A tool path compensation method, medium and program product for finishing of integral blade disk

Abstract

The present invention discloses a tool path compensation method, medium and program product for finishing of an integral blade disk, and relates to the technical field of tool path compensation, including: collecting historical integral blade disk processing tool data, establishing a tool wear prediction model, estimating the wear prediction value of the tool in different time periods according to the established wear model, fitting the wear prediction value with time, and drawing a tool wear time curve; according to the influence of tool wear on the finishing of the integral blade disk, combined with the tool wear time curve, dividing the tool wear into several time periods, and establishing a processing compensation model for different time periods, and at the same time performing blade grouping on the current integral blade disk. The present invention ensures that the processing accuracy and efficiency of the integral blade disk can be effectively controlled and improved at different stages of tool wear by accurately predicting and compensating tool wear, correcting tool compensation in combination with product detection results, and reasonably dividing the processing time periods.

Description

Technical Field

[0001] The present invention belongs to the field of mechanical manufacturing and precision machining technology, and relates to tool path compensation and integral blade disc finishing technology. Specifically, it relates to an integral blade disc finishing tool path compensation method, medium and program product for improving the machining accuracy, consistency and efficiency of integral blade disc blades. Background Art

[0002] As the core component of advanced turbomachinery, the machining accuracy and surface quality of the blisk directly affect its aerodynamic performance and service life. The machining of the blisk is usually divided into rough machining, semi-finishing and finishing stages. Among them, finishing is usually the machining of the blade part of the blisk to ensure that the blade size, shape and surface quality of the blisk meet the high-precision size and surface roughness required by the drawing. The finishing tool is the tool used in the finishing stage. It is a tool that performs fine cutting on the workpiece after removing a large amount of excess material in rough machining to further improve the size accuracy, shape accuracy and surface quality of the workpiece. However, the tool will wear to varying degrees after long-term use. During the milling process, the wear of the tool will cause machining errors (such as feed error, cutting depth error, etc.), and even unqualified machining quality. Therefore, how to compensate the tool error to maintain the machining accuracy as much as possible has always been an important technical issue in the field of blisk manufacturing.

[0003] Existing error compensation is often a dynamic compensation performed during the tool processing of a blade, which is used to adjust the processing path or parameters to compensate for the error caused by tool wear. For example, some methods use sensors to measure tool wear in real time and feed back the measurement value to the CNC system to dynamically adjust the tool path. However, the accuracy of this method is limited by the measurement accuracy and real-time performance of the sensor. Especially in cases of high-speed cutting and complex surface processing, it is difficult to accurately measure the tool wear, thereby limiting the accuracy of compensation. In addition, since the tool maintenance cycle is different from the processing time of a blade, multiple blades may be processed during one tool processing, resulting in difficulty in tracing the source after tool compensation problems occur. At the same time, under different wear thresholds, the required compensation model is also different. For example, in the early stage of tool wear, its impact on machining accuracy is small, and a simple compensation strategy can be adopted. In the later stage of tool wear, its impact on machining accuracy is greater, and a more complex compensation strategy is required. The existing tool path compensation methods often fail to distinguish between different wear stages of the tool. The use of a unified compensation strategy can easily lead to reduced machining consistency of the entire blade disc, and the path compensation cannot be corrected in time through the entire blade disc processing product, and the path compensation cannot be corrected according to the detection results. There are problems with low practicality and functionality.

[0004] Taking the “A tool cutting path processing error compensation method based on online monitoring” disclosed in the Chinese invention patent CN115562161B as an example, this method calculates the transient error of the system architecture, the theoretical error of the workpiece and the tool wear error through online monitoring of force, vibration and other signals, and compensates for them. However, this method mainly focuses on the processing error compensation of a single part, and does not consider the consistency problem between multiple blades in the processing of the integral blade disk. In addition, the method has a relatively simple compensation strategy for tool wear, does not adopt different compensation models for different wear stages, and does not consider the grouping of the blades of the integral blade disk and the division of the processing area, which makes it difficult to meet the high consistency requirements of the integral blade disk finishing.

[0005] In summary, the existing tool path compensation methods have many limitations when applied to the finishing of blisks, such as difficulty in tracing, single compensation strategy, lack of feedback correction mechanism, etc., which make it difficult to meet the processing requirements of high precision, high consistency and high efficiency. No effective solutions have been proposed for the problems in related technologies. Therefore, it is a technical problem to be solved urgently to develop a tool path compensation method that can effectively solve these problems and improve the machining quality of blisk blades. Summary of the invention

[0006] 1. Purpose of the invention

[0007] In view of the above defects and deficiencies in the prior art, and in order to solve at least one of the above and other technical problems in the prior art, the present invention proposes a tool path compensation method, medium and program product for finishing of an integral blade disk, which achieves accurate prediction and compensation of tool wear, improves the consistency of blade processing, and facilitates problem tracing by establishing a tool wear prediction model, dividing the wear time period and establishing a corresponding compensation model, grouping and isochronous processing of blades, and feedback correction of the compensation model based on the measurement results. At the same time, the present invention can dynamically adjust the processing parameters according to the tool wear conditions, ensuring that the processing accuracy and efficiency of the integral blade disk can be effectively controlled and improved at different stages of tool wear, and finally obtaining a high-quality integral blade disk product.

[0008] (II) Technical solution

[0009] In order to achieve the purpose of the invention and solve the technical problems, the present invention adopts the following technical solutions:

[0010] The first invention object of the present invention is to provide a tool path compensation method for finishing an integral blisk, which is used to improve the dimensional accuracy, shape accuracy and surface quality of the integral blisk blade processing, solve the processing error problem caused by tool wear, and ensure processing consistency and processing efficiency. The method at least includes the following steps when implemented:

[0011] SS1. Collect historical processing data and establish a tool wear prediction model

[0012] Collect multiple sets of historical tool data for processing integral blade disks of the same material. Each set of historical tool data includes at least tool wear value, normal load, cutting speed, processing time and workpiece material hardness parameters. Calculate the tool wear coefficient of each group and average the tool wear coefficients of multiple groups. Combined with the processing parameters, establish the current tool wear prediction model. Then, estimate the wear prediction value of the current tool in different time periods based on the current tool wear prediction model. At the same time, fit the wear prediction value with time to obtain the wear time prediction curve of the current tool. The algorithm formula of the current tool wear prediction model is expressed as follows:

[0013]

[0014] in, V is the wear prediction value of the current tool, represents the mean of multiple groups of tool wear coefficients, W , v , t Respectively represent the normal load, cutting speed and processing time of the current tool, H Indicates the hardness of the workpiece material;

[0015] SS2. Divide tool wear time periods and establish a machining compensation model

[0016] According to the influence of tool wear on the finishing of the integral blade disk, combined with the current tool wear time curve fitted in step SS1, the tool wear is divided into several wear time periods, and the historical processing error data in different wear time periods are collected. Based on the feed error and cutting depth error in the historical processing error data, the processing compensation model in different wear time periods is obtained through convolutional neural network training. The processing compensation model is used to adjust the processing parameters to compensate for the influence of tool wear on the processing accuracy in different wear time periods, and then the blades of the currently processed integral blade disk are grouped;

[0017] SS3. Implement isochronous machining and record machining areas

[0018] Based on the wear time periods divided in step SS2 and the blade grouping result of the integral blade disk, tool processing is performed in a manner of wear time period by wear time period and blade group by blade group, a corresponding blade group processing task is assigned to each wear time period, and each tool wear time period is divided isochronously based on the number of blades in the corresponding blade group, and wear time period processing is performed based on the grouping order of the blades of the integral blade disk, and the same processing area of ​​the blades of the same blade group isochronously processed in sequence in each wear time period, and tool path compensation is performed by matching the processing compensation model of the corresponding wear time period;

[0019] SS4. Detect machining errors and correct compensation models

[0020] After finishing, the machining parameters of different areas of the blisk blade are measured, and standard threshold comparison is performed based on different divided areas. When the machining parameters of a divided area exceed the standard threshold, the current machining compensation model corresponding to the divided area is corrected based on the part deviation between the measured value and the standard threshold. If the machining parameters are not greater than the standard threshold, there is no need to correct the machining compensation model.

[0021] A second inventive object of the present invention is to provide a computer program product, comprising computer instructions for executing the above-mentioned integral blisk finishing tool path compensation method.

[0022] The third inventive object of the present invention is to provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned integral blade finishing tool path compensation method is implemented.

[0023] (III) Technical Effect

[0024] Compared with the prior art, the tool path compensation method, medium and program product of the present invention for finishing the integral blade disk have the following beneficial and significant technical effects:

[0025] (1) The present invention establishes a wear prediction model for the tool and divides the tool processing time into time periods in combination with the wear threshold, thereby matching different processing compensation models for different processing time periods. The corresponding processing compensation model automatically adjusts the processing parameters to control the surface roughness back within the qualified range, thereby effectively ensuring the dimensional accuracy and surface quality of each part of the integral blade disk;

[0026] (2) The present invention groups and pairs the blades of the blisk being processed, and divides the different wear time periods of the current processing tool into equal parts based on the number of blades in each group. The blades in the same group are processed in sequence and isochronously based on the processing time after equal division, thereby enhancing the consistency of the processing of the blisk blades and improving the quality uniformity of the blade part of the entire blisk.

[0027] (3) The present invention records and marks the blade areas in a group processed within the same wear time period, and divides the processing of each blade in a group into clear time intervals and corresponding tool wear stages. Once a processing quality problem occurs, it can quickly and accurately locate which group of blades was processed and which wear time period of the corresponding tool the problem occurred in, which facilitates subsequent problem tracing and enhances functionality.

[0028] (4) The present invention performs isochronous machining according to the blade grouping and the tool wear time period being equally divided, which can ensure that the blades in the same group are machined at approximately the same tool wear stage and using the same compensation strategy, so that the machining conditions faced by each blade are more uniform, thereby effectively improving the consistency of blade machining in the same group and better meeting the requirements of the drawings on blade size, shape and surface quality.

[0029] (5) The present invention predicts and compensates for tool wear, and reasonably divides different blade processing time periods according to different tool wear time pairs, thereby ensuring that the processing accuracy and efficiency of the entire blade disk can be effectively controlled and improved at different stages of tool wear. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0031] Figure 1 is a flow chart of a tool path compensation method for finishing a blisk according to an embodiment of the present invention;

[0032] Figure 2 is an implementation flow chart of step SS1 in an embodiment of the present invention;

[0033] Figure 3 It is an implementation flow chart of step SS2 in the embodiment of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solutions and advantages of the implementation of the present invention clearer, the embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be understood as limiting the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. According to an embodiment of the present invention, a tool path compensation method for finishing an integral blisk is provided, which aims to improve the dimensional accuracy, shape accuracy and surface quality of the machining of the blades of the integral blisk, solve the problem of machining errors caused by tool wear, and ensure machining consistency and machining efficiency. The present invention is further described in conjunction with the accompanying drawings and specific embodiments:

[0035] Example 1

[0036] like Figure 1 As shown, the tool path compensation method for finishing of an integral blade disk provided by an embodiment of the present invention mainly includes the following steps when implemented:

[0037] SS1. Collect historical processing data and establish a tool wear prediction model

[0038] Collect multiple sets of historical tool data for machining integral blades of the same material. Each set of historical tool data includes at least tool wear value, normal load, cutting speed, machining time and workpiece material hardness parameters. Calculate the tool wear coefficient of each group and average the tool wear coefficients of multiple groups. Combined with the machining parameters, establish the current tool wear prediction model. Then, estimate the wear prediction value of the current tool in different time periods based on the current tool wear prediction model. At the same time, fit the wear prediction value with time to obtain the wear time prediction curve of the current tool. The algorithm formula of the current tool wear prediction model is expressed as follows:

[0039]

[0040] in, V is the wear prediction value of the current tool, represents the mean of multiple groups of tool wear coefficients, W , v , t Respectively represent the normal load, cutting speed and processing time of the current tool, H Indicates the hardness of the workpiece material.

[0041] SS2. Divide tool wear time periods and establish a machining compensation model

[0042] According to the influence of tool wear on the finishing of the integral blade disk, combined with the current tool wear time curve fitted and drawn in step SS1, the tool wear is divided into several wear time periods, and the historical processing error data under different wear time periods are collected. Based on the feed error and cutting depth error in the historical processing error data, the processing compensation model under different wear time periods is obtained through convolutional neural network training. The processing compensation model is used to adjust the processing parameters to compensate for the influence of tool wear on the processing accuracy under different wear time periods, and then the blades of the currently processed integral blade disk are grouped.

[0043] SS3. Implement isochronous machining and record machining areas

[0044] Based on the wear time periods divided in step SS2 and the blade grouping results of the entire impeller, tool processing is performed in a wear time period-by-wear time period and blade group-by-blade manner, a corresponding blade group processing task is assigned to each wear time period, and each tool wear time period is divided isochronously based on the number of blades in the corresponding blade group. Processing is performed in each wear time period based on the grouping order of the blades of the entire impeller, and in each wear time period, the same processing areas of the blades of the same blade group are sequentially isochronously processed, and tool path compensation is performed by matching the processing compensation model of the corresponding wear time period.

[0045] SS4. Detect machining errors and correct compensation models

[0046] After finishing, the machining parameters of different areas of the blisk blade are measured, and standard threshold comparison is performed based on different divided areas. When the machining parameters of a divided area exceed the standard threshold, the current machining compensation model corresponding to the divided area is corrected based on the part deviation between the measured value and the standard threshold. If the machining parameters are not greater than the standard threshold, there is no need to correct the machining compensation model.

[0047] As a preferred embodiment, the above-mentioned integral blade disk finishing tool path compensation method of the present invention also includes step SS5 of problem tracing and processing consistency control, which realizes rapid tracing of processing problems through the processing time period and tool wear stage information recorded in the blade file, and at the same time, optimizes the blade grouping strategy and processing path to improve the processing consistency and quality uniformity of the integral blade disk in response to errors occurring during the processing.

[0048] Through the above steps, the present invention can achieve accurate prediction and real-time compensation of tool wear, and significantly improve the accuracy, consistency and efficiency of overall blade disk processing.

[0049] Embodiment 2:

[0050] Based on the above embodiment 1, as a further optimization of its content, this embodiment 2 provides a preferred implementation method, such as Figure 2 As shown, step SS1 includes at least the following sub-steps when constructing the wear prediction model of the current tool:

[0051] SS11. Collect historical blisk machining tool data

[0052] collect N The tool data for machining the blisk of the same material in the group history, including N is an integer and satisfies N ≥2, the collected historical tool data are derived from actual processing records, processing experiments or numerical simulations during the processing, and each set of historical tool data includes at least tool wear value, normal load, cutting speed, processing time and workpiece material hardness, where the tool wear value represents the volume or value of tool wear during the processing, the normal load represents the force on the tool in the vertical direction of the processing surface, and the cutting speed represents the relative movement speed between the tool and the workpiece.

[0053] SS12. Calculate tool wear coefficient and average it

[0054] Based on the collected historical tool data, the algorithm formula Calculate the tool wear coefficient corresponding to each set of historical tool data one by one, where k i For the i Tool wear coefficient corresponding to the historical tool data of the group, V i , W i , v i , t i , H i Respectively represent i Tool wear values, normal load, cutting speed, machining time and workpiece material hardness in the group historical tool data, i is the number of each history tool data set and i =1,2,…, N , then based on the algorithm formula The tool wear coefficient corresponding to each group of data is averaged to obtain the mean value of the tool wear coefficient , in order to eliminate random errors and individual differences in the data.

[0055] SS13. Establish current tool wear prediction model

[0056] Mean value of tool wear coefficient after averaging , combined with the current tool processing parameters, establish the current tool wear prediction model, the wear prediction model algorithm formula is as follows:

[0057]

[0058] in, V Represents the wear prediction value of the current tool, W , v , t Respectively represent the normal load, cutting speed and processing time under the current tool processing conditions, H Indicates the hardness of the workpiece material.

[0059] Further preferably, step SS1 includes at least the following sub-steps when fitting and drawing the wear time curve of the current tool:

[0060] SS14. Divide tool processing maintenance cycle

[0061] Determine the current blisk finishing tool maintenance cycle T , using arithmetic progression to divide it into multiple equal time intervals, each time interval is Δt , from the first time interval point t Starting from 1, increase the same time interval successively Δt, until the last time interval point t n = T End, get the time series { t 1,.., t j , ..., t n},in t j For the j time intervals and j =1,2,…, n , n Maintenance cycle for tool processing T The number of time intervals to be divided.

[0062] SS15. Calculate the tool wear prediction value at each moment

[0063] Based on each time interval t j And the normal load corresponding to that moment W j , cutting speed v j , Processing time t ' j and workpiece material hardness H , through the wear prediction model of the current tool established in sub-step SS13, the tool wear prediction value corresponding to the current interval is calculated respectively:

[0064]

[0065] in, V j Time interval point t j Corresponding tool wear prediction value, processing time t ' j satisfy t ' j = t j .

[0066] SS16. Fitting the relationship between tool wear and machining time

[0067] The least squares method is used to linearly fit the tool wear prediction value and the processing duration. First, based on the algorithm formula and Calculate the time variables separately x Variables with tool wear prediction values y The average , then based on the algorithm formula and Calculate the slope of the tool wear time relationship equation m With intercept b .

[0068] SS17. Plotting tool wear versus time

[0069] According to the slope calculated in step SS16 m With intercept b , establish the tool wear time relationship equation , through the plane rectangular coordinate system, with time variable x is the horizontal axis, wear prediction value variable y The vertical axis is used to draw the tool wear time prediction curve based on the tool wear time relationship equation.

[0070] It should be noted that during the finishing process of the integral blade disk, the cutting speed is usually set to a constant value to maintain stable cutting parameters to obtain high-precision dimensions and good surface quality. At the same time, since the cutting depth and feed rate during finishing are relatively small and stable, the component of the cutting force in the direction perpendicular to the machining surface, that is, the normal load, is also relatively stable. The tool wear time curve is obtained through the tool wear time relationship equation, which is convenient for the subsequent division of the tool wear time period.

[0071] Embodiment 3:

[0072] Based on the above embodiment, as a further optimization of its content, this embodiment 3 provides a further preferred implementation method, such as Figure 3 As shown, step SS2 includes at least the following sub-steps when constructing the machining compensation model under different wear time periods:

[0073] SS21. Divide tool wear into time periods

[0074] Based on the tool wear time prediction curve obtained by fitting in step SS1, the tool wear threshold sequence is set based on historical processing experience or workpiece processing accuracy requirements. i 1,..., i k ,..., i n′},in i 1 represents the wear threshold of the tool during the initial run-in phase. i n′ The wear threshold that represents the tool replacement criterion, i k Indicates k wear threshold and k =1,2,…, n ′, n′ is the number of wear thresholds, and is based on the tool wear threshold sequence { i 1,..., i k ,..., i n′}, combined with the tool wear time prediction curve obtained by fitting in step SS1, based on the algorithm formula Get the corresponding time node sequence { x 1,..., x k ,..., x n′}, thus shortening the entire processing and maintenance cycle T Divide into n ′+1 wear time period ;

[0075] SS22. Collect historical machining error data at different wear time periods

[0076] For each wear time period divided in step SS21, collect the historical machining error data corresponding to the tool wear change, including the feed error Δf Error with cutting depth Δd , where the feed error Δf Indicates the deviation between the actual feed amount and the set feed amount of the tool during the processing, and the cutting depth error Δd Indicates the deviation between the actual cutting depth and the theoretical depth of the workpiece surface after processing, and divides the historical processing error data collected in each wear time period into a training data set and a verification data set;

[0077] SS23. Training machining compensation model based on historical machining error data

[0078] The training data set is trained through a convolutional neural network, and a corresponding machining compensation model regarding tool wear, feed rate, and cutting depth compensation is established for each wear time period. The corresponding machining compensation model is further verified and optimized through a verification data set.

[0079] It should be noted that the tool wear threshold { i 1,..., i k ,..., i n′It is obtained by conducting tool processing experiments on the current material. That is, when the tool wear reaches a certain level, the feed error and cutting depth error generated change significantly. The relevant wear amount can be defined as the tool wear threshold. Computer-aided engineering (CAE) software can also be used to establish a numerical simulation model of the cutting process of the integral blade disk. In the model, the material properties, geometric shapes of the tool and workpiece, and the kinematic and dynamic parameters of the cutting process are accurately set. The tool wear condition is quantitatively modeled to simulate the cutting process of the tool under different wear amounts and obtain the tool wear threshold.

[0080] Further preferably, step SS2 comprises at least the following sub-steps when blade grouping is performed on the currently processed blisk:

[0081] SS24. Blade No.

[0082] According to the 3D drawing of the blisk being processed, the number of blades to be finished is determined by traversing the current 3D drawing, a blade is selected as the starting point, and all blades are numbered sequentially;

[0083] SS25. Leaf Grouping

[0084] pass MySQL Create a blade file, divide the blades of the whole blade disk with an even number into equal groups, and divide the blades of the whole blade disk with an odd number into unequal groups.

[0085] It should be noted that for an even number of integral blade disks, the blades are equally grouped, i.e., the blades of the integral blade disk are paired in groups of even numbers, and for an odd number of integral blade disks, the blades are unequally grouped, i.e., the blades are first paired in groups of even numbers, and the remaining blades after grouping are grouped separately.

[0086] Embodiment 4:

[0087] Based on the above embodiment, as a further optimization of its content, this embodiment 4 provides a further preferred implementation mode, and step SS3 includes the following sub-steps during implementation:

[0088] SS31. Equal division of wear time periods

[0089] Based on the wear time period obtained by the tool wear threshold in step SS2, combined with the blade grouping result of the current integral blade disk, a corresponding blade group processing task is assigned to each wear time period, and based on the number of component blades in the blade group, the corresponding wear time period is equally divided based on the number of blades in the group, ensuring that the processing time of each blade in the same group is the same;

[0090] SS32. Processing area marking

[0091] Based on the equally divided processing time periods in each wear time period, the processing area of ​​each blade in the current tool processing group in the same wear time period is marked, and the processing area is recorded in the file of the corresponding blade;

[0092] SS33. Tool Path Compensation

[0093] The blades of the same group are processed isochronously in sequence in equally divided time periods in each tool wear time period, and the corresponding processing compensation model regarding feed rate and cutting depth is matched based on the current wear time period, and path compensation is performed during tool finishing.

[0094] It should be noted that the processing area is recorded in the file of the corresponding blade, that is, it is recorded as A , the processing area of ​​the blade disk in the group during the wear period is recorded as A 1. A 2. … A l ,in l The number of blades in the current group is recorded in the corresponding area in the blade file, which can facilitate problem tracing. At the same time, the processing consistency of the blades of the blade disk is enhanced by isochronous processing of the blades in the same group in each equally divided interval.

[0095] Embodiment 5:

[0096] Based on the above embodiment, as a further optimization of its content, this embodiment 5 provides a further preferred implementation mode, and step SS4 includes the following sub-steps during implementation:

[0097] SS41. Checking blade processing parameters

[0098] Based on the processing areas recorded in the blade files, after each set of blisk blades is processed, the processing parameters of different processing areas of the blades are measured by line laser measuring equipment, and standard thresholds are set based on the drawing requirements;

[0099] SS42. Determine the error compensation range

[0100] The processing parameters measured in different processing areas are compared with the standard threshold. When the processing parameters are less than or equal to the standard threshold, it means that the compensation model of the current processing area is reasonable and does not need to be corrected. When the processing parameters are greater than the standard threshold, it means that the compensation model of the current processing area is unreasonable and needs to be corrected.

[0101] SS43. Modified machining compensation model

[0102] For the processing area that needs to be corrected, based on the difference between the measured value and the standard threshold Δx , a regularization term is added to the loss function of the original convolutional neural network to correct the original processing compensation model of the current processing area. The algorithm formula is: ,in represents the loss function, MSE , l They represent the mean square error loss and the regularization coefficient respectively. According to the modified loss function, back propagation is performed to update the convolutional neural network parameters to obtain the machining compensation model of the corrected feed amount and cutting depth of the current machining area to replace the original machining compensation model of the feed amount and cutting depth.

[0103] It should be noted that the regularization coefficient is used in machine learning and deep learning to balance the complexity of the model and the degree of fit of the training data to prevent overfitting. Through cross-validation, the data set is divided into multiple subsets, and the model is repeatedly trained and verified to select the optimal regularization coefficient.

[0104] In summary, the present invention establishes a wear prediction model for the tool, and divides the tool processing time into time periods in combination with the wear threshold, so as to match different processing compensation models for different processing time periods. The corresponding processing compensation model will automatically adjust the processing parameters to control the surface roughness back within the qualified range; the blades of the integral impeller being processed are grouped and paired, and based on the number of blades in each group, the different wear time periods of the current processing tool are equally divided, and based on the equally divided processing time, the blades in the same group are processed in sequence at the same time, thereby enhancing the consistency of the integral impeller blade processing and improving the quality uniformity of the entire integral impeller blade part. Isochronous machining during the wear time period can ensure that the blades in the same group are processed at approximately the same tool wear stage and with the same compensation strategy, making the machining conditions faced by each blade more uniform, thereby effectively improving the consistency of blade machining in the same group and better meeting the requirements of the drawings for blade size, shape and surface quality. By recording and marking the blade areas in the group that are machined during the same wear time period, the machining of each blade in the group is divided into clear time intervals and corresponding tool wear stages. Once a machining quality problem occurs, it can be quickly and accurately located to which group of blades the problem occurred during machining and in which wear time period of the corresponding tool, facilitating subsequent problem tracing.

[0105] Through the above embodiments, the purpose of the present invention is fully and effectively achieved. Those skilled in the art can understand that the present invention includes but is not limited to the contents described in the drawings and the above specific embodiments. Although the present invention has been described with respect to the most practical and preferred embodiments currently considered, it should be understood that the present invention is not limited to the disclosed embodiments, and any modification that does not deviate from the functional and structural principles of the present invention will be included in the scope of the claims.

Claims

1. A tool path compensation method for finishing of an integral blade disk, characterized in that: The method comprises at least the following steps when implemented: SS1. Collect multiple sets of historical tool data for processing integral blades of the same material. Each set of historical tool data includes at least tool wear value, normal load, cutting speed, processing time and workpiece material hardness parameters. Calculate the tool wear coefficient of each group and average the tool wear coefficients of multiple groups. Combined with the processing parameters, establish the current tool wear prediction model. Then, estimate the wear prediction value of the current tool in different time periods based on the current tool wear prediction model. At the same time, fit the wear prediction value with time to obtain the wear time prediction curve of the current tool. The algorithm formula of the current tool wear prediction model is expressed as follows: ,in V is the wear prediction value of the current tool, represents the mean of multiple groups of tool wear coefficients, W , v , t are the normal load, cutting speed and processing time of the current tool respectively. H Indicates the hardness of the workpiece material; SS2. According to the influence of tool wear on the finishing of the blisk, combined with the current tool wear time prediction curve fitted in step SS1, the tool wear is divided into several wear time periods, and the historical machining error data in different wear time periods are collected. Based on the feed error and cutting depth error in the historical machining error data, a machining compensation model in different wear time periods is obtained through convolutional neural network training. The machining compensation model is used to adjust the machining parameters to compensate for the influence of tool wear on machining accuracy in different wear time periods, and then the blades of the currently processed blisk are grouped; SS3. Based on the wear time periods divided in step SS2 and the blade grouping results of the blisk, tool processing is performed in a wear time period-by-wear time period-by-blade group manner, a corresponding blade group processing task is assigned to each wear time period, and each tool wear time period is divided isochronously based on the number of blades in the corresponding blade group, and processing is performed in each wear time period based on the grouping order of the blades of the blisk, and the same processing area of ​​the blades of the same blade group isochronously processed in sequence in each wear time period, and tool path compensation is performed by matching the processing compensation model of the corresponding wear time period; SS4. Measure the machining parameters of different areas of the blisk blade after finishing, and compare the standard thresholds based on different divided areas. When the machining parameters of a divided area exceed the standard threshold, the machining compensation model corresponding to the divided area is corrected based on the part deviation between the measured value and the standard threshold. If the machining parameters are not greater than the standard threshold, there is no need to correct the machining compensation model. Wherein, step SS4 at least includes the following sub-steps when implemented: SS41. Based on the processing areas recorded in the blade files, after each set of blisk blades is processed, the processing parameters of different processing areas of the blades are measured by line laser measurement equipment, and standard thresholds are set based on the drawing requirements; SS42. Compare the machining parameters measured in different machining areas with the standard threshold. If the machining parameters are less than or equal to the standard threshold, it means that the machining compensation model of the current machining area is reasonable and does not need to be corrected. If the machining parameters are greater than the standard threshold, it means that the machining compensation model of the current machining area is unreasonable and needs to be corrected. SS43. For the processing area that needs to be corrected, based on the difference between the measured value and the standard threshold Δx , a regularization term is added to the loss function of the original convolutional neural network to correct the original processing compensation model of the current processing area. The algorithm formula is: ,in represents the loss function, MSE , λ Represent the mean square error loss and the regularization coefficient respectively. According to the modified loss function, back propagation is performed to update the convolutional neural network parameters to obtain the machining compensation model of the corrected feed amount and cutting depth of the current machining area to replace the original machining compensation model of the feed amount and cutting depth. SS5. Rapidly trace the source of machining problems through the machining time period and tool wear stage information recorded in the blade archive. At the same time, optimize the blade grouping strategy and machining path to improve the machining consistency and quality uniformity of the entire blade disk in response to errors that occur during the machining process.

2. The tool path compensation method for finishing of an integral blade disk according to claim 1, characterized in that: Step SS1 includes at least the following sub-steps when constructing the wear prediction model of the current tool: SS11. Collection N The tool data for machining the blisk of the same material in the group history, including N is an integer and satisfies N ≥2, each set of historical tool data collected is derived from actual processing records, processing experiments and / or numerical simulations during the processing, and each set of historical tool data includes at least tool wear value, normal load, cutting speed, processing time and workpiece material hardness, where tool wear value represents the volume or value of tool wear during the processing, normal load represents the force on the tool in the vertical direction of the processing surface, and cutting speed represents the relative movement speed between the tool and the workpiece; SS12. Based on the collected historical tool data, the algorithm formula Calculate the tool wear coefficient corresponding to each set of historical tool data one by one, where k i For the i Tool wear coefficient corresponding to the historical tool data of the group, V i , W i , v i , t i , H i Respectively represent i Tool wear values, normal load, cutting speed, machining time and workpiece material hardness in the group historical tool data, i is the number of each history tool data set and i =1,2,…, N , then based on the algorithm formula The tool wear coefficient corresponding to each group of data is averaged to obtain the mean value of the tool wear coefficient , to eliminate random errors and individual differences in the data; SS13. Mean value of tool wear coefficient after averaging , combined with the current tool processing parameters, the current tool wear prediction model is established, and its algorithm formula is as follows: in, V Represents the wear prediction value of the current tool, W , v , t Respectively represent the normal load, cutting speed and processing time of the current tool, H Indicates the hardness of the workpiece material.

3. A tool path compensation method for finishing of an integral blade disk according to claim 2, characterized in that: Step SS1 includes the following sub-steps when fitting the wear time curve of the current tool: SS14. Determine the current maintenance cycle of the blisk finishing tool T , using arithmetic progression to divide it into multiple equal time intervals, each time interval is Δt , from the first time interval point t Starting from 1, increase the same time interval successively Δt , until the last time interval point t n = T End, get the time series { t 1,.., t j , ..., t n },in t j For the j time intervals and j =1,2,…, n , n Maintenance cycle for tool processing T the number of time intervals divided; SS15. Based on each time interval and the corresponding normal load, cutting speed, machining time and workpiece material hardness at that moment. H , through the current tool wear prediction model established in sub-step SS13, the tool wear prediction values ​​corresponding to each time interval are calculated based on the following algorithm formula: in, V j , W j , v j , t ' j The time intervals are t j The corresponding tool wear prediction value, normal load, cutting speed, processing time, and processing time t ' j satisfy t ' j = t j ; SS16. The least square method is used to linearly fit the tool wear prediction value and the processing duration. First, based on the algorithm formula and Calculate the time variables separately x Variables with tool wear prediction values y The average , then based on the algorithm formula and Calculate the slope of the tool wear time relationship equation m With intercept b ; SS17. Based on the calculated slope m With intercept b , establish the tool wear time relationship equation , through the plane rectangular coordinate system, with time variable x is the horizontal axis, wear prediction value variable y The vertical axis is used to draw the tool wear time prediction curve based on the tool wear time relationship equation.

4. The tool path compensation method for finishing of an integral blade disk according to claim 1, characterized in that: Step SS2 includes: SS21. Divide tool wear into time periods Based on the tool wear time prediction curve obtained by fitting, the tool wear threshold sequence is set based on historical processing experience or workpiece processing accuracy requirements. θ 1,..., θ k ,..., θ n′ },in θ 1 represents the wear threshold of the tool during the initial running-in phase. θ n′ The wear threshold that represents the tool replacement criterion, θ k Indicates k wear threshold and k =1,2,…, n ′, n ′ is the number of wear thresholds, and is based on the tool wear threshold sequence { θ 1,..., θ k ,..., θ n′ }, combined with the tool wear time prediction curve fitted in step SS1, the corresponding time node sequence { x 1,..., x k ,..., x n′ }, thus shortening the entire processing and maintenance cycle T Divide into n ′+1 wear time period; SS22. For each wear period, collect historical machining error data corresponding to tool wear changes, including feed error Δf Error with cutting depth Δd , where the feed error Δf Indicates the deviation between the actual feed amount and the set feed amount of the tool during the processing, and the cutting depth error Δd Indicates the deviation between the actual cutting depth and the theoretical depth of the workpiece surface after processing, and divides the historical processing error data collected in each wear time period into a training data set and a verification data set; SS23. The training data set is trained by convolutional neural network, and the corresponding machining compensation model about tool wear, feed rate and cutting depth compensation is established for each wear time period. The corresponding machining compensation model is further verified and optimized by the verification data set.

5. A tool path compensation method for finishing of an integral blade disk according to claim 4, characterized in that: Step SS2 includes: SS24. According to the 3D drawing of the blisk being processed, determine the number of blades to be finished by traversing the 3D drawing, select a blade as the starting point, and number all the blades in sequence; SS25. Create a blade file, divide the blades into equal groups for the whole blade disks with an even number, and divide the blades into unequal groups for the whole blade disks with an odd number.

6. A tool path compensation method for finishing a blisk according to claim 5, characterized in that: Step SS3 at least includes the following sub-steps when implemented: SS31. Based on the wear time period obtained by the tool wear threshold in step SS2, combined with the blade grouping result of the current blisk, a corresponding blade group processing task is assigned to each wear time period, and based on the number of blades in the blade group, the corresponding wear time period is equally divided based on the number of blades in the group, ensuring that the processing time of each blade in the same group is the same; SS32. Based on the equally divided processing time periods in each wear time period, mark the processing area of ​​each blade in the current tool processing group in the same wear time period, and record the processing area in the file of the corresponding blade; SS33. The blades of the same group are processed in equal time periods in each tool wear time period, and the corresponding processing compensation model about feed rate and cutting depth is matched based on the current wear time period, and path compensation is performed during tool finishing.

7. A computer program product comprising computer instructions, characterized in that The computer instructions are used to execute the tool path compensation method for finishing the integral blisk as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the tool path compensation method for finishing the integral blisk as described in any one of claims 1 to 6 is implemented.

Images