Intelligent coal cutting planning method combining machine vision and cutting history curve

Abstract

The application provides an intelligent coal cutting planning method combining machine vision and cutting history curve, which comprises the following steps: creating a first-time working coal cutting working path, controlling the advancing speed and working direction of the heading machine through machine vision and manual adjustment, and inputting the position data of the heading machine into a database; determining the daily coal cutting working path according to the position data and storing the daily coal cutting working path into the database; the heading machine advances according to the daily coal cutting working path, and the heading machine corrects the working path in real time through machine vision analysis; obtaining the real-time working condition of the heading machine and the coal and rock collection image of the fully-mechanized working face, and transmitting the coal and rock collection image to a background monitoring interface and correcting the working speed and working direction of the heading machine in real time; updating the database, determining the daily coal cutting working path and storing the daily coal cutting working path into the database for calling in the next operation. The application solves the problem that the planning of the working path of the memory cutting technology of the coal mining machine in the prior art excessively depends on manual operation or does not have the planning ability to adapt to the change of the actual working environment.

Description

Technical Field

[0001] This invention relates to the field of tunneling machine cutting path planning technology, and more specifically, to an intelligent coal cutting planning method that combines machine vision and cutting process curves. Background Technology

[0002] Coal, as my country's basic energy source and industrial raw material, provides strong support for social progress and industrial development. According to relevant data, my country's coal reserves account for 12.8% of the world's total, reaching 1145 × 10⁸ tons. However, coal accounts for 57.9% of my country's energy consumption, indicating that it remains the dominant component of my country's energy structure. A crucial aspect of industrial upgrading is the construction of intelligent mines. In recent years, with the advent of the intelligent era, various industries have entered the internet age. However, due to the complexity of the environment and geological limitations, coal mines have only implemented intelligent technologies in certain areas. The core production area of ​​the mine—the working face—often has a low level of intelligence. Therefore, accelerating the intelligentization of coal mines can expedite their integration with the times, thereby promoting industry transformation and development. Coal mine development has gone through three stages: manual excavation, blasting mining, and mechanized mining. From the initial manual mining, it has gradually moved towards mechanized mining, and resource recovery rates have increased with technological advancements. After entering the mechanized mining stage, coal mining machines have gone through four generations: fixed drum coal mining machines, single-rocker drum coal mining machines, double-rocker drum coal mining machines, and electrically traction coal mining machines. This demonstrates that the production efficiency of coal resources is inextricably linked to the technology at a given stage. As of 2020, the mechanization rates of coal mining and tunneling in my country had reached 78.5% and 60.4%, respectively. However, the current level of mechanization is gradually failing to meet the needs of industrial development, and the coal industry's technological innovation capabilities are somewhat insufficient, hindering its prospects for high-quality development.

[0003] Unmanned coal mining is a mining technology that is being pursued both domestically and internationally. With the promotion of various emerging information technologies, such as the Internet of Things, big data, industrial internet, artificial intelligence, and cloud computing, these technologies have been applied in different production stages of mines. Among them, coal mining machines, as the basic equipment for mine production and operation, have complex structures, operate in confined spaces, and face variable environments. Currently, information technology is widely used in China.

[0004] Traditional path planning methods are broadly categorized into geometric control, artificial potential field, unit partitioning, and intelligent planning. Geometric control involves identifying the starting and ending points of the target route, mapping obstacles onto a spatial weighted graph based on an environmental model, and then weighting the route in the weighted graph according to the starting and ending points to obtain the final target route. Yang Hai et al. used a strapdown inertial navigation coordinate system to position and determine the attitude of a coal mining machine, applying trigonometric geometry to solve the coal cutting curve of the machine's drum, thus obtaining the machine's trajectory. The artificial potential field method simulates the attractive and repulsive forces of a natural gravitational field, both of which jointly control the object. Dong Gang et al. planned the path of the drum curve on a coal mining machine by establishing a gravitational field, demonstrating the effectiveness of the artificial potential field method; however, their environmental model uses a virtual coal-rock interface, resulting in often low accuracy. The unit partitioning method divides the spatial model into sets of simple units, with obstacles as full units and no obstacles as empty units, determining the path by calculating the relationships between adjacent units. Intelligent planning methods are classified into three categories based on their principles: fuzzy logic control, neural network optimization, and genetic algorithm optimization. Quan Guotong et al. used particle swarm optimization to optimize cubic spline curves, obtaining smooth and reliable coal cutting curves. Si Lei proposed a method for rationally planning the cutting path of a coal mining machine based on improved DS evidence theory and multiple neural networks. The above-mentioned coal mining machine cutting path planning methods have high algorithmic complexity and are difficult to implement.

[0005] In recent years, with the development of new technologies such as "transparent geology," people have gained increasingly accurate knowledge of the coal seam occurrence conditions in fully mechanized mining faces. However, no matter how accurate, it is still a prediction. The occurrence conditions of coal seams underground are extremely complex, and it is impossible to completely predict the coal seam occurrence conditions in fully mechanized mining faces. The working location of the fully mechanized mining face is constantly changing. Therefore, the coal seam occurrence conditions encountered at different locations during the advance of the fully mechanized mining unit, especially the coal body cutting resistance, will not be exactly the same. The heterogeneity of the coal body, the wear of the drum cutting teeth, the condition of floating coal on the floor, and the unevenness of the floor all have a great deal of randomness, even chance. Numerous studies have proven that during the coal cutting process, the instantaneous traction speed of the coal mining machine, the drum cutting speed, the drum cutting height, and the cutting depth are all random variables that follow a certain distribution. That is, the coal cutting process of the coal mining machine is a random process. Similarly, the advancement process of the working face support and scraper conveyor is also a random process. During the mining of fully mechanized mining faces, the undulation of the floor, the coal wall, and the unevenness of the hydraulic support and scraper conveyor tend to intensify on their own. The production process of a fully mechanized mining face is a self-opening, non-repetitive process, which is significantly different from the repetitive processes on a typical industrial automated production line.

[0006] The current coal mining machine memory cutting technology is actually similar to the repetitive process on a general automated industrial production line, which is not compatible with the fully mechanized mining production process. At the same time, its work path planning relies too much on manual labor or does not have the planning ability to adapt to changes in the actual working environment. Summary of the Invention

[0007] The main objective of this invention is to provide an intelligent coal cutting planning method that combines machine vision and cutting process curves, so as to at least solve the problem that the planning of the working path of the coal mining machine memory cutting technology in the prior art relies too much on manual labor or does not have the planning ability to adapt to changes in the actual working environment.

[0008] To achieve the above objectives, this invention provides an intelligent coal cutting planning method combining machine vision and cutting process curves, comprising: Step S102: Creating a coal cutting work path for the initial operation, controlling the forward speed and working direction of the tunneling machine through machine vision and manual adjustment, and recording the position data of the tunneling machine into a database; Step S104: After completing the initial operation, determining the daily coal cutting work path based on the position data and storing it in the database; Step S106: During daily operations, the tunneling machine tunnels according to the daily coal cutting work path, and the tunneling machine corrects the work path in real time through machine vision analysis; Step S108: Acquiring the real-time working status of the tunneling machine and coal and rock acquisition images of the fully mechanized mining face and transmitting them to the background monitoring interface; correcting the working speed and working direction of the tunneling machine in real time based on the real-time working status of the tunneling machine and the coal and rock acquisition images; Step S110: After completing the daily operation, updating the database and determining the daily coal cutting work path and storing it in the database for future use.

[0009] Further, step S102 includes: acquiring coal and rock condition information of the fully mechanized mining face; controlling the working direction of the tunneling machine through a machine vision recognition algorithm based on the coal and rock condition information, while continuously adjusting the forward speed and working direction of the tunneling machine; acquiring the actual underground positioning of the tunneling machine as the longitudinal cutting process, and acquiring the actual forward direction of the tunneling machine as the transverse cutting process; storing the longitudinal cutting process, the transverse cutting process, and the corresponding time points into the database.

[0010] Further, step S104 includes: fitting a three-dimensional coal cutting model for the first operation based on the longitudinal cutting history and transverse cutting history stored in the database, using the three-dimensional model to determine the daily coal cutting work path and storing it in the database for future operation.

[0011] Further, step S106 includes: selecting the daily coal cutting work path in the database and controlling the tunneling machine to move forward according to the daily coal cutting work path; when the tunneling machine is operating automatically, it analyzes the coal-rock ratio of the fully mechanized mining face in real time through machine vision to correct the work path. If the correction is small, it is included in the work log and fed back to the background; if the correction is large, it slows down and issues a warning in the background for manual processing.

[0012] Furthermore, step S108 includes: recording the working status of each part of the tunneling machine in real time; when an abnormality is detected in the working status of the tunneling machine, the tunneling machine slows down and issues a warning in the background, waiting for manual handling; when the tunneling machine is operating automatically, it transmits its working status and coal and rock acquisition images of the fully mechanized mining face back to the background monitoring interface; and the management personnel correct the working speed and working direction of the tunneling machine in real time based on the working status and coal and rock acquisition images.

[0013] Furthermore, during automatic operation, the tunneling machine transmits its working status back to the back-end monitoring interface in real time in the form of graphic data, and the coal and rock images collected from the fully mechanized mining face are transmitted back to the back-end monitoring interface in real time via image transmission.

[0014] Further, step S110 includes: determining the error of the actual working path, manually corrected trajectory data and predicted working path stored in the database, selecting and deleting data with large errors in a weighted manner to update the database, cleaning and updating the database, forming the coal cutting working path for the next daily work based on the database prediction and storing it in the database for the next operation.

[0015] The intelligent coal cutting planning method combining machine vision and cutting process curves of this invention includes: Step S102: Creating the coal cutting work path for the first job, controlling the forward speed and working direction of the tunneling machine through machine vision and manual adjustment, and recording the position data of the tunneling machine into the database; Step S104: After completing the first job, determining the daily coal cutting work path based on the position data and storing it in the database; Step S106: During daily work, the tunneling machine tunnels according to the daily coal cutting work path, and the tunneling machine corrects the work path in real time through machine vision analysis; Step S108: Acquiring the real-time working status of the tunneling machine and coal and rock acquisition images of the fully mechanized mining face and transmitting them to the background monitoring interface; correcting the working speed and working direction of the tunneling machine in real time based on the real-time working status of the tunneling machine and the coal and rock acquisition images; Step S110: After completing the daily work, updating the database and determining the daily coal cutting work path and storing it in the database for future use.

[0016] This invention achieves intelligent planning of coal cutting work paths by combining machine vision and cutting process curves.

[0017] On the one hand, the reduced number of operators for tunneling machines in fully mechanized mining faces lowers the potential for underground work hazards and losses, increases personnel utilization, and reduces labor costs. On the other hand, the reduced contact between workers and the front-line workers lowers the risk to their health, aligning with the people-oriented and safety-first safety philosophy of my country's coal mining industry.

[0018] On the other hand, the tunneling machine's cutting path planning is more rational, improving coal production efficiency, reducing the proportion of rock extracted during mining, lowering the wear rate of cutting teeth, increasing energy utilization, and thus significantly reducing mining costs. Simultaneously, dust levels during mining are reduced due to the lower proportion of extracted rock, leading to less air and solid pollutant generation and emissions, which aligns with my country's green mining development philosophy.

[0019] Furthermore, this invention truly achieves a balance of control between humans and machines. Workers have the highest authority over the tunneling machine's trajectory planning, while the machine simultaneously learns from manually corrected trajectories by recording them in its database for self-cleaning and updating. During operation, the system integrates the planning experience of workers, continuously improving their decision-making capabilities. The machine vision's coal and rock recognition function also corrects and prompts both the system's predicted and manually planned paths, improving the system's real-time performance and accuracy. Workers can also gain a better understanding of the actual situation on-site, obtaining more information from the front-line working face, thus enhancing the accuracy and scientific rigor of their judgments.

[0020] This invention improves the scientific rigor, accuracy, and real-time performance of tunneling machine path planning, thereby increasing coal production efficiency, reducing coal mine safety accidents, and effectively implementing my country's development concepts of smart and green mines. This invention addresses the problem that existing coal mining machine memory cutting technology relies too heavily on manual labor for path planning or lacks the ability to adapt to changes in the actual working environment. Attached Figure Description

[0021] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0022] Figure 1 This is a flowchart of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is an optional embodiment of the present invention.

[0023] Figure 2 This is a block diagram of the elements of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0024] Figure 3This is a flowchart of the first operation of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0025] Figure 4 This is a flowchart of the post-operation standby mode of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0026] Figure 5 This is a flowchart of a non-first-time operation (daily use) intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0027] Figure 6 This is a block diagram of the cutting process curve composition structure of an intelligent coal cutting planning method that combines machine vision and cutting process curve, which is optional according to an embodiment of the present invention.

[0028] Figure 7 This is a schematic diagram of an improved PSPNET network structure, which is an optional intelligent coal cutting planning method combining machine vision and cutting process curves according to an embodiment of the present invention.

[0029] Figure 8 This is a non-infrared image defogging time comparison chart of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0030] Figure 9 This is an infrared image defogging time comparison chart of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0031] Figure 10 This is a piezoelectric sensor workflow diagram of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention.

[0032] Figure 11 This is a design flowchart of an intelligent coal cutting planning method that combines machine vision and cutting process curves, which is optional according to an embodiment of the present invention. Detailed Implementation

[0033] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0034] like Figure 1As shown, an intelligent coal cutting planning method combining machine vision and cutting process curves includes: Step S102: Creating the coal cutting work path for the first job, controlling the forward speed and working direction of the tunneling machine through machine vision and manual adjustment, and recording the position data of the tunneling machine into the database; Step S104: After completing the first job, determining the daily coal cutting work path based on the position data and storing it in the database; Step S106: During daily work, the tunneling machine tunnels according to the daily coal cutting work path, and the tunneling machine corrects the work path in real time through machine vision analysis; Step S108: Acquiring the real-time working status of the tunneling machine and coal and rock acquisition images of the fully mechanized mining face and transmitting them to the background monitoring interface; correcting the working speed and working direction of the tunneling machine in real time based on the real-time working status of the tunneling machine and the coal and rock acquisition images; Step S110: After completing the daily work, updating the database and determining the daily coal cutting work path and storing it in the database for future use.

[0035] As an optimized solution of the present invention, step S102 includes: acquiring coal and rock condition information of the fully mechanized mining face; controlling the working direction of the tunneling machine through a machine vision recognition algorithm based on the coal and rock condition information, while continuously adjusting the forward speed and working direction of the tunneling machine; acquiring the actual underground positioning of the tunneling machine as the longitudinal cutting process, and acquiring the actual forward direction of the tunneling machine as the transverse cutting process; storing the longitudinal cutting process, the transverse cutting process and the corresponding time points into the database.

[0036] As an optimized solution of the present invention, step S104 includes: fitting a three-dimensional coal cutting model for the first operation based on the longitudinal cutting process and the transverse cutting process stored in the database, using the three-dimensional model to determine the daily coal cutting work path and storing it in the database for future operation.

[0037] As an optimized solution of the present invention, step S106 includes: selecting the daily coal cutting work path in the database and controlling the tunneling machine to advance according to the daily coal cutting work path; when the tunneling machine is operating automatically, it analyzes the coal-rock ratio of the fully mechanized mining face in real time through machine vision to correct the work path. If the correction is small, it is included in the work log and fed back to the background; if the correction is large, it slows down and issues a warning in the background for manual processing.

[0038] As an optimized solution of the present invention, step S108 includes: recording the working status of each part of the tunneling machine in real time; when an abnormality is detected in the working status of the tunneling machine, the tunneling machine slows down and issues a warning in the background, awaiting manual handling; during automatic operation, the tunneling machine transmits its working status and coal and rock acquisition images of the fully mechanized mining face back to the background monitoring interface; the management personnel adjust the working speed and working direction of the tunneling machine in real time based on the working status and coal and rock acquisition images. During automatic operation, the tunneling machine transmits its working status back to the background monitoring interface in the form of graphic data in real time, and the coal and rock acquisition images of the fully mechanized mining face are transmitted back to the background monitoring interface in real time via image transmission.

[0039] As an optimized solution of the present invention, step S110 includes: performing error judgment on the actual working path, manually corrected trajectory data and predicted working path stored in the database, selecting and deleting data with large errors in a weighted manner to update the database, and after cleaning and updating the database, forming the coal cutting working path for the next daily work according to the database prediction and storing it in the database for the next operation.

[0040] like Figure 2 As shown, the intelligent coal cutting planning method combining machine vision and cutting process curves provided by this invention realizes intelligent planning of coal cutting work paths by combining machine vision and cutting process curves; the coal and rock identification of machine vision provides operators with information on the coal and rock conditions of the front-line fully mechanized mining face, and machine vision can analyze the coal and rock ratio of the fully mechanized mining face in real time to correct the work path; at the same time, this invention utilizes the mode of a three-party database of cutting process, which records the actual coal cutting path, assists in correcting the path with manual assistance, and uses prediction algorithms to generate predicted paths for the next work.

[0041] like Figure 3 As shown, during the initial operation, the machine vision coal and rock recognition algorithm controls the working direction of the tunneling machine. The operator is responsible for monitoring the working situation at the front of the tunneling machine, adjusting the tunneling machine's forward speed, and continuously assisting in correcting the working direction. After working for a period of time, the first reasonable longitudinal cutting process and transverse cutting process in this mine are recorded and stored in the database according to the time points to form the initial cutting curve.

[0042] like Figure 4As shown, after each operation, the tunneling machine enters a standby mode. During the standby phase following the initial operation, a 3D model of the coal cutting process for that underground operation is fitted based on the existing longitudinal and transverse cutting history curves in the database. This 3D model is then used to predict the coal cutting path for the next operation, and the prediction is stored in the database for future reference. In the standby state after the tunneling machine completes subsequent operations, the actual working path, manually corrected trajectory data, and predicted working path stored in the database are evaluated for errors. Data with errors significantly deviating from the average are weighted and deleted for data cleaning. Then, the existing data in the database is used to predict the next working trajectory, and this prediction is stored in the database.

[0043] like Figure 5 As shown, during routine operations (not the initial setup), the tunneling machine does not require frontline staff to be online. Instead, the safety backend monitors the coal and rock identification data and the tunneling machine's operating status transmitted from the frontline. Management personnel select the predicted coal cutting path from the database, confirm the working environment is safe, and then send a "start operation" command from the backend. The tunneling machine will then proceed along the predicted path. During automatic operation, the tunneling machine uses machine vision to analyze the coal-rock ratio of the fully mechanized mining face in real time to adjust the working path. If the adjustment is small, it is recorded in the work log and fed back to the backend; if the adjustment is large, it slows down and issues a warning in the backend, awaiting manual intervention. During automatic operation, the tunneling machine transmits its real-time operating status back to the backend big data display interface in the form of graphic data. Images of coal and rock collected from the fully mechanized mining face are also transmitted back to the backend monitoring interface in real time via image transmission. Management personnel can adjust the tunneling machine's speed and direction in real time based on the current working status and the existing operating experience of the staff. As the tunneling machine advances along the predicted coal cutting path, the working status of each part of the machine is recorded in real time. If any abnormality is detected, the machine will immediately slow down and issue a warning in the background, awaiting manual intervention.

[0044] like Figure 6 As shown, the process of predicting the working path for the next operation using the database is as follows: After the tunneling machine has been working for a period of time, the first reasonable longitudinal cutting path for this mine is recorded in the database. At the same time, the mine positioning system will transmit its transverse cutting path, which is also stored in the database according to the time point. During the standby phase after the tunneling machine completes the current operation, the actual working path, manually corrected trajectory data, and predicted working path stored in the database are simultaneously evaluated for errors. Data with errors significantly deviating from the average value are weighted and deleted for data cleaning. After data cleaning, a three-dimensional coal cutting model for the current underground operation is fitted based on the existing longitudinal and transverse cutting path curves in the database. The three-dimensional model is used to predict the coal cutting path for the next operation and is stored in the database for future use.

[0045] This invention achieves intelligent planning of coal cutting work paths by combining machine vision and cutting process curves.

[0046] like Figure 11 As shown, the method provided by the present invention has the following characteristics:

[0047] 1. Machine vision output function of coal and rock recognition algorithm based on image feature enhancement technology with specific wavelength polarization characteristics.

[0048] 2. The process of segmenting the data into a three-party database (human intervention input, machine vision planning, and prediction algorithm generation).

[0049] 3. Self-updating and self-cleaning functions of the three-party databases of the truncation process (human intervention input, machine vision planning, and prediction algorithm generation).

[0050] 4. The function of combining machine vision with coal cutting path planning.

[0051] On the one hand, the reduced number of operators for tunneling machines in fully mechanized mining faces lowers the potential for underground work hazards and losses, increases personnel utilization, and reduces labor costs. On the other hand, the reduced contact between workers and the front-line workers lowers the risk to their health, aligning with the people-oriented and safety-first safety philosophy of my country's coal mining industry.

[0052] On the other hand, the tunneling machine's cutting path planning is more rational, improving coal production efficiency, reducing the proportion of rock extracted during mining, lowering the wear rate of cutting teeth, increasing energy utilization, and thus significantly reducing mining costs. Simultaneously, dust levels during mining are reduced due to the lower proportion of extracted rock, leading to less air and solid pollutant generation and emissions, which aligns with my country's green mining development philosophy.

[0053] Furthermore, this invention truly achieves a balance of control between humans and machines. Workers have the highest authority over the tunneling machine's trajectory planning, while the machine simultaneously learns from manually corrected trajectories by recording them in its database for self-cleaning and updating. During operation, the system integrates the planning experience of workers, continuously improving their decision-making capabilities. The machine vision's coal and rock recognition function also corrects and prompts both the system's predicted and manually planned paths, improving the system's real-time performance and accuracy. Workers can also gain a better understanding of the actual situation on-site, obtaining more information from the front-line working face, thus enhancing the accuracy and scientific rigor of their judgments.

[0054] This invention improves the scientific rigor, accuracy, and real-time performance of tunneling machine path planning, thereby increasing coal production efficiency, reducing coal mine safety accidents, and effectively implementing my country's development concepts of smart and green mines. This invention addresses the problem that existing coal mining machine memory cutting technology relies too heavily on manual labor for path planning or lacks the ability to adapt to changes in the actual working environment.

[0055] This invention provides an intelligent coal cutting planning method that combines machine vision and cutting process curves. The main contents of the method are illustrated below with reference to a specific embodiment:

[0056] (1) Upon initial use, the first coal cutting working path is created. The machine vision coal and rock recognition technology provides operators with information on the coal and rock conditions of the fully mechanized mining face. At this time, the machine vision coal and rock recognition algorithm controls the working direction of the tunneling machine. The operator is responsible for monitoring the working conditions of the tunneling machine, adjusting the tunneling machine's forward speed, and continuously correcting the working direction. After working for a period of time, the first reasonable longitudinal and transverse cutting process obtained by combining the underground positioning system is recorded in the database and stored in the database according to the time points. The transverse cutting process comes from the cutting path of the working face corresponding to the current working point of the tunneling machine, and the longitudinal cutting process comes from the displacement path of the tunneling face working point of the underground positioning system.

[0057] The machine vision coal and rock recognition technology of this invention has the machine vision output function of a coal and rock recognition algorithm based on image feature enhancement technology with specific wavelength polarization characteristics. The hardware uses a domestically produced ARM architecture chip, and utilizes the chip's unique NPU neural network processing unit to improve the running speed of the machine vision algorithm. It employs an improved DeeplabV3+ algorithm of PSPNET, adding a CBAM attention mechanism. The improved PSPNET network structure is as follows: Figure 7 As shown.

[0058] To achieve better image clarity and overcome image interference caused by water mist and coal dust in underground mining environments, as well as to remove noise and unnecessary information, the following methods were compared on infrared and non-infrared images: channel prior method, multi-scale Retinex, Gaussian dehazing, median dehazing, adaptive color level contrast dehazing, and adaptive histogram dehazing. Ultimately, the dark channel dehazing algorithm was selected. The dark channel dehazing algorithm utilizes the dark channel prior principle, the mathematical principle of which is as follows:

[0059] The fog formation formula is expressed as:

[0060]

[0061] Where J is the dark channel in the image.

[0062] The relevant calculation formula for the image after dehazing is expressed as follows:

[0063] I(x)=J(x)t(x)+A(1-t(x))

[0064] Where I(x) is the original image, J(x) is the image after dehazing, A is the global atmospheric light composition, and t(x) is the projective rate.

[0065] Through a series of mathematical transformations, the formula for calculating J(x) can be obtained as follows:

[0066]

[0067] like Figure 8 and Figure 9 As shown in the figure, the time consumption of the three dehazing techniques—dark channel, adaptive histogram, and adaptive levels—is similar. However, analysis of the processed images reveals that the dark channel algorithm achieves better dehazing results than the adaptive histogram and adaptive levels algorithms. It is worth noting that the adaptive histogram and adaptive levels algorithms make significant adjustments to the image's color tone, which may alter the semantics of the image. The dark channel algorithm, on the other hand, better preserves the image's detail information. Therefore, the dark channel dehazing algorithm was ultimately selected for preprocessing.

[0068] After preprocessing with the dark channel defogging algorithm, the underground image is clearer, demonstrating excellent defogging performance. The recognition parameters obtained using the defogging technology are all higher than the original parameters. The defogging algorithm increases the clarity of coal in the image and improves processing and recognition performance in the network. Specific performance indicators are shown in Tables 1 and 2.

[0069] Table 1 Performance indicators of different network models before defogging

[0070]

[0071] Table 2 Performance indicators of different network models after defogging

[0072]

[0073] Machine vision route planning uses a depth-first traversal search algorithm to obtain the current distribution location of the coal face and construct an optimal working route.

[0074] (2) During the standby phase after the tunneling machine completes the current operation, a three-dimensional coal cutting model for the current underground operation is fitted based on the existing longitudinal and transverse cutting process curves in the database. The three-dimensional model is used to predict the coal cutting path for the next operation and is stored in the database for future use.

[0075] The transverse cutting process curve is acquired using monocular camera ranging technology and dynamically stored as a two-dimensional array to fit the two-dimensional shape of the currently mined coal seam. A three-dimensional model, representing the distribution of the coal seam, is derived using a Cartesian coordinate system to ensure that the coal seam can be mined using this three-dimensional model even in fault conditions, allowing mining to proceed beyond the fault and reach the coal seam beyond the fault. The output of the semantic segmentation coal and rock recognition algorithm is fully utilized to fit a three-dimensional model of the current coal and rock area. The coal and rock feature information database in the system is used to estimate the current coal and rock reserves in the area.

[0076] (3) When the next operation starts, the tunneling machine does not require front-line staff to be online. Instead, the safety backend monitors the coal and rock identification data and the tunneling machine's working status transmitted from the front-line staff. The management personnel select the predicted coal cutting path in the database, and after confirming that the working environment is safe, they send the "start operation" command in the backend. The tunneling machine will then proceed along the predicted path.

[0077] (4) When the tunneling machine advances along the predicted coal cutting path, the working status of each part of the tunneling machine is recorded in real time. If any abnormality is found, the speed is reduced immediately and a warning is issued in the background, waiting for manual handling.

[0078] When the tunneling machine is moving forward, the following data needs to be recorded: 1. Electrical system operating status: working power supply voltage, working power supply current, oil pump power, conveyor current, and spray dust suppression system power. 2. Hydraulic system operating status: oil level, oil pressure, oil temperature, and flow rate. 3. Working environment conditions: dust concentration, gas concentration, air temperature, and humidity.

[0079] This invention introduces a piezoelectric sensor into a tunneling machine, solving the problem of damage to coal mining machinery caused by contact between the machinery and non-coal seams due to single-path planning. Its workflow is as follows: Figure 10 As shown.

[0080] (5) When the tunneling machine is operating automatically, it will use machine vision to analyze the coal and rock ratio of the fully mechanized mining face in real time to correct the working path. If the correction is small, it will be included in the work log and fed back to the background; if the correction is large, it will slow down and issue a warning in the background for manual processing.

[0081] Work logs can be directly exported as tables, which can be Excel spreadsheets that can be automatically read by the database.

[0082] (6) When the tunneling machine is operating automatically, it transmits its real-time working status back to the background big data display interface in the form of graphic data. The coal and rock acquisition images of the fully mechanized mining face are also transmitted back to the background monitoring interface in real time via image transmission. Managers can adjust the working speed and direction of the tunneling machine in real time based on the current working status and the existing operating experience of the staff.

[0083] (7) When the tunneling machine finishes a non-first operation, it enters the standby state. At the same time, it performs error judgment on the actual working path, manually corrected trajectory data and predicted working path stored in the database. The error with the larger value is selected for deletion by weighting. The data is cleaned. Then, the existing data in the database is used to predict the working trajectory for the next operation and stored in the database.

[0084] Data cleaning is the process of using existing technologies, including data mining, mathematical statistics, and machine learning, to process dirty data. Through data cleaning, high-quality datasets that meet requirements are obtained, thus preparing the data for subsequent data mining. Data cleaning employs a backtracking approach, analyzing from the source of the dirty data, examining each process the dataset undergoes, and extracting data cleaning rules and strategies. Finally, these rules and strategies are applied to identify and clean dirty data. The strength of these cleaning rules and strategies determines the quality of data cleaning.

[0085] Data cleaning includes cleaning erroneous records, cleaning incomplete records, and cleaning duplicate records.

[0086] For data cleaning of erroneous records, a common sense database and a business rule database are established based on records of normal working paths. By comparing erroneous data with the rule database, non-existent or completely erroneous data is identified and corrected. By analyzing the overall distribution of data in the database, data objects that deviate from the distribution are identified, and statistical information is used to correct these erroneous data. Alternatively, the relationships or constraints between attributes can be utilized to identify and correct non-compliant erroneous attributes.

[0087] There are three methods for cleaning incomplete records: (1) Determine the missing values ​​of a field based on the relationships between fields. (2) Fill in the missing value field using the average, minimum, maximum, or mode of the field, or by using a probability function to calculate the most likely result. The accuracy varies greatly depending on the specific circumstances. (3) Manually fill in the missing values ​​based on experience in the field. (4) Directly delete records with missing values.

[0088] For data cleaning involving duplicate records, duplicate records are commonly found during the merging of multiple data sources, where the same record may appear differently in different data sources. The process for handling duplicate records generally includes comparison and merging. By comparing the datasets, identical or similar data is identified, and then a decision is made, based on the actual situation verified manually, whether to retain the data or merge it into a single record.

[0089] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart coal cutting planning method combining machine vision and cutting process curves, characterized in that, include: Step S102: Create the coal cutting working path for the first operation, control the forward speed and working direction of the tunneling machine through machine vision and manual adjustment, and enter the position data of the tunneling machine into the database; Step S104: After completing the initial work, determine the daily coal cutting work path based on the location data and store it in the database; wherein, step S104 includes: obtaining the actual underground positioning of the tunneling machine as the longitudinal cutting process, obtaining the actual forward direction of the tunneling machine as the lateral cutting process, storing the longitudinal cutting process, the lateral cutting process and the corresponding time points in the database; fitting a three-dimensional coal cutting model for the initial work based on the longitudinal cutting process and the lateral cutting process stored in the database, using the three-dimensional model to determine the daily coal cutting work path and storing it in the database for future use; Step S106: During daily operation, the tunneling machine advances according to the daily coal cutting work path. The tunneling machine corrects the work path in real time through machine vision analysis. Step S106 includes selecting the daily coal cutting work path from the database and controlling the tunneling machine to advance along that path. During automatic operation, the tunneling machine uses machine vision to analyze the coal-rock ratio of the fully mechanized mining face in real time to correct the work path. If the correction is small, it is recorded in the work log and fed back to the backend; if the correction is large, it slows down and issues a warning in the backend, awaiting manual intervention. Step S108: Acquire the real-time working status of the tunneling machine and the coal and rock acquisition images of the fully mechanized mining face and transmit them to the background monitoring interface; adjust the working speed and working direction of the tunneling machine in real time according to the real-time working status of the tunneling machine and the coal and rock acquisition images. Step S110: After completing daily work, update the database and determine the daily coal cutting work path and store it in the database for future use; wherein, step S110 includes: judging the error of the actual work path, manually corrected trajectory data and predicted work path stored in the database, selecting and deleting data with large errors in a weighted manner to update the database, and after cleaning and updating the database, predicting the coal cutting work path for the next daily work based on the database and storing it in the database for future use.

2. The intelligent coal cutting planning method combining machine vision and cutting process curves according to claim 1, characterized in that, Step S102 includes: Obtain information on the coal and rock conditions of the fully mechanized mining face; Based on the coal and rock conditions, the machine vision recognition algorithm controls the working direction of the tunneling machine and continuously adjusts its forward speed and working direction.

3. The intelligent coal cutting planning method combining machine vision and cutting process curves according to claim 1, characterized in that, Step S108 includes: The working status of each part of the tunneling machine is recorded in real time. When an abnormality is detected in the working status of the tunneling machine, the tunneling machine slows down and issues a warning in the background, waiting for manual handling. When the tunneling machine is operating automatically, it transmits its working status and coal and rock images of the fully mechanized mining face back to the background monitoring interface; the management personnel adjust the working speed and working direction of the tunneling machine in real time based on the working status and the coal and rock images.

4. The intelligent coal cutting planning method combining machine vision and cutting process curves according to claim 3, characterized in that, When the tunneling machine is operating automatically, it transmits its working status back to the background monitoring interface in the form of graphic data in real time. The coal and rock images collected from the fully mechanized mining face are also transmitted back to the background monitoring interface in real time via image transmission.

Images