A method, system, device and medium for predicting concrete temperature in a shaft construction

Abstract

The application provides a concrete temperature prediction method, system, device and medium in shaft construction, and relates to the technical field of shaft construction.The method comprises the following steps: acquiring shaft concrete construction videos of different depth sections; determining a shaft concrete temperature risk depth section by using a risk depth section processing model based on the shaft concrete construction videos of different depth sections; determining a plurality of preliminary measurement points based on the shaft concrete construction videos of the concrete temperature risk depth section; acquiring concrete temperature data of the plurality of preliminary measurement points; acquiring temperature data of a plurality of supplementary measurement points; and generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction videos of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, temperature data of a plurality of temperature inspection risk points and the temperature data of the plurality of supplementary measurement points.The method can efficiently and accurately determine temperature distribution information of the shaft concrete temperature risk depth section.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of shaft construction, and particularly relates to a concrete temperature prediction method, system, device and medium in shaft construction. BACKGROUND

[0002] In the shaft construction process, concrete temperature control is a key link to ensure the engineering quality and structural safety. The traditional method mainly relies on manual experience to select a limited number of temperature measurement points at a specific depth, which has inherent defects of strong randomness and insufficient coverage. Not only is it time-consuming and labor-intensive, but also the detection is difficult to comprehensively and accurately reflect the real distribution of the concrete temperature inside the complex structure of the shaft. Especially for super-deep shafts, the temperature field is influenced by the comprehensive effects of geological conditions, cement hydration heat and environmental factors, and the distribution is extremely uneven. It is difficult to effectively capture potential local high temperature or temperature gradient abnormal areas with only sparse point measurement data. This prediction blind area based on limited information is prone to cause inaccurate judgment of the real thermal state inside the concrete structure, and cannot timely find the temperature stress concentration point, thereby increasing the risk of concrete cracking and other durability problems, and restricting the reliability and effectiveness of temperature monitoring in the construction process.

[0003] Therefore, how to efficiently and accurately determine the temperature distribution information of the temperature risk depth section of the shaft concrete is a problem to be solved at present. SUMMARY

[0004] The technical problem solved by the present application is how to efficiently and accurately determine the temperature distribution information of the temperature risk depth section of the shaft concrete.

[0005] According to a first aspect, the present application provides a method for predicting concrete temperature in shaft construction, comprising: obtaining shaft concrete construction videos of different depth sections; determining a shaft concrete temperature risk depth section using a risk depth section processing model based on the shaft concrete construction videos of the different depth sections, and obtaining shaft concrete construction videos of the concrete temperature risk depth section; determining a plurality of preliminary measurement points based on the shaft concrete construction videos of the concrete temperature risk depth section; obtaining concrete temperature data of the plurality of preliminary measurement points; generating a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction videos of the concrete temperature risk depth section and the concrete temperature data of the plurality of preliminary measurement points; determining a plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; obtaining temperature data of the plurality of temperature inspection risk points; determining a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and obtaining temperature data of the plurality of supplementary measurement points; and generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction videos of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points.

[0006] In a possible implementation, the determining of the plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section comprises: determining a plurality of concrete temperature anomaly points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; determining a plurality of concrete temperature regions based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the plurality of concrete temperature anomaly points; determining N temperature inspection risk points of a prediction distribution map of each concrete temperature region based on the prediction distribution map of each concrete temperature region, and obtaining the plurality of temperature inspection risk points by aggregating the N temperature inspection risk points of the prediction distribution map of each concrete temperature region.

[0007] In a possible implementation, the determining of the plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points comprises: constructing a temperature inspection risk graph, the temperature inspection risk graph comprising a plurality of temperature inspection risk nodes and a plurality of edges between the plurality of temperature inspection risk nodes, the node features of the temperature inspection risk nodes comprising the temperature data of the temperature inspection risk points, the concrete temperature prediction distribution map of the concrete temperature risk depth section, and the temperature difference between the temperature inspection risk nodes; and obtaining the plurality of supplementary measurement points by processing the temperature inspection risk graph based on a graph neural network.

[0008] In a possible implementation, the risk depth section processing model is a recurrent neural network.

[0009] According to a second aspect, the present application provides a concrete temperature prediction system in shaft construction, comprising: an acquisition module configured to acquire shaft concrete construction videos of different depth sections; a risk depth section processing module configured to determine a shaft concrete temperature risk depth section using a risk depth section processing model based on the shaft concrete construction videos of the different depth sections, and acquire shaft concrete construction videos of the concrete temperature risk depth section; a preliminary measurement point determination module configured to determine a plurality of preliminary measurement points based on the shaft concrete construction videos of the concrete temperature risk depth section; a preliminary temperature data acquisition module configured to acquire concrete temperature data of the plurality of preliminary measurement points; a prediction distribution map generation module configured to generate a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction videos of the concrete temperature risk depth section and the concrete temperature data of the plurality of preliminary measurement points; a test risk point determination module configured to determine a plurality of temperature test risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; a test temperature data acquisition module configured to acquire temperature data of the plurality of temperature test risk points; a supplementary measurement point determination module configured to determine a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature test risk points, and acquire temperature data of the plurality of supplementary measurement points; and a target distribution map generation module configured to generate a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction videos of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature test risk points, and the temperature data of the plurality of supplementary measurement points.

[0010] In a possible implementation, the test risk point determination module is further configured to: determine a plurality of concrete temperature abnormal points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; determine a plurality of concrete temperature regions based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the plurality of concrete temperature abnormal points; determine N temperature test risk points of a prediction distribution map of each concrete temperature region based on the prediction distribution map of each concrete temperature region, and aggregate the N temperature test risk points of the prediction distribution map of each concrete temperature region to obtain the plurality of temperature test risk points.

[0011] In a possible implementation, the supplementary measurement point determination module is further configured to: construct a temperature inspection risk graph, the temperature inspection risk graph comprising a plurality of temperature inspection risk nodes and a plurality of edges between the plurality of temperature inspection risk nodes, a node feature of a temperature inspection risk node comprising temperature data of the temperature inspection risk node, a concrete temperature prediction distribution map of the concrete temperature risk depth section, and an edge between temperature inspection risk nodes being a temperature difference value between the temperature inspection risk nodes; and process the temperature inspection risk graph based on a graph neural network to obtain a plurality of supplementary measurement points.

[0012] In a possible implementation, the risk depth section processing model is a recurrent neural network.

[0013] According to a third aspect, embodiments of the present application provide an electronic device, comprising: a processor; a memory; and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement a method as previously described, the method comprising: obtaining a shaft concrete construction video of different depth sections; determining a shaft concrete temperature risk depth section based on the shaft concrete construction video of the different depth sections using a risk depth section processing model, and obtaining a shaft concrete construction video of the concrete temperature risk depth section; determining a plurality of preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section; obtaining concrete temperature data of the plurality of preliminary measurement points; generating a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the plurality of preliminary measurement points; determining a plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; obtaining temperature data of the plurality of temperature inspection risk points; determining a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and obtaining temperature data of the plurality of supplementary measurement points; and generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points.

[0014] According to a fourth aspect, the embodiments provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the aforementioned method for predicting concrete temperature in shaft construction, the method comprising: obtaining shaft concrete construction videos of different depth sections; determining a shaft concrete temperature risk depth section using a risk depth section processing model based on the shaft concrete construction videos of the different depth sections, and obtaining a shaft concrete construction video of the concrete temperature risk depth section; determining a plurality of preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section; obtaining concrete temperature data of the plurality of preliminary measurement points; generating a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the plurality of preliminary measurement points; determining a plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; obtaining temperature data of the plurality of temperature inspection risk points; determining a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and obtaining temperature data of the plurality of supplementary measurement points; and generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points.

[0015] The application provides a concrete temperature prediction method, system, device and medium in shaft construction, which comprises the following steps: acquiring shaft concrete construction videos of different depth sections; determining a shaft concrete temperature risk depth section by using a risk depth section processing model based on the shaft concrete construction videos of the different depth sections, and acquiring a shaft concrete construction video of the concrete temperature risk depth section; determining a plurality of preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section; acquiring concrete temperature data of the plurality of preliminary measurement points; generating a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the plurality of preliminary measurement points; determining a plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section; acquiring temperature data of the plurality of temperature inspection risk points; determining a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and acquiring temperature data of the plurality of supplementary measurement points; and generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points and the temperature data of the plurality of supplementary measurement points, so that the temperature distribution information of the shaft concrete temperature risk depth section can be determined efficiently and accurately. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a concrete temperature prediction method in shaft construction is provided for the embodiments of the application.

[0017] Figure 2 A schematic diagram of a shaft is provided for the embodiments of the application.

[0018] Figure 3 A flowchart of determining a plurality of temperature inspection risk points is provided for the embodiments of the application.

[0019] Figure 4 A flowchart of determining a plurality of supplementary measurement points is provided for the embodiments of the application.

[0020] Figure 5 A schematic diagram of a concrete temperature prediction system in shaft construction is provided for the embodiments of the application. DETAILED DESCRIPTION

[0021] The application will be described in further detail below with specific reference to the drawings. Like elements in different embodiments are denoted by like reference numerals. In the following description, numerous specific details are described to provide a thorough understanding of the application. However, it will be apparent to one skilled in the art that the present application can be practiced without these specific details. In other instances, well-known methods have not been described in detail in order not to unnecessarily obscure the present application. In the following description, numerous specific details are described to provide a thorough understanding of the application. However, one skilled in the art will recognize that the application can be practiced without these specific details. In some instances, well-known structures have not been described in detail in order not to unnecessarily obscure the present application.

[0022] In the embodiments of the present application, a concrete temperature prediction method in shaft construction is provided, as shown in the figure. Figure 1 The concrete temperature prediction method in shaft construction comprises steps S1-S9.

[0023] In step S1, a shaft concrete construction video of different depth sections is acquired.

[0024] The shaft is a cylindrical or rectangular shaft structure vertically excavated in underground engineering. The shaft can be used for personnel passage, material transportation, ventilation and drainage during underground engineering construction, and is the core operation space for construction operations such as concrete pouring. Figure 2 A schematic diagram of a shaft is provided for the embodiments of the present application.

[0025] The shaft concrete construction video of different depth sections is a video captured by a high-definition camera installed in the shaft construction area, covering the concrete construction process of different depth ranges of the shaft.

[0026] The shaft concrete construction video of different depth sections can be a continuous video stream or a video segmented according to the construction progress, for example, a fixed high-definition camera installed at different heights of the shaft inner wall real-time captures the construction process of the corresponding depth section and generates a continuous video.

[0027] The shaft concrete construction video of different depth sections can record the dynamic information of the whole process of concrete construction at different depth positions of the shaft, including concrete pouring operation, construction environment change, and equipment running state of each depth section.

[0028] In step S2, based on the shaft concrete construction video of different depth sections, a risk depth section processing model is used to determine the concrete temperature risk depth section of the shaft, and a shaft concrete construction video of the concrete temperature risk depth section is acquired.

[0029] The risk depth section processing model is a recurrent neural network. The input of the risk depth section processing model is the shaft concrete construction video of different depth sections, and the output of the risk depth section processing model is a shaft concrete temperature risk depth section.

[0030] A recurrent neural network (RNN) is a neural network capable of processing sequence data. The recurrent neural network has feedback connections inside the network, so that the output at the previous moment can be used as the input at the next moment, thereby giving the model the ability to remember previous information. The recurrent neural network can capture the time dependence, time sequence features and dynamic change rules in the video through hidden states or memory cells.

[0031] The shaft concrete temperature risk depth section is a specific shaft depth range determined by the risk depth section processing model, in which, during construction in different depth section ranges, temperature changes may be abnormal, temperature cracks may be prone to occur, and other quality problems may occur.

[0032] The shaft concrete construction video of the concrete temperature risk depth section is the construction video corresponding to the shaft concrete temperature risk depth section part in the shaft concrete construction video of different depth sections.

[0033] The shaft concrete construction video of different depth sections can directly record the whole process information of shaft concrete construction at different depths, including the speed, time, environmental changes and other time sequence characteristics of concrete pouring. The recurrent neural network has the ability to process sequence data, so it can analyze the frame sequence of the shaft concrete construction video of different depth sections frame by frame. The model can store the construction information in the previous video frames, such as pouring speed, vibration frequency, environmental changes, etc. through the internal memory unit, and can capture the time sequence feature rules in the construction process of each depth section. The model can compare the changes in the state of concrete reflected in the construction videos of different depth sections to identify the depth sections that may cause temperature abnormalities during construction, i.e. determine the shaft concrete temperature risk depth section.

[0034] Step S3, determining a plurality of preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section.

[0035] In some embodiments, a measurement point analysis model can be used to determine a plurality of preliminary measurement points. The measurement point analysis model is a recurrent neural network. The input of the measurement point analysis model is the shaft concrete construction video of the concrete temperature risk depth section, and the output of the measurement point analysis model is a plurality of preliminary measurement points.

[0036] The multiple preliminary measurement points are determined by analyzing the shaft concrete construction video of the concrete temperature risk depth section through a measurement point analysis model, and are internal and surface position points of the concrete in the concrete temperature risk depth section for preliminary temperature measurement.

[0037] The shaft concrete construction video of the concrete temperature risk depth section accurately records the dynamic information of the whole process of the corresponding concrete construction in the risk depth section. These information can be identified by the model as key structure areas and changes in construction operation influence, to serve as the basis for determining the multiple preliminary measurement points.

[0038] The recurrent neural network can perform frame-level image analysis on the shaft concrete construction video of the concrete temperature risk depth section, and then extract spatial features related to temperature distribution, such as thickness variation areas of the concrete, positions of embedded parts, density of steel bars, and layered interfaces of pouring, etc. Through the time series memory capability of the recurrent unit, the recurrent neural network can track dynamic features in the pouring process, such as the exposure time of the concrete in a certain area, the change of curing cover, etc. The recurrent neural network can learn the correlation between visual features and potential key points in the shaft concrete construction video of the concrete temperature risk depth section, such as positions prone to temperature gradient or heat accumulation, such as the place with the maximum thickness, the place with the worst heat dissipation condition, or the place with a sudden change in continuous pouring volume, etc. The model can calculate the temperature sensitivity score of each point in the risk depth section according to the comprehensive analysis of these spatial and time series features. For example, the area with larger thickness will get a higher sensitivity score, because its heat dissipation path is longer. The model can select representative coordinate positions from the areas with higher scores as the multiple preliminary measurement points based on the sensitivity scores and the preset uniformity and coverage principles.

[0039] In some embodiments, determining the multiple preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section includes steps S21-S23:

[0040] Step S21, determining the concrete pouring form variation sequence, key structure position, and vibrating equipment operation path based on the shaft concrete construction video of the concrete temperature risk depth section.

[0041] In some embodiments, a recurrent neural network can be used to determine the concrete pouring form variation sequence, key structure position, and vibrating equipment operation path.

[0042] The concrete pouring form variation sequence is the continuous process information of the dynamic changes of the pouring range, thickness, layered interface, etc. of the concrete in the risk depth section over time, which is determined by the recurrent neural network.

[0043] The key structural position is a structural part sensitive to temperature change determined by the recurrent neural network, including a steel bar dense area, a periphery of a pre-embedded part, a pouring joint, and the like fixed positions.

[0044] The vibrating equipment operation path is a moving track and a stay area of the vibrating equipment in the concrete pouring process determined by the recurrent neural network.

[0045] The recurrent neural network can analyze the construction video of the concrete temperature risk depth section frame by frame, thereby capturing the time sequence correlation between frames. The recurrent neural network can track the pouring range, thickness, and layered interface change at different time points by memorizing the pouring state of the previous frame, thereby extracting a continuous concrete pouring shape change sequence. For the key structural position, the recurrent neural network can extract the stable performance of the steel bar dense area, the periphery of the pre-embedded part, and the like in multiple frames in the video, thereby accurately positioning these fixed parts. When determining the vibrating equipment operation path, the recurrent neural network can record the position of the vibrating equipment in each frame by virtue of the processing capability of the time sequence data, thereby forming a moving track and a stay area.

[0046] In step S22, a plurality of high heat dissipation difficulty positions and a plurality of easy temperature difference change positions are determined based on the concrete pouring shape change sequence, the key structural position, and the vibrating equipment operation path.

[0047] In some embodiments, a deep neural network can be used to determine the plurality of high heat dissipation difficulty positions and the plurality of easy temperature difference change positions.

[0048] The deep neural network (DNN) is a neural network model with multiple hidden layers. The deep neural network can enhance the feature learning and expression capability of the model by increasing the depth of the network. The deep neural network can learn the complex deep features of the data through the weight connection between the neurons of each layer and the nonlinear transformation of multiple layers of input data.

[0049] The high heat dissipation difficulty position is a specific position where internal heat is difficult to dissipate due to large concrete thickness and structural closure, such as a region suddenly thickened in the pouring shape change.

[0050] The easy temperature difference change position is a position where local temperature rises sharply or the temperature difference with the periphery is too large due to uneven vibration and long pouring interval.

[0051] The deep neural network has strong multi-feature fusion and complex relationship mining capabilities. The deep neural network can capture features such as sudden increase in concrete thickness and structure closure by analyzing the pouring shape change sequence, so as to determine the area where heat is difficult to dissipate, and then determine the high heat dissipation difficulty position. At the same time, the deep neural network can combine the fixed attribute of the key structure position, and superimpose the information such as vibration uniformity and stay time reflected by the vibration equipment operation path, so as to identify the area where the local temperature rises sharply or the temperature difference is too large due to uneven vibration and too long pouring interval, and finally determine multiple temperature difference change positions.

[0052] Step S23, determining a plurality of preliminary measurement points based on the key structure position, the plurality of high heat dissipation difficulty positions, and the plurality of temperature difference change positions.

[0053] In some embodiments, a deep neural network can be used to determine a plurality of preliminary measurement points.

[0054] The deep neural network can integrate the key structure position, the high heat dissipation difficulty position, and the temperature difference change position, and through multi-layer feature extraction and weight distribution, to accurately screen out preliminary measurement points. The model can preferentially focus on the fixed parts sensitive to temperature in the key structure position, and at the same time, perform risk level assessment on the high heat dissipation difficulty position and the temperature difference change position, and then combine the spatial distribution characteristics of various positions to eliminate repeated or low-risk areas, and finally determine representative measurement points in the intersection of the key structure position, the high heat dissipation difficulty position, and the temperature difference change position, to ensure that these points can cover the key parts of the structure, and monitor the high-risk heat dissipation and temperature difference areas.

[0055] Step S4, obtaining concrete temperature data of the plurality of preliminary measurement points.

[0056] The concrete temperature data of the plurality of preliminary measurement points is the temperature value information of the concrete at each preliminary measurement point collected in real time by installing temperature sensors at the plurality of preliminary measurement points.

[0057] The concrete temperature data of the plurality of preliminary measurement points can directly reflect the temperature changes of the concrete positions corresponding to the preliminary measurement points at different time points.

[0058] Step S5, generating a concrete temperature prediction distribution map of the concrete temperature risk deep section based on the shaft concrete construction video of the concrete temperature risk deep section and the concrete temperature data of the plurality of preliminary measurement points.

[0059] In some embodiments, the first temperature distribution model can be used to generate a concrete temperature prediction distribution map of the concrete temperature risk depth section. The first temperature distribution model is a generative adversarial network. The input of the first temperature distribution model is the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, and the output of the first temperature distribution model is the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0060] A generative adversarial network (GAN) is a generative model composed of two neural networks, a generator and a discriminator, which can be optimized through adversarial training. The generator is used to learn the distribution of real data and generate new samples, while the discriminator is used to evaluate the authenticity of the samples generated by the generator. Through the mutual game between the generator and the discriminator, the generative adversarial network can achieve accurate simulation of complex data distribution and high-resolution image generation.

[0061] The concrete temperature prediction distribution map of the concrete temperature risk depth section is a visualization chart for predicting the spatial distribution of concrete temperature in the concrete temperature risk depth section, which is generated by the first temperature distribution model.

[0062] The concrete temperature prediction distribution map of the concrete temperature risk depth section can intuitively display the predicted temperature values of concrete at different positions in the risk depth section, as well as the temperature change gradient and distribution law.

[0063] The shaft concrete construction video of the concrete temperature risk depth section contains the dynamic process and environmental information of the concrete construction in the depth section, and can reflect the influence of construction operation on the distribution of concrete temperature. The concrete temperature data of the plurality of preliminary measurement points are the actual temperature observation values of the discrete plurality of preliminary measurement points, and can provide the actual temperature standard of specific points in the risk depth section for the model.

[0064] The generator in the generative adversarial network can perform frame sequence analysis on the input vertical shaft concrete construction video of the concrete temperature risk depth section, and extract key features in the construction process, such as pouring area, vibrating range, and environmental heat dissipation conditions. At the same time, the generative adversarial network can encode the concrete temperature data of multiple preliminary measurement points and obtain the numerical features and trend features of the temperature data. The generator can fuse these extracted features and then perform mapping transformation through a multi-layer neural network to generate a preliminary concrete temperature prediction distribution map of the concrete temperature risk depth section. The discriminator can compare the generated prediction distribution map with the theoretical temperature distribution features derived based on the preliminary measurement point temperature data, and then judge the rationality and accuracy of the distribution map generated by the generator, and feed back the discrimination result to the generator. The generator can adjust the generated prediction distribution map according to the feedback information to continuously optimize the prediction distribution map, until the discriminator cannot effectively distinguish the generated distribution map from the theoretically reasonable distribution, and finally a reasonable concrete temperature prediction distribution map of the concrete temperature risk depth section can be obtained.

[0065] Step S6, determining a plurality of temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0066] In some embodiments, Figure 3 A flowchart for determining a plurality of temperature inspection risk points is provided for the embodiments of the present application, and the determination of the plurality of temperature inspection risk points includes steps S31-S33:

[0067] Step S31, determining a plurality of concrete temperature abnormal points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0068] In some embodiments, an abnormal point analysis model can be used to determine the plurality of concrete temperature abnormal points. The abnormal point analysis model is a convolutional neural network. The input of the abnormal point analysis model is the concrete temperature prediction distribution map of the concrete temperature risk depth section, and the output of the abnormal point analysis model is the plurality of concrete temperature abnormal points.

[0069] A convolutional neural network (CNN) is a deep learning model that is good at processing grid structure data. The core structure of the convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. The convolutional layer can extract local features of the data through a convolution kernel. The pooling layer can perform dimension reduction processing on the extracted features to retain key information. The fully connected layer can map the extracted features to the output layer. The convolutional neural network can automatically learn hierarchical features in the data and gradually build high-level global features from low-level local features, so as to realize classification, recognition, or positioning of the data.

[0070] The multiple concrete temperature abnormal points are determined by analyzing the concrete temperature prediction distribution map of the concrete temperature risk depth section through the abnormal point analysis model, and are a set of specific point positions whose temperature values exceed the normal and reasonable range and whose temperature changes do not conform to the conventional rules.

[0071] The concrete temperature prediction distribution map of the concrete temperature risk depth section directly shows the spatial distribution of the concrete temperature in the depth section, which contains the numerical distribution of the temperature, the change gradient and other key information. These information can identify the position reflecting the temperature anomaly through the model.

[0072] The convolutional neural network can perform sliding convolution operation on the concrete temperature prediction distribution map data of the concrete temperature risk depth section through the convolution kernel in the convolution layer, so as to extract local features in the temperature distribution map, such as edge features of temperature mutation area and aggregation features of high temperature value area. The convolutional neural network can use a sliding window mechanism to scan the temperature prediction distribution map. If the local temperature features in the sliding window, such as the difference between the central temperature and the surrounding average temperature and the significant deviation of the temperature variance in the region from the normal distribution statistical characteristics in the training set, are significantly different from the normal distribution statistical characteristics in the training set, the position corresponding to the region will be assigned a higher abnormal score. The model will pay special attention to the regions whose predicted temperature exceeds the critical threshold and the boundary regions whose temperature gradient change rate exceeds the set safety standard. Finally, the model can determine multiple pixel or grid coordinates with high risk features, i.e. multiple concrete temperature abnormal points, according to the abnormal score threshold.

[0073] In step S32, multiple concrete temperature regions are determined based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the multiple concrete temperature abnormal points.

[0074] In some embodiments, a region positioning model can be used to determine the multiple concrete temperature regions. The region positioning model is a deep neural network. The input of the region positioning model is the concrete temperature prediction distribution map of the concrete temperature risk depth section and the multiple concrete temperature abnormal points, and the output of the region positioning model is the multiple concrete temperature regions.

[0075] The multiple concrete temperature regions are multiple sub-regions located in the concrete temperature risk depth section of the shaft and composed of continuous regions with similar temperature characteristics, which are determined by the region positioning model. Each concrete temperature region contains one or more concrete temperature abnormal points, and the concrete temperature prediction value range and the temperature change trend are consistent.

[0076] The concrete temperature prediction distribution map of the concrete temperature risk depth section provides continuous temperature values of the temperature field within the concrete temperature risk depth section. The multiple concrete temperature abnormal points provide accurate coordinate information of local risk concentration. The regional positioning model can determine the accurate spatial boundaries and coverage of the multiple concrete temperature regions by combining the two types of input information. For one or a group of adjacent concrete temperature abnormal points, the deep neural network can refer to the temperature attenuation or isotherm trend centered on the abnormal point in the prediction distribution map. The deep neural network can identify the reasonable boundaries of the abnormal point influence range by learning the temperature diffusion and heat conduction rules, i.e. the boundaries where the temperature features change significantly in space or the temperature gradient decreases to a normal level. The model can use the idea of semantic segmentation or instance segmentation to perform pixel-level classification or bounding box regression on the prediction distribution map, thereby aggregating continuous pixel regions with similar high-temperature features or in the same temperature difference band. The model can ensure that the internal temperature features of the multiple concrete temperature regions divided in the training process remain high consistency, and the feature difference between each concrete temperature region is obvious.

[0077] In step S33, N temperature inspection risk points of the prediction distribution map of each concrete temperature region are determined based on the prediction distribution map of each concrete temperature region, and the N temperature inspection risk points of the prediction distribution map of each concrete temperature region are aggregated to obtain multiple temperature inspection risk points.

[0078] The prediction distribution map of each concrete temperature region is a sub-distribution map formed by independently extracting the corresponding part of the single concrete temperature region from the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0079] In some embodiments, the N temperature inspection risk points of the prediction distribution map of each concrete temperature region can be determined using an inspection risk point determination model. The inspection risk point determination model is a convolutional neural network. The input of the inspection risk point determination model is the prediction distribution map of each concrete temperature region, and the output of the inspection risk point determination model is the N temperature inspection risk points of the prediction distribution map of each concrete temperature region.

[0080] The N temperature inspection risk points of the prediction distribution map of each concrete temperature region are a set of N specific points selected from the region for accurately verifying the temperature prediction accuracy of the region after the inspection risk point determination model performs fine analysis on the prediction distribution map of each concrete temperature region.

[0081] The N temperature inspection risk points of the predicted distribution map of each concrete temperature region are used for high-precision temperature verification in the corresponding concrete temperature region. The N temperature inspection risk points of the predicted distribution map of each concrete temperature region not only cover the identified concrete temperature abnormal points in the corresponding concrete temperature region, but also include key nodes of temperature gradient changes, characteristic points of temperature distribution boundaries, and other core positions that can fully characterize the temperature law of the region, so as to ensure that through temperature monitoring of these temperature inspection risk points, the reliability and accuracy of the temperature prediction distribution map of the region can be fully verified.

[0082] N is a positive integer preset by the inspection risk point determination model according to the size of the region and the complexity of the temperature, and each concrete temperature region corresponds to an independent N value.

[0083] The plurality of temperature inspection risk points are a point set formed by aggregating the N temperature inspection risk points of the predicted distribution map of each concrete temperature region.

[0084] The predicted distribution map of each concrete temperature region details the specific distribution of the concrete temperature in the region, including temperature values, change trends, positions of abnormal points, and other information. These information can provide the inspection risk point determination model with identification basis for key positions in the region, so as to ensure that the temperature inspection risk points selected by the model can accurately reflect the temperature risk conditions in the region.

[0085] The convolutional neural network can extract local features from the predicted distribution map of each concrete temperature region, and the convolution kernel of the convolutional neural network can capture key information such as the surrounding features of the temperature abnormal points and the temperature gradient change features in each concrete temperature region. Based on the actual size of the concrete temperature region and the complexity of the temperature distribution, the model can determine the N value corresponding to the region. The larger the region and the more dramatic the temperature gradient fluctuation, the higher the N value. The model can identify the key positions in each concrete temperature region that are most sensitive to temperature changes and have the highest abnormal risk, such as the center position of the temperature abnormal point, the point with the maximum temperature gradient, the point with the highest or lowest temperature in the region, etc. Finally, according to the determined N value, the model can select N representative points from the identified key positions as the N temperature inspection risk points of the predicted distribution map of the concrete temperature region.

[0086] Step S7, acquiring temperature data of the plurality of temperature inspection risk points.

[0087] The temperature data of the plurality of temperature inspection risk points is the temperature value information of the concrete at each temperature inspection risk point collected in real time by installing temperature sensors at the plurality of temperature inspection risk points.

[0088] Step S8, determining a plurality of supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and acquiring temperature data of the plurality of supplementary measurement points.

[0089] The plurality of supplementary measurement points are a set of specific points for supplementary measurement of concrete temperature of the N temperature inspection risk points of the predicted distribution map of the plurality of concrete temperature regions.

[0090] The plurality of supplementary measurement points can be used to fill in the monitoring blanks not covered by the temperature inspection risk points to improve the monitoring of the temperature distribution of the concrete temperature risk depth section.

[0091] The temperature data of the plurality of supplementary measurement points are temperature value information of the concrete at each supplementary measurement point collected in real time by installing temperature sensors at the plurality of supplementary measurement points.

[0092] In some embodiments, Figure 4 A flowchart for determining a plurality of supplementary measurement points is provided for embodiments of the present application, and the determination of the plurality of supplementary measurement points includes steps S41-S42.

[0093] Step S41, constructing a temperature inspection risk map, the temperature inspection risk map including a plurality of temperature inspection risk nodes and a plurality of edges between the plurality of temperature inspection risk nodes, the node features of the temperature inspection risk points including temperature data of the temperature inspection risk points and the concrete temperature prediction distribution map of the concrete temperature risk depth section, and the edges between the temperature inspection risk nodes being temperature difference values between the temperature inspection risk nodes.

[0094] The temperature inspection risk map is a map data constructed based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and the temperature inspection risk map includes a plurality of temperature inspection risk nodes and a plurality of edges connecting the nodes.

[0095] The node features of the temperature inspection risk points include temperature data of the temperature inspection risk points and the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0096] The edges between the nodes represent temperature difference values between the corresponding temperature inspection risk nodes, and the edges between the nodes can reflect the degree of temperature correlation between the nodes.

[0097] Step S42, processing the temperature inspection risk map based on a graph neural network to obtain a plurality of supplementary measurement points.

[0098] The graph neural network is a neural network model for processing graph data, which can learn the representation of nodes by aggregating and transforming the feature information of nodes and their neighbors. The graph neural network can effectively capture the complex topological relationship between nodes in the graph. The input of the graph neural network is the temperature test risk graph, and the output of the graph neural network is a plurality of supplementary measurement points.

[0099] By constructing the temperature test risk graph, the temperature data of the discrete plurality of temperature test risk points can be clearly integrated with the background information of the concrete temperature prediction distribution graph of the continuous concrete temperature risk depth section and the temperature difference relationship between points. The temperature test risk graph can explicitly express the temperature relationship and spatial topological structure between each temperature test risk point. This structural information is very important for the model to identify the local and global uncertainty of the temperature field, because the areas that are not fully measured are often located around the high temperature difference edge or in the topological position far from the existing measurement points. Using the temperature data of the temperature test risk points and the concrete temperature prediction distribution graph of the concrete temperature risk depth section as node features, and using the temperature difference between the temperature test risk nodes as edge features, the measured data, the prediction background and the relationship information between points can be more comprehensively utilized. This helps the graph neural network model to better understand the dependence and propagation pattern of temperature in space, thereby improving the accuracy of identifying the area where the prediction error is concentrated. Based on the graph neural network, the temperature test risk graph can be processed to effectively learn the complex relationship and information transmission between nodes, thereby more accurately mining the areas that are not covered or have high uncertainty in the temperature field. Compared with the traditional distance-based or interpolation-based supplementary point determination method, the graph neural network has better representation and learning ability when processing such structured data with relationship information.

[0100] The graph neural network can initialize the embedding of the plurality of temperature test risk nodes in the temperature test risk graph and their features, including the temperature data of the temperature test risk points and the background features in the concrete temperature prediction distribution graph of the concrete temperature risk depth section. In each round of message passing and aggregation process, the graph neural network can calculate the influence weight of adjacent nodes on the center node feature according to the edges between the temperature test risk nodes. If the temperature difference between two adjacent nodes is small, it means that they are in a similar temperature environment, so the edge weight connecting them is larger and the feature information is easier to aggregate. By analyzing the node features and topological relationships, the graph neural network can identify the blank areas in temperature monitoring, the areas where the temperature change is unclear, and the areas where the temperature difference is large but lack intermediate monitoring points. For these areas, the model can generate new nodes as supplementary measurement points to ensure that the supplementary measurement points can effectively fill the monitoring blank, thereby improving the comprehensiveness and accuracy of temperature monitoring.

[0101] Step S9, generating a target temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points.

[0102] In some embodiments, the target temperature prediction distribution map of the concrete temperature risk depth section can be generated using a second temperature distribution model. The second temperature distribution model is a generative adversarial network. The input of the second temperature distribution model is the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points. The output of the second temperature distribution model is the target temperature prediction distribution map of the concrete temperature risk depth section.

[0103] The target temperature prediction distribution map of the concrete temperature risk depth section is a visual chart that can accurately reflect the spatial distribution of the concrete temperature in the concrete temperature risk depth section, which is generated by the second temperature distribution model to make a high-precision and high-reliability prediction of the temperature field inside the concrete temperature risk depth section.

[0104] The target temperature prediction distribution map of the concrete temperature risk depth section is an optimization and correction of the concrete temperature prediction distribution map of the concrete temperature risk depth section, and can more accurately show the temperature values, change gradients, and distribution rules of the concrete at different positions in the risk depth section.

[0105] The shaft concrete construction video of the concrete temperature risk depth section can reflect the influence of the construction process on the temperature distribution, and the concrete temperature data of the plurality of preliminary measurement points provides a basic temperature benchmark. On this basis, the temperature data of the plurality of temperature inspection risk points verifies the actual situation of the temperature abnormal area, and the temperature data of the plurality of supplementary measurement points further fills in the monitoring blank of the plurality of temperature inspection risk points.

[0106] The generator in the generative adversarial network can perform frame sequence feature extraction on the input shaft concrete construction video of the concrete temperature risk depth section, so as to obtain features of construction operations, environmental conditions and the like which affect temperature distribution. Meanwhile, the generator can also fuse and encode concrete temperature data of multiple preliminary measurement points, temperature data of multiple temperature inspection risk points and temperature data of multiple supplementary measurement points, so as to obtain comprehensive temperature value features and change trend features. The generator can perform deep fusion on the construction features and the temperature features, and perform mapping and transformation through a multi-layer neural network, so as to generate a preliminary target temperature prediction distribution map of the concrete temperature risk depth section. The discriminator can compare the generated target prediction distribution map with theoretical temperature distribution features derived based on all input data, so as to judge the accuracy and rationality of the generated distribution map, and feed back the judgment result to the generator. The generator can adjust parameters according to the feedback information, and continuously optimize the generated target prediction distribution map, until the discriminator cannot effectively distinguish the generated distribution map from the theoretical reasonable distribution, and finally obtain a target temperature prediction distribution map which can accurately reflect the temperature spatial distribution in the concrete temperature risk depth section.

[0107] Based on the same inventive concept, Figure 5 A concrete temperature prediction system in shaft construction provided by an embodiment of the present application is shown in a schematic diagram, and the concrete temperature prediction system in shaft construction comprises:

[0108] An acquisition module 51 is configured to acquire shaft concrete construction videos of different depth sections.

[0109] A risk depth section processing module 52 is configured to determine a shaft concrete temperature risk depth section based on the shaft concrete construction videos of the different depth sections by using a risk depth section processing model, and acquire a shaft concrete construction video of the concrete temperature risk depth section.

[0110] A preliminary measurement point determination module 53 is configured to determine multiple preliminary measurement points based on the shaft concrete construction video of the concrete temperature risk depth section.

[0111] A preliminary temperature data acquisition module 54 is configured to acquire concrete temperature data of the multiple preliminary measurement points.

[0112] A prediction distribution map generation module 55 is configured to generate a concrete temperature prediction distribution map of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the multiple preliminary measurement points.

[0113] An inspection risk point determination module 56 is configured to determine multiple temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section.

[0114] The temperature data acquisition module 57 is configured to acquire temperature data of the plurality of temperature inspection risk points.

[0115] The supplementary measurement point determination module 58 is configured to determine a plurality of supplementary measurement points based on the concrete temperature prediction distribution of the concrete temperature risk depth section and the temperature data of the plurality of temperature inspection risk points, and acquire temperature data of the plurality of supplementary measurement points.

[0116] The target distribution generation module 59 is configured to generate a target temperature prediction distribution of the concrete temperature risk depth section based on the shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the plurality of preliminary measurement points, the temperature data of the plurality of temperature inspection risk points, and the temperature data of the plurality of supplementary measurement points.

[0117] It should be noted that the foregoing description of the embodiments of the present specification is sometimes indicative of aspects of various embodiments, to facilitate an understanding of one or more of the inventive aspects. However, such descriptions are not intended to be, and are not to be construed as providing any kind of limitation of the inventive aspects. In fact, it is intended to cover also other potential and / or obvious aspects within the scope of the inventive concept.

[0118] Finally, it is to be understood that the embodiments of the present specification are merely illustrative of the principles of this specification. Other variations can occur to those skilled in the art in light of the above descriptions. Accordingly, alternatives and equivalents of the embodiments of the present specification could be employed as would be understood by one of ordinary skill in the art. The embodiments of the present specification are to be construed as not necessarily limited to the particular exemplifications introduced above but are amenable to any changes and physically possible embodiments interposed equivalents.

Claims

1. A method for predicting concrete temperature during shaft construction, characterized in that, include: Obtain videos of vertical shaft concrete construction at different depths; Based on the vertical shaft concrete construction videos of different depth segments, the risk depth segment of vertical shaft concrete temperature risk is determined using the risk depth segment processing model, and vertical shaft concrete construction videos of the concrete temperature risk depth segment are obtained. Based on the concrete construction video of the vertical shaft at the concrete temperature risk depth range, multiple preliminary measurement points are determined, including: determining the concrete pouring morphology change sequence, key structural locations, and vibration equipment operation paths based on the concrete pouring morphology change sequence, the key structural locations, and the vibration equipment operation paths; determining multiple locations with high heat dissipation difficulty and multiple locations prone to temperature difference changes based on the concrete pouring morphology change sequence, the key structural locations, and the vibration equipment operation paths; and determining multiple preliminary measurement points based on the key structural locations, the multiple locations with high heat dissipation difficulty, and the multiple locations prone to temperature difference changes. Obtain concrete temperature data from multiple preliminary measurement points; Based on the vertical shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the multiple preliminary measurement points, a concrete temperature prediction distribution map of the concrete temperature risk depth section is generated. Based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment, multiple temperature inspection risk points are determined, including: determining multiple concrete temperature anomaly points based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment; determining multiple concrete temperature regions based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment and the multiple concrete temperature anomaly points; determining N temperature inspection risk points for each concrete temperature region based on the prediction distribution map of each concrete temperature region; and summing up the N temperature inspection risk points for each concrete temperature region to obtain multiple temperature inspection risk points. Acquire temperature data from multiple temperature inspection risk points; Based on the concrete temperature risk depth segment predicted distribution map and the temperature data of the multiple temperature inspection risk points, multiple supplementary measurement points are determined, and the temperature data of the multiple supplementary measurement points are obtained. Based on the vertical shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the multiple preliminary measurement points, the temperature data of the multiple temperature inspection risk points, and the temperature data of the multiple supplementary measurement points, a target temperature prediction distribution map of the concrete temperature risk depth section is generated.

2. The method for predicting concrete temperature during shaft construction as described in claim 1, characterized in that, The concrete temperature prediction distribution map based on the concrete temperature risk depth segment identifies multiple temperature inspection risk points, including: Based on the concrete temperature risk depth segment, a number of concrete temperature anomaly points were identified. Based on the concrete temperature risk depth segment predicted distribution map and the multiple concrete temperature anomaly points, multiple concrete temperature zones are determined. Based on the predicted distribution map of each concrete temperature zone, N temperature inspection risk points are determined for each concrete temperature zone. The N temperature inspection risk points of each concrete temperature zone are then summarized to obtain multiple temperature inspection risk points.

3. The method for predicting concrete temperature during shaft construction as described in claim 1, characterized in that, The concrete temperature prediction distribution map based on the concrete temperature risk depth segment and the temperature data of the multiple temperature inspection risk points determine multiple supplementary measurement points, including: A temperature inspection risk map is constructed, which includes multiple temperature inspection risk nodes and multiple edges between the multiple temperature inspection risk nodes. The node features of the temperature inspection risk nodes include the temperature data of the temperature inspection risk nodes and the predicted distribution map of the concrete temperature risk depth segment. The edges between the temperature inspection risk nodes are the temperature differences between the temperature inspection risk nodes. Multiple supplementary measurement points were obtained by processing the temperature inspection risk map using a graph neural network.

4. The method for predicting concrete temperature during shaft construction as described in claim 1, characterized in that, The risk depth segment processing model is a recurrent neural network.

5. A concrete temperature prediction system for vertical shaft construction, characterized in that, include: The acquisition module is used to acquire videos of vertical shaft concrete construction at different depths. The risk depth segment processing module is used to determine the vertical shaft concrete temperature risk depth segment based on the vertical shaft concrete construction videos of the different depth segments, and to obtain the vertical shaft concrete construction videos of the concrete temperature risk depth segments. The preliminary measurement point determination module is used to determine multiple preliminary measurement points based on the vertical shaft concrete construction video of the concrete temperature risk depth section, including: determining the concrete pouring morphology change sequence, key structural locations, and vibration equipment operation path based on the concrete pouring morphology change sequence, the key structural locations, and the vibration equipment operation path; determining multiple locations with high heat dissipation difficulty and multiple locations prone to temperature difference changes based on the concrete pouring morphology change sequence, the key structural locations, and the vibration equipment operation path; and determining multiple preliminary measurement points based on the key structural locations, the multiple locations with high heat dissipation difficulty, and the multiple locations prone to temperature difference changes. The preliminary temperature data acquisition module is used to acquire concrete temperature data from multiple preliminary measurement points; The prediction distribution map generation module is used to generate a predicted distribution map of concrete temperature in the concrete temperature risk depth section based on the vertical shaft concrete construction video of the concrete temperature risk depth section and the concrete temperature data of the multiple preliminary measurement points. The inspection risk point determination module is used to determine multiple temperature inspection risk points based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment. This includes: determining multiple concrete temperature anomaly points based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment; determining multiple concrete temperature regions based on the concrete temperature prediction distribution map of the concrete temperature risk depth segment and the multiple concrete temperature anomaly points; determining N temperature inspection risk points for each concrete temperature region based on its prediction distribution map; and summing the N temperature inspection risk points for each concrete temperature region to obtain multiple temperature inspection risk points. The temperature data acquisition module is used to acquire temperature data from multiple temperature inspection risk points. The supplementary measurement point determination module is used to determine multiple supplementary measurement points based on the concrete temperature prediction distribution map of the concrete temperature risk depth section and the temperature data of the multiple temperature inspection risk points, and to acquire the temperature data of the multiple supplementary measurement points. The target distribution map generation module is used to generate a target temperature prediction distribution map of the concrete temperature risk depth section based on the vertical shaft concrete construction video of the concrete temperature risk depth section, the concrete temperature data of the multiple preliminary measurement points, the temperature data of the multiple temperature inspection risk points, and the temperature data of the multiple supplementary measurement points.

6. The concrete temperature prediction system during shaft construction as described in claim 5, characterized in that, The inspection risk point determination module is also used for: Based on the concrete temperature risk depth segment, a number of concrete temperature anomaly points were identified. Based on the concrete temperature risk depth segment predicted distribution map and the multiple concrete temperature anomaly points, multiple concrete temperature zones are determined. Based on the predicted distribution map of each concrete temperature zone, N temperature inspection risk points are determined for each concrete temperature zone. The N temperature inspection risk points of each concrete temperature zone are then summarized to obtain multiple temperature inspection risk points.

7. The concrete temperature prediction system for shaft construction as described in claim 5, characterized in that, The supplementary measurement point determination module is also used for: A temperature inspection risk map is constructed, which includes multiple temperature inspection risk nodes and multiple edges between the multiple temperature inspection risk nodes. The node features of the temperature inspection risk nodes include the temperature data of the temperature inspection risk nodes and the predicted distribution map of the concrete temperature risk depth segment. The edges between the temperature inspection risk nodes are the temperature differences between the temperature inspection risk nodes. Multiple supplementary measurement points were obtained by processing the temperature inspection risk map using a graph neural network.

8. The concrete temperature prediction system during shaft construction as described in claim 5, characterized in that, The risk depth segment processing model is a recurrent neural network.

9. An electronic device, characterized in that, include: processor; Memory; And a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method for predicting concrete temperature during shaft construction as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for predicting concrete temperature during shaft construction as described in any one of claims 1 to 4.

Images