Mass concrete temperature monitoring and early warning method

Abstract

A mass concrete temperature monitoring and early warning method comprises the following steps: acquiring and preprocessing historical data, and constructing input and output; a temperature prediction model and a rainfall prediction model are constructed and trained, the temperature prediction model and the rainfall prediction model both adopt neural network models, and an attention mechanism is built in the neural network models; concrete temperature distribution data and meteorological data are collected, the future concrete temperature is predicted through the temperature prediction model, and if the predicted temperature exceeds an early warning set value, early warning point position information is sent; determining a rainfall prediction grade by using the rainfall prediction model, and sending rainfall prediction grade information; real-time monitoring of the temperature and the rainfall condition in the concrete can be realized, and the accuracy and the timeliness of monitoring are improved.

Description

Technical Field

[0001] The invention relates to the technical field of mass concrete construction, and in particular to a mass concrete temperature monitoring and early warning method. Background Art

[0002] Temperature control is a crucial link in the construction process of large-volume concrete. Traditional temperature monitoring methods often rely on manual handheld thermometers for fixed-point measurement, which is not only inefficient but also difficult to achieve real-time monitoring of the internal temperature of concrete. In addition, rainfall also has a significant impact on the construction and hardening process of concrete, but existing monitoring systems often ignore this factor and lack comprehensive monitoring methods.

[0003] Specifically, the prior art has the following disadvantages:

[0004] 1. Temperature monitoring is not real-time and comprehensive: Traditional manual measurement methods cannot achieve continuous and real-time monitoring of the internal temperature of concrete, and the measurement points are limited, making it difficult to fully reflect the temperature distribution of concrete.

[0005] 2. Insufficient assessment of rainfall impact: Rainfall has an important impact on the construction and hardening process of concrete, but the existing monitoring systems often lack real-time monitoring and early warning functions for rainfall, making it difficult for construction units to adjust construction plans in a timely manner, affecting project quality.

[0006] 3. Limited data processing and analysis capabilities: The data collected by traditional monitoring methods are often scattered and lack systematic data processing and analysis capabilities, making it difficult to provide a scientific basis for construction decisions. Summary of the invention

[0007] In view of the above problems existing in the prior art, this application proposes a large-volume concrete temperature monitoring and early warning method. It aims to achieve real-time and comprehensive monitoring of the internal temperature of concrete, and simultaneously monitor rainfall conditions, to provide timely and accurate construction decision support for construction units. It is easy to install, highly applicable, and low cost.

[0008] A method for monitoring and early warning temperature of large volume concrete, characterized by comprising the following steps:

[0009] Acquire historical data and preprocess it to construct input and output; the acquired data includes concrete temperature distribution data and meteorological data;

[0010] Build and train temperature prediction models and rainfall prediction models. Both the temperature prediction model and rainfall prediction model use neural network models with built-in attention mechanisms.

[0011] Collect concrete temperature distribution data and meteorological data, use the temperature prediction model to predict the future concrete temperature, and compare the predicted temperature with the warning set value; if the predicted temperature exceeds the warning set value, send warning point information;

[0012] The rainfall prediction model is used to compare the future predicted rainfall with the set level, determine the rainfall prediction level, and send the rainfall prediction level information.

[0013] The beneficial effects of the present invention are:

[0014] Compared with the existing technology, it can realize real-time monitoring of the internal temperature of concrete and rainfall conditions, improving the accuracy and timeliness of monitoring.

[0015] Comprehensive monitoring: By arranging multiple temperature sensors and rainfall sensors, this system can comprehensively reflect the temperature distribution and rainfall conditions of concrete, providing a scientific basis for construction decisions.

[0016] Early warning function: The system can generate early warning information based on preset thresholds to remind construction units to take timely measures to avoid quality problems and construction delays.

[0017] Data processing and analysis capabilities: The system has powerful data processing and analysis capabilities, and can conduct in-depth mining and analysis of monitoring data, providing strong support for construction optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is the overall system architecture diagram;

[0019] Figure 2 It is a data processing flow chart;

[0020] In the figure: 11 concrete internal temperature data, 12 meteorological data, 21 temperature prediction model, 22 rainfall prediction model, 31 predicted temperature, 32 predicted rainfall level. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] This application is based on the Internet of Things technology. It collects concrete internal temperature and rainfall data in real time through traditional temperature sensors placed inside the concrete and rainfall sensors placed at the construction site. The data is sent to the on-site Bluetooth 5G gateway via Bluetooth / zigbee transmission, and transmitted to the cloud server via 5G communication. The server processes and analyzes the data to generate intuitive monitoring reports and early warning information. Construction unit users can view monitoring data in real time through mobile terminal APP or computer terminal, and adjust the construction plan according to the data to perform concrete cooling and maintenance.

[0023] An intelligent large-volume concrete online temperature control and rainfall monitoring system, the system mainly consists of a temperature sensor array, a cooling water pipe array, a rainfall sensor array, a data measurement collection module, a Bluetooth 5G gateway, a cloud server and a mobile phone APP / desktop program.

[0024] The temperature sensor array and cooling water pipe array are arranged inside the concrete to monitor the internal temperature of the concrete in real time. The traditional 3.5MM plug-in interface is used for measurement, which is low-cost.

[0025] According to the shape and size of the concrete casting to be cast and the layout requirements of the temperature sensor and the cooling water pipe in the concrete casting, the layout position information of the temperature sensor and the cooling water pipe in the concrete casting is calculated, including the number, number of columns, number of rows, spacing and important point layout information of the temperature sensors;

[0026] During installation and layout, the on-site installation and layout of the temperature sensor and the cooling water pipe are completed according to the layout position information. First, the steel frame of the concrete pouring is built, and then the temperature sensor and the cooling water pipe are fixedly installed to the corresponding positions on the steel frame according to the layout position information.

[0027] Rainfall sensor arrays are deployed at construction sites to monitor rainfall conditions in real time.

[0028] The rainfall sensor array periodically collects local meteorological elements around the monitored area, including wind speed and direction sensors, rainfall sensors, ambient temperature sensors and other meteorological element sensors. The above sensors are connected through the Internet of Things, making it faster for the Internet of Things gateway to collect data. The frequency of meteorological element data collection is once every 15 minutes. The Bluetooth 5G gateway is used to process the communication between the rainfall sensor array and the cloud server, and centrally transmits data to the cloud server through a wireless network (which can be through LoRa / NB-IoT / 4G / 5G network).

[0029] The data measurement and collection module is used to collect the data collected by the sensor and transmit it to the Bluetooth 5G gateway via Bluetooth / zigbee.

[0030] The Bluetooth 5G gateway aggregates the data from each data measurement device and uploads it to the cloud server via the 5G network.

[0031] The cloud server has a preset temperature prediction model, which is mainly used to predict future temperatures. The cloud server processes and analyzes the temperature sensor array data, generates a monitoring report, and generates temperature warning information through the temperature prediction model.

[0032] The temperature prediction model consists of an encoder, a decoder and an attention mechanism. It is divided into two sub-layers: the encoder and the decoder. The decoder is connected to the encoder using the attention mechanism.

[0033] The input layer in the encoder is a fully connected layer, which maps the temperature sequence to a vector-dimensional model, and then passes through the self-attention sublayer and the fully connected feedforward sublayer to obtain a vector as the output;

[0034] The encoder encodes the input N-step data, obtains the feature relationship between the attributes at similar time points, the contextual connection, and the change trend of this time period, and finally forms a numerical vector to input into the decoder.

[0035] The encoder consists of 6 identical layers, each of which consists of two sub-layers, namely the multi-head attention mechanism and the fully connected feedforward neural network. Residual connections and regularization are added between each sub-layer.

[0036] The data that is fed into the encoder first passes through a self-attention layer, which helps the encoder focus on the data in the input sequence as it encodes each piece of data. The output of the self-attention layer is passed into a feed-forward neural network. The feed-forward neural network is exactly the same for every word in every position. Here, every word in the input sequence has its own unique path into the encoder. In the self-attention layer, there are dependencies between these paths. The feed-forward layer does not have these dependencies. Therefore, various paths can be executed in parallel in the feed-forward layer.

[0037] The decoder decomposes the input vector and outputs N predicted feature vectors.

[0038] The input layer in the decoder maps the last node input of the input encoder, and then passes through four same-frequency decoding layers for decoding processing, in which the decoder inserts a third sublayer to apply a self-attention mechanism on the encoder output, and finally maps the output to the target time series through the output layer to display the predicted temperature curve. When the prediction reaches the warning set value, an early warning message will be issued.

[0039] The decoder and encoder have a similar structure, but the decoder has an additional attention sublayer, which is to calculate the self-attention of the input.

[0040] The data processing flow includes the following steps.

[0041] 1) Obtain and preprocess data to construct input and output;

[0042] Factors affecting concrete temperature include meteorological factors, which include temperature, relative humidity, atmospheric pressure, wind speed, etc.

[0043] The acquired data generally includes concrete temperature distribution data and weather data, which need to be processed into input sequence and output sequence. Generally, the input sequence includes concrete temperature data and weather data over a period of time. Let D = {x, y} be the processed data set; where x is the input sequence, i.e., historical data, including concrete temperature data and weather data. For each input sequence x∈R S×Q , whose length is S, i.e., historical data of the past S hours, and has Q features, i.e., concrete temperature data and various weather data; for each target sequence y∈R T , whose length is T. The target y can include multiple, i.e., concrete temperature data and rainfall data for the next T hours. R is a data set.

[0044] The meteorological data is external data, including: ambient temperature, air pressure, humidity, wind speed, and rainfall.

[0045] 2) Split the data into training data and test data.

[0046] The samples obtained in step 2 are divided into training data and test data. The training data is used to train the model, and the test data is used to test the effect of the model.

[0047] 3) Use the training data to train the temperature prediction model.

[0048] 4) Early warning judgment: compare the predicted temperature with the early warning set value. The early warning set value is calculated based on the design requirements of the concrete pouring, concrete mix ratio, construction conditions and the empirical formula of temperature change during concrete pouring.

[0049] If the predicted temperature exceeds the warning set value, the on-site monitoring system sends warning point information to the remote monitoring terminal;

[0050] Remote monitoring terminals include mobile phones, tablets, computers and spliced ​​large screens. Mobile phones, tablets, computers and spliced ​​large screens can provide user-friendly interfaces, which are convenient for construction units to view monitoring data and receive early warning information in real time. Construction units log in to the cloud server through mobile phone APP or computer terminals to view monitoring data and receive early warning information in real time. According to the monitoring data and early warning information, they can adjust the construction plan in time to ensure the quality of the project.

[0051] In addition, automatic cooling can be implemented, and the cooling module is set to obtain cooling instructions and warning point information and perform corresponding actions to cool the warning point. The specific steps of automatic cooling are: if the predicted temperature exceeds the warning setting value, when the on-site monitoring system sends the warning point information to the remote monitoring terminal, the remote monitoring terminal confirms and generates a cooling instruction, which is sent to the cooling module. When the cooling module receives the cooling instruction, the industrial computer controls the water pump to start, and the water pump pumps the cooling water in the water tank into the cooling water pipe array to cool the warning point. When the cooling module receives the stop instruction, the industrial computer controls the water pump to shut down and stop cooling the warning point, so as to better ensure that the temperature difference between the surface and the core of the large-volume concrete does not exceed the specification requirements.

[0052] The cooling module includes an industrial computer, a cooling water pipe, a water pump and a water tank. The industrial computer is used to receive cooling instructions, stop instructions and warning point information sent by the instruction sending module, as well as calculate and analyze and control the start or shutdown of the water pump and control the power adjustment of the water pump. The cooling water pipe is connected to the water pump, and the water pump is located at the bottom of the water tank cavity.

[0053] The cloud server also has a rainfall prediction model that can accurately predict the future short-term rainfall level through the processed sample data set. It can effectively process the time domain data analysis of regional meteorological information, extract the characteristics of rainfall level to predict future data characteristics, and then classify the obtained future data characteristics to finally obtain the rainfall level.

[0054] The rainfall prediction model has the same model architecture as the temperature prediction model, and its training meteorological data includes: ambient temperature, air pressure, humidity, wind speed, and rainfall. The rainfall prediction model is trained using historical training data.

[0055] When making early warning judgments, the predicted rainfall is compared with the set level, which includes 7 levels: sporadic light rain, light rain, moderate rain, heavy rain, rainstorm, heavy rainstorm, and extremely heavy rainstorm. The on-site monitoring system sends the rainfall forecast level information to the remote monitoring terminal.

[0056] In specific applications:

[0057] 1. Before concrete construction, determine the layout and quantity of temperature sensors according to the construction drawings and construction plan.

[0058] Fix the temperature sensor on the concrete formwork, making sure the sensor is in close contact with the concrete.

[0059] 2. Install meteorological sensors at appropriate locations on the construction site.

[0060] 3. Connect the data measurement collection module to ensure that the measurement module can collect all input data.

[0061] 4. Test the Bluetooth 5G gateway to see if it can reliably connect to each data measurement and collection module, whether there is message loss, and confirm the gateway networking status and whether the data can be transmitted to the cloud server in real time.

[0062] 5. The construction unit logs in to the cloud server through a mobile phone APP or computer terminal to view monitoring data and receive early warning information in real time.

[0063] 6. Based on monitoring data and early warning information, the construction unit shall promptly adjust the construction plan to ensure project quality.

[0064] For example, in a large bridge project, the construction unit uses this system to monitor the concrete temperature in real time. During the construction process, the system promptly discovered that the temperature inside the concrete in a certain area was abnormally high, and predicted that the temperature would rise to the set value. An early warning message was immediately generated and sent to the construction unit via a mobile phone APP. After receiving the early warning, the construction unit started the cooling module to cool down the concrete, avoiding concrete quality problems caused by excessive temperature. At the same time, during a rainfall process, the system predicted the rainfall trend in real time. According to the rainfall trend early warning information provided by the system, the construction unit can take corresponding measures in time to avoid concrete quality problems caused by rainfall hazards.

[0065] Finally, it should be noted that the above description is only an explanation of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail, those skilled in the art can still modify the technical solutions described above or replace some of the technical features with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for monitoring and early warning temperature of large volume concrete, characterized in that: The following steps are involved: Acquire historical data and preprocess it to construct input and output; the acquired data includes concrete temperature distribution data and meteorological data; Build and train temperature prediction models and rainfall prediction models. Both the temperature prediction model and rainfall prediction model use neural network models with built-in attention mechanisms. Collect concrete temperature distribution data and meteorological data, use the temperature prediction model to predict the future concrete temperature, and compare the predicted temperature with the warning set value; if the predicted temperature exceeds the warning set value, send warning point information; The rainfall prediction model is used to compare the future predicted rainfall with the set level, determine the rainfall prediction level, and send the rainfall prediction level information.

2. A method for monitoring and early warning temperature of mass concrete according to claim 1, characterized in that: The acquired historical data needs to be processed into input sequence and output sequence. The input sequence includes concrete temperature data and weather data for a period of time. Let D = {x, y} be the processed data set. x is the input sequence, i.e., historical data, including concrete temperature data and weather data. For each input sequence x∈R S×Q , whose length is S, i.e., historical data of the past S hours, and has Q features, i.e., concrete temperature data and various weather data; for each target sequence y∈R T , whose length is T, the target y may include multiple, namely, concrete temperature data and rainfall data for the next T hours; R is the data set.

3. A method for monitoring and early warning temperature of mass concrete according to claim 2, characterized in that: The neural network model consists of an encoder, a decoder, and an attention mechanism. The decoder is connected to the encoder using the attention mechanism. The input layer in the encoder is a fully connected layer, which maps the input sequence to a vector-dimensional model, and then passes through the self-attention sublayer and the fully connected feedforward sublayer to obtain a vector as the output; The input layer in the decoder maps the last node input of the input encoder, which is then decoded through four on-frequency decoding layers, where the decoder inserts a third sublayer to apply a self-attention mechanism on the encoder output, and finally mapped to the target time series through the output layer output.

4. A method for monitoring and early warning temperature of mass concrete according to claim 3, characterized in that: The encoder encodes the input N-step data, obtains the characteristic relationship between the attributes at similar time points, the contextual connection, and the change trend of this time period, and finally forms a numerical vector to input into the decoder; The encoder consists of 6 identical layers, each of which consists of two sub-layers, namely the multi-head attention mechanism and the fully connected feedforward neural network; residual connections and regularization are added between each sub-layer.

5. A method for monitoring and early warning temperature of mass concrete according to claim 4, characterized in that: The data input to the encoder first passes through a self-attention layer, which helps the encoder focus on the data in the input sequence when encoding each piece of data; the output of the self-attention layer is passed to the feedforward neural network; the feedforward neural network corresponding to the word at each position is exactly the same; here, the word at each position in the input sequence has its own unique path flowing into the encoder; in the self-attention layer, there are dependencies between the paths; while in the feedforward layer, there are no path dependencies.

6. A method for monitoring and early warning temperature of mass concrete according to claim 4, characterized in that: The decoder decomposes the input vector and outputs N predicted feature vectors; The decoder and encoder have a similar structure, but the decoder has an additional attention sublayer, which is to calculate the self-attention of the input.

7. A method for monitoring and early warning temperature of mass concrete according to claim 1, characterized in that: The meteorological data include: ambient temperature, air pressure, humidity, wind speed, and rainfall.

8. The method for monitoring and early warning temperature of mass concrete according to claim 1, characterized in that: The setting levels include scattered light rain, light rain, moderate rain, heavy rain, torrential rain, heavy rain and extremely heavy rain.

9. The method for monitoring and early warning temperature of mass concrete according to claim 1, characterized in that: If the predicted temperature exceeds the warning set value, the warning point information is sent to the remote monitoring terminal, and a cooling instruction is generated through the remote monitoring terminal and sent to the cooling module; when the cooling module receives the cooling instruction, it controls the cooling water to be pumped into the cooling water pipe array to cool the warning point location.

10. A method for monitoring and early warning temperature of mass concrete according to claim 1, characterized in that: The system predicts rainfall trends in real time; based on the rainfall trend warning information provided by the system, appropriate measures are taken to avoid concrete quality problems caused by rainfall hazards.

Images