Geotechnical engineering construction safety monitoring method and device based on Internet of Things
Through IoT technology combined with BP network and genetic algorithm to optimize convolutional neural network, combined with sensors and remote sensing data, the foundation settlement situation in geotechnical engineering construction is predicted, which solves the problem of untimely monitoring in the existing technology and improves construction safety and reliability, early warning and strategy formulation of foundation settlement are achieved.
Patent Information
- Application Number
- CN202411457107.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-18
- Publication Date
- 2025-01-07
AI Technical Summary
During geotechnical engineering construction, the existing technology cannot monitor foundation settlement and geological changes in a timely manner, resulting in low construction safety and reliability. Especially when there are consolidation settlement and creep phenomena, it cannot effectively respond, affecting construction safety.
Using an Internet of Things method, the convolutional neural network is optimized through BP network and genetic algorithm, combined with sensor data and remote sensing images, predict foundation settlement conditions, generate settlement information, and improve prediction accuracy and reliability through feature fusion to achieve early warning and Strategy development.
It improves the timeliness and reliability of geotechnical engineering construction monitoring results, reduces the impact of foundation settlement on construction, ensures the safety and stability of the construction process, and enhances the accuracy and reliability of the monitoring results.
Smart Images
Figure CN119273307A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and specifically to a method and device for monitoring the safety of geotechnical engineering construction based on the Internet of Things. Background Art
[0002] With the continuous expansion of infrastructure construction, geotechnical engineering is increasingly widely used in modern construction, transportation, water conservancy and other fields. In the process of geotechnical engineering construction, the complexity and instability of geological conditions in the construction area bring many challenges to the construction, especially problems such as foundation settlement, slope stability and groundwater level changes, which are directly related to the safety and long-term stability of the project. Therefore, geotechnical construction safety monitoring has become an important means to ensure project quality, protect personnel safety and reduce economic losses. In the prior art, various types of sensors (such as inclinometers, strain gauges, water level gauges, etc.) can be installed to monitor the key parameters of foundation settlement in real time. However, the settlement data obtained in real time has a certain lag. When there are consolidation settlement, creep phenomena and other conditions in the construction area, effective countermeasures cannot be taken in time, resulting in low accuracy and reliability of safety monitoring results, affecting the safety of geotechnical engineering construction.
[0003] Therefore, how to overcome the above-mentioned technical problems and defects becomes a key issue that needs to be solved. Summary of the invention
[0004] The purpose of this application is to provide a method and device for geotechnical engineering construction safety monitoring based on the Internet of Things to solve the problems raised in the above-mentioned background technology, improve the accuracy and reliability of geotechnical engineering construction monitoring results, and ensure the safety of geotechnical engineering construction.
[0005] The technical solution of the embodiment of the present application is implemented as follows: The present application provides a method for monitoring geotechnical engineering construction safety based on the Internet of Things, the method comprising: Collect sensor data acquired by multiple monitoring points in the construction area in a first construction period to obtain first data; the first construction period is a current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions; Based on the first data and the first model, determining the change of the settlement influencing factors in the first construction period to obtain the sensing data characteristics; the first model includes a second model, and the second model is constructed based on a back propagation (BP) network; Based on the sensor data features, the foundation settlement of the construction area during the first construction period is predicted to obtain first settlement information.
[0006] In the above solution, the first model also includes a third model, and the third model is obtained by optimizing a convolutional neural network (CNN) based on a genetic algorithm (GA); the first settlement information is obtained by predicting the foundation settlement of the construction area in the first construction period based on the sensor data characteristics, including: Continuously acquiring remote sensing images of the foundation in the construction area to obtain second data; Determine the terrain change of the construction area during the first construction period by using the second data and a third model in the first model to obtain image features; The sensor data features and the image features are fused to obtain a predicted situation of foundation settlement in the construction area during the first construction period, thereby generating first settlement information.
[0007] In the above solution, the settlement influencing factors also include a second influencing factor determined based on the construction environment dimension.
[0008] In the above scheme, the first influencing factors include groundwater level, soil cohesion and soil internal friction angle; the second influencing factors include construction time, construction area and construction depth.
[0009] In the above solution, based on the first data and the first model, determining the change of the settlement influencing factors in the first construction period to obtain the sensing data features includes: Quantifying and assigning values to the first sensor data and the second sensor data to obtain a quantization result; the first sensor data includes sensor data related to the first influencing factor in the first data, and the second sensor data includes sensor data related to the second influencing factor in the first data; Based on the quantization result, generating a multi-dimensional feature vector; The multidimensional feature vector is used as the input of the second model in the first model, and the second model is used to determine the change of the settlement influencing factors in the first construction period to obtain the sensor data features.
[0010] In the above scheme, the method further includes: Generate an initial population based on initial configuration information pre-configured for the CNN; the initial configuration information includes multiple CNN configuration strategies; Performing iterative optimization processing on the initial population based on GA to obtain a target population; Based on the target population, determining target configuration information of the CNN; Based on the target configuration information, a third model is constructed.
[0011] In the above scheme, the iterative optimization process based on GA on the initial population to obtain the target population includes: Selecting at least two second individuals from a plurality of first individuals in the initial population based on a fitness proportional selection algorithm; Cross-pairing the at least two second individuals based on a mating pool to obtain a pairing result; the size of the mating pool is different from the size of the initial population; Based on the pairing results, a target population is generated.
[0012] In the above scheme, when the at least two second individuals are cross-paired based on the mating pool, parents are randomly selected to produce an offspring through cross-pairing, and the parents are returned to the mating pool.
[0013] In the above scheme, the method further includes: Determine second settlement information of the construction area in the first construction period based on third data obtained from the plurality of monitoring points; the third data is actually measured foundation settlement data; determining a difference between predicted settlement information and actual settlement information based on the first settlement information and the second settlement information; Based on the difference between the predicted settlement information and the actual settlement information, the model parameters of the first model are optimized.
[0014] The embodiment of the present application also provides a geotechnical engineering construction safety monitoring device based on the Internet of Things, the device comprising: A data acquisition unit is used to collect sensor data acquired by multiple monitoring points in a construction area in a first construction period to obtain first data; the first construction period is a current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions; A calculation unit, used to determine the change of the settlement influencing factors in the first construction period based on the first data and the first model, and obtain the sensing data characteristics; the first model includes a second model, and the second model is constructed based on a BP network; The processing unit is used to predict the foundation settlement of the construction area in the first construction period based on the characteristics of the sensor data to obtain first settlement information.
[0015] The Internet of Things-based geotechnical engineering construction safety monitoring method and device provided in the embodiment of the present application determine the changes in geological factors by acquiring sensor data of factors affecting foundation settlement, so that foundation settlement can be predicted according to the changes in geological factors, and early warning of foundation settlement can be achieved, so that preventive strategies can be formulated in advance to reduce the impact of foundation settlement on geotechnical construction, which not only improves the timeliness of safety monitoring results, but also can improve the reliability and accuracy of the formulated strategies in the process of formulating geotechnical construction strategies based on safety monitoring results, thereby ensuring the safe and smooth progress of the geotechnical engineering construction process; further, due to the combination of the advantages of GA and CNN, the structure and hyperparameters of CNN are automatically optimized by GA, thereby improving the prediction accuracy and generalization ability of the model, thereby improving the accuracy and reliability of the safety monitoring results. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A schematic diagram of a flow chart of a geotechnical engineering construction safety monitoring method based on the Internet of Things provided in an embodiment of the present application; Figure 2 This is a flow chart of training the third model in the method for geotechnical engineering construction safety monitoring based on the Internet of Things according to an embodiment of the present application; Figure 3 A schematic structural diagram of a geotechnical engineering construction safety monitoring device based on the Internet of Things provided in an embodiment of the present application. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0018] The present application embodiment provides a geotechnical engineering construction safety monitoring method based on the Internet of Things, which is applied to electronic equipment, and can be specifically applied to industrial computers, cloud servers, etc. Figure 1 As shown, the method may include S101 to S103; S101 to S103 are described in detail below.
[0019] S101: Collect sensor data acquired from multiple monitoring points in the construction area during the first construction period to obtain first data; the first construction period is the current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions.
[0020] In actual application, the data of sensors installed at each monitoring point within a construction period can be continuously obtained according to the preset sampling frequency.
[0021] In actual applications, the sampling frequencies of different sensors can be the same or different; the corresponding sampling frequencies can be configured according to the changing characteristics of sensor data; specifically, the sampling frequencies can be pre-configured f 1 to f n , n is an integer greater than 1, f 1 to f n Decrease in sequence; according to the data change speed of all sensors, divide them into n The data groups are sorted from fast to slow according to the change speed, and each data group is numbered from A1 to A1 according to the arrangement order. n Then select the preset sampling frequency corresponding to the data group number as the sampling frequency of the sensor data in the data group. For example, the sampling frequency of data group A1 is f 1 .
[0022] In practical applications, the corresponding sampling frequency can also be determined according to the maximum frequency of change of the sensor data; specifically, the minimum sampling frequency can be determined according to the maximum frequency of change of the data based on the Nyquist-Shannon sampling theorem.
[0023] For example, the maximum frequency of change of the groundwater level at the construction site is one unit length (0.1m) per hour, and the maximum frequency of change of the groundwater level is calculated as follows: ; (1) in, Indicates the maximum frequency of change of the sampled data; The sampling frequency of groundwater level can be expressed as: ; (2) in, represents the sampling frequency; according to formula (2), the minimum sampling frequency of groundwater level can be calculated to be 5.56×10 -5 Hz.
[0024] In practical application, considering that foundation settlement is affected not only by the geological conditions of the soil, but also by external construction parameters, the settlement influencing factors include not only the first influencing factors determined based on the geological dimension, but also the second influencing factors determined based on the construction environment dimension.
[0025] Here, the settlement influencing factors can be understood as factors that have an impact on foundation settlement. The first influencing factor can be understood as the parameters in the soil mechanical parameters that affect foundation settlement. The second influencing factor can be understood as the construction parameters that affect foundation settlement during the construction process. The foundation settlement can be reflected through changes in the soil structure, that is, it is related to the first influencing factor, and it can also be reflected through external construction conditions, that is, it is related to the second influencing factor.
[0026] In one embodiment, the first influencing factor may include groundwater level, soil cohesion and soil internal friction angle.
[0027] In practical applications, the height of the groundwater level can be expressed as the height relative to the ground, and is quantified in meters. For example, if the groundwater level is 2 meters underground, it is recorded as -2; the soil cohesion can be expressed as the bonding strength between soil particles, usually in kilopascals (kPa); the internal friction angle of the soil can be expressed as the maximum friction angle between soil particles, usually in degrees; the soil cohesion and the internal friction angle of the soil can be measured experimentally or the specific values can be obtained from or referenced.
[0028] In one embodiment, the second influencing factors may include construction time, construction area and construction depth.
[0029] In actual application, the construction time may indicate the number of days of actual construction in the current cycle; the construction area may indicate the total area excavated or constructed in the current cycle; and the construction depth may indicate the maximum construction depth reached in the current cycle.
[0030] S102: Based on the first data and the first model, determine the change of the settlement influencing factors in the first construction period to obtain sensor data characteristics; the first model includes a second model, and the second model is constructed based on a BP network.
[0031] In actual application, after acquiring the sensor data (ie, the first data), data features, namely, sensing data features, may be extracted from the sensor data; here, the sensing data features may be understood as settlement features determined based on the sensing data.
[0032] Here, after acquiring the first data, the first data may be preprocessed first; the preprocessing process may include: first performing missing value processing on the first data to remove outliers and noise and process missing data; then, performing outlier detection and processing; then, normalizing the first data so that different sensor data are normalized to the same numerical range.
[0033] In practical applications, missing value processing can be achieved through interpolation methods. Specifically, for missing data points, linear interpolation, spline interpolation and other methods can be used to fill them. Data cleaning can also be achieved through deletion methods. Specifically, if there are too many missing data in certain time periods, the data in that time period can be deleted.
[0034] In practical applications, outlier detection and processing can be achieved through the Z-score method, that is, calculating the Z-score of each data point and removing data points that exceed the threshold; it can also be achieved through the interquartile range (IQR) method, that is, using the IQR to identify and remove outliers.
[0035] In practical applications, normalization can be achieved using methods such as Min-Max.
[0036] After the first data is preprocessed, feature extraction may be performed on the first data, and quantization assignment may be performed during the feature extraction process.
[0037] Based on this, in one embodiment, the first data and the first model determine the change of the settlement influencing factors in the first construction period to obtain the sensing data features, which may include: Quantifying and assigning values to the first sensor data and the second sensor data to obtain a quantization result; the first sensor data includes sensor data related to the first influencing factor in the first data, and the second sensor data includes sensor data related to the second influencing factor in the first data; Based on the quantization result, generating a multi-dimensional feature vector; The multidimensional feature vector is used as the input of the second model in the first model, and the second model is used to determine the change of the settlement influencing factors in the first construction period to obtain the sensor data features.
[0038] In practical applications, a multidimensional feature vector is generated based on the quantization result. It can be understood that the first sensor data and the second sensor data are quantized and feature extracted to obtain numerical features, and then all the extracted numerical features are combined into a multidimensional feature vector.
[0039] In practical applications, feature extraction can specifically include the extraction of the following three features: Time series feature extraction; specifically, extracting time series features such as mean, variance, and autocorrelation coefficient of sensor data; Frequency domain feature extraction; specifically, frequency domain features are extracted through Fourier Transform (FT) and other methods; Statistical feature extraction: specifically, calculating the statistical features of sensor data, such as maximum value, minimum value, peak value, etc.
[0040] S103: Based on the sensor data features, predict the foundation settlement of the construction area in the first construction period to obtain first settlement information.
[0041] In practical applications, sensor data provides quantitative and high-precision measurements, which can directly reflect the key factors affecting foundation settlement. In order to improve the feature representation capability of the prediction model and thus improve the prediction accuracy of the prediction model, image data can also be used for prediction.
[0042] Based on this, in one embodiment, the first model may also include a third model, and the third model is obtained by optimizing CNN based on GA.
[0043] Based on this, in one implementation, the method may further include: Continuously acquiring remote sensing images of the foundation in the construction area to obtain second data; The second data and the third model in the first model are used to determine the terrain changes in the construction area during the first construction period to obtain image features.
[0044] In practical applications, the remote sensing image can be acquired by a drone.
[0045] In practical applications, after acquiring remote sensing images, remote sensing images from different periods can be used to determine the terrain differences at different periods, thereby extracting the image features of foundation settlement and identifying abnormal terrain changes.
[0046] Based on this, in one embodiment, the use of the second data and the third model in the first model to determine the terrain change of the construction area during the first construction period to obtain image features may include: Using the third model in the first model, based on the difference between the second data and the baseline data, the terrain changes in the construction area during the first construction period are determined to obtain image features; the baseline data represents the remote sensing image of the construction area acquired during the second construction period, and the second construction period is the N construction periods before the first construction period, where N is an integer greater than 0.
[0047] In actual application, the first model can also be called a multimodal fusion model; here, sensor data can provide high-precision measurement values, such as groundwater level, construction depth, soil moisture, etc., which can directly reflect the key factors affecting foundation settlement; by continuously acquiring sensor data, the dynamic changes of various geological influencing parameters in the process of foundation settlement can be captured; and remote sensing image data can reflect the global perspective of the construction site, so as to present the spatial distribution of foundation settlement, and can combine time and space characteristics for more accurate and comprehensive predictions; because it combines image-based model data and sensor data-based models, the first model can make full use of the advantages of different types of data to improve the accuracy and reliability of foundation settlement prediction results.
[0048] In practical applications, after obtaining the sensor data features and image features, the two types of features can be combined for prediction.
[0049] Based on this, predicting the foundation settlement of the construction area in the first construction period based on the sensor data characteristics to obtain first settlement information may include: The sensor data features and the image features are fused to obtain a predicted situation of foundation settlement in the construction area during the first construction period, thereby generating first settlement information.
[0050] In actual application, the sensor data features and the image features are feature fused to obtain the predicted foundation settlement of the construction area in the first construction period, and the first settlement information is generated. It can be understood that the feature vectors output by the second model and the third model are spliced together, that is, the sensor data features and the image features are feature fused, and then the final prediction result, i.e., the first settlement information, is generated through the fully connected layer; the multimodal fusion model can integrate this information at different levels through feature-level fusion, thereby generating a more comprehensive and powerful feature representation, thereby improving the robustness and generalization ability of the model.
[0051] In actual application, before executing step 101, the prediction model, that is, the first model, can be generated by optimizing training first; here, the sub-models in the first model can be trained separately, that is, the second model and the third model can be trained separately.
[0052] In actual application, before training the third model, GA can be used to optimize the model structure and hyperparameters, determine the configuration scheme of the CNN structure, and then use the optimized configuration scheme to train the third model.
[0053] Based on this, in one embodiment, the method may further include training a third model; Figure 2 As shown, the process of training the third model may include S201 to S204. The process of training the third model is described in detail below.
[0054] S201: Generate an initial population based on initial configuration information pre-configured for CNN; the initial configuration information includes multiple CNN configuration strategies.
[0055] In practical applications, the structure and hyperparameters of CNN can be encoded into a chromosome, and a certain number of initial chromosomes are randomly generated to form an initial population; each chromosome in the initial population is an individual, and each chromosome represents a possible CNN configuration.
[0056] Illustratively, a chromosome may include the following genes: The number of convolutional layers; The number of filters in each convolutional layer; The kernel size of each convolutional layer; The location and type of pooling layers; The number of fully connected layers and the number of nodes in each layer; Learning rate; Batch size; Optimizer type.
[0057] S202: performing iterative optimization processing on the initial population based on GA to obtain a target population.
[0058] In practical applications, through selection operations, individuals in the initial population can be sorted according to the fitness function, and individuals with higher fitness can be selected as parents; selection methods such as roulette selection and tournament selection can be used for individual selection.
[0059] Based on this, in one embodiment, the iterative optimization process based on GA on the initial population to obtain the target population may include: Selecting at least two second individuals from a plurality of first individuals in the initial population based on a fitness proportional selection algorithm; Cross-pairing the at least two second individuals based on a mating pool to obtain a pairing result; the size of the mating pool is different from the size of the initial population; Based on the pairing results, a target population is generated.
[0060] In one embodiment, when the at least two second individuals are cross-paired based on the mating pool, parents are randomly selected to produce an offspring through cross-pairing, and the parents are returned to the mating pool.
[0061] In practical applications, the second individual can be understood as the above-mentioned parent generation. After the parent generation is selected, two individuals are selected from the parent generation for crossover processing to generate a new offspring. The crossover point can be any position in the chromosome. For example, the number of convolutional layers and the number of nodes in the fully connected layer can be selected for crossover to generate a new CNN configuration.
[0062] In actual application, the embodiment of the present application can adopt an elite genetic algorithm; since the mating pool size is different from the initial population size, the most adaptable individuals (such as the first ten individuals), i.e., the elites, will directly enter the next generation. Therefore, it can be ensured that the best individuals in each generation can be directly passed on to the next generation, thereby improving the convergence speed and stability of the algorithm and preventing excellent individuals from being lost during the evolution process.
[0063] In practical applications, after a crossover operation is performed to generate new offspring, the offspring can be mutated to introduce new changes. Here, mutation can be achieved by randomly changing certain genes in the chromosomes. For example, mutation can be achieved by randomly increasing or decreasing the number of convolutional layers, or changing the learning rate.
[0064] In actual application, after selection, crossover and mutation operations, new individuals are combined with some old individuals to form a new generation of population. This is an iterative process; the iterative process is repeated, that is, the above selection, crossover, mutation and new population generation process is repeated until the predetermined number of iterations is reached or the fitness is no longer significantly improved, and the final generation of population, that is, the target population, is obtained.
[0065] S203: Determine target configuration information of CNN based on the target population.
[0066] In practical applications, after obtaining the final generation of population, that is, the target population, the individual with the highest fitness is selected from the target population as the optimal solution. The CNN configuration corresponding to the individual is the optimized model structure and hyperparameters, that is, the target configuration information.
[0067] S204: Construct a third model based on the target configuration information.
[0068] Here, GA is a global optimization algorithm that can effectively search for the optimal parameter combination, thereby improving the accuracy and efficiency of time series prediction; CNN optimized by GA can be called GA-CNN, and its full English name is Convolutional Neural Networks Optimized By Genetic Algorithm; the CNN model optimized by the GA algorithm can find a better structure and hyperparameter configuration while maintaining a high accuracy, thereby improving the accuracy of foundation settlement prediction.
[0069] In practical applications, during the application of the prediction model, the actual settlement data obtained by the sensor can be used to optimize the parameters of the prediction model.
[0070] Based on this, in one embodiment, the method may further include: Determine second settlement information of the construction area in the first construction period based on third data obtained from the plurality of monitoring points; the third data is actually measured foundation settlement data; determining a difference between predicted settlement information and actual settlement information based on the first settlement information and the second settlement information; Based on the difference between the predicted settlement information and the actual settlement information, the model parameters of the first model are optimized.
[0071] In practical applications, a displacement sensor can be used to detect the actual settlement data of the foundation, that is, the third data.
[0072] In summary, the Internet of Things-based geotechnical engineering construction safety monitoring method provided in the embodiment of the present application determines the changes in geological factors by acquiring sensor data of factors affecting foundation settlement, so that foundation settlement can be predicted according to the changes in geological factors, and early warning of foundation settlement can be achieved, so that preventive strategies can be formulated in advance to reduce the impact of foundation settlement on geotechnical construction, which not only improves the timeliness of safety monitoring results, but also can improve the reliability and accuracy of the formulated strategies in the process of formulating geotechnical construction strategies based on safety monitoring results, thereby ensuring the safe and smooth progress of the geotechnical engineering construction process; further, due to the combination of the advantages of GA and CNN, the structure and hyperparameters of CNN are automatically optimized by GA, thereby improving the prediction accuracy and generalization ability of the model, thereby improving the accuracy and reliability of the safety monitoring results.
[0073] In order to implement the geotechnical engineering construction safety monitoring method based on the Internet of Things of the present application, the embodiment of the present application also provides a geotechnical engineering construction safety monitoring device based on the Internet of Things, which is arranged on an electronic device, such as Figure 3 As shown, the device may include: The data acquisition unit 301 is used to collect sensor data acquired by multiple monitoring points in the construction area in a first construction period to obtain first data; the first construction period is the current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions; The calculation unit 302 is used to determine the change of the settlement influencing factors in the first construction period based on the first data and the first model to obtain the sensing data characteristics; the first model includes a second model, and the second model is constructed based on the BP network; The processing unit 303 is used to predict the foundation settlement of the construction area in the first construction period based on the sensor data characteristics to obtain first settlement information.
[0074] In one embodiment, the first model further includes a third model, and the third model is obtained by optimizing CNN based on a genetic algorithm; the processing unit 303 may be specifically configured to: Continuously acquiring remote sensing images of the foundation in the construction area to obtain second data; Determine the terrain change of the construction area during the first construction period by using the second data and a third model in the first model to obtain image features; The sensor data features and the image features are fused to obtain a predicted situation of foundation settlement in the construction area during the first construction period, thereby generating first settlement information.
[0075] In one embodiment, the settlement influencing factors also include a second influencing factor determined based on the construction environment dimension.
[0076] In one embodiment, the first influencing factors include groundwater level, soil cohesion and soil internal friction angle; the second influencing factors include construction time, construction area and construction depth.
[0077] In one embodiment, the calculation unit 302 may be specifically configured to: Quantifying and assigning values to the first sensor data and the second sensor data to obtain a quantization result; the first sensor data includes sensor data related to the first influencing factor in the first data, and the second sensor data includes sensor data related to the second influencing factor in the first data; Based on the quantization result, generating a multi-dimensional feature vector; The multidimensional feature vector is used as the input of the second model in the first model, and the second model is used to determine the change of the settlement influencing factors in the first construction period to obtain the sensor data features.
[0078] In one embodiment, the apparatus further comprises a training unit; the training unit is configured to: Generate an initial population based on initial configuration information pre-configured for the CNN; the initial configuration information includes multiple CNN configuration strategies; Performing iterative optimization processing on the initial population based on a genetic algorithm to obtain a target population; Based on the target population, determining target configuration information of the CNN; Based on the target configuration information, a third model is constructed.
[0079] In one embodiment, the training unit is specifically used to: Selecting at least two second individuals from a plurality of first individuals in the initial population based on a fitness proportional selection algorithm; Cross-pairing the at least two second individuals based on a mating pool to obtain a pairing result; the size of the mating pool is different from the size of the initial population; Based on the pairing results, a target population is generated.
[0080] In one embodiment, when the at least two second individuals are cross-paired based on the mating pool, parents are randomly selected to produce an offspring through cross-pairing, and the parents are returned to the mating pool.
[0081] In one embodiment, the device further comprises an optimization unit, wherein the optimization unit is configured to: Determine second settlement information of the construction area in the first construction period based on third data obtained from the plurality of monitoring points; the third data is actually measured foundation settlement data; determining a difference between predicted settlement information and actual settlement information based on the first settlement information and the second settlement information; Based on the difference between the predicted settlement information and the actual settlement information, the model parameters of the first model are optimized.
[0082] It should be noted that: the geotechnical engineering construction safety monitoring device based on the Internet of Things provided in the above embodiment only uses the division of the above program modules as an example when performing geotechnical engineering construction safety monitoring based on the Internet of Things. In actual applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device is divided into different program modules to complete all or part of the processing described above. In addition, the geotechnical engineering construction safety monitoring device based on the Internet of Things provided in the above embodiment and the geotechnical engineering construction safety monitoring method embodiment based on the Internet of Things belong to the same concept. The specific implementation process is detailed in the method embodiment, which will not be repeated here.
[0083] It should be noted that: "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0084] In addition, the technical solutions described in the embodiments of the present application can be combined arbitrarily without conflict.
[0085] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application.
Claims
1. A geotechnical engineering construction safety monitoring method based on the Internet of Things, characterized in that: The method comprises: Collect sensor data acquired by multiple monitoring points in the construction area in a first construction period to obtain first data; the first construction period is a current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions; Based on the first data and the first model, determining the change of the settlement influencing factors in the first construction period to obtain the sensing data characteristics; the first model includes a second model, and the second model is constructed based on a back propagation network; Based on the sensor data features, the foundation settlement of the construction area during the first construction period is predicted to obtain first settlement information.
2. The method according to claim 1, characterized in that The first model also includes a third model, and the third model is obtained by optimizing a convolutional neural network based on a genetic algorithm; the first settlement information is obtained by predicting the foundation settlement of the construction area in the first construction period based on the sensor data characteristics, including: Continuously acquiring remote sensing images of the foundation in the construction area to obtain second data; Determine the terrain change of the construction area during the first construction period by using the second data and a third model in the first model to obtain image features; The sensor data features and the image features are fused to obtain a predicted situation of foundation settlement in the construction area during the first construction period, thereby generating first settlement information.
3. The method according to claim 2, characterized in that The settlement influencing factors also include a second influencing factor determined based on the construction environment dimension.
4. The method according to claim 3, characterized in that The first influencing factors include groundwater level, soil cohesion and soil internal friction angle; the second influencing factors include construction time, construction area and construction depth.
5. The method according to claim 3, characterized in that: The determining, based on the first data and the first model, changes in the settlement influencing factors during the first construction period to obtain sensing data features includes: Quantifying and assigning values to the first sensor data and the second sensor data to obtain a quantization result; the first sensor data includes sensor data related to the first influencing factor in the first data, and the second sensor data includes sensor data related to the second influencing factor in the first data; Based on the quantization result, generating a multi-dimensional feature vector; The multidimensional feature vector is used as the input of the second model in the first model, and the second model is used to determine the change of the settlement influencing factors in the first construction period to obtain the sensor data features.
6. The method according to any one of claims 3 to 5, characterized in that The method further comprises: Generate an initial population based on initial configuration information pre-configured for the convolutional neural network; the initial configuration information includes multiple convolutional neural network configuration strategies; Performing iterative optimization processing on the initial population based on a genetic algorithm to obtain a target population; Based on the target population, determining target configuration information of the convolutional neural network; Based on the target configuration information, a third model is constructed.
7. The method according to claim 6, characterized in that The iterative optimization process based on the genetic algorithm on the initial population to obtain the target population includes: Selecting at least two second individuals from a plurality of first individuals in the initial population based on a fitness proportional selection algorithm; Cross-pairing the at least two second individuals based on a mating pool to obtain a pairing result; the size of the mating pool is different from the size of the initial population; Based on the pairing results, a target population is generated.
8. The method according to claim 7, characterized in that When the at least two second individuals are cross-paired based on the mating pool, parents are randomly selected to produce an offspring through cross-pairing, and the parents are returned to the mating pool.
9. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Determine second settlement information of the construction area in the first construction period based on third data obtained from the plurality of monitoring points; the third data is actually measured foundation settlement data; determining a difference between predicted settlement information and actual settlement information based on the first settlement information and the second settlement information; Based on the difference between the predicted settlement information and the actual settlement information, the model parameters of the first model are optimized.
10. A geotechnical engineering construction safety monitoring device based on the Internet of Things, characterized in that: The device comprises: A data acquisition unit is used to collect sensor data acquired by multiple monitoring points in a construction area in a first construction period to obtain first data; the first construction period is a current construction period; the first data includes sensor data associated with settlement influencing factors, and the settlement influencing factors include first influencing factors determined based on geological dimensions; A calculation unit, configured to determine the change of the settlement influencing factors in the first construction period based on the first data and the first model, and obtain the sensing data characteristics; the first model includes a second model, and the second model is constructed based on a back propagation network; The processing unit is used to predict the foundation settlement of the construction area in the first construction period based on the characteristics of the sensor data to obtain first settlement information.