A method and system for monitoring the condition and diagnosing anomalies of hydraulic supports in building construction.
By constructing a hydraulic support monitoring matrix and multi-dimensional state fingerprint matching, the problem of distinguishing between sensor anomalies and engineering anomalies in existing technologies has been solved, enabling accurate monitoring and safe management of hydraulic supports in building construction.
Patent Information
- Application Number
- CN202611610797.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-19
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies cannot effectively distinguish between sensor anomalies and engineering anomalies in the monitoring of hydraulic supports in building construction. They lack multi-parameter correlation analysis, have poor adaptability to fixed thresholds, and cannot accurately identify anomaly types and expansion trends.
A hydraulic support monitoring matrix is constructed. Data is collected from multiple sources to form an overall state matrix. Time synchronization and reliability evaluation are performed, the evaluation benchmark is dynamically adjusted, multi-dimensional state fingerprint matching and physical correlation analysis are conducted, and a diagnostic report is generated.
It enables accurate classification and diagnosis of sensor anomalies and engineering anomalies, improves the accuracy and adaptability of monitoring, provides situational awareness capabilities from single points to areas, and significantly enhances construction safety management.
Smart Images

Figure CN123116028A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent monitoring technology for hydraulic supports, specifically relating to a method and system for monitoring the status and diagnosing anomalies of hydraulic supports in building construction. Background Technology
[0002] Hydraulic support systems in building construction are subjected to dynamically changing construction loads during operations such as concrete pouring. Their stress and deformation states exhibit significant time-varying characteristics. Continuous monitoring of key parameters such as hydraulic pressure, support displacement, and support inclination angle is an important means to ensure construction safety.
[0003] Currently, existing technologies typically use pressure sensors, displacement sensors, and tilt sensors to acquire construction support data and trigger alarms for exceeding limits using preset thresholds. In recent years, monitoring solutions based on multi-sensor data acquisition and cloud platform management have emerged, along with technological explorations for intelligent early warning based on construction stages. However, these existing technologies still have the following shortcomings: First, they lack the credibility assessment of the sensor data itself; sensor anomalies such as communication interruptions, occasional jumps, and long-term drift cannot be effectively distinguished from actual engineering anomalies, leading to frequent false alarms. Second, there is a lack of correlation analysis between various monitoring parameters; when pressure is abnormal, it is impossible to determine whether the cause is load change, support settlement, hydraulic leakage, or sensor malfunction. Third, using fixed thresholds for alarm judgment is difficult to adapt to the differentiated requirements of support status at different construction stages such as installation, pouring, curing, and formwork removal. Fourth, anomaly diagnosis is limited to a single measuring point, lacking synchronous analysis of changes in the status of adjacent supports, making it impossible to determine whether the anomaly is a local problem or has an expanding trend.
[0004] Therefore, there is an urgent need for a new monitoring and anomaly diagnosis method and system that can comprehensively evaluate the reliability of monitoring data, perform correlation diagnosis of multi-source parameters, adaptively identify the construction stage, and accurately classify and spatially trace abnormal states. Summary of the Invention
[0005] In view of this, the purpose of the present invention is to provide a method and system for monitoring the status and diagnosing anomalies of hydraulic supports in building construction.
[0006] To achieve the above objectives, the present invention provides a method for monitoring and diagnosing the condition of hydraulic supports in building construction, comprising the following steps: A hydraulic support monitoring matrix is constructed, and multi-source sensors are configured for each support unit to collect data on pressure, displacement, height, tilt angle and operating status of each support, forming an overall status matrix; The multi-source monitoring data in the overall state matrix are synchronized and aligned under a time reference to obtain a standardized data sequence. The credibility of each channel data in the standardized data sequence is evaluated, and the evaluation results are integrated into a comprehensive credibility coefficient. Based on the comparison results between the comprehensive credibility coefficient and the preset threshold value, the data in the standardized data sequence is divided into high-credibility data and low-credibility data; Low-confidence data is classified, processed, and reconstructed, and the processed data is combined with high-confidence data to form a dataset to be diagnosed. The current construction stage is automatically identified based on the temporal characteristics of the dataset to be diagnosed, and the evaluation benchmark is dynamically adjusted accordingly. Extract multidimensional state fingerprints of each support from the dataset to be diagnosed, match them with a pre-stored fingerprint template library, and identify the current state category of each support; When the status category is determined to be abnormal, the same supporting parameter data is extracted from the dataset to be diagnosed for physical correlation analysis, and the abnormality type is classified and diagnosed. Based on the diagnostic results, adjacent support data are extracted from the overall state matrix for spatial comparison to determine the scope of impact and expansion trend of the anomaly. A diagnostic report is generated and output based on the comprehensive credibility coefficient, status category, classification diagnosis results, and expansion trend.
[0007] Preferably, constructing the hydraulic support monitoring matrix and forming the overall state matrix includes: Each hydraulic support unit is assigned a unique number and its spatial coordinates are bound; Pressure sensors, displacement sensors, height sensors, and tilt sensors are installed on each support unit to collect data on hydraulic pressure, support displacement, support height change, and support tilt angle of each support, respectively. All monitoring data from the same support unit are aggregated into a single support state vector for that support. The single support state vectors of all support units are arranged according to spatial coordinates to form the overall state matrix.
[0008] Preferably, the time-based synchronization and alignment processing of multi-source monitoring data in the overall state matrix includes: Using a unified timing signal as a time reference, time tags are added to the data from each sensor in the overall state matrix; Resampling is performed according to a pre-defined unified sampling time sequence. For channels that do not have original sampled values at the target time, the corresponding values at the target time are calculated by interpolation using the effective values at adjacent times and then backfilled into the overall state matrix to form a standardized data sequence.
[0009] Preferably, the credibility of each channel data in the standardized data sequence is evaluated, and the evaluation results are integrated into a comprehensive credibility coefficient, including: The data in each channel of the standardized data sequence were evaluated for continuity, abruptness, stability, and multi-parameter consistency, resulting in four scores. The continuity score is determined based on the data packet reception integrity rate and whether the value is stuck for a long time; the abrupt change score is determined based on the comparison between the rate of change at adjacent time points and the upper limit of the reasonable rate of change; the stability score is determined based on the coefficient of variation of the data within the window; and the consistency score is determined based on the degree of response matching between pressure changes, displacement changes, and tilt angle changes of the same support. The overall credibility coefficient of each channel is obtained by weighted summation of the four scores.
[0010] Preferably, the process of dividing the data in the standardized data sequence into high-confidence data and low-confidence data based on the comparison result of the comprehensive confidence coefficient and the preset threshold value, and classifying and reconstructing the low-confidence data includes: Channel data whose comprehensive credibility coefficient reaches the threshold value are marked as high-credibility data and retained. Channel data that does not reach the threshold value is marked as low-confidence data. Its time series features are extracted for pattern recognition. When it is determined to be short-term data missing, interpolation compensation is performed. When it is determined to be isolated mutation, it is removed and replaced with neighboring values. When it is determined to be long-term drift, it is marked as sensor drift and a calibration reminder is generated. When it is determined to be multi-parameter synchronous response, it is determined to be a real engineering anomaly and is retained. The processed data is merged with the high-reliability data to form a dataset to be diagnosed.
[0011] Preferably, automatically identifying the current construction stage based on the temporal characteristics of the dataset to be diagnosed and dynamically adjusting the evaluation benchmark accordingly includes: Extract the pressure change rate and displacement change of all supports from the dataset to be diagnosed and determine the stage. When the pressure change rate and displacement change of each support meet the preset installation stage characteristics, it is determined to be the installation stage; when it meets the pouring stage characteristics, it is determined to be the pouring stage; when it meets the curing stage characteristics, it is determined to be the curing stage; when it meets the demolding stage characteristics, it is determined to be the demolding stage. Adjust the reasonable rate of change threshold and feature weight coefficients for subsequent evaluation stages based on the identified construction stages.
[0012] Preferably, the process of extracting multidimensional state fingerprints of each support from the dataset to be diagnosed and matching them with a pre-stored fingerprint template library to identify the state category, and reading parameter data of the same support from the dataset to be diagnosed for physical correlation analysis to classify and diagnose the anomaly type when the state category is determined to be abnormal, includes: The pressure change rate, displacement change rate, and tilt angle change of each support are extracted from the dataset to be diagnosed, and the average pressure change rate of adjacent supports is extracted from the overall state matrix, and combined into a multi-dimensional state fingerprint. The multidimensional state fingerprint is weighted and the typical state templates in the pre-stored fingerprint template library are calculated. The state category corresponding to the closest template is selected as the diagnostic conclusion. The fingerprint template library contains at least standard templates corresponding to normal state and multiple abnormal states. When an anomaly is detected, the pressure, displacement, and tilt data of the support within the same time period are read from the standardized data sequence. The direction and magnitude of the change of each parameter are analyzed. When the pressure decreases and the displacement increases and the tilt changes, it is determined to be support settlement. When the pressure decreases but the displacement and tilt do not change and the adjacent support does not respond, it is determined to be hydraulic leakage. When the data of a single channel changes abruptly and other channels of the same support do not respond, it is determined to be sensor failure. When the pressure of a certain support increases and the pressure of the adjacent support decreases, it is determined to be load transfer.
[0013] Preferably, based on the diagnostic results, adjacent support data are extracted from the overall state matrix for spatial comparison to determine the scope and expansion trend of the anomaly, including: Read the spatial coordinates of all supports from the overall state matrix and calculate the distance between them. Mark the support pairs with a distance less than a preset threshold as adjacent and establish an adjacency relationship. After obtaining the abnormal support number based on the classification and diagnosis results, all adjacent supports of the support are found through adjacency relationship. The state data of adjacent supports in the same time period are extracted from the overall state matrix. By comparing the magnitude ratio of load change rate and displacement change rate, it is determined whether the load has undergone spatial transfer and the direction of transfer. Expand the analysis scope to all supports in the same construction area to determine whether the anomaly is localized, regional, or widespread, and determine the risk level accordingly.
[0014] To achieve the above objectives, the present invention also provides a system for monitoring and diagnosing the condition of hydraulic supports in building construction, comprising: A multi-source data acquisition module is deployed in each hydraulic support unit to collect multi-source monitoring data from each support and aggregate it into an overall status matrix; The data preprocessing and time synchronization module is used to perform time alignment and standardization on the overall state matrix and output a standardized data sequence. The data credibility evaluation module is used to evaluate the credibility of standardized data sequences in multiple dimensions and output the comprehensive credibility coefficient of each channel. The abnormal data reconstruction module is used to classify and process low-confidence data based on the comprehensive confidence coefficient and output the dataset to be diagnosed. The construction phase identification module is used to extract time-series features from the dataset to be diagnosed and automatically identify the current construction phase. The state fingerprint matching and anomaly diagnosis module is used to extract multidimensional state fingerprints from the dataset to be diagnosed and match them with a pre-stored fingerprint template library to identify the state category. When an anomaly is determined, physical correlation analysis is performed to output the specific anomaly type. The spatial tracing and risk assessment module is used to analyze the anomaly propagation path and expansion trend based on the spatial coordinates in the overall state matrix. The human-computer interaction and output module is used to output a comprehensive report that includes diagnostic conclusions and treatment recommendations.
[0015] The beneficial effects of this invention are as follows: 1. This invention addresses the problems of unreliable data and high false alarm rates caused by sensor malfunctions or communication interference in existing technologies by establishing a data reliability screening mechanism that includes four dimensions of evaluation: continuity, abrupt change, stability, and consistency. By constructing a multi-dimensional state fingerprint of pressure, displacement, tilt angle, and adjacent supports and performing pattern matching with a typical state template library, it achieves accurate classification and diagnosis of different types of anomalies such as support settlement, hydraulic leakage, sensor malfunction, and load transfer, solving the problem that traditional technologies can only trigger alarms for exceeding limits but cannot determine the cause of the anomaly. By automatically identifying the construction stage based on the temporal characteristics of monitoring data and dynamically adjusting the evaluation benchmark, it solves the technical challenge of high false alarm rates at different construction stages with fixed thresholds. Through adjacent support state comparison and influence area expansion analysis based on spatial adjacency matrices, it achieves an upgrade from single-point monitoring to regional situational awareness, providing a complete intelligent solution for construction safety management, from data governance, state identification, anomaly diagnosis to risk tracing, significantly improving the accuracy and engineering practicality of hydraulic support safety monitoring. Attached Figure Description
[0016] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration: Figure 1 This is the overall flowchart of the method for monitoring and diagnosing the condition of hydraulic supports in building construction in this invention; Figure 2 This is a flowchart of the low-reliability data classification, processing, and reconstruction process in this invention; Figure 3 This is a flowchart illustrating the dynamic adjustment of the construction phase identification and evaluation criteria in this invention. Figure 4 This is a flowchart of the abnormal state physical association classification and diagnosis process in this invention; Figure 5 This is a schematic diagram of the support plane arrangement in this invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0018] like Figure 1 As shown, the present invention provides a method for monitoring and diagnosing the condition of hydraulic supports in building construction, comprising the following steps: constructing a hydraulic support monitoring matrix, configuring multi-source sensors for each support unit and collecting pressure, displacement, height, tilt angle, and operating status data of each support to form an overall condition matrix; performing synchronous alignment processing on the multi-source monitoring data in the overall condition matrix under a time reference to obtain a standardized data sequence; evaluating the reliability of each channel data in the standardized data sequence and merging the evaluation results into a comprehensive reliability coefficient; based on the comparison result of the comprehensive reliability coefficient and a preset threshold value, dividing the data in the standardized data sequence into high-reliability data and low-reliability data; classifying and reconstructing the low-reliability data. The processed data is merged with high-reliability data to form a dataset for diagnosis. The current construction stage is automatically identified based on the temporal characteristics of the dataset for diagnosis, and the evaluation benchmark is dynamically adjusted accordingly. Multidimensional state fingerprints of each support are extracted from the dataset for diagnosis and matched with a pre-stored fingerprint template library to identify the current state category of each support. When the state category is determined to be abnormal, physical correlation analysis is performed on the parameter data of the same support extracted from the dataset for diagnosis to classify and diagnose the anomaly type. Based on the diagnosis results, adjacent support data are extracted from the overall state matrix for spatial comparison to determine the scope of influence and expansion trend of the anomaly. A diagnostic report is generated and output based on the comprehensive reliability coefficient, state category, classification diagnosis results, and expansion trend.
[0019] In this embodiment, constructing a hydraulic support monitoring matrix involves uniformly arranging all hydraulic support units within the construction area according to their spatial locations. Each support unit is assigned a unique number, which is then bound to its spatial coordinates. Multiple types of sensors are installed on each support unit to continuously collect data on the hydraulic pressure exerted on the support, the amount of expansion and contraction displacement, the absolute height change of the support, and the tilt angle of the support in two orthogonal directions within the horizontal plane. Simultaneously, the hydraulic pump station controller reads the hydraulic valve opening signals and pump operation status signals corresponding to each support. All monitoring data for the same support unit are aggregated into a single support state vector. The single support state vectors of all support units, arranged in the order of their spatial coordinates, together constitute the overall state matrix. This matrix records the number, spatial location, and corresponding multi-source monitoring data for each support, providing a fundamental data source for all subsequent analysis stages.
[0020] In this embodiment, the synchronization and alignment of multi-source monitoring data in the overall state matrix under a time reference means that each sensor independently collects data according to its own preset sampling frequency: the pressure sensor has a sampling frequency of 10 Hz, the displacement sensor has a sampling frequency of 5 Hz, and the altitude and tilt sensors both have a sampling frequency of 2 Hz. Since the sampling frequencies of each sensor are different, directly using unaligned raw data for correlation analysis would result in time misalignment. Therefore, the system uses the synchronization pulse signal emitted by the satellite timing module as a unified time reference. Data collected by each sensor is tagged with a time stamp accurate to the millisecond level upon arrival at the data acquisition card. Subsequently, resampling is performed according to a preset unified sampling time sequence. For sensor channels without original sampling values at the target time, the corresponding value at the target time is calculated by interpolation using the effective acquisition values of adjacent times for that channel. After the above processing, all data in the overall state matrix is mapped to the same set of time coordinates, forming a standardized data sequence after time synchronization, ensuring that data from sensors with different sampling frequencies are comparable at the same time.
[0021] In this embodiment, the reliability evaluation of each channel data in the standardized data sequence and the fusion into a comprehensive reliability coefficient refers to the system performing continuity evaluation, abrupt change evaluation, stability evaluation, and multi-parameter consistency evaluation on the data of each sensor channel in the standardized data sequence within each preset time window, resulting in four scores. The continuity score is determined based on the ratio of the number of valid data packets actually received by the channel within the time window to the theoretically expected number, while also checking for any "freezing" phenomena where values remain completely unchanged for multiple consecutive sampling periods. A higher data loss rate or a more severe freezing phenomenon results in a lower score. The abrupt change score is determined by comparing the parameter change rate at adjacent time points with the upper limit of the physically reasonable change rate dynamically set according to the current construction stage. A higher proportion of abrupt change points exceeding the threshold within the window results in a lower score. The stability score is determined based on the coefficient of variation of the data within the window—that is, the ratio of the standard deviation to the mean. A smaller coefficient of variation indicates more stable data and a higher score. The consistency score is determined based on the degree of matching between the physical response of pressure changes, displacement changes, and tilt angle changes of the same support. A higher score is achieved when displacement and tilt angle respond synchronously according to physical laws when pressure changes, and a lower score is achieved when pressure changes abruptly but other parameters do not respond. The four scores are weighted and summed according to preset weights to obtain a comprehensive credibility coefficient with a value between zero and one. The comprehensive credibility coefficients of all channels together constitute the credibility evaluation result set.
[0022] In this embodiment, the specific rules for dividing data in the standardized data sequence into high-confidence data and low-confidence data based on the comparison result of the comprehensive confidence coefficient and the preset threshold value are as follows: The preset threshold value is a fixed value between 0.6 and 0.8. The comprehensive confidence coefficient of each channel is read from the confidence evaluation result set and compared with the threshold value one by one. When the comprehensive confidence coefficient of a certain channel is greater than or equal to the threshold value, the monitoring data corresponding to that channel is extracted from the standardized data sequence, marked as high-confidence data, and completely retained, allowing it to enter the subsequent status identification and anomaly diagnosis stages. When the comprehensive confidence coefficient of a certain channel is less than the threshold value, the monitoring data corresponding to that channel is extracted from the standardized data sequence, marked as low-confidence data, prohibited from directly entering the diagnosis stage, and sent to the data reconstruction module for classification processing.
[0023] In this embodiment, classifying and reconstructing low-confidence data involves extracting time-series features from each channel's data and performing pattern recognition. When short-term data loss is identified, if the loss duration is within a preset compensation range, interpolation is performed using valid data before and after the loss. If the loss duration exceeds the compensation range, the time period is marked as invalid, and a communication anomaly alert is sent to management. When isolated mutations are identified, single-point, pulse-like deviations from the normal sequence are identified, the mutation point is removed, and the mean of the preceding and following times is used as a replacement. An additional data verification sampling is triggered to confirm that subsequent acquisitions have returned to normal. When long-term drift is identified, a trend deviation exhibiting monotonous, slow changes over a long time window with no corresponding response from other related channels is identified, marked as sensor drift, and a calibration reminder is generated without triggering an engineering safety alarm. When multi-parameter synchronous response is identified, it is confirmed that the pressure, displacement, and tilt data of the same support change synchronously and the direction of change conforms to physical laws. This group of data is identified as a real engineering anomaly signal, retained, and marked as engineering anomaly data. The processed data is merged with high-confidence data to form a dataset for diagnosis.
[0024] In this embodiment, the specific implementation method for automatically identifying the current construction stage based on the temporal characteristics of the dataset to be diagnosed and dynamically adjusting the evaluation benchmark accordingly is as follows: Pressure and displacement data of all supports are read from the dataset to be diagnosed, and the pressure change rate and displacement change of each support are calculated. When the average pressure change rate of all supports is less than 0.5 MPa per hour and this stable state lasts for more than two hours, it is determined to be the installation stage; when the pressure of each support rises synchronously and the average change rate exceeds 2 MPa per hour, while the displacement continues to increase, it is determined to be the concrete pouring stage; when the pressure change rate drops to below 0.2 MPa per hour and the hourly displacement change of each support is less than one millimeter, it is determined to be the curing stage; when the pressure of each support decreases synchronously at a rate exceeding 1 MPa per minute, it is determined to be the formwork removal stage. Based on the identified different construction stages, the system automatically adjusts the upper limit threshold of the physical reasonable change rate in subsequent evaluation stages and the weight coefficients of each feature component in the state fingerprint matching. The same monitoring value is assigned different engineering meanings and evaluation standards at different stages.
[0025] In this embodiment, the specific implementation of extracting multidimensional state fingerprints of each support from the dataset to be diagnosed and matching them with a pre-stored fingerprint template library to identify the state category is as follows: The pressure change rate, displacement change rate, and tilt angle change in two orthogonal directions of each support within a preset time window are extracted from the dataset to be diagnosed. The average pressure change rate of all adjacent supports within the same time window is read from the overall state matrix. These five feature components are combined to form the multidimensional state fingerprint of the support. During real-time matching, a weighted Euclidean distance calculation method is used. The current multidimensional state fingerprint is weighted and the feature vectors of each typical state template in the fingerprint template library are calculated one by one. The state mode corresponding to the closest template is selected as the diagnostic conclusion for the current state category of the support. The weight coefficients of each feature component are dynamically adjusted according to the currently identified construction stage. The fingerprint template library contains at least six typical state modes and their corresponding standard feature vectors: normal load-bearing, uniform loading, local overload, support settlement, hydraulic leakage, and sensor failure.
[0026] In this embodiment, when the state category is determined to be abnormal, the specific logic for extracting the parameter data of the same support from the dataset to be diagnosed and performing physical correlation analysis to classify and diagnose the abnormality type is as follows: The pressure, displacement, and tilt angle data of the support determined to be abnormal within the same time window are read from the dataset to be diagnosed, and their respective change directions and magnitudes are calculated. When a pressure decrease is detected while displacement increases and tilt angle changes, it is determined to be support foundation settlement or support deformation. The physical mechanism is that the foundation undergoes compression deformation under load, causing the support to sink. Part of the load is released by the foundation deformation, and the sinking causes the support to tilt. When a pressure decrease is detected while displacement and tilt angle remain unchanged, and the pressure data of adjacent supports read from the overall state matrix also show no response, it is determined to be hydraulic system leakage or actuator abnormality. The physical mechanism is that oil leakage in the hydraulic circuit causes a pressure decrease, but the mechanical position of the support has not changed. When a drastic change in data of a certain channel is detected while other channel data of the same support read from the dataset to be diagnosed show no response, it is determined that the sensor corresponding to that channel has malfunctioned. When an increase in the pressure of a support is detected while the pressure of adjacent supports read from the overall state matrix decreases synchronously, it is determined that a load redistribution has occurred within the support group. Various diagnostic conclusions and their confidence levels are output. When the confidence level is lower than a preset threshold, it is marked as a state awaiting review.
[0027] In this embodiment, the specific implementation method for extracting adjacent support data from the overall state matrix based on the diagnostic results and performing spatial comparison to determine the scope and expansion trend of the anomaly is as follows: The planar coordinates of all supports are read from the overall state matrix, the horizontal distance between any two supports is calculated, and support pairs with a distance less than a preset adjacency threshold are marked as spatially adjacent. Based on this, an adjacency relationship matrix containing all support adjacency information is constructed. After the classification diagnosis outputs the number of the abnormal support, a list of all adjacent support numbers for that abnormal support is obtained by querying the adjacency relationship matrix. The pressure and displacement data of these adjacent supports during the anomaly occurrence period are extracted from the overall state matrix, and their respective load change rate and displacement change rate are calculated. The calculation results are compared with the corresponding data of the abnormal support to determine whether the load has undergone spatial transfer and the direction and magnitude of the transfer. Subsequently, the analysis scope is expanded to all supports within the same construction zone, and their state data are extracted one by one from the overall state matrix to check whether the boundary of the abnormal area continues to expand outward. If the anomaly is limited to a single support, it is judged as a local anomaly; if it involves multiple adjacent supports, it is judged as a regional anomaly; if the scope of impact continues to expand over time, it is judged as an extended anomaly. The risk level is dynamically adjusted based on the judgment results, and the risk level and scope of impact information are written into the diagnostic report.
[0028] In this embodiment, the final diagnostic report includes the following structured information: the anomaly support number and its spatial location, current status description, anomaly type classification, data credibility assessment value, risk level classification, anomaly cause analysis, a list of related supports, development trend prediction, and handling recommendations. Through the above ten-step processing flow, this invention achieves a complete intelligent closed loop from data collection, credibility evaluation, stage identification, status matching, anomaly diagnosis to spatial tracing.
[0029] Example 2: As described in Example 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction involves constructing a hydraulic support monitoring matrix and forming an overall state matrix. This includes: assigning a unique number to each hydraulic support unit and binding its spatial coordinates; installing pressure sensors, displacement sensors, height sensors, and tilt sensors on each support unit to collect data on hydraulic pressure, support displacement, support height changes, and support tilt angles of each support; and converging all monitoring data of the same support unit into a single support state vector for that support. The single support state vectors of all support units are arranged according to their spatial coordinates to form the overall state matrix.
[0030] In this embodiment, the unique number is assigned based on the support layout diagram of the construction area. A permanent electronic code is established for each support unit according to row and column numbering or sequential numbering. This code cannot be changed during system operation. Spatial coordinate binding refers to measuring the precise planar position coordinates (X, Y) and elevation position Z of each support unit on the construction site using a total station or laser rangefinder, and storing the coordinate data in the system database in association with the number, establishing a one-to-one mapping relationship between the number and the spatial position.
[0031] In this embodiment, the pressure sensor is a strain gauge or piezoresistive hydraulic pressure sensor, with its range determined according to the design load of the support, typically 0 to 60 MPa, and is installed at the oil inlet of the support cylinder. The displacement sensor is a wire-type displacement sensor, with a range of 0 to 200 mm and an accuracy of 0.1 mm, installed between the support base and the movable end. The height sensor is an ultrasonic or laser rangefinder sensor, with a range of 0.5 to 10 meters, installed on the top of the support. The tilt sensor is a MEMS dual-axis tilt sensor, with a range of ±15 degrees and an accuracy of 0.01 degrees, installed in two orthogonal directions on the support cylinder.
[0032] In this embodiment, the single-support state vector is represented as: in, Indicates the first One support in Hydraulic pressure at all times Indicates support displacement. Indicates changes in support height. and They respectively represent the support at direction and The angle of inclination of the direction, This indicates the device's operating status code. (All) The individual support state vectors are arranged according to spatial coordinates to form the overall state matrix: Example 3: As in Example 1, a method for monitoring and diagnosing the status of hydraulic support in building construction includes the following steps for synchronizing and aligning multi-source monitoring data in the overall status matrix under a time reference: using a unified timing signal as the time reference, time-labeling the data of each sensor in the overall status matrix; resampling according to a preset unified sampling time sequence; for channels that do not have original sampling values at the target time, calculating the corresponding values at the target time using the effective values of adjacent times through interpolation and backfilling them into the overall status matrix to form a standardized data sequence.
[0033] In this embodiment, a unified timing signal is generated by a GPS or BeiDou satellite timing module. This module outputs a synchronization pulse signal every second. The system uses the rising edge of this pulse signal as the reference for the whole second. All sensor data are tagged with a time stamp accurate to the millisecond level when they arrive at the data acquisition card, ensuring that the alignment accuracy of the data of each channel on the time axis is better than ±0.1 seconds.
[0034] In this embodiment, the resampling process uses a linear interpolation method for the target time. If a certain sensor channel is in The original sampled value does not exist at any given time, but adjacent time values do exist. and ( The original sampled value is then used to calculate the corresponding value at the target time using the following formula: All interpolated values are backfilled into the corresponding positions in the overall state matrix, so that all data in the overall state matrix are evenly distributed in the time dimension, forming a standardized data sequence.
[0035] Example 4: As described in Example 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction involves evaluating the reliability of data from each channel in a standardized data sequence. The evaluation results are then integrated into a comprehensive reliability coefficient. This includes: evaluating the continuity, abrupt change, stability, and multi-parameter consistency of data from each channel in the standardized data sequence, resulting in four scores. The continuity score is determined based on the data packet reception integrity rate and whether the values are stuck for extended periods. The abrupt change score is determined based on a comparison of the rate of change between adjacent time points and the upper limit of a reasonable rate of change. The stability score is determined based on the coefficient of variation of data within a window. The consistency score is determined based on the degree of response matching between pressure changes, displacement changes, and tilt angle changes of the same support. The four scores are then weighted and summed to obtain the comprehensive reliability coefficient for each channel.
[0036] In this embodiment, continuous scoring The method for determining this is: statistical preset time window Inner Number of valid data packets actually received by each channel Compared with the theoretical number of data packets to be received The ratio, while detecting whether there is continuity. A stuck phenomenon where the value remains completely unchanged for each sampling period. The continuity score calculation formula is: in This is the pass rate threshold. If consecutive [items] appear within the window... More than one identical value, Reset directly to zero.
[0037] In this embodiment, mutation score The method for determining this is: calculate the rate of change of parameters at adjacent time points. ,in This is the sampling interval. If... ( If the maximum reasonable rate of change of the physical condition is dynamically set based on the current construction stage, then this point is marked as a mutation point. The percentage of mutation points within the window is... ,but: In this embodiment, stability score The method for determining this is: calculate the coefficient of variation of the data within the window. ,in The mean of the window. The standard deviation is given. The transformed scoring rules are as follows: In this embodiment, consistency scoring The method for determining this is as follows: calculate the Pearson correlation coefficient between pressure change and displacement change. And the correlation coefficient between pressure changes and tilt angle changes. : Similarly, calculate The consistency score is the larger of the two absolute values: In this embodiment, the overall credibility coefficient of each channel is obtained by weighted summation of the four scores. : Where the weight coefficients satisfy In this embodiment, we take , , , .
[0038] Example 5: As in Example 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction involves dividing data in a standardized data sequence into high-confidence data and low-confidence data based on the comparison results of the comprehensive confidence coefficient and the preset threshold value. Furthermore, the low-confidence data is categorized, processed, and reconstructed for reference. Figure 2 This includes: marking channel data whose comprehensive confidence coefficient reaches a threshold value as high-confidence data and retaining it; marking channel data that does not reach the threshold value as low-confidence data, extracting its time series features for pattern recognition, performing interpolation compensation when it is determined to be short-term data missing, removing it and replacing it with neighboring values when it is determined to be isolated mutation, marking it as sensor drift and generating a calibration reminder when it is determined to be long-term drift, and determining it as a real engineering anomaly when it is determined to be multi-parameter synchronous response and retaining it; merging the processed data with the high-confidence data into a dataset to be diagnosed.
[0039] In this embodiment, the overall credibility coefficient With preset threshold value The comparison rules are as follows: Set as an interval The fixed value between these two values is taken in this embodiment. .when At this time, the data in this channel is marked as high-confidence data and retained in its entirety, allowing direct entry into subsequent diagnostic stages. At that time, the data in this channel is marked as low-confidence data and is prohibited from directly entering the diagnostic process.
[0040] In this embodiment, the criterion for determining short-term data loss is: a certain channel has continuous... No valid data arrived within a sampling period, and (This embodiment takes) (seconds). The compensation method is linear interpolation: .like If the data for that period is invalid, a communication error message will be sent.
[0041] In this embodiment, the criterion for determining an isolated mutation is: the data value at a certain moment. satisfy However, the average of the data within the two seconds before and after is and The difference is no more than 2%. The solution is to remove the mutation point and replace it with the average of the time points before and after.
[0042] In this embodiment, the criterion for determining long-term drift is: within a long time window of 30 minutes, The data exhibits a monotonically slow changing trend (rate of change less than 0.01 MPa / s), but reading other channel data (displacement, tilt angle) of the same support and pressure data of adjacent supports from the overall state matrix shows no significant response (change amplitude less than 1%). The solution is to mark sensor drift and generate a calibration reminder.
[0043] In this embodiment, the criterion for determining multi-parameter synchronous response is: pressure drop, displacement increase, and tilt angle change occur simultaneously on the same support, and the direction of change conforms to physical laws (pressure drop is accompanied by displacement increase and tilt angle increase). The processing method is to retain this set of data and mark it as engineering abnormal data.
[0044] Example 6: As exemplified in Embodiment 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction automatically identifies the current construction stage based on the temporal characteristics of the dataset to be diagnosed and dynamically adjusts the evaluation benchmark accordingly. Figure 3 This includes: extracting the pressure change rate and displacement change of all supports from the dataset to be diagnosed for stage determination; determining the installation stage when the pressure change rate and displacement change of each support meet the preset installation stage characteristics, the pouring stage when they meet the pouring stage characteristics, the curing stage when they meet the curing stage characteristics, and the demolding stage when they meet the demolding stage characteristics; and adjusting the reasonable change rate threshold and feature weight coefficient of subsequent evaluation links based on the identified construction stage.
[0045] In this embodiment, the determination criterion for the installation phase is: the average pressure change rate of all supports. The pressure is measured in megapascals per hour, and this stable state lasts for more than 2 hours, while the displacement change of each support is less than 0.5 mm / hour. At this point, the system will set the upper limit of the abrupt change rate for subsequent evaluation stages. Set to 2 MPa / min.
[0046] In this embodiment, the criterion for determining the concrete pouring stage is: the pressure of each support increases synchronously and the average rate of change is... The pressure is measured in megapascals per hour, and simultaneously, the displacement sensor shows that the compression of each support is continuously increasing at a rate greater than 1 millimeter per hour. At this point, the system will... The pressure was increased to 5 MPa / min, and the weighting coefficient of the pressure change rate feature in state fingerprint matching was increased from 0.3 to 0.4.
[0047] In this embodiment, the criteria for determining the curing stage are: the pressure change rate of all supports drops below 0.2 MPa / hour, and the displacement change of each support is less than 1 mm within 1 hour. At this time, the system will... Reduce to 1 MPa / min and focus on the 24-hour pressure decay rate of each support level.
[0048] In this embodiment, the condition for determining the demolding stage is: the pressure of each support decreases synchronously at a rate exceeding 1 MPa / min. At this time, the system will... The pressure was set to 3 MPa / min, and the weighting coefficient of the displacement change rate characteristic was increased to focus on the synchronicity of the unloading process.
[0049] Example 7: As illustrated in Embodiment 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction involves extracting multidimensional condition fingerprints from the dataset to be diagnosed and matching them with a pre-stored fingerprint template library to identify condition categories. Furthermore, when a condition category is determined to be abnormal, physical correlation analysis is performed on the parameter data of the same support from the dataset to be diagnosed to classify and diagnose the anomaly type. Figure 4 The process includes: extracting the pressure change rate, displacement change rate, and tilt angle change of each support from the dataset to be diagnosed, and extracting the average pressure change rate of adjacent supports from the overall state matrix, combining them into a multidimensional state fingerprint; calculating the weighted distance between the multidimensional state fingerprint and each typical state template in the pre-stored fingerprint template library, and selecting the state category corresponding to the closest template as the diagnostic conclusion. The fingerprint template library contains at least standard templates corresponding to normal states and multiple abnormal states; when an abnormality is determined, reading the pressure, displacement, and tilt angle data of the support in the same time period from the dataset to be diagnosed, analyzing the direction and magnitude of change of each parameter, determining support settlement when pressure decreases accompanied by displacement increase and tilt angle change, determining hydraulic leakage when pressure decreases but displacement and tilt angle do not change and adjacent supports do not respond, determining sensor failure when single-channel data changes abruptly and other channels of the same support do not respond, and determining load transfer when the pressure of a support increases while the pressure of adjacent supports decreases.
[0050] In this embodiment, the multidimensional state fingerprint Its composition is as follows: in, For the first The normalized rate of change of pressure for each support, This is the normalized rate of change of displacement. and They are respectively Xianghe Change in tilt angle This represents the average rate of change of pressure across all adjacent supports.
[0051] In this embodiment, the formula for calculating the weighted Euclidean distance is: in For the first The weight coefficients of each feature component satisfy the following conditions: Each weight coefficient is dynamically adjusted according to the current construction stage. The state pattern corresponding to the template with the smallest distance is selected as the diagnostic conclusion, and the matching similarity is output simultaneously. .
[0052] In this embodiment, the typical state patterns and their corresponding feature vector templates pre-stored in the fingerprint template library are shown in the following table: In this embodiment, when the status fingerprint matching outputs an abnormal status category, the system reads the pressure data of the support within the same time period from the dataset to be diagnosed. Displacement data and tilt angle data Calculate the direction and magnitude of change for each parameter. The specific diagnostic logic is as follows: When detected and and At that time, it was determined to be support settlement, among which , , These are the threshold values for the changes in each parameter.
[0053] When detected and and And the rate of change of pressure of adjacent supports At that time, it was determined to be a hydraulic leak.
[0054] When the rate of change of data in a certain channel is detected (That is, the mutation amount exceeds 5 times the upper limit of the normal change rate) while the change rate of other channels of data in the same support If this occurs, it is determined that the sensor corresponding to that channel is faulty.
[0055] When a certain rate of change in support pressure is detected And the rate of change of pressure of adjacent supports When this occurs, it is determined to be a load transfer.
[0056] All diagnostic conclusions are accompanied by a confidence score. ,when The time stamp is marked as pending review and prompts for manual intervention for confirmation.
[0057] Example 8: As illustrated in Example 1, a method for monitoring and diagnosing the condition of hydraulic supports in building construction involves extracting adjacent support data from the overall condition matrix based on the diagnostic results for spatial comparison to determine the scope and expansion trend of the anomaly. (Refer to...) Figure 5 The process includes: reading the spatial coordinates of all supports from the overall state matrix and calculating the distances between them; marking support pairs with a distance less than a preset threshold as adjacent and establishing adjacency relationships; obtaining the numbers of abnormal supports based on the classification and diagnosis results; finding all adjacent supports of the abnormal support through adjacency relationships; extracting the state data of adjacent supports in the same time period from the overall state matrix; determining whether the load has undergone spatial transfer and the direction of transfer by comparing the magnitude ratio of the load change rate and the displacement change rate; expanding the analysis scope to all supports in the same construction area; determining whether the anomaly is local, regional, or widespread; and determining the risk level accordingly.
[0058] In this embodiment, the spatial adjacency relationship is established as follows: All states are read from the overall state matrix. Supporting planar coordinates Calculate any two supports and Horizontal distance between: like (This embodiment takes) (meters), then mark the support. With support For spatial adjacency, in the adjacency matrix Set the corresponding position to 1, otherwise set it to 0: In this embodiment, when the abnormal support number is When the system queries the adjacency matrix, the first... Okay, get the set of all adjacent support numbers. Extract from the overall state matrix All supports during the period of abnormal occurrence Using internal pressure and displacement data, calculate the load variation rate of each adjacent support. and displacement change rate : in By comparing the ratio of the load change rate of the abnormal support to that of the adjacent support. Determine the direction and magnitude of load transfer.
[0059] In this embodiment, the method for determining the expansion trend is as follows: the analysis scope is expanded from directly adjacent supports to all supports within the same construction zone, and the boundary of the abnormal area is checked to see if it continues to expand outward. If the number of supports affected by the anomaly... It was determined to be a localized anomaly, with a risk level of three. If Furthermore, since the affected area does not expand over time, it is classified as a regional anomaly with a risk level of Level 2. If Or the range of influence in terms of rate The number of [units] per hour continues to increase, which is determined to be an expansion anomaly, with a risk level of Level 1, requiring immediate action.
[0060] Example 9: A system for monitoring and diagnosing the condition of hydraulic supports in building construction includes: a multi-source data acquisition module deployed in each hydraulic support unit to collect multi-source monitoring data from each support and aggregate it into an overall condition matrix; a data preprocessing and time synchronization module to perform time alignment and standardization on the overall condition matrix and output a standardized data sequence; a data credibility evaluation module to perform multi-dimensional credibility evaluation on the standardized data sequence and output a comprehensive credibility coefficient for each channel; an anomaly data reconstruction module to classify and handle low-credibility data based on the comprehensive credibility coefficient and output a dataset to be diagnosed; a construction stage identification module to extract temporal features from the dataset to be diagnosed to automatically identify the current construction stage; a condition fingerprint matching and anomaly diagnosis module to extract multi-dimensional condition fingerprints from the dataset to be diagnosed and match them with a pre-stored fingerprint template library to identify condition categories, and perform physical correlation analysis when an anomaly is determined to be present to output the specific anomaly type; a spatial tracing and risk assessment module to analyze the anomaly propagation path and expansion trend based on the spatial coordinates in the overall condition matrix; and a human-computer interaction and output module to output a comprehensive report containing diagnostic conclusions and handling suggestions.
[0061] In this embodiment, the multi-source data acquisition module includes pressure sensors, wire-type displacement sensors, ultrasonic height sensors, MEMS dual-axis tilt sensors deployed in each hydraulic support unit, and an equipment status acquisition unit that reads hydraulic pump station and valve opening signals via a PLC interface. Each sensor independently collects data according to a preset sampling frequency and transmits the data to the field data aggregation terminal via wired or wireless means. After data packaging is completed at the field data aggregation terminal, the data is uploaded to the system server.
[0062] In this embodiment, the data preprocessing and time synchronization module includes a GPS or BeiDou satellite timing unit and a data resampling submodule. The timing unit outputs a synchronization pulse signal every second to ensure that the time stamp accuracy of the data in each channel is better than ±0.1 seconds. The resampling submodule performs linear interpolation resampling on the original data at a uniform sampling frequency of 10 Hz, and outputs a time-synchronized standardized data sequence.
[0063] In this embodiment, the data credibility evaluation module triggers an evaluation calculation once every second, performs four scoring calculations on the data of each channel of the standardized data sequence within the past 60-second time window, including continuity, mutability, stability and consistency, and sums them according to preset weights to output the comprehensive credibility coefficient of each channel. At the same time, the calculation results are stored in the system cache for subsequent modules to call.
[0064] In this embodiment, the abnormal data reconstruction module receives the comprehensive credibility coefficient output by the data credibility evaluation module, sends the channel data with a coefficient lower than the preset threshold value into the reconstruction processing queue, and performs corresponding reconstruction operations according to four categories: short-term missing data, isolated mutation, long-term drift, and multi-parameter synchronous response. After reconstruction, the processed data is merged with the high-credibility data, and the dataset to be diagnosed is output to the subsequent modules.
[0065] In this embodiment, the construction stage identification module extracts the pressure change rate and displacement change of all supports from the dataset to be diagnosed, performs stage identification according to the judgment conditions of installation stage, concrete pouring stage, curing stage and demolding stage, and outputs the identification results to the state fingerprint matching and anomaly diagnosis module in the form of stage code.
[0066] In this embodiment, the state fingerprint matching and anomaly diagnosis module has a built-in fingerprint template library, which is pre-built based on a large amount of engineering test data and numerical simulation results. The module extracts multi-dimensional state fingerprints of each support from the dataset to be diagnosed, and performs weighted distance matching with the template library using weighted coefficients adjusted according to the current construction stage, outputting the state category and its matching similarity. When the state category is determined to be abnormal, the module further reads the parameter data of each support from the dataset to be diagnosed for physical correlation analysis, and outputs the specific anomaly type and confidence level.
[0067] In this embodiment, the spatial tracing and risk assessment module constructs an adjacency relationship matrix based on the spatial coordinates in the overall state matrix. After receiving the abnormal support number, it obtains the set of adjacent supports through matrix query, extracts the state data of adjacent supports from the overall state matrix for spatial comparison, judges the impact range and expansion trend of the abnormality by comparing the spatial distribution characteristics of load change rate and displacement change rate, and outputs the risk level according to the impact range of the abnormality.
[0068] In this embodiment, the human-computer interaction and output module includes a graphical display interface and a report generation unit. The graphical display interface shows the status category and risk level of each support in real time in the form of a support distribution heatmap, with color indicating the location and impact range of abnormal supports. The report generation unit generates a structured diagnostic report according to a preset template, including the abnormal support number and spatial location, status diagnosis conclusion, anomaly classification and confidence level, credibility assessment value, risk level, related support list, development trend prediction and handling suggestions. The report is output in electronic document format and supports printing.
[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for monitoring and diagnosing abnormality of a state of a hydraulic support for construction, characterized by, Includes the following steps: A hydraulic support monitoring matrix is constructed, and multi-source sensors are configured for each support unit to collect data on pressure, displacement, height, tilt angle and operating status of each support, forming an overall status matrix; The multi-source monitoring data in the overall state matrix are synchronized and aligned under a time reference to obtain a standardized data sequence. The credibility of each channel data in the standardized data sequence is evaluated, and the evaluation results are integrated into a comprehensive credibility coefficient. Based on the comparison results between the comprehensive credibility coefficient and the preset threshold value, the data in the standardized data sequence is divided into high-credibility data and low-credibility data; Low-confidence data is classified, processed, and reconstructed, and the processed data is combined with high-confidence data to form a dataset to be diagnosed. The current construction stage is automatically identified based on the temporal characteristics of the dataset to be diagnosed, and the evaluation benchmark is dynamically adjusted accordingly. Extract multidimensional state fingerprints of each support from the dataset to be diagnosed, match them with a pre-stored fingerprint template library, and identify the current state category of each support; When the status category is determined to be abnormal, the same supporting parameter data is extracted from the dataset to be diagnosed for physical correlation analysis, and the abnormality type is classified and diagnosed. Based on the diagnostic results, adjacent support data are extracted from the overall state matrix for spatial comparison to determine the scope of impact and expansion trend of the anomaly. A diagnostic report is generated and output based on the comprehensive credibility coefficient, status category, classification diagnosis results, and expansion trend.
2. The method according to claim 1, wherein Constructing a hydraulic support monitoring matrix and forming an overall state matrix includes: Each hydraulic support unit is assigned a unique number and its spatial coordinates are bound; Pressure sensors, displacement sensors, height sensors, and tilt sensors are installed on each support unit to collect data on hydraulic pressure, support displacement, support height change, and support tilt angle of each support, respectively. All monitoring data from the same support unit are aggregated into a single support state vector for that support. The single support state vectors of all support units are arranged according to spatial coordinates to form the overall state matrix.
3. The method according to claim 1, wherein the method is characterized by, Synchronization and alignment of multi-source monitoring data in the overall state matrix under a time base includes: Using a unified timing signal as a time reference, time tags are added to the data from each sensor in the overall state matrix; Resampling is performed according to a pre-defined unified sampling time sequence. For channels that do not have original sampled values at the target time, the corresponding values at the target time are calculated by interpolation using the effective values at adjacent times and then backfilled into the overall state matrix to form a standardized data sequence.
4. The method of claim 1, wherein the method further comprises: The credibility of each channel data in the standardized data sequence is evaluated, and the evaluation results are integrated into a comprehensive credibility coefficient, including: The data in each channel of the standardized data sequence were evaluated for continuity, abruptness, stability, and multi-parameter consistency, resulting in four scores. The continuity score is determined based on the data packet reception integrity rate and whether the value is stuck for a long time; the abrupt change score is determined based on the comparison between the rate of change at adjacent time points and the upper limit of the reasonable rate of change; the stability score is determined based on the coefficient of variation of the data within the window; and the consistency score is determined based on the degree of response matching between pressure changes, displacement changes, and tilt angle changes of the same support. The overall credibility coefficient of each channel is obtained by weighted summation of the four scores.
5. The method of claim 1, wherein the method further comprises: Based on the comparison results between the comprehensive credibility coefficient and the preset threshold value, the data in the standardized data sequence is divided into high-credibility data and low-credibility data. The low-credibility data is then classified, processed, and reconstructed, including: Channel data whose comprehensive credibility coefficient reaches the threshold value are marked as high-credibility data and retained. Channel data that does not reach the threshold value is marked as low-confidence data. Its time series features are extracted for pattern recognition. When it is determined to be short-term data missing, interpolation compensation is performed. When it is determined to be isolated mutation, it is removed and replaced with neighboring values. When it is determined to be long-term drift, it is marked as sensor drift and a calibration reminder is generated. When it is determined to be multi-parameter synchronous response, it is determined to be a real engineering anomaly and is retained. The processed data is merged with the high-reliability data to form a dataset for diagnosis.
6. The method of claim 1, wherein the method further comprises: Automatically identify the current construction stage based on the temporal characteristics of the dataset to be diagnosed, and dynamically adjust the evaluation benchmark accordingly, including: Extract the pressure change rate and displacement change of all supports from the dataset to be diagnosed and determine the stage. When the pressure change rate and displacement change of each support meet the preset installation stage characteristics, it is determined to be the installation stage; when it meets the pouring stage characteristics, it is determined to be the pouring stage; when it meets the curing stage characteristics, it is determined to be the curing stage; when it meets the demolding stage characteristics, it is determined to be the demolding stage. Adjust the reasonable rate of change threshold and feature weight coefficients for subsequent evaluation stages based on the identified construction stages.
7. The method according to claim 1, wherein the method is characterized by, The process of extracting multidimensional state fingerprints for each support from the dataset to be diagnosed and matching them with a pre-stored fingerprint template library to identify state categories, and reading parameter data of the same support from the dataset to be diagnosed for physical correlation analysis to classify and diagnose the anomaly type when the state category is determined to be abnormal, includes: The pressure change rate, displacement change rate, and tilt angle change of each support are extracted from the dataset to be diagnosed, and the average pressure change rate of adjacent supports is extracted from the overall state matrix, and combined into a multi-dimensional state fingerprint. The multidimensional state fingerprint is weighted and the typical state templates in the pre-stored fingerprint template library are calculated. The state category corresponding to the closest template is selected as the diagnostic conclusion. The fingerprint template library contains at least standard templates corresponding to normal state and multiple abnormal states. When an anomaly is detected, the pressure, displacement, and tilt data of the support within the same time period are read from the standardized data sequence. The direction and magnitude of the change of each parameter are analyzed. When the pressure decreases and the displacement increases and the tilt changes, it is determined to be support settlement. When the pressure decreases but the displacement and tilt do not change and the adjacent support does not respond, it is determined to be hydraulic leakage. When the data of a single channel changes abruptly and other channels of the same support do not respond, it is determined to be sensor failure. When the pressure of a certain support increases and the pressure of the adjacent support decreases, it is determined to be load transfer.
8. The method of claim 1, wherein the method further comprises: Based on the diagnostic results, adjacent support data are extracted from the overall state matrix for spatial comparison to determine the scope and expansion trend of the anomaly, including: Read the spatial coordinates of all supports from the overall state matrix and calculate the distance between them. Mark the support pairs with a distance less than a preset threshold as adjacent and establish an adjacency relationship. After obtaining the abnormal support number based on the classification and diagnosis results, all adjacent supports of the support are found through adjacency relationship. The state data of adjacent supports in the same time period are extracted from the overall state matrix. By comparing the magnitude ratio of load change rate and displacement change rate, it is determined whether the load has undergone spatial transfer and the direction of transfer. Expand the analysis scope to all supports in the same construction area to determine whether the anomaly is localized, regional, or widespread, and determine the risk level accordingly.
9. A system for monitoring the state of a hydraulic support for construction and diagnosing abnormalities, for implementing the method according to any one of claims 1 to 8, characterized in that the system include: A multi-source data acquisition module is deployed in each hydraulic support unit to collect multi-source monitoring data from each support and aggregate it into an overall status matrix; The data preprocessing and time synchronization module is used to perform time alignment and standardization on the overall state matrix and output a standardized data sequence. The data credibility evaluation module is used to evaluate the credibility of standardized data sequences in multiple dimensions and output the comprehensive credibility coefficient of each channel. The abnormal data reconstruction module is used to classify and process low-confidence data based on the comprehensive confidence coefficient and output the dataset to be diagnosed. The construction phase identification module is used to extract time-series features from the dataset to be diagnosed and automatically identify the current construction phase. The state fingerprint matching and anomaly diagnosis module is used to extract multidimensional state fingerprints from the dataset to be diagnosed and match them with a pre-stored fingerprint template library to identify the state category. When an anomaly is determined, physical correlation analysis is performed to output the specific anomaly type. The spatial tracing and risk assessment module is used to analyze the anomaly propagation path and expansion trend based on the spatial coordinates in the overall state matrix. The human-computer interaction and output module is used to output a comprehensive report that includes diagnostic conclusions and treatment recommendations.
