A method, device and medium for repairing a shield machine roller cutter

Abstract

The application discloses a kind of shield machine cutter repair method, equipment and medium, belong to mechanical repair technical field, to solve the technical problems that the existing shield machine cutter repair method cannot be repaired according to the damage type of cutter Adaptive selection repair parameter, so that the utilization efficiency of repaired cutter gradient needs to be improved.The method comprises the following steps: obtaining the appearance data of the cutter to be repaired;The appearance data of the cutter to be repaired is input into the machine learning model to determine the damage type of the cutter to be repaired;Based on the corresponding relationship between the damage type and the repair parameter, determine the repair parameter corresponding to the damage type;Compare the appearance data of the cutter to be repaired with the appearance data of the new cutter to determine the repair area;Based on the repair parameter and the repair area, the cutter to be repaired is repaired.The quality and utilization rate of the repaired cutter are improved by the above method.

Description

Technical Field

[0001] This invention relates to the field of mechanical repair technology, and in particular to a method, equipment and medium for repairing the cutterhead of a tunnel boring machine. Background Technology

[0002] Tunnel boring machines (TBMs) are crucial engineering equipment for tunnel excavation and mining, ensuring safety during construction, minimizing environmental impact, and significantly improving construction efficiency. The cutterheads of a TBM are the most easily worn and frequently replaced components. Quantitative and qualitative analysis of the cutterheads not only provides a basis for decision-making regarding the optimization of cutterhead distribution and tunneling parameters but also offers vital data support for cutterhead repair.

[0003] Cutterhead repair can extend its service life and reduce construction costs, which is of great significance for improving the tunneling efficiency of tunnel boring machines and reducing the overall project cost. However, existing cutterhead repair methods cannot adaptively select repair materials and other repair parameters based on the type of damage to the cutterhead; therefore, the efficiency of the repaired cutterhead's reuse needs to be improved. Summary of the Invention

[0004] This invention provides a method, equipment, and medium for repairing tunnel boring machine cutterheads, which solves the technical problem that existing tunnel boring machine cutterhead repair methods cannot adaptively select repair parameters according to the damage type of the cutterheads, thus the efficiency of the repaired cutterheads for repeated use needs to be improved.

[0005] In a first aspect, the present invention provides a method for repairing the cutterhead of a tunnel boring machine, comprising:

[0006] Obtain the appearance data of the hob to be repaired;

[0007] The appearance data of the hob to be repaired is input into a machine learning model to determine the damage type of the hob to be repaired;

[0008] Based on the correspondence between damage type and repair parameters, the repair parameters corresponding to the damage type are determined;

[0009] By comparing the appearance data of the hob to be repaired with the appearance data of the new hob, the area to be repaired is determined;

[0010] Based on the repair parameters and the area to be repaired, the hobbing cutter to be repaired is repaired.

[0011] In one feasible implementation, before acquiring the appearance data of the hob to be repaired, the method further includes:

[0012] The presence of internal damage to the cutter to be repaired was determined using an ultrasonic flaw detector.

[0013] If the cutter to be repaired has internal damage, then it is determined that the cutter will not be repaired.

[0014] In one feasible implementation, the appearance data of the hob to be repaired includes a three-dimensional image of the hob. The appearance data of the hob to be repaired is input into a machine learning model to determine the damage type of the hob to be repaired, including:

[0015] Based on the three-dimensional image and projection method, at least two projected images of the hob to be repaired are obtained;

[0016] The at least two projected images are respectively input into a convolutional neural network to obtain the corresponding image features;

[0017] The image features are input into a classification network to determine the damage type of the hob to be repaired.

[0018] In one feasible implementation, the image features are input into a classification network to determine the damage type of the hobbing cutter to be repaired, including:

[0019] The image features are fused using an attention mechanism layer to obtain fused features;

[0020] The fused features are input into the classification layer to determine the damage type of the hob to be repaired.

[0021] In one feasible implementation, after performing feature fusion on the image features through an attention mechanism layer to obtain fused features, the method further includes:

[0022] The fused features are input into the Softmax layer to determine the probability of each damage type.

[0023] In one feasible implementation, after inputting the appearance data of the hob to be repaired into a machine learning model to determine the damage type of the hob to be repaired, the method further includes:

[0024] When the damage type of the hob to be repaired is determined to be fracture, it is determined that the hob to be repaired will not be repaired.

[0025] In one feasible implementation, the area to be repaired is determined by comparing the appearance data of the hob to be repaired with the appearance data of a new hob, including:

[0026] By comparing and analyzing the 3D scanning data of the hob to be repaired and the 3D scanning data of the new hob, the wear data and the location coordinates of the damage location of the hob to be repaired are determined, and the 3D model of the area to be repaired is obtained through size calculation.

[0027] In one feasible implementation, the hobbing cutter to be repaired is repaired based on the repair parameters and the area to be repaired, including:

[0028] The repair parameters and the area to be repaired are input into the laser cladding equipment to repair the roller cutter.

[0029] Secondly, the present invention provides a tunnel boring machine cutterhead repair device, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to execute a tunnel boring machine cutterhead repair method as described in any of the above embodiments.

[0030] Thirdly, the present invention provides a non-volatile computer storage medium, wherein the storage medium is a non-volatile computer-readable storage medium, the non-volatile computer-readable storage medium stores at least one program, each program including instructions, and the instructions, when executed by a terminal, cause the terminal to perform a tunnel boring machine cutter repair method as described in any of the above embodiments.

[0031] The present invention provides a method, equipment, and medium for repairing the cutterhead of a tunnel boring machine, which has the following beneficial technical effects compared with the prior art:

[0032] (1) The repair parameters for repairing the hob to be repaired in this invention are determined according to the damage type of the hob to be repaired. By combining the repair parameters and the area to be repaired, the hob to be repaired can be repaired in a targeted and appropriate manner, thereby improving the quality and utilization rate of the repaired hob.

[0033] (2) In this invention, the convolutional neural network focuses on local features to extract image features of each projected image of the cutter to be repaired; on this basis, the attention mechanism layer further weights and fuses the above image features, so that the classification network can pay attention to more important features; the combination of convolutional neural network and attention mechanism can accurately evaluate the damage type of the cutter to be repaired.

[0034] (3) The present invention compares the 3D scanning data of the hob to be repaired with the 3D scanning data of the new hob to obtain the three-dimensional coordinates of the wear amount and damage location of the hob repair, and obtains the 3D model of the hob to be repaired; the repair parameters are determined according to the damage type of the hob, and the 3D model of the area to be repaired is combined with the laser cladding equipment to achieve precise repair of the hob to be repaired.

[0035] (4) When the ultrasonic flaw detector determines that there is damage inside the cutter to be repaired, or when the damage type is determined to be fracture, the present invention determines that the cutter to be repaired should not be repaired, thus avoiding unnecessary repair of cutters that have no repair value and improving repair efficiency. Attached Figure Description

[0036] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0037] Figure 1 A flowchart of a method for repairing the cutter head of a tunnel boring machine provided by the present invention;

[0038] Figure 2 A flowchart of another method for repairing the cutter head of a tunnel boring machine provided by the present invention;

[0039] Figure 3 A schematic diagram of a six-view projection image of a hob to be repaired, provided by the present invention;

[0040] Figure 4 This is a schematic diagram of the structure of a hobbing cutter damage classification network provided by the present invention;

[0041] Figure 5 A flowchart illustrating the process of determining a 3D model of the area to be repaired, provided by the present invention;

[0042] Figure 6 A schematic diagram illustrating the correspondence between damage types and repair parameters provided by the present invention;

[0043] Figure 7 This is a structural schematic diagram of a tunnel boring machine cutterhead repair device provided by the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0045] Tunnel boring machines (TBMs) are crucial engineering equipment for tunnel excavation and mining, ensuring safety during construction, minimizing environmental impact, and significantly improving construction efficiency. The cutterheads of a TBM are the most easily worn and frequently replaced components. Quantitative and qualitative analysis of the cutterheads not only provides a basis for decision-making regarding the optimization of cutterhead distribution and tunneling parameters but also offers vital data support for cutterhead repair.

[0046] Cutterhead repair can extend its service life and reduce construction costs, which is of great significance for improving the tunneling efficiency of tunnel boring machines (TBMs) and reducing the overall project cost. Typically, before repairing the cutterheads, damage assessment is required. Currently, this process mainly relies on visual observation and manual measurement. While the human eye can distinguish significant surface damage and defects, it easily misses minor defects and damage, and internal damage is impossible to observe visually. Furthermore, for cutterheads with damage and defects, manual measurement is necessary to assess the damage and provide a basis for subsequent repair. However, manual measurement methods have inherent drawbacks such as low efficiency and poor accuracy, and the measurement data cannot establish a precise mapping relationship with the cutterhead's coordinate system, which may lead to significant deviations in repair accuracy. Finally, in the cutterhead repair stage, electric welding is mostly used, resulting in low precision in repair and an inability to adaptively select repair materials based on the type of damage. Therefore, the reuse efficiency of repaired cutterheads needs improvement.

[0047] To address the aforementioned problems, this invention proposes an automated repair method for tunnel boring machine cutterheads based on data-driven technology and laser cladding. First, the cutterhead to be repaired is scanned using an ultrasonic flaw detector to obtain its internal structural state. If internal damage is found, the cutterhead is not suitable for repair and is discarded. If the internal condition of the cutterhead is good, a 3D scanning model of the cutterhead is created using a 3D scanner, and the state of each working surface of the cutterhead is obtained using a six-view projection method. A cutterhead damage classification network is established based on convolutional neural networks and attention mechanisms to accurately assess the type of cutterhead damage. Furthermore, the 3D scanning data of the cutterhead to be repaired is compared with the 3D scanning data of a new cutterhead to obtain the wear amount and three-dimensional coordinates of the damage location. Repair parameters for 3D laser cladding of the cutterhead are determined based on the damage type and wear data. Finally, laser cladding technology is used to precisely repair the cutterhead. The technical solution proposed in this invention is described in detail below with reference to the accompanying drawings.

[0048] Figure 1 A flowchart illustrating a method for repairing the cutterhead of a tunnel boring machine provided by this invention. Figure 1 As shown, the method includes the following execution steps:

[0049] S101. Obtain the appearance data of the hob to be repaired.

[0050] In this invention, the appearance data of the hob to be repaired can be a three-dimensional image or three-dimensional scan data obtained through 3D scanning.

[0051] As one possible implementation, ultrasonic internal flaw detection can quickly sort repairable hobs. If internal damage is found, the hob is not worth repairing and needs to be replaced with a new one. If there is no internal damage, the hob can continue to be repaired. Before obtaining the external data of the hob to be repaired, the method also includes: using an ultrasonic flaw detector to determine whether there is internal damage to the hob to be repaired; if there is internal damage, it is determined that the hob to be repaired will not be repaired. If it is determined that there is no internal damage to the hob to be repaired, then S101 is executed.

[0052] S102. Input the appearance data of the hob to be repaired into the machine learning model to determine the damage type of the hob to be repaired.

[0053] In this invention, the appearance data of the hob to be repaired can reflect whether there is damage to the appearance of the hob to be repaired, and what kind of damage exists. Through machine learning models, the appearance data of the hob to be repaired can be classified to determine the damage type of the hob to be repaired.

[0054] In some embodiments, the damage type of the hob to be repaired includes at least one of uneven wear, chipping, cracking, and fracture.

[0055] As one possible implementation, the appearance data of the hob to be repaired includes a three-dimensional image. S102 inputs the appearance data of the hob to be repaired into a machine learning model to determine the damage type of the hob to be repaired, including:

[0056] S1021. Based on three-dimensional images and projection methods, obtain at least two projection images of the hob to be repaired.

[0057] This invention can obtain a three-dimensional image of the hob to be repaired by 3D scanning, and obtain projected images on different faces of the three-dimensional image by projection method. For example, based on six-face projection, the projected images of the six faces of the three-dimensional image of the hob to be repaired are obtained by upward projection, downward projection, left projection, right projection, forward projection and backward projection.

[0058] S1022. Input at least two projected images into a convolutional neural network to obtain the corresponding image features.

[0059] This invention can extract image features from projected images using a convolutional neural network. For example, for projected images of six sides (top, bottom, left, right, front, and back), the upward-projected image can be input into the convolutional neural network to obtain image features corresponding to the upward-projected image; the downward-projected image can be input into the convolutional neural network to obtain image features corresponding to the downward-projected image; the left-projected image can be input into the convolutional neural network to obtain image features corresponding to the left-projected image; the right-projected image can be input into the convolutional neural network to obtain image features corresponding to the right-projected image; the forward-projected image can be input into the convolutional neural network to obtain image features corresponding to the forward-projected image; and the backward-projected image can be input into the convolutional neural network to obtain image features corresponding to the backward-projected image.

[0060] S1023. Input the image features into the classification network to determine the damage type of the hob to be repaired.

[0061] As one possible implementation, S1023 inputs image features into a classification network to determine the damage type of the cutter to be repaired, including: performing feature fusion on the image features through an attention mechanism layer to obtain fused features; and inputting the fused features into a classification layer to determine the damage type of the cutter to be repaired.

[0062] In this invention, the convolutional neural network focuses on local features to extract image features from each projected image of the cutter to be repaired. On this basis, the attention mechanism layer further weights the above image features, so that the classification network can focus on more important features. Combining the convolutional neural network and the attention mechanism can accurately assess the damage type of the cutter to be repaired.

[0063] As a possible implementation, after fusing image features through an attention mechanism layer to obtain fused features, the method further includes: inputting the fused features into a Softmax layer to determine the probability of each damage type.

[0064] As one possible approach, when the damage type of the hob to be repaired is determined to be fracture, since a fractured hob has no repair value, it is determined not to repair the hob.

[0065] When an ultrasonic flaw detector determines that there is damage inside the cutter to be repaired, or when the damage type is determined to be fracture, the present invention determines that the cutter to be repaired should not be repaired. This avoids unnecessary repair of cutters that are not worth repairing and improves repair efficiency.

[0066] S103. Based on the correspondence between damage type and repair parameters, determine the repair parameters corresponding to the damage type.

[0067] Different repair parameters are suitable for different damage types. This invention allows for the selection of appropriate repair parameters based on the damage type, thereby improving repair quality and increasing the utilization rate of the repaired hobbing cutter. The repair parameters include at least one of the following: material parameters, laser parameters, powder parameters, gas protection parameters, process parameters, and material performance requirements.

[0068] For example, for laser cladding repair, the material parameters of the repair material include material type, powder particle size, and preset powder temperature; laser parameters include laser power, spot diameter, scanning speed, and overlap rate; powder parameters include powder feeding efficiency and powder carrier gas flow rate; gas protection parameters include protective gas type, gas flow rate, lateral gas flow rate, nozzle distance, and gas purity; process parameters include substrate preheating temperature, interlayer temperature, and interpass cooling time; and material performance requirements include hardness requirements, bonding strength, and maximum cladding thickness.

[0069] S104. Compare the appearance data of the hob to be repaired with the appearance data of the new hob to determine the area to be repaired.

[0070] In this invention, the new hob is free of damage. By comparing the appearance data of the hob to be repaired with the appearance data of the new hob, the repairable area of ​​the hob to be repaired can be determined based on the differences between the two.

[0071] As one possible approach, the 3D scanning data of the hob to be repaired and the 3D scanning data of the new hob are compared and analyzed to determine the wear data and the location coordinates of the damage position of the hob to be repaired, and a 3D model of the area to be repaired is obtained through size calculation.

[0072] S105. Based on the repair parameters and the area to be repaired, perform repair processing on the hobbing tool to be repaired.

[0073] In this invention, the repair parameters are determined based on the damage type of the hob to be repaired. Combined with the area to be repaired, the hob to be repaired can be subjected to targeted and appropriate repair treatment, thereby improving the quality and utilization rate of the repaired hob.

[0074] As one possible approach, the repair parameters and the area to be repaired are input into the laser cladding equipment to perform repair processing on the roller cutter.

[0075] This invention determines repair parameters based on the damage type of the hobbing cutter, and, combined with the area to be repaired, can achieve precise repair of the hobbing cutter using laser cladding equipment.

[0076] This invention proposes a method for repairing tunnel boring machine (TBM) cutterheads. This method is an automated repair approach based on data-driven technology and laser cladding, designed to improve the quality and efficiency of TBM cutterhead repair and reduce construction costs. Figure 2 The flowchart of another method for repairing the cutterhead of a tunnel boring machine provided by the present invention specifically includes the following steps.

[0077] Step 1: First, use an ultrasonic flaw detector to test the cutter to be repaired. If damage is found inside the cutter, the cutter is not worth repairing and needs to be replaced with a new one.

[0078] If there is no internal damage to the hob, it can continue to be repaired. In this way, ultrasonic internal flaw detection can quickly sort repairable hobs, improve repair efficiency, and avoid unnecessary repairs to hobs that are not worth repairing, thus avoiding waste.

[0079] Step 2: Obtain a 3D image of the hobbing cutter using 3D scanning. Establish a hobbing cutter damage classification network (HCDCNet) by combining a convolutional neural network and an attention mechanism. The convolutional neural network focuses on local features to extract image features from each projected image of the hobbing cutter to be repaired. Based on this, the attention mechanism layer further weights these image features, allowing the classification network to focus on more important features. Combining the convolutional neural network and the attention mechanism can accurately assess the damage type of the hobbing cutter to be repaired, specifically including the following steps.

[0080] Step 2.1: After acquiring the three-dimensional information of the hob to be repaired using 3D scanning, the hob is projected onto six planes based on six-view projections: upward projection, downward projection, forward projection, backward projection, left projection, and right projection. This is known as six-view projection. Figure 3 The image shown is a schematic diagram of a six-view projection image of a hobbing cutter to be repaired according to the present invention. The projected image is normalized, and the image size is uniformly adjusted to 280×280×3.

[0081] Step 2.2: Based on convolutional neural networks and attention mechanisms, an HCDCNet rolling cutter damage classification network is established. The entire rolling cutter damage classification network contains 64 layers and has a total learnable parameter count of 652.3K. Figure 4 The diagram shown is a structural schematic of a hobbing cutter damage classification network provided by the present invention.

[0082] Step 2.2.1: The input to HCDCNet is 280×280×3×6, where 280×280 represents the size of the projected image in each projection direction, 3 represents an RGB color image, and 6 represents six projection image inputs in six projection directions. Figure 4 The projection image input 1, projection image input 2... projection image input 6 are the projection images of the six faces of the hob to be repaired.

[0083] Step 2.2.2: Feature extraction network structure is the same for the projection images of the hob in these 6 directions. The feature extraction network for the projection image of the hob in each direction includes an input layer, two convolutional layers, two max pooling layers, a batch normalization layer, a dropout layer, an activation layer and a global pooling layer.

[0084] The input layer of the projected image has a size of 280×280×3.

[0085] The size of the first convolutional layer is 140×140×32, where 32 represents 32 convolutional kernels, the convolution stride is [2,2], and the padding method is same.

[0086] The first max pooling layer has a size of 140×140×32, a pooling window size of 5×5, a convolution stride of [1,1], and a padding method of "same".

[0087] The second convolutional layer has a size of 70×70×64, where 64 represents 64 convolutional kernels, a stride of [2,2], and a padding method of "same".

[0088] The second max pooling layer has a size of 70×70×64, a pool size of 5×5, a convolution stride of [1,1], and a padding method of "same".

[0089] The third convolutional layer has a size of 35×35×128, where 128 represents 128 convolutional kernels, a stride of [2,2], and a padding method of "same".

[0090] The size of the batch normalization layer is 35×35×128, where the mean attenuation (MeanDecay) and attenuation noise variance (VarianceDecay) are both set to 0.1.

[0091] The size of the discard layer is 35×35×128, and the discard probability is set to 0.5.

[0092] After the discard layer, there is an activation layer and a global max pooling layer. The size of the activation layer is 35×35×128, and the size of the global normalization layer is 1×1×128.

[0093] The feature extraction process for the hobbing projection images in the other five directions is consistent with the above scheme.

[0094] Step 2.2.3: After each projection image of the rolling cutter is transformed into a 1×1×128 feature vector by a convolutional neural network, the image features in 6 directions are fused through an attention mechanism layer. The size of the attention mechanism layer is 6×1×128, where 6 represents the image features in 6 directions.

[0095] After the attention mechanism layer, there is a fully connected layer, a softmax layer, and a classification layer.

[0096] The Softmax layer is used to predict the probability of each type of hob damage. The size of the classification layer is 1×1×4, where 4 represents the four types of hob damage, including uneven wear, chipping, cracking, and breakage.

[0097] For hobs with fracture damage, this type of damage is not worth repairing. It is determined that the hob will not be repaired and a new hob needs to be replaced.

[0098] Step 3: Compare and analyze the 3D scan data of the hob to be repaired and the 3D scan data of the new hob to obtain the wear data and the location coordinates of the damage location of the hob to be repaired. Obtain the 3D model of the area to be repaired through dimensional calculations, such as... Figure 5 The diagram shown is a flowchart illustrating a method for determining a 3D model of an area to be repaired, as provided by the present invention.

[0099] Step 4: Determine the repair parameters of the hob to be repaired based on the damage type of the hob.

[0100] Figure 6 A schematic diagram illustrating the correspondence between damage types and repair parameters provided by this invention, as shown below. Figure 6 As shown, the damage types include edge chipping damage, eccentric wear damage, and crack damage. Different damage types correspond to different repair parameters. The repair parameters include the material parameters of the repair material, the laser parameters of the 3D laser cladding equipment, the powder parameters, the gas protection parameters, the process parameters, and the material performance requirements.

[0101] The material parameters for the repair material include material type, powder particle size, and preset powder temperature; laser parameters include laser power, spot diameter, scanning speed, and overlap rate; powder parameters include powder feeding efficiency and powder carrier gas flow rate; gas protection parameters include protective gas type, gas flow rate, lateral gas flow rate, nozzle distance, and gas purity; process parameters include substrate preheating temperature, interlayer temperature, and interpass cooling time; and material performance requirements include hardness requirements, bonding strength, and maximum cladding thickness.

[0102] The 3D laser cladding equipment repairs the area to be repaired on the roller cutter according to the repair parameters, resulting in a repaired roller cutter, such as... Figure 5 As shown.

[0103] Corresponding to the above embodiments, the present invention also provides a tunnel boring machine cutterhead repair device. Figure 7 This is a schematic diagram of a tunnel boring machine cutterhead repair device provided in an embodiment of the present invention. The device may include: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, so that the at least one processor can execute a tunnel boring machine cutterhead repair method as described in the above embodiment.

[0104] In one possible implementation of the present invention, the aforementioned at least one processor is capable of performing the following actions: acquiring the appearance data of the hob to be repaired; inputting the appearance data of the hob to be repaired into a machine learning model to determine the damage type of the hob to be repaired; determining the repair parameters corresponding to the damage type based on the correspondence between the damage type and the repair parameters; comparing the appearance data of the hob to be repaired with the appearance data of a new hob to determine the area to be repaired; and performing repair processing on the hob to be repaired based on the repair parameters and the area to be repaired.

[0105] In a specific implementation, the present invention may also be provided as a non-volatile computer storage medium, wherein the storage medium is a non-volatile computer-readable storage medium storing at least one program, each program including instructions, which, when executed by a terminal, cause the terminal to execute a tunnel boring machine cutter repair method as described in the above embodiments.

[0106] In one possible implementation of the present invention, the aforementioned terminal executes the following steps: acquiring the appearance data of the hob to be repaired; inputting the appearance data of the hob to be repaired into a machine learning model to determine the damage type of the hob to be repaired; determining the repair parameters corresponding to the damage type based on the correspondence between the damage type and the repair parameters; comparing the appearance data of the hob to be repaired with the appearance data of a new hob to determine the area to be repaired; and performing repair processing on the hob to be repaired based on the repair parameters and the area to be repaired.

[0107] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0108] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0109] The above description is merely an 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 principle of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for repairing the cutterhead of a tunnel boring machine, characterized in that, The method includes: Obtain the appearance data of the hob to be repaired; The appearance data of the hob to be repaired is input into a machine learning model to determine the damage type of the hob to be repaired; Based on the correspondence between damage type and repair parameters, the repair parameters corresponding to the damage type are determined; By comparing the appearance data of the hob to be repaired with the appearance data of the new hob, the area to be repaired is determined; Based on the repair parameters and the area to be repaired, the hobbing cutter to be repaired is repaired. The appearance data of the hob to be repaired includes a three-dimensional image of the hob. The appearance data of the hob to be repaired is input into a machine learning model to determine the damage type of the hob to be repaired, including: Based on the three-dimensional image and projection method, at least two projected images of the hob to be repaired are obtained; The at least two projected images are respectively input into a convolutional neural network to obtain the corresponding image features; The image features are input into a classification network to determine the damage type of the hobbing cutter to be repaired; The image features are input into a classification network to determine the damage type of the hobbing cutter to be repaired, including: The image features are fused using an attention mechanism layer to obtain fused features; The fused features are input into the classification layer to determine the damage type of the hobbing cutter to be repaired; By comparing the appearance data of the hob to be repaired with the appearance data of the new hob, the area to be repaired is determined, including: The 3D scanning data of the hob to be repaired and the 3D scanning data of the new hob are compared and analyzed to determine the wear data and the location coordinates of the damage position of the hob to be repaired. The 3D model of the area to be repaired is obtained through size calculation. The method involves using 3D scanning to obtain three-dimensional images of the hobbing cutter, and then establishing a hobbing cutter damage classification network by combining convolutional neural networks and attention mechanisms. Specifically, this includes: After obtaining the three-dimensional information of the hob to be repaired by 3D scanning, the projected images of the hob on the six sides are obtained based on the six-sided projection: upward projection, downward projection, forward projection, backward projection, left projection and right projection. The projected image is then normalized. Based on convolutional neural networks and attention mechanisms, an HCDCNet rolling cutter damage classification network was established. The entire rolling cutter damage classification network contains 64 layers and has a total of 652.3K learnable parameters. The input to HCDCNet is 280×280×3×6, where 280×280 represents the size of the projected image in each projection direction, 3 represents the RGB color image, and 6 represents the input of projected images in 6 projection directions. The feature extraction network structure is the same for the projected images in these 6 directions. The feature extraction network for the projected image of the hob in each direction includes an input layer, two convolutional layers, two max pooling layers, a batch normalization layer, a dropout layer, an activation layer, and a global pooling layer. Each projected image of the rolling cutter is transformed into a 1×1×128 feature vector by a convolutional neural network. Then, the image features in 6 directions are fused through an attention mechanism layer, where the size of the attention mechanism layer is 6×1×128, and 6 represents the image features in 6 directions.

2. The method for repairing the cutterhead of a tunnel boring machine according to claim 1, characterized in that, Before acquiring the appearance data of the hob to be repaired, the method further includes: The presence of internal damage to the cutter to be repaired was determined using an ultrasonic flaw detector. If the cutter to be repaired has internal damage, then it is determined that the cutter will not be repaired.

3. The method for repairing the cutterhead of a tunnel boring machine according to claim 1, characterized in that, After fusing the image features through an attention mechanism layer to obtain fused features, the method further includes: The fused features are input into the Softmax layer to determine the probability of each damage type.

4. A method for repairing the cutterhead of a tunnel boring machine according to claim 1, characterized in that, After inputting the appearance data of the hob to be repaired into a machine learning model to determine the damage type of the hob to be repaired, the method further includes: When the damage type of the hob to be repaired is determined to be fracture, it is determined that the hob to be repaired will not be repaired.

5. A method for repairing the cutterhead of a tunnel boring machine according to claim 1, characterized in that, Based on the repair parameters and the area to be repaired, the hobbing cutter to be repaired is repaired, including: The repair parameters and the area to be repaired are input into the laser cladding equipment to repair the roller cutter.

6. A tunnel boring machine cutterhead repair device, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, enabling the at least one processor to perform a tunnel boring machine cutterhead repair method according to any one of claims 1-5.

7. A non-volatile computer storage medium, characterized in that, The storage medium is a non-volatile computer-readable storage medium that stores at least one program, each program including instructions that, when executed by a terminal, cause the terminal to perform a tunnel boring machine cutterhead repair method according to any one of claims 1-5.

Images