Tunnel ventilation system control method and system based on artificial intelligence

Abstract

The invention provides a tunnel ventilation system control method and system based on artificial intelligence, and relates to the technical field of artificial intelligence. The method comprises the following steps: firstly, determining blast hole position distribution, blast hole size, blast hole detonation sequence and a rock stratum structure of a tunnel face on the basis of tunnel face image information; secondly, generating a tunnel face characteristic spectrum based on blast hole position distribution, blast hole sizes, a blast hole detonating sequence, a rock stratum structure of the tunnel face and the quantity of the explosive arranged in each blast hole on the tunnel face; then, inputting the tunnel face characteristic spectrum into a pre-trained blasting dust prediction model, and outputting the dust amount generated by tunnel face blasting by the blasting dust prediction model; and finally, the fan is controlled to work at the target output power based on the dust amount generated by tunnel face blasting. According to the scheme, the discharging time of dust generated by blasting can be accurately controlled, constructors are prevented from waiting for entering a site for a long time, and the tunnel construction efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based tunnel ventilation system control method and system. Background Art

[0002] In tunnel construction, blasting is often used to advance the heading face. After the blasting, a large amount of dust will be generated. In the relatively closed environment of the tunnel, this dust is difficult to be discharged, resulting in the tunnel construction environment being unable to quickly meet the requirements for construction personnel to enter the site, which will seriously affect the efficiency of tunnel construction.

[0003] In the prior art, ventilation ducts are often used to speed up the discharge of dust after blasting. After the blasting, on-site construction workers control the output power of the fan according to the amount of dust generated after the blasting to speed up the rapid discharge of the dust, so that the environment of the tunnel can meet the requirements for the construction workers to enter the site. However, human control relies heavily on personal experience, and there is a situation where inaccurate control causes the environment in the tunnel to still fail to meet the entry requirements after the preset time. The waiting time for entry becomes longer, affecting the efficiency of tunnel construction, and forcing construction workers to enter the site for work will cause irreversible damage to the bodies of construction workers. Summary of the invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, in a first aspect, the present invention provides a tunnel ventilation system control method based on artificial intelligence, which is applied to a computer device in a tunnel ventilation system, wherein the tunnel ventilation system further comprises an image acquisition device and a fan communicatively connected to the computer device, wherein the image acquisition device is arranged in the tunnel for acquiring image information of the tunnel face, and the fan is arranged in a ventilation duct of the tunnel, wherein the method comprises: Acquire the tunnel face image information acquired by the image acquisition device, and determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock layer structure of the tunnel face based on the tunnel face image information, wherein the rock layer structure includes the rock layer structure distribution around each of the blastholes; Acquire the amount of explosives set in each blasthole in the tunnel face, and generate a tunnel face feature map based on the blasthole position distribution, blasthole size, blasthole detonation sequence, stratum structure of the tunnel face, and the amount of explosives set in each blasthole, wherein the tunnel face feature map includes blasthole feature vectors corresponding to each blasthole, and the blasthole feature vectors include the position coordinates of the blasthole on the tunnel face, the blasthole size, blasthole detonation sequence, stratum structure distribution around the blasthole, and the amount of explosives in the blasthole; Inputting the tunnel face characteristic map into a pre-trained blasting dust prediction model, the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting; The fan is controlled to operate at a target output power based on the amount of dust generated by the face blasting, wherein the fan controls the dust concentration in the tunnel after the face blasting to be below a preset dust concentration after a preset time at the target output power.

[0005] In a possible implementation, the step of acquiring the tunnel face image information acquired by the image acquisition device, and determining the blasthole position distribution, blasthole size, blasthole detonation sequence and rock structure of the tunnel face based on the tunnel face image information includes: receiving the tunnel face image information acquired by the image acquisition device, and determining the actual size of the tunnel face based on the distance between the image acquisition device and the tunnel face; A coordinate system is constructed with the geometric center of the tunnel face as the origin to obtain the position coordinates of each blast hole on the tunnel face; Determining the detonation order of the blastholes based on the position coordinates of the blastholes; According to the ratio of the blasthole to the tunnel face and the actual size of the tunnel face, the size of the blasthole is obtained; Image recognition is performed on the tunnel face image information to determine the rock layer structure of the tunnel face.

[0006] In a possible implementation, the step of obtaining the amount of explosives set in each blast hole in the tunnel face and generating a tunnel face characteristic map based on the blast hole position distribution, blast hole size, blast hole detonation sequence, stratum structure of the tunnel face and the amount of explosives set in each blast hole includes: The amount of explosives set in each blast hole in the tunnel face is obtained by performing image analysis on the tunnel face image information or from the tunnel face blasting design plan; Constructing a blasthole feature vector of each blasthole based on the position coordinates of each blasthole on the tunnel face, the blasthole size, the blasthole detonation sequence, the rock formation structure distribution around the blasthole, and the amount of explosives in the blasthole; The tunnel face feature map is obtained according to the blasthole feature vectors corresponding to all the blastholes on the tunnel face.

[0007] In a possible implementation, before the step of inputting the tunnel face feature map into a pre-trained blasting dust prediction model, wherein the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting, the method further includes a step of training the blasting dust prediction model, which includes: Creating a training sample set, the training sample set comprising a face feature map sample and a corresponding dust amount sample value, wherein the face feature map sample is a face feature map corresponding to a blasted face, the dust amount sample value is the amount of dust actually generated by the blasted face, and the dust amount sample value comprises a first dust amount sample value and a second dust amount sample value; Inputting the face feature map samples in the training sample set into a neural network for training, and outputting the dust amount training results generated by blasting based on the face feature map samples, wherein the dust amount training results include a first dust amount training value generated by the face rock formation during blasting and a second dust amount training value generated by the explosive during blasting, and the neural network includes a feature extraction subnetwork, a feature fusion subnetwork, a first prediction subnetwork, and a second prediction subnetwork, wherein the first prediction subnetwork is used to predict the first dust amount training value generated by the face rock formation during blasting, and the second prediction subnetwork is used to predict the second dust amount training value generated by the explosive during blasting; Calculating a loss function value of the neural network based on the first dust amount sample value, the second dust amount sample value, the first dust amount training value, and the second dust amount value, and adjusting network parameters of the feature extraction subnetwork, the feature fusion subnetwork, the first prediction subnetwork, and the second prediction subnetwork when the loss function value is greater than a preset loss function value; Repeatedly inputting the face feature map samples in the training sample set into the neural network for training until the loss function value is no greater than the preset loss function value, and the training is completed. The neural network obtained by training is used as the blasting dust prediction model.

[0008] In a possible implementation, the step of inputting the tunnel face feature map samples in the training sample set into a neural network for training, and outputting the training results of the amount of dust generated by blasting based on the tunnel face feature map samples includes: Inputting the face feature map samples in the training sample set into the feature extraction subnetwork to extract features, thereby obtaining extracted sample features; Inputting the sample features into the feature fusion subnetwork for fusion, thereby obtaining a first fusion feature obtained by fusion of all sample features and a second fusion feature obtained by fusion of some sample features; The first fused feature is input into the first prediction sub-network for prediction to obtain the first dust amount training value, the second fused feature is input into the second prediction sub-network for prediction to obtain the second dust amount training value, and the dust amount training result is obtained from the first dust amount training value and the second dust amount training value.

[0009] In a possible implementation, before the step of controlling the fan to operate at a target output power based on the amount of dust generated by the tunnel face blasting, the method further includes: Pre-configure the mapping relationship between dust volume range and output power; The step of controlling the fan to operate at a target output power based on the amount of dust generated by the tunnel face blasting comprises: Comparing the dust amount generated by the tunnel face blasting with a pre-configured dust amount range to determine a target dust amount range within which the dust amount generated by the tunnel face blasting falls; The output power corresponding to the target dust amount range is used as the target output power, and the fan is controlled to operate at the target output power.

[0010] In a possible implementation, the tunnel ventilation system further includes a portable vital sign detector communicatively connected to the computer device, wherein the portable vital sign detector is used to detect physical characteristic information of the wearer, and the method further includes: Acquiring physical characteristic information detected by the portable physical sign detector, wherein the physical characteristic information includes heart rate information and blood oxygen saturation information; Determine whether the physical characteristic information is lower than a preset physical characteristic information threshold; if the physical characteristic information is lower than the preset physical characteristic information threshold, increase the output power of the fan.

[0011] In a possible implementation, the method further includes: If the physical characteristic information is lower than the preset physical characteristic information threshold, and the current output power of the fan has reached its maximum output power, an alarm message is generated; The alarm information is sent to the portable vital sign detector, wherein the alarm information includes warning content to remind the wearer to evacuate the tunnel as soon as possible.

[0012] In a second aspect, the present invention further provides a tunnel ventilation system, the tunnel ventilation system comprising a computer device, an image acquisition device and a fan in communication connection; The image acquisition device is used to acquire tunnel face image information and send the tunnel face image information to the computer device; The computer device is used to determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock formation structure of the tunnel face based on the tunnel face image information, wherein the rock formation structure includes the rock formation structure distribution around each blasthole; Acquire the amount of explosives set in each blasthole in the tunnel face, and generate a tunnel face feature map based on the blasthole position distribution, blasthole size, blasthole detonation sequence, stratum structure of the tunnel face, and the amount of explosives set in each blasthole, wherein the tunnel face feature map includes blasthole feature vectors corresponding to each blasthole, and the blasthole feature vectors include the position coordinates of the blasthole on the tunnel face, the blasthole size, blasthole detonation sequence, stratum structure distribution around the blasthole, and the amount of explosives in the blasthole; Inputting the tunnel face characteristic map into a pre-trained blasting dust prediction model, the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting; generating a control signal based on the amount of dust generated by the tunnel face blasting; The fan is used to operate with a target output power based on the control signal, wherein the fan controls the dust concentration in the tunnel after the face blasting to be below a preset dust concentration after a preset time at the target output power.

[0013] In a third aspect, the present invention further provides a computer device, comprising a processor, a computer-readable storage medium and a communication interface, wherein the computer-readable storage medium, the communication interface and the processor are connected via a bus system, the computer-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the computer-readable storage medium to execute the artificial intelligence-based tunnel ventilation system control method described in any possible implementation of the first aspect.

[0014] In a fourth aspect, the present invention further provides a computer-readable storage medium, in which instructions are stored, which, when executed, enable a computer to execute the artificial intelligence-based tunnel ventilation system control method in any possible implementation of the first aspect.

[0015] In the above scheme provided by the present invention, first, the blasthole position distribution, blasthole size, blasthole detonation sequence and rock structure of the face are determined based on the tunnel face image information; then, the face feature map is generated based on the blasthole position distribution, blasthole size, blasthole detonation sequence, rock structure of the face and the amount of explosives set in each blasthole; then, the face feature map is input into the pre-trained blasting dust prediction model, and the blasting dust prediction model outputs the amount of dust generated by the face blasting; finally, the fan is controlled to work at the target output power based on the amount of dust generated by the face blasting. The above scheme can predict the amount of dust generated by blasting according to the blasthole position distribution on the face, the amount of explosives in the blastholes and the rock structure of the face, and by controlling the output power of the fan, the tunnel can reach the requirement of being able to enter the site after a preset time after the blasting, so that the discharge time of the dust generated by the blasting can be accurately controlled, the waiting time for entering the site can be avoided, and the efficiency of tunnel construction can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0017] Figure 1 A block diagram of a tunnel ventilation system provided by an embodiment of the present invention; Figure 2 A schematic diagram of a flow chart of a tunnel ventilation system control method based on artificial intelligence provided by an embodiment of the present invention; Figure 3 for Figure 2 A schematic flow chart of sub-steps of step S10 in FIG. Figure 4 A schematic diagram of the steps of training a blasting dust prediction model provided by an embodiment of the present invention; Figure 5 Another block diagram of a tunnel ventilation system provided by an embodiment of the present invention; Figure 6 A partial flow chart of a tunnel ventilation system control method based on artificial intelligence provided by an embodiment of the present invention; Figure 7 A schematic diagram of the structural framework of a computer device for implementing the above-mentioned artificial intelligence-based tunnel ventilation system control method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention is described in detail below with reference to the accompanying drawings. The specific operation method in the method embodiment can also be applied to the device embodiment or the system embodiment.

[0019] Before introducing the specific solution provided by this embodiment, the application scenario of the tunnel ventilation system to which the specific solution is applicable is first introduced. Figure 1 In this embodiment, the tunnel ventilation system 1 includes a computer device 10, an image acquisition device 20 and a fan 30 that are communicatively connected, wherein the image acquisition device 20 is installed in the tunnel to at least acquire image information of the face, and the fan 30 is arranged in the ventilation duct of the tunnel, and the fan 30 provides power for the air in the ventilation duct to enable the air to move rapidly in the ventilation duct. The image acquisition device 20 sends the acquired video image information to the computer device 10, and the computer device 10 performs analysis and processing based on the above video image information. The image acquisition device 20 includes a camera with adjustable image acquisition direction.

[0020] Combine the following Figure 1 The application scenario shown is an exemplary description of the tunnel ventilation system control method based on artificial intelligence provided in the embodiment of the present application. Figure 2 The artificial intelligence-based tunnel ventilation system control method provided in the embodiment of the present application can be executed by the aforementioned computer device 10. In other embodiments, the order of some steps in the artificial intelligence-based tunnel ventilation system control method in the embodiment of the present application can be interchanged according to actual needs, or some of the steps can be omitted or deleted. The detailed steps of the artificial intelligence-based tunnel ventilation system control method executed by the computer device 10 are introduced as follows.

[0021] Step S10, obtaining tunnel face image information acquired by an image acquisition device, and determining blast hole position distribution, blast hole size, blast hole detonation sequence and rock layer structure on the tunnel face based on the tunnel face image information.

[0022] The tunnel face is a working face that is continuously advanced in the tunnel excavation. Before blasting the tunnel face, blast holes need to be opened on the tunnel face and explosives need to be filled in the blast holes. Before blasting the tunnel face, the tunnel face image information is acquired through the image acquisition device 20, and the tunnel face image information is analyzed to determine the blast hole position distribution, blast hole size, blast hole detonation sequence and rock layer structure of the tunnel face.

[0023] Step S20, obtaining the amount of explosives set in each blast hole in the tunnel face, and generating a tunnel face characteristic map based on the blast hole position distribution, blast hole size, blast hole detonation sequence, rock layer structure of the tunnel face and the amount of explosives set in each blast hole.

[0024] In this embodiment, the tunnel face feature map includes a blasthole feature vector corresponding to each blasthole, and the blasthole feature vector includes the position coordinates of the blasthole on the tunnel face, the size of the blasthole, the blasthole detonation sequence, the rock layer structure distribution around the blasthole, and the amount of explosives in the blasthole. Exemplarily, the tunnel face feature map can be expressed as {Q1, Q2, Q3...Qi...Qn}, where Qi represents the blasthole feature vector of the blasthole numbered i, and the blasthole number is a positive integer. The blasthole characteristic vector Qi=((xi,yi), ci, ai, bi, di), wherein (xi,yi) represents the position coordinates of the blasthole numbered i on the tunnel face; ci represents the size of the blasthole numbered i, illustratively, the size of the blasthole may be the diameter of the blasthole; ai represents the detonation order of the blasthole numbered i, which may be represented by a specific numerical value, such as a value of 1 represents the first detonation, a value of 2 represents the second detonation, and so on; bi represents the rock formation structure distribution around the blasthole numbered i, wherein the rock formation structure distribution includes different rock formation structure classifications, such as granite, basalt and quartzite; di represents the amount of explosives in the blasthole numbered i.

[0025] Step S30, inputting the tunnel face feature map into a pre-trained blasting dust prediction model, and the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting.

[0026] Among them, the blasting dust prediction model can predict the amount of dust generated by the corresponding face after blasting based on the input face characteristic map. Specifically, the blasting dust prediction model predicts the amount of dust generated by the face after blasting based on the position coordinates of the blastholes on the face, the size of the blastholes, the blasthole detonation sequence, the rock structure distribution around the blastholes, and the amount of explosives in the blastholes.

[0027] Step S40, controlling the fan to operate at a target output power based on the amount of dust generated by the blasting at the tunnel face.

[0028] In this embodiment, the computer device 10 can determine the target output power of the fan 30 according to the amount of dust generated by the face blasting, and send a control instruction to the fan 30 to control the fan 30 to operate according to the target output power. When the fan 30 operates with the target output power, the dust concentration in the tunnel after the face blasting can be controlled below the preset dust concentration after a preset time, that is, the dust concentration in the tunnel after the face blasting can be reduced to a dust concentration range within the specified time that allows construction personnel to enter the site.

[0029] The above technical scheme can predict the amount of dust generated by blasting based on the distribution of blasthole positions on the tunnel face, the amount of explosives in the blastholes, and the rock structure of the tunnel face, and control the output power of the fan so that the tunnel can reach the requirements for entering the site after a preset time after blasting. In this way, the discharge time of dust generated by blasting can be accurately controlled, avoiding a long waiting time for entering the site and improving the efficiency of tunnel construction.

[0030] For further information, please refer to Figure 3 In this embodiment, step S10 can be implemented in the following manner.

[0031] Sub-step S101, receiving image information of a tunnel face acquired by an image acquisition device, and determining the actual size of the tunnel face based on the distance between the image acquisition device and the tunnel face.

[0032] For example, the actual size of the tunnel face can be determined based on the distance between the image acquisition device 20 and the tunnel face, the depth of field of the tunnel face in the tunnel tunnel face image information, and the device parameters of the image acquisition device 20. It can be understood that in other embodiments, other methods can also be used to determine the actual size of the tunnel face.

[0033] Sub-step S102, constructing a coordinate system with the geometric center of the tunnel face as the origin, and obtaining the position coordinates of each blast hole on the tunnel face.

[0034] For example, a coordinate system can be constructed with the geometric center of the tunnel face as the origin to determine the position coordinates of each blasthole on the tunnel face. The blastholes in the coordinate system can also be numbered. The specific numbering rule can be to number them in order from top to bottom and from left to right.

[0035] Sub-step S103, determining the detonation sequence of the blast holes based on the position coordinates of the blast holes.

[0036] For example, the type of blastholes can be determined based on the position coordinates of the blastholes on the tunnel face and the position distribution of different types of blastholes, wherein the blasthole types include slot holes, caving holes, and peripheral holes. The detonation order of the blastholes is determined based on the types of blastholes and the position coordinates of the corresponding blastholes on the tunnel face. For example, the detonation can be performed in the order of slot holes, caving holes, and peripheral holes. The detonation intervals of different types of blastholes can be set according to construction requirements, for example, an interval of 200ms.

[0037] Sub-step S104, obtaining the blasthole size according to the ratio of the blasthole to the tunnel face and the actual size of the tunnel face.

[0038] Exemplarily, the blasthole size includes the aperture of the blasthole, and the aperture of the blasthole can be obtained based on the ratio of the blasthole to the tunnel face and the actual size of the tunnel face. Furthermore, the blasthole size can also include the depth of the blasthole. Unless otherwise specified, the depth of the blasthole can be pre-set according to the actual construction requirements of the blasthole.

[0039] Sub-step S105, performing image recognition on the tunnel face image information to determine the rock layer structure of the tunnel face.

[0040] In this step, the tunnel face image information can be input into the trained rock structure recognition model for recognition, wherein the rock structure recognition model can automatically recognize the rock structure existing in the image based on the input image, and the rock structure recognition model can be trained by a large number of training samples with rock structures.

[0041] Further, in this embodiment, step S20 can be implemented in the following manner.

[0042] Firstly, the amount of explosives set in each blast hole in the tunnel face is obtained by performing image analysis on the tunnel face image information or from the tunnel face blasting design plan.

[0043] That is, in this embodiment, the amount of explosives set in each blast hole in the tunnel face can be obtained by performing image analysis on the tunnel face image information, or the amount of explosives set in each blast hole in the tunnel face can be obtained from the tunnel face blasting design plan. Among them, a detonator can be used as a basic unit of explosives, and the amount of explosives can be the number of detonators.

[0044] Then, the blasthole feature vector of each blasthole is constructed based on the position coordinates of each blasthole on the tunnel face, the blasthole size, the blasthole detonation sequence, the rock structure distribution around the blasthole, and the amount of explosives in the blasthole.

[0045] Exemplarily, the blasthole feature vector can be expressed as Qi=((xi,yi), ci, ai, bi, di), where the meanings of the parameters are the same as described before and will not be repeated here.

[0046] Then, the tunnel face feature map is obtained according to the blasthole feature vectors corresponding to all blastholes on the tunnel face.

[0047] Optionally, the tunnel face characteristic map may be represented as {Q1, Q2, Q3...Qi...Qn}, wherein Qi represents a blasthole characteristic vector of a blasthole numbered i.

[0048] Before step S30, the tunnel ventilation system control method based on artificial intelligence provided in this embodiment also includes the step of training a blasting dust prediction model. Figure 4, this step can be achieved by the following steps.

[0049] Step S201: Create a training sample set, where the training sample set includes a tunnel face feature map sample and a corresponding dust volume sample value.

[0050] The tunnel face characteristic map sample is the tunnel face characteristic map corresponding to the blasted tunnel face, the dust amount sample value is the actual dust amount generated by the blasted tunnel face, and the dust amount sample value includes a first dust amount sample value and a second dust amount sample value, wherein the first dust amount sample value corresponds to the rock dust generated by the blasted tunnel face, and the second dust amount sample value corresponds to the explosive dust generated by the explosives when the tunnel face is blasted, wherein the explosive dust includes toxic gases generated by the explosion of explosives.

[0051] Step S202: inputting the tunnel face characteristic map samples in the training sample set into a neural network for training, and outputting the training results of the dust amount generated by blasting based on the tunnel face characteristic map samples.

[0052] In this embodiment, the dust amount training result includes a first dust amount training value generated by the rock formation at the tunnel face during blasting and a second dust amount training value generated by the explosives during blasting.

[0053] The neural network includes a feature extraction subnetwork, a feature fusion subnetwork, a first prediction subnetwork, and a second prediction subnetwork, wherein the first prediction subnetwork is used to predict a first dust amount training value generated by the face rock formation during blasting, and the second prediction subnetwork is used to predict a second dust amount training value generated by the explosives during blasting.

[0054] Optionally, in this embodiment, step S202 may be implemented in the following manner.

[0055] Firstly, the face feature map samples in the training sample set are input into the feature extraction subnetwork for feature extraction to obtain the extracted sample features.

[0056] Next, the sample features are input into the feature fusion subnetwork for fusion to obtain a first fusion feature obtained by fusion of all sample features and a second fusion feature obtained by fusion of some sample features.

[0057] Then, the first fused feature is input into the first prediction sub-network for prediction to obtain a first dust amount training value, and the second fused feature is input into the second prediction sub-network for prediction to obtain a second dust amount training value. The dust amount training result is obtained from the first dust amount training value and the second dust amount training value.

[0058] It can be understood that the process of predicting the amount of dust generated by the tunnel face blasting in step S30 is the same as the above process, and will not be repeated here.

[0059] Step S203, based on the first dust amount sample value, the second dust amount sample value, the first dust amount training value and the second dust amount value, the loss function value of the neural network is calculated, and when the loss function value is greater than the preset loss function value, the network parameters of the feature extraction subnetwork, the feature fusion subnetwork, the first prediction subnetwork and the second prediction subnetwork are adjusted.

[0060] Exemplarily, a first loss function value can be obtained based on a first dust amount sample value and a first dust amount training value, a second loss function value can be obtained based on a second dust amount sample value and a second dust amount training value, and a loss function value can be obtained based on the first loss function value and the second loss function value, wherein the loss function value can be a weighted sum of the first loss function value and the second loss function value. When the loss function value is greater than a preset loss function value, the network parameters of each sub-network in the neural network are adjusted.

[0061] Step S204, repeatedly inputting the face feature map samples in the training sample set into the neural network for training until the loss function value is no greater than the preset loss function value, and the training is completed, and the neural network obtained by the training is used as the blasting dust prediction model.

[0062] Furthermore, in this embodiment, before step S40, the tunnel ventilation system control method based on artificial intelligence provided by this embodiment also includes the following steps.

[0063] The mapping relationship between the dust amount range and the output power is preconfigured. Specifically, the mapping relationship between the dust amount range and the output power is preconfigured in the computer device 10. With such a design, the output power of the fan 30 can be accurately controlled based on the dust amount generated by the blasting, so as to ensure that the environment in the tunnel can meet the requirements of the construction personnel entering the site for construction after a preset time after the blasting. On the other hand, the use of matching output power can also make the fan 30 work at the minimum output power that meets the above requirements, thereby reducing power consumption.

[0064] In this embodiment, step S40 can be specifically implemented in the following manner.

[0065] Firstly, the dust amount generated by the tunnel face blasting is compared with a pre-configured dust amount range to determine the target dust amount range within which the dust amount generated by the tunnel face blasting falls.

[0066] Next, the output power corresponding to the target dust amount range is used as the target output power, and the fan is controlled to operate at the target output power.

[0067] Further, in this embodiment, please refer to Figure 5The tunnel ventilation system 1 further includes a portable vital sign detector 40 that is communicatively connected to the computer device 10. The portable vital sign detector 40 is used to detect the wearer's physical characteristic information, wherein the physical characteristic information includes heart rate information and blood oxygen saturation information. Exemplarily, the portable vital sign detector 40 may be a smart watch or a smart bracelet. Figure 6 , the artificial intelligence-based tunnel ventilation system control method provided in this embodiment also includes the following steps.

[0068] Step S50, obtaining physical characteristic information detected by the portable physical sign detector.

[0069] The portable vital sign detector 40 detects and sends the real-time detected physical characteristic information to the computer device 10 .

[0070] Step S60, determining whether the physical characteristic information is lower than a preset physical characteristic information threshold value, if the physical characteristic information is lower than the preset physical characteristic information threshold value, increasing the output power of the fan.

[0071] Exemplarily, the threshold of the physical characteristic information may be pre-configured in the computer device 10 in advance, wherein the threshold of the physical characteristic information may be a fixed value or may not be a fixed value. For example, when the threshold of the physical characteristic information is a fixed value, it may be a heart rate exceeding 120 beats / minute, or a blood oxygen saturation lower than 90%, where the blood oxygen saturation is arterial blood oxygen saturation. For another example, when the threshold of the physical characteristic information is not a fixed value, it may be a heart rate exceeding 20% ​​of the average heart rate, or a blood oxygen saturation lower than 10% of the average blood oxygen saturation. The specific threshold may vary from person to person.

[0072] When the physical characteristic information is lower than the preset physical characteristic information threshold, the output power of the fan 30 is increased by one gear to increase the output power of the fan so as to increase the supply of fresh air in the tunnel, reduce the impact of the adverse environment of the tunnel on the human body, and ensure the health of the construction workers in the tunnel.

[0073] Further, please refer again to Figure 6 , the artificial intelligence-based tunnel ventilation system control method provided in this embodiment also includes the following steps.

[0074] Step S70: If the physical characteristic information is lower than a preset physical characteristic information threshold and the current output power of the wind turbine has reached its maximum output power, an alarm message is generated.

[0075] In this case, either the fan 30 is not enough to change the environment in the tunnel, and the environment in the tunnel is not suitable for construction; or the construction workers themselves are unwell and need to rest or seek medical treatment.

[0076] Step S80, sending the alarm information to the portable vital sign detector.

[0077] The computer device 10 generates an alarm message and sends it to the portable vital sign detector 40, wherein the alarm message includes a warning content to remind the wearer (construction worker) to evacuate the tunnel as soon as possible, thereby reminding the construction workers to evacuate the tunnel in time, avoiding tunnel accidents and ensuring the safety of the construction workers.

[0078] Please refer again Figure 1 This embodiment provides a tunnel ventilation system. The tunnel ventilation system 1 includes a computer device 10, an image acquisition device 20 and a fan 30 that are communicatively connected.

[0079] The image acquisition device 20 is used to acquire tunnel face image information and send the tunnel face image information to the computer device 10 .

[0080] The computer device 10 is used to determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock formation structure of the tunnel face based on the tunnel face image information, wherein the rock formation structure includes the rock formation structure distribution around each blasthole; Obtain the amount of explosives set in each blast hole in the tunnel face, and generate a tunnel face feature map based on the blast hole position distribution on the tunnel face, the blast hole size, the blast hole detonation sequence, the stratum structure of the tunnel face, and the amount of explosives set in each blast hole. The tunnel face feature map includes a blast hole feature vector corresponding to each blast hole. The blast hole feature vector includes the position coordinates of the blast hole on the tunnel face, the blast hole size, the blast hole detonation sequence, the stratum structure distribution around the blast hole, and the amount of explosives in the blast hole; The tunnel face feature map is input into a pre-trained blasting dust prediction model, and the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting; A control signal is generated based on the amount of dust generated by blasting at the tunnel face.

[0081] The fan 30 is used to operate with a target output power based on the control signal, wherein the fan controls the dust concentration in the tunnel after the face blasting to be below a preset dust concentration after a preset time at the target output power.

[0082] Please refer to Figure 7 , Figure 7 FIG. 1 is a schematic diagram of the hardware structure of a computer device 10 for implementing the above-mentioned tunnel ventilation system control method based on artificial intelligence provided by an embodiment of the present disclosure. The computer device 10 can be implemented on a cloud server. Figure 7 As shown, the computer device 10 may include a processor 101 , a computer-readable storage medium 102 , a bus 103 , and a communication interface 104 .

[0083] In the specific implementation process, at least one processor 101 executes computer-executable instructions stored in a computer-readable storage medium 102, so that the processor 101 can execute the tunnel ventilation system control method based on artificial intelligence in the above method embodiment, wherein the processor 101, the computer-readable storage medium 102 and the communication interface 104 are connected via a bus 103, and the processor 101 can be used to control the sending and receiving actions of the communication interface 104.

[0084] The specific implementation process of the processor 101 can refer to the various method embodiments executed by the above-mentioned computer device 10. The implementation principles and technical effects are similar and will not be repeated in this embodiment.

[0085] The computer-readable storage medium 102 may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk storage.

[0086] The bus 103 can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present invention is not limited to only one bus or one type of bus.

[0087] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above-mentioned tunnel ventilation system control method based on artificial intelligence is implemented.

[0088] In summary, the technical solution provided by the embodiment of the present invention is, first, to determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock structure of the face on the face based on the tunnel face image information; then, to generate a face feature map based on the blasthole position distribution, blasthole size, blasthole detonation sequence, rock structure of the face and the amount of explosives set in each blasthole; then, to input the face feature map into a pre-trained blasting dust prediction model, and the blasting dust prediction model outputs the amount of dust generated by the face blasting; finally, to control the fan to work at the target output power based on the amount of dust generated by the face blasting. The above scheme can predict the amount of dust generated by blasting according to the blasthole position distribution on the face, the amount of explosives in the blastholes and the rock structure of the face, and by controlling the output power of the fan, the tunnel can reach the requirement of being able to enter the site after a preset time after the blasting, so that the discharge time of the dust generated by the blasting can be accurately controlled, the waiting time for entering the site can be avoided, and the efficiency of tunnel construction can be improved.

[0089] The above is a description of a specific embodiment of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A tunnel ventilation system control method based on artificial intelligence, characterized in that: A computer device applied to a tunnel ventilation system, wherein the tunnel ventilation system further comprises an image acquisition device and a fan communicatively connected to the computer device, wherein the image acquisition device is arranged in the tunnel for acquiring image information of the tunnel face, and the fan is arranged in a ventilation duct of the tunnel, wherein the method comprises: Acquire the tunnel face image information acquired by the image acquisition device, and determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock layer structure of the tunnel face based on the tunnel face image information, wherein the rock layer structure includes the rock layer structure distribution around each of the blastholes; Acquire the amount of explosives set in each blasthole in the tunnel face, and generate a tunnel face feature map based on the blasthole position distribution, blasthole size, blasthole detonation sequence, stratum structure of the tunnel face, and the amount of explosives set in each blasthole, wherein the tunnel face feature map includes blasthole feature vectors corresponding to each blasthole, and the blasthole feature vectors include the position coordinates of the blasthole on the tunnel face, the blasthole size, blasthole detonation sequence, stratum structure distribution around the blasthole, and the amount of explosives in the blasthole; Inputting the tunnel face characteristic map into a pre-trained blasting dust prediction model, the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting; The fan is controlled to operate at a target output power based on the amount of dust generated by the face blasting, wherein the fan controls the dust concentration in the tunnel after the face blasting to be below a preset dust concentration after a preset time at the target output power.

2. The tunnel ventilation system control method based on artificial intelligence according to claim 1, characterized in that: The step of acquiring the tunnel face image information acquired by the image acquisition device, and determining the blasthole position distribution, blasthole size, blasthole detonation sequence and rock layer structure of the tunnel face based on the tunnel face image information comprises: receiving the tunnel face image information acquired by the image acquisition device, and determining the actual size of the tunnel face based on the distance between the image acquisition device and the tunnel face; A coordinate system is constructed with the geometric center of the tunnel face as the origin to obtain the position coordinates of each blast hole on the tunnel face; Determining the detonation order of the blastholes based on the position coordinates of the blastholes; According to the ratio of the blasthole to the tunnel face and the actual size of the tunnel face, the size of the blasthole is obtained; Image recognition is performed on the tunnel face image information to determine the rock layer structure of the tunnel face.

3. The tunnel ventilation system control method based on artificial intelligence according to claim 1, characterized in that: The step of obtaining the amount of explosives set in each blast hole in the tunnel face and generating a tunnel face characteristic map based on the blast hole position distribution, blast hole size, blast hole detonation sequence, stratum structure of the tunnel face and the amount of explosives set in each blast hole includes: The amount of explosives set in each blast hole in the tunnel face is obtained by performing image analysis on the tunnel face image information or from the tunnel face blasting design plan; Constructing a blasthole feature vector of each blasthole based on the position coordinates of each blasthole on the tunnel face, the blasthole size, the blasthole detonation sequence, the rock formation structure distribution around the blasthole, and the amount of explosives in the blasthole; The tunnel face feature map is obtained according to the blasthole feature vectors corresponding to all the blastholes on the tunnel face.

4. The tunnel ventilation system control method based on artificial intelligence according to claim 1, characterized in that: Before the step of inputting the tunnel face feature map into a pre-trained blasting dust prediction model, wherein the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting, the method further includes the step of training the blasting dust prediction model, which step includes: Creating a training sample set, the training sample set comprising a face feature map sample and a corresponding dust amount sample value, wherein the face feature map sample is a face feature map corresponding to a blasted face, the dust amount sample value is the amount of dust actually generated by the blasted face, and the dust amount sample value comprises a first dust amount sample value and a second dust amount sample value; Inputting the face feature map samples in the training sample set into a neural network for training, and outputting the dust amount training results generated by blasting based on the face feature map samples, wherein the dust amount training results include a first dust amount training value generated by the face rock formation during blasting and a second dust amount training value generated by the explosive during blasting, and the neural network includes a feature extraction subnetwork, a feature fusion subnetwork, a first prediction subnetwork, and a second prediction subnetwork, wherein the first prediction subnetwork is used to predict the first dust amount training value generated by the face rock formation during blasting, and the second prediction subnetwork is used to predict the second dust amount training value generated by the explosive during blasting; Calculating a loss function value of the neural network based on the first dust amount sample value, the second dust amount sample value, the first dust amount training value, and the second dust amount value, and adjusting network parameters of the feature extraction subnetwork, the feature fusion subnetwork, the first prediction subnetwork, and the second prediction subnetwork when the loss function value is greater than a preset loss function value; Repeatedly inputting the face feature map samples in the training sample set into the neural network for training until the loss function value is no greater than the preset loss function value, and the training is completed. The neural network obtained by training is used as the blasting dust prediction model.

5. The tunnel ventilation system control method based on artificial intelligence according to claim 4, characterized in that: The step of inputting the tunnel face characteristic map samples in the training sample set into a neural network for training, and outputting the training results of the dust amount generated by blasting based on the tunnel face characteristic map samples, comprises: Inputting the face feature map samples in the training sample set into the feature extraction subnetwork to extract features, thereby obtaining extracted sample features; Inputting the sample features into the feature fusion subnetwork for fusion, thereby obtaining a first fusion feature obtained by fusion of all sample features and a second fusion feature obtained by fusion of some sample features; The first fused feature is input into the first prediction sub-network for prediction to obtain the first dust amount training value, the second fused feature is input into the second prediction sub-network for prediction to obtain the second dust amount training value, and the dust amount training result is obtained from the first dust amount training value and the second dust amount training value.

6. The tunnel ventilation system control method based on artificial intelligence according to any one of claims 1 to 5, characterized in that: Before the step of controlling the fan to operate at a target output power based on the amount of dust generated by the tunnel face blasting, the method further includes: Pre-configure the mapping relationship between dust volume range and output power; The step of controlling the fan to operate at a target output power based on the amount of dust generated by the tunnel face blasting comprises: Comparing the dust amount generated by the tunnel face blasting with a pre-configured dust amount range to determine a target dust amount range within which the dust amount generated by the tunnel face blasting falls; The output power corresponding to the target dust amount range is used as the target output power, and the fan is controlled to operate at the target output power.

7. The tunnel ventilation system control method based on artificial intelligence according to any one of claims 1 to 5, characterized in that: The tunnel ventilation system further comprises a portable vital sign detector communicatively connected to the computer device, wherein the portable vital sign detector is used to detect physical characteristic information of the wearer, and the method further comprises: Acquiring physical characteristic information detected by the portable physical sign detector, wherein the physical characteristic information includes heart rate information and blood oxygen saturation information; Determine whether the physical characteristic information is lower than a preset physical characteristic information threshold; if the physical characteristic information is lower than the preset physical characteristic information threshold, increase the output power of the fan.

8. The tunnel ventilation system control method based on artificial intelligence according to claim 7, characterized in that: The method further comprises: If the physical characteristic information is lower than the preset physical characteristic information threshold, and the current output power of the fan has reached its maximum output power, an alarm message is generated; The alarm information is sent to the portable vital sign detector, wherein the alarm information includes warning content to remind the wearer to evacuate the tunnel as soon as possible.

9. A tunnel ventilation system, characterized in that: The tunnel ventilation system includes a computer device, an image acquisition device and a fan connected in communication; The image acquisition device is used to acquire tunnel face image information and send the tunnel face image information to the computer device; The computer device is used to determine the blasthole position distribution, blasthole size, blasthole detonation sequence and rock formation structure of the tunnel face based on the tunnel face image information, wherein the rock formation structure includes the rock formation structure distribution around each blasthole; Acquire the amount of explosives set in each blasthole in the tunnel face, and generate a tunnel face feature map based on the blasthole position distribution, blasthole size, blasthole detonation sequence, stratum structure of the tunnel face, and the amount of explosives set in each blasthole, wherein the tunnel face feature map includes blasthole feature vectors corresponding to each blasthole, and the blasthole feature vectors include the position coordinates of the blasthole on the tunnel face, the blasthole size, blasthole detonation sequence, stratum structure distribution around the blasthole, and the amount of explosives in the blasthole; Inputting the tunnel face characteristic map into a pre-trained blasting dust prediction model, the blasting dust prediction model outputs the amount of dust generated by the tunnel face blasting; generating a control signal based on the amount of dust generated by the blasting of the tunnel face; The fan is used to operate with a target output power based on the control signal, wherein the fan controls the dust concentration in the tunnel after the face blasting to be below a preset dust concentration after a preset time at the target output power.

10. A computer device, characterized in that: The computer device includes a processor, a computer-readable storage medium and a communication interface, wherein the computer-readable storage medium, the communication interface and the processor are connected via a bus system, the computer-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the computer-readable storage medium to execute the artificial intelligence-based tunnel ventilation system control method according to any one of claims 1 to 8.

Images