CN-121982428-A - Rotary drilling bit wear state detection method and system based on image recognition

CN121982428ACN 121982428 ACN121982428 ACN 121982428ACN-121982428-A

Abstract

The invention belongs to the technical field of image processing, and particularly relates to a method and a system for detecting the abrasion state of a rotary drilling bit based on image recognition, wherein the method comprises the steps of obtaining scalar brightness field and global potential flow gradient amplitude field data of a bit surface image; calculating a local potential well gentle index and a potential flow divergence coefficient based on scalar brightness and potential flow gradient amplitude values in a local window of a pixel point, performing top hat transformation on a scalar brightness field to obtain a physical protrusion weight of the pixel point, combining the local potential well gentle index and the potential flow divergence coefficient weight to obtain a degradation retention weight, mapping all the pixel points into a graph network node, calculating migration transition probability of the node to any adjacent node to construct a full graph transition probability matrix, performing PageRank iteration until convergence based on the transition probability matrix, and obtaining the retention probability of the pixel point, thereby judging the abrasion state of a drill bit. The invention effectively inhibits the slurry interference and improves the detection accuracy.

Inventors

ZHANG SUCHEN
WANG WEI
WU WENMIAO
CHEN QUAN
GENG BING
YANG ZHIFU

Assignees

陕西天地地质有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260330

Claims (10)

1. The method for detecting the abrasion state of the rotary drilling bit based on image recognition is characterized by comprising the following steps of: Scalar brightness field data of the bit surface image are obtained, and a global potential flow gradient amplitude field is calculated to obtain potential flow gradient amplitude of each pixel point; Calculating the variance of the potential flow gradient amplitude values of all the pixel points in the local window of each pixel point to obtain a potential flow divergence coefficient; performing top hat transformation on the scalar brightness field to obtain the entity protrusion weight of each pixel point; weighting the entity protrusion weight according to the local potential well gentle index and the potential flow divergence coefficient, and calculating degradation retention weight; mapping all the pixel points into nodes of a graph network, calculating the pixel migration transition probability of each node to any adjacent node according to the distance between each node and any adjacent node and the degradation retention weight of the pixel point corresponding to the adjacent node, and constructing a full graph transition probability matrix; And performing PageRank iteration until convergence based on the transition probability matrix, acquiring retention probability of each pixel point, and judging the abrasion state of the drill bit based on a statistical result that the normalized value of the retention probability of all the pixel points is larger than a set threshold value.
2. The image recognition-based method for detecting the wear state of the rotary drill bit according to claim 1, wherein the local potential well flattening index satisfies the expression: ; In the formula, Is the first A local potential well flattening index of each pixel point; 、 Is the first Maximum and average values of all scalar brightness values in a local window of each pixel point; Is a natural exponential function; Is a preset amplification factor.
3. The method for detecting the wear state of the rotary drill bit based on image recognition according to claim 1, wherein the calculating the global potential flow gradient magnitude field comprises: and obtaining the gradient amplitude of each pixel point by square sum and root number of the horizontal gradient and the vertical gradient, and forming a global potential flow gradient amplitude field.
4. The method for detecting the wear state of the rotary drill bit based on image recognition according to claim 1, wherein the potential flow divergence coefficient is equal to a normalized value of a variance of potential flow gradient magnitudes of all pixel points within a local window of each pixel point.
5. The method for detecting the wear state of the rotary drill bit based on image recognition according to claim 1, wherein the step of obtaining the physical protrusion weight of each pixel point comprises the following steps: And carrying out top hat transformation on the scalar brightness field by utilizing a filter check with a preset fixed size to obtain a convex distribution diagram, carrying out Gaussian filtering on the convex distribution diagram and carrying out maximum and minimum normalization processing on the convex distribution diagram to obtain a physical convex weight matrix, wherein the value of a corresponding pixel point in the matrix is the convex weight of the pixel point.
6. The image recognition-based rotary drill bit wear state detection method according to claim 1, wherein the degradation retention weight satisfies an expression: ; In the formula, Is the first Degradation retention weights for individual pixels; to be corresponding to the first in the entity convex weight matrix A convex weight value of each pixel point; Is the first A local potential well flattening index of each pixel point; Is the first Potential flow divergence coefficients for the individual pixel points; is a standard normalization function.
7. The method for detecting the wear state of the rotary drill bit based on image recognition according to claim 1, wherein the acquiring manner of the adjacent node comprises: all nodes in the local window of each node are acquired as adjacent nodes.
8. The method for detecting the wear state of the rotary drill bit based on image recognition according to claim 1, wherein the probability of the pixel migration of each node to any one of the adjacent nodes satisfies the expression: ; In the formula, Is a node Transfer to the first Pixel migration transition probabilities of the adjacent nodes; Is a node And the first thereof Euclidean distance between adjacent nodes; 、 Index value and total number for adjacent nodes; Is a node Is the first of (2) The adjacent nodes correspond to degradation retention weights of the pixel points; The distance attenuation coefficient is preset; the method comprises the steps of presetting an intervention coefficient; As a natural exponential function.
9. The method for detecting the wear state of the rotary drilling bit based on image recognition according to claim 1, wherein the determining the wear state of the drilling bit comprises: calculating the percentage of the total number of the pixel points of which the normalized value of the detention probability is larger than the set threshold value to the total number of the pixel points of the whole image, and judging that the abrasion state is serious if the percentage exceeds the process safety proportion.
10. An image recognition-based rotary drill bit wear state detection system, comprising a processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the image recognition-based rotary drill bit wear state detection method according to any one of claims 1-9.

Description

Rotary drilling bit wear state detection method and system based on image recognition Technical Field The invention relates to the technical field of image processing. More particularly, the invention relates to a method and a system for detecting the abrasion state of a rotary drilling bit based on image recognition. Background Along with the large-scale development of underground space, the ultra-long hollow pile filling pile is widely applied in deep foundation pit engineering, and in the pore-forming drilling stage, the dynamic cutting reaction force generated by the uneven stratum is extremely easy to cause serious physical grinding or breaking of alloy drilling teeth of the rotary drilling bit, and once the drilling teeth fail, the drilling efficiency is reduced, the stress unbalance loading and deflection of the drilling bit are caused, and extremely serious pile position deviation quality accidents are caused. Currently, visual detection of drill bit abrasion is mostly dependent on PageRank algorithm, the core logic is to construct pixel points in an image as network nodes, and the probability of migration transition of virtual particles among pixels is statically allocated only by means of two-dimensional geometric distance and surface brightness similarity among the nodes, so as to try to gather the particles and search for a significant abrasion region with strong contrast. However, in the ultra-long hollow pile rotary drilling hole forming scene, thick and uneven slurry and sediment are inevitably adhered to the surface of a drill bit when the drill bit is lifted out of a hole, a large number of false highlight plaques are generated due to the random fluid characteristics of the slurry and the reflection of a surface water film, the true mechanical abrasion form of drilling teeth is covered, the bottom layer migration transfer network of the traditional PageRank algorithm is seriously disturbed, virtual migration particles are extremely easy to be trapped by a large-area highly-reflective continuous slurry area, the actually flattened abrasion area is extremely low in transfer probability due to smooth wrapping of the slurry, and the abrasion state is completely missed by the algorithm, so that the detection accuracy of the abrasion state is extremely low, and reliable early warning cannot be provided for engineering construction. Disclosure of Invention In order to solve the technical problem that the tradition PageRank algorithm wander transfer network fails and the bit abrasion state detection is inaccurate due to the existence of thick mud, the invention provides the following schemes. The invention provides a method for detecting the abrasion state of a rotary drilling bit based on image identification, which comprises the steps of obtaining scalar brightness field data of a bit surface image, calculating a global potential flow gradient amplitude field to obtain potential flow gradient amplitude of each pixel point, calculating a local potential well mild index according to the difference between the maximum value and the average value of scalar brightness in a local window of each pixel point, calculating variances of potential flow gradient amplitude of all pixel points in the local window of each pixel point to obtain potential flow divergence coefficients, performing top hat transformation on the scalar brightness field to obtain entity protrusion weights of each pixel point, weighting the entity protrusion weights according to the local potential well mild index and the potential flow divergence coefficients, calculating degradation weights, mapping all pixel points to nodes of a graph network, calculating pixel migration transfer probability of each node to any adjacent node according to the distance between each node and any adjacent node and degradation retention weight of the adjacent node corresponding to the pixel point, constructing a full graph transfer probability matrix, performing Pank iteration on the basis of the transfer probability matrix to obtain retention probability values of each pixel point, and setting the retention probability value of each pixel point to be normalized on the basis of the retention probability value of the bit retention state of the pixel point. According to the invention, the physical grinding and breaking characteristics of the drilling teeth are evaluated by calculating the local potential well gentle index and the potential flow divergence coefficient, the physical protrusion weight is extracted by combining the morphological top cap transformation, the degradation retention weight is obtained by calculation, the bottom Markov transition probability matrix is reconstructed by utilizing the degradation retention weight, the static transition rule is broken, the migration particles are forcedly guided to converge towards the real defect, the migration bottom noise of complex slurry is effectively restrained, and the detection accuracy under th