Concrete beam stress intelligent monitoring method, device and equipment and storage medium

Abstract

The invention provides a concrete beam stress intelligent monitoring method and device, equipment and a storage medium. The method comprises the steps that concrete beam monitoring data are collected in real time through a sensor; analyzing data by using a deep learning model, and outputting overload and crack state probability values; after the two are converted into basic probability distribution, D-S evidence theory fusion is executed, an evidence conflict value is detected, and comprehensive probability distribution is generated through a conflict resolution rule; and triggering graded early warning accordingly. According to the application, sensors, deep learning and a D-S evidence theory are integrated, real-time accurate monitoring and intelligent early warning of stress of the concrete beam are realized, visual decision support is combined, the safety management efficiency and reliability of the structure are effectively improved, and the risk of safety accidents is reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of beam stress monitoring, and particularly relates to a concrete beam stress intelligent monitoring method, device, equipment and storage medium. BACKGROUND

[0002] In the daily operation of pier wharf, the safety and stability of prestressed concrete beams as the key load-bearing structure directly relate to the operation efficiency and safety of the entire wharf. However, with the continuous growth of water transportation volume and the increase of wharf service life, these concrete beams are facing increasingly severe tests.

[0003] Traditionally, the monitoring of prestressed concrete beams mainly relies on various types of sensors such as stress sensors, strain sensors and displacement sensors, which can collect various data of the beam body under different stress states. However, the data format, frequency and precision collected by these sensors differ significantly, making data integration and utilization difficult. More seriously, historical monitoring data is often stored in different databases, lacking a systematic integration mechanism, resulting in low data utilization. When processing these data, staff not only need to manually perform a large amount of raw data processing work, but also are prone to data loss, error matching and other problems, seriously affecting the efficiency and accuracy of monitoring. In addition, due to the lack of unified processing standards, multi-source heterogeneous data cannot be directly used for structural health analysis, further limiting the intelligent level of the monitoring system.

[0004] In terms of structural damage identification, existing methods mostly rely on manual inspection and experience judgment, making it difficult to detect early cracks, stress concentration and other subtle damages in real time and accurately. This situation easily leads to missed detection and misjudgment, making it difficult to discover and handle structural hazards in a timely manner, thereby increasing the risk of safety accidents. At the same time, in the face of identified damage problems, how to develop a reasonable maintenance plan is also a problem. Traditional methods require manual review of a large number of specifications and cases, the process is tedious and the adaptability and effectiveness of the plan are difficult to guarantee, making it impossible to achieve scientific and efficient maintenance management. In addition, monitoring data and maintenance information are mainly transmitted through manual recording and reporting, lacking intuitive and visual display means. This makes it difficult for managers to comprehensively and real-time grasp the structural health status and maintenance progress, and the decision-making lacks data support, seriously affecting the safety and economy of wharf operation. SUMMARY

[0005] The application aims to overcome the defects in the prior art and provide a concrete beam stress intelligent monitoring method, device, equipment and storage medium.

[0006] The application provides a concrete beam stress intelligent monitoring method, comprising:

[0007] Collect real-time monitoring data of the concrete beam through a sensor;

[0008] Analyze the real-time monitoring data based on a deep learning model, and output an overload state probability value and a crack state probability value;

[0009] Convert the overload state probability value into a first basic probability assignment, and convert the crack state probability value into a second basic probability assignment;

[0010] Perform a D-S evidence theory fusion operation on the first basic probability assignment and the second basic probability assignment, including: detecting an evidence conflict value between the first basic probability assignment and the second basic probability assignment; based on the evidence conflict value, fusing the first basic probability assignment and the second basic probability assignment through a conflict resolution rule to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a hierarchical early warning signal according to a preset numerical interval.

[0011] Optionally, the collecting real-time monitoring data of the concrete beam through a sensor comprises:

[0012] Deploying a strain sensor at a mid-span position of the concrete beam;

[0013] Deploying a crack width sensor at a support position of the concrete beam;

[0014] Collecting beam strain data through the strain sensor;

[0015] Collecting crack propagation data through the crack width sensor;

[0016] Taking the beam strain data and the crack propagation data as real-time monitoring data.

[0017] Optionally, the analyzing the real-time monitoring data based on a deep learning model comprises:

[0018] Constructing a CNN-LSTM hybrid neural network;

[0019] Inputting the real-time monitoring data into the CNN-LSTM hybrid neural network;

[0020] Extracting spatial feature distribution through a convolutional layer;

[0021] Capturing temporal feature changes through a long short-term memory layer;

[0022] Outputting an overload state probability value and a crack state probability value.

[0023] Optionally, the converting the overload state probability value into a first basic probability assignment comprises:

[0024] Mapping the overload state probability value into a basic probability assignment of an overload event;

[0025] The converting the crack state probability value into a second basic probability assignment comprises: mapping the crack state probability value into a basic probability assignment of a crack event.

[0026] Optionally, the detecting an evidence conflict value between the first basic probability assignment and the second basic probability assignment comprises:

[0027] calculating an intersection empty event pair of the first basic probability assignment and the second basic probability assignment;

[0028] for the intersection empty event pair, calculating a sum of probability assignment products as an evidence conflict value.

[0029] Optionally, the fusing the first basic probability assignment and the second basic probability assignment by a conflict resolution rule based on the evidence conflict value comprises:

[0030] calculating a sum of probability assignment products of non-conflict events;

[0031] dividing the sum of probability assignment products of the non-conflict events by a normalization factor, the normalization factor being 1 minus the evidence conflict value, to generate a comprehensive probability distribution.

[0032] Optionally, the triggering a hierarchical early warning signal according to a preset numerical interval based on the comprehensive probability distribution comprises:

[0033] when the comprehensive probability distribution value is in an interval [0.4, 0.6], triggering a yellow early warning signal;

[0034] when the comprehensive probability distribution value is in an interval (0.6, 0.8], triggering an orange early warning signal;

[0035] when the comprehensive probability distribution value is greater than 0.8, triggering a red early warning signal.

[0036] The application further provides a concrete beam stress intelligent monitoring device, comprising:

[0037] a monitoring module, which collects real-time monitoring data of a concrete beam through a sensor;

[0038] an analysis module, which analyzes the real-time monitoring data based on a deep learning model and outputs an overload state probability value and a crack state probability value;

[0039] a conversion module, which converts the overload state probability value into a first basic probability assignment and converts the crack state probability value into a second basic probability assignment;

[0040] The fusion module performs a D-S evidence theory fusion operation on the first basic probability assignment and the second basic probability assignment, including: detecting an evidence conflict value between the first basic probability assignment and the second basic probability assignment; based on the evidence conflict value, fusing the first basic probability assignment and the second basic probability assignment through a conflict resolution rule to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a hierarchical early warning signal according to a preset numerical interval.

[0041] The application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method.

[0042] The application further provides a computer readable storage medium, which stores a computer program, and the computer program, when executed in a computer, causes the computer to execute the method.

[0043] The application has the following beneficial effects:

[0044] The application provides a concrete beam stress intelligent monitoring method, comprising: collecting real-time monitoring data of a concrete beam through a sensor; analyzing the real-time monitoring data based on a deep learning model to output an overload state probability value and a crack state probability value; converting the overload state probability value into a first basic probability assignment and converting the crack state probability value into a second basic probability assignment; performing a D-S evidence theory fusion operation on the first basic probability assignment and the second basic probability assignment, including: detecting an evidence conflict value between the first basic probability assignment and the second basic probability assignment; based on the evidence conflict value, fusing the first basic probability assignment and the second basic probability assignment through a conflict resolution rule to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a hierarchical early warning signal according to a preset numerical interval. The application realizes real-time and accurate monitoring, intelligent early warning and visual decision support of the concrete beam stress by integrating the sensor, the deep learning and the D-S evidence theory, effectively improves the structural safety management efficiency and reliability, and reduces the risk of safety accidents. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 FIG. 1 is a schematic diagram of a concrete beam stress intelligent monitoring process in the application. DETAILED DESCRIPTION

[0046] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that various forms implement the present disclosure and should not be limited by the embodiments set forth herein. On the contrary, the embodiments are provided to enable a more thorough understanding of the present disclosure and to convey the scope of the present disclosure to those skilled in the art.

[0047] Referring to Figure 1 The present application provides a concrete beam stress intelligent monitoring method, based on deep learning technology, realizes the intelligent monitoring of the stress state of the pier type wharf prestressed concrete beam. The method comprises the steps of real-time data acquisition, deep learning analysis, probability distribution transformation, D-S evidence theory fusion and hierarchical early warning triggering.

[0048] S101, collecting real-time monitoring data of the concrete beam by the sensor;

[0049] The acquisition process of real-time monitoring data ensures that the data comprehensively reflects the stress state of the beam body.

[0050] Through the sensor deployed at the key position of the beam body, the data is collected in real time at a frequency of 10Hz. The sensor includes: a strain sensor deployed at the mid-span position of the concrete beam, used to capture beam strain data; a crack width sensor deployed at the support position of the concrete beam, used to monitor crack propagation data. The crack width sensor adopts the crack width sensor produced by Tokyo Measuring Instrument Co., Ltd. (model: PD-2021), with a resolution of 0.01mm and a measurement accuracy of ±0.05mm; the strain sensor is a fiber grating surface strain gauge, with a range of ±2500με, an accuracy level of 0.1 level, and a sampling frequency set to 10Hz. Both types of sensors are deployed on the surface of the concrete beam after IP68 waterproof sealing treatment, and the strain sensor at the mid-span position needs to be pasted parallel to the main reinforcement to ensure cooperative deformation with the beam body.

[0051] Strain data is collected by a fiber grating surface strain gauge, which is pasted on the surface and internal main reinforcement of the beam body, with the characteristics of anti-electromagnetic interference and high precision, and can stably monitor strain distribution. Crack width data is obtained in real time by a crack width sensor, which accurately measures crack size changes.

[0052] The collected data also includes vertical load force and horizontal load force, but the real-time monitoring data focuses on beam strain data and crack propagation data, providing a basis for subsequent analysis.

[0053] The collected data needs to be preprocessed: first, detect outliers, and use the 3σ principle to process abnormal data. Specifically, if the stress data at a certain time exceeds 3 times the standard deviation of its mean value, it is determined as an outlier, and the weighted average value of adjacent time points is used to replace it.

[0054] Subsequently, the data is standardized by using the normalization formula:

[0055]

[0056] wherein, represents the normalized data value, x represents the original data value, Min value of training set data, Max value of training set data. This step ensures that the data format is consistent with the deep learning model training, eliminating the influence of dimensional differences. The pretreated beam strain data and crack propagation data are input as real-time monitoring data into the next stage.

[0057] S102, based on the deep learning model, analyzing the real-time monitoring data, outputting overload state probability value and crack state probability value;

[0058] Using a deep learning model to process real-time monitoring data, outputting a probabilistic analysis result.

[0059] The deep learning model uses a CNN-LSTM hybrid neural network architecture to adapt to the complex features of multi-source data. Model construction includes: the convolutional neural network (CNN) part contains 3 layers of convolutional layers, with a convolution kernel size of 3x3 and a step size of 1, used to extract the spatial feature distribution between different monitoring points of the prestressed concrete beam, such as the distribution of strain data in the beam structure; The long short-term memory (LSTM) part is set to 128 memory units, which captures the time series feature changes through the gating mechanism, such as the periodic fluctuations of load data. After dimensionality reduction of the features, the global average pooling layer is connected to the fully connected layer, and the Softmax activation function is used to output the probability value.

[0060] The CNN-LSTM hybrid neural network input layer time step is set to 60 sampling points, containing 3 layers of convolutional layers (the number of convolutional kernels in each layer is 32, 64, and 128 respectively), and the convolutional kernel size is 3x1; The number of hidden units in the LSTM layer is set to 128, and the dropout rate is set to 0.25 to avoid overfitting. The optimizer for training this network is Adam, the initial learning rate is set to 0.001, the batch size is 32, the number of training rounds is 200, and the size of the max pooling layer connected after the convolutional layer is set to 2x1 to compress the feature dimension.

[0061] The analysis process is as follows: input the pretreated real-time monitoring data into the CNN-LSTM hybrid neural network. Extract spatial feature distribution through convolutional layer, and mine local patterns in data; capture time series feature changes through LSTM layer, such as the rate trend of crack propagation. The model output includes overload state probability value and crack state probability value. The overload state probability value is output based on the classification task, and the Softmax function is used to calculate:

[0062]

[0063] Where K is the total number of categories, including "normal", "minor damage", "severe damage", "overload", etc. represents the probability output value of the jth category; represents the unactivated raw output value in the model output layer corresponding to the jth class; represents the unactivated raw output value in the output layer for the kth class; e is a natural constant for exponential transformation mapping to a positive number range.

[0064] The crack state probability value is also calculated by this formula, but for the crack-related class. The output result directly reflects the possibility of the abnormal state of the beam body.

[0065] S103, convert the overload state probability value into a first basic probability distribution, and convert the crack state probability value into a second basic probability distribution;

[0066] The conversion of probability values to probability distributions is for the preparation of evidence fusion. The conversion process is based on the principle of direct mapping: the overload state probability value is converted into the first basic probability distribution, representing the degree of trust in the overload event; the crack state probability value is converted into the second basic probability distribution, representing the degree of trust in the crack event.

[0067] The specific operation is: if the model predicts the overload state probability value to be 0.7, the first basic probability distribution is m({overload})=0.7, that is, the support degree of evidence 1 to the "overload" proposition; similarly, if the crack state probability value is 0.6, the second basic probability distribution is m({crack})=0.6, indicating the support degree of evidence 2 to the "crack" proposition. This conversion ensures that the probability result can be directly used in D-S evidence theory, preserving the uncertainty information.

[0068] S104, performing D-S evidence theory fusion operation on the first basic probability distribution and the second basic probability distribution, including: detecting the evidence conflict value between the first basic probability distribution and the second basic probability distribution; based on the evidence conflict value, fusing the first basic probability distribution and the second basic probability distribution through a conflict resolution rule to generate a comprehensive probability distribution.

[0069] By D-S evidence theory, multiple evidences are fused to generate a robust comprehensive probability.

[0070] The fusion operation is divided into two steps:

[0071] First, detect the evidence conflict value. Calculate the intersection of the first basic probability distribution and the second basic probability distribution as the event pair with empty intersection, that is, when the propositions of the two evidences have no overlap. For example, if evidence 1 supports "overload" and evidence 2 supports "crack", and the intersection is empty, then calculate the sum of the product of the probability distributions of these event pairs as the evidence conflict value. The formula is:

[0072]

[0073] Where A and B are different evidence propositions, for example, A can be "overload" and B can be "crack"; is the first basic probability assignment value, representing the support degree of evidence 1 to proposition A; is the second basic probability assignment value, representing the support degree of evidence 2 to proposition B; represents a pair of propositions whose intersection is empty. The above formula is used for summation operation to calculate the conflict coefficient.

[0074] Secondly, based on the evidence conflict value, the probability assignment is fused by the conflict resolution rule. The sum of the probability assignment products of non-conflict events is calculated, that is, when the proposition intersection of two evidences is not empty. The formula is:

[0075]

[0076] Where C is the integrated proposition after fusion, for example, "dangerous state of beam body"; A∩B=C represents a pair of propositions whose intersection is equal to proposition C. The above formula is used for summation operation. Then, the non-conflict is divided by the normalization factor (1 minus the conflict value) to generate the integrated probability distribution:

[0077]

[0078] Where m(c) represents the integrated probability distribution value about the final state of the beam body; the normalization factor in the denominator is used to eliminate the influence of evidence conflict. This fusion enhances the reliability of decision-making, for example, when the overload and crack evidences conflict, the system can still output consistent integrated probability.

[0079] S105, based on the integrated probability distribution, triggering a hierarchical early warning signal according to a preset numerical interval.

[0080] According to the integrated probability value, a multi-level early warning is triggered to realize risk grading response.

[0081] The early warning mechanism is divided into three levels: when the integrated probability distribution value is in the interval [0.4, 0.6], a yellow early warning signal is triggered, indicating that the potential risk is low, and the system suggests increasing the monitoring frequency; when the integrated probability distribution value is in the interval (0.6, 0.8], an orange early warning signal is triggered, indicating a medium risk, which requires professional personnel to check in detail; when the integrated probability distribution value is greater than 0.8, a red early warning signal is triggered, indicating a high risk, which requires immediate comprehensive inspection and emergency repair. The early warning signal is sent through multiple channels such as SMS, email and sound and light alarm to ensure that the staff respond in time. At the same time, the system records the warning time, the integrated probability value and the processing result, which is used for subsequent model optimization.

[0082] The application also provides a concrete beam stress intelligent monitoring device, comprising:

[0083] The monitoring module collects real-time monitoring data of the concrete beam through the sensor;

[0084] The analysis module analyzes the real-time monitoring data based on a deep learning model, and outputs an overload state probability value and a crack state probability value;

[0085] The conversion module converts the overload state probability value into a first basic probability assignment and converts the crack state probability value into a second basic probability assignment;

[0086] The fusion module performs a D-S evidence theory fusion operation on the first basic probability assignment and the second basic probability assignment, including: detecting an evidence conflict value between the first basic probability assignment and the second basic probability assignment; based on the evidence conflict value, fusing the first basic probability assignment and the second basic probability assignment through a conflict resolution rule to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a hierarchical early warning signal according to a preset numerical interval.

[0087] The application also provides an electronic device, including a memory and a processor, the memory stores a computer program, and the processor executes the computer program to realize the method described above.

[0088] The application also provides a computer readable storage medium, which stores a computer program, and when the computer program is executed in a computer, the computer executes the method described above.

[0089] The above description of the embodiments is to facilitate the understanding and application of the application by those skilled in the art. Those skilled in the art will obviously make various modifications to the above embodiments, and apply the general principles described herein to other embodiments without having to go through creative labor. Therefore, the application is not limited to the above embodiments, and those skilled in the art should make improvements and modifications to the application within the scope of the application.

Claims

1. A method for intelligent monitoring of concrete beam stress, characterized in that: include: Collect real-time monitoring data of concrete beams through sensors; Analyze the real-time monitoring data based on a deep learning model and output an overload state probability value and a crack state probability value; Converting the overload state probability value into a first basic probability distribution, and converting the crack state probability value into a second basic probability distribution; Performing a DS evidence theory fusion operation on the first basic probability distribution and the second basic probability distribution includes: detecting an evidence conflict value between the first basic probability distribution and the second basic probability distribution; based on the evidence conflict value, fusing the first basic probability distribution and the second basic probability distribution through conflict resolution rules to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a graded warning signal according to a preset numerical range.

2. The method according to claim 1, characterized in that The real-time monitoring data of the concrete beam is collected by the sensor, including: Deploy strain sensors at the mid-span of the concrete beam; Deploy crack width sensors at concrete beam supports; Collecting beam body strain data through the strain sensor; Collecting crack extension data by means of the crack width sensor; The beam body strain data and the crack extension data are used as real-time monitoring data.

3. The method according to claim 1, characterized in that The analyzing the real-time monitoring data based on the deep learning model includes: Construct a CNN-LSTM hybrid neural network; Inputting the real-time monitoring data into the CNN-LSTM hybrid neural network; Extract spatial feature distribution through convolutional layers; Capture temporal feature changes through long short-term memory layers; Output the overload state probability value and crack state probability value.

4. The method according to claim 1, wherein The converting the overload state probability value into a first basic probability distribution includes: Mapping the overload state probability value to the basic probability distribution of overload events; The converting the fracture state probability value into a second basic probability distribution includes: mapping the fracture state probability value into a basic probability distribution of a fracture event.

5. The method according to claim 1, wherein The detecting the evidence conflict value between the first basic probability distribution and the second basic probability distribution includes: Calculating event pairs for which the intersection of the first basic probability distribution and the second basic probability distribution is empty; For event pairs whose intersection is empty, the sum of their probability distribution products is calculated as the evidence conflict value.

6. The method according to claim 1, characterized in that The fusing the first basic probability distribution and the second basic probability distribution by conflict resolution rules based on the evidence conflict value includes: Calculate the sum of the products of the probability distributions of non-conflicting events; The sum of the products of the probability distributions of the non-conflicting events is divided by a normalization factor, where the normalization factor is 1 minus the evidence conflict value, to generate a comprehensive probability distribution.

7. The method according to claim 1, characterized in that The triggering of a graded warning signal according to a preset numerical range based on the comprehensive probability distribution includes: When the comprehensive probability distribution value is in the interval [0.4, 0.6], a yellow warning signal is triggered; When the comprehensive probability distribution value is in the range of (0.6, 0.8], an orange warning signal is triggered; When the comprehensive probability distribution value is greater than 0.8, a red warning signal is triggered.

8. An intelligent monitoring device for concrete beam stress, characterized in that: include: Monitoring module, which collects real-time monitoring data of concrete beams through sensors; An analysis module, which analyzes the real-time monitoring data based on a deep learning model and outputs an overload state probability value and a crack state probability value; a conversion module, converting the overload state probability value into a first basic probability distribution, and converting the crack state probability value into a second basic probability distribution; A fusion module performs a DS evidence theory fusion operation on the first basic probability distribution and the second basic probability distribution, including: detecting an evidence conflict value between the first basic probability distribution and the second basic probability distribution; based on the evidence conflict value, fusing the first basic probability distribution and the second basic probability distribution through conflict resolution rules to generate a comprehensive probability distribution; based on the comprehensive probability distribution, triggering a graded warning signal according to a preset numerical range.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

Images