Mine airflow sensing system and method based on image recognition

Abstract

The invention provides a mine airflow sensing system and method based on image recognition, and relates to the field of mine ventilation. The system comprises a sensing module, an industrial control module, a dustproof module, a data generation module and a data acquisition module, according to the technical scheme provided by the invention, a two-dimensional unsteady-state temperature field is generated through a data generation module, video image data is acquired by using a data acquisition module, the video image data is preprocessed by using a sensing module, and mine airflow information in the video image data is stored and extracted; in the process, the industrial control module accurately controls cooperative work among the data generation module, the data acquisition module and the dustproof module; the dustproof module is used for preventing dust from accumulating on the data generation module and the data acquisition module. According to the invention, the problem that the anti-knock property of the sensor is insufficient in the existing technology for monitoring the mine air flow by using the sensor is solved, and secondary correction of parameters and reproduction and inversion of the mine air flow information can be carried out by using data directly or indirectly related to the mine air flow information.

Description

Technical Field

[0001] The present invention relates to the field of mine ventilation, and in particular to a mine wind flow sensing system and method based on image recognition. Background Art

[0002] In underground mines, a good ventilation system is an important guarantee for the personal safety of workers and safe production in mines. A good ventilation system is not only to provide fresh air to the underground space, but more importantly, it is to monitor and control the underground airflow and grasp the status of the underground airflow in real time, so as to detect abnormalities in time and make adjustments. This will avoid the risks brought by poor mine ventilation and further improve the production environment.

[0003] Mine wind flow is invisible and difficult to form video image data, so the current monitoring of mine wind flow mostly relies on various sensors; although they have the advantage of high precision, the sensors are usually more precise and fragile. The current underground mining of minerals still cannot do without blasting operations, especially in non-coal mines. The shock waves generated by blasting operations contain a lot of energy, so the mine wind flow monitoring sensors are easily affected by the shock waves generated by blasting operations, causing them to malfunction or even be damaged. This is especially true for sensors that are arranged close to the mining site. Due to the precision of their manufacturing and the fragility of their materials, once the sensors are damaged, they will be difficult to repair, and usually they can only be replaced with new sensors, which is costly.

[0004] In addition, when various sensors are used to monitor and control the mine wind flow, usually only the corrected direct readings of the mine wind flow information, such as wind flow speed and temperature, can be obtained. However, it is impossible to obtain the most original data directly or indirectly related to the mine wind flow information, making it difficult to perform secondary corrections to the mine wind flow information, and even more impossible to reproduce and invert the mine wind flow information. Therefore, it is very necessary to develop a mine wind flow sensing system and method that can not only withstand the shock wave energy generated by underground blasting operations without being damaged, but also form colorless and invisible mine wind flow into video image data, and at the same time obtain the most original data directly or indirectly related to the mine wind flow information. Summary of the invention

[0005] Aiming at the difficulty that mine wind flow is colorless and invisible and difficult to form video image data, the characteristics that various existing sensors used for mine wind flow monitoring are easily damaged by underground blasting operations and the deficiency that sensors are difficult to obtain data directly or indirectly related to mine wind flow information, the present invention discloses a mine wind flow perception system and method based on image recognition, which can not only collect colorless and invisible mine wind flow information in the form of video image data, but also improve the ability of mine wind flow monitoring equipment to resist damage, and at the same time can obtain the most original data directly or indirectly related to the mine wind flow information, which not only solves the problem that mine wind flow is colorless and invisible and difficult to form video image data and the problem that the sensor itself is insufficiently resistant to explosion in the existing mine wind flow monitoring technology using sensors, but also can use the obtained data directly or indirectly related to the mine wind flow information to perform secondary correction of parameters, reproduction and inversion of mine wind flow information.

[0006] The technical solution of the present invention is:

[0007] On one hand, the present invention provides a mine wind flow perception system based on image recognition, including a perception module, an industrial control module, a dust prevention module, a data generation module and a data acquisition module;

[0008] The data generation module is used to generate a two-dimensional non-steady-state temperature field containing mine wind flow information under the control of the industrial control module; the mine wind flow information includes wind flow direction, wind speed, wind flow temperature and air humidity;

[0009] The data acquisition module is used to acquire the two-dimensional non-stable temperature field containing the mine wind flow information generated by the data generation module in the form of video image data under the control of the industrial control module, and simultaneously obtain the real-time temperature information of the data generation module;

[0010] The perception module is used to pre-process, store and extract mine wind flow information from the video image data collected by the data collection module;

[0011] The industrial control module is used to control the collaborative work between the data generation module, the data acquisition module and the dust prevention module; supervise the operation status of the data generation module, the data acquisition module and the dust prevention module, and control the power supply of the data generation module, the data acquisition module and the dust prevention module; and transmit the video image data collected by the data acquisition module to the perception module;

[0012] The dust prevention module is arranged above the data generation module and the data acquisition module, and is used to generate a jet airflow under the control of the industrial control module to prevent dust from accumulating on the data generation module and the data acquisition module;

[0013] Furthermore, the sensing module and the industrial control module are connected by using a transmission bus;

[0014] The data generation module and the industrial control module are connected by using a control bus and a data generation module control branch line;

[0015] The data acquisition module and the industrial control module are connected by using a control bus and a data acquisition module control branch line;

[0016] The dustproof module and the industrial control module are connected by using a control bus and a dustproof module control branch line;

[0017] Furthermore, the perception module includes an image preprocessing module, an image database and a deep learning neural network model;

[0018] The image preprocessing module is used to simplify the video image data collected by the data acquisition module, eliminate information in the video image data that is irrelevant to the mine air flow information, and obtain preprocessed video image data;

[0019] The image database is used to store the original video image data collected by the data collection module and the video image data preprocessed by the image preprocessing module;

[0020] The deep learning neural network model is used to extract features from the processed video image data, identify and output the mine wind flow information contained in the video image data;

[0021] Furthermore, the industrial control module includes a data generation module controller, a dust prevention module controller, a data transmission controller and a data acquisition module controller;

[0022] The dustproof module controller is used to control the power supply of the dustproof module, the opening or closing of the jet airflow, and the size of the generated jet airflow;

[0023] The data generation module controller is used to control the power supply of the data generation module and set a constant temperature value of the data generation module;

[0024] The data acquisition module controller is used to control the power supply of the data acquisition module and the acquisition of video image data;

[0025] The data transmission controller is used to receive real-time video image data collected by the data acquisition module and transmit it to the perception module;

[0026] Furthermore, the data generation module includes a constant temperature heating panel, whose power supply and constant temperature value are controlled by the data generation module controller in the industrial control module. After power is turned on, the constant temperature heating panel is heated evenly to a set constant temperature value and maintained constant, and then the power supply of the constant temperature heating panel is cut off. The constant temperature heating panel is naturally cooled down under the action of the mine wind flow to generate a two-dimensional non-steady-state temperature field;

[0027] Further, the data acquisition module includes a thermal imager for acquiring real-time temperature information of the constant temperature heating panel in the data generation module, and acquiring video image data of a two-dimensional non-steady-state temperature field generated by the constant temperature heating panel in the data generation module;

[0028] Further, the dustproof module includes an air curtain generator, an air door switch, a blocking air door, a panel air duct, a lens air duct, a lens air outlet and a panel air outlet;

[0029] The lens air outlet is aligned with the front lens of the thermal imager in the data acquisition module, the panel air outlet is aligned with the constant temperature heating panel, the panel air duct is arranged above the lens air duct, the panel air duct and the lens air duct are connected with the wind curtain generator, and the air inlets of the panel air duct and the lens air duct are arranged inside the wind curtain generator; the damper switch is arranged at the air inlet of the panel air duct inside the wind curtain generator, and a blocking damper is arranged on the damper switch;

[0030] The wind curtain generator generates a jet airflow, which enters the panel air duct and the lens air duct through the air inlets of the panel air duct and the lens air duct, and flows out from the panel air outlet and the lens air outlet to form a lens dust-proof wind curtain and a panel dust-proof wind curtain; the damper switch is turned on, the blocking damper is opened, and the airflow generated by the wind curtain generator flows into the panel air duct; the damper switch is closed, the blocking damper and the panel air duct are closed, and the airflow is blocked from flowing into the panel air duct; the power supply and working state of the wind curtain generator and the damper switch are controlled by the dust-proof module controller in the industrial control module;

[0031] Another aspect of the present invention provides a mine wind flow perception method based on image recognition, comprising the following steps:

[0032] Step 1: Obtain video image data under different mine wind flow information conditions, construct a video image dataset and divide the video image dataset into a training set, a validation set and a test set according to a set ratio;

[0033] Step 1.1: Deploy industrial control modules, data generation modules, and data acquisition modules in the laboratory. Under different mine airflow information conditions, the constant temperature heating panel in the data generation module reaches the set constant temperature value and stops power supply, and naturally cools down under the action of airflow. The thermal imager in the data acquisition module is used to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel during the natural cooling process under the action of airflow, and obtain several laboratory original video image data;

[0034] Step 1.2: deploy the industrial control module, data generation module and data acquisition module in different mine tunnels, measure the mine wind flow information in the mine tunnels, the constant temperature heating panel in the data generation module reaches the set constant temperature value and stops power supply, and naturally cools down under the action of wind flow, and uses the thermal imager in the data acquisition module to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel in the process of natural cooling under the action of wind flow, and obtain a number of measured original video image data;

[0035] Step 1.3: Use the image preprocessing module in the perception module to preprocess the laboratory original video image data and the measured original video image data respectively, use the preprocessed laboratory original video image data and the measured original video image data as sample feature data, and set the corresponding mine wind flow information as the label of the sample feature data, store it in the image database in the perception module, and obtain the laboratory original data set and the measured original data set respectively;

[0036] Step 1.4: Divide the original laboratory data set into a training set and a validation set according to the set ratio, and use the measured original data set as the test set;

[0037] Step 2: Use the training set to train the deep learning neural network model in the perception module to obtain a trained deep learning neural network model, use the validation set to evaluate the performance of the trained deep learning neural network model and adjust its parameters, and finally use the test set to evaluate the perception accuracy of the trained deep learning neural network model in the measured data of the mine wind flow information to obtain the final deep learning neural network model;

[0038] Step 3: Deploy the mine airflow sensing system in the mine tunnel to be tested;

[0039] Specifically, the sensing module is deployed on the ground, the industrial control module, the data generation module, the data acquisition module and the dust prevention module are all deployed in the mine tunnel, the sensing module is connected to the industrial control module through a transmission bus; the data generation module controls the branch line through the data generation module, the data acquisition module controls the branch line through the data acquisition module, and the dust prevention module controls the branch line through the dust prevention module and is connected to the industrial control module via the control bus;

[0040] Step 4: Collect real-time video image data in the mine tunnel;

[0041] Step 4.1: The dustproof module controller in the industrial control module supplies power to the dustproof module and controls the air curtain generator in the dustproof module to generate a jet airflow, controls the air door switch to open, blocks the air door and the panel air duct from opening, and the jet airflow passes through the panel air duct and the lens air duct, and flows out from the panel air outlet and the lens air outlet, forming a lens dustproof air curtain and a panel dustproof air curtain;

[0042] Step 4.2: The data generation module controller in the industrial control module sets a constant temperature value for the constant temperature heating panel in the data generation module and supplies power. If the data generation module starts normally, the normal startup status information is transmitted to the dustproof module controller in the industrial control module, and step 4.3 is executed; if the data generation module cannot run normally, the abnormal operation information is transmitted to the dustproof module controller in the industrial control module, and the data generation module state is determined to be abnormal. The dustproof module controller and the data generation module controller in the industrial control module control to cut off the power supply of the dustproof module and the data generation module, and re-execute step 4.1 after eliminating the abnormality;

[0043] Step 4.3: After receiving the normal startup status information fed back by the data generation module controller, the dustproof module controller in the industrial control module controls the closing of the damper switch, blocks the damper and the panel air duct from closing, blocks the jet air flow from passing through the panel air duct, closes the panel dustproof air curtain, and sends damper switch closing information to the data acquisition module controller;

[0044] Step 4.4: After receiving the damper switch closing information, the data acquisition module controller in the industrial control module supplies power to the thermal imager in the data acquisition module. After the thermal imager starts to run, the data acquisition module controller controls the thermal imager in the data acquisition module to obtain the real-time temperature information of the constant temperature heating panel in the data generation module;

[0045] Step 4.5: The constant temperature heating panel evenly heats up to the set constant temperature value and keeps the temperature constant. After the thermal imager in the data acquisition module obtains the real-time temperature information of the constant temperature heating panel and has reached the set constant temperature value, the data acquisition module controller sends a power-off signal to the data generation module controller;

[0046] Step 4.6: After receiving the power-off signal transmitted by the data acquisition module controller, the data generation module controller cuts off the power supply of the data generation module and feeds back the normal stop information of the data generation module to the data acquisition module controller. The constant temperature heating panel in the data generation module is naturally cooled down under the action of the mine tunnel wind flow;

[0047] Step 4.7: After receiving the normal stop information fed back by the data generation module controller, the data acquisition module controller controls the thermal imager in the data acquisition module to acquire video image data, and transmits the video image data to the data transmission controller in the industrial control module;

[0048] Step 4.8: After the data transmission controller receives the video image data, if the video image data has been successfully received, it sends a data reception success signal to the dustproof module controller, and transmits the video image data to the sensing module via the transmission bus, and then executes step 4.9; if the video image data reception fails, the data transmission controller sends a data reception failure signal to the data acquisition module controller, and the data acquisition module transmits the video image data to the data transmission controller in the industrial control module again; if the video image data reception fails after exceeding the set number of times, it is determined that the data transmission controller is abnormal, and the industrial control module controls to cut off the power supply of the dustproof module, the data generation module and the data acquisition module, and re-executes step 4.1 after eliminating the abnormality;

[0049] Step 4.9: After receiving the data receiving success signal fed back by the data transmission controller, the dustproof module controller controls the opening of the damper switch, opens the blocking damper, and allows the jet airflow to pass through the panel air duct and flow out from the panel air outlet, thus forming a panel dustproof air curtain again;

[0050] Step 5: Use the perception module to perceive the mine wind flow information of the mine and obtain the mine wind flow information;

[0051] Specifically: the perception module receives real-time video image data transmitted by the data transmission controller in the industrial control module, and after pre-processing by the image pre-module, it is stored in the image database. After the final deep learning neural network model in the perception module obtains the real-time video image data, it extracts the mine wind flow information in the video image data to realize the perception of mine wind flow information.

[0052] Compared with the prior art, the beneficial effects of the present invention are:

[0053] 1. Compared with sensor technology, the constant temperature heating panel in the data generation module and the thermal imager in the data acquisition module in the technical solution provided by the present invention have strong explosion resistance and can effectively resist the impact of shock waves generated by underground blasting operations.

[0054] 2. Mine wind flow is colorless and invisible, and it is difficult to form video image data. The technical solution provided by the present invention generates a two-dimensional non-steady-state temperature field through a data generation module, and uses a data acquisition module to collect it as video image data. It is the most original data directly or indirectly related to the mine wind flow information; it can not only collect colorless and invisible mine wind flow information in the form of video image data, but also can be used for the reproduction and inversion of mine wind flow information.

[0055] 3. The mine wind flow information perception technology based on image recognition provided by the present invention is realized through a deep learning neural network model in a perception module. The accuracy of mine wind flow information perception can be improved by adjusting the model or increasing the amount of video image data, and secondary correction of the mine wind flow information perception results can also be performed.

[0056] 4. The mine airflow perception system based on image recognition is equipped with an independent dust prevention module, which can prevent dust accumulation on the data generation module and the data acquisition module, reduce the impact of dust accumulation on data acquisition, and ensure the quality of the collected video image data. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Schematic diagram of the system structure of a mine wind flow perception system based on image recognition in an embodiment of the present invention;

[0058] Figure 2 Schematic diagram of the device composition of the mine wind flow sensing system based on image recognition in an embodiment of the present invention;

[0059] Figure 3 A schematic diagram of the structure of a blocking damper in an embodiment of the present invention;

[0060] Figure 4 This is a schematic diagram of an embodiment of an underground deployment of a mine wind flow sensing system based on image recognition in an embodiment of the present invention;

[0061] Figure 5 This is a partial enlarged view of a mine wind flow sensing system based on image recognition deployed in an embodiment of the present invention;

[0062] Among them, Ⅰ-sensing module; Ⅱ-industrial control module; Ⅲ-dustproof module; Ⅳ-data generation module; Ⅴ-data acquisition module; 3-transmission bus; 4-control bus; 5-dustproof module control branch; 6-data acquisition module control branch; 7-data generation module control branch; 8-constant temperature heating panel; 9-thermal imager; 10-wind curtain generator; 11-damper switch; 12-panel air duct; 13-lens air duct; 14-lens air outlet; 15-panel air outlet; 16-blocking damper; 17-mine tunnel. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to describe the technical solutions in the embodiments of the present invention in detail and completely. It should be stated that the embodiments and drawings described below are only used to help understand the method and core ideas of the present invention, and are only partial embodiments of the present invention; for other technical personnel in the field, according to the ideas provided by the present invention, there will be changes in the specific implementation methods and application scopes; therefore, the content of this specification should not be understood as limiting the present invention; all other embodiments obtained by ordinary technical personnel in the field without creative work are within the scope of protection of the present invention.

[0064] Reference to "embodiments" in the present invention means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described in the present invention may be combined with other embodiments.

[0065] The conception basis of the present invention is:

[0066] 1) The present invention proposes to use equipment with strong explosion-proof performance as data generation and collection devices. Compared with various sensors with precise structures but relatively fragile, constant temperature heating panels and thermal imagers have better explosion-proof performance; even if they are deployed in mine tunnels close to the mining site, they can operate normally and maintain good working conditions during underground blasting operations.

[0067] 2) The present invention proposes to use a thermal imager to collect video image data of the natural cooling of a constant temperature heating panel under the action of mine wind flow over a period of time. When the ambient temperature is relatively stable and there is no air flow, if the temperature of an object is higher than the ambient temperature, its natural cooling process must have certain rules; if a certain air flow rate is applied in a single direction, the cooling speed of the object will be accelerated, and the cooling process must also have certain rules. Therefore, the constant temperature heating panel that has reached a constant temperature can be cut off from its power supply and placed under the mine wind flow, so that the constant temperature heating panel is cooled down because the heat on the surface is carried away by the mine wind flow; and a thermal imager is used to collect video image data of the natural cooling of the constant temperature heating panel under the action of mine wind flow; these video image data contain relevant information about the mine wind flow, and then image recognition technology is used to extract this information.

[0068] like Figure 1 , Figure 2 As shown, the mine wind flow perception system based on image recognition includes a perception module I, an industrial control module II, a dust prevention module III, a data generation module IV and a data acquisition module V;

[0069] The data generation module IV is used to generate a two-dimensional non-steady-state temperature field containing mine wind flow information under the control of the industrial control module II; the mine wind flow information includes but is not limited to wind flow direction, wind speed, wind flow temperature and air humidity;

[0070] The data acquisition module V is used to acquire the two-dimensional non-stable temperature field containing the mine wind flow information generated by the data generation module IV in the form of video image data under the control of the industrial control module II, and simultaneously obtain the real-time temperature information of the data generation module IV;

[0071] The sensing module I is used to pre-process, store and extract the mine wind flow information in the video image data collected by the data collection module V;

[0072] The industrial control module II is used to accurately control the collaborative work between the data generation module IV, the data acquisition module V and the dust prevention module III; supervise the operation status of the data generation module IV, the data acquisition module V and the dust prevention module III, and control the power supply of the data generation module IV, the data acquisition module V and the dust prevention module III; transmit the video image data collected by the data acquisition module V to the perception module I;

[0073] The dustproof module III is arranged above the data generation module IV and the data acquisition module V, and is used to generate a jet airflow under the control of the industrial control module II to prevent dust from accumulating on the data generation module IV and the data acquisition module V;

[0074] One sensing module I can be equipped with several sets of industrial control modules II, data generation modules IV, data acquisition modules V and dust prevention modules III;

[0075] The sensing module I and the industrial control module II are connected via a transmission bus 3;

[0076] The data generation module IV and the industrial control module II are connected by using the control bus 4 and the data generation module control branch line 7;

[0077] The data acquisition module V and the industrial control module II are connected by using the control bus 4 and the data acquisition module control branch line 6;

[0078] The dustproof module III and the industrial control module II are connected by using the control bus 4 and the dustproof module control branch line 5;

[0079] The perception module I includes an image preprocessing module, an image database and a deep learning neural network model;

[0080] The image preprocessing module is used to simplify the video image data collected by the data collection module V, eliminate the information in the video image data that is irrelevant to the mine wind flow information, enhance the detectability of the mine wind flow information in the image, and obtain the preprocessed video image data;

[0081] The image database is used to store the original video image data collected by the data collection module V and the video image data preprocessed by the image preprocessing module;

[0082] The deep learning neural network model includes but is not limited to a convolutional neural network, which is used to extract features from the processed video image data, identify and output the mine wind flow information contained in the video image data;

[0083] The industrial control module II includes a data generation module controller, a dust prevention module controller, a data transmission controller and a data acquisition module controller;

[0084] The dustproof module controller is used to control the power supply of the dustproof module III, the opening or closing of the jet airflow, and the size of the jet airflow generated;

[0085] The data generation module controller is used to control the power supply of the data generation module IV and set a constant temperature value of the data generation module IV;

[0086] The data acquisition module controller is used to control the power supply of the data acquisition module V and the acquisition of video image data;

[0087] The data transmission controller is used to receive real-time video image data collected by the data acquisition module V and transmit it to the perception module I;

[0088] The data generation module IV includes a constant temperature heating panel 8, whose power supply and constant temperature value are controlled by the data generation module controller in the industrial control module II. After power is turned on, the constant temperature heating panel 8 is heated evenly to a set constant temperature value and kept constant, and then the power supply of the constant temperature heating panel 8 is cut off. The constant temperature heating panel 8 is naturally cooled down under the action of the mine wind flow to generate a two-dimensional non-steady-state temperature field;

[0089] The data acquisition module V includes a thermal imager 9, which is used to obtain real-time temperature information of the constant temperature heating panel 8 in the data generation module IV, and collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel 8 in the data generation module IV; its power supply and working state are controlled by the data generation module controller in the industrial control module II;

[0090] The dustproof module III includes an air curtain generator 10, an air door switch 11, a blocking air door 16, a panel air duct 12, a lens air duct 13, a lens air outlet 14 and a panel air outlet 15;

[0091] The lens air outlet 14 is aligned with the front lens of the thermal imager 9 in the data acquisition module V, the panel air outlet 15 is aligned with the constant temperature heating panel 8, the panel air duct 12 is arranged above the lens air duct 13, the panel air duct 12 and the lens air duct 13 are connected with the wind curtain generator 10, and the air inlets of the panel air duct 12 and the lens air duct 13 are arranged inside the wind curtain generator 10; the damper switch 11 is arranged at the air inlet of the panel air duct 12 inside the wind curtain generator 10, and a blocking damper 16 is arranged on the damper switch 11;

[0092] The wind curtain generator 10 generates a jet airflow, which enters the panel air duct 12 and the lens air duct 13 through the air inlets of the panel air duct 12 and the lens air duct 13, and flows out from the panel air outlet 15 and the lens air outlet 14 to form a lens dust-proof wind curtain and a panel dust-proof wind curtain. The two dust-proof wind curtains can prevent dust accumulation on the data generation module IV and the data acquisition module V, and reduce the impact of dust accumulation on data acquisition. Figure 3 As shown, the damper switch 11 is turned on, the blocking damper 16 is opened, and the airflow generated by the air curtain generator 10 flows into the panel air duct 12. The damper switch 11 is closed, the blocking damper 16 and the panel air duct 12 are closed, and the airflow is blocked from flowing into the panel air duct 12. The power supply and working state of the air curtain generator 10 and the damper switch 11 are controlled by the dustproof module controller in the industrial control module II.

[0093] Figure 5 for Figure 4 A partial enlarged view of Figure 4 , Figure 5 In the exemplary embodiment of the mine wind flow perception system based on image recognition shown, the mine wind flow perception method based on image recognition includes the following steps:

[0094] Step 1: Obtain video image data under different mine wind flow information conditions, construct a video image dataset and divide the video image dataset into a training set, a validation set and a test set according to a set ratio;

[0095] Step 1.1: Deploy industrial control module II, data generation module IV and data acquisition module V in the laboratory. Under different mine wind flow information conditions, the constant temperature heating panel 8 in the data generation module IV reaches the set constant temperature value and stops power supply, and naturally cools down under the action of wind flow. The thermal imager 9 in the data acquisition module V is used to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel 8 during the natural cooling process under the action of wind flow, and obtain a number of laboratory original video image data;

[0096] In this embodiment, the industrial control module II, the data generation module IV and the data acquisition module V are deployed in the laboratory, and different mine wind flow information conditions (including but not limited to wind flow direction, wind flow speed, wind flow temperature and humidity, etc.) are set for the wind flow by changing the working state of the fan (such as running direction, blade installation angle and power, etc.) and air temperature control measures (such as resistance heating, ice cooling, etc.), so that the constant temperature heating panel 8 in the data generation module IV reaches a constant temperature value and stops power supply and cools down naturally under the action of wind flow, and uses the thermal imager 9 in the data acquisition module V to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel 8 in the process of natural cooling under the action of wind flow;

[0097] Step 1.2: deploy the industrial control module II, data generation module IV and data acquisition module V in different mine tunnels 17, measure the mine wind flow information in the mine tunnel, the constant temperature heating panel 8 in the data generation module IV reaches the set constant temperature value and stops power supply, and naturally cools down under the action of wind flow, and uses the thermal imager 9 in the data acquisition module V to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel 8 in the process of natural cooling under the action of wind flow, and obtain a number of measured original video image data;

[0098] In this embodiment, the industrial control module II, the data generation module IV and the data acquisition module V are temporarily deployed in different mine tunnels 17 to measure ventilation parameters (including but not limited to wind direction, wind speed, wind temperature and humidity, etc.) on site, so that the constant temperature heating panel 8 in the data generation module IV reaches the set constant temperature value and stops power supply to cool down naturally under the action of wind flow, and the thermal imager 9 in the data acquisition module V is used to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel 8 during the natural cooling process under the action of wind flow;

[0099] Step 1.3: Use the image preprocessing module in the perception module I to preprocess the laboratory original video image data and the measured original video image data respectively, use the preprocessed laboratory original video image data and the measured original video image data as sample feature data, and set the corresponding mine wind flow information as the label of the sample feature data, store it in the image database in the perception module I, and obtain the laboratory original data set and the measured original data set respectively;

[0100] Step 1.4: Divide the original laboratory data set into a training set and a validation set according to the set ratio, and use the measured original data set as the test set;

[0101] In this embodiment, the training set is used to train the deep learning neural network model in the perception module I, the validation set is used to verify the performance of the deep learning neural network model and the accuracy of mine wind flow information perception, and the test set is used to test the accuracy of mine wind flow information perception of the deep learning neural network model in measured data.

[0102] Step 2: Use the training set to train the deep learning neural network model in the perception module I to obtain a trained deep learning neural network model, use the validation set to evaluate the performance of the trained deep learning neural network model and adjust the parameters, and finally use the test set to evaluate the perception accuracy of the trained deep learning neural network model in the mine wind flow information in the measured data to obtain the final deep learning neural network model;

[0103] Step 3: Deploy the mine airflow sensing system in the mine tunnel to be tested;

[0104] Specifically, the sensing module I is deployed on the ground. Figure 4 , Figure 5 As shown, the industrial control module II, data generation module IV, data acquisition module V and dust prevention module III are all deployed in the mine tunnel 17, and the sensing module I is connected to the industrial control module II through the transmission bus 3; the data generation module IV is connected to the industrial control module II through the data generation module control branch line 7, the data acquisition module V is connected to the industrial control module II through the data acquisition module control branch line 6, and the dust prevention module III is connected to the industrial control module II through the dust prevention module control branch line 5 via the control bus 4;

[0105] Step 4: Collecting real-time video image data in the mine tunnel 17;

[0106] Step 4.1: The dustproof module controller in the industrial control module II supplies power to the dustproof module III and controls the air curtain generator 10 in the dustproof module III to generate a jet airflow, controls the opening of the damper switch 11, blocks the damper 16 and the panel air duct from being opened, and the jet airflow passes through the panel air duct 12 and the lens air duct 13, and flows out from the panel air outlet 15 and the lens air outlet 24, forming a lens dustproof air curtain and a panel dustproof air curtain;

[0107] Step 4.2: The data generation module controller in the industrial control module II sets a constant temperature value for the constant temperature heating panel 8 in the data generation module IV and supplies power. If the data generation module IV starts normally, the normal startup state information is transmitted to the dustproof module controller in the industrial control module II, and step 4.3 is executed; if the data generation module IV cannot run normally, the abnormal operation information is transmitted to the dustproof module controller in the industrial control module II, and the data generation module IV is judged to be abnormal. The industrial control module II controls to cut off the power supply of the dustproof module III and the data generation module IV, and re-execute step 4.1 after eliminating the abnormality;

[0108] Step 4.3: After receiving the normal startup status information of the data generation module IV fed back by the data generation module controller, the dust prevention module controller in the industrial control module II controls to close the damper switch 11, blocks the damper 16 and the panel air duct 12 from closing, blocks the jet air flow from passing through the panel air duct 12, closes the panel dust prevention air curtain, and sends damper switch closing information to the data acquisition module controller;

[0109] Step 4.4: After receiving the damper switch closing information, the data acquisition module controller in the industrial control module II supplies power to the thermal imager 9 in the data acquisition module V. After the thermal imager 9 starts to run, the data acquisition module controller controls the thermal imager 9 to obtain the real-time temperature information of the constant temperature heating panel 8 in the data generation module IV.

[0110] Step 4.5: The constant temperature heating panel 8 evenly heats up to the set constant temperature value and keeps the temperature constant. After the thermal imager 9 in the data acquisition module V obtains the real-time temperature information of the constant temperature heating panel 8 and has reached the set constant temperature value, the data acquisition module controller sends a power-off signal to the data generation module controller;

[0111] Step 4.6: After receiving the power-off signal transmitted by the data acquisition module controller, the data generation module controller cuts off the power supply of the data generation module IV and feeds back the normal stop information of the data generation module IV to the data acquisition module controller, and the constant temperature heating panel 8 is naturally cooled down under the action of the mine tunnel wind flow;

[0112] Step 4.7: After receiving the normal stop information fed back by the data generation module controller, the data acquisition module controller controls the thermal imager 9 to collect video image data, and transmits the video image data to the data transmission controller in the industrial control module II;

[0113] Step 4.8: After the data transmission controller receives the video image data, if the video image data has been successfully received, it sends a data reception success signal to the dustproof module controller, and transmits the video image data to the sensing module I via the transmission bus 3, and then executes step 4.9; if the video image data reception fails, the data transmission controller sends a data reception failure signal to the data acquisition module controller, and the data acquisition module transmits the video image data to the data transmission controller in the industrial control module again; if the video image data reception fails multiple times, it is determined that the data transmission controller is abnormal, and the industrial control module II controls to cut off the power supply of the dustproof module III, the data generation module IV and the data acquisition module V, and re-executes step 4.1 after eliminating the abnormality;

[0114] Step 4.9: The dustproof module controller receives the data receiving success signal fed back by the data transmission controller, controls to open the damper switch 11, opens the blocking damper 16, allows the jet airflow to pass through the panel air duct 12, and flows out from the panel air outlet 15, thereby forming a panel dustproof air curtain again;

[0115] Step 5: Use the perception module I to perceive the mine wind flow information of the mine and obtain the mine wind flow information;

[0116] Specifically: the perception module I receives the real-time video image data transmitted by the data transmission controller in the industrial control module II, and after being pre-processed by the image pre-module, it is stored in the image database. After the final deep learning neural network model in the perception module I obtains the real-time video image data, it extracts the characteristic information related to the mine airflow in the video image data, and finally outputs the mine airflow information (including but not limited to airflow direction, airflow speed, airflow temperature and humidity, etc.) to realize the perception of mine airflow information.

Claims

1. A mine wind flow perception system based on image recognition, characterized in that: Including perception module, industrial control module, dust prevention module, data generation module and data acquisition module; The data generation module is used to generate a two-dimensional non-steady-state temperature field containing mine wind flow information under the control of the industrial control module; the mine wind flow information includes wind flow direction, wind speed, wind flow temperature and air humidity; The data acquisition module is used to acquire the two-dimensional non-stable temperature field containing the mine wind flow information generated by the data generation module in the form of video image data under the control of the industrial control module, and simultaneously obtain the real-time temperature information of the data generation module; The perception module is used to pre-process, store and extract mine wind flow information from the video image data collected by the data collection module; The industrial control module is used to control the collaborative work between the data generation module, the data acquisition module and the dust prevention module; Monitor the operating status of the data generation module, data collection module and dust prevention module, and control the power supply of the data generation module, data collection module and dust prevention module; Transmitting the video image data collected by the data acquisition module to the perception module; The dustproof module is arranged above the data generation module and the data acquisition module, and is used to generate a jet airflow under the control of the industrial control module to prevent dust from accumulating on the data generation module and the data acquisition module.

2. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The sensing module and the industrial control module are connected via a transmission bus; The data generation module and the industrial control module are connected by using a control bus and a data generation module control branch line; The data acquisition module and the industrial control module are connected by using a control bus and a data acquisition module control branch line; The dustproof module and the industrial control module are connected by using a control bus and a dustproof module control branch line.

3. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The perception module includes an image preprocessing module, an image database and a deep learning neural network model; The image preprocessing module is used to simplify the video image data collected by the data acquisition module, eliminate information in the video image data that is irrelevant to the mine air flow information, and obtain preprocessed video image data; The image database is used to store the original video image data collected by the data collection module and the video image data preprocessed by the image preprocessing module; The deep learning neural network model is used to extract features from the processed video image data, and identify and output the mine wind flow information contained in the video image data.

4. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The industrial control module includes a data generation module controller, a dust prevention module controller, a data transmission controller and a data acquisition module controller; The dustproof module controller is used to control the power supply of the dustproof module, the opening or closing of the jet airflow, and the size of the generated jet airflow; The data generation module controller is used to control the power supply of the data generation module and set a constant temperature value of the data generation module; The data acquisition module controller is used to control the power supply of the data acquisition module and the acquisition of video image data; The data transmission controller is used to receive real-time video image data collected by the data acquisition module and transmit it to the perception module.

5. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The data generation module includes a constant temperature heating panel, whose power supply and constant temperature value are controlled by the data generation module controller in the industrial control module. After power is turned on, the constant temperature heating panel heats up evenly to the set constant temperature value and remains constant, and then the power supply of the constant temperature heating panel is cut off. The constant temperature heating panel naturally cools down under the action of the mine wind flow to generate a two-dimensional non-steady-state temperature field.

6. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The data acquisition module includes a thermal imager, which is used to obtain real-time temperature information of the constant temperature heating panel in the data generation module, and to collect video image data of a two-dimensional non-steady-state temperature field generated by the constant temperature heating panel in the data generation module.

7. The mine wind flow sensing system based on image recognition according to claim 1 is characterized in that: The dustproof module includes an air curtain generator, an air door switch, a blocking air door, a panel air duct, a lens air duct, a lens air outlet and a panel air outlet; The lens air outlet is aligned with the front lens of the thermal imager in the data acquisition module, the panel air outlet is aligned with the constant temperature heating panel, the panel air duct is arranged above the lens air duct, the panel air duct and the lens air duct are connected with the wind curtain generator, and the air inlets of the panel air duct and the lens air duct are arranged inside the wind curtain generator; the damper switch is arranged at the air inlet of the panel air duct inside the wind curtain generator, and a blocking damper is arranged on the damper switch; The wind curtain generator generates a jet airflow, which enters the panel air duct and the lens air duct through the air inlets of the panel air duct and the lens air duct, and flows out from the panel air outlet and the lens air outlet to form a lens dust-proof wind curtain and a panel dust-proof wind curtain; the damper switch is turned on, the blocking damper is opened, and the airflow generated by the wind curtain generator flows into the panel air duct; the damper switch is closed, the blocking damper and the panel air duct are closed, and the airflow is blocked from flowing into the panel air duct; the power supply and working status of the wind curtain generator and the damper switch are controlled by the dustproof module controller in the industrial control module.

8. A mine wind flow perception method based on image recognition, implemented based on the mine wind flow perception system based on image recognition according to claim 1, characterized in that: The steps include: Step 1: Obtain video image data under different mine wind flow information conditions, construct a video image dataset and divide the video image dataset into a training set, a validation set and a test set according to a set ratio; Step 2: Use the training set to train the deep learning neural network model in the perception module to obtain a trained deep learning neural network model, use the validation set to evaluate the performance of the trained deep learning neural network model and adjust its parameters, and finally use the test set to evaluate the perception accuracy of the trained deep learning neural network model in the measured data of the mine wind flow information to obtain the final deep learning neural network model; Step 3: Deploy the mine airflow sensing system in the mine tunnel to be tested; Specifically, the sensing module is deployed on the ground, the industrial control module, the data generation module, the data acquisition module and the dust prevention module are all deployed in the mine tunnel, and the sensing module is connected to the industrial control module through a transmission bus; The data generation module controls the branch line through the data generation module, the data acquisition module controls the branch line through the data acquisition module, and the dust prevention module controls the branch line through the dust prevention module and is connected to the industrial control module via the control bus; Step 4: Collect real-time video image data in the mine tunnel; Step 5: Use the perception module to perceive the mine wind flow information of the mine and obtain the mine wind flow information; Specifically: the perception module receives real-time video image data transmitted by the data transmission controller in the industrial control module, and after pre-processing by the image pre-module, it is stored in the image database. After the final deep learning neural network model in the perception module obtains the real-time video image data, it extracts the mine wind flow information in the video image data to realize the perception of mine wind flow information.

9. The mine wind flow perception method based on image recognition according to claim 8, characterized in that: Step 1 specifically includes: Step 1.1: Deploy industrial control modules, data generation modules, and data acquisition modules in the laboratory. Under different mine airflow information conditions, the constant temperature heating panel in the data generation module reaches the set constant temperature value and stops power supply, and naturally cools down under the action of airflow. The thermal imager in the data acquisition module is used to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel during the natural cooling process under the action of airflow, and obtain several laboratory original video image data; Step 1.2: deploy the industrial control module, data generation module and data acquisition module in different mine tunnels, measure the mine wind flow information in the mine tunnels, the constant temperature heating panel in the data generation module reaches the set constant temperature value and stops power supply, and naturally cools down under the action of wind flow, and uses the thermal imager in the data acquisition module to collect video image data of the two-dimensional non-steady-state temperature field generated by the constant temperature heating panel in the process of natural cooling under the action of wind flow, and obtain a number of measured original video image data; Step 1.3: Use the image preprocessing module in the perception module to preprocess the laboratory original video image data and the measured original video image data respectively, use the preprocessed laboratory original video image data and the measured original video image data as sample feature data, and set the corresponding mine wind flow information as the label of the sample feature data, store it in the image database in the perception module, and obtain the laboratory original data set and the measured original data set respectively; Step 1.4: Divide the original laboratory data set into a training set and a validation set according to the set ratio, and use the measured original data set as the test set.

10. The mine wind flow perception method based on image recognition according to claim 8, characterized in that: Step 4 specifically includes: Step 4.1: The dustproof module controller in the industrial control module supplies power to the dustproof module and controls the air curtain generator in the dustproof module to generate a jet airflow, controls the air door switch to open, blocks the air door and the panel air duct from opening, and the jet airflow passes through the panel air duct and the lens air duct, and flows out from the panel air outlet and the lens air outlet, forming a lens dustproof air curtain and a panel dustproof air curtain; Step 4.2: The data generation module controller in the industrial control module sets a constant temperature value for the constant temperature heating panel in the data generation module and supplies power. If the data generation module starts normally, the normal startup status information is transmitted to the dustproof module controller in the industrial control module, and step 4.3 is executed; if the data generation module cannot run normally, the abnormal operation information is transmitted to the dustproof module controller in the industrial control module, and the data generation module state is determined to be abnormal. The dustproof module controller and the data generation module controller in the industrial control module control to cut off the power supply of the dustproof module and the data generation module, and re-execute step 4.1 after eliminating the abnormality; Step 4.3: After receiving the normal startup status information fed back by the data generation module controller, the dustproof module controller in the industrial control module controls the closing of the damper switch, blocks the damper and the panel air duct from closing, blocks the jet air flow from passing through the panel air duct, closes the panel dustproof air curtain, and sends damper switch closing information to the data acquisition module controller; Step 4.4: After receiving the damper switch closing information, the data acquisition module controller in the industrial control module supplies power to the thermal imager in the data acquisition module. After the thermal imager starts to run, the data acquisition module controller controls the thermal imager in the data acquisition module to obtain the real-time temperature information of the constant temperature heating panel in the data generation module; Step 4.5: The constant temperature heating panel evenly heats up to the set constant temperature value and keeps the temperature constant. After the thermal imager in the data acquisition module obtains the real-time temperature information of the constant temperature heating panel and has reached the set constant temperature value, the data acquisition module controller sends a power-off signal to the data generation module controller; Step 4.6: After receiving the power-off signal transmitted by the data acquisition module controller, the data generation module controller cuts off the power supply of the data generation module and feeds back the normal stop information of the data generation module to the data acquisition module controller. The constant temperature heating panel in the data generation module is naturally cooled down under the action of the mine tunnel wind flow; Step 4.7: After receiving the normal stop information fed back by the data generation module controller, the data acquisition module controller controls the thermal imager in the data acquisition module to acquire video image data, and transmits the video image data to the data transmission controller in the industrial control module; Step 4.8: After the data transmission controller receives the video image data, if the video image data has been successfully received, it sends a data reception success signal to the dustproof module controller, and transmits the video image data to the sensing module via the transmission bus, and then executes step 4.9; if the video image data reception fails, the data transmission controller sends a data reception failure signal to the data acquisition module controller, and the data acquisition module transmits the video image data to the data transmission controller in the industrial control module again; if the video image data reception fails after exceeding the set number of times, it is determined that the data transmission controller is abnormal, and the industrial control module controls to cut off the power supply of the dustproof module, the data generation module and the data acquisition module, and re-executes step 4.1 after eliminating the abnormality; Step 4.9: After receiving the data reception success signal fed back by the data transmission controller, the dustproof module controller controls the damper switch to open, opens the blocking damper, and allows the jet airflow to pass through the panel air duct and flow out from the panel air outlet, thereby forming a panel dustproof air curtain again.

Images