Grinding wheel grinding machine error compensation method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a grinding wheel grinding machine error compensation method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring a first process parameter of a grinding wheel grinding machine corresponding to a to-be-detected workpiece; inputting the process parameters into an error prediction model, and obtaining an error prediction result output by the error prediction model; the error prediction result is determined based on the prediction geometric error and the prediction thermal error; the error prediction model is generated by training a second process parameter of the grinding wheel grinding machine corresponding to the sample workpiece and a real geometric error and a real thermal error of the sample workpiece as a sample set; and error compensation is conducted on the relative position of the grinding wheel grinding machine according to the error prediction result, wherein the relative position represents the positions of the grinding wheel grinding machine and the workpiece to be measured. By adopting the method, the workpiece grinding accuracy of the grinding wheel grinding machine can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of grinder error analysis, and particularly to a method, device, computer device, computer-readable storage medium, and computer program product for compensating grinding wheel dust errors. Background Art

[0002] With the development of machine learning technology, machine learning models have emerged in the field of machine tools.

[0003] If geometric errors and thermal errors can be controlled simultaneously, the machining accuracy can be significantly improved. However, the mapping relationship between the geometric accuracy of the grinding wheel grinder and the machine tool errors is not clear, making it difficult to perform precise error compensation. When one of them is compensated while the other is not controlled, the error compensation effect is poor.

[0004] However, the current error compensation methods still have the problem of poor accuracy after compensation. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a method, device, computer device, computer-readable storage medium, and computer program product for compensating grinding wheel grinder errors that can improve accuracy.

[0006] In a first aspect, the present application provides a method for compensating grinding wheel grinder errors, the method comprising:

[0007] Obtaining first process parameters of a grinding wheel grinder corresponding to a workpiece to be measured;

[0008] Inputting the process parameters into an error prediction model, and obtaining an error prediction result output by the error prediction model; the error prediction result is determined based on a predicted geometric error and a predicted thermal error; the error prediction model is generated after being trained with a sample set including second process parameters of a grinding wheel grinder corresponding to a sample workpiece, the true geometric error, and the true thermal error of the sample workpiece;

[0009] Performing error compensation on the relative position of the grinding wheel grinder according to the error prediction result, where the relative position represents the position of the grinding wheel grinder and the workpiece to be measured.

[0010] In one embodiment, the way of training the error prediction model includes:

[0011] Obtaining, through an input layer, the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinder under different working conditions, and the second process parameters of each sample workpiece; wherein, the second process parameters include the grinding wheel speed, the feed rate, and the cutting depth;

[0012] Extract the eigenvalue of the second process parameter of each sample workpiece through the input layer to obtain a feature vector;

[0013] Determine a first to-be-output feature vector and a second to-be-output feature vector through the hidden layer according to the target weight matrix, the target bias matrix, and the feature vector; wherein, the first to-be-output feature vector is a vector indicating the predicted geometric error; the second to-be-output feature vector is a vector indicating the predicted thermal error;

[0014] Convert the first to-be-output feature vector and the second to-be-output feature vector into an error prediction result through the output layer;

[0015] Calculate the difference between the error prediction result and the true error through the loss function; wherein, the true error includes the true geometric error and the true thermal error; when the loss function meets the preset accuracy, obtain the error prediction model.

[0016] In one embodiment, the method for determining the target weight matrix and the target bias matrix includes:

[0017] Obtain an initial weight matrix and an initial bias matrix;

[0018] Optimize the initial weight matrix and the initial bias matrix through a genetic algorithm to obtain a target solution; wherein, the target solution includes the target weight matrix and the target bias matrix.

[0019] In one embodiment, the obtaining of the initial weight matrix and the initial bias matrix includes:

[0020] Obtain the mean of the Gaussian distribution, the standard deviation of the Gaussian distribution, and a constant initialization value;

[0021] Call a first function to determine the initial weight matrix according to the mean of the Gaussian distribution and the standard of the Gaussian distribution;

[0022] Call a second function to determine the initial bias matrix based on the constant initialization value.

[0023] In one embodiment, the optimizing of the initial weight matrix and the initial bias matrix through a genetic algorithm to obtain a target solution includes:

[0024] Encode the initial weight matrix to obtain an initial weight encoding; encode the initial bias matrix to obtain an initial bias encoding;

[0025] Randomly generate an initial solution set according to the initial weight encoding and the bias encoding, the initial solution set includes at least one initial solution, and each initial solution represents a set of weight matrix and bias matrix of the error prediction model;

[0026] Perform fitness calculation on each of the initial solutions through a fitness function, and perform selection, crossover, and mutation processing according to the calculation results to generate a new set of initial solutions; wherein, the fitness function represents the difference between the error prediction result output under the weight matrix and the bias matrix corresponding to each initial solution and the true error.

[0027] Until a preset condition is satisfied, use the new set of initial solutions that meet the preset condition as the candidate solution set.

[0028] Determine the target solution from the candidate solution set according to the selection rule.

[0029] In one embodiment, the update method of the target weight matrix includes:

[0030] Obtain the historical weight matrix, learning rate, and loss function.

[0031] Determine the target weight matrix according to the historical weight matrix, the learning rate, and the loss function.

[0032] The update method of the target bias matrix includes:

[0033] Obtain the historical bias matrix, learning rate, and loss function.

[0034] Determine the target bias matrix according to the bias matrix, the learning rate, and the loss function.

[0035] In a second aspect, the present application also provides a grinding wheel grinder error compensation device, and the device includes:

[0036] An acquisition module, configured to acquire first process parameters of a grinding wheel grinder corresponding to a workpiece to be measured.

[0037] A processing module, configured to input the process parameters into an error prediction model, and obtain an error prediction result output by the error prediction model; the error prediction result is determined based on a predicted geometric error and a predicted thermal error; the error prediction model is generated after being trained with second process parameters of a grinding wheel grinder corresponding to a sample workpiece, the true geometric error, and the true thermal error of the sample workpiece as a sample set.

[0038] A compensation module, configured to perform error compensation on the relative position of the grinding wheel grinder according to the error prediction result, where the relative position represents the position of the grinding wheel grinder and the workpiece to be measured.

[0039] In a third aspect, the present application also provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the above method are implemented.

[0040] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned method are implemented.

[0041] In a fifth aspect, the present application further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method are implemented.

[0042] For the above-mentioned grinding wheel grinder error compensation method, device, computer device, computer-readable storage medium and computer program product, first, obtain the first process parameters of the grinding wheel grinder corresponding to the workpiece to be measured; secondly, input the process parameters into the error prediction model, and obtain the error prediction result output by the error prediction model; the error prediction result is determined based on the predicted geometric error and the predicted thermal error; the error prediction model is generated after being trained with the second process parameters of the grinding wheel grinder corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as a sample set; the error prediction model takes into account both geometric error and thermal error factors and is generated after being trained with the true geometric error and the true thermal error of the sample workpiece as a sample set; finally, perform error compensation on the relative position of the grinding wheel grinder according to the error prediction result, and the error compensation is based on the compensation considering both geometric error and thermal error factors. Compared with only considering a single factor of geometric error or thermal error, the compensation effect is significant, thereby improving the accuracy of the grinding wheel grinder for grinding workpieces to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required to be used in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0044] Figure 1 It is an application environment diagram of the grinding wheel grinder error compensation method in an embodiment;

[0045] Figure 2 It is a flowchart of the grinding wheel grinder error compensation method in an embodiment;

[0046] Figure 3 It is a flowchart of the way of training the error prediction model in an embodiment;

[0047] Figure 4 It is a flowchart of the way of determining the target weight matrix and the target bias matrix in an embodiment;

[0048] Figure 5Schematic diagram of the process for obtaining the initial weight matrix and the initial bias matrix in an embodiment;

[0049] Figure 6 Schematic diagram of the process for obtaining the target solution through a genetic algorithm in an embodiment;

[0050] Figure 7 Schematic diagram of the process for updating the target weight matrix and the target bias matrix in an embodiment;

[0051] Figure 8 Structural block diagram of a grinding wheel grinder error compensation device in an embodiment;

[0052] Figure 9 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0053] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0054] The grinding wheel grinder error compensation method provided by the embodiments of the present application can be applied to an application environment as Figure 1 shown. Among them, the grinding wheel grinder 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or on other network servers. The server obtains the first process parameters of the grinding wheel grinder corresponding to the workpiece to be measured; inputs the process parameters into the error prediction model, and obtains the error prediction result output by the error prediction model; the error prediction result is determined based on the predicted geometric error and the predicted thermal error; the error prediction model is generated after being trained with the second process parameters of the grinding wheel grinder corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as a sample set; performs error compensation on the relative position of the grinding wheel grinder according to the error prediction result, where the relative position represents the position of the grinding wheel grinder and the workpiece to be measured. The server 104 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The grinding wheel grinder error compensation method involved in the embodiments of the present application can be applied to a grinding wheel grinder error compensation device, and the communication signal type recognition device can be the above-mentioned grinding wheel grinder.

[0055] In an exemplary embodiment, as Figure 2 shown, a grinding wheel grinder error compensation method is provided. Taking the method applied to the Figure 1 server as an example, it includes the following steps S202 to step S206. Among them:

[0056] Step S202: Obtain the first process parameters of the grinding wheel grinding machine corresponding to the workpiece to be measured.

[0057] Among them, the first process parameters are the process parameters for machining the workpiece to be measured, such as the grinding wheel speed, feed rate, cutting depth, material thermal expansion coefficient, etc.

[0058] Optionally, the server obtains the first process parameters of the grinding wheel grinding machine corresponding to the workpiece to be measured, as shown in Table 1.

[0059] Table 1 First process parameters of the workpiece to be measured

[0060]

[0061] Step S204: Input the first process parameters into the error prediction model, and obtain the error prediction result output by the error prediction model; the error prediction result is determined based on the predicted geometric error and the predicted thermal error; the error prediction model is generated after being trained with the second process parameters of the grinding wheel grinding machine corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as a sample set.

[0062] Among them, the error prediction model can be a neural network model.

[0063] Optionally, before inputting the first process parameters into the error prediction model, the server generates the initial error prediction model after training with the second process parameters of the grinding wheel grinding machine corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as samples.

[0064] Optionally, the server inputs the first process parameters into the trained error prediction model, and obtains the error prediction result output by the error prediction model. Among them, the error prediction model includes two output neurons, one neuron outputs the predicted geometric error, and the other neuron outputs the predicted thermal error, and the error prediction result is determined based on the predicted geometric error and the predicted thermal error.

[0065] Step S206: Perform error compensation on the relative position of the grinding wheel grinding machine according to the error prediction result.

[0066] Among them, the relative position characterizes the position of the grinding wheel grinding machine and the workpiece to be measured.

[0067] Optionally, the server performs error compensation on the relative position of the grinding wheel grinding machine according to the error prediction result, such as adjusting the controller parameters of the grinding wheel grinding machine to complete the error compensation.

[0068] In the above-mentioned error compensation method for a grinding wheel grinding machine, first, obtain the first process parameters of the grinding wheel grinding machine corresponding to the workpiece to be measured; secondly, input the process parameters into the error prediction model, and obtain the error prediction result output by the error prediction model; the error prediction result is determined based on the predicted geometric error and the predicted thermal error; the error prediction model is generated after training with the second process parameters of the grinding wheel grinding machine corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as a sample set; the error prediction model takes into account both geometric error and thermal error factors, and is generated after training with the true geometric error and the true thermal error of the sample workpiece as a sample set; finally, perform error compensation on the relative position of the grinding wheel grinding machine according to the error prediction result. The error compensation is based on the compensation considering both geometric error and thermal error factors. Compared with only considering a single factor of geometric error or thermal error, the compensation effect is significant, thereby improving the accuracy of the grinding wheel grinding machine for grinding workpieces to a certain extent.

[0069] Continuing from the above-mentioned embodiments, as Figure 3 shown, the method for training the error prediction model includes steps S302 to S310. Among them:

[0070] Step S302, obtain the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinding machine under different working conditions, and the second process parameters of each sample workpiece through the input layer.

[0071] Among them, the second process parameters are the parameters corresponding to the sample workpiece, including the grinding wheel speed, feed speed, cutting depth, material thermal expansion coefficient, etc.

[0072] The error prediction model includes an input layer, a hidden layer, and an output layer.

[0073] Optionally, collect the geometric error and thermal error data of the grinding wheel grinding machine under different working conditions, as well as the corresponding second process parameters (such as grinding wheel speed, feed speed, cutting depth, etc.) and temperature data. The server obtains the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinding machine under different working conditions, and the second process parameters of each sample workpiece. As shown in Table 2, input the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinding machine under different working conditions, and the second process parameters of each sample workpiece into the input layer.

[0074] Among them, for the measurement of the true geometric error, use high-precision measuring equipment (such as a laser rangefinder, a coordinate measuring machine, etc.) to measure the geometric accuracy of the machine tool. The measurement items include: the radial runout and axial runout of the grinding wheel spindle, the parallelism between the grinding wheel and the worktable, the coaxiality between the headstock and the tailstock, etc. Record the measurement data and calculate the magnitudes of various geometric errors.

[0075] Measurement of real thermal error: After the machine tool has run for a period of time, use temperature measurement equipment such as infrared thermometers or thermocouples to measure the temperature of key parts of the machine tool. The measurement locations include: the grinding area of the grinding wheel, the spindle bearing, the workbench guide rail, the coolant tank, etc. Record the temperature data and analyze the temperature distribution and change trend.

[0076] Record the machine tool parameters such as the current grinding wheel speed, feed speed, cutting depth, etc. of the grinding machine. When recording the measurement, external conditions such as ambient temperature and humidity can also be considered so as to take these factors into account when analyzing the data for subsequent error analysis.

[0077] Table 2 Second process parameters of the sample workpieces

[0078]

[0079] Step S304: Extract the eigenvalue of the second process parameter of each sample workpiece through the input layer to obtain the feature vector.

[0080] Optionally, the server extracts the eigenvalue of the second process parameter of each sample workpiece through the input layer to obtain the feature vector so that the hidden layer can perform mathematical operations on the feature vector.

[0081] Step S306: Determine the first to-be-output feature vector and the second to-be-output feature vector through the hidden layer according to the target weight matrix, the target bias matrix, and the feature vector.

[0082] Among them, the first to-be-output feature vector is the vector indicating the predicted geometric error; the second to-be-output feature vector is the vector indicating the predicted thermal error.

[0083] In practical applications, the hidden layer processes the feature vector through a series of complex mathematical operations (such as weighted summation and activation functions) to extract useful features. These features may include various higher-level representations related to the process. The relationship between the second process parameter, the real geometric error, and the real thermal error transformed into the feature vector is reflected through the weights and biases of the neural network and the nonlinear transformation of the activation function. These relationships are learned and optimized during the training process of the neural network to better predict and explain various phenomena in the process.

[0084] Optionally, the server performs a linear transformation on the feature vector of the input layer through the target weight matrix and the target bias matrix of the hidden layer to obtain the input value of each neuron. The input value after the linear transformation undergoes a nonlinear transformation through the activation function to obtain the output value of each neuron, that is, the first to-be-output feature vector and the second to-be-output feature vector.

[0085] Step S308: Transform the first to-be-output feature vector and the second to-be-output feature vector into the error prediction result through the output layer.

[0086] Optionally, the server jointly determines an error prediction result based on the first output feature vector and the second output feature vector through an output layer.

[0087] Step S310: Calculate the difference between the error prediction result and the true error through a loss function; when the loss function meets a preset accuracy, an error prediction model is obtained.

[0088] Wherein, the true error includes a true geometric error and a true thermal error.

[0089] Optionally, calculate the difference between the error prediction result and the true error through a loss function. A cross-entropy loss function can be used, as shown in formulas (1) and (2). When the loss function meets a preset accuracy, an error prediction model is obtained. When the loss function does not meet the preset accuracy, adjust the parameters until it is satisfied.

[0090]

[0091] In the formula, N is the number of samples, y i is the true error, is the predicted error.

[0092] In this embodiment, by training an error prediction model with the eigenvalue of the second process parameter of each sample workpiece and the true error, a trained error prediction model can be obtained, and the error prediction model is used for error prediction, improving the accuracy of error prediction.

[0093] In an exemplary embodiment, as Figure 4 shown, the method for determining the target weight matrix and the target bias matrix includes steps S402 to S404. Wherein:

[0094] Step S402: Obtain an initial weight matrix and an initial bias matrix.

[0095] Optionally, before obtaining the initial weight matrix and the initial bias matrix, first initialize the weight matrix and the bias matrix. Among them, the initialization of the weight matrix can be through random initialization. For example, it can be initialized through a Gaussian distribution (normal distribution): the weights are randomly sampled from a Gaussian distribution with a mean of zero and a standard deviation of σ. Commonly used are the standard normal distribution (mean of 0, standard deviation of 1) and the scaled normal distribution (mean of 0, standard deviation of 1 / √n, where n is the number of input neurons). It can also be initialized through a uniform distribution: the weights are randomly selected from a uniform distribution. For example, it can be uniformly distributed in the interval [-1,1], or uniformly distributed in the interval [-1 / √n,1 / √n], where n is the number of input neurons.

[0096] The initialization of the bias weight matrix can be done by initializing the bias weight matrix to 0 or using small random values to avoid some neurons from being inhibited in the initial stage of training.

[0097] Optionally, the server obtains the initialized weight matrix and bias matrix, that is, obtains the initial weight matrix and initial bias matrix.

[0098] Step S404, optimize the initial weight matrix and initial bias matrix through a genetic algorithm to obtain a target solution; where the target solution includes a target weight matrix and a target bias matrix.

[0099] Optionally, the server can optimize the initial weight matrix and initial bias matrix through a single-objective genetic algorithm or a multi-objective genetic algorithm to obtain a target solution; where the target solution includes a target weight matrix and a target bias matrix.

[0100] In this embodiment, through the genetic algorithm, the target weight matrix and target bias matrix of the error prediction model can be gradually optimized, so as to find a set of parameters that can better map the relationship between the input features (second process parameters) and the output targets (true geometric error and true thermal error), thereby improving the accuracy of the error prediction model.

[0101] In practical applications, the initialization of the initial weight matrix and initial bias matrix is not fixed. Obtaining the initial weight matrix and initial bias matrix may also include steps S502 to S506. Wherein:

[0102] Step S502, obtain the mean of the Gaussian distribution, the standard deviation of the Gaussian distribution, and the constant initialization value.

[0103] In practical applications, for example, initialize a weight matrix of shape (X, Y) for the error prediction model, and it is desired that the weights are drawn from a Gaussian distribution with a mean (μ) of 0 and a standard deviation (σ) of 0.01. This initialization method helps to break symmetry at the start of training and helps the neural network to learn better. Where X and Y are natural numbers greater than 0.

[0104] Optionally, the server obtains the mean (μ) of the Gaussian distribution and the standard deviation (σ) of the Gaussian distribution, and the constant initialization value, such as any natural number.

[0105] Step S504, call the first function to determine the initial weight matrix according to the mean of the Gaussian distribution and the standard of the Gaussian distribution.

[0106] Optionally, the server can call the first function np.random.normal to generate random numbers from the specified Gaussian distribution, and return the initial weight matrix according to the mean (μ) of the Gaussian distribution and the standard deviation (σ) of the Gaussian distribution.

[0107] Step S506: Call the second function to determine the initial bias matrix based on the constant initialization value.

[0108] Optionally, the server calls the second function bias_variable to return a bias matrix (or vector) with all elements initialized to the constant initialization value.

[0109] In this embodiment, initialization by Gaussian distribution (normal distribution) and small constant initialization can accelerate the convergence rate of the loss function gradient descent and reduce the training time.

[0110] In an exemplary embodiment, as Figure 6 shown, the initial weight matrix and the initial bias matrix are optimized by a genetic algorithm to obtain the target solution, including steps S602 to S610. Among them:

[0111] Step S602: Encode the initial weight matrix to obtain the initial weight encoding; encode the initial bias matrix to obtain the initial bias encoding.

[0112] Optionally, the server encodes the initial weight matrix to obtain the initial weight encoding; encodes the initial bias matrix to obtain the initial bias encoding. For example, real number encoding can be used to represent the initial weight matrix and the initial bias matrix as chromosomes (individuals) in the genetic algorithm.

[0113] Step S604: Randomly generate an initial solution set according to the initial weight encoding and the bias encoding.

[0114] Among them, the initial solution set includes at least one initial solution. Each initial solution represents a set of weight matrix and bias matrix of the error prediction model. Each initial solution is a chromosome individual.

[0115] Optionally, the server determines the optimization objective and fitness function of the target genetic algorithm according to the instruction. For example, the fitness function represents the difference between the error prediction result output under the weight matrix and bias matrix corresponding to each initial solution and the true error. The population is initialized according to the scores corresponding to each performance index by the target genetic algorithm, and the initial solution set is randomly generated. The server determines the gene encoding of each chromosome individual (initial solution). For example, bit strings or real number encoding are used to represent different configuration parameters.

[0116] Step S606: Perform adaptability calculation on each initial solution through the fitness function, and perform selection, crossover, and mutation processing according to the calculation results to generate a new initial solution set.

[0117] Among them, the fitness function characterizes the difference between the error prediction result output under the weight matrix and bias matrix corresponding to each initial solution and the true error. For example, the mean square error (MSE) is used as the measurement criterion of the fitness function. The smaller the fitness function is, the higher the fitness of the individual is.

[0118] Optionally, the server scores all the objectives of each chromosome individual (initial solution) according to the fitness function, and calculates the Pareto ranking and non-dominated rank of each chromosome individual (initial solution). According to the Pareto ranking and non-dominated rank of each chromosome individual (initial solution), and other selection pressures (such as crowding distance), excellent chromosome individuals (initial solutions) are selected to enter the next generation. The selected excellent chromosome individuals (initial solutions) are paired, and crossover operations are applied to generate new offspring chromosome individuals (initial solutions). Mutation operations are applied to the new offspring chromosome individuals (initial solutions) to introduce new genetic diversity and generate a new set of initial solutions.

[0119] Step S608, until a preset condition is met, the new set of initial solutions that meet the preset condition is used as the candidate solution set.

[0120] Among them, the preset conditions include a predetermined number of iterations or other termination conditions.

[0121] Optionally, the server performs fitness calculation on the new set of initial solutions through the fitness function, the Pareto ranking and non-dominated rank of each chromosome individual (initial solution), and selection, crossover and mutation processing are performed according to the Pareto ranking and non-dominated rank to generate a new set of initial solutions. Until a preset condition is met, such as whether the predetermined number of iterations is reached or whether other termination conditions are met (such as the change in the Pareto front is less than a certain threshold), the new set of initial solutions that meet the preset condition is used as the candidate solution set. Among them, the candidate solution set includes at least one candidate solution.

[0122] Step S610, determine the target solution from the candidate solution set according to the selection rule.

[0123] Optionally, the server determines a unique target solution from at least one candidate solution in the candidate solution set. Among them, the selection rule is determined according to the minimum error, such as the minimum variance.

[0124] In practical applications, when the preset number of iterations is reached or the population diversity is reduced to a certain extent, the optimal solution or a set of excellent solutions is terminated and output. The individual with the highest fitness value is extracted from the final population as the combination of the target weight matrix and target bias matrix of the optimal error prediction model.

[0125] Assume that the initial population size is 50 and the number of iterations is 100. Randomly generate 50 sets of weight and bias values of the neural network as the initial population, and calculate the fitness value of each individual using the training set data. Then, select excellent individuals according to the fitness value for crossover and mutation operations to generate a new population. After several generations of evolution, extract the individual with the highest fitness value as the optimal solution.

[0126] In this embodiment, optimizing the weight matrix and bias matrix through the genetic algorithm can improve the accuracy of the error prediction model.

[0127] In one embodiment, as Figure 7 shown, the update method of the target weight matrix includes steps S702 to S704. Among them:

[0128] Step S702, obtain the historical weight matrix, learning rate, and loss function.

[0129] Optionally, the server obtains the historical weight matrix W old 、learning rate lr, and loss function Loss.

[0130] Step S704, determine the target weight matrix according to the historical weight matrix, learning rate, and loss function.

[0131] Optionally, the server determines the target weight matrix W old according to the historical weight matrix W new as shown in formula (3).

[0132]

[0133] The update method of the target bias matrix includes steps S706 to S708. Among them:

[0134] Step S706, obtain the historical bias matrix, learning rate, and loss function.

[0135] Optionally, obtain the historical bias matrix B old 、learning rate lr, and loss function Loss.

[0136] Step S708, determine the target bias matrix according to the bias matrix, learning rate, and loss function.

[0137] Optionally, the server determines the target bias matrix B old according to the historical bias matrix B new as shown in formula (4).

[0138]

[0139] It should be noted that each neuron in the hidden layer is connected to the neurons in the input layer through a target weight matrix, and each neuron has a target bias matrix.

[0140] In this embodiment, the target weight matrix and the target bias matrix are determined according to the historical weight matrix, the historical bias matrix, the learning rate, and the loss function, thereby improving the accuracy of the error prediction model.

[0141] In an exemplary embodiment, the server obtains the first process parameters of the grinding wheel grinder corresponding to the workpiece to be measured, as shown in Table 1. Before inputting the first process parameters into the error prediction model, the server uses the second process parameters of the grinding wheel grinder corresponding to the sample workpiece, the true geometric error and the true thermal error of the sample workpiece as samples to train the initial error prediction model and then generates it. The server inputs the first process parameters into the trained error prediction model and obtains the error prediction result output by the error prediction model. Among them, the error prediction model includes two output neurons, one neuron outputs the predicted geometric error, and the other neuron outputs the predicted thermal error. The error prediction result is determined based on the predicted geometric error and the predicted thermal error. The server performs error compensation on the relative position of the grinding wheel grinder, such as adjusting the controller parameters of the grinding wheel grinder to complete the error compensation.

[0142] The method for training the error prediction model includes:

[0143] Collect the geometric error and thermal error data of the grinding wheel grinder under different working conditions, as well as the corresponding second process parameters (such as grinding wheel speed, feed speed, cutting depth, etc.) and temperature data. The server obtains the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinder under different working conditions, as well as the second process parameters of each sample workpiece, as shown in Table 2, and inputs the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinder under different working conditions, as well as the second process parameters of each sample workpiece into the input layer.

[0144] Among them, for the measurement of the true geometric error, a high-precision measurement device (such as a laser rangefinder, a coordinate measuring machine, etc.) is used to measure the geometric accuracy of the machine tool. The measurement items include: the radial runout and axial runout of the grinding wheel spindle, the parallelism between the grinding wheel and the workbench, the coaxiality between the headstock and the tailstock, etc. Record the measurement data and calculate the magnitudes of various geometric errors.

[0145] For the measurement of the true thermal error, after the machine tool has been running for a period of time, a temperature measurement device such as an infrared thermometer or a thermocouple is used to measure the temperature of the key parts of the machine tool. The measurement parts include: the grinding area of the grinding wheel, the spindle bearing, the workbench guide rail, the coolant tank, etc. Record the temperature data and analyze the temperature distribution and change trend.

[0146] Record the machine tool parameters such as the current settings of the grinding wheel speed, feed rate, cutting depth, etc. of the grinding wheel grinding machine. When recording the measurement, external conditions such as ambient temperature and humidity can also be considered to take into account the influence of these factors on the error during subsequent data analysis. The server extracts the eigenvalue of the second process parameter of each sample workpiece through the input layer to obtain a feature vector for the hidden layer to perform mathematical operations on the feature vector. In practical applications, the hidden layer processes the feature vector through a series of complex mathematical operations (such as weighted summation and activation functions) to extract useful features. These features may include various higher-level representations related to the process. The relationship between the second process parameter converted into a feature vector, the true geometric error, and the true thermal error is reflected through the weights and biases of the neural network and the non-linear transformation of the activation function. These relationships are learned and optimized during the training process of the neural network to better predict and explain various phenomena in the process. The server performs a linear transformation on the feature vector of the input layer through the target weight matrix and the target bias matrix of the hidden layer to obtain the input value of each neuron. The input value after the linear transformation undergoes a non-linear transformation through the activation function to obtain the output value of each neuron, that is, the first to-be-output feature vector and the second to-be-output feature vector. The server determines the error prediction result through the output layer based on the first to-be-output feature vector and the second to-be-output feature vector. Calculate the difference between the error prediction result and the true error through the loss function. The cross-entropy loss function can be used, as shown in formulas (1) and (2). When the loss function meets the preset accuracy, the error prediction model is obtained. When the loss function does not meet the preset accuracy, adjust the parameters until it is satisfied.

[0147]

[0148] In the formula, N is the number of samples, y i is the true error, is the predicted error.

[0149] In practical applications, for example, initialize a weight matrix of shape (X, Y) for an error prediction model, and it is desired that the weights are drawn from a Gaussian distribution with a mean (μ) of 0 and a standard deviation (σ) of 0.01. This initialization method helps to break symmetry at the start of training and helps the neural network learn better. Here, X and Y are natural numbers greater than 0. The server obtains the mean (μ) of the Gaussian distribution and the standard deviation (σ) of the Gaussian distribution, and a constant initialization value, such as any natural number. Random numbers can be generated from the specified Gaussian distribution by calling the first function np.random.normal. The server can call the first function np.random.normal to generate random numbers from the specified Gaussian distribution and return the initial weight matrix according to the mean (μ) of the Gaussian distribution and the standard deviation (σ) of the Gaussian distribution. Call the second function bias_variable to return a bias matrix (or vector) with all elements initialized to the constant initialization value.

[0150] The server encodes the initial weight matrix to obtain the initial weight encoding; encodes the initial bias matrix to obtain the initial bias encoding. For example, real number encoding can be used to represent the initial weight matrix and the initial bias matrix as chromosomes (individuals) in a genetic algorithm. The server determines the optimization objective and fitness function of the target genetic algorithm according to the instruction. For example, the fitness function characterizes the difference between the error prediction result output under the weight matrix and bias matrix corresponding to each initial solution and the true error. The population is initialized according to the scores corresponding to each performance index by the target genetic algorithm, and an initial solution set is randomly generated. The server determines the gene encoding of each chromosome individual (initial solution). For example, bit strings or real number encoding are used to represent different configuration parameters. Among them, the fitness function characterizes the difference between the error prediction result output under the weight matrix and bias matrix corresponding to each initial solution and the true error. For example, the mean square error (MSE) is used as the measurement criterion of the fitness function. The smaller the fitness function is, the higher the fitness of the individual is. The server scores all the objectives of each chromosome individual (initial solution) according to the fitness function, and calculates the Pareto ranking and non-dominated rank of each chromosome individual (initial solution). According to the Pareto ranking and non-dominated rank of each chromosome individual (initial solution), as well as other selection pressures (such as crowding distance), excellent chromosome individuals (initial solutions) are selected to enter the next generation. The selected excellent chromosome individuals (initial solutions) are paired, and crossover operations are applied to generate new offspring chromosome individuals (initial solutions). Mutation operations are applied to the new offspring chromosome individuals (initial solutions) to introduce new genetic diversity and generate a new initial solution set. The server performs fitness calculation on the new initial solution set through the fitness function, the Pareto ranking and non-dominated rank of each chromosome individual (initial solution), and selection, crossover, and mutation processing are performed according to the Pareto ranking and non-dominated rank to generate a new initial solution set. Until the preset conditions are met, such as whether the predetermined number of iterations is reached or whether other termination conditions are satisfied (such as the change in the Pareto front is less than a certain threshold), the new initial solution set that meets the preset conditions is used as the candidate solution set. Among them, the candidate solution set includes at least one candidate solution. The server determines the unique target solution from at least one candidate solution in the candidate solution set according to the selection rule. Among them, the selection rule is determined according to the minimum error, such as the minimum variance. Assume that the size of the initial population is 50 and the number of iterations is 100. 50 groups of weight and bias values of the neural network are randomly generated as the initial population, and the fitness value of each individual is calculated using the training set data. Then, excellent individuals are selected according to the fitness value for crossover and mutation operations to generate a new population. After several generations of evolution, the individual with the highest fitness value is extracted as the optimal solution.

[0151] The update method of the target weight matrix includes: the server obtains the historical weight matrix W old, learning rate lr and loss function Loss. The server determines the target weight matrix W based on the historical weight matrix W old , learning rate lr and loss function Loss, and determines the target weight matrix W new , as shown in formula (3).

[0152]

[0153] The update method of the target bias matrix includes: obtaining the historical bias matrix B old , learning rate lr and loss function Loss. The server determines the target bias matrix B based on the historical bias matrix B old , learning rate lr and loss function Loss, and determines the target bias matrix B new , as shown in formula (4).

[0154]

[0155] It should be noted that each neuron in the hidden layer is connected to the neurons in the input layer through the target weight matrix, and each neuron has a target bias matrix.

[0156] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the indications of the arrows, these steps do not necessarily have to be executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily have to be executed at the same time, but can be executed at different times. The execution order of these steps or stages does not necessarily have to be sequential, but can be executed alternately or alternately with at least a part of other steps or steps in other steps.

[0157] Based on the same inventive concept, the embodiments of the present application also provide a grinding wheel grinding machine error compensation device for implementing the above-mentioned grinding wheel grinding machine error compensation method. The implementation solutions provided by this device to solve problems are similar to the implementation solutions described in the above method. Therefore, the specific limitations in one or more embodiments of the grinding wheel grinding machine error compensation device provided below can refer to the limitations on the grinding wheel grinding machine error compensation method in the above text, and will not be repeated here.

[0158] In an exemplary embodiment, as Figure 8 shown, a grinding wheel grinding machine error compensation device is provided, including: an acquisition module 801, a processing module 802, and a compensation module 803, where:

[0159] An acquisition module 801, configured to acquire first process parameters of a grinding wheel grinding machine corresponding to a workpiece to be measured.

[0160] A processing module 802, configured to input the process parameters into an error prediction model, and acquire an error prediction result output by the error prediction model; the error prediction result is determined based on a predicted geometric error and a predicted thermal error; the error prediction model is generated after being trained with second process parameters of a grinding wheel grinding machine corresponding to a sample workpiece, the true geometric error, and the true thermal error of the sample workpiece as a sample set.

[0161] A compensation module 803, configured to perform error compensation on the relative position of the grinding wheel grinding machine according to the error prediction result, where the relative position represents the position of the grinding wheel grinding machine and the workpiece to be measured.

[0162] In an exemplary embodiment, a grinding wheel grinding machine error compensation device further includes: a training module, configured to acquire the true geometric error and the true thermal error of each sample workpiece of the grinding wheel grinding machine under different working conditions, and the second process parameters of each sample workpiece through an input layer; where the second process parameters include a grinding wheel rotation speed, a feed speed, and a cutting depth; extract eigenvalue of the second process parameters of each sample workpiece through the input layer to obtain a feature vector; determine a first to-be-output feature vector and a second to-be-output feature vector through a hidden layer according to a target weight matrix, a target bias matrix, and the feature vector; where the first to-be-output feature vector is a vector indicating a predicted geometric error; the second to-be-output feature vector is a vector indicating a predicted thermal error; convert the first to-be-output feature vector and the second to-be-output feature vector into an error prediction result through an output layer; calculate a difference between the error prediction result and a true error through a loss function; where the true error includes a true geometric error and a true thermal error; and obtain an error prediction model when the loss function meets a preset accuracy.

[0163] In an exemplary embodiment, a grinding wheel grinding machine error compensation device further includes: a matrix determination module, configured to acquire an initial weight matrix and an initial bias matrix; optimize the initial weight matrix and the initial bias matrix through a genetic algorithm to obtain a target solution; where the target solution includes a target weight matrix and a target bias matrix.

[0164] In an exemplary embodiment, the matrix determination module includes: an acquisition and determination unit, configured to acquire a mean of a Gaussian distribution, a standard deviation of the Gaussian distribution, and a constant initialization value; call a first function to determine the initial weight matrix according to the mean of the Gaussian distribution and the standard of the Gaussian distribution; and call a second function to determine the initial bias matrix based on the constant initialization value.

[0165] In an exemplary embodiment, the matrix determination module includes: a matrix determination unit configured to encode an initial weight matrix to obtain an initial weight encoding; encode an initial bias matrix to obtain an initial bias encoding; randomly generate an initial solution set according to the initial weight encoding and the bias encoding, the initial solution set including at least one initial solution, and each initial solution representing a set of weight matrices and bias matrices of an error prediction model; perform fitness calculation on each initial solution through a fitness function, and perform selection, crossover, and mutation processing according to the calculation results to generate a new initial solution set; wherein the fitness function represents the difference between the error prediction result output under the weight matrix and bias matrix corresponding to each initial solution and the true error; until a preset condition is met, use the new initial solution set that meets the preset condition as a candidate solution set; determine a target solution from the candidate solution set according to a selection rule.

[0166] In an exemplary embodiment, a grinding wheel grinder error compensation device further includes: an update module configured to obtain a historical weight matrix, a learning rate, and a loss function; determine a target weight matrix according to the historical weight matrix, the learning rate, and the loss function; obtain a historical bias matrix, the learning rate, and the loss function; determine a target bias matrix according to the bias matrix, the learning rate, and the loss function.

[0167] Each module in the above grinding wheel grinder error compensation device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0168] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as Figure 9 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store error data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. The computer program, when executed by the processor, implements a grinding wheel grinder error compensation method.

[0169] Those skilled in the art can understand that Figure 9 the structure shown in Figure 9 is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0170] In one embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0171] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0172] In one embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0173] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0174] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in the present application.

[0175] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for compensating a grinding wheel grinder error, characterized in that: The method comprises: Acquire a first process parameter of a grinding wheel grinder corresponding to the workpiece to be measured; The first process parameter is input into an error prediction model, and an error prediction result output by the error prediction model is obtained; the error prediction result is determined based on a predicted geometric error and a predicted thermal error; the error prediction model is generated after training with a second process parameter of a grinding wheel grinder corresponding to a sample workpiece, and a true geometric error and a true thermal error of the sample workpiece as a sample set; The relative position of the grinding wheel grinder is error compensated according to the error prediction result, wherein the relative position represents the position of the grinding wheel grinder and the workpiece to be measured.

2. The method according to claim 1, characterized in that: The error prediction model training method includes: Acquire the true geometric error and the true thermal error of each of the sample workpieces of the grinding wheel grinder under different working conditions, and the second process parameters of each of the sample workpieces through the input layer; wherein the second process parameters include grinding wheel rotation speed, feed speed, and cutting depth; Extracting a characteristic value of the second process parameter of each sample workpiece through the input layer to obtain a characteristic vector; Determine a first feature vector to be output and a second feature vector to be output according to the target weight matrix, the target bias matrix and the feature vector through a hidden layer; wherein the first feature vector to be output indicates a vector of predicted geometric errors; and the second feature vector to be output indicates a vector of predicted thermal errors; Converting the first feature vector to be output and the second feature vector to be output into an error prediction result through an output layer; The difference between the error prediction result and the true error is calculated by a loss function; wherein the true error includes the true geometric error and the true thermal error; and when the loss function satisfies a preset accuracy, the error prediction model is obtained.

3. The method according to claim 2, characterized in that The method for determining the target weight matrix and the target bias matrix includes: Get the initial weight matrix and initial bias matrix; The initial weight matrix and the initial bias matrix are optimized by a genetic algorithm to obtain a target solution; wherein the target solution includes the target weight matrix and the target bias matrix.

4. The method according to claim 3, characterized in that The obtaining of the initial weight matrix and the initial bias matrix comprises: Get the mean of the Gaussian distribution, the standard deviation of the Gaussian distribution, and the constant initialization value; Calling a first function to determine the initial weight matrix according to the mean of the Gaussian distribution and the standard of the Gaussian distribution; A second function is called to determine the initial bias matrix based on the constant initialization value.

5. The method according to claim 3, characterized in that: The step of optimizing the initial weight matrix and the initial bias matrix by a genetic algorithm to obtain a target solution includes: Encoding the initial weight matrix to obtain an initial weight code; encoding the initial bias matrix to obtain an initial bias code; Randomly generate an initial solution set according to the initial weight code and the bias code, the initial solution set includes at least one initial solution, each of the initial solutions represents a set of weight matrices and bias matrices of the error prediction model; Performing fitness calculation on each of the initial solutions through a fitness function, and performing selection, crossover and mutation processing according to the calculation results to generate a new set of initial solutions; wherein the fitness function represents the difference between the error prediction result and the true error output under the weight matrix and the bias matrix corresponding to each of the initial solutions; Until the preset conditions are met, a new initial solution set that meets the preset conditions is used as a candidate solution set; The target solution is determined from the candidate solution set according to a selection rule.

6. The method according to any one of claims 2 to 5, characterized in that: The target weight matrix is ​​updated in the following manner: Get the historical weight matrix, learning rate and loss function; Determining the target weight matrix according to the historical weight matrix, the learning rate and the loss function; The target bias matrix is ​​updated by: Get the historical bias matrix, learning rate and loss function; The target bias matrix is ​​determined according to the bias matrix, the learning rate, and the loss function.

7. An error compensation device for a grinding wheel grinder, characterized in that: The device comprises: An acquisition module, used for acquiring a first process parameter of a grinding wheel grinder corresponding to a workpiece to be tested; A processing module, used for inputting the process parameters into an error prediction model, and obtaining an error prediction result output by the error prediction model; the error prediction result is determined based on a predicted geometric error and a predicted thermal error; the error prediction model is generated after training with a second process parameter of a grinding wheel grinder corresponding to a sample workpiece, and a true geometric error and a true thermal error of the sample workpiece as a sample set; A compensation module is used to perform error compensation on the relative position of the grinding wheel grinder according to the error prediction result, wherein the relative position represents the position of the grinding wheel grinder and the workpiece to be measured.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images