A multi-dimensional parameter analysis diamond bit quality detection method

Abstract

The application discloses a kind of multi-dimensional parameter analysis diamond drill bit quality detection method, it is related to quality detection technical field, the method includes: by function loss correlation analysis construction function detection tree, wherein root node is appearance detection.According to root node, directional data acquisition is carried out, obtains single-dimensional detection data, and carries out quality detection analysis.If function detection passes, according to detection tree, continue to carry out the data acquisition and analysis of secondary child node, output secondary detection result.According to secondary detection result, judge whether to continue to execute tertiary child node detection.The detection result of each level determines the next execution, so that the quality defect information set of diamond drill bit is gradually output, redundant detection is avoided, and the detection efficiency is improved.The technical problem that the existing diamond drill bit quality detection process is complicated, detection resource is wasted and not accurate is solved, and the technical effect of improving detection efficiency and accuracy is achieved.

Description

TECHNICAL FIELD

[0001] The application relates to the quality detection technical field, in particular to a diamond drill bit quality detection method based on multi-dimensional parameter analysis. BACKGROUND

[0002] The diamond drill bit is widely used in drilling and cutting fields due to its high strength, wear resistance and impact resistance, and the quality of the diamond drill bit directly affects the drilling efficiency and safety. The traditional quality detection method usually relies on a single detection item, such as appearance inspection, hardness test and cutting performance test, and these methods are cumbersome and inefficient when facing complex drill bit structures and multi-functional detection. The existing technology has the problems of time and resource waste in the detection process, especially in the functional loss detection, and the detection process cannot be accurate and efficient. How to improve the detection accuracy and efficiency while reducing unnecessary detection and resource waste has become a key challenge in the production process of the diamond drill bit.

[0003] At present, there are technical problems of cumbersome quality detection process, waste of detection resources and inaccuracy in the related art of the diamond drill bit. SUMMARY

[0004] The application provides a diamond drill bit quality detection method based on multi-dimensional parameter analysis, which solves the technical problems of cumbersome quality detection process, waste of detection resources and inaccuracy of the existing diamond drill bit quality detection process.

[0005] The application provides a diamond drill bit quality detection method based on multi-dimensional parameter analysis, which includes:

[0006] The functional loss correlation analysis of the diamond drill bit is performed, and a functional detection tree is constructed based on the analysis result, wherein the root node of the functional detection tree is the appearance detection; the root node is taken as the starting point of the quality detection, the detection data of the diamond drill bit is collected according to the functional detection item of the root node, and the single-dimensional detection data is obtained; the quality detection analysis of the diamond drill bit is performed according to the single-dimensional detection data, and the functional detection result is output; if the functional detection result is passed, the detection data collection of N secondary nodes is performed under the guidance of the functional detection tree, and the quality detection analysis of the diamond drill bit is performed based on the obtained N-dimensional detection data, and N secondary detection results are output; the out-degree connection relationship between the N secondary nodes and M tertiary nodes in the functional detection tree is taken as the judgment basis for whether to continue the quality detection, and the functional detection of the M tertiary nodes is performed according to the pass identification of the N secondary detection results, wherein N and M are positive integers, and N

[0007] The application provides a multi-dimensional parameter analysis diamond drill bit quality detection method. A function detection tree is constructed through function loss correlation analysis, wherein a root node is appearance detection. Single-dimensional detection data is obtained through directional data collection according to the root node, and quality detection analysis is performed. If the function detection is passed, data collection and analysis of a secondary node are continued according to the detection tree, and a secondary detection result is output. Whether to continue to perform a tertiary node detection is determined according to the secondary detection result. The detection result of each level determines the next execution, so that a quality defect information set of the diamond drill bit is gradually output, redundant detection is avoided, detection efficiency is improved, the detection process is optimized through function loss correlation analysis, unnecessary detection items are reduced, and technical effects of improving detection efficiency and precision are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings of the embodiments of the application will be briefly introduced below. In the present application, a flowchart is used to illustrate the operations performed by the system according to the embodiments of the application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously according to needs. Meanwhile, other operations can be added to these processes, or a step or several steps can be removed from these processes.

[0009] Figure 1 A flowchart of a multi-dimensional parameter analysis diamond drill bit quality detection method provided by the embodiments of the application.

[0010] Figure 2 A detection model activation tree construction flowchart of a multi-dimensional parameter analysis diamond drill bit quality detection method provided by the embodiments of the application. DETAILED DESCRIPTION

[0011] The above description is only a summary of the technical solutions of the application. In order to more clearly understand the technical means of the application, the application can be implemented according to the content of the specification, and in order to make the above and other purposes, characteristics and advantages of the application more obvious and easy to understand, the following specific embodiments of the application are described.

[0012] In order to make the purposes, technical solutions and advantages of the application more clear, the application will be further described in detail below with reference to the drawings. The described embodiments should not be regarded as limiting the application. All other embodiments obtained by those skilled in the art without making creative efforts fall within the scope of protection of the application.

[0013] In the following description, "some embodiments" are referred, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subset of all possible embodiments, and can be combined with each other without conflict, and the term "first\second" referred to only distinguishes similar objects, and does not represent a specific order for the objects. The terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0014] The embodiments of the present application provide a multi-dimensional parameter analysis diamond bit quality detection method, as shown in Figure 1 The method comprises the following steps:

[0015] In step S100, the function loss correlation of the diamond bit is analyzed, and a function detection tree is constructed based on the analysis result, wherein the root node of the function detection tree is appearance detection. Specifically, when constructing the function detection tree of the diamond bit, the appearance detection is first determined as the root node because it is intuitive and simple, and can preliminarily evaluate the quality of the bit. Then the function loss correlation is analyzed, for example, hardness decline will affect wear resistance, and inaccurate cutting edge angle will lead to reduced cutting efficiency. Then according to the correlation, the secondary nodes are determined, for example, the appearance defects are related to the internal structure integrity and the coating integrity is related to the corrosion resistance, so the internal structure integrity detection and the corrosion resistance detection are set as the secondary nodes. Then the subsequent levels are constructed, for example, the unqualified internal structure integrity detection will affect the stress distribution uniformity, and the stress distribution uniformity detection can be used as the tertiary node of the internal structure integrity detection; poor corrosion resistance will affect the service life, and the service life detection can be used as the tertiary node of the corrosion resistance detection. Finally, by setting the logical relationship between the nodes "if not passed, then do not perform subsequent detection", the detection efficiency is ensured, and unnecessary detection is avoided.

[0016] In a possible implementation, the diamond drill bit is subjected to function loss correlation analysis, and a function detection tree is constructed based on the analysis result, wherein a root node of the function detection tree is appearance detection, and step S100 further includes step S110 of obtaining the diamond drill bit from a drill bit production line by system sampling. Specifically, a rule and interval of system sampling are determined. For example, a representative diamond drill bit sample is selected from the drill bit production line in a manner of every certain number (such as 1 sample for every 50 drill bits produced). Randomness and uniformity of sampling can be ensured, so that the sample can reflect product quality characteristics of the entire production line. During sampling, it is ensured that the sampling environment and operation conform to the specification, so as to avoid introducing additional errors or biases due to improper sampling, thereby providing reliable basic data for subsequent detection and analysis.

[0017] Step S120: an interaction is performed to obtain a batch production use scenario of the drill bit production line, and a use scenario setting is obtained by collecting limit use data of the batch production use scenario. Specifically, communication and exchange are performed with multiple parties such as a production department, a sales department, and an actual use customer, and use information of the drill bit under different working conditions is collected, including but not limited to geological conditions (such as soft rock, hard rock, and mixed stratum), drilling depth, drilling fluid type, rotation speed and torque requirement, and the like. Then, the batch production use scenario is sorted and analyzed to find out the most extreme use scenario. For example, if the drill bit of this model is used for drilling in soft soil stratum and mining in hard rock stratum, the use scenario of the hard rock stratum with high rotation speed and high torque is determined as the limit use scenario. For the limit use scenario, a data collection device and a sensor are used to monitor and record various performance parameters (such as stress condition, wear degree, temperature change, and the like) of the drill bit under this scenario in real time, so as to obtain detailed limit use data. Through sorting and induction of the limit use data, a use scenario setting capable of covering the most severe working condition is determined, so as to ensure strictness and comprehensiveness of subsequent detection, so that the detection result can reflect performance of the drill bit under various actual use conditions, and provide a strong basis for product quality evaluation.

[0018] Step S130, according to the use scene setting, the detection item set of the diamond drill bit is obtained by calling the network data, wherein the detection item set includes H function detection items, wherein H is a positive integer greater than M. Specifically, according to the determined use scene setting, the relevant database or network platform is connected, and the platform contains the standard detection items in the industry, the detection experience data of similar products in the past, and the information resources such as the technical reports published by professional research institutions. By inputting the key parameters of the use scene (such as formation hardness, drilling depth, etc.) as search conditions, the detection item information closely related to the current diamond drill bit in the use scene is screened out from the massive data. For example, in the extreme hard rock stratum use scene, hardness detection, wear resistance detection, impact resistance detection and other items will be screened out; for the scene with high speed requirement, cutting performance detection and dynamic balance detection and other items will also be included. After integration and screening, the detection item set containing H function detection items is finally obtained, and the detection items will comprehensively cover various performance challenges that the drill bit may face in the set use scene, provide detailed detection indexes and directions for subsequent quality detection, ensure that all functions of the product can be fully evaluated and verified to meet the actual use requirements, and improve the reliability and applicability of the product.

[0019] Step S140, the historical function defect record is called as a constraint with the detection item set. Specifically, the detection item set obtained in the foregoing is taken as a key constraint condition, and the enterprise internal quality database, customer feedback record, and industry fault case library and other information sources are accessed again. In the data source, according to the category and specific requirements of the detection items, the historical function defect record related to the H function detection items of the current diamond drill bit is searched and extracted. For example, if the detection item set contains hardness detection, the records of the problems, defect cause analysis and solving measures of the past products in the hardness aspect are searched from the historical data; for cutting performance detection, the case information of poor cutting performance caused by problems such as cutting edge design, material selection or processing technology in the past is collected. Through targeted query and screening, detailed historical function defect records are obtained, which will provide valuable experience data and reference basis for subsequent function loss correlation analysis, help to identify potential risks and weak links that may exist in different functions of the product, so that effective prevention and improvement measures can be taken in the quality detection and improvement process, and the overall quality level of the product is improved.

[0020] Step S150, according to the history function defect record on the set of detection items for loss of function correlation analysis, and based on the analysis results to build the function detection tree. Specifically, the collected history function defect record is analyzed statistically and data mining. For example, by analyzing a large number of hardness defect records, it is found that when the hardness is insufficient, it is often accompanied by a significant decrease in wear resistance and cutting performance, indicating that there is a strong loss of function correlation between the hardness detection item and the wear resistance detection item and the cutting performance detection item. The history defect records of other function detection items are analyzed similarly to find the mutual influence and correlation mode between the detection items, and a function loss probability matrix or correlation diagram is constructed. On this basis, the function loss correlation is determined as the basis for determining the level and order of the detection items, and the function detection tree is constructed. The detection items in the function loss chain, such as appearance detection, because appearance defects often indicate potential problems in internal structure or other functions, are used as root nodes or upper nodes, and detection items with strong correlation with the root nodes are used as the next level of child nodes (such as internal structure integrity detection, coating quality detection, etc. associated with appearance detection), and so on. Gradually, a complete function detection tree reflecting the logical relationship and hierarchical structure of each function detection item is constructed. Through the function detection tree, the subsequent quality detection process can be clearly guided, and the detection results of the previous node are used to determine whether to perform the detection of the next node, thereby improving the detection efficiency and accuracy, and also helping to more comprehensively and deeply understand the quality status of the product, providing strong support for the quality improvement and optimization of the product.

[0021] In one possible implementation, according to the history function defect record on the set of detection items for loss of function correlation analysis, and based on the analysis results to build the function detection tree, step S150 further comprises step S151, by based on the history function defect record on the H kind of function detection item for loss of function synchronism analysis, construct to generate function loss probability matrix. Specifically, first, based on the collected history function defect record is analyzed. Suppose we have H kind of function detection item, respectively marked as A, B, C, …, E. For each pair of function detection items, for example, A and B, analyze the frequency of function B when function A is lost. By statistical analysis of a large number of historical function defect records, the probability that function B is lost when function A is lost is calculated as . Similarly, the loss probability of all pairs of functions is calculated, thereby constructing a function loss probability matrix . For example, the matrix represents the probability of loss of function A itself (in practice, it may be caused by structural damage, etc.), and represents the probability that function A is lost when function B is lost, and so on.

[0022] Step S152: A transition probability association threshold is preset. Appearance detection is used as the root node, and the function loss probability matrix is ​​traversed using appearance detection and the aforementioned transition probability association threshold to obtain N function detection items. These N function detection items are then used as the N second-level child nodes, where the root node is connected to the N second-level child nodes by their out-degree connections. Specifically, a transition probability association threshold is set. (This threshold is determined based on experience or actual needs). Assuming appearance inspection corresponds to function A (this is just an assumption for ease of explanation), then we can examine the probability matrix... The first line (corresponding to function A). Iterate through each probability value in this line, for example... If a certain probability value is greater than or equal to ,for example and Then, the detection items corresponding to functions B and C are selected as secondary child nodes, with a quantity of N (here N=2, that is, the detection items corresponding to functions B and C). The appearance inspection (root node) establishes an out-degree connection relationship with these N secondary child nodes (the detection items corresponding to functions B and C), indicating that the appearance inspection can transition to these N secondary child nodes for inspection.

[0023] Step S153: Using the N secondary child nodes and the transition probability association threshold, traverse the function loss probability matrix to obtain N sets of function detection items. Specifically, for the N secondary child nodes obtained in the previous step, assume they are detection items corresponding to functions B and C. For the detection items corresponding to function B, examine the probability matrix. The row corresponding to B (the second row) is associated with the threshold based on the transition probability. Perform a filter. For example, if and Then the detection items corresponding to functions D and E form a group. For the detection items corresponding to function C, look at the row corresponding to C (the third row), assuming... and Then the detection items corresponding to functions F and G form another group. In this way, N groups of function detection items are obtained.

[0024] Step S154, construct N groups of third-level child nodes of the N groups of function detection items, and connect the N groups of second-level child nodes with the N groups of third-level child nodes. The N groups of third-level child nodes include the M third-level child nodes. Specifically, for the N groups of function detection items obtained in the previous step, each group is constructed as a third-level child node. For example, the function detection items corresponding to function D and function E screened from function B form a group of third-level child nodes, and the function detection items corresponding to function F and function G screened from function C form another group of third-level child nodes. Then, the second-level child nodes (the function detection items corresponding to function B and function C) are connected with the third-level child nodes corresponding to them (for example, the function detection items corresponding to function D and function E for function B, and the function detection items corresponding to function F and function G for function C). The total number of all third-level child nodes is M (in the above example, M = 4, i.e., the function detection items corresponding to function D, function E, function F, and function G).

[0025] Step S155, and so on, until the nodes in the function detection tree cover all the H function detection items, and the construction of the function detection tree is completed. Specifically, according to the above method, the third-level child nodes are further screened. For example, for the function detection item corresponding to function D (assuming it is one of the third-level child nodes), the corresponding row in the probability matrix is checked again, and the next-level child nodes are screened according to the transition probability association threshold. This process is repeated until the nodes in the function detection tree can cover all the H function detection items. In this way, the construction of the entire function detection tree is completed, and each node in the tree represents a function detection item, and the connection relationship between the nodes reflects the detection order and logic based on the function loss probability.

[0026] In a possible implementation, the function loss synchronization analysis of the H function detection items based on the historical function defect records is performed to construct the function loss probability matrix. Step S151 further includes step S1511 of obtaining a first function association information set of a first function detection item in the detection item set from the historical function defect records based on the function loss synchronization. Specifically, there are historical function defect records, and the records contain result data of detection of the H function detection items. Assuming that the first function detection item is A, all records when function A is lost are screened from the historical records. For example, when function A fails in the past detection, the conditions of other function detection items are also recorded. The combination of the conditions of other function detection items when function A is lost constitutes the first function association information set of the first function detection item A. For example, when A is lost, there are multiple records showing that functions B and C cannot also detect normal results, and the combination of (B, C) is part of the first function association information set.

[0027] Step S1512, the first set of functional correlation information H-1 kind of function detection item frequency statistics of repeated, get the first group of loss transfer probability, wherein the first group of loss transfer probability includes H-1 function loss transfer probability. Specifically, for the first functional correlation information set obtained above, the appearance frequency of other H-1 kind of function detection item is counted. Assuming that in addition to the first function detection item A, there are functions B, C, …, E (H-1 in total). Check each record in the first functional correlation information set, count the number of times function B appears in these records , and the number of times function C appears is counted in the same way . Then calculate the first group of loss transfer probability. For example, the loss transfer probability of function A to function B , where The sum of the number of times other functions appear in the first functional correlation information set except A. Get H-1 function loss transfer probability, that is, the first group of loss transfer probability.

[0028] Step S1513, in the same way, based on the historical function defect record, the function loss synchronism analysis is carried out on the H kind of function detection item, and H group of function loss transfer probability is obtained. Specifically, according to the above steps, we analyze each function detection item in turn. When analyzing function B, find out the combination of other function detection items when function B is lost from the historical function defect record, and get the function correlation information set of function B. Then the similar frequency statistics is carried out on this information set, and the loss transfer probability of function B to other functions is calculated, and the second group of loss transfer probability is obtained. Repeat this process until all H kind of function detection item is analyzed in this way, so as to obtain H group of function loss transfer probability.

[0029] Step S1514, according to the H kind of function detection item and H group of function loss transfer probability, the function loss probability matrix is constructed. Specifically, we have H kind of function detection item, which is marked as A, B, C, …, E respectively. H group of function loss transfer probability is obtained from the previous step, each group corresponds to the transfer probability of one function detection item to other function detection item. The function loss probability matrix is constructed , the element in the matrix represents the loss transfer probability from function to function . For example, is one of the first group of loss transfer probability, located in the first row and the second column of the matrix, which represents the loss transfer probability of function A to function B. Similarly, is located in the second row and the first column, which represents the loss transfer probability of function B to function A. In this way, according to H kind of function detection item and H group of function loss transfer probability, the function loss probability matrix is constructed .

[0030] Step S200, taking the root node as the quality detection starting point, collecting detection data of the diamond drill bit according to the functional detection items of the root node, and obtaining single-dimensional detection data. Specifically, when taking the root node as the quality detection starting point, the functional detection items of the root node (assuming appearance detection) are first determined, including surface integrity, shape and size accuracy, and coating quality (if there is a coating) and the like. Detection data is collected in a targeted manner according to the items. For example, for surface integrity, surface cracks and scratches can be observed and recorded by optical or electronic microscopes, or visual grading can be recorded; for shape and size accuracy, high-precision measuring tools are used to measure key dimensions such as drill bit diameter multiple times, and an angle measuring instrument is used to measure the cutting edge angle; for coating quality (if any), the thickness and uniformity of the coating are measured by an optical microscope, and the adhesion of the coating is detected by a grid test. Through the above operations, single-dimensional detection data related only to the dimension of appearance detection is obtained, which will serve as the basis for subsequent quality judgment.

[0031] Step S300, quality detection analysis is performed on the diamond drill bit according to the single-dimensional detection data, and a functional detection result is output. Specifically, the quality standards corresponding to the single-dimensional detection data are determined, such as the surface integrity aspect, the size and number of defects such as crack length, depth, scratches, pits and protrusions are strictly limited; for shape and size accuracy, the cutting edge angle, drill bit diameter and length have specific design requirements and tolerance ranges; the thickness and uniformity of the coating quality (if any) also have corresponding indicators. Then the detection data is compared and analyzed with these standards one by one. For example, if the surface crack length exceeds the standard value, the surface integrity does not meet the requirements; if the cutting edge angle deviation is at the edge of the allowable range, it is in a critical state; if the coating thickness is not in the standard range or the adhesion does not meet the standard, the coating quality is unqualified, etc. Finally, the analysis results of each aspect are comprehensively determined to determine whether the appearance functional detection of the diamond drill bit passes, and the results are output in the form of a report, which details the data, standards and determination of each detection item, and also gives the overall functional detection result, such as "pass" or "not pass" and the corresponding unqualified item record, so that subsequent measures can be taken.

[0032] In a possible implementation manner, as Figure 2As shown, according to the single-dimensional detection data, the quality of the diamond drill bit is detected and analyzed, and a functional detection result is output, and step S300 further includes step S310 of interactively obtaining H reference data requirements of the H functional detection items. Specifically, a team composed of quality engineers, production technicians and industry experts is organized, and in-depth discussion and analysis are carried out for each functional detection item. Taking hardness detection as an example, considering the working characteristics and material properties of the diamond drill bit, data such as the overall hardness, the hardness difference of different parts (such as cutting edges, drill bit bodies, etc.), and the hardness change with temperature are determined, and these specific data indicators constitute the reference data requirements of the hardness detection item. For cutting performance detection, the reference data range and precision requirements of cutting force, cutting speed, chip shape, etc. are determined; for wear resistance detection, the collection standards and methods of relevant data such as wear amount, wear rate, wear uniformity are determined, so as to obtain detailed reference data requirements corresponding to each of the H functional detection items, and provide accurate guidance for subsequent data acquisition and model establishment.

[0033] Step S320, according to the H reference data requirements and the model information of the diamond drill bit, network data calling is performed, and H quality abnormal information sets are obtained, wherein each quality abnormal information set includes a plurality of sample reference data and a plurality of sample abnormal detection results. Specifically, using data acquisition software and network interface, connect to enterprise internal quality database, industry standard database and research achievement database of related research institutions, etc. When calling data, first input the specific model information of the diamond drill bit, so as to filter out the data highly related to the model. Combined with the H reference data requirements determined before, using search algorithm and data filtering tool, extract the required information from the massive data. For example, for the appearance detection item, according to the detection requirements of surface cracks, pores, defects and wear in the reference data requirements, find and collect a plurality of appearance image data of the drill bit of this model in different production batches and different use stages from the database as sample reference data, and the corresponding crack size, defect area ratio and other sample abnormal detection results, which together constitute the quality abnormal information set of the appearance detection item. Repeat this process for each functional detection item, and finally obtain H complete quality abnormal information sets, which provide rich and targeted actual data support for subsequent model training.

[0034] Step S330, H standard anomaly detection models are pre-configured according to the H detection feature differences of the H functional detection items. Specifically, based on the analysis of the H functional detection item data, the unique detection features and data patterns of each functional detection item are analyzed. Taking hardness detection as an example, the detection features are the distribution range of hardness value, the relationship between hardness value and material structure, etc. According to the above features, a suitable machine learning algorithm is selected, such as an anomaly detection algorithm based on statistical analysis, to build a standard anomaly detection model for hardness detection. For cutting performance detection, since its detection features involve complex factors such as dynamic changes of cutting force and morphological characteristics of chips, a convolutional neural network in deep learning algorithm is used to build the model, so as to better capture the complex patterns and abnormal situations in the cutting process. For wear resistance detection, the time series characteristics of the wear data are analyzed, and a recurrent neural network or time series analysis method is selected to build the corresponding standard anomaly detection model. Through the differentiated model selection and pre-configuration for different detection features, H initial standard anomaly detection models are tailor-made for each functional detection item, laying a foundation for subsequent model optimization.

[0035] Step S340, the H quality anomaly information sets are used as training data for the directional parameter optimization of the H standard anomaly detection models, and H quality anomaly detection models are obtained. Specifically, for each pre-configured standard anomaly detection model, the corresponding quality anomaly information set is subjected to data preprocessing operations, including data cleaning (removing noise data, handling outliers, etc.), data normalization (converting different scale data to a unified numerical range), and data feature engineering (extracting key features, dimension reduction, etc.), to improve the quality and usability of the data. The processed quality anomaly information set is divided into a training set and a validation set according to a certain proportion, and the training set is used to train the standard anomaly detection model. During the training process, by adjusting the internal parameters of the model (such as the weights and biases of the neural network, the parameter estimates of the statistical model, etc.), the difference between the model prediction results and the actual sample anomaly detection results is minimized as the target, and a suitable optimization algorithm (such as gradient descent method, stochastic gradient descent method, etc.) is used for iterative optimization. The performance of the model is evaluated and verified using the validation set to prevent overfitting. After several training and verification cycles, the model parameters are adjusted continuously until the performance of the model on the validation set reaches the expected indicators, thereby obtaining H optimized quality anomaly detection models for the H functional detection items. The model can more accurately identify and predict the quality anomaly of the diamond drill bit in different functional detection items.

[0036] Step S350, according to the H kind of function detection item in the functional detection tree hierarchical relationship, construct the activation hierarchy of the H quality anomaly detection model, complete the establishment of detection model activation tree. Specifically, analyze the H kind of function detection item in the functional detection tree in the hierarchical position and logical relationship of the already constructed. For example, in the functional detection tree, appearance detection is usually in the root node or upper layer position, because appearance defect often can intuitively reflect the possible existence of other potential quality problems of the drill bit. If the appearance detection passes, then the subsequent associated function detection items such as internal structure integrity detection, coating quality detection and the like are necessary to further detect. Based on the hierarchical relationship, the quality anomaly detection model corresponding to the appearance detection is set as the model activated first at the beginning of the detection process. If the appearance detection model is determined to pass, then according to the logic of the functional detection tree, the quality anomaly detection models corresponding to the internal structure integrity detection and the coating quality detection are activated, and so on, to construct the activation hierarchy of the H quality anomaly detection models. In this structure, the activation order and condition of the model are determined according to the order and logical dependence relationship of the function detection items in the functional detection tree, thereby completing the establishment of the detection model activation tree, so that when the quality of the diamond drill bit is detected, the corresponding quality anomaly detection model can be called in a reasonable order, the detection efficiency and accuracy are improved, unnecessary detection resource waste is avoided, and the quality problem of the drill bit can be more comprehensively and systematically found and diagnosed, thereby providing strong technical support for the improvement and improvement of product quality.

[0037] In a possible implementation, according to the single-dimensional detection data, the quality of the diamond drill bit is analyzed, and a functional detection result is output. Step S300 further includes step S360 of calling a first reference data requirement from the H reference data requirements according to the functional detection item of the root node. Specifically, the functional detection item of the root node is determined, and it is assumed that the root node is appearance detection, which may include surface crack inspection, shape and size measurement, coating integrity evaluation, etc. Then, the first reference data requirement related to the appearance detection item is selected from the pre-determined H reference data requirements. For example, if the appearance detection focuses on surface cracks, the first reference data requirement may involve specific data standards and collection methods for the length, width, number and distribution position of the cracks; for shape and size measurement, the tolerance range and measurement accuracy requirement of the drill bit diameter, length and cutting edge angle, etc. are included; for coating integrity, the allowable range of coating thickness and the roughness index of the coating surface are specified. Through matching, the first reference data requirement dedicated to appearance detection is extracted from the numerous reference data requirements, which provides clear guidance and specification for subsequent data collection.

[0038] Step S370, according to the first reference data requirements, the diamond drill bit detection data directional collection, obtain the single dimension detection data. Specifically, according to the above-mentioned first reference data requirements obtained, organize professional detection personnel and equipped with appropriate detection equipment to carry out data collection work. For surface crack inspection, if the first reference data requirements stipulate that the optical microscope is used for detection, the detection personnel will carry out comprehensive and detailed scanning on the drill bit surface under the specified illumination conditions from the specified observation angle, and record the relevant data of all the cracks found, such as length accurate to millimeter, width accurate to micron, and the specific position coordinates of the cracks, etc. In the shape and size measurement, according to the requirements, high-precision caliper, micrometer or optical measuring instrument is used to measure the diameter of the drill bit at multiple positions, to ensure the accuracy and repeatability of the measurement results, the length and cutting edge angle are also measured accurately according to the standard measurement method and recorded. For coating integrity evaluation, coating thickness gauge is used to measure the coating thickness, roughness meter is used to detect the roughness of the coating surface, and data collection is carried out according to the pre-set sampling points and measurement path to ensure that the key areas of the entire coating surface are covered. Through targeted detection operation, single dimension detection data related to appearance detection only is collected, which will be used as the basis for subsequent quality evaluation.

[0039] Step S380, according to the root node in the detection model activation tree mapping activates the first quality abnormality detection model. Specifically, the root node is known as appearance detection, and the corresponding node in the pre-constructed detection model activation tree is found. The detection model activation tree is constructed based on the logical relationship and importance of functional detection items, and appearance detection as the starting node has potential association and triggering conditions with other functional detection items (such as hardness detection, cutting performance detection, etc.). When it is determined that the appearance detection is performed, according to the mapping mechanism of the activation tree, the first quality abnormality detection model specially designed for appearance detection is found and activated. The model is trained and optimized based on a large number of historical appearance detection data and quality abnormality cases, and can quickly and accurately analyze the collected appearance detection data to judge whether there is an abnormal situation, for example, through the comprehensive analysis of surface crack data, shape and size deviation data and coating integrity data, whether the appearance of the drill bit meets the quality standard and whether there is potential quality risk.

[0040] Step S390, load the single-dimensional detection data to the first quality anomaly detection model for quality detection analysis, output the function detection result, wherein when the function detection result is not passed, the function detection result further includes a function anomaly detection result. Specifically, the single-dimensional detection data collected in the previous step is loaded according to the format and input method required by the first quality anomaly detection model. The model will analyze these data according to the built-in algorithm and the parameters obtained by training. For example, for surface crack data, the model will compare the crack feature distribution of normal drill bits in historical data to determine whether the number, length and distribution of the cracks detected currently are beyond the normal range; for shape and size data, it will check whether the measured values are within the predetermined tolerance range, if they are beyond the tolerance range, it is determined to be abnormal; for coating integrity data, it analyzes whether the deviation of coating thickness and roughness index meet the quality requirements. After comprehensive analysis by the model, if all data meet the quality standards, the function detection result output is "passed", indicating that the diamond drill bit has reached the expected quality level in appearance detection and can continue the subsequent detection process (if necessary); if the model finds that any data does not meet the requirements, it is determined that the function detection result is "not passed", at this time, in addition to outputting the overall result of "not passed", the function anomaly detection result is also output, which clearly indicates which specific detection items are abnormal, for example, "the surface crack length exceeds the standard value, the maximum crack length reaches 5mm, the standard requirement is not more than 3mm" or "the cutting edge angle deviation is too large, the measured angle is 125°, the standard range is 118°~122°", etc., so that the subsequent quality improvement and problem troubleshooting work can quickly locate the problem, take targeted measures for repair or adjustment, and at the same time, according to the "not passed" result, stop unnecessary detection processes such as hardness test, cutting performance test, etc. according to the pre-set rules, avoid waste of resources and consumption of time, and improve the overall detection efficiency and cost-effectiveness.

[0041] At step S400, if the function detection result is passed, the detection data collection of N secondary nodes is performed under the guidance of the function detection tree, and the quality detection analysis of the diamond drill bit is performed based on the obtained N-dimensional detection data, and N secondary detection results are output. Specifically, after the function detection result is passed, the N secondary nodes and their corresponding function detection items are determined under the guidance of the function detection tree, such as the hardness detection which needs to determine the measurement position, method, precision and collection point number, the coating adhesion detection which needs to determine the detection standard, area and quantitative evaluation index, and the internal structure integrity detection which involves flaw detection technical means and defect information collection requirements. Detection data collection is performed respectively, such as the hardness detection which measures and records the hardness value at each key position multiple times, the coating adhesion detection which makes a pattern by grid method, evaluates and records the falling characteristics after the adhesive tape is torn, and the internal structure integrity detection which obtains an image by X-ray flaw detection and analyzes the defect parameters. Based on the collected N-dimensional detection data, the quality detection analysis is performed, the hardness condition is judged by comparing the hardness data with the standard through statistical analysis, whether the coating adhesion is good is analyzed according to the coating adhesion rating and the falling characteristics, whether there is a problem in the internal structure is judged according to the defect information of the internal structure integrity detection, and N secondary detection results are output.

[0042] In a possible implementation, if the function detection result is passed, the detection data collection of the N secondary nodes is performed under the guidance of the function detection tree, and the diamond drill bit is detected and analyzed based on the obtained N-dimensional detection data, and N secondary detection results are output. Step S400 further includes step S410. If the function detection result is passed, the N reference data requirements of the N secondary nodes are called from the H reference data requirements under the guidance of the function detection tree. Specifically, when it is confirmed that the function detection result of the root node (for example, appearance detection) is "passed" (indicated by 1 to represent the pass state), the next operation is performed based on the constructed function detection tree. The function detection tree is constructed based on the association relationship and importance level of each function of the diamond drill bit, and the logical path from the root node to each secondary node is determined. It is known that there are H function detection items and corresponding H reference data requirements. According to the guidance of the function detection tree, the function detection items corresponding to the N secondary nodes are accurately positioned. For example, if the secondary nodes include hardness detection, cutting performance detection and wear resistance detection, the detailed reference data requirements corresponding to each of the three detection items are selected from the H reference data requirements. For example, the reference data requirements for hardness detection cover the hardness value range of different parts, the standard method of hardness test and the allowed hardness deviation, etc.; the reference data requirements for cutting performance detection involve the maximum and minimum values of cutting force, the reasonable interval of cutting speed, the shape and size standard of cutting chip, etc.; and the reference data requirements for wear resistance detection include the limit value of wear amount, the wear uniformity index, and the cycle number and conditions of wear resistance test, etc., thereby clearly defining the direction and standard for subsequent detection data collection.

[0043] Step S420, according to the N reference data requirements, the diamond drill bit is detected data directional collection, obtain the N-dimensional detection data. Specifically, according to the reference data requirements of the N secondary nodes called in the last step, organize professional detection personnel and equip corresponding high-precision detection equipment to carry out data collection work. Taking hardness detection as an example, if the reference data requirements stipulate that the hardness of the cutting edge, side surface and center of the drill bit is measured, and the Rockwell hardness test method is used, the detection personnel will use the Rockwell hardness tester to select a plurality of uniformly distributed test points (such as 5 points) on the cutting edge, apply the specified load according to the standard test procedure, and record the hardness value of each point; a suitable number of test points (such as 3 points each) are also selected on the side surface and the center, and the same operation is performed to obtain a series of detection data about hardness. For cutting performance detection, set the cutting experiment conditions according to the reference data requirements, such as using a specific cutting tool, fixed cutting speed and feed rate, in the simulation of actual work cutting process, through the sensor installed on the equipment, real-time collection of cutting force, cutting temperature and other data, and measurement and recording of the shape and size of the chip. For wear resistance detection, install the drill bit on a special wear testing equipment, perform the experiment according to the predetermined wear test cycle and loading conditions, measure the wear amount of the drill bit regularly, observe the uniformity of the wear, and record the data of each measurement. Through targeted detection operation of each secondary node detection item, multi-dimensional detection data related to N secondary nodes, i.e. N-dimensional detection data, are collected, which will be used as the basis for subsequent quality detection analysis.

[0044] Step S430, according to the hierarchical characteristics of the N secondary nodes, map and activate N quality anomaly detection models in the detection model activation tree. Specifically, the detection model activation tree is constructed according to the hierarchical relationship and logical order of the functional detection items, and is related to the functional detection tree but focuses on the activation control of the model. Each secondary node has a corresponding node position and activation condition in the detection model activation tree. When the N secondary nodes are determined, the model activation operation is performed according to the hierarchical characteristics in the detection model activation tree. For example, the quality anomaly detection model corresponding to hardness detection may be trained by machine learning algorithms (such as support vector machine, neural network, etc.) based on a large amount of historical hardness detection data and known quality anomaly cases, when hardness detection is triggered as a secondary node, through the mapping mechanism of the detection model activation tree, the quality anomaly detection model for hardness detection is found and activated, so that it is in the state of being ready to receive data for analysis. For the secondary nodes of cutting performance detection and wear resistance detection, the corresponding quality anomaly detection models are also activated, and the models are pre-trained and optimized to efficiently and accurately analyze the anomalies and evaluate the quality of the corresponding detection data, providing intelligent support for subsequent quality detection analysis.

[0045] Step S440, by loading the N-dimensional detection data to the N quality anomaly detection models for quality detection analysis, outputting the N secondary detection results. Specifically, the N-dimensional detection data collected in the previous step is arranged and loaded according to the format and data preprocessing method required by the respective quality anomaly detection model. For example, for hardness detection data, data normalization is required to convert the hardness value to a specific numerical interval to meet the requirements of model input; for cutting performance detection data, feature engineering processing is required for cutting force, cutting temperature and other multiple variables, key features are extracted and combined into a vector form suitable for model input. The processed N-dimensional detection data is input into the corresponding N quality anomaly detection models. The model will analyze and identify the input data according to the built-in algorithm and the parameters obtained by training. Taking the hardness detection quality anomaly detection model as an example, the input hardness data is compared with the distribution characteristics of historical normal data to determine whether there is hardness anomaly, such as whether there is local hardness too low or uneven hardness; the cutting performance detection quality anomaly detection model analyzes the fluctuation of cutting force, the change trend of cutting temperature and the abnormal characteristics of chip shape, and comprehensively judges whether the cutting performance meets the quality requirements; the wear resistance detection quality anomaly detection model evaluates whether the wear resistance of the drill bit meets the standard according to the growth trend of the wear amount, the distribution of the wear and the difference between the expected wear resistance. After independent analysis of each model, N secondary detection results are output, each result clearly indicates whether the corresponding secondary node detection item passes the quality detection, for example, "hardness detection passes" "cutting performance detection does not pass, cutting force is too large" "wear resistance detection passes" and the like, the secondary detection results provide detailed and accurate basis for further detection decision and quality evaluation, guiding the subsequent investigation of diamond drill bit quality problems and the formulation of improvement measures.

[0046] Step S500, the out-degree connection relationship between the N secondary nodes and the M tertiary nodes in the function detection tree is taken as the judgment basis for whether to continue the quality detection, and the function detection of the M tertiary nodes is performed according to the pass identification of the N secondary detection results, wherein N and M are positive integers and N < M. Specifically, the out-degree connection relationship between the N secondary nodes and the M tertiary nodes in the function detection tree is determined, and this relationship is constructed based on the deep understanding of the internal logic and mutual influence of each function of the diamond drill bit. For example, there is a correlation between "drill bit hardness detection" and "cutting performance detection", and a correlation between "coating quality detection" and "corrosion resistance performance detection". The connection relationship is set as the judgment basis for whether to continue the quality detection. After the detection of the N secondary nodes is completed and the corresponding secondary detection results with pass identification (such as 1 for pass and 0 for fail) are obtained, the detection execution of the tertiary nodes is determined according to the judgment basis. If the secondary node detection passes and has an out-degree connection with the tertiary node, such as "drill bit hardness detection" passing, the "cutting performance detection" is started, and the detection personnel will detect and analyze the cutting force and other indicators according to the established standard and method. If the secondary node detection fails, such as "coating quality detection" failing, the "corrosion resistance performance detection" and other tertiary node detections associated therewith are directly skipped. In this way, the detection can be comprehensive and invalid detection can be avoided, thereby effectively improving the efficiency and pertinence of the quality detection of the diamond drill bit.

[0047] Step S600, by analogy, the quality detection execution is judged according to the detection results of the previous level nodes in the function detection tree, and the quality defect information set of the diamond drill bit is output. Specifically, the detection is started from the root node (such as appearance detection) of the function detection tree, the detection is implemented according to the established process and standard, and the results are obtained. If the root node fails, the defect details related to the appearance (such as crack position and size deviation value) are recorded as the initial content of the quality defect information set. If the root node passes, the secondary nodes are detected according to the tree structure, and the results (pass or fail) are obtained. The detection of the tertiary nodes is determined according to the detection results of the secondary nodes and the out-degree connection relationship between the secondary nodes and the tertiary nodes. If the secondary node fails, the detection of the associated tertiary node is skipped. If the secondary node passes, the corresponding detection is started, such as "hardness detection" passing and then "cutting performance detection". In each level of detection, if a node fails, the functional abnormality and data (such as excessive cutting force, abnormal chip shape and measurement data when "cutting performance detection" fails) are recorded in detail and are included in the quality defect information set. In this way, the subsequent level detection is promoted according to the results of the previous level, the quality defect information is continuously updated and summarized, and all necessary level detections are completed. Finally, the quality defect information set comprehensively covers all kinds of quality problems of the detected diamond drill bit, and provides accurate basis for subsequent work to improve product quality.

[0048] In a possible implementation, by analogy, the quality detection execution judgment is performed according to the detection result of the previous level sub-node in the function detection tree, and the quality defect information set of the diamond drill bit is output, and step S600 further includes step S610, that is, whether the quality defect information set covers the H function detection items. Specifically, the existing quality defect information set is obtained, and the information set is accumulated in the process of gradually detecting the diamond drill bit, and contains various quality problems and related data that have been found, and each problem corresponds to a specific function detection item. The function detection items involved in the information set are compared with the total H function detection items one by one. For example, the existing quality defect information set records the problems found in appearance detection, hardness detection and cutting performance detection, and the total function detection items include appearance detection, hardness detection, cutting performance detection, wear resistance detection, impact resistance detection and the like (assuming H=5). Through checking, it is confirmed whether the quality defect information set covers the five function detection items, and this step is a key basis for subsequent decision, which can clearly indicate the completeness of the current detection and whether further detection operation is needed.

[0049] Step S620, if the quality defect information set does not cover the H function detection items, a detection item deviation set is output. Specifically, when it is found through the above judgment that the quality defect information set does not cover all the H function detection items, it is necessary to determine which function detection items are missed. The items in the H function detection items that do not appear in the quality defect information set are extracted to form the detection item deviation set. For example, if the quality defect information set lacks information related to wear resistance detection and impact resistance detection, then the detection item deviation set includes the two items. The purpose of outputting the deviation set is to clearly indicate the direction that needs to be focused on in subsequent detection work, so as to carry out supplementary detection in a targeted manner, thereby improving the comprehensive understanding of the quality status of the diamond drill bit and ensuring that important function quality problems are not ignored.

[0050] Step S630, after sampling the same batch of sample drill bits of the diamond drill bit through system sampling, the quality detection of the sample drill bits is carried out according to the functional detection tree and the detection model activation tree, and the quality defect information set is updated based on the detection result until the quality defect information set covers the H functional detection items. Specifically, once the detection item deviation set is determined, further sampling detection needs to be carried out on the same batch of diamond drill bits. The system sampling method is adopted, and the sample drill bits are selected from the same batch of drill bits according to certain sampling rules (such as taking one every N drill bits) to ensure the randomness and representativeness of sampling. For the sample drill bits extracted, the quality detection work is carried out according to the previously constructed functional detection tree and detection model activation tree. The functional detection tree provides a clear hierarchical structure and logical order for the detection process. Starting from the root node, whether to detect the child nodes of the next level is determined according to the detection result of the previous level, and the detection process of each functional detection item is gradually promoted. The detection model activation tree activates the corresponding quality abnormality detection model in time according to the characteristics and hierarchical relationship of the functional detection items, and efficiently and accurately analyzes the collected detection data. For example, when an abnormality of a functional detection item of a sample drill bit is detected, the abnormality and its related data are recorded in detail and included in the quality defect information set, and the information set is updated. Then, it is judged again whether the updated quality defect information set covers the H functional detection items. If it is still not completely covered, system sampling and sample drill bit detection are continued, and the above process is repeated, and the quality defect information set is continuously supplemented and improved until it finally covers all the H functional detection items. At this time, it can be considered that the quality detection of the batch of diamond drill bits has reached a relatively comprehensive and in-depth level, which can provide sufficient basis for subsequent quality evaluation, improvement measure formulation, etc., and ensure the quality stability and reliability of the product.

[0051] The embodiment of the present application adopts the construction of a functional detection tree through functional loss correlation analysis, wherein the root node is appearance detection. According to the root node, directional data acquisition is carried out to obtain single-dimensional detection data, and quality detection analysis is carried out. If the functional detection passes, data acquisition and analysis of the secondary child nodes are continued according to the detection tree, and the secondary detection result is output. It is judged whether to continue to execute the tertiary child node detection according to the secondary detection result. The detection result of each level determines the next execution, so as to gradually output the quality defect information set of the diamond drill bit, avoid redundant detection, improve the detection efficiency, optimize the detection process through functional loss correlation analysis, reduce unnecessary detection items, and achieve the technical effects of improving the detection efficiency and precision.

[0052] The foregoing DETAILED DESCRIPTION, including the above section titled "Detailed Description," is not to be taken as limiting the scope of the application. Various modifications, combinations, and equivalents can be apparent to those skilled in the art and can be made once the nature of the application is understood. Any modification, combination, or equivalent, which falls within the principles and the scope of the present application, is intended to be included in the present application. In some instances, the actions or steps can be performed in different order from those described herein, and still achieve desirable results. Additionally, the process depicted in the figures can not necessarily require the particular order shown or sequential order to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

Claims

1. A method for detecting the quality of a diamond drill bit by multi-dimensional parameter analysis, characterized in that, The method comprises: Performing loss-of-function correlation analysis on the diamond drill bit, and constructing a function detection tree based on the analysis result, wherein the root node of the function detection tree is appearance detection; Taking the root node as the starting point of quality detection, collecting detection data of the diamond drill bit according to the function detection project of the root node, and obtaining single-dimensional detection data; Performing quality detection analysis on the diamond drill bit according to the single-dimensional detection data, and outputting a function detection result; If the function detection result is passed, collecting detection data of N secondary nodes guided by the function detection tree, and performing quality detection analysis on the diamond drill bit based on the obtained N-dimensional detection data, and outputting N secondary detection results; Taking the out-degree connection relationship between the N secondary nodes and M tertiary nodes in the function detection tree as the judgment basis for whether to continue quality detection, and performing function detection of the M tertiary nodes according to the pass identification of the N secondary detection results, wherein N and M are positive integers and N < M; In this way, the quality detection execution judgment is performed according to the detection result of the previous level node in the function detection tree, and the quality defect information set of the diamond drill bit is outputted; Performing loss-of-function correlation analysis on the diamond drill bit, and constructing a function detection tree based on the analysis result, the method further comprises: Obtaining the diamond drill bit from the drill bit production line by system sampling; Interactively obtaining the batch production use scene of the drill bit production line, and obtaining the use scene setting by collecting limit use data of the batch production use scene; According to the use scene setting, network data calling is performed to obtain a detection item set of the diamond drill bit, wherein the detection item set includes H function detection items, and H is a positive integer greater than M; The detection item set is called as a constraint to obtain historical function defect records; According to the historical function defect records, the detection item set is analyzed for loss-of-function correlation, and the function detection tree is constructed based on the analysis result, the method further comprises: By analyzing the H function detection items for loss-of-function synchronicity based on the historical function defect records, a function loss probability matrix is constructed; A transition probability correlation threshold is preset, appearance detection is taken as the root node, and appearance detection and the transition probability correlation threshold are used to traverse the function loss probability matrix to obtain N function detection items, and the N function detection items are taken as the N secondary nodes, wherein the root node and the N secondary nodes are out-degree connected; The N secondary nodes and the transition probability correlation threshold are used to traverse the function loss probability matrix to obtain N groups of function detection items; N groups of tertiary nodes of the N groups of function detection items are constructed, and the N secondary nodes and the N groups of tertiary nodes are out-degree connected, wherein the N groups of tertiary nodes collectively include the M tertiary nodes; In this way, until the nodes in the function detection tree cover the H function detection items, the construction of the function detection tree is completed.

2. The method for detecting the quality of the diamond drill bit by multi-dimensional parameter analysis according to claim 1, wherein, The method further comprises: Based on the history function defect record, the H function detection items are analyzed for function loss synchronism to construct a generated function loss probability matrix. Based on the history function defect record, the H function detection items are analyzed for function loss synchronism to construct a generated function loss probability matrix. According to the H function detection items and the H group of function loss transition probabilities, the function loss probability matrix is constructed and generated. According to the single-dimensional detection data, the diamond drill bit is analyzed for quality detection, and a function detection result is output.

3. The method for quality inspection of diamond drill bits using multidimensional parameter analysis as described in claim 1, characterized in that, H reference data requirements of the H function detection items are interactively obtained. According to the H reference data requirements and the model information of the diamond drill bit, network data calling is performed to obtain H quality abnormal information sets, wherein each quality abnormal information set includes multiple sample reference data and multiple sample abnormal detection results. H standard abnormal detection models are preconfigured according to the H detection feature differences of the H function detection items. The H quality abnormal information sets are used as training data to perform directional parameter optimization of the H standard abnormal detection models to obtain H quality abnormal detection models. According to the hierarchical relationship of the H function detection items in the function detection tree, an activation hierarchical structure of the H quality abnormal detection models is constructed to complete the establishment of a detection model activation tree. According to the single-dimensional detection data, the diamond drill bit is analyzed for quality detection, and a function detection result is output.

4. The method for quality inspection of diamond drill bits using multidimensional parameter analysis as described in claim 3, characterized in that, According to the function detection item of the root node, a first reference data requirement is called from the H reference data requirements. According to the first reference data requirement, detection data of the diamond drill bit is directionally collected to obtain the single-dimensional detection data. According to the root node, a first quality abnormal detection model is activated in the detection model activation tree. The single-dimensional detection data is loaded into the first quality abnormal detection model for quality detection analysis, and the function detection result is output, wherein when the function detection result is passed, the function detection result further includes a function abnormal detection result. If the function detection result is passed, detection data of N secondary nodes is collected according to the function detection tree, and quality detection analysis is performed on the diamond drill bit based on the obtained N-dimensional detection data to output N secondary detection results, and the method further comprises:

5. The method for quality inspection of diamond drill bits using multidimensional parameter analysis as described in claim 4, characterized in that, If the function detection result is passed, N reference data requirements of the N secondary nodes are called from the H reference data requirements according to the function detection tree. According to the N reference data requirements, detection data of the diamond drill bit is directionally collected to obtain the N-dimensional detection data. ​ According to the hierarchical characteristics of the N secondary nodes, the detection model activation tree is mapped to activate N quality anomaly detection models; By loading the N-dimensional detection data into the N quality anomaly detection models for quality detection analysis, the N secondary detection results are output.

6. The method of claim 3, wherein the multi-dimensional parameter analysis is performed by using a neural network. The method comprises: determining whether the quality defect information set covers the H functional detection items; If the quality defect information set does not cover the H functional detection items, a detection item deviation set is output. After system sampling is performed on the same batch of sample drill bits of the diamond drill bit, quality detection of the sample drill bits is performed according to the functional detection tree and the detection model activation tree, and the quality defect information set is updated based on the detection results until the quality defect information set covers the H functional detection items.

Images