Grinding wheel wear degree recognition method and device based on adaptive neural fuzzy inference

Abstract

The application discloses a grinding wheel wear degree recognition method and device based on adaptive neural fuzzy reasoning, and the method comprises the following steps: acquiring a grinding temperature signal, a grinding force and the surface quality of a workpiece; determining the wear stage of the grinding wheel according to the surface quality, the grinding force and the grinding temperature; extracting the total number of thermal pulses in the grinding temperature signal as the dynamic effective abrasive grain number; acquiring the surface topography of the grinding wheel, identifying and distinguishing the grinding dust and abrasive grain wear in the surface topography of the grinding wheel through an HSV segmentation method, and calculating the area rate to obtain the grinding dust proportion and the abrasive grain wear proportion; taking the grinding dust proportion, the abrasive grain wear proportion and the dynamic effective abrasive grain number as the input, taking the wear stage as the output, training the adaptive neural fuzzy reasoning system, and obtaining the trained adaptive neural fuzzy reasoning system to recognize the wear degree. The application utilizes the adaptive neural fuzzy reasoning technology, comprehensively considers the multi-dimensional characteristic signals in the grinding process, and can accurately and conveniently judge the wear condition of the grinding wheel.

Description

Technical Field

[0001] This invention relates to the field of precision machining, and in particular to a method and apparatus for identifying the wear degree of grinding wheels based on adaptive neural fuzzy reasoning. Background Technology

[0002] In modern manufacturing, precision machining processes place increasingly higher demands on surface quality, geometric accuracy, and processing efficiency, especially in fields such as aerospace, automotive, and mold making.

[0003] Diamond wheel grinding is widely used for machining cemented carbide, ceramic materials, and other difficult-to-machine materials. While wheel grinding offers high speed and efficiency, it also involves the continuous wear of the grinding wheel itself. This wear manifests as abrasive grain shedding, passivation, and chip clogging, gradually causing the grinding wheel to lose its original machining capabilities. Failure to monitor the wear level of the grinding wheel in a timely manner can not only affect machining quality but also the dimensional accuracy and surface integrity of the workpiece. Therefore, identifying the wear condition of the grinding wheel to determine when it needs dressing or replacement is crucial for ensuring production efficiency and product quality.

[0004] Currently, grinding wheel wear detection technologies can be broadly categorized into two types: direct detection and indirect detection. Direct detection technologies primarily determine the wear state by measuring the amount of wear or changes in the morphology of the grinding wheel surface (e.g., stylus method, imprint method, optical cross-section method, and electron microscopy). While these methods are intuitive, they require high-precision testing equipment and typically necessitate stopping the machine for testing, thus making them unsuitable for online real-time monitoring. Indirect monitoring technologies, on the other hand, indirectly infer the wear state of the grinding wheel by analyzing various physical quantities during the grinding process (e.g., acoustic emission signals, vibration signals, current, voltage, temperature, etc.). This method allows for online monitoring, but due to the complexity of the grinding process, measuring a single physical quantity often fails to comprehensively and accurately reflect the wear condition of the grinding wheel. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a method and apparatus for identifying the wear degree of grinding wheels based on adaptive neural fuzzy reasoning, so as to improve the above-mentioned problems.

[0006] This invention provides a method for identifying the wear level of grinding wheels based on adaptive neurofuzzy inference, comprising:

[0007] S101, during the continuous grinding of the workpiece by the grinding wheel, acquire the grinding temperature signal, grinding force and surface quality of the workpiece at each grinding stage;

[0008] S102, determine the wear stage of the grinding wheel based on the surface quality of the workpiece, grinding force and grinding temperature;

[0009] S103, extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains;

[0010] S104: Obtain the surface morphology image of the grinding wheel after each grinding. Use the HSV segmentation method to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel and calculate the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear.

[0011] S105, with the proportion of grinding debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output, the constructed adaptive neurofuzzy inference system is trained to obtain the trained adaptive neurofuzzy inference system.

[0012] S106, Identify the degree of wear of the grinding wheel based on the trained adaptive neurofuzzy inference system.

[0013] Preferably, in step S101, a type k thermocouple is used to measure the grinding temperature signal and grinding force, wherein the type k thermocouple is used to grind to 4-10 micrometers.

[0014] Preferably, in step S102, the wear stages of the grinding wheel are divided into initial, intermediate, severe, and dull stages.

[0015] Preferably, step S103 includes:

[0016] The range of noise signal in the grinding temperature signal is obtained, and the noise threshold is obtained;

[0017] The thermal pulse counting program was used to analyze multiple sets of temperature signals at the same wear stage, and the average number of thermal pulses greater than the noise threshold in the grinding temperature signal was used as the dynamic effective number of abrasive particles.

[0018] Preferably, step S104 includes:

[0019] Multiple shooting areas are evenly divided in the circumferential direction of the grinding wheel, and three shooting areas (front, middle, and rear) are divided in the axial direction. Multiple images of the grinding wheel surface morphology are obtained by taking pictures in the shooting areas using a camera.

[0020] The surface topography image of the grinding wheel is converted from the RGB color space to the HSV color space to obtain an HSV image;

[0021] Color feature extraction technology was used to obtain the H, S, and V channel information of HSV images, and histogram analysis was performed on each channel.

[0022] Based on the results of histogram analysis, the threshold ranges for wear debris blockage and abrasive wear in the H, S, and V channels are determined, and a mask is created based on the threshold ranges to mark pixels on the HSV image that are within the specified HSV value range.

[0023] The HSV image is filtered by applying the mask to obtain the total number of pixels N1 and N2 of the wear debris blockage and wear area.

[0024] The wear debris percentage is calculated based on the total number of pixels N in the HSV image. The proportion of abrasive wear is

[0025] Preferably, step S105 includes:

[0026] A sample dataset is generated based on the wear debris ratio, abrasive wear ratio, and dynamic effective wear number, and the sample dataset is randomly divided into independent training dataset and test dataset.

[0027] Define fuzzy sets: set 2 to 8 types of initial membership functions to be determined and 2 to 5 numbers of fuzzy subsets to be determined;

[0028] Define the fuzzy system structure: Generate the fuzzy inference system using the grid partitioning method;

[0029] Model Training: Set the number of iterations, and use a hybrid neural network algorithm of backpropagation and least squares to train the fuzzy inference system, so that the output of the fuzzy inference system continuously approaches the training results of the training dataset. When the preset conditions are met, the training stops, and an adaptive neural network fuzzy inference system model is obtained.

[0030] Preferably, it further includes:

[0031] Model optimization: The membership function and the number of fuzzy subsets are determined based on the root mean square error and the number of iterations in training and testing. Cross-validation is used to determine whether the training is overfitting.

[0032] The formulas for calculating the root mean square error and cross-validation are as follows:

[0033] Suppose there are k subsets, each of size n. i There are a total of n samples, of which

[0034]

[0035] For the i-th subset, let its prediction result be... Then the root mean square error:

[0036]

[0037] Where y ij It is the actual value of the j-th sample in the i-th subset. It is the predicted value of the j-th sample in the i-th subset;

[0038] Finally, calculate the average RMSE:

[0039]

[0040] This yields the model's training error and the mean square error of cross-validation. Training is stopped when the training error is less than 5%, and if the mean square error of cross-validation differs from the training error by less than 2.5%, the model is considered well-fitted; otherwise, it is considered overfitting.

[0041] This invention also provides a grinding wheel wear degree recognition device based on adaptive neural fuzzy reasoning, which includes:

[0042] The acquisition unit is used to acquire the grinding temperature signal, grinding force, and surface quality of the workpiece at each grinding stage during the continuous grinding process of the grinding wheel.

[0043] The wear stage determination unit is used to determine the wear stage of the grinding wheel based on the surface quality of the workpiece, the grinding force, and the grinding temperature.

[0044] The extraction unit is used to extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains.

[0045] The proportion acquisition unit is used to acquire the surface morphology image of the grinding wheel after each grinding. It uses the HSV segmentation method to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel and calculates the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear.

[0046] The training unit is used to train the constructed adaptive neurofuzzy inference system with the proportion of wear debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output, so as to obtain the trained adaptive neurofuzzy inference system.

[0047] The recognition unit is used to identify the wear level of the grinding wheel based on the trained adaptive neurofuzzy inference system.

[0048] In summary, this invention utilizes computer image processing and adaptive neuro-fuzzy reasoning technology to analyze the wear of the grinding wheel by integrating multi-dimensional feature signals (grind wheel surface state, dynamic effective number of abrasive grains) during the grinding process. It can accurately and conveniently determine the wear condition of the grinding wheel without disassembling it. Attached Figure Description

[0049] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1This is a flowchart illustrating the grinding wheel wear degree recognition method based on adaptive neurofuzzy reasoning provided in the first embodiment of the present invention.

[0051] Figure 2 A flowchart of a thermal pulse counting program provided in an embodiment of the present invention;

[0052] Figure 3 A schematic diagram of the image acquisition area of ​​the grinding wheel surface provided in the embodiment;

[0053] Figure 4 The wear diagrams of the fixed-point grinding wheel under different material removal volumes V are provided in the embodiments.

[0054] Figure 5 A flowchart illustrating the adaptive neuro-fuzzy inference system provided in this embodiment;

[0055] Figures 6(a) and 6(b) are structural framework diagrams of the adaptive neural fuzzy reasoning system of the present invention;

[0056] Figure 7 This is a schematic diagram of the structure of the grinding wheel wear degree recognition device based on adaptive neural fuzzy reasoning provided in the second embodiment of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figure 1 This invention provides a method for identifying the wear level of grinding wheels based on adaptive neural fuzzy reasoning, comprising:

[0059] S101 acquires the grinding temperature signal, grinding force, and surface quality of the workpiece at each grinding stage during the continuous grinding process of the grinding wheel.

[0060] Specifically, in this embodiment, a wear experiment can be conducted on the workpiece being ground by a grinding wheel, and the grinding temperature, grinding force, and workpiece surface quality at each stage of the diamond grinding wheel wear process can be recorded.

[0061] Among these methods, type K thermocouples can be used to measure grinding temperature and grinding force. To reduce measurement errors, the thermocouple thickness needs to be ground to 4–10 micrometers. The recorded grinding temperature and grinding force signals are filtered and processed in Dewesoft software, and then the data is exported.

[0062] S102, determine the wear stage of the grinding wheel based on the surface quality of the workpiece, grinding force, and grinding temperature.

[0063] The wear stages of the grinding wheel can be divided according to the changes in grinding force at different grinding stages. For example, they can be divided into initial, intermediate, heavy, and dull stages. Of course, the wear stages can also be set according to actual needs, and this invention does not impose specific limitations.

[0064] S103, extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains.

[0065] Specifically, such as Figure 2 As shown, the number of dynamic effective abrasive particles can be counted using a thermal pulse counting program:

[0066] First, the grinding temperature signal is imported into the program as data. Each row is a data point, totaling i rows, and each column is the number of grinding temperature signal samples, totaling j columns. Each column of data is traversed by nested for loops, and adjacent data points are subtracted to obtain the data fluctuation value Td. Then, values ​​of Td greater than the noise threshold are saved in the Wt array and counted. Finally, the total number of thermal pulses of c samples is stored in the Pt array in the form of 1 row and c columns and output. The average value of the total number of thermal pulses of c samples will be used as the dynamic effective number of abrasive grains that pass through the thermocouple in the grinding arc area.

[0067] S104. Obtain the surface morphology image of the grinding wheel after each grinding. Use the HSV segmentation method to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel and calculate the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear.

[0068] Specifically, S104 includes:

[0069] First, multiple shooting areas are evenly divided in the circumferential direction of the grinding wheel, and three shooting areas (front, middle, and rear) are divided in the axial direction. Multiple images of the grinding wheel surface morphology are obtained by taking pictures in the shooting areas using a camera.

[0070] like Figure 3 As shown, the grinding wheel is evenly divided into 6 to 20 shooting areas along its circumference and into three shooting areas along its axial direction: front, middle, and rear. Markings are made at 6 sampling points for fixed-point shooting. Figure 4 This is a graph showing the wear effect of a grinding wheel at a fixed point under different material removal rates V.

[0071] Then, the surface topography image of the grinding wheel is converted from the RGB color space to the HSV color space to obtain an HSV image.

[0072] The process involves importing the photographed image of the grinding wheel surface morphology into a computer, then using the imread function in image processing software such as MATLAB to read the image, and using the rgb2hsv function to convert the original image from RGB to HSV color space to obtain an HSV image.

[0073] Next, based on the results of histogram analysis, the threshold ranges for wear debris blockage and abrasive wear in the H, S, and V channels are determined, and a mask is created based on the threshold ranges to mark pixels on the HSV image that are within the specified HSV value range.

[0074] Specifically, the hsv Image function is used to obtain the H, S, and V channel information of the original image through indices (:,:,1), (:,:,2), and (:,:,3), respectively. Then, the imshow function is used to display the image of each channel and perform histogram analysis on each channel. Through these analyses, the threshold range of wear debris and wear particles can be determined, thereby creating a mask to mark pixels that meet the conditions and performing pixel filtering on the original HSV image.

[0075] Next, the mask is applied to filter the pixels of the HSV image to obtain the total number of pixels N1 and N2 of the wear area and the area of ​​wear debris blockage.

[0076] Finally, the wear debris percentage is calculated based on the total number of pixels N in the HSV image. The proportion of abrasive wear is

[0077]

[0078] S105 uses the proportion of grinding debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output to train the constructed adaptive neurofuzzy inference system, thus obtaining the trained adaptive neurofuzzy inference system.

[0079] like Figure 5 As shown, in this embodiment, after the data is fully prepared, the adaptive neurofuzzy inference system can be constructed. The adaptive neurofuzzy inference system (ANFIS) combines the self-learning ability of neural networks with the reasoning ability of fuzzy logic, enabling it to build relatively accurate models in complex nonlinear systems and possessing strong fault tolerance and robustness. The construction process specifically includes:

[0080] First, the previously obtained dataset (the proportion of grinding debris clogging and the proportion of abrasive wear are used as inputs, and the degree of grinding wheel wear is used as the output) is divided into a training set and a test set.

[0081] Under the grid partitioning method, select the type of membership function (MF), such as trimf (MF1), trapmf (MF2), gbe l lmf l (MF3), gaussmf (MF4), etc., and input the number of MFs (i.e., the number of corresponding fuzzy subsets) to establish the FIS system and perform fuzzification processing on the input data.

[0082] Then, a hybrid neural network algorithm of backpropagation and least squares was selected to train the model, with 40 iterations. During this process, the membership function parameters and rule parameters in the system are automatically optimized. Training stops when the model output continuously approaches the training results of the training dataset and reaches the preset condition (in this embodiment, the error is less than 5%). Since the optimal number of iterations varies depending on the membership function type and the number of fuzzy subsets, and the number of iterations is related to the program running time, it can be used as a reference for selecting the optimal model.

[0083] As shown in Figures 6(a) and 6(b), the ANFI S structure used in this embodiment consists of 5 layers: a fuzzification layer, a basic rule layer, a normalization layer, a defuzzification layer, and an output layer. The number of fuzzy rules is obtained by multiplying the number of fuzzy subsets of each input. For example, in Figure 6(a), the number of fuzzy rules is 2x2x2 = 8 (3 input variables, each variable has 2 fuzzy subsets); while in Figure 6(b), the number of fuzzy rules is 3x3x3 = 27 (3 input variables, each variable has 3 fuzzy subsets).

[0084] The form of the fuzzy rule is: lfxis A and yis B, then zis C, where x and y are input variables, z is the output variable, and A, B, and C are different fuzzy sets;

[0085] During the modeling process, the number of fuzzy subsets for each input variable can be set freely as needed. However, considering the running speed and modeling requirements, the number of MFs for each input parameter in this case is selected to be in the range of 2 to 5.

[0086] Since the parameters of the input variables are fixed, the quality of the model mainly depends on the types of commonly used MFs (4 types) and the number of MFs corresponding to each input parameter (this paper selects 2 to 5 cases). To facilitate model selection, this embodiment uses the root mean square error (RMSE) obtained after running the ANF IS system to initially screen the models, which is used to measure the accuracy of prediction. The smaller the RMSE value, the higher the prediction accuracy. To evaluate the stability and generalization ability of the model, and to avoid overfitting on a specific training set, k-fold cross-validation is introduced. Through multiple training and evaluations, a more comprehensive understanding of the model's performance can be obtained.

[0087] The formulas for calculating the root mean square error and cross-validation are as follows:

[0088] Suppose there are k subsets, each of size n. i There are a total of n samples, of which

[0089]

[0090] For the i-th subset, let its prediction result be... Then the root mean square error:

[0091]

[0092] Where y ij It is the actual value of the j-th sample in the i-th subset. It is the predicted value of the j-th sample in the i-th subset;

[0093] Finally, calculate the average RMSE:

[0094]

[0095] By training the model with different MF types and numbers, this embodiment obtained the data shown in Table 1. First, the root mean square error (RMSE) of training, testing, and validation was controlled to be within 10%. Parameters exceeding 10% were excluded. Then, the average RMSE of cross-validation was compared. If the average RMSE was greater than 2.5% of the training error, the model was considered overfitted. Finally, the validation error and the number of iterations were compared. Smaller validation errors and iterations indicate higher model prediction accuracy and shorter prediction time. Through a series of judgments, the optimal parameters for this case model were determined to be: trapezoidal membership function and 4 fuzzy subsets.

[0096] Table 1 Comparison of RMSE, cross-validation, and number of iterations for different MF types and MF numbers.

[0097]

[0098] S106, Identify the degree of wear of the grinding wheel based on the trained adaptive neurofuzzy inference system.

[0099] In this embodiment, based on the obtained adaptive neural fuzzy inference system, the system predicts the dynamic effective number of abrasive particles collected from the grinding wheel, the proportion of abrasive debris after image processing, and the wear ratio of abrasive particles. The wear degree of the grinding wheel can be obtained through the output of the system.

[0100] In summary, this invention utilizes computer image processing and adaptive neuro-fuzzy reasoning technology to analyze the wear of the grinding wheel by integrating multi-dimensional feature signals (grind wheel surface state, dynamic effective number of abrasive grains) during the grinding process. It can accurately and conveniently determine the wear condition of the grinding wheel without disassembling it.

[0101] Please see Figure 7 The second embodiment of the present invention also provides a grinding wheel wear degree recognition device based on adaptive neural fuzzy reasoning, which includes:

[0102] The acquisition unit 210 is used to acquire the grinding temperature signal, grinding force and surface quality of the workpiece at each grinding stage during the continuous grinding process of the grinding wheel.

[0103] Wear stage determination unit 220 is used to determine the wear stage of the grinding wheel based on the surface quality of the workpiece, grinding force and grinding temperature;

[0104] Extraction unit 230 is used to extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains;

[0105] The proportion acquisition unit 240 is used to acquire the surface morphology image of the grinding wheel after each grinding, and to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel by HSV segmentation method and calculate the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear.

[0106] Training unit 250 is used to train the constructed adaptive neurofuzzy inference system with the proportion of wear debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output, so as to obtain the trained adaptive neurofuzzy inference system.

[0107] The recognition unit 260 is used to identify the degree of wear of the grinding wheel based on the trained adaptive neurofuzzy inference system.

[0108] The third embodiment of the present invention also provides a grinding wheel wear degree recognition device based on adaptive neurofuzzy reasoning, which includes a memory and a processor. The memory stores a computer program, which can be executed by the processor to realize the grinding wheel wear degree recognition method based on adaptive neurofuzzy reasoning as described above.

[0109] In the several embodiments provided in this invention, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus and method embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0110] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0111] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. It should be noted that, in this document, 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. In the absence of further restrictions, an element defined by the phrase "comprising a..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0112] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying the wear degree of grinding wheels based on adaptive neural fuzzy reasoning, characterized in that, include: S101, during the continuous grinding of the workpiece by the grinding wheel, acquire the grinding temperature signal, grinding force and surface quality of the workpiece at each grinding stage; S102, determine the wear stage of the grinding wheel based on the surface quality of the workpiece, grinding force and grinding temperature signals; S103, extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains; S104: Obtain the surface morphology image of the grinding wheel after each grinding. Use the HSV segmentation method to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel and calculate the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear. S105, with the proportion of grinding debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output, the constructed adaptive neurofuzzy inference system is trained to obtain the trained adaptive neurofuzzy inference system. S106, Identify the wear level of the grinding wheel based on the trained adaptive neuro-fuzzy inference system; wherein, step S103 includes: The range of noise signal in the grinding temperature signal is obtained, and the noise threshold is obtained; The temperature signals of multiple sets of the same wear stage are analyzed using a thermal pulse counting program, and the average number of thermal pulses greater than the noise threshold in the grinding temperature signal is taken as the dynamic effective number of abrasive particles; step S104 includes: Multiple shooting areas are evenly divided in the circumferential direction of the grinding wheel, and three shooting areas (front, middle, and rear) are divided in the axial direction. Multiple images of the grinding wheel surface morphology are obtained by taking pictures in the shooting areas using a camera. The surface topography image of the grinding wheel is converted from the RGB color space to the HSV color space to obtain an HSV image; Color feature extraction technology was used to obtain the H, S, and V channel information of HSV images, and histogram analysis was performed on each channel. Based on the results of histogram analysis, the threshold ranges for wear debris blockage and abrasive wear in the H, S, and V channels are determined, and a mask is created based on the threshold ranges to mark pixels on the HSV image that are within the specified HSV value range. The HSV image is filtered by applying the mask to obtain the total number of pixels N1 and N2 of the wear debris blockage and wear area. The wear debris percentage is calculated based on the total number of pixels N in the HSV image. The proportion of abrasive wear is Step S105 includes: A sample dataset is generated based on the wear debris ratio, abrasive wear ratio, and dynamic effective wear number, and the sample dataset is randomly divided into independent training dataset and test dataset. Define fuzzy sets: set 2 to 8 types of initial membership functions to be determined and 2 to 5 numbers of fuzzy subsets to be determined; Define the fuzzy system structure: Generate the fuzzy inference system using the grid partitioning method; Model Training: Set the number of iterations, and use a hybrid neural network algorithm of backpropagation and least squares to train the fuzzy inference system, so that the output of the fuzzy inference system continuously approaches the training results of the training dataset. When the preset conditions are met, the training stops, and an adaptive neural network fuzzy inference system model is obtained.

2. The grinding wheel wear degree recognition method based on adaptive neurofuzzy reasoning according to claim 1, characterized in that, In step S101, a type k thermocouple is used to measure the grinding temperature signal and grinding force, wherein the type k thermocouple is used to grind to 4~10 micrometers.

3. The grinding wheel wear degree recognition method based on adaptive neurofuzzy reasoning according to claim 1, characterized in that, In step S102, the wear stages of the grinding wheel are divided into initial, intermediate, severe, and dull stages.

4. The grinding wheel wear degree recognition method based on adaptive neurofuzzy reasoning according to claim 1, characterized in that: Also includes: Model optimization: The membership function and the number of fuzzy subsets are determined based on the root mean square error and the number of iterations in training and testing. Cross-validation is used to determine whether the training is overfitting. The formulas for calculating the root mean square error and cross-validation are as follows: Suppose there are k subsets, each of size k. There are a total of n samples, of which ； For the i-th subset, let its prediction result be... Then the root mean square error: = ； in It is the actual value of the j-th sample in the i-th subset. It is the predicted value of the j-th sample in the i-th subset; Finally, calculate the average RMSE: Average RMSE= ; This yields the model's training error and the mean square error of cross-validation. Training is stopped when the training error is less than 5%, and if the mean square error of cross-validation differs from the training error by less than 2.5%, the model is considered well-fitted; otherwise, it is considered overfitting.

5. A grinding wheel wear degree recognition device based on adaptive neuro-fuzzy reasoning, used to implement the grinding wheel wear degree recognition method based on adaptive neuro-fuzzy reasoning as described in any one of claims 1 to 4, characterized in that, include: The acquisition unit is used to acquire the grinding temperature signal, grinding force, and surface quality of the workpiece at each grinding stage during the continuous grinding process of the grinding wheel. The wear stage determination unit is used to determine the wear stage of the grinding wheel based on the surface quality of the workpiece, grinding force, and grinding temperature signals. The extraction unit is used to extract the total number of thermal pulses in the grinding temperature signal as the dynamic effective number of abrasive grains. The proportion acquisition unit is used to acquire the surface morphology image of the grinding wheel after each grinding. It uses the HSV segmentation method to identify and distinguish the grinding debris and abrasive wear in the surface morphology image of the grinding wheel and calculates the area ratio to obtain the proportion of grinding debris and the proportion of abrasive wear. The training unit is used to train the constructed adaptive neurofuzzy inference system with the proportion of wear debris, the proportion of abrasive wear, and the number of dynamic effective abrasive particles as inputs and the wear stage as output, so as to obtain the trained adaptive neurofuzzy inference system. The recognition unit is used to identify the wear level of the grinding wheel based on the trained adaptive neurofuzzy inference system.

Images