Slag leakage construction method and system based on deep learning algorithm

Abstract

The invention discloses a slag leakage construction method and system based on a deep learning algorithm, and belongs to the technical field of data acquisition and processing. Geological data are acquired through a sensor integrated on excavator bucket teeth, the geological data are identified by adopting a first deep learning algorithm, lithologic stratification data are determined, and the lithologic stratification data are used as a basis; the excavation speed of the excavator is intelligently adjusted, the excavator is controlled to excavate the leaked slag according to the adjusted excavation speed, collapse accidents can be effectively avoided, the construction safety of the leaked slag is guaranteed, the leaked slag is transported to the outside through the vertical slag leakage channel after the leaked slag is excavated, and the construction efficiency is improved. And a stroboscopic camera is adopted at the slag leakage outlet to collect a real-time slag leakage image, a second deep learning algorithm is adopted to predict the prediction probability of blockage of the slag leakage outlet caused by the current slag leakage amount on the basis of the real-time slag leakage image, and the opening size of the slag leakage outlet is adjusted according to the prediction probability, so that the problem of much raised dust or blockage is avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of data acquisition and processing, and particularly relates to a slag leakage construction method and system based on a deep learning algorithm. Background Art

[0002] In traditional layered excavation and vertical transportation technologies, when conducting deep and large foundation pit excavations, a layered excavation method is usually adopted. The layer thickness of the excavation is manually controlled, and vertical transportation equipment such as gantry cranes, tower cranes, and conveyor tracks, which are inclined or vertical, is set up to lift the slag leakage to the ground for stacking and external transportation. Currently, this method is widely used in the construction of deep foundation pits for urban subways and urban buildings; traditional deep and large foundation pit excavations are usually carried out according to a fixed layered thickness (such as 2m / layer), and the construction in an environment with interbedded sandstone and mudstone, fractures, and bedding contacts has not been considered yet. Work such as site clearance and tunneling speed often needs to be judged by on-site workers based on engineering experience. When encountering complex strata conditions, such as strata with a wide distribution of soft rock, it is prone to collapse, posing a relatively large engineering hazard. Moreover, the slag leakage outlet is manually set to a fixed opening size, which is prone to problems such as too large or too small an opening, resulting in more dust and blockage problems. Summary of the Invention

[0003] The present invention provides a slag leakage construction method and system based on a deep learning algorithm to solve the problems existing in the prior art.

[0004] In a first aspect, the present invention provides a slag leakage construction method based on a deep learning algorithm, including: When constructing a subway station ventilation shaft foundation pit, geological data is collected through sensors integrated on the teeth of an excavator bucket, and the first deep learning algorithm is used to identify the geological data to determine lithological stratification data; Based on the lithological stratification data, the excavation rate of the excavator is intelligently adjusted, and the excavator is controlled to conduct slag leakage excavation according to the adjusted excavation rate; After the slag leakage excavation, the slag leakage is transported to the outside through a vertical slag leakage channel, and a stroboscopic camera is used to collect real-time slag leakage images at the slag leakage outlet; Based on the real-time slag leakage images, the second deep learning algorithm is used to predict the probability of slag leakage outlet blockage caused by the current slag leakage volume, and the opening size of the slag leakage outlet is adjusted according to the prediction probability to complete the slag leakage construction.

[0005] In a possible implementation manner, it further includes: The slag leakage dust concentration at the slag leakage outlet is collected through a dust sensor, and when the slag leakage dust concentration is greater than a preset concentration threshold, the atomizing nozzles arranged at the slag leakage outlet are turned on for spraying to perform dust reduction treatment.

[0006] In a possible implementation, when constructing the foundation pit of the ventilation shaft of a subway station, geological data is collected through sensors integrated on the bucket teeth of an excavator, including: When constructing the foundation pit of the ventilation shaft of a subway station, pressure sensing data is collected through pressure sensors integrated on the bucket teeth of the excavator; Based on the pressure sensing data, the lithologic stratification data is preliminarily determined; wherein, the lithologic stratification data includes that the current tunneling layer is a sandstone layer, a mudstone layer or a plain fill layer; When the current tunneling layer is a sandstone layer or a mudstone layer, fracture density data is collected through a ground penetrating radar integrated on the bucket teeth of the excavator, and based on the fracture density data, the lithologic stratification data is corrected by using a first deep learning algorithm to obtain the corrected lithologic stratification data.

[0007] In a possible implementation, based on the lithologic stratification data, the tunneling rate of the excavator is intelligently adjusted, and the excavator is controlled to perform slag leakage excavation according to the adjusted tunneling rate, including: Dispatch the preset tunneling rate corresponding to the lithologic stratification data from the database, and control the excavator to perform slag leakage excavation according to the adjusted tunneling rate.

[0008] In a possible implementation, during the process of controlling the excavator to perform slag leakage excavation according to the adjusted tunneling rate, it further includes: continuously identifying the lithologic stratification data to determine the number of tunneling layers; wherein, when the lithologic stratification data changes once, the number of tunneling layers is incremented by one; Transmit the lithologic stratification data and the number of tunneling layers to the staff at the same time, so that the staff can manage the construction site of the foundation pit of the ventilation shaft of the subway station.

[0009] In a possible implementation, based on the real-time slag leakage image, a second deep learning algorithm is used to predict the prediction probability that the current slag leakage amount causes the slag leakage outlet to be blocked, and the opening size of the slag leakage outlet is adjusted according to the prediction probability, including: Obtain the slag particle size data in the real-time slag leakage image; Based on the slag particle size data, the slag leakage amount and the opening degree of the slag leakage outlet, a second deep learning algorithm is used to predict the prediction probability that the current slag leakage amount causes the slag leakage outlet to be blocked; When the prediction probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold, a feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet.

[0010] In a possible implementation, when the prediction probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold, a feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet, including: When the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold, the opening size of the slag leakage outlet is adjusted by a preset adjustment amount to obtain the adjusted opening size of the slag leakage outlet; Based on the adjusted opening size of the slag leakage outlet, the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is obtained again, and it is judged whether the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold. If so, return to the previous step; otherwise, continue to monitor.

[0011] In a possible implementation manner, it further includes: when the slag particle size data satisfies that the proportion of mudstone powder debris exceeds the preset quantity threshold, the opening size of the slag leakage outlet is directly adjusted to the preset opening size value.

[0012] In a possible implementation manner, the first deep learning algorithm is set as the RF algorithm; the second deep learning algorithm is set as the CNN-LSTM algorithm.

[0013] In a possible implementation manner, before using the second deep learning algorithm, the optimization is as follows: an intelligent optimization algorithm is used to train the hyperparameters of the second deep learning algorithm.

[0014] In a second aspect, the present invention provides a slag leakage construction system based on a deep learning algorithm, including: a field data acquisition module, an excavation rate control module, a slag leakage image acquisition module, and an intelligent slag discharging control module; The field data acquisition module is used to collect geological data through sensors integrated on the excavator bucket teeth during the construction of the subway station ventilation shaft foundation pit, and use the first deep learning algorithm to identify the geological data to determine the lithology stratification data; The excavation rate control module is used to intelligently adjust the excavation rate of the excavator based on the lithology stratification data, and control the excavator to perform slag leakage excavation according to the adjusted excavation rate; The slag leakage image acquisition module is used to transport the slag to the outside through the vertical slag leakage channel after the slag leakage excavation, and collect real-time slag leakage images with a stroboscopic camera at the slag leakage outlet; The intelligent slag discharging control module is used to predict the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked based on the real-time slag leakage image, and adjust the opening size of the slag leakage outlet according to the predicted probability to complete the slag leakage construction.

[0015] A slag leakage construction method and system based on a deep learning algorithm provided by the present invention collect geological data through sensors integrated on the bucket teeth of an excavator, and use a first deep learning algorithm to identify the geological data to determine lithological stratification data. Based on the lithological stratification data, the excavation rate of the excavator is intelligently adjusted, and the slag leakage excavation is controlled according to the adjusted excavation rate, which can effectively avoid the occurrence of collapse accidents, ensure the safety of slag leakage construction, and after the slag leakage excavation, the slag is transported to the outside through a vertical slag leakage channel, and a stroboscopic camera is used at the slag leakage outlet to collect real-time slag leakage images. Based on the real-time slag leakage images, a second deep learning algorithm is used to predict the probability of blockage of the slag leakage outlet caused by the current slag leakage volume, and the opening size of the slag leakage outlet is adjusted according to the prediction probability, avoiding the problems of more dust or blockage. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present invention and, together with the specification, are used to explain the principles of the present invention.

[0017] Figure 1 It is a flowchart of a slag leakage construction method based on a deep learning algorithm provided by an embodiment of the present invention.

[0018] Figure 2 It is a schematic structural diagram of a slag leakage construction system based on a deep learning algorithm provided by an embodiment of the present invention.

[0019] Among them, 201 - on-site data acquisition module, 202 - excavation rate control module, 203 - slag leakage image acquisition module, 204 - intelligent slag discharging control module.

[0020] Through the above-mentioned accompanying drawings, specific embodiments of the present invention have been shown, and there will be more detailed descriptions hereinafter. These drawings and written descriptions are not intended to limit the scope of the inventive concept in any way, but to illustrate the concept of the present invention to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] Here, the exemplary embodiments will be described in detail, and the examples are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0022] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Such as Figure 1As shown in the figure, an embodiment of the present invention provides a slag leakage construction method based on a deep learning algorithm, including: S101. When constructing the foundation pit of the ventilation shaft of the subway station, collect geological data through the sensors integrated on the excavator bucket teeth, and use the first deep learning algorithm to identify the geological data to determine the lithological stratification data; S102. Based on the lithological stratification data, intelligently adjust the excavation rate of the excavator, and control the excavator to perform slag leakage excavation according to the adjusted excavation rate; S103. After performing slag leakage excavation, transport the slag to the outside through the vertical slag leakage channel, and collect real-time slag leakage images at the slag leakage outlet using a stroboscopic camera; S104. Based on the real-time slag leakage images, use the second deep learning algorithm to predict the prediction probability of the current slag leakage volume causing blockage at the slag leakage outlet, and adjust the opening size of the slag leakage outlet according to the prediction probability to complete the slag leakage construction.

[0024] In a possible implementation manner, it further includes: Collect the dust concentration of the slag leakage at the slag leakage outlet through a dust sensor, and when the dust concentration of the slag leakage is greater than the preset concentration threshold, turn on the atomizing nozzles arranged at the slag leakage outlet for spraying to perform dust reduction treatment.

[0025] In a possible implementation manner, when constructing the foundation pit of the ventilation shaft of the subway station, collecting geological data through the sensors integrated on the excavator bucket teeth includes: When constructing the foundation pit of the ventilation shaft of the subway station, collect pressure sensing data through the pressure sensors integrated on the excavator bucket teeth; Based on the pressure sensing data, preliminarily determine the lithological stratification data; wherein, the lithological stratification data includes that the current excavation layer is a sandstone layer, a mudstone layer or a plain fill layer; Different pressure regions can be set for the sandstone layer, the mudstone layer or the plain fill layer, and by judging which pressure interval the pressure sensing data is in, the current lithological stratification data can be determined.

[0026] When the current excavation layer is a sandstone layer or a mudstone layer, collect fracture density data through the ground penetrating radar integrated on the excavator bucket teeth, and based on the fracture density data, use the first deep learning algorithm to correct the lithological stratification data to obtain the corrected lithological stratification data.

[0027] For example, the first deep learning algorithm can be used to classify the lithological stratification data, so as to further determine the lithological stratification data and improve the accuracy of the lithological stratification data.

[0028] In a possible implementation manner, based on the lithologic stratification data, the excavation rate of the excavator is intelligently adjusted, and the excavator is controlled to perform slag leakage excavation according to the adjusted excavation rate, including: Dispatch the preset excavation rate corresponding to the lithologic stratification data from the database, and control the excavator to perform slag leakage excavation according to the adjusted excavation rate.

[0029] By intelligently adjusting the excavation rate of the excavator, the collapse accident can be effectively avoided, and the engineering hidden danger is reduced.

[0030] In a possible implementation manner, during the process of controlling the excavator to perform slag leakage excavation according to the adjusted excavation rate, it further includes: continuously identifying the lithologic stratification data to determine the number of excavation layers; wherein, when the lithologic stratification data changes once, the number of excavation layers is incremented by one; Transmit the lithologic stratification data and the number of excavation layers to the staff at the same time, so that the staff can manage the construction site of the subway station ventilation shaft foundation pit.

[0031] For example, the staff can select different reinforcement schemes according to the different lithologic stratification data and the number of excavation layers to improve the construction safety.

[0032] In a possible implementation manner, based on the real-time slag leakage image, a second deep learning algorithm is used to predict the prediction probability of the current slag leakage volume causing the slag leakage outlet to be blocked, and the opening size of the slag leakage outlet is adjusted according to the prediction probability, including: Obtain the slag particle size data in the real-time slag leakage image; During the free fall of the slag, the camera can continuously scan at a line frequency of 172 kHz. In order to reduce the trailing phenomenon caused by the camera stroboscopic effect during the falling of the muck, the stroboscopic frequency of the light source is the same as the movement speed of the muck to eliminate the trailing. The particle contours in the real-time slag leakage image can be extracted, and the mutually adhered particles can be separated to improve the segmentation accuracy, and then the area equivalent diameter is used to generate the particle size distribution curve of the discharged muck.

[0033] Based on the slag particle size data, the slag leakage volume and the opening degree of the slag leakage outlet, a second deep learning algorithm is used to predict the prediction probability of the current slag leakage volume causing the slag leakage outlet to be blocked; Among them, the slag leakage volume can be represented by the working rate / working power of the slag leakage extraction device or the transportation device, and then the slag particle size data, the slag leakage volume and the opening degree of the slag leakage outlet can be used as the input data of the second deep learning algorithm, so as to realize the probability prediction of blockage.

[0034] When the prediction probability of the current slag leakage volume causing the slag leakage outlet to be blocked is greater than the preset probability threshold, a feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet.

[0035] What the second deep learning algorithm outputs are essentially the probabilities of blockage and non-blockage. The blockage probability can be used as the prediction probability. The greater the prediction probability, the greater the likelihood that the current slag leakage amount causes blockage at the slag leakage outlet.

[0036] In a possible implementation, when the prediction probability that the current slag leakage amount causes blockage at the slag leakage outlet is greater than a preset probability threshold, a feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet, including: When the prediction probability that the current slag leakage amount causes blockage at the slag leakage outlet is greater than the preset probability threshold, the opening size of the slag leakage outlet is adjusted by a preset adjustment amount to obtain the adjusted opening size of the slag leakage outlet; Based on the adjusted opening size of the slag leakage outlet, the prediction probability that the current slag leakage amount causes blockage at the slag leakage outlet is obtained again, and it is judged whether the prediction probability that the current slag leakage amount causes blockage at the slag leakage outlet is greater than the preset probability threshold. If so, return to the previous step; otherwise, continue to monitor.

[0037] Optionally, in addition to adjusting the slag leakage outlet, the angle of the slag discharge channel can also be adjusted to better control the slag discharge efficiency.

[0038] In addition to using the feedback adjustment algorithm to adjust the opening size of the slag leakage outlet, direct adjustment can also be performed. For example, when the proportion of large particle sandstone (>300mm) is between 30% and 50%, the opening and closing size of the slag leakage outlet is controlled at 70%, and the angle of the slag leakage channel is 45° with the horizontal direction. When the proportion of large particle sandstone (>300mm) is between 50% and 70%, the opening and closing size of the slag leakage outlet is controlled at 80%, and the angle of the slag leakage channel is 50° with the horizontal direction to accelerate the slag discharge efficiency and prevent slag discharge blockage. When the proportion of large particle sandstone (>300mm) exceeds 70%, the opening and closing size of the slag leakage outlet is controlled at 90%, and the angle of the slag leakage channel is 60° with the horizontal direction to accelerate the slag discharge efficiency and prevent slag discharge blockage. In other cases, blockage is not likely to occur and the dust is also small, and the slag can be discharged with a preset opening and closing size.

[0039] In a possible implementation, it further includes: when the slag particle size data satisfies that the proportion of mudstone powder exceeds a preset quantity threshold, the opening size of the slag leakage outlet is directly adjusted to a preset opening size value.

[0040] For example, slag with a particle size data less than 10mm can be considered as mudstone powder. When the proportion of mudstone powder exceeds 70%, the opening and closing size of the slag leakage outlet can be reduced to 60%, and the angle of the slag discharge channel can also be controlled to be reduced to 40% to control the slag discharge efficiency and reduce dust pollution at the same time.

[0041] In a possible implementation manner, the first deep learning algorithm is set as the RF (Random Forest) algorithm; the second deep learning algorithm is set as the CNN-LSTM (Convolutional Neural Network-Long Short-Term Memory) algorithm.

[0042] It should be noted that, in addition to using the above deep learning algorithms, other deep learning algorithms can also be used as the first deep learning algorithm and the second deep learning algorithm.

[0043] In a possible implementation manner, before the second deep learning algorithm is used, the optimization is as follows: an intelligent optimization algorithm is used to train the hyperparameters of the second deep learning algorithm.

[0044] In the embodiment of the present invention, using an intelligent optimization algorithm to train the hyperparameters of the second deep learning algorithm may include: A1. Initialize the hyperparameters of the second deep learning algorithm using a random initialization method or a chaotic mapping method to obtain multiple hyperparameter individuals; Wherein, the hyperparameter individual includes all or part of the hyperparameters to be trained of the second deep learning algorithm; A2. Obtain the loss function value corresponding to each hyperparameter individual, and determine the hyperparameter individual with the smallest loss function value as the optimal individual; The loss function value can be obtained through a root mean square loss function or a cross-entropy loss function. When obtaining the loss function value corresponding to each hyperparameter individual, a data pair of training data and training labels is required. The training data can be the slag leakage particle size data, slag leakage amount, and slag leakage outlet opening degree at historical moments, while the training label can be the actual occurrence of blockage under the condition of the training data. Blockage is set to 1, and non-blockage is set to 0.

[0045] A3. According to the optimal individual, perform a single search on the hyperparameter individuals, and the hyperparameter individuals after a single search are obtained as:

[0046]

[0047] Wherein, represents the t th training process and the n th hyperparameter individual, represents the n th hyperparameter individual after a single update, n i = 1, 2, …, N, where N represents the total number of hyperparameter individuals, represents the inertia weight corresponding to the n th hyperparameter individual, represents the optimal individual, u represents the first spiral constant, v represents the second spiral constant, e represents the natural constant, represents a random angle between (0, 2 π ), sin represents the sine function, cos represents the cosine function, and exp represents the exponential function with the natural constant e as the base. represents the n th fitness corresponding to the hyperparameter individual, represents the fitness corresponding to the optimal individual, and the fitness is obtained by taking the reciprocal of the loss function value; Optionally, when obtaining the fitness, to avoid a zero denominator, a preset constant can be added to the loss function value first. To avoid the influence of this preset constant on the fitness result, this preset constant can be set to 0.0001.

[0048] This one-time search process can fuse the original hyperparameter individuals with the optimal individual, occupying a new position in the solution space. It can not only effectively avoid collisions during the search process but also prevent the algorithm from falling into local optima. Moreover, adaptive fusion based on fitness can enable the algorithm to have higher convergence accuracy in the later stage. Finally, a spiral search function is introduced to make the search progress along an irregular path, enhancing the ability to find the global optimum.

[0049] A4. Perform a secondary search on the hyperparameter individuals after the one-time search to obtain the hyperparameter individuals after the secondary search as:

[0050] where, represents the t th hyperparameter individual after the h th one-time search during the th training process, h represents the information fusion individual randomly matched with the th hyperparameter individual after the one-time search. The information fusion individual is h different from the hyperparameter individual, = 1, 2,..., N, and M represents a positive integer generated by Lévy flight, represents a random number between (0, 1) generated by Lévy flight, represents the floor function, represents the first random number between (0, 1), represents the second random number between (0, 1); This secondary search process can randomly integrate the position information of other hyperparameter individuals and introduce an exponential function for control, which can effectively improve the search ability for unfamiliar regions. Meanwhile, it searches around the optimal position. When a hyperparameter individual strays outside the group, it will accelerate towards it; otherwise, it turns to fine search to enhance the search ability of the algorithm.

[0051] A5. Perform a tertiary search on the hyperparameter individuals after the secondary search, and the hyperparameter individuals after the tertiary search are:

[0052]

[0053] Among them, represents the t th hyperparameter individual after the secondary search in the k th training process, represents the k th hyperparameter individual after the tertiary search, k = 1, 2, …, N, represents the third random number between (0, 1), represents the fourth random number between (0, 1), represents pi, represents the random hyperparameter individual with a smaller loss function value randomly matched to the k th hyperparameter individual after the secondary search. When is the optimal individual, then is set as the hyperparameter individual with the second smallest loss function value; represents the optimal information learning factor, represents the fifth random number between (0, 1), represents the preset maximum number of training times; This tertiary search process enables hyperparameter individuals to learn the position information of the optimal individual in the solution space at an adaptive rate while learning the information of other better positions, improving the search accuracy and search speed of the algorithm.

[0054] A6. Perform a quaternary search on the hyperparameter individuals after the tertiary search, and the hyperparameter individuals after the quaternary search are:

[0055] Among them, represents the t th hyperparameter individual after the tertiary search in the q th training process, represents the q th hyperparameter individual after the quaternary search, q = 1, 2, …, N; Denote the worst individual, i.e., the hyperparameter individual with the largest loss function value; Denote the sixth random number between (0, 1).

[0056] This four - time search process can perform mutually exclusive information fusion, enabling the hyperparameter individuals to randomly move under the attraction of the optimal individual and the repulsion of the worst individual, thereby realizing the update of positions, improving the diversity of the population, and enabling the algorithm to jump out of the local optimal solution.

[0057] Optionally, the four - time search can be accepted only when the loss function value of the hyperparameter individual after the four - time search decreases, otherwise the four - time search is rejected. By introducing the greedy strategy, the convergence ability and convergence speed of the algorithm can be effectively guaranteed.

[0058] A7. Determine whether the current number of training times is greater than or equal to the preset maximum number of training times. If so, re - determine the optimal individual according to the hyperparameter individual after the four - time search, and use the hyperparameters in the re - determined optimal individual as the final hyperparameters of the second deep learning algorithm; otherwise, return to the step of determining the optimal individual.

[0059] The hyperparameter training of the second deep learning algorithm using the intelligent optimization algorithm provided by the embodiments of the present invention has a powerful global search ability compared with the prior art, is not easily trapped in the local optimum, effectively improves the training speed and training accuracy, ensures that the second deep learning algorithm after training can effectively make predictions, and finally improves the prediction accuracy of the current slag leakage amount causing the slag leakage outlet to be blocked, and prevents the occurrence of slag leakage blockage.

[0060] A slag leakage construction method based on a deep learning algorithm provided by the present invention collects geological data through sensors integrated on the bucket teeth of an excavator, and uses a first deep learning algorithm to identify the geological data to determine the lithology stratification data. Based on the lithology stratification data, the excavator tunneling rate is intelligently adjusted, and the excavator is controlled to perform slag leakage excavation according to the adjusted tunneling rate, which can effectively avoid the occurrence of collapse accidents, ensure the safety of slag leakage construction. After the slag leakage excavation, the slag is transported to the outside through a vertical slag leakage channel, and a stroboscopic camera is used to collect real - time slag leakage images at the slag leakage outlet. Based on the real - time slag leakage images, a second deep learning algorithm is used to predict the probability of the current slag leakage amount causing the slag leakage outlet to be blocked, and the opening size of the slag leakage outlet is adjusted according to the prediction probability, avoiding problems such as more dust or blockage.

[0061] As Figure 2 shown, an embodiment of the present invention provides a slag leakage construction system based on a deep learning algorithm, including: a field data acquisition module 201, a tunneling rate control module 202, a slag leakage image acquisition module 203, and an intelligent slag discharge control module 204; The on-site data acquisition module 201 is used to collect geological data through sensors integrated on the excavator bucket teeth during the construction of the ventilation shaft foundation pit of the subway station, and use the first deep learning algorithm to identify the geological data to determine the lithology stratification data; The excavation rate control module 202 is used to intelligently adjust the excavation rate of the excavator based on the lithology stratification data, and control the excavator to perform slag leakage excavation according to the adjusted excavation rate; The slag leakage image acquisition module 203 is used to transport the slag leakage to the outside through the vertical slag leakage channel after the slag leakage excavation, and collect real-time slag leakage images with a stroboscopic camera at the slag leakage outlet; The intelligent slag discharge control module 204 is used to predict the prediction probability of the slag leakage outlet being blocked caused by the current slag leakage volume based on the real-time slag leakage image, and adjust the opening size of the slag leakage outlet according to the prediction probability to complete the slag leakage construction.

[0062] Figure 2 The shown slag leakage construction system based on the deep learning algorithm can execute the above method, and its principle and beneficial effects are similar, so they will not be elaborated here.

[0063] Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0064] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0065] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one process Figure 1 one process or a plurality of processes and / or blocks Figure 1 one block or a plurality of blocks.

[0066] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process Figure 1 one process or a plurality of processes and / or blocks Figure 1 one block or a plurality of blocks.

[0067] Those of ordinary skill in the art will understand that all or part of the steps in implementing the above facts and methods can be completed by instructing relevant hardware through a program. The program involved or the described program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are introduced. The storage medium can be ROM / RAM, magnetic disk, optical disk, etc.

[0068] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A slag leakage construction method based on deep learning algorithms, characterized in that, Including: When constructing the foundation pit of the ventilation shaft of the subway station, geological data is collected through sensors integrated on the bucket teeth of the excavator, and the first deep learning algorithm is used to identify the geological data to determine the lithological stratification data; Based on the lithological stratification data, the excavation rate of the excavator is intelligently adjusted, and the excavator is controlled to perform slag leakage excavation according to the adjusted excavation rate; After the slag leakage excavation, the slag is transported to the outside through the vertical slag leakage channel, and a stroboscopic camera is used to collect real-time slag leakage images at the slag leakage outlet; Based on the real-time slag leakage image, the second deep learning algorithm is used to predict the probability of slag leakage outlet blockage caused by the current slag leakage volume, and the opening size of the slag leakage outlet is adjusted according to the predicted probability to complete the slag leakage construction.

2. The slag leakage construction method based on the deep learning algorithm according to claim 1, characterized in that It also includes: The dust sensor is used to collect the dust concentration of the slag leakage at the slag leakage outlet, and when the dust concentration of the slag leakage is greater than the preset concentration threshold, the atomizing nozzles arranged at the slag leakage outlet are turned on for spraying to perform dust reduction treatment.

3. The slag leakage construction method based on the deep learning algorithm according to claim 1, characterized in that When constructing the foundation pit of the ventilation shaft of the subway station, geological data is collected through sensors integrated on the bucket teeth of the excavator, including: When constructing the foundation pit of the ventilation shaft of the subway station, pressure sensing data is collected through the pressure sensors integrated on the bucket teeth of the excavator; Based on the pressure sensing data, the lithological stratification data is preliminarily determined; among them, the lithological stratification data includes that the current excavation layer is a sandstone layer, a mudstone layer or a plain fill layer; When the current excavation layer is a sandstone layer or a mudstone layer, the fracture density data is collected through the ground penetrating radar integrated on the bucket teeth of the excavator, and based on the fracture density data, the first deep learning algorithm is used to correct the lithological stratification data to obtain the corrected lithological stratification data, and the first deep learning algorithm is set as the RF algorithm.

4. The slag leakage construction method based on the deep learning algorithm according to claim 1, characterized in that Based on the lithological stratification data, the excavation rate of the excavator is intelligently adjusted, and the excavator is controlled to perform slag leakage excavation according to the adjusted excavation rate, including: The preset excavation rate corresponding to the lithological stratification data is scheduled from the database, and the excavator is controlled to perform slag leakage excavation according to the adjusted excavation rate.

5. The slag leakage construction method based on the deep learning algorithm according to claim 4, characterized in that During the process of controlling the excavator to perform slag leakage excavation according to the adjusted excavation rate, it also includes: continuously identifying the lithological stratification data to determine the number of excavation layers; among them, when the lithological stratification data changes once, the number of excavation layers is incremented by one; The lithological stratification data and the number of excavation layers are simultaneously transmitted to the staff to enable the staff to manage the construction site of the foundation pit of the ventilation shaft of the subway station.

6. The slag leakage construction method based on a deep learning algorithm according to claim 3, wherein Based on the real-time slag leakage image, the second deep learning algorithm is used to predict the probability of slag leakage outlet blockage caused by the current slag leakage volume, and the opening size of the slag leakage outlet is adjusted according to the predicted probability, including: Obtain the slag particle size data in the real-time slag leakage image; Based on the slag particle size data, the slag leakage volume and the opening degree of the slag leakage outlet, the second deep learning algorithm is used to predict the probability of slag leakage outlet blockage caused by the current slag leakage volume, and the second deep learning algorithm is set as the CNN-LSTM algorithm; When the predicted probability of slag leakage outlet blockage caused by the current slag leakage volume is greater than the preset probability threshold, the feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet.

7. The slag leakage construction method based on the deep learning algorithm according to claim 5, characterized in that, When the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold, a feedback adjustment algorithm is used to adjust the opening size of the slag leakage outlet, including: When the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold, the opening size of the slag leakage outlet is adjusted by a preset adjustment amount to obtain the adjusted opening size of the slag leakage outlet; Based on the adjusted opening size of the slag leakage outlet, the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is obtained again, and it is judged whether the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked is greater than the preset probability threshold. If so, return to the previous step; otherwise, continue to monitor.

8. The slag leakage construction method based on the deep learning algorithm according to claim 6, characterized in that, It also includes: When the slag particle size data satisfies that the proportion of mudstone powder exceeds the preset quantity threshold, the opening size of the slag leakage outlet is directly adjusted to the preset opening size value.

9. The slag leakage construction method based on the deep learning algorithm according to claim 6, characterized in that, Before using the second deep learning algorithm, it is optimized as: an intelligent optimization algorithm is used to train the hyperparameters of the second deep learning algorithm.

10. A slag leakage construction system based on a deep learning algorithm, characterized in that, It includes: A field data acquisition module, a digging rate control module, a slag leakage image acquisition module, and an intelligent slag discharging control module; The field data acquisition module is used to collect geological data through sensors integrated on the digging bucket teeth during the construction of the subway station ventilation shaft foundation pit, and use the first deep learning algorithm to identify the geological data to determine the lithology stratification data; The digging rate control module is used to intelligently adjust the digging rate of the excavator based on the lithology stratification data, and control the excavator to carry out slag leakage excavation according to the adjusted digging rate; The slag leakage image acquisition module is used to transport the slag to the outside through the vertical slag leakage channel after the slag leakage excavation, and collect real-time slag leakage images with a stroboscopic camera at the slag leakage outlet; The intelligent slag discharging control module is used to predict the predicted probability that the current slag leakage amount causes the slag leakage outlet to be blocked by using the second deep learning algorithm based on the real-time slag leakage image, and adjust the opening size of the slag leakage outlet according to the predicted probability to complete the slag leakage construction.

Images