Steel enclosing purlin stress real-time monitoring method based on Internet of Things and storage medium

Abstract

The invention relates to the technical field of stress measurement, in particular to a steel enclosing purlin stress real-time monitoring method based on the Internet of Things and a storage medium, and the method comprises the steps: obtaining the local neighborhood similarity of stress values of different stress monitoring points on a steel enclosing purlin at different moments; the method is used for classifying the stress values of all monitoring points at all moments before the current moment, obtaining a plurality of sets, merging the stress values of all the stress monitoring points at the current moment into each set, obtaining a plurality of update sets and the abnormal probability of the stress value of each stress monitoring point at the current moment, and calculating the abnormal probability of the stress values of all the stress monitoring points at the current moment. And performing data stripping on the update set to obtain a plurality of secondary update sets so as to obtain the abnormal probability of the stress value of each stress monitoring point at the next moment at the current moment, thereby judging whether to send out an early warning prompt or not. According to the method, the abnormal data can be accurately identified through comparison between the real-time data and the historical data.

Description

Technical Field

[0001] The present invention relates to the technical field of stress measurement, and in particular to a real-time monitoring method and storage medium for steel purlin stress based on the Internet of Things. Background Art

[0002] Steel purlins, also known as steel fences or steel fences, are typically constructed by splicing two or more sections of the same type and length. They are typically composed of steel plates and sections. In foundation pit support projects, steel purlins serve as key components connecting support piles to steel supports. Their stress state directly affects foundation pit stability, making real-time monitoring of steel purlin stress crucial. Furthermore, to meet rapid response requirements such as monitoring and early warning, real-time monitoring based on IoT technology is essential.

[0003] Existing Problem: For real-time stress monitoring of steel purlins, existing technology primarily relies on comparing real-time data with historical data. However, stress on steel purlins exhibits varying distributions, resulting in multiple distribution characteristics in historical data, making direct comparative analysis ineffective. Furthermore, as construction progresses, the stress behavior of steel purlins changes, and thus the data characteristics change. This discrepancy between the characteristics of real-time stress data and historical data can lead to erroneous analysis and judgment of real-time monitoring data. Summary of the Invention

[0004] The present invention provides a real-time monitoring method and storage medium for steel purlin stress based on the Internet of Things to solve existing problems.

[0005] The present invention's Internet of Things-based real-time monitoring method for steel purlin stress and storage medium adopt the following technical solutions:

[0006] One embodiment of the present invention provides a real-time monitoring method for steel purlin stress based on the Internet of Things, the method comprising the following steps:

[0007] Obtain the position characteristic sequence of each stress monitoring point on the steel purlin, the sequence of neighboring stress monitoring points, and the stress value at each moment;

[0008] Based on the differences in the positional feature sequences of different stress monitoring points and the differences in the stress values ​​of the neighboring stress monitoring point sequences at different times, the local neighborhood similarity of the stress values ​​of different stress monitoring points at different times is determined, and the stress values ​​of all monitoring points at all times before the current time are classified to obtain several sets.

[0009] According to the magnitude of the local neighborhood similarity, the stress values ​​of all stress monitoring points at the current moment are merged into respective sets, and a plurality of updated sets and an abnormal probability of the stress value of each stress monitoring point at the current moment are obtained;

[0010] According to the difference between stress values ​​in the update set and the difference between the stress values ​​at corresponding moments and the local neighborhood similarity between the stress values, combined with the abnormal probability, the update set is subjected to data stripping to obtain several secondary update sets;

[0011] According to the method for obtaining the abnormal probability of the stress value at the current moment, based on the secondary update set and the stress value of each stress monitoring point at the next moment after the current moment, the abnormal probability of the stress value of each stress monitoring point at the next moment after the current moment is obtained to determine whether to issue an early warning prompt.

[0012] Furthermore, the determination of the local neighborhood similarity of stress values ​​of different stress monitoring points at different times includes the following specific steps:

[0013] Obtain a sequence A consisting of stress values ​​of all stress monitoring points in a neighborhood stress monitoring point sequence of stress monitoring point a at the i-th moment, and a sequence B consisting of stress values ​​of all stress monitoring points in a neighborhood stress monitoring point sequence of stress monitoring point b at the x-th moment;

[0014] Determine, based on the difference between the sequence A and the sequence B, the similarity S of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment;

[0015] In the position feature sequence of stress monitoring points a and b, the absolute value of the difference of the position features under the same ordinal value is calculated, and the mean ΔW of the absolute value of the difference of the position features under all the same ordinal values ​​is obtained. The ratio of the similarity S to the mean ΔW is recorded as the local neighborhood similarity of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment.

[0016] Furthermore, the method of determining the similarity S of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment based on the difference between the sequence A and the sequence B includes the following specific steps:

[0017] In sequence A and sequence B, the absolute value of the difference between the stress values ​​under the same ordinal value is calculated, and the mean ΔH of the absolute value of the difference between the stress values ​​under all the same ordinal values ​​is obtained. Then, the Pearson correlation coefficient between sequence A and sequence B is obtained, and the ratio of the normalized value of the Pearson correlation coefficient to the mean ΔH is recorded as the similarity S of the stress values ​​of stress monitoring point a at the i-th moment and stress monitoring point b at the x-th moment.

[0018] Furthermore, the specific steps of obtaining the multiple update sets and the abnormal probability of the stress value of each stress monitoring point at the current moment are as follows:

[0019] Determining the merging of the stress value of each stress monitoring point at the jth moment with each set according to the magnitude of the local neighborhood similarity;

[0020] For all stress monitoring points, obtain a set corresponding to the stress value of each stress monitoring point at the jth moment and the maximum value in the mergeability of all sets, merge the stress value of each stress monitoring point at the jth moment into the set corresponding to the maximum value, and obtain several updated sets;

[0021] Obtain the update set in which the stress value of the y-th stress monitoring point at the j-th moment is located, record it as the target update set, and take the inversely proportional normalized value of the merging property of the stress value of the y-th stress monitoring point at the j-th moment and the target update set as the abnormal probability of the stress value of the y-th stress monitoring point at the j-th moment; the j-th moment is the current moment.

[0022] Furthermore, the method of determining the compatibility of the stress value of each stress monitoring point at the jth moment with each set according to the magnitude of the local neighborhood similarity includes the following specific steps:

[0023] A threshold value N is preset. In any set, the mean of the local neighborhood similarity between each stress value and all other stress values ​​is obtained as the consistency of each stress value. The top N stress values ​​with the largest consistency are recorded as feature data.

[0024] The mean of the local neighborhood similarities between the stress value of the y-th stress monitoring point at the j-th moment and all characteristic data in the t-th set is obtained as the merging property of the stress value of the y-th stress monitoring point at the j-th moment and the t-th set.

[0025] Furthermore, the obtaining of several secondary update sets includes the following specific steps:

[0026] The probability of whether to perform data stripping on each update set is determined based on the difference between the stress values ​​in each update set and the difference between the stress values ​​at the corresponding moments, as well as the local neighborhood similarity between the stress values.

[0027] The update set whose probability of data stripping is greater than a preset probability threshold is recorded as a stripping update set;

[0028] The update set whose probability of data stripping is less than or equal to the preset probability threshold is recorded as the secondary update set;

[0029] Determine the probability of each stress value being stripped based on the time interval between the corresponding moment of each stress value and the corresponding moments of all other stress values ​​in each stripping update set, the local neighborhood similarity between each stress value and all other stress values, and the abnormal probability of each stress value;

[0030] In any stripping update set, the stress values ​​whose stripping probability is greater than a preset stripping threshold are deleted to obtain a secondary update set.

[0031] Furthermore, the method of determining the probability of whether to perform data stripping on each update set based on the difference between the stress values ​​in each update set and the difference between the stress values ​​at corresponding moments and the local neighborhood similarity between the stress values ​​includes the following specific steps:

[0032] In any update set, obtain the normalized value of the variance σ1 of all stress values, obtain the normalized value of the variance σ2 of all stress values ​​at the corresponding moment, and then obtain the inversely proportional normalized value of the mean G of the local neighborhood similarity of all any two stress values, calculate the sum of the normalized value of the variance σ1 and the normalized value of the variance σ2, and record the normalized value of the product of the sum and the inversely proportional normalized value of the mean G as the probability of whether data stripping is performed in any update set.

[0033] Furthermore, the stripping probability of each stress value is determined based on the time interval between the corresponding moment of each stress value and the corresponding moments of all other stress values ​​in each stripping update set, the local neighborhood similarity between each stress value and all other stress values, and the abnormal probability of each stress value, including the following specific steps:

[0034] In any stripping update set, obtain the mean of the time intervals between the moment corresponding to the zth stress value and the moments corresponding to all other stress values, recorded as the first mean, obtain the inversely proportional normalized value of the mean M of the local neighborhood similarity between the zth stress value and all other stress values, and use the normalized value of the product of the first mean, the inversely proportional normalized value of the mean M, and the abnormal probability of the zth stress value as the stripping probability of the zth stress value.

[0035] Furthermore, the determination of whether to issue an early warning prompt includes the following specific steps:

[0036] When the abnormal probability of the stress value of any stress monitoring point at any moment is greater than the preset abnormal threshold, an early warning prompt is issued.

[0037] The present invention also proposes a storage medium for real-time monitoring of steel purlin stress based on the Internet of Things, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned method for real-time monitoring of steel purlin stress based on the Internet of Things.

[0038] The beneficial effects of the technical solution of the present invention are:

[0039] In an embodiment of the present invention, the local neighborhood similarity of stress values ​​at different stress monitoring points on a steel purlin at different times is obtained to classify the stress values ​​of all monitoring points at all times before the current time, and obtain several sets. Here, the local performance of the stress distribution of the steel purlin is taken into account, so the local distribution relationship of the monitoring data is used for comparative analysis. At the same time, considering the distribution forms of the steel purlin stress, such as concentrated distribution and uniform distribution, the historical data is integrated into sets to facilitate the refined comparison of real-time data and improve the accuracy of the comparative analysis. The stress values ​​of all stress monitoring points at the current time are merged into each set to obtain several updated sets and the abnormal probability of the stress value of each stress monitoring point at the current time. The updated sets are then stripped to obtain several secondary updated sets. Here, the changes in the stress distribution form of the steel purlin caused by the advancement of the construction progress are taken into account, and the data in the historical data set is stripped, so that the set data more accurately represents the normal distribution of the current steel purlin stress, and effectively reduces the amount of data involved in the calculation, thereby improving the calculation speed. The abnormal probability of the stress value of each stress monitoring point at the next moment is obtained to determine whether to issue an early warning. Thus, the present invention accurately identifies abnormal data by comparing real-time data with historical data. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 This is a flowchart of the steps of the real-time monitoring method of steel purlin stress based on the Internet of Things of the present invention;

[0042] Figure 2 Schematic diagram of steel purlin. DETAILED DESCRIPTION

[0043] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of the IoT-based real-time stress monitoring method and storage medium for steel purlins proposed by the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0045] The following describes in detail the method for real-time monitoring of steel purlin stress and the specific solution of the storage medium based on the Internet of Things provided by the present invention with reference to the accompanying drawings.

[0046] See also Figure 1 , which shows a flowchart of a method for real-time monitoring of steel purlin stress based on the Internet of Things provided by one embodiment of the present invention, the method comprising the following steps:

[0047] Step S001: Obtain the position feature sequence of each stress monitoring point on the steel purlin, the neighborhood stress monitoring point sequence, and the stress value at each moment.

[0048] What needs to be explained is that steel purlins are key supporting components in fields such as foundation pit support and bridge engineering. Their structural design is based on high-strength steel and has the characteristics of modularity and splicing. The specific components are as follows: (1) Main structural materials: (1.1) Steel type: Commonly used I-beams (such as I20a, I45b), H-beams or channel steels, with "I" or "C" cross-sections, use the high bending stiffness of steel to withstand horizontal loads. (1.2) Combination form: Single or multiple steel sections are spliced ​​into a combined section (such as double I-beams) through steel plates and bolts to enhance the overall bending and torsional resistance. (2) Connection node structure: (2.1) Connection with support piles: Fixed to the outside of the support piles (such as bored piles, steel sheet piles) by pre-buried steel plates, bolts or welding, and stiffening ribs are set at the nodes to prevent local deformation. (2.2) Connection with steel supports: Weld corbels or flanges on the purlins and rigidly connect them to horizontal steel supports (such as steel pipes, H-shaped steel) through bolts to form a "support pile-purlin-support" force system. (2.3) Splicing nodes: Steel plates are used to splice multiple purlins. The bolt hole spacing is arranged according to the design specifications to ensure force continuity. Schematic diagram of steel purlins, as shown Figure 2 shown.

[0049] Obtain the position characteristic sequence of each stress monitoring point on the steel purlin, the sequence of neighboring stress monitoring points, and the stress value at each moment.

[0050] It should be noted that when monitoring the stress of steel purlins in real time, it is first necessary to use sensors to collect the stress of steel purlins in real time. Generally, the stress monitoring of steel purlins is long-term monitoring, so a high-stability surface strain gauge is selected as the sensor for collecting data. The installation process is as follows: (1) The steel purlins mainly support the side walls of the foundation pit, and the steel purlins need to be spliced ​​and welded. The sensor installation position is near the steel purlins at the corners of the foundation pit, the steel purlin interfaces, and the support points. The maximum distance between steel purlin sensors on the same horizontal plane is 3 meters. (2) Sensor inspection: calibrate the zero point of the strain gauge, confirm that the cable insulation is good, and the matching reading instrument functions normally. (3) Surface treatment: Polish the installation position of the steel purlin, remove rust and oil stains, ensure the surface is flat (roughness ≤ 200 microns), and mark the installation center point. (4) Mechanical installation of the sensor: (4.1) Drill the mounting hole at the marked point (the hole diameter matches the strain gauge base, and the depth is based on the surface of the base after being embedded and flush with the steel purlin). The hole position deviation is ≤2 mm. (4.2) Use high-strength bolts (such as 8.8 grade bolts) to fasten the strain gauge base to the surface of the steel purlin, ensuring that the base fits tightly with the steel purlin and the pre-tightening force is uniform (it can be controlled by a torque wrench, and the torque value is based on the design requirements, generally ≥50 Newton meters (N·m)). (4.3) Insert the strain gauge body into the base slot, adjust the direction so that the strain gauge axis is consistent with the main force direction of the steel purlin (deviation ≤5 degrees), and lock it with a fixing bolt. At the same time, check whether the cable route is smooth to avoid entanglement or stretching. (4.4) After installation, apply waterproof glue (such as silicone rubber) to the connection between the strain gauge and the base and the cable interface, and wrap it with waterproof tape to seal it. Finally, install a metal protective sleeve (such as a corrugated pipe) to cover the cable to prevent damage during construction collision.

[0051] It should be further explained that after the sensor is installed, real-time stress data is collected. The data collection system includes: (1) hardware equipment: surface stress gauge; data acquisition instrument (such as vibrating wire reader, strain acquisition instrument), which supports real-time reading and storage; communication module (such as 4G, WiFi, optical fiber), which is used to remotely transmit data to the monitoring platform. (2) Collection method: Automated real-time collection, the acquisition instrument automatically reads the stress gauge data at a set frequency (such as every 10 minutes). The data is transmitted to the server via 4G or optical fiber. (3) Data recording: Record the stress value collected at each stress monitoring point at each moment. For each stress monitoring point, the monitoring position is recorded, including the shortest distance to the steel purlin interface, the shortest distance to the steel purlin support point, and the horizontal height. The three position data constitute the position feature sequence of the stress monitoring point. In this embodiment, the minimum and maximum standard method is used to normalize the data of each dimension in the position feature sequence to unify the dimension. This is a well-known technology and the specific method will not be introduced here. The stress monitoring points on the same steel purlin are distributed horizontally in a circular pattern. In this case, the neighborhood of a single stress monitoring point is defined as the adjacent monitoring points in the horizontal direction. In this embodiment, a first threshold value of 6 is preset. For this example, the six nearest stress monitoring points to each stress monitoring point in the horizontal direction of each steel purlin are obtained to form a neighborhood stress monitoring point sequence for each stress monitoring point.

[0052] Step S002: Based on the differences in the positional feature sequences of different stress monitoring points and the differences in the stress values ​​of the neighborhood stress monitoring point sequences at different times, the local neighborhood similarities of the stress values ​​of different stress monitoring points at different times are determined, so as to classify the stress values ​​of all monitoring points at all times before the current time and obtain several sets.

[0053] It should be noted that the stress distribution of steel purlins is primarily characterized by uniform and concentrated distribution, so stress monitoring primarily relies on the distribution of stress within a local area. However, normal stress distribution is inherently distributed across multiple locations and time points within the steel purlin, meaning that the monitoring data contains numerous similar stress distribution relationships. Real-time monitoring of steel purlin stress can also be performed by constructing distribution relationships between monitoring points and analyzing the consistency between the corresponding distribution relationships within the monitored data.

[0054] It should be further explained that stress monitoring of steel purlins primarily focuses on local stress distribution. The resulting local neighborhoods can represent the local distribution of stress. Local areas with similar distributions are primarily characterized by similarity between stresses in local neighborhoods and similarity between distribution relationships among monitoring points within the local area. The greater the difference in stress values ​​among monitoring points in a local neighborhood, the less similar the stresses are among them; and the more consistent the stress distribution trends among monitoring points in a local neighborhood, the greater the similarity among the stresses in the local neighborhood.

[0055] Preferably, in one embodiment of the present invention, the method for obtaining the local neighborhood similarity and set of stress values ​​of different stress monitoring points at different times includes:

[0056] Taking any two moments, i.e., moment i and moment x, as an example, for any two stress monitoring points a and b, obtain sequence A consisting of the stress values ​​of all stress monitoring points in the stress monitoring point a's neighborhood stress monitoring point sequence at moment i, and sequence B consisting of the stress values ​​of all stress monitoring points in the stress monitoring point b's neighborhood stress monitoring point sequence at moment x. In sequence A and sequence B, calculate the absolute value of the difference between the stress values ​​under the same ordinal value, obtain the mean ΔH of the absolute value of the difference between the stress values ​​under all the same ordinal values, then obtain the Pearson correlation coefficient between sequence A and sequence B, and record the ratio of the normalized value of the Pearson correlation coefficient to the mean ΔH as the similarity S of the stress values ​​between stress monitoring point a at moment i and stress monitoring point b at moment x.

[0057] It should be noted that the Pearson correlation coefficient is a well-known technique, and the specific method is not described here. The value range of the Pearson correlation coefficient is between -1 and 1. Therefore, in this embodiment, the ratio of the difference between 1 and the Pearson correlation coefficient and 2 is used as the normalized value of the Pearson correlation coefficient. Furthermore, the smaller the difference in the positional characteristics of two monitoring points, the higher the similarity of the stress distribution in the corresponding local neighborhood.

[0058] In the position feature sequence of any two stress monitoring points a and b, the absolute value of the difference between the position features under the same ordinal value is calculated, and the mean ΔW of the absolute value of the difference between the position features under all the same ordinal values ​​is obtained. The ratio of the similarity S of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment to the mean ΔW is recorded as the local neighborhood similarity of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment.

[0059] The current moment is recorded as the jth moment.

[0060] The stress values ​​of all monitoring points at all times before the jth moment are used as input (excluding the jth moment). According to the local neighborhood similarity of any two stress values, the Louvain algorithm is used for classification and several sets are output.

[0061] It should be noted that since the current stress monitoring of the steel purlin is real-time, the current calculation is limited to the data from the same monitoring session. That is, the above calculation is performed on the data collected from the start of monitoring to obtain the true distribution of the current steel purlin stress. In this case, the time range is set to 30 minutes, that is, 30 minutes of stress data are collected to obtain the true distribution of the current steel purlin stress. Therefore, in this embodiment, the jth moment is the 40th minute after the start of data collection. Since the data collection frequency is 10 minutes, the data for the first 30 minutes after the start of data collection is classified. This example is used for description. A threshold is directly used to determine whether the data for the first 30 minutes after the start of data collection is abnormal. Specifically, the design stress of the steel purlin is obtained from the technical data sheet provided by the material supplier. The ratio of the stress value to the design stress is recorded as the abnormal probability of the stress value. The judgment threshold is set to 0.95. For this example, when the abnormal probability is greater than 0.95, it is determined to be abnormal and an alarm is issued. The following analysis is performed on the data collected 30 minutes after the data collection begins to determine the anomaly, that is, starting from the 40th minute after the data collection begins, subsequent real-time anomaly judgment is performed. Due to the uncontrollable construction environment, the stress magnitude and stress distribution of steel purlins at different positions or heights are different. This difference value is real and allowed, so there are different types of real stress distributions. At this time, it is necessary to determine multiple distribution sets based on the similarity between the local neighborhoods of the monitoring points. Among them, the Louvain algorithm (Louvain Method for Community Detection) is a well-known technology, and the specific method will not be introduced here. The local neighborhood similarity between all stress values ​​in the divided set is relatively large.

[0062] Step S003: Based on the magnitude of the local neighborhood similarity, the stress values ​​of all stress monitoring points at the current moment are merged into various sets, and several updated sets and the abnormal probability of the stress value of each stress monitoring point at the current moment are obtained.

[0063] What needs to be explained is: Based on the above analysis, the actual multiple stress distribution sets in the current steel purlin stress monitoring are determined. Then in the subsequent real-time monitoring, the real-time stress data needs to be classified based on the consistent relationship between the real-time stress data and the existing stress distribution set. That is, the real-time stress data is merged into the existing set. The above-mentioned determined set contains multiple monitoring data, and there are also differences between the monitoring data. When merging the real-time monitored stress data into the existing set, it is necessary to judge the consistent relationship between the real-time monitored stress data and the existing set, so it is necessary to integrate the different monitoring data in the set and determine the characteristic performance of the set. Because the data characteristics of the set contain multiple dimensional features, direct calculation is more complicated, so you can select feature data from several and use the feature performance of the feature data as the feature performance of the current set.

[0064] Preferably, in one embodiment of the present invention, the method for obtaining the abnormal probability of the stress value of the update set and each stress monitoring point at the current moment includes:

[0065] The preset quantity threshold N is 5, and this is used as an example for description.

[0066] In any set, the mean of the local neighborhood similarities between each stress value and all other stress values ​​is obtained as the consistency of each stress value, and the top N stress values ​​with the greatest consistency are recorded as feature data. Feature data is used to represent the characteristics of the corresponding set.

[0067] It should be noted that during real-time monitoring of steel purlin stress, the collected data requires real-time processing. Based on the obtained set and its characteristic representation, the feature consistency relationship between the real-time stress data and the set is analyzed, and the real-time data is merged into the set. The similarity between the real-time stress data and the set's characteristic data directly reflects the merging relationship between the real-time stress data and the set. The similarity between the real-time stress data and the set's characteristic data is calculated in the local neighborhood between the real-time stress data and multiple characteristic data in a set to determine the similarity between the real-time stress data and the set.

[0068] The mean of the local neighborhood similarities between the stress value of the y-th stress monitoring point at the j-th moment and all characteristic data in the t-th set is obtained as the merging property of the stress value of the y-th stress monitoring point at the j-th moment and the t-th set.

[0069] Obtain the set corresponding to the stress value of the y-th stress monitoring point at the j-th moment and the maximum value among all sets of mergeability, record it as the target set, and merge the stress value of the y-th stress monitoring point at the j-th moment into the target set. In this way, the stress values ​​of all stress monitoring points at the j-th moment are sequentially merged into the corresponding set with the maximum mergeability, and several updated sets are obtained.

[0070] It should be noted that after the real-time monitoring data is merged into the corresponding set, the anomaly probability of the real-time monitoring data is calculated. The higher the similarity between the real-time monitoring data and the merged set, the lower the anomaly probability.

[0071] Obtain the update set in which the stress value of the y-th stress monitoring point at the j-th moment is located, record it as the target update set, and take the inversely proportional normalized value of the merging degree F of the stress value of the y-th stress monitoring point at the j-th moment and the target update set as the abnormal probability of the stress value of the y-th stress monitoring point at the j-th moment.

[0072] It should be noted that: in this embodiment, according to the compatibility of the stress value of the y-th stress monitoring point at the j-th moment with the t-th set, the compatibility F of the stress value of the y-th stress monitoring point at the j-th moment with the target update set is obtained, and 1-norm(F) is used as the inverse proportional normalization value of F, where norm() is a linear normalization function used to normalize the data value to between 0 and 1.

[0073] Step S004: Based on the stress value difference and the stress value corresponding time difference in the update set and the local neighborhood similarity between the stress values, combined with the abnormal probability, data stripping is performed on the update set to obtain several secondary update sets.

[0074] It should be noted that as monitoring progresses, the amount of data contained in the system becomes larger and larger, and the real-time monitoring process requires continuous call of aggregate data, which in turn reduces the computing speed. Furthermore, as monitoring continues, the stress distribution of the steel purlins continues to change as the progress of foundation pit construction progresses. Therefore, during the monitoring process, the data distribution in the aggregate has a large deviation, and some data in the aggregate cannot fully represent the distribution of the currently monitored stress data. Therefore, during the real-time stress monitoring process, it is necessary to consider the time difference of stress distribution and strip the data in the aggregate to ensure that the aggregate can represent the distribution of real-time stress data.

[0075] It should be further explained that when stripping the data in the set, the distribution deviation in the set is taken into consideration. However, not every stress monitoring and monitoring data merger will cause a distribution deviation in the set. Therefore, after each monitoring data merger, the deviation of the data characteristics in the set is used to determine whether to perform data stripping. The greater the difference in the data characteristics in the set, the more necessary it is to perform data stripping. The difference in data characteristics is mainly manifested in: the size of the stress value in the set, the collection time corresponding to the stress data, and the difference in the local neighborhood of the monitoring data. That is, the smaller the local neighborhood similarity, the greater the local neighborhood difference.

[0076] Preferably, in one embodiment of the present invention, the method for obtaining the secondary update set includes:

[0077] In any update set, obtain the normalized value of the variance σ1 of all stress values, obtain the normalized value of the variance σ2 of all stress values ​​at the corresponding moment, and then obtain the inversely proportional normalized value of the mean G of the local neighborhood similarity of all any two stress values. Calculate the sum of the normalized value of the variance σ1 and the normalized value of the variance σ2, and record the normalized value of the product U of the sum and the inversely proportional normalized value of the mean G as the probability of whether data stripping is performed in any update set.

[0078] It should be noted that: in this embodiment, norm(σ1), norm(σ2) and norm(U) are used as the normalized values ​​of σ1, σ2 and U respectively, and 1-norm(G) is used as the inversely proportional normalized value of G, where norm() is a linear normalization function used to normalize data values ​​to between 0 and 1.

[0079] The preset probability threshold is 0.8, which is used as an example for description.

[0080] An update set whose probability of data stripping is greater than a preset probability threshold is recorded as a stripping update set.

[0081] An update set for which the probability of data stripping is less than or equal to a preset probability threshold is recorded as a secondary update set, ie, data stripping is not performed.

[0082] It should be noted that the data that is stripped is primarily data that does not represent the stress distribution characteristics of the steel purlins for the current time period. This is specifically manifested in the following ways: the monitoring time interval between a single data point and the rest of the data set, and the local neighborhood difference between a single data point and the rest of the data set. The longer the monitoring time interval between a single data point and the rest of the data set, the greater the probability of being stripped; the greater the local neighborhood difference between a single data point and the rest of the data set, the greater the probability of being stripped; and the greater the probability of anomaly in a single data point, the greater the probability of being stripped.

[0083] In any stripping update set, obtain the mean of the time intervals between the moment corresponding to the z-th stress value and the moments corresponding to all other stress values, recorded as the first mean, obtain the inversely proportional normalized value of the mean M of the local neighborhood similarity between the z-th stress value and all other stress values, and use the normalized value of the product K of the first mean, the inversely proportional normalized value of the mean M, and the abnormal probability of the z-th stress value as the stripping probability of the z-th stress value.

[0084] It should be noted that: in this embodiment, norm(K) is used as the normalized value of K, and 1-norm(M) is used as the inversely proportional normalized value of M, where norm() is a linear normalization function used to normalize data values ​​to between 0 and 1.

[0085] The preset peeling threshold is 0.7, which is used as an example for description.

[0086] In any stripping update set, the stress values ​​whose stripping probability is greater than a preset stripping threshold are deleted to obtain a secondary update set.

[0087] Step S005: According to the method for obtaining the abnormal probability of the stress value at the current moment, based on the secondary update set and the stress value of each stress monitoring point at the next moment after the current moment, the abnormal probability of the stress value of each stress monitoring point at the next moment after the current moment is obtained to determine whether to issue an early warning prompt.

[0088] Preferably, in one embodiment of the present invention, the method for determining whether to issue an early warning prompt includes:

[0089] According to the method for obtaining the abnormal probability of the stress value of each stress monitoring point at the jth moment, the abnormal probability of the stress value of each stress monitoring point at the j+1th moment is obtained according to the secondary update set and the stress value of each stress monitoring point at the j+1th moment.

[0090] Specifically, the stress value of each stress monitoring point at the jth moment is incorporated into the set, and the stress value of each stress monitoring point at the j+1th moment is incorporated into the secondary update set, so as to obtain the abnormal probability of the stress value of each stress monitoring point at the j+1th moment (the next moment of the current moment).

[0091] Therefore, according to the above method, the set after data stripping of the secondary update set is obtained to obtain the abnormal probability of the stress value of each stress monitoring point at the j+2th moment (the next moment after the current moment), and so on, to obtain the abnormal probability of the real-time stress value of each stress monitoring point.

[0092] The default abnormal threshold is 0.75, which is used as an example for description.

[0093] When the abnormal probability of the stress value of any stress monitoring point at any moment is greater than the preset abnormal threshold, an early warning prompt is issued to remind the staff to conduct an inspection.

[0094] It should be noted that in the real-time monitoring of steel purlin stress, the storage scheme for data collection, merging and feature screening processes must take into account real-time performance, reliability and data processing efficiency. Based on this, in this embodiment, a layered storage mode of "edge computing node + cloud database" is adopted to achieve data preprocessing and core data persistence: (1) Edge node cache (real-time processing layer): (1.1) Storage medium: Use SSD solid-state drive or memory database (such as Redis) to support high-speed reading and writing. (1.2) Storage logic: Temporarily store the raw data collected in real time, sort by timestamp to form a data stream, and when new data is connected, directly merge it into the real-time data set (such as hash table or time series queue) through memory operation to avoid disk IO delay. (2) Cloud database (persistence layer): (2.1) Database selection: Select a time series database (InfluxDB or TimescaleDB), which is optimized for time series data, supports efficient writing and querying of massive real-time data, and is suitable for storing stress data with timestamps. (2.2) Storage Strategy: The edge node writes the cached valid data to the cloud in batches, excluding all data from the cached data set, every 24 hours to reduce network transmission pressure. These operations are well-known and the specific methods are not described here. In this embodiment, when calculating the ratio, if the denominator is 0, it is set to 1 to ensure that the ratio holds. This example is used for description.

[0095] The present invention also provides a storage medium for real-time monitoring of steel purlin stress based on the Internet of Things, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. The processor executes the computer program stored in the memory to implement the steps of the aforementioned method for real-time monitoring of steel purlin stress based on the Internet of Things.

[0096] So far, the present invention is completed.

[0097] In summary, in an embodiment of the present invention, the position feature sequence of each stress monitoring point on the steel purlin and the sequence of neighborhood stress monitoring points and the stress value at each moment are obtained, and the local neighborhood similarity of the stress values ​​of different stress monitoring points at different moments is determined to classify the stress values ​​of all monitoring points at all moments before the current moment, obtain several sets, merge the stress values ​​of all stress monitoring points at the current moment into each set, obtain several update sets and the abnormal probability of the stress value of each stress monitoring point at the current moment, then perform data stripping on the update sets, obtain several secondary update sets, and thus obtain the abnormal probability of the stress value of each stress monitoring point at the next moment after the current moment, so as to determine whether to issue an early warning prompt. The present invention accurately identifies abnormal data by comparing real-time data with historical data.

[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A real-time monitoring method for steel purlin stress based on the Internet of Things, characterized in that: The method comprises the following steps: Obtain the position characteristic sequence of each stress monitoring point on the steel purlin, the sequence of neighboring stress monitoring points, and the stress value at each moment; Based on the differences in the positional feature sequences of different stress monitoring points and the differences in the stress values ​​of the neighboring stress monitoring point sequences at different times, the local neighborhood similarity of the stress values ​​of different stress monitoring points at different times is determined, and the stress values ​​of all monitoring points at all times before the current time are classified to obtain several sets. According to the magnitude of the local neighborhood similarity, the stress values ​​of all stress monitoring points at the current moment are merged into respective sets, and a plurality of updated sets and an abnormal probability of the stress value of each stress monitoring point at the current moment are obtained; According to the difference between stress values ​​in the update set and the difference between the stress values ​​at corresponding moments and the local neighborhood similarity between the stress values, combined with the abnormal probability, the update set is subjected to data stripping to obtain several secondary update sets; According to the method for obtaining the abnormal probability of the stress value at the current moment, based on the secondary update set and the stress value of each stress monitoring point at the next moment after the current moment, the abnormal probability of the stress value of each stress monitoring point at the next moment after the current moment is obtained to determine whether to issue an early warning prompt.

2. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 1 is characterized in that: The specific steps of determining the local neighborhood similarity of stress values ​​of different stress monitoring points at different times are as follows: Obtain a sequence A consisting of stress values ​​of all stress monitoring points in a neighborhood stress monitoring point sequence of stress monitoring point a at the i-th moment, and a sequence B consisting of stress values ​​of all stress monitoring points in a neighborhood stress monitoring point sequence of stress monitoring point b at the x-th moment; Determine, based on the difference between the sequence A and the sequence B, the similarity S of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment; In the position feature sequence of stress monitoring points a and b, the absolute value of the difference of the position features under the same ordinal value is calculated, and the mean ΔW of the absolute value of the difference of the position features under all the same ordinal values ​​is obtained. The ratio of the similarity S to the mean ΔW is recorded as the local neighborhood similarity of the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment.

3. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 2 is characterized in that: The determining, based on the difference between the sequence A and the sequence B, of the similarity S between the stress values ​​of the stress monitoring point a at the i-th moment and the stress monitoring point b at the x-th moment comprises the following specific steps: In sequence A and sequence B, the absolute value of the difference between the stress values ​​under the same ordinal value is calculated, and the mean ΔH of the absolute value of the difference between the stress values ​​under all the same ordinal values ​​is obtained. Then, the Pearson correlation coefficient between sequence A and sequence B is obtained, and the ratio of the normalized value of the Pearson correlation coefficient to the mean ΔH is recorded as the similarity S of the stress values ​​of stress monitoring point a at the i-th moment and stress monitoring point b at the x-th moment.

4. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 1 is characterized in that: The specific steps of obtaining the multiple update sets and the abnormal probability of the stress value of each stress monitoring point at the current moment are as follows: Determining the merging of the stress value of each stress monitoring point at the jth moment with each set according to the magnitude of the local neighborhood similarity; For all stress monitoring points, obtain a set corresponding to the stress value of each stress monitoring point at the jth moment and the maximum value in the mergeability of all sets, merge the stress value of each stress monitoring point at the jth moment into the set corresponding to the maximum value, and obtain several updated sets; Obtain the update set in which the stress value of the y-th stress monitoring point at the j-th moment is located, record it as the target update set, and take the inversely proportional normalized value of the merging property of the stress value of the y-th stress monitoring point at the j-th moment and the target update set as the abnormal probability of the stress value of the y-th stress monitoring point at the j-th moment; the j-th moment is the current moment.

5. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 4 is characterized in that: The specific steps of determining the merging of the stress value of each stress monitoring point at the jth moment with each set according to the magnitude of the local neighborhood similarity are as follows: A threshold value N is preset. In any set, the mean of the local neighborhood similarity between each stress value and all other stress values ​​is obtained as the consistency of each stress value. The top N stress values ​​with the largest consistency are recorded as feature data. The mean of the local neighborhood similarities between the stress value of the y-th stress monitoring point at the j-th moment and all characteristic data in the t-th set is obtained as the merging property of the stress value of the y-th stress monitoring point at the j-th moment and the t-th set.

6. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 1 is characterized in that: The specific steps of obtaining several secondary update sets are as follows: The probability of whether to perform data stripping on each update set is determined based on the difference between the stress values ​​in each update set and the difference between the stress values ​​at the corresponding moments, as well as the local neighborhood similarity between the stress values. The update set whose probability of data stripping is greater than a preset probability threshold is recorded as a stripping update set; The update set whose probability of data stripping is less than or equal to the preset probability threshold is recorded as the secondary update set; Determine the probability of each stress value being stripped based on the time interval between the corresponding moment of each stress value and the corresponding moments of all other stress values ​​in each stripping update set, the local neighborhood similarity between each stress value and all other stress values, and the abnormal probability of each stress value; In any stripping update set, the stress values ​​whose stripping probability is greater than a preset stripping threshold are deleted to obtain a secondary update set.

7. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 6 is characterized in that: The method of determining the probability of whether to perform data stripping on each update set based on the difference between the stress values ​​in each update set and the difference between the stress values ​​at corresponding moments and the local neighborhood similarity between the stress values ​​includes the following specific steps: In any update set, obtain the normalized value of the variance σ1 of all stress values, obtain the normalized value of the variance σ2 of all stress values ​​at the corresponding moment, and then obtain the inversely proportional normalized value of the mean G of the local neighborhood similarity of all any two stress values, calculate the sum of the normalized value of the variance σ1 and the normalized value of the variance σ2, and record the normalized value of the product of the sum and the inversely proportional normalized value of the mean G as the probability of whether data stripping is performed in any update set.

8. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 6 is characterized in that: The stripping probability of each stress value is determined based on the time interval between the corresponding moment of each stress value and the corresponding moments of all other stress values ​​in each stripping update set, the local neighborhood similarity between each stress value and all other stress values, and the abnormal probability of each stress value, including the following specific steps: In any stripping update set, obtain the mean of the time intervals between the moment corresponding to the zth stress value and the moments corresponding to all other stress values, recorded as the first mean, obtain the inversely proportional normalized value of the mean M of the local neighborhood similarity between the zth stress value and all other stress values, and use the normalized value of the product of the first mean, the inversely proportional normalized value of the mean M, and the abnormal probability of the zth stress value as the stripping probability of the zth stress value.

9. The method for real-time monitoring of steel purlin stress based on the Internet of Things according to claim 1, characterized in that: The specific steps of determining whether to issue an early warning prompt are as follows: When the abnormal probability of the stress value of any stress monitoring point at any moment is greater than the preset abnormal threshold, an early warning prompt is issued.

10. A storage medium for real-time monitoring of steel purlin stress based on the Internet of Things, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the real-time monitoring method of steel purlin stress based on the Internet of Things are implemented as described in any one of claims 1 to 9.

Images