acquisition trend reflects the change in slurry viscosity between these two acquisition time points; viscosity changes affect bubble movement. Constructing a multi-scale bubble motion feature sequence comprehensively considers both bubble positional and viscosity changes, describing bubble motion characteristics at both short-term and long-term scales. Short-term positional changes describe the bubble's movement between adjacent acquisition time points, while long-term motion trends reflect the overall direction and velocity changes of the bubble across multiple acquisition time points. This sequence covers all cross-node matching content for each bubble, and analysis of this sequence provides a more comprehensive understanding of the bubble's motion patterns.

[0118] Specifically, for each bubble pair in the cross-node bubble matching tag set, the position coordinates of its preceding and following nodes are first extracted. For example, if the position of the preceding bubble in a bubble pair is (x1, y1, z1) and the position of the following bubble is (x2, y2, z2), the position difference (Δx=x2-x1, Δy=y2-y1, Δz=z2-z1) can be calculated to represent the short-term position change. Simultaneously, viscosity data between corresponding acquisition time points is obtained, and its trend is analyzed, such as whether the viscosity is increasing, decreasing, or remaining stable. Based on the viscosity change trend and short-term position change, combined with information from multiple adjacent bubble pairs, the long-term motion trend of the bubbles is gradually constructed, for example, by fitting position change curves to determine their overall motion direction and velocity changes. Combining the short-term position changes and long-term motion trends forms a multi-scale motion feature sequence of the bubbles.

[0119] Step S424: The missing node content in the multi-scale motion feature sequence is filled in by interpolation using the motion trend of adjacent nodes, generating a complete continuous sequence of position changes for each bubble, so that the sequence covers all acquisition time nodes from the initial appearance of the bubble to its disappearance, with no missing content.

[0120] In actual data acquisition, various factors (such as data acquisition errors and signal loss) may lead to missing nodes in the multi-scale motion feature sequence. Interpolation using the motion trends of adjacent nodes is an effective data processing method. It estimates the content of the missing node by using the motion information (such as position change trends and speeds) of adjacent nodes before and after the missing node. The generated complete position change continuous sequence for each bubble is a sequence containing position information of all acquisition time points from the initial appearance to the disappearance of the bubble. Through interpolation, it is ensured that no content is missing in this sequence, thus allowing for a more accurate depiction of the bubble's motion trajectory.

[0121] Specifically, for multi-scale motion feature sequences containing missing nodes, the motion trends of adjacent nodes can be analyzed. For example, if the positions of two adjacent nodes before and after the missing node exhibit a linear change trend, linear interpolation can be used to estimate the position of the missing node. Assuming the position of the preceding node is (x1, y1, z1) and the position of the following node is (x2, y2, z2), the missing node lies between these two nodes with a uniform time interval. The position coordinates of the missing node can be calculated using the linear interpolation formula. For more complex motion trends, methods such as polynomial interpolation can also be used.

[0122] Step S425: Integrate all the complete position change continuous sequences of all bubbles in the order of acquisition time to generate an initial trajectory set of continuous bubble position change. Perform multi-scale feature fusion on each trajectory in the set to eliminate the trajectory differences at different scales and obtain the fused bubble trajectory content.

[0123] Integrating the continuous sequence of complete bubble position changes according to the acquisition time involves arranging the complete position information of each bubble from its initial appearance to its disappearance in chronological order, forming a set containing the motion trajectories of all bubbles—the initial trajectory set of continuous bubble position changes. Multi-scale feature fusion processes each trajectory in the set. Since the description of the trajectory may differ at different scales (such as short-term and long-term), feature fusion can eliminate these differences, making the trajectory smoother and more accurate. The final fused bubble trajectory content is trajectory information that more accurately reflects the true movement of the bubbles after processing. Specifically, the continuous sequence of complete bubble position changes is first sorted according to the acquisition time, and then all sorted sequences of bubbles are combined to form the initial trajectory set of continuous bubble position changes. For each trajectory in the set, a multi-scale feature fusion method is used. For example, a weighted average of short-term and long-term motion features can be applied, with weights set according to actual conditions to balance the influence of features at different scales. Wavelet transform and other methods can also be used to decompose and reconstruct the trajectory, eliminating noise and differences at different scales.

[0124] Step S426: Map the fused bubble trajectory content to the viscosity acquisition content in the viscosity-bubble information time domain binding data, verify the correspondence between trajectory changes and viscosity changes, remove trajectories whose correspondence does not conform to the preset content, and obtain a set of continuous bubble position change trajectories.

[0125] Mapping the fused bubble trajectory data to the viscosity data collected in the viscosity-bubble information time-domain binding data involves associating the fused trajectory of each bubble with the slurry viscosity value at the same acquisition time node to analyze the relationship between trajectory changes and viscosity changes. Verifying the correspondence between trajectory changes and viscosity changes involves checking whether the trajectory changes of each bubble conform to the viscosity changes according to pre-defined rules (such as the correlation between trajectory change direction and viscosity change trend). Removing trajectories whose correspondence does not conform to the pre-defined rules is to eliminate those trajectories that may not conform to physical laws due to measurement errors or abnormal conditions. The final set of continuously changing bubble positions contains trajectories that have a reasonable correspondence with viscosity changes, more accurately reflecting the actual movement of bubbles in the slurry.

[0126] Specifically, for each time point in the fused bubble trajectory, the viscosity data collected at the same time point in the viscosity-bubble information time-domain binding data is identified. The relationship between the trajectory changes at that time point (such as changes in position direction, velocity, etc.) and viscosity changes is analyzed. The preset condition is that when viscosity increases, the bubble's velocity should decrease or its direction of motion should change. If the trajectory change of a bubble does not conform to this preset relationship with viscosity changes, the bubble's trajectory is removed from the set.

[0127] Step S430: Perform time-domain continuous processing on each trajectory content in the set of continuously changing bubble positions to generate a complete movement path description for each bubble, and integrate all description content to obtain the bubble migration trajectory.

[0128] Temporal continuity processing further processes each trajectory in the set of continuously changing bubble positions, making the trajectory more continuous and smooth in the time domain. Since the actual collected data may have some discreteness, temporal continuity processing can more accurately reflect the continuous movement of bubbles over time. The generated complete movement path description for each bubble is a detailed description of its movement path in the slurry, including its position information at different time points. Integrating all descriptions yields the bubble migration trajectory, which is essentially summarizing the complete movement path descriptions of all bubbles to form a comprehensive information describing the migration of all bubbles in the slurry.

[0129] Specifically, for each trajectory in the set of continuously changing bubble positions, a smoothing algorithm (such as spline interpolation) is used to process the data points on the trajectory, making the trajectory more continuous and smooth in the time domain. Spline interpolation can generate a smooth curve to fit the existing data points, thereby inserting some virtual data points in the middle to make the trajectory continuous. After performing this processing on each trajectory, the complete movement path description of each bubble is obtained. The movement path description of each bubble can be organized according to the bubble's unique identifier, and finally, all descriptions are integrated to obtain the bubble migration trajectory.

[0130] Step S440: Perform continuous analysis on the bubble size distribution in the viscosity-bubble information time-domain binding data, mark the collection time nodes where bubble size merges and changes, and generate a bubble merging time node mark set.

[0131] In one implementation, step S440 may include the following steps S441-S446:

[0132] Step S441: Read the bubble size distribution and the corresponding time-frequency characteristics of the single bubble sensing response signal unit in the viscosity-bubble information time-domain binding data. Bind the bubble size description of each acquisition time node with the corresponding time-frequency characteristics to generate bubble size-time-frequency characteristic association data, ensuring that each bubble size corresponds to a unique time-frequency characteristic, covering the bubble size content of all acquisition time nodes.

[0133] The bubble size distribution information is the number or proportion of bubbles of different sizes recorded at each acquisition time point in the viscosity-bubble information time-domain binding data. The time-frequency characteristic of the single bubble sensing response signal unit describes the time and frequency characteristics of a single bubble sensing signal. Binding the bubble size description at each acquisition time point to the corresponding time-frequency characteristic establishes a one-to-one correspondence between bubble size and its time-frequency characteristic. The generated bubble size-time-frequency characteristic association data is a dataset containing this correspondence, where each bubble size corresponds to a unique time-frequency characteristic, and this dataset covers the bubble size content of all acquisition time points. This helps to more accurately analyze the changes and aggregation of bubble size through time-frequency characteristics.

[0134] Step S442: Perform differential comparison of bubble time-frequency features at adjacent acquisition time nodes in the bubble size-time-frequency feature correlation data, mark new time-frequency features that meet the merging of multiple bubble features from the previous node in the later node, and generate a set of suspected clustered bubble feature markers. Each marker contains the bubble identifier and feature merging description of the preceding and following nodes.

[0135] Differential comparison compares the bubble time-frequency features of adjacent acquisition time nodes and calculates the differences between them. A new time-frequency feature appearing in a subsequent node that matches the merging of multiple bubble features from previous nodes is identified because when multiple bubbles merge, the time-frequency features of their sensed signals change, forming a new time-frequency feature. This new feature may be formed by merging multiple bubble features from previous nodes. The generated set of suspected merged bubble feature tags is a collection recording bubble feature information for all possible merging scenarios. Each tag includes bubble identifiers from the preceding and following nodes (used to identify bubbles participating in the merging) and a description of the feature merging (such as the specific changes in the merged time-frequency features).

[0136] Specifically, for bubble size-time-frequency feature correlation data, the bubble time-frequency features of adjacent acquisition time nodes are compared sequentially according to the acquisition time order. Difference calculation or similarity calculation methods are used for differential comparison. Assuming that feature vectors are used to represent time-frequency features, the similarity between a certain time-frequency feature vector of a later node and the combined vector of multiple time-frequency feature vectors of a previous node is calculated. If the similarity is greater than a preset threshold (e.g., 0.8), it is considered that the time-frequency feature of the later node conforms to the case of multiple bubble features from the previous node being combined. This case is marked, and the bubble identifiers of the preceding and following nodes, as well as the specific feature merging description, such as which time-frequency feature components have changed, are recorded.

[0137] Step S443: For each marker in the suspected clustered bubble feature marker set, extract the instantaneous position description of the bubble corresponding to the preceding and following nodes, construct a bubble position association map, which includes the spatial correspondence of bubble positions, and verify whether the position of the suspected clustered bubble meets the preset clustering spatial conditions.

[0138] For each marker in the suspected clustered bubble feature marker set, the instantaneous bubble position description of the corresponding preceding and following nodes is extracted from the viscosity-bubble information time-domain binding data, assuming the bubble position is represented by three-dimensional coordinates (x, y, z). Based on this position information, a bubble position association map is constructed. The bubble positions can be graphically represented in three-dimensional space, and the relationships between them can be indicated by lines or other methods. The positions of suspected clustered bubbles are verified to meet preset clustering space conditions. For example, the Euclidean distance between two suspected clustered bubbles is calculated. If the distance is less than a preset threshold (e.g., 5 mm), their positions are considered to meet the clustering space conditions; otherwise, they are considered not to meet the conditions.

[0139] Step S444: Mark the suspected clustering bubble markers that have passed the location association map verification, and map them to the multi-resolution viscosity subsequence set. Verify the correspondence between viscosity changes and suspected clustering at different resolutions, mark the suspected clustering markers that meet the preset content at all resolutions, and generate a cross-resolution clustering marker set.

[0140] Mapping suspected bubble clustering markers that pass location correlation mapping to a multi-resolution viscosity subsequence set involves associating suspected bubble clustering markers whose locations meet the spatial conditions for clustering with viscosity data in the multi-resolution viscosity subsequence set to analyze the relationship between viscosity changes and suspected clustering events. Verifying the correspondence between viscosity changes and suspected clustering at different resolutions is beneficial because different resolutions allow observation of viscosity changes at different time scales, providing a more comprehensive analysis of the impact of viscosity on bubble clustering. Suspected clustering markers that meet preset criteria across all resolutions are then labeled. These preset criteria are conditions established based on experimental and theoretical analysis regarding the correspondence between viscosity changes and bubble clustering, such as a certain trend in viscosity change before and after the bubble clustering event. The generated cross-resolution clustering marker set records all markers whose viscosity changes and suspected clustering meet the preset criteria at different resolutions; these markers are more likely to represent actual bubble clustering events.

[0141] Specifically, for each suspected clustering bubble marker that passes the location correlation map verification, its corresponding data acquisition time point is identified. Then, viscosity data at different resolutions for that time point are found in the multi-resolution viscosity subsequence set. The viscosity changes before and after that time point at different resolutions are analyzed. For example, at a certain high resolution, a slow increase in viscosity is observed before the suspected clustering moment, followed by a rapid decrease in viscosity after the clustering moment; a similar viscosity change trend is also observed at low resolution, and these trends conform to preset criteria. If the viscosity changes of a suspected clustering marker conform to preset criteria at all resolutions, then that marker is marked.

[0142] Step S445: Assign each marker in the cross-resolution clustering marker set to the bubble size distribution content, verify whether the bubble size of the next node matches the merged content of multiple bubble sizes of the previous node, generate a set of size merging verification results, and each result corresponds to the verification status of a suspected clustering marker.

[0143] Mapping each marker in the cross-resolution clustering marker set to bubble size distribution data involves associating suspected clustering markers verified by viscosity changes at different resolutions with bubble size distribution data to further verify whether bubble clustering has occurred. Verifying whether the bubble size of the subsequent node matches the merging of multiple bubble sizes from the previous node is based on the physical phenomenon of bubble clustering—that multiple small bubbles merge to form a larger bubble—by comparing the bubble sizes of the preceding and following nodes to determine if this merging pattern holds. The generated particle size merging verification result set records the verification status of each suspected clustering marker; each result corresponds to the verification result of a suspected clustering marker, providing more direct evidence for ultimately confirming the bubble clustering event.

[0144] Specifically, for each marker in the cross-resolution clustering marker set, the bubble size distribution at the corresponding preceding and following acquisition time nodes is identified. Assume the preceding node has two bubbles with sizes d1 and d2, and the following node introduces a new, larger bubble d3. If d3 equals d1 + d2 within a certain error range (considering actual measurement errors and energy losses during the clustering process, an error range, such as ±10%, can be set), then the bubble size at the following node is considered to match the combined size distribution of multiple bubbles from the preceding node, and the marker's verification result is considered passed; otherwise, the verification result is considered failed.

[0145] Step S446: Based on the particle size merging verification result set, remove the suspected merging markers that failed the verification, summarize the collection time nodes corresponding to the remaining markers, and generate a bubble merging time node marker set to ensure that each node corresponds to a real bubble merging event and has no erroneous markers.

[0146] Based on the particle size merging verification result set, suspected agglomeration markers that failed the verification were removed. This was to exclude markers that, while exhibiting suspected agglomeration characteristics in terms of location and viscosity changes, did not conform to the actual bubble particle size merging, thus ensuring that the final markers were all genuine bubble agglomeration events. The collection time nodes corresponding to the remaining markers were summarized to generate a bubble agglomeration time node marker set. Each node in this set corresponds to a real bubble agglomeration event, with no erroneous markers.

[0147] Specifically, the particle size merging verification result set is traversed. For each verification result, if the result is unsuccessful, the corresponding suspected clustering marker is removed from the cross-resolution clustering marker set. The acquisition time nodes corresponding to the remaining markers are extracted and summarized. For example, these time nodes are arranged in ascending order to form an ordered list of time nodes, and finally, a bubble clustering time node marker set is generated.

[0148] Step S450: Map the bubble coalescence time node marker set to the continuous collection content of viscosity data in the multi-source slurry monitoring dataset, extract the viscosity collection content of the corresponding node, generate the viscosity content set corresponding to bubble coalescence, and obtain the critical condition for bubble coalescence.

[0149] Specifically, for each time node in the bubble coalescence time node marker set, the corresponding time point is found in the continuous collection of viscosity data in the multi-source slurry monitoring dataset, and the viscosity value at that time point is extracted. For example, the bubble coalescence time node marker set includes time nodes t1, t2, t3, etc., and the viscosity values ​​v1 at time t1, v2 at time t2, and v3 at time t3 are found in the viscosity data. These viscosity values ​​are then aggregated to generate a viscosity content set corresponding to bubble coalescence. Statistical analysis is performed on the viscosity values ​​in this set, such as calculating the mean, standard deviation, and other statistics, while observing the distribution range of the viscosity values. If it is found that the viscosity values ​​at the time of most bubble coalescence events are concentrated in a certain range, then this range can be defined as the critical condition for bubble coalescence, for example, the critical condition for bubble coalescence is a viscosity value between 0.5 and 1.0 Pa·s.

[0150] Step S460: Based on the bubble migration trajectory, track the number of collection time nodes experienced by each bubble from the first detection to the disappearance of its signal or movement to the boundary of the preset monitoring area, convert it into time length content, and integrate all content to obtain the bubble escape time information; among them, the determination of the bubble signal disappearance or arrival at the boundary needs to be combined with its movement trend and force state for comprehensive judgment.

[0151] Specifically, for each bubble in the bubble migration trajectory, counting begins from the first acquisition time point, continuously tracking its position and signal status. When a bubble signal disappears or its position reaches the boundary of a preset monitoring area, a judgment is made based on the bubble's motion trend (such as velocity direction, acceleration, etc.) and force state (such as buoyancy, drag direction and magnitude, etc.). For example, if the bubble's velocity direction is towards the boundary and the force causes it to continue moving towards the boundary, while the signal disappears, it can be determined that the bubble has escaped; if the bubble reaches the boundary but its velocity direction shows a tendency to turn back, and the force direction is not conducive to its continued escape, it is not determined to have escaped. The number of acquisition time points during which each bubble escapes (is determined to have escaped) is recorded. Assuming each acquisition time interval is Δt seconds, the number of acquisition time points is multiplied by Δt to obtain the escape time of each bubble.

[0152] Step S500: Correlation characterization processing is performed on the information on bubble migration trajectory, coalescence critical conditions and escape time with the impermeability and frost resistance of concrete to obtain the monitoring results of bubble evolution state related to concrete performance.

[0153] In one implementation, step S500 may specifically include the following steps S510-S560:

[0154] Step S510: Read all the information on bubble migration trajectory, coalescence critical conditions and escape time, organize it in the order of collection time, and generate a continuous time-domain sequence of bubble evolution information.

[0155] Reading all the information about bubble migration trajectories, coalescence critical conditions, and escape time involves collecting all the information about bubbles obtained in the previous steps. Organizing this information chronologically according to the acquisition time ensures continuity over time. The resulting continuous temporal sequence of bubble evolution information is a sequence that integrates bubble migration trajectories, coalescence critical conditions, and escape time information in chronological order, providing a clearer view of bubble evolution at different points in time.

[0156] Specifically, bubble migration trajectories, clustering critical conditions, and escape time information can be read from previously stored data. The bubble migration trajectory itself contains the position information of each bubble at different time points; the clustering critical conditions are associated with time through a set of bubble clustering time node markers; and the escape time information records the time from detection to escape for each bubble. This information is sorted according to the acquisition time, for example, using time as an index to map different types of information to corresponding time points. Finally, a continuous temporal sequence of bubble evolution information is generated, where each time point contains relevant information such as the bubble migration trajectory, clustering status, and escape time at that moment.

[0157] Step S520: Match the bubble migration trajectory content in the time-domain continuous sequence of bubble evolution information with the preset concrete impermeability performance related content to generate migration trajectory-impermeability performance time-domain related content.

[0158] In one implementation, step S520 may specifically include the following steps S521-S526:

[0159] Step S521: Read all the contents of the bubble migration trajectory, extract the complete movement path description of each bubble, and generate a bubble movement path description set.

[0160] Reading the complete bubble migration trajectory involves extracting all information about the bubble migration trajectory from the continuous time-domain sequence of bubble evolution information. Extracting the complete movement path description for each bubble involves refining and organizing the entire movement path information of each bubble from its initial appearance to its final state (escape or cessation of movement, etc.). The generated bubble movement path description set is a collection containing detailed descriptions of all bubble movement paths, where each description corresponds to the complete movement path of one bubble, providing a concrete information foundation for subsequent matching with content related to concrete impermeability.

[0161] Specifically, for the bubble migration trajectory portion of the continuous temporal sequence of bubble evolution information, the movement path of each bubble is extracted according to its unique identifier. Assuming the bubble's movement path is represented by a series of three-dimensional coordinates (x, y, z) at different time points, these coordinates for each bubble are organized into a complete path description in chronological order. For example, the movement path description of bubble A could be [(x1, y1, z1), (x2, y2, z2), ..., (xn, yn, zn)], representing its position at different time points as these coordinates.

[0162] Step S522: Match each path description in the bubble movement path description set with the preset concrete impermeability performance related content to generate a single bubble path-impermeability related description.

[0163] Matching each path description in the bubble movement path description set with preset concrete impermeability performance correlation content involves searching for the corresponding concrete impermeability performance based on the path description characteristics (such as path length, path direction, and areas traversed). The preset concrete impermeability performance correlation content is likely derived from extensive experiments and research, recording the relationship between different path characteristics and concrete impermeability performance. For example, bubbles with longer paths that pass through preset areas may reduce concrete impermeability performance. The generated single bubble path-impermeability correlation description is information describing the correspondence between each bubble's movement path and concrete impermeability performance. It includes a specific description of the bubble path's impact on concrete impermeability performance, such as the degree to which it improves or decreases impermeability.

[0164] Specifically, for each path description in the set of bubble movement path descriptions, its characteristics are analyzed, such as calculating the path length and determining the area traversed by the path. Then, matching information is searched in the preset concrete impermeability performance related content.

[0165] Step S523: Integrate all individual bubble paths and anti-seepage association descriptions for the same time node according to the collection time node, and generate a summary of anti-seepage associations for the same time node.

[0166] Integrating the path-impermeability correlation descriptions of all individual bubbles at the same data collection time point summarizes the impact of all bubbles on the impermeability of concrete at that time. Since multiple bubbles may exist at the same time point, their impact on concrete impermeability needs to be considered comprehensively. The generated summary of impermeability correlations at the same time point provides comprehensive information describing the combined impact of all bubbles on the impermeability of concrete at a given data collection time point. For example, a numerical value can represent the overall change in concrete impermeability at that time point, or a more detailed description can illustrate the degree of influence of different types of bubbles.

[0167] Step S524: Arrange all the anti-seepage correlation summary contents at the same time node in the order of collection time to generate a continuous time domain sequence of anti-seepage performance correlation.

[0168] Arranging all the summary data on seepage resistance at the same time point in chronological order of collection time involves arranging the summary data on seepage resistance at each time point obtained previously in chronological order, forming a temporally continuous sequence. The generated continuous temporal sequence of seepage resistance performance correlation can intuitively display the changes in concrete seepage resistance over time, and the relationship between these changes and the bubble migration trajectory. Specifically, the summary data on seepage resistance at each time point is sorted according to the order of collection time; for example, time can be used as an index to arrange the summary data from different time points sequentially. The final result is a sequence with time as the horizontal axis and concrete seepage resistance performance correlation information as the vertical axis. This sequence reflects the dynamic changes in concrete seepage resistance over time throughout the monitoring process, and has a temporal correspondence with the bubble migration trajectory.

[0169] Step S525: Perform time-domain difference analysis on the contents of the continuous time-domain sequence associated with anti-permeability performance, mark the collection time nodes where the associated contents change, and generate a set of marker nodes for changes in anti-permeability performance.

[0170] Temporal difference analysis (TDA) calculates the differences between adjacent data points in a continuous temporal sequence related to concrete impermeability performance. By analyzing these differences, changes in the data can be identified. Marking data collection time points where changes occur is indicated when the calculated difference is non-zero, signifying a change in the concrete impermeability performance at that time point.

[0171] Specifically, for each data point in the continuous time-domain sequence of anti-seepage performance correlation, the difference between it and the previous data point is calculated. For example, assuming the continuous time-domain sequence of anti-seepage performance correlation is [v1,v2,v3,…,vn], calculate Δv1=v2-v1, Δv2=v3-v2,…,Δvn-1=vn-vn-1. If a certain difference Δvi is not zero, it is considered that the anti-seepage performance correlation content has changed at time node i+1, and that time node is marked.

[0172] Step S526: Integrate the temporal continuous sequence of anti-seepage performance with the set of nodes marked by changes in anti-seepage performance to generate migration trajectory-time domain correlation content of anti-seepage performance.

[0173] Specifically, the continuous time-domain sequence of permeability resistance performance is used as the basic data, and information from the set of permeability resistance performance change node markers is appended to the corresponding time nodes. For example, in the continuous time-domain sequence of permeability resistance performance, time nodes that have changed can be specially marked (e.g., represented by different colors or symbols), or annotations can be added next to them to explain the changes.

[0174] Step S530: Match the bubble coalescence critical condition content in the time-domain continuous sequence of bubble evolution information with the preset concrete frost resistance performance related content to generate coalescence condition-frost resistance performance time-domain related content.

[0175] The pre-defined correlation content for concrete frost resistance is a pre-established database or model that records the relationship between different bubble coalescence critical conditions and concrete frost resistance. Matching the bubble coalescence critical condition content in the continuous time-domain sequence of bubble evolution information with this correlation content is to identify the concrete frost resistance corresponding to each bubble coalescence critical condition. The generated coalescence condition-frost resistance time-domain correlation content is an information set that records the temporal correspondence between bubble coalescence critical conditions and concrete frost resistance, which can intuitively show the impact of bubble coalescence on concrete frost resistance at different time points.

[0176] Specifically, for the critical conditions for bubble coalescence in the continuous time-domain sequence of bubble evolution information, their characteristics are analyzed, such as the viscosity range and coalescence time point. Then, matching information is searched in the preset concrete frost resistance performance related content. This matching operation is performed on all bubble coalescence critical conditions, and these related information are organized according to the acquisition time sequence to finally generate the coalescence condition-frost resistance performance time-domain related content.

[0177] Step S540: Match the bubble escape time information in the bubble evolution information time domain continuous sequence with the preset concrete impermeability and frost resistance related content to generate escape time-dual performance time domain related content.

[0178] The pre-defined correlation between concrete impermeability and frost resistance is a pre-established comprehensive database or model that records the relationship between different bubble escape times and concrete impermeability and frost resistance. Matching the bubble escape time information in the continuous time-domain sequence of bubble evolution information with this correlation is to find the concrete impermeability and frost resistance corresponding to each bubble escape time. The generated escape time-dual performance time-domain correlation content is an information set that records the temporal correspondence between bubble escape time and concrete impermeability and frost resistance, which can simultaneously demonstrate the impact of bubble escape time on these two properties of concrete.

[0179] Step S550: Align the migration trajectory-permeability performance time-domain related content, the aggregation condition-freeze resistance performance time-domain related content, and the escape time-dual performance time-domain related content according to the collection time node to generate performance-related time-domain integrated data.

[0180] In one implementation, step S550 may specifically include the following steps S551-S556:

[0181] Step S551: Read all the content of the migration trajectory-permeability resistance time domain correlation content, extract the correlation description content of each acquisition time node, and generate the permeability resistance resistance correlation time domain sequence.

[0182] Reading the complete content of the time-domain correlation between the migration trajectory and impermeability performance involves retrieving all information from the previously generated data on this correlation. Extracting the correlation description for each acquisition time node involves extracting specific descriptive information about the correlation between the bubble migration trajectory and the concrete impermeability performance at each time node, forming a new sequence. The generated impermeability performance correlation time-domain sequence is indexed by time nodes and records the correlation between the concrete impermeability performance and the bubble migration trajectory at each time point, providing a single type of time series data for subsequent time alignment with other correlation content.

[0183] Specifically, for the migration trajectory-permeability performance time-domain correlation content, the correlation description content of each time node is extracted sequentially according to the collection time node order. For example, if the correlation description at time node t1 is a 5% reduction in concrete permeability performance, this description is recorded as the data of the permeability performance correlation time-domain sequence at time node t1. This extraction operation is performed on all collection time nodes, ultimately generating the permeability performance correlation time-domain sequence.

[0184] Step S552: Read all the content of the aggregation condition-antifreeze performance time domain correlation content, extract the correlation description content of each collection time node, and generate the antifreeze performance correlation time domain sequence.

[0185] Reading the complete content of the time-domain correlation between bubble aggregation conditions and concrete frost resistance means obtaining all information from this correlation data. Extracting the correlation description content for each acquisition time node involves extracting the specific descriptive information about the correlation between bubble aggregation conditions and concrete frost resistance at each time node, forming a new sequence. The generated time-domain sequence of frost resistance correlation is indexed by time nodes and records the correlation between concrete frost resistance and bubble aggregation conditions at each time point, providing a single type of time series data for subsequent time alignment with other correlation content.

[0186] Specifically, for the time-domain correlation content between the aggregation condition and the frost resistance performance, the correlation description content of each time node is extracted sequentially according to the order of the collection time nodes. For example, if the correlation description at time node t2 is that the frost resistance performance of concrete is improved by 3%, this description is recorded as the data of the frost resistance performance correlation time-domain sequence at time node t2.

[0187] Step S553: ​​Read all the content of the escape time-dual performance time domain correlation content, extract the correlation description content of each acquisition time node, and generate the dual performance correlation time domain sequence.

[0188] Reading the complete content of the bubble escape time-dual performance time-domain correlation data involves retrieving all information from this correlation data. Extracting the correlation description content for each acquisition time node involves extracting specific descriptive information about the correlation between bubble escape time and concrete impermeability and frost resistance at each time node, forming a new sequence. The generated dual performance correlation time-domain sequence is a sequence that records the correlation between concrete impermeability and frost resistance and bubble escape time at each time point, indexed by time nodes.

[0189] Step S554: Align the anti-permeability performance correlation time-domain sequence and the antifreeze performance correlation time-domain sequence point by point according to the acquisition time node to generate anti-permeability-antifreeze performance correlation alignment data.

[0190] Aligning the time-domain sequences related to impermeability and frost resistance according to their acquisition time points establishes a temporal correspondence between the two sequences, allowing the information on impermeability and frost resistance at the same time point to be combined. The resulting aligned impermeability-frost resistance data is a comprehensive set of information containing the correlation between concrete impermeability, frost resistance, and air bubbles at the same time point. This alignment operation provides a more intuitive observation of the correlation changes between these two properties over the same period.

[0191] Step S555: Bind the anti-permeability and anti-freeze performance correlation alignment data and the dual performance correlation time domain sequence point by point according to the acquisition time node to generate the initial data for performance correlation integration.

[0192] Binding the alignment data of impermeability and frost resistance performance to the dual-performance correlation time-domain sequence point by point according to the acquisition time node is to re-correlate and combine the previously generated alignment data with the dual-performance correlation time-domain sequence in time. The generated performance correlation integration initial data is a more comprehensive set of information containing the correlation between concrete impermeability, frost resistance performance and bubble migration trajectory, aggregation conditions, and escape time at the same time node. Through this binding operation, the three different correlation information are integrated together.

[0193] Specifically, for the alignment data of impermeability-freezing performance and the time-domain sequence of dual performance correlation, a one-to-one correspondence is established according to the collection time node. For example, at time node t, the alignment data of impermeability-freezing performance shows a 5% decrease in impermeability and a 3% increase in freezing performance, while the data of the dual performance correlation time-domain sequence shows a 2% decrease in impermeability and a 4% decrease in freezing performance. This information is comprehensively processed (e.g., by taking an average or weighted average), and the processed result is used as the initial data for performance correlation integration at time node t. This binding operation is performed for all time nodes, ultimately generating the initial data for performance correlation integration.

[0194] Step S556: Perform node-by-node verification on the initial data of performance correlation integration, remove conflicting node data, supplement missing node correlation descriptions, and obtain performance correlation time-domain integrated data.

[0195] Performing node-by-node verification on the initial data for performance correlation integration involves checking the rationality of the data at each time point and identifying any conflicts or contradictions. Removing conflicting node data involves deleting illogical or contradictory data from the data sequence to ensure accuracy and reliability. Supplementing missing node correlation descriptions involves using appropriate methods (such as interpolation or estimation based on preceding and following node data) to fill in the missing information when certain time points in the data sequence are found to have missing correlation descriptions.

[0196] Specifically, for each time point in the initial data integration of performance correlation, the rationality of the data is checked. For example, if the correlation description of concrete impermeability at a certain time point shows both improvement and decrease, this is a case of content conflict, and the data for that time point is removed. For time points with missing correlation descriptions, if the data of the preceding and following nodes have a certain regularity, linear interpolation can be used to estimate the correlation description content of the missing node.

[0197] Step S560: Perform node-by-node consistency verification on the content of the performance-related time-domain integrated data to generate monitoring results of the bubble evolution state of concrete performance correlation containing continuous correlation descriptions.

[0198] Performing node-by-node consistency checks on the integrated time-domain data related to performance involves verifying whether the correlation description at each time node in the data sequence is consistent and logical with the preceding and following nodes. If the correlation description at a certain time node differs significantly from the preceding and following nodes or does not conform to physical laws, an inconsistency is considered to exist, requiring adjustment or correction. Generating monitoring results of the bubble evolution state of concrete performance correlations, containing continuous correlation descriptions, yields a continuous and logical information set after consistency checks. This comprehensively and accurately demonstrates the correlation between bubble evolution and the impermeability and frost resistance of concrete, providing important evidence for evaluating and improving concrete performance.

[0199] Specifically, for each time node in the integrated time-domain data of performance correlation, the relationship between its correlation description and the correlation descriptions of adjacent nodes is analyzed. Methods such as difference analysis and trend analysis can be used to determine consistency. For example, if the correlation description of concrete impermeability at a certain time node suddenly changes from a 5% decrease to a 10% increase without a reasonable explanation (such as no significant change in bubble evolution), then inconsistency is considered to exist. In this case, adjustments can be made by referring to the data of adjacent nodes and considering the actual situation of bubble evolution to make the correlation description of that time node consistent with the overall trend. This consistency verification and adjustment operation is performed on all time nodes, ultimately generating monitoring results of the bubble evolution status of concrete performance correlation containing continuous correlation descriptions.

[0200] Please see Figure 2This is a schematic diagram of a computer system provided in an embodiment of the present invention. The computer system includes at least a processor 101, a communication interface 102, and a memory 103. The processor 101, communication interface 102, and memory 103 can be connected via a bus or other means. The processor 101 (or Central Processing Unit, CPU) is the computing and control core of the computer system, capable of parsing various instructions and processing various data within the computer system. The communication interface 102 may optionally include standard wired interfaces or wireless interfaces (such as Wi-Fi, mobile communication interfaces, etc.), and can be used for sending and receiving data under the control of the processor 101; the communication interface 102 can also be used for data transmission and interaction within the computer system. The memory 103 is a memory device in the computer system used to store programs and data. It is understood that the memory 103 here can include the computer system's built-in memory, or it can include extended memory supported by the computer system. The memory 103 provides storage space, which stores the computer system's operating system; this invention does not limit this. The processor 101 executes the real-time monitoring method for concrete bubbles in water conservancy projects provided in the above embodiments of the present invention by running the computer program in the memory 103.

Claims

1. A method for real-time monitoring of air bubbles in concrete applied to water conservancy projects, characterized in that, The method includes: The ultrasonic-laser co-path sensing signal, slurry shear rate data, and viscosity data of hydraulic concrete slurry are collected simultaneously to obtain a multi-source slurry monitoring dataset containing time-series synchronization markers. The ultrasonic-laser co-path sensing signal, slurry shear rate data, and viscosity data in the multi-source slurry monitoring dataset correspond to the same acquisition time node. By combining the slurry shear rate data and viscosity data in the multi-source slurry monitoring dataset, interference components are removed from the ultrasonic-laser co-path sensing signal in the multi-source slurry monitoring dataset to obtain the interference-free pure bubble sensing signal. The interference-free pure bubble sensing signal is analyzed to obtain information on bubble size distribution, instantaneous position, and stress state. Specifically, this includes: reading the time-domain sampling content of the interference-free pure bubble sensing signal; performing time-frequency joint transformation processing on the time-domain sampling content to generate a time-frequency transformation spectrum containing the correspondence between time, frequency, and amplitude. Each data point in the time-frequency transformation spectrum corresponds to amplitude acquisition content within a preset time window and frequency range, thus covering the time and frequency dimensions of all time-domain sampling content; dividing the time-frequency transformation spectrum into windows according to preset multi-resolution rules to generate multiple sets of continuous time-frequency windows of different durations. Each window contains time-frequency transformation spectrum content within a fixed time range and the entire frequency range, ensuring a preset proportion of time overlap between windows, covering the time dimension of the entire time-frequency transformation spectrum; performing phase calibration processing on each window in the multi-resolution time-frequency window set, adjusting the phase acquisition content of all data points within the window to a unified reference range, generating a calibrated time-frequency window set, ensuring that different data points within the same window have the same phase acquisition content. The phase acquisition content within the frequency range maintains a consistent baseline correspondence. Features are extracted from each window in the calibrated time-frequency window set. The energy distribution, phase coherence, and scattering spectrum characteristics of the preset frequency band within the window are analyzed. The extracted features are matched with a preset bubble acoustic and optical scattering feature library. Windows with a matching degree meeting a preset threshold are marked, generating a candidate time-frequency window set containing bubble response content. Each window corresponds to a possible bubble sensing response time interval. The time-frequency content of each window in the candidate time-frequency window set is matched with preset bubble particle size corresponding content to generate bubble particle size description content for each window. All bubble particle size description content is integrated in the order of acquisition time to obtain the bubble particle size distribution. The time-frequency content of each window in the candidate time-frequency window set is matched with preset bubble instantaneous position corresponding content and bubble force state corresponding content, respectively. All matched bubble instantaneous position description content and bubble force state description content are integrated to obtain bubble instantaneous position and force state information. By combining the dynamic changes of viscosity data in the multi-source slurry monitoring dataset, the evolution law of the bubble particle size distribution, instantaneous position and stress state information is derived to obtain bubble migration trajectory, coalescence critical condition and escape time information; The correlation characterization process is performed on the bubble migration trajectory, coalescence critical conditions, and escape time information with the concrete impermeability and frost resistance properties to obtain the bubble evolution state monitoring results related to concrete performance.

2. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 1, characterized in that, The process involves combining the slurry shear rate data and viscosity data from the multi-source slurry monitoring dataset to remove interfering components from the ultrasonic-laser copath sensing signal, resulting in a pure bubble sensing signal after interference removal. This includes: Read the time-domain sampling content of the ultrasonic-laser co-path sensing signal in the multi-source slurry monitoring dataset, bind the continuous acquisition content of the slurry shear rate data in the multi-source slurry monitoring dataset with the time-domain sampling content point by point according to the acquisition time node, generate shear rate-sensing signal time-domain binding data, so that each sensing signal sampling point corresponds to a unique shear rate acquisition content; Read the continuous acquisition content of viscosity data in the multi-source slurry monitoring dataset, bind the shear rate-sensing signal time-domain binding data with the continuous acquisition content of viscosity data point by point according to the acquisition time node, generate triple time-domain associated sensing data, so that each data node is simultaneously associated with sensing signal, shear rate and viscosity acquisition content; The continuous acquisition of slurry shear rate data and viscosity data in the triple time-domain correlated sensing data is subjected to time-domain differential processing. The time-domain intervals in which the differential results of the acquisition content conform to the preset content are marked, and a set of time-domain intervals of state change is generated. The set of time-domain intervals of state change is mapped to the ultrasonic-laser co-path sensing signal part in the triple time-domain associated sensing data, and the sensing signal content in the corresponding time-domain interval is extracted to generate a set of suspected interference sensing signal content. Each signal segment in the suspected interference sensing signal content set is compared with the sensing signal segment in the adjacent non-abrupt time domain interval in the frequency domain. Signal segments whose frequency domain content differences match the preset content are marked, and a determined interference sensing signal content set is generated. Remove the set of determined interference sensing signals from the ultrasonic-laser copath sensing signals in the multi-source slurry monitoring dataset, and perform time-domain completion and splicing on the remaining sensing signals to obtain the interference-free pure bubble sensing signals.

3. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 2, characterized in that, The continuous acquisition of slurry shear rate data and viscosity data from the triple time-domain correlated sensing data is subjected to time-domain differential processing. Time-domain intervals where the differential results of the acquired content conform to preset content are marked, generating a set of time-domain intervals for state abrupt changes, including: The continuously acquired content of slurry shear rate data in the triple time-domain correlated sensing data is read and divided into multiple groups of continuous subsequences of different durations according to a preset time-domain resolution division strategy, to obtain a multi-resolution shear rate subsequence set, each subsequence containing a fixed number of continuously acquired contents. Each subsequence in the multi-resolution shear rate subsequence set is subjected to temporal domain difference processing, the difference between adjacent acquired contents is calculated, the shear rate difference sequence of the corresponding subsequence is generated, and the difference sequences of all subsequences are integrated to obtain the shear rate multi-resolution difference sequence set. Read the continuously collected viscosity data from the triple time-domain correlated sensing data, and divide it into multiple groups of continuous subsequences of different durations using the same time-domain resolution partitioning strategy to obtain a multi-resolution viscosity subsequence set; Each subsequence in the multi-resolution viscosity subsequence set is subjected to temporal domain difference processing, the difference between adjacent acquired contents is calculated, the viscosity difference sequence of the corresponding subsequence is generated, and the difference sequences of all subsequences are integrated to obtain a viscosity multi-resolution difference sequence set. The shear rate multi-resolution difference sequence set and the viscosity multi-resolution difference sequence set are bound point by point according to the corresponding resolution. The bound content is cross-validated, and the time domain intervals in which both difference sequences conform to the preset content are marked to obtain the cross-resolution abrupt change interval set. The overlapping time-domain intervals in the cross-resolution mutation interval set are merged, the merged intervals are validated for duration, and intervals whose durations do not conform to the preset content are removed to obtain the state mutation time-domain interval set.

4. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 2, characterized in that, The step of comparing each signal segment in the suspected interference sensing signal content set with the sensing signal segments in adjacent non-abrupt time domain intervals in the frequency domain, marking signal segments whose frequency domain content differences match preset content, and generating a determined interference sensing signal content set includes: Read a single suspected interference signal segment from the suspected interference sensing signal content set, perform time-frequency joint transformation on the time-domain sampling content of the suspected interference signal segment, and generate a time-frequency feature map containing time, frequency and amplitude information, wherein each grid corresponds to the amplitude content of a preset time window and frequency range; Read the sensing signal segment in the non-abrupt time domain interval adjacent to the suspected interference signal segment in the triple time domain associated sensing data, process it using the same time-frequency joint transformation strategy, and generate a reference time-frequency feature map containing the same time and frequency dimensions; The time-frequency feature map is compared with the reference time-frequency feature map grid by grid, and the difference in amplitude content of each grid is recorded to generate a grid difference set containing descriptions of all grid differences, with each description corresponding to the difference of one grid. A continuity analysis is performed on the content in the set of grid differences, and the length of the grids that continuously meet the preset difference content in the time dimension and the coverage in the frequency dimension are statistically analyzed to generate a description of the continuity and concentration of differences. The description of the difference continuity and concentration is compared with the preset interference feature content, and suspected interference signal segments whose description content completely matches the preset content are marked. All marked suspected interference signal segments are summarized, duplicate signal segments are investigated and removed, and the remaining signal segments are integrated to obtain a set of confirmed interference sensing signal contents.

5. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 1, characterized in that, The step of dividing the time-frequency conversion spectrum into windows according to a preset multi-resolution rule to generate multiple sets of continuous time-frequency windows of different durations includes: Read the time dimension range of the time-frequency conversion spectrum, divide the time dimension range into multiple time intervals of different lengths according to a preset multi-resolution ratio, and obtain a set of multi-resolution time intervals. The duration of each interval increases sequentially according to a preset ratio, covering the entire time dimension range. For each time interval in the set of multi-resolution time intervals, all frequency dimension contents of the corresponding time range are extracted from the time-frequency conversion graph to generate the initial content of a single-resolution time-frequency window. Each initial content corresponds to the time-frequency conversion graph content of a time interval. The initial content of the single-resolution time-frequency window is adjusted to ensure that the window content of adjacent time intervals has a preset proportion of time overlap. The start and end time points of the adjusted window are moved according to the overlap ratio to cover the entire time dimension range. The initial content of the adjusted single-resolution time-frequency window is classified according to resolution, generating multiple sets of time-frequency window sequences with the same resolution. Each set of sequences contains all consecutive time-frequency windows at the same resolution, and the windows in each sequence are arranged in chronological order. For each set of time-frequency window sequences with the same resolution, perform content verification to check whether the frequency dimension content of each window completely covers the entire frequency range, remove windows with incomplete frequency dimension content, and generate a set of multi-resolution time-frequency window verification sequences. Integrate all windows in the multi-resolution time-frequency window verification sequence set, arrange all windows of different resolutions in chronological order, and generate multiple continuous time-frequency window sets of different durations.

6. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 5, characterized in that, The step of adjusting the boundaries of the initial content of the single-resolution time-frequency window to ensure that the window content of adjacent time intervals has a preset proportion of time overlap, and the start and end time points of the adjusted window are moved according to the overlap proportion to cover the entire time dimension range, including: Read the start and end time points of the time interval of the initial content of the single-resolution time-frequency window, generate the time boundary content of the single window, and each content corresponds to the start and end time markers of a window; Calculate the duration of the initial content of the single-resolution time-frequency window, calculate the overlap time length of adjacent windows according to the preset overlap ratio, and generate window overlap duration content, which is a preset proportion of the total window duration. Subtract the window overlap duration from the end time of the current window to obtain the start time of the next window. The end time of the next window is the start time plus the total window duration, generating the adjusted window time boundary content. The adjusted window time boundary content is mapped to the time-frequency conversion graph, and all frequency dimension content of the corresponding time range is extracted to generate the adjusted single-resolution time-frequency window content, with each content corresponding to an adjusted window; The content of the adjusted single-resolution time-frequency window is checked for overlap with the content of the previous window. The frequency dimension content within the overlapping time range is checked to see if they are completely consistent, and a window overlap check result is generated. Based on the window overlap verification results, the start and end time points of the windows are fine-tuned to ensure that the content of the overlapping parts is completely consistent and without time dimension gaps, covering the entire time dimension range, and generating a single-resolution time-frequency window set with adjusted boundaries.

7. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 1, characterized in that, The dynamic changes in viscosity data from the multi-source slurry monitoring dataset are combined to deduce the evolutionary patterns of bubble size distribution, instantaneous position, and stress state information, resulting in bubble migration trajectories, coalescence critical conditions, and escape time information, including: Read the continuous collection content of viscosity data in the multi-source slurry monitoring dataset, and bind the continuous collection content with the bubble particle size distribution, instantaneous position and stress state information point by point according to the collection time node to generate viscosity-bubble information time domain binding data, so that each data node is simultaneously associated with viscosity collection content and bubble information; The instantaneous position of bubbles in the viscosity-bubble information time-domain bound data is continuously tracked, and the position change of each bubble at adjacent acquisition time nodes is recorded to generate a set of continuous bubble position change trajectories. Each trajectory in the set of continuously changing bubble positions is processed in the time domain to generate a complete description of the movement path of each bubble. All descriptions are then integrated to obtain the bubble migration trajectory. Continuous analysis is performed on the bubble size distribution in the viscosity-bubble information time-domain bound data, and the collection time nodes where bubble size merges and changes are marked, generating a bubble merging time node mark set; The bubble coalescence time node marker set is mapped to the continuous collection content of viscosity data in the multi-source slurry monitoring dataset. The viscosity collection content of the corresponding node is extracted to generate the viscosity content set corresponding to bubble coalescence, and the critical condition for bubble coalescence is obtained. Based on the bubble migration trajectory, the number of collection time nodes experienced by each bubble from the first detection to the disappearance of its signal or movement to the boundary of the preset monitoring area is tracked, converted into time length content, and all content is integrated to obtain the bubble escape time information; among them, the determination of the disappearance of bubble signal or arrival at the boundary needs to be combined with its movement trend and force state for comprehensive judgment.

8. The method for real-time monitoring of concrete air bubbles in water conservancy projects as described in claim 7, characterized in that, The continuous tracking of the instantaneous bubble position content in the viscosity-bubble information time-domain bound data, recording the position change of each bubble at adjacent acquisition time nodes, and generating a set of continuous bubble position change trajectories includes: Read the instantaneous bubble position content and the corresponding time-frequency characteristics of the pure bubble sensing signal after interference removal from the viscosity-bubble information time-domain binding data, bind the bubble position description content of each acquisition time node with the corresponding time-frequency characteristics, generate bubble time-frequency characteristic-position association data, so that each bubble position corresponds to a unique time-frequency characteristic description, covering the bubble position content of all acquisition time nodes; The consistency of bubble time-frequency features at adjacent acquisition time nodes in the bubble time-frequency feature-location association data is verified. Bubble pairs whose features match the preset correspondence are marked, and a cross-node bubble matching mark set is generated. Each mark contains a unique bubble identifier and feature consistency description of the preceding and following nodes, covering the bubble comparison content of all adjacent nodes. For each bubble pair in the cross-node bubble matching tag set, the position descriptions of the preceding and following nodes are extracted. Combined with the changing trends of the corresponding viscosity data, a multi-scale motion feature sequence of the bubble is constructed. The sequence includes short-term position changes and long-term motion trends, covering all cross-node matching data for each bubble. The missing node content in the multi-scale motion feature sequence is filled in by interpolation using the motion trend of adjacent nodes, generating a complete continuous sequence of position changes for each bubble, so that the sequence covers all acquisition time nodes from the initial appearance of the bubble to its disappearance, with no missing content. All the complete continuous sequence of bubble position changes are integrated in the order of acquisition time to generate an initial trajectory set of continuous bubble position changes. Multi-scale feature fusion is performed on each trajectory in the set to eliminate the trajectory differences at different scales and obtain the fused bubble trajectory content. The fused bubble trajectory content is mapped to the viscosity acquisition content in the viscosity-bubble information time-domain binding data. The correspondence between trajectory changes and viscosity changes is verified, and trajectories whose correspondence does not conform to the preset content are removed to obtain a set of continuously changing bubble positions trajectories.

9. A computer system, characterized in that, include: A memory, wherein a computer program is stored; A processor is configured to load the computer program to implement the real-time monitoring method for concrete air bubbles applied to water conservancy projects as described in any one of claims 1-8.

Images