Continuous mining and cutting control method for heterogeneous hard rock strata based on neural network

Abstract

The present invention discloses a method for controlling continuous mining and cutting of heterogeneous hard rock formations based on a neural network, including the establishment and training of a deep neural network model, the establishment and training of a convolutional neural network model, the testing of a pre-drilling fracturing operation, the extraction and updating of rock formation characteristics after fracturing, the prediction and control of cutting ratio energy consumption, the pre-drilling fracturing operation and the continuous mining operation, the visual recognition of an open-pit mine working face and the updating of a second deep neural network model; the pre-drilling fracturing operation is controlled and adjusted according to a cutting ratio energy consumption index to ensure that the cuttability of the heterogeneous hard rock formation meets the requirements; the actual cutting ratio energy consumption value is monitored in real time, the deep neural network model is updated, the threshold of the cutting ratio energy consumption is adjusted, and then the technical parameters of the cutting mechanism are adjusted, thereby achieving real-time control and adjustment of the continuous mining operation, and improving the economy and efficiency of the continuous mining operation of the open-pit mine under the heterogeneous hard rock formation environment.

Description

Technical Field

[0001] The invention belongs to the technical field of neural network models and relates to a continuous excavation and cutting control method for a heterogeneous hard rock layer based on a neural network. Background Art

[0002] The neural network model is a mathematical model of distributed parallel information processing built by simulating the working principle of neurons in the human brain. Its core idea is to achieve feature extraction and abstract expression of input data through layer-by-layer learning of multi-layer neurons and gradual adjustment of network weights. Multi-layer neural network technology represented by deep neural network models and convolutional neural network models realizes the automatic learning of high-order complex features and the expression of complex relationships, broadening the application scope of neural network model technology. With the rapid development of computing power, neural network model technology is widely used in various industries, especially in the mining industry, which has rich practical experience but relatively scarce theoretical results.

[0003] Mechanized mining of hard rock formations in open-pit mines is a development trend in the current mining industry. This mining technology avoids the adverse effects of rock instability, safety accidents and environmental damage that may be caused by blasting operations and tunnel excavation operations, and is conducive to the realization of continuous mining of hard rock formations without blasting, safety, economy and environmental protection. In mechanized mining operations of hard rock formations in open-pit mines, pre-drilling fracturing technology is usually used to assist in rock breaking, that is, after drilling a test hole, non-blasting auxiliary rock breaking methods are used to achieve fracturing of the rock formations around the test hole, resulting in expansion cracks in the hard rock formations around the test hole, thereby reducing the difficulty of cutting, so that open-pit continuous mining equipment can complete mining operations safely, economically and efficiently. However, on the one hand, under the condition that the technical parameters of pre-drilling and fracturing remain unchanged, the heterogeneity of hard rock formations, the nonlinearity of rock formation mechanical properties and the randomness of microcrack expansion will lead to a large degree of discreteness in the strength characteristics of rock formations after pre-drilling and fracturing, which will in turn affect the stability of continuous mining operations. Inhomogeneous hard rock formations may have parts that are difficult to cut and peel off, which is likely to cause difficulties in cutting and damage to picks in open-pit continuous mining equipment. The current method of identifying rock formation cracks after pre-drilling and fracturing operations by naked eyes is neither accurate nor efficient, and it is even more impossible to predict and control the cuttable characteristics of rock formations after pre-drilling and fracturing operations.

[0004] On the other hand, the cuttability of rock formations is usually measured by compressive strength, Procter & Gamble hardness or specific energy consumption for cutting. Specifically, specific energy consumption for cutting refers to the energy consumed when cutting a unit volume of rock formation. Compared with compressive strength or Procter & Gamble hardness, which can only reflect the firmness of the rock formation, specific energy consumption for cutting can quantitatively reflect the cuttability of the rock formation in terms of rock breaking efficiency and economy.

[0005] Therefore, it is particularly important to monitor the cutting specific energy consumption in real time during open-pit continuous mining of heterogeneous hard rock formations, and to predict and control the cutting characteristics of the rock formation after pre-drilling and fracturing operations. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a neural network-based continuous mining and cutting control method for heterogeneous hard rock strata, which is suitable for the automatic control and adjustment of the cutting operation of open-pit continuous mining equipment in a heterogeneous hard rock environment, which is beneficial to improving the mining efficiency and reducing the energy loss of the cutting mechanism, and realizing the control of the cutting operation and the automation, safety and efficiency of the continuous mining operation of the hard rock strata.

[0007] The present application discloses a control method for continuous excavation and cutting of heterogeneous hard rock layers based on a neural network, the control method comprising the following steps:

[0008] S100, establishment and training of a deep neural network model: establishing an initial deep neural network model, including a data input layer, a plurality of hidden layers and an output layer; screening first historical data accumulated and stored in past mining operations to obtain first mining history data under heterogeneous hard rock conditions; using the drilling technical parameters in the first mining history data as input values ​​and the rock formation characteristic parameters after fracture as output values, training the initial deep neural network model to obtain a first deep neural network model; using the drilling technical parameters, rock formation characteristic parameters after fracture and cutting mechanism technical parameters in the first mining history data as input values ​​and the cutting specific energy consumption as output values, training the initial deep neural network model to obtain a second deep neural network model;

[0009] S200, establishment and training of a convolutional neural network model: establishing an initial convolutional neural network model, including an image input layer, a convolutional layer, a pooling layer, and a fully connected layer; screening the second historical data accumulated and stored in the past mining operations to obtain the second mining history data under the condition of heterogeneous hard rock formation; using the image of the rock formation around the test hole in the second mining history data as an input value and the characteristic parameters of the rock formation after fracture as an output value, training the initial convolutional neural network model to obtain a first convolutional neural network model; using the image of the mining situation of the open-pit mine working face in the second mining history data as an input value and the geometric morphology parameters of the open-pit mine working face as an output value, training the initial convolutional neural network model to obtain a second convolutional neural network model;

[0010] S300, testing of pre-drilling fracturing operation: dividing the open-pit mine working face into several zones, selecting any zone to drill a test hole, and monitoring the first drilling technical parameters during the drilling process; then using non-blasting auxiliary rock breaking technology to fracture the surrounding rock formations of the test hole, so that expansion cracks appear in the heterogeneous hard rock formations of the open-pit mine working face along the periphery of the test hole;

[0011] S400, extracting and updating the characteristics of the rock formation after fracturing: obtaining an image of the rock formation around the test hole through a visual sensor device mounted on the open-pit continuous mining equipment, inputting the image into the first convolutional neural network model, and obtaining characteristic parameters of the rock formation after fracturing; then inputting the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the first deep neural network model and updating them, thereby strengthening the correspondence between the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing;

[0012] S500, prediction and control of cutting ratio energy consumption: inputting the technical parameters of the cutting mechanism, the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the second deep neural network model, and predicting the cutting ratio energy consumption of the rock formation around the test hole; when the predicted value of the cutting ratio energy consumption exceeds the first threshold value of the cutting ratio energy consumption, adjusting the first drilling technical parameters to obtain the second drilling technical parameters, re-predicting the characteristic parameters of the rock formation after fracturing, and then re-obtaining the predicted value of the cutting ratio energy consumption, until the predicted value of the cutting ratio energy consumption is lower than the first threshold value, and then using the second drilling technical parameters at this time as the drilling technical parameters adopted in the pre-drilling fracturing operation of the partition;

[0013] S600, pre-drilling fracturing operation and continuous mining operation: repeating steps S300 to S500 until the second drilling technical parameters of all partitions of the open-pit mine working face are determined, and then performing pre-drilling fracturing operation; then operating the open-pit continuous mining equipment to perform continuous mining operation according to the cutting mechanism technical parameters;

[0014] S700, visual recognition of open-pit mine working face: obtaining images of the mining situation of the open-pit mine working face in real time through a visual sensor device mounted on the open-pit continuous mining equipment, and extracting geometric parameters of the open-pit mine working face from the images using the second convolutional neural network model, thereby establishing and continuously updating a three-dimensional model of the open-pit mine working face;

[0015] S800, update of the second deep neural network model: t The volume reduction of the three-dimensional model of the open-pit mine working face during the period is t The volume of rock cut during the period V t ; then the actual cutting specific energy consumption E As shown in the following expression:

[0016]

[0017] in P is the cutting mechanism power, V t for t The volume of rock formation cut during the time period;

[0018] The second drilling technical parameters, the rock formation characteristic parameters after fracturing, the cutting mechanism technical parameters and the actual cutting specific energy consumption are calculated. E Input into the second deep neural network model and update it to improve the accuracy of the cutting specific energy consumption prediction.

[0019] Preferably, in step S100, the data input layer is the first layer of the initial deep neural network model, used to receive input data; the output layer is the last layer of the initial deep neural network model, used to generate prediction results; the hidden layer is a layer located between the input layer and the output layer, used for data processing and feature extraction.

[0020] Preferably, in step S200, the image input layer is the first layer of the initial convolutional neural network model, which is used to receive image data; the convolutional layer is located after the image input layer, which is used to extract local features in the image data; the pooling layer is located after the convolutional layer, which is used to reduce the spatial dimension of the feature map and retain important feature information; the fully connected layer is located at the end of the initial convolutional neural network model, which is used for the integration and classification of feature information.

[0021] Preferably, in step S300, the drilling technical parameters include hole depth, hole diameter, hole spacing, drilling speed, drilling energy consumption and original rock formation characteristic parameters obtained from geological exploration data.

[0022] Preferably, in step S300, the non-explosion assisted rock breaking technology includes a carbon dioxide phase change fracturing technology; the carbon dioxide phase change fracturing technology utilizes the physical expansion characteristics generated during the phase change of carbon dioxide to cause fractures in the surrounding rock formations, and its specific operating steps are: sealing liquid carbon dioxide in a dedicated pipe and placing it in a test hole, heating the liquid carbon dioxide with a heating device to cause it to undergo a vaporization phase change and expand rapidly, and then forming carbon dioxide high-pressure gas, which is discharged from the dedicated pipe and generates a tensile stress field in the rock formations around the test hole, thereby achieving rock formation fracturing.

[0023] Preferably, in step S400, the characteristic parameters of the fractured rock formation include spatial distribution and geometric size parameters of cracks, joints and layers.

[0024] Preferably, in step S500, the technical parameters of the cutting mechanism include the cutting mechanism linear velocity, the cutting mechanism angular velocity, the cutting mechanism output power and the cutting depth.

[0025] Preferably, during the continuous mining operation, the measured value of the power of the cutting mechanism is monitored in real time. When the measured value of the power of the cutting mechanism exceeds a threshold value, the open-pit continuous mining equipment is controlled to stop the continuous mining operation, and a second threshold value of the cutting specific energy consumption is calculated. The second threshold value of the cutting specific energy consumption is [ E 2 ] satisfies the following expression:

[0026]

[0027] in,[ P ] is the threshold of cutting mechanism power, V t for t The volume of rock formation cut during the time period;

[0028] By adjusting the technical parameters of the cutting mechanism and using the second deep neural network model to re-obtain the predicted value of the cutting energy consumption ratio until it is lower than the second threshold of the cutting energy consumption ratio, the adjusted technical parameters of the cutting mechanism are then used as the cutting mechanism technical parameters adopted in the continuous mining operation, and the continuous mining operation is restarted.

[0029] Compared with the prior art, the beneficial effects of the present invention are: for the automatic control and adjustment of the cutting operation of open-pit continuous mining equipment in a heterogeneous hard rock environment, a neural network-based heterogeneous hard rock continuous mining and cutting control method is disclosed, the control method includes the establishment and training of a deep neural network model, the establishment and training of a convolutional neural network model, the testing of pre-drilling and fracturing operations, the extraction and updating of rock formation characteristics after fracturing, the prediction and control of cutting specific energy consumption, pre-drilling and fracturing operations and continuous mining operations, visual recognition of open-pit mine working faces and the updating of a second deep neural network model; wherein, while improving the cuttability of heterogeneous hard rock formations through pre-drilling and fracturing operations, a corresponding relationship between drilling technical parameters and rock formation characteristic parameters after fracturing is established. , the correspondence between the characteristic parameters of the rock formation after fracturing and the cutting energy consumption ratio, so as to control and adjust the pre-drilling fracturing operation according to the cutting energy consumption ratio index, to ensure that the cuttability of the heterogeneous hard rock formation meets the requirements of continuous mining operation for economy and efficiency; to monitor the actual cutting energy consumption ratio value in real time through the visual recognition method of the open-pit mine working face and update the deep neural network model to improve the accuracy of the prediction of the cutting energy consumption ratio value; to avoid overload operation of open-pit continuous mining equipment by monitoring and controlling the power value of the cutting mechanism, and then adjust the threshold of the cutting energy consumption ratio and then adjust the technical parameters of the cutting mechanism, so as to realize real-time control and adjustment of continuous mining operations, and improve the economy and efficiency of open-pit continuous mining operations in heterogeneous hard rock environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a flow chart of the method for controlling continuous mining and cutting of heterogeneous hard rock layers based on neural network of the present invention;

[0031] Figure 2 A schematic diagram of the structure of an open-pit continuous mining device provided by an embodiment of the present invention;

[0032] Figure 3 A schematic flow chart of a method for controlling continuous mining and cutting of heterogeneous hard rock layers based on a neural network provided in one embodiment of the present invention;

[0033] Figure numerals: 11 - cutting mechanism, 12 - open-pit mine working face, 13 - test hole. DETAILED DESCRIPTION

[0034] The following is combined with Figure 1-3 The embodiments of the present invention are described in more detail with reference to the accompanying drawings so that a person skilled in the art can implement the present invention after reading the present description. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0035] This application discloses Figures 1 to 3The neural network-based continuous mining and cutting control method for heterogeneous hard rock formations shown in the figure specifically includes the following steps:

[0036] S100, establishment and training of a deep neural network model: establishing an initial deep neural network model, including a data input layer, a plurality of hidden layers and an output layer; screening first historical data accumulated and stored in past mining operations to obtain first mining history data under heterogeneous hard rock conditions; using the drilling technical parameters in the first mining history data as input values ​​and the rock formation characteristic parameters after fracture as output values, training the initial deep neural network model to obtain a first deep neural network model; using the drilling technical parameters, rock formation characteristic parameters after fracture and cutting mechanism technical parameters in the first mining history data as input values ​​and the cutting specific energy consumption as output values, training the initial deep neural network model to obtain a second deep neural network model;

[0037] In a specific implementation, the data input layer is the first layer of the initial deep neural network model, which is used to receive input data; the output layer is the last layer of the initial deep neural network model, which is used to generate prediction results; the hidden layer is a layer located between the input layer and the output layer, which is used for data processing and feature extraction;

[0038] S200, establishment and training of a convolutional neural network model: establishing an initial convolutional neural network model, including an image input layer, a convolutional layer, a pooling layer, and a fully connected layer; screening the second historical data accumulated and stored in the past mining operations to obtain the second mining history data under the condition of heterogeneous hard rock formation; using the image of the rock formation around the test hole in the second mining history data as an input value and the characteristic parameters of the rock formation after fracture as an output value, training the initial convolutional neural network model to obtain a first convolutional neural network model; using the image of the mining situation of the open-pit mine working face in the second mining history data as an input value and the geometric morphology parameters of the open-pit mine working face as an output value, training the initial convolutional neural network model to obtain a second convolutional neural network model;

[0039] In a specific implementation, the image input layer is the first layer of the initial convolutional neural network model, which is used to receive image data and pass the image data to subsequent layers after preprocessing; the convolution layer is located after the image input layer, and extracts local features in the input image data by applying a convolution kernel to slide on the image data and calculating the dot product between the convolution kernel and the local area of ​​the image; the pooling layer is located after the convolution layer, and is used to reduce the spatial dimension of the feature map and retain important feature information; the fully connected layer is located at the end of the initial convolutional neural network model, and maps the extracted features to the final classification or regression task through weighted summation of neurons and activation function processing, thereby realizing the integration and classification of feature information;

[0040] S300, testing of pre-drilling fracturing operation: dividing the open-pit mine working face into several zones, selecting any zone to drill a test hole, and monitoring the first drilling technical parameters during the drilling process; then using non-blasting auxiliary rock breaking technology to fracture the surrounding rock formations of the test hole, so that expansion cracks appear in the heterogeneous hard rock formations of the open-pit mine working face along the periphery of the test hole;

[0041] In a specific implementation, the drilling technical parameters include hole depth, hole diameter, hole spacing, drilling speed, drilling energy consumption and original rock formation characteristic parameters obtained from geological exploration data;

[0042] In specific implementation, the non-explosion-assisted rock breaking technology adopts carbon dioxide phase change fracturing technology; the carbon dioxide phase change fracturing technology uses the physical expansion characteristics generated during the phase change of carbon dioxide to cause fractures in the surrounding rock formations, and its specific operation steps are: sealing liquid carbon dioxide in a special pipe and placing it in a test hole, using a heating device to heat the liquid carbon dioxide to cause it to undergo a vaporization phase change and expand rapidly, and then forming carbon dioxide high-pressure gas, which is discharged from the special pipe and generates a tensile stress field in the rock formations around the test hole, thereby achieving rock formation fracturing;

[0043] S400, extracting and updating the characteristics of the rock formation after fracturing: obtaining an image of the rock formation around the test hole through a visual sensor device mounted on the open-pit continuous mining equipment, inputting the image into the first convolutional neural network model, and obtaining characteristic parameters of the rock formation after fracturing; then inputting the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the first deep neural network model and updating them, so as to strengthen the correspondence between the drilling technical parameters and the characteristic parameters of the rock formation after fracturing;

[0044] In a specific implementation, the characteristic parameters of the fractured rock formation include the spatial distribution and geometric size parameters of cracks, joints, and layers;

[0045] S500, prediction and control of cutting ratio energy consumption: inputting the technical parameters of the cutting mechanism, the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the second deep neural network model, and predicting the cutting ratio energy consumption of the rock formation around the test hole; when the predicted value of the cutting ratio energy consumption exceeds the first threshold value of the cutting ratio energy consumption, adjusting the first drilling technical parameters to obtain the second drilling technical parameters, re-predicting the characteristic parameters of the rock formation after fracturing, and then re-obtaining the predicted value of the cutting ratio energy consumption, until the predicted value of the cutting ratio energy consumption is lower than the first threshold value, and then using the second drilling technical parameters at this time as the drilling technical parameters adopted in the pre-drilling fracturing operation of the partition;

[0046] In a specific implementation, the technical parameters of the cutting mechanism include the cutting mechanism linear velocity, the cutting mechanism angular velocity, the cutting mechanism output power and the cutting depth;

[0047] In the specific implementation, in order to improve the economy and efficiency of continuous mining operations, the first threshold of cutting specific energy consumption [ E 1 ] is taken as 0.4 kW·h·m -3 ;

[0048] S600, pre-drilling fracturing operation and continuous mining operation: repeating steps S300 to S500 until the second drilling technical parameters of all partitions of the open-pit mine working face are determined, and then performing pre-drilling fracturing operation, that is, after drilling the test hole, using a non-blasting auxiliary rock breaking method to achieve fracturing of the rock formation around the test hole; then operating the open-pit continuous mining equipment to perform continuous mining operation according to the technical parameters of the cutting mechanism;

[0049] S700, visual recognition of open-pit mine working face: obtaining images of the mining situation of the open-pit mine working face in real time through a visual sensor device mounted on the open-pit continuous mining equipment, and extracting geometric parameters of the open-pit mine working face from the images using the second convolutional neural network model, thereby establishing and continuously updating a three-dimensional model of the open-pit mine working face;

[0050] In the specific implementation, the initial open-pit mine working face 3D model dataset M 0 for[ M x0 , M y0 , M z0 ], t 3D model dataset of open-pit mine working face after the period M t for[ M xt , M yt , M zt ];

[0051] S800, update of the second deep neural network model: t The volume reduction of the three-dimensional model of the open-pit mine working face during the period is t The volume of rock cut during the period V t ; then the actual cutting specific energy consumption E As shown in the following expression:

[0052] (1)

[0053] inP is the cutting mechanism power, V t for t The volume of rock formation cut during the time period;

[0054] The drilling technical parameters, the rock formation characteristic parameters after fracturing, the cutting mechanism technical parameters and the actual cutting specific energy consumption are calculated. E Inputting into the second deep neural network model and updating it, so as to improve the accuracy of the prediction of the cutting specific energy consumption;

[0055] In specific implementation, t The volume of rock cut during the period V t 3D model dataset for open pit working face M 0 and M t The volume of the difference of .

[0056] In a specific implementation, the measured value of the power of the cutting mechanism is monitored in real time during the continuous mining operation. When the measured value of the power of the cutting mechanism exceeds its threshold, the open-pit continuous mining equipment is controlled to stop the continuous mining operation, and a second threshold of the cutting specific energy consumption is calculated. The second threshold of the cutting specific energy consumption [ E 2 ] satisfies the following expression:

[0057] (2)

[0058] in,[ P ] is the threshold of cutting mechanism power, V t for t The volume of rock formation cut during the time period;

[0059] By adjusting the technical parameters of the cutting mechanism and using the second deep neural network model to re-obtain the predicted value of the cutting energy consumption ratio until it is lower than the second threshold of the cutting energy consumption ratio, the adjusted technical parameters of the cutting mechanism are then used as the cutting mechanism technical parameters adopted in the continuous mining operation, and the continuous mining operation is restarted.

[0060] It can be seen that while the cuttability of heterogeneous hard rock formations is improved through pre-drilling fracturing operations, the corresponding relationship between drilling technical parameters and characteristic parameters of rock formations after fracturing, and the corresponding relationship between characteristic parameters of rock formations after fracturing and cutting energy consumption ratio are established, so that the pre-drilling fracturing operations are controlled and adjusted according to the cutting energy consumption ratio index to ensure the cuttability of heterogeneous hard rock formations; the actual cutting energy consumption ratio value is monitored in real time through the visual recognition method of the open-pit mine working face and the deep neural network model is updated to improve the accuracy of the prediction of the cutting energy consumption ratio value; by monitoring and controlling the power value of the cutting mechanism, the overload operation of the open-pit continuous mining equipment is avoided, and then the threshold of the cutting energy consumption ratio is adjusted to adjust the technical parameters of the cutting mechanism, so as to realize real-time control and adjustment of the continuous mining operation, and improve the economy and efficiency of the continuous mining operation of the open-pit mine under the heterogeneous hard rock environment.

[0061] The above are one or more embodiments of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

Claims

1. A method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on a neural network, characterized in that: The following steps are involved: S100, establishment and training of a deep neural network model: establishing an initial deep neural network model, including a data input layer, a plurality of hidden layers and an output layer; screening first historical data accumulated and stored in past mining operations to obtain first mining history data under heterogeneous hard rock conditions; using the drilling technical parameters in the first mining history data as input values ​​and the rock formation characteristic parameters after fracture as output values, training the initial deep neural network model to obtain a first deep neural network model; using the drilling technical parameters, rock formation characteristic parameters after fracture and cutting mechanism technical parameters in the first mining history data as input values ​​and the cutting specific energy consumption as output values, training the initial deep neural network model to obtain a second deep neural network model; S200, establishment and training of a convolutional neural network model: establishing an initial convolutional neural network model, including an image input layer, a convolutional layer, a pooling layer, and a fully connected layer; screening the second historical data accumulated and stored in the past mining operations to obtain the second mining history data under the condition of heterogeneous hard rock formation; using the image of the rock formation around the test hole in the second mining history data as an input value and the characteristic parameters of the rock formation after fracture as an output value, training the initial convolutional neural network model to obtain a first convolutional neural network model; using the image of the mining situation of the open-pit mine working face in the second mining history data as an input value and the geometric morphology parameters of the open-pit mine working face as an output value, training the initial convolutional neural network model to obtain a second convolutional neural network model; S300, testing of pre-drilling fracturing operation: dividing the open-pit mine working face into several zones, selecting any zone to drill a test hole, and monitoring the first drilling technical parameters during the drilling process; Then, non-blasting assisted rock breaking technology is used to fracture the surrounding rock formations of the test hole; S400, extracting and updating the characteristics of the rock formation after fracturing: obtaining an image of the rock formation around the test hole through a visual sensing device mounted on the open-pit continuous mining equipment, inputting the image into the first convolutional neural network model, and obtaining characteristic parameters of the rock formation after fracturing; then inputting the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the first deep neural network model and updating them; S500, prediction and control of cutting specific energy consumption: inputting the technical parameters of the cutting mechanism, the first drilling technical parameters and the characteristic parameters of the rock formation after fracturing into the second deep neural network model to predict the cutting specific energy consumption of the rock formation around the test hole; When the predicted value of the cutting ratio energy consumption exceeds the first threshold value of the cutting ratio energy consumption, the second drilling technical parameter is obtained by adjusting the first drilling technical parameter, the characteristic parameter of the rock formation after fracturing is re-predicted, and then the predicted value of the cutting ratio energy consumption is re-obtained, until the predicted value of the cutting ratio energy consumption is lower than the first threshold value, and then the second drilling technical parameter at this time is used as the drilling technical parameter adopted in the pre-drilling fracturing operation of the partition; S600, pre-drilling fracturing operation and continuous mining operation: repeating steps S300 to S500 until the second drilling technical parameters of all partitions of the open-pit mine working face are determined, and then performing pre-drilling fracturing operation; then operating the open-pit continuous mining equipment to perform continuous mining operation according to the cutting mechanism technical parameters; S700, visual recognition of open-pit mine working face: obtaining images of the mining situation of the open-pit mine working face in real time through a visual sensor device mounted on the open-pit continuous mining equipment, and extracting geometric parameters of the open-pit mine working face from the images using the second convolutional neural network model, thereby establishing and continuously updating a three-dimensional model of the open-pit mine working face; S800, Update of the second deep neural network model: actual cutting specific energy consumption E As shown in the following expression: in P is the cutting mechanism power, V t for t The volume of rock formation cut during the time period; The second drilling technical parameters, the rock formation characteristic parameters after fracturing, the cutting mechanism technical parameters and the actual cutting specific energy consumption are calculated. E Input into the second deep neural network model and update it.

2. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: In step S100, the data input layer is the first layer of the initial deep neural network model, which is used to receive input data; the output layer is the last layer of the initial deep neural network model, which is used to generate prediction results; the hidden layer is a layer located between the input layer and the output layer, which is used for data processing and feature extraction.

3. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: In step S200, the image input layer is the first layer of the initial convolutional neural network model, which is used to receive image data; the convolutional layer is located after the image input layer, which is used to extract local features in the image data; the pooling layer is located after the convolutional layer, which is used to reduce the spatial dimension of the feature map and retain important feature information; the fully connected layer is located at the end of the initial convolutional neural network model, which is used for the integration and classification of feature information.

4. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: In step S300, the drilling technical parameters include hole depth, hole diameter, hole spacing, drilling speed, drilling energy consumption and original rock formation characteristic parameters obtained from geological exploration data.

5. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: In step S300, the non-explosion assisted rock breaking technology includes a carbon dioxide phase change fracturing technology; the carbon dioxide phase change fracturing technology uses the physical expansion characteristics generated during the phase change of carbon dioxide to cause fractures in the surrounding rock formations, and its specific operating steps are: sealing liquid carbon dioxide in a special pipe and placing it in a test hole, using a heating device to heat the liquid carbon dioxide to cause it to undergo a vaporization phase change and expand rapidly, and then forming carbon dioxide high-pressure gas, which is discharged from the special pipe and generates a tensile stress field in the rock formations around the test hole, thereby achieving rock formation fracturing.

6. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: In step S400, the characteristic parameters of the fractured rock formation include the spatial distribution and geometric size parameters of cracks, joints and layers.

7. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1, characterized in that: In step S500, the technical parameters of the cutting mechanism include the cutting mechanism linear velocity, the cutting mechanism angular velocity, the cutting mechanism output power and the cutting depth.

8. The method for controlling continuous excavation and cutting of heterogeneous hard rock formations based on neural network according to claim 1 is characterized in that: During the continuous mining operation, the measured value of the power of the cutting mechanism is monitored in real time. When the measured value of the power of the cutting mechanism exceeds its threshold, the open-pit continuous mining equipment is controlled to stop the continuous mining operation, and a second threshold of the cutting specific energy consumption is calculated. The second threshold of the cutting specific energy consumption is [ E 2 ] satisfies the following expression: in,[ P ] is the threshold of cutting mechanism power, V t for t The volume of rock formation cut during the time period; By adjusting the technical parameters of the cutting mechanism and using the second deep neural network model to re-obtain the predicted value of the cutting energy consumption ratio until it is lower than the second threshold of the cutting energy consumption ratio, the adjusted technical parameters of the cutting mechanism are then used as the cutting mechanism technical parameters adopted in the continuous mining operation, and the continuous mining operation is restarted.

Images