An online evaluation method for CFRP hole making process quality

Abstract

An online evaluation method for CFRP hole making process quality relates to the technical field of composite material mechanical processing, including using a power analyzer to collect original processing power data of hole making equipment; using an electron microscope to collect hole quality data of CFRP materials under corresponding power data to form a sample database; after cleaning the power data in the sample database, recombining the characteristics of the power data with the hole quality data to form a power quality data set; using the recombined data in the power quality data set to train the constructed model to generate a hole making quality evaluation model; using the obtained hole making quality evaluation model to online evaluate the hole making power data generated in real time to obtain the quality of the current drilling, etc., so as to achieve the effect of online evaluation of the processing quality of CFRP hole making directly in the processing process without using additional hardware in the hole making process.

Description

Technical Field

[0001] The invention relates to the technical field of composite material machining, and in particular to an online evaluation method for CFRP hole making process quality. Background Art

[0002] Carbon fiber reinforced polymer composites (CFRP) have been widely used in a variety of fields such as aerospace, automobiles, infrastructure and sports and leisure for more than 50 years due to their high specific strength and specific modulus. Their high fatigue, toughness and high temperature wear resistance, lightweight and oxidation resistance make these materials an excellent choice for engineering applications.

[0003] Carbon fiber composites have been widely used in the aerospace industry since the 1990s, and in the 2010s, they were used again in the wind energy, automotive and other industries on a larger scale. In the document "Zhang J, Lin G, Vaidya U, et al. Past, present and future prospect of global carbon fiber composite developments and applications [J]. Composites Part B: Engineering, 2023, 250: 110463," Zhang J et al. reviewed the use of carbon fiber composites. Over the past two decades, the compound annual growth rate of global carbon fiber composites has been about 12.5%.

[0004] Drilling is one of the most common operations in the processing of carbon fiber composite materials, and the hole quality after drilling is the key factor to ensure that the processed material can meet the application requirements. There are many methods to evaluate the quality of hole making. Many scholars at home and abroad have conducted many studies on the quality control of carbon fiber composite drilling. The mainstream CFRP drilling quality evaluation methods currently used include: measuring the roundness error after the hole is processed; optical inspection methods based on machine vision and ultrasonic non-destructive evaluation of the inner wall of the hole. Most of these methods require the processed holes to undergo a series of processing, such as taking pictures with an optical microscope, C scanning, etc. to detect and distinguish defects, and then evaluate the quality of the hole. Many studies have shown that the spindle speed, feed speed, cutting force, torque and other conditions during CFRP hole making are related to the quality of the hole, and optimizing the processing parameters is helpful to improve the quality of the hole; the method for monitoring the drilling quality of traditional metal materials is generally achieved by collecting and monitoring vibration signals or force signals. Although it can provide a reference for the monitoring and evaluation of the drilling quality of composite materials, due to the anisotropy of composite materials, the collection of force, vibration and other signals during the processing process is more difficult and complex than that of traditional metal materials.

[0005] Since the quality of CFRP hole making is directly related to the spindle speed, feed speed, cutting force, torque and other factors during processing, the change of these factors will cause the change of the spindle power of the machine tool during hole making, and at the same time cause the change of hole making quality. Based on this, a method for online prediction of processing quality based on CFRP hole making processing power information is proposed. During the processing process, only the processing power information needs to be collected, and the proposed integrated decision tree model can be used to evaluate whether the processing quality of the hole meets the requirements. Summary of the invention

[0006] The embodiment of the present invention provides an online evaluation method for the process quality of CFRP hole making, which can directly realize the online evaluation of the processing quality of CFRP hole making during the processing.

[0007] An online evaluation method for CFRP hole making process quality, comprising: using a power analyzer to collect original processing power data of a hole making device; and using an electron microscope to collect hole quality data of CFRP materials under corresponding power data to form a sample database;

[0008] After cleaning the power data in the sample database, the features of the power data and the hole quality data are recombined to form a power quality data set;

[0009] The constructed model is trained using the reorganized data in the power quality data set to generate a hole making quality evaluation model;

[0010] The obtained hole-making quality evaluation model is used to perform online evaluation on the real-time hole-making power data to obtain the quality of the current hole.

[0011] Furthermore, the process of cleaning the power data includes the following steps:

[0012] By comparing the data in the sample database, duplicate items are deleted;

[0013] The average value, root square amplitude, standard deviation, peak value and time domain peak-to-valley ratio of the power data during the entire processing process were extracted, and the characteristic data were normalized.

[0014] Furthermore, the collected hole quality data includes image data of the hole entrance, hole exit and hole wall.

[0015] Further, it also includes a hole quality identifier, wherein the hole quality identifier includes a standard database, a feature extraction unit and a quality calibration unit;

[0016] The standard database has preset hole standard feature metadata, including standard feature metadata of the hole entrance, exit and hole wall;

[0017] The feature extraction unit is suitable for extracting features from hole quality data;

[0018] The quality calibration unit identifies the extracted features based on the hole standard feature metadata and judges the hole quality data;

[0019] If the hole quality data is judged to be qualified, its label is defined as 1, and if the hole quality data is judged to be unqualified, its label is defined as 0;

[0020] The corresponding power characteristic data and hole quality labels are reorganized to form a power quality data set.

[0021] Furthermore, the process of cleaning the power data also includes the identification of outliers, and the steps are as follows:

[0022] Construct a corresponding hole-making equipment model according to the parameters of the hole-making equipment;

[0023] Under the same working conditions of the hole-making device and the hole-making device model, a power parameter a of the hole-making device model and a power parameter b of the hole-making device are obtained to form a power parameter a data set and a power parameter b data set;

[0024] Set a threshold and compare the corresponding data in the power parameter a data set with the power parameter b data set;

[0025] The power parameters b exceeding the threshold are taken out, and the remaining power parameters b are output and cleaned.

[0026] Further,

[0027] The steps to generate a hole quality evaluation model include:

[0028] Randomly extract n groups of training data from the sample data set in the power quality data set, and assign weights to each group of data. The calculation formula is:

[0029]

[0030] Among them, Q n (i) is the weight corresponding to each set of power data;

[0031] Establish the corresponding C4.5 tree model according to the characteristics of the training data, and initialize the weight of each node in the model;

[0032] Construct a weak evaluation learner and specify the number of cycles required by the model m = 1, 2, ..., M, thereby generating M C4.5 weak evaluation learners H m ;

[0033] Calculate the weight of the weak evaluation learner. In order to effectively combine the classification results of all weak evaluation learners, the AdaBoost algorithm sets a weight value α for each weak evaluation learner and calculates the weight value based on the error rate of each weak evaluation learner. The weight value is organically combined with the classification results of the C4.5 weak evaluation learner. The calculation formula is:

[0034]

[0035] Among them, ε is the error rate of the C4.5 weak evaluation learner, which is calculated as follows:

[0036] Adjust the weight distribution of the training samples. According to the weights of each weak evaluation learner calculated above, the weights of the training sample data of the next cycle are updated accordingly, thereby reducing the weight D of the correctly classified samples and increasing the weight D of the incorrectly classified samples.

[0037] The formula is as follows: m+1 =w m+1,1 ,w m+1,2 ,w m+1,3 ,…,w m+1,i ,…,w m+1,N ;

[0038]

[0039] Among them, D m+1 is the set of all sample weights, w m+1,i is the updated weight of the i-th power quality sample. If the sample is correctly classified, y i =1, if it is misclassified, then y i = -1;

[0040] Construct a strong evaluation learner. When the weight D is updated, the Adaboost algorithm enters the next round of cyclic iteration. In this process, the training sample data will be repeatedly trained and the sample weight will be updated until the number of wrong evaluations reaches the threshold or the number of weak evaluation learners has reached the maximum number previously set. Combining these weak evaluation learners will form a strong evaluation learner, whose formula is:

[0041]

[0042] Among them, F (x) is a strong evaluation learner; finally, a CFRP hole quality evaluation model is formed;

[0043] After completing the learning phase, the formed model is packaged and the hole making quality is evaluated online.

[0044] Furthermore, a power analyzer is used to monitor the processing power of the hole-making equipment, and a time window is intercepted through the start processing signal and the end processing signal of the hole-making equipment, so that only the power information in the window is processed and calculated. The processing power data is passed to the established hole-making quality evaluation model to output the quality evaluation result of the current drilling.

[0045] The beneficial effects of the above technical solution provided by the embodiment of the present invention include at least:

[0046] The present invention obtains the original processing power data of the hole-making equipment and the corresponding hole quality data, and builds a quality evaluation model after preprocessing for training to obtain a hole-making quality evaluation model, thereby achieving the effect of evaluating the hole-making quality through the power data of the hole-making equipment during the hole-making equipment processing process; that is, the hole-making quality results of CFRP can be quickly obtained without the need for additional equipment, and at the same time, it can provide decision support for subsequent hole-making processing plans, which is of great significance.

[0047] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0048] The technical solution of the present invention is further described in detail below through the drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0050] Figure 1 A schematic flow chart of an online evaluation method for CFRP hole making process quality disclosed in an embodiment of the present invention;

[0051] Figure 2 It is a framework diagram of the hole making quality evaluation method disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0053] The invention content (overall framework, technical solution) of the present invention is as follows:

[0054] The technical problem to be solved by the present invention is to address the shortcomings of the hole quality evaluation method for carbon fiber reinforced polymer composite (CFRP) hole making processing, and propose an online hole quality evaluation method based on hole making processing power. The method can be used to obtain hole quality information of the composite material after hole making processing in a timely manner.

[0055] like Figure 2 As shown in FIG. 1 , the method includes two stages: learning and online application. In the learning stage, the established hole-making quality evaluation model is trained using the historical data collected by the power analyzer; in the online application stage, the trained hole-making quality evaluation model directly uses the power data from the machining system to evaluate the hole-making quality without using additional hardware measurements (such as electron microscopes).

[0056] The learning phase is carried out offline, with the purpose of using historical data to train a hole-making quality evaluation model. Figure 2 As shown in the figure, this stage includes several modules, including data collection, data cleaning, data reorganization and hole making quality evaluation modeling. First, the original processing power data and hole quality data are collected by using a power analyzer and an electron microscope. After outlier processing, feature extraction and normalization of the power data, the features of the power data are reorganized with the hole quality data; then the reorganized data is used to train the constructed model and generate a hole making quality evaluation model.

[0057] In the second stage, the hole-making quality evaluation model obtained in the first stage is used to perform online identification on the real-time hole-making power data to obtain the quality of the current hole. In this way, the collected processing power data is continuously input into the generated model to form an online evaluation of the hole-making quality.

[0058] Figure 1 The flowchart of the online evaluation method for CFRP hole making process quality disclosed in an embodiment of the present invention is shown, which includes the following steps:

[0059] 1. Adjust the corresponding parameters of the power analyzer to obtain the power data information of the hole making process of the hole making equipment, and use an electron microscope to collect the hole quality data of the CFRP material under the corresponding power data to form a sample database.

[0060] The collected hole quality data includes image data of the hole entrance, hole exit and hole wall.

[0061] According to different processing conditions such as spindle speed and feed speed, different processing power data are obtained. The power analyzer collects the power data generated by the hole-making equipment.

[0062] 2. Preprocess the power data in the sample database, including outlier processing, extract the average value, root square amplitude, standard deviation, peak value and time domain peak-to-valley ratio of the power data during the entire processing process, and normalize the feature data.

[0063] The process of preprocessing (cleaning) the power data includes the following steps:

[0064] 2.1. Compare the data in the sample database and delete the duplicates;

[0065] 2.2. Extract the average value, root square amplitude, standard deviation, peak value and time domain peak-to-valley ratio of the power data during the entire processing process, and normalize the characteristic data.

[0066] The process of cleaning power data also includes the identification of outliers. The steps are as follows:

[0067] Construct a corresponding hole-making equipment model according to the parameters of the hole-making equipment;

[0068] Under the same working conditions of the hole-making device and the hole-making device model, a power parameter a of the hole-making device model and a power parameter b of the hole-making device are obtained to form a power parameter a data set and a power parameter b data set;

[0069] Set a threshold and compare the corresponding data in the power parameter a data set with the power parameter b data set;

[0070] The power parameters b exceeding the threshold are taken out, and the remaining power parameters b are output and cleaned.

[0071] 3. Perform quality calibration on the processed holes, with qualified labels as 1 and unqualified labels as 0; reorganize the corresponding power characteristic data and hole quality labels to form a power quality data set.

[0072] The above also includes a hole quality identifier, wherein the hole quality identifier includes a standard database, a feature extraction unit and a quality calibration unit;

[0073] The standard database has preset hole standard feature metadata, including standard feature metadata of the hole entrance, exit and hole wall;

[0074] The feature extraction unit is suitable for extracting features from hole quality data;

[0075] The quality calibration unit identifies the extracted features based on the hole standard feature metadata and judges the hole quality data;

[0076] If the hole quality data is judged to be qualified, its label is defined as 1, and if the hole quality data is judged to be unqualified, its label is defined as 0;

[0077] The corresponding power characteristic data and hole quality labels are reorganized to form a power quality data set.

[0078] 4. Randomly extract n groups of training data from the power quality data set and assign weights to each group of data. The calculation formula is:

[0079]

[0080] Among them, Q n (i) is the weight corresponding to each set of power data;

[0081] Establish the corresponding C4.5 tree model according to the characteristics of the training data, and initialize the weight of each node in the model;

[0082] 5. Construct a weak evaluation learner and specify the number of cycles required by the model m = 1, 2, ..., M, thereby generating M C4.5 weak evaluation learners H m ;

[0083] 6. Calculate the weight of the weak evaluation learner. In order to effectively combine the classification results of all weak evaluation learners, the AdaBoost algorithm sets a weight value α for each weak evaluation learner and calculates the weight value based on the error rate of each weak evaluation learner. The weight value is organically combined with the classification results of the C4.5 weak evaluation learner. The calculation formula is:

[0084]

[0085] Among them, ε is the error rate of the C4.5 weak evaluation learner, which is calculated as follows:

[0086] 7. Adjust the weight distribution of the training samples. According to the weights of each weak evaluation learner calculated above, the weights of the training sample data of the next cycle are updated accordingly, thereby reducing the weight D of the correctly classified samples and increasing the weight D of the incorrectly classified samples.

[0087] The formula is as follows: m+1 =w m+1,1 ,w m+1,2 ,w m+1,3 ,…,w m+1,i ,…,w m+1,N ;

[0088]

[0089] Among them, D m+1 is the set of all sample weights, w m+1,i is the updated weight of the i-th power quality sample. If the sample is correctly classified, y i=1, if it is misclassified, then y i = -1;

[0090] 8. Construct a strong evaluation learner. When the weight D is updated, the Adaboost algorithm enters the next round of cyclic iteration. In this process, the training sample data will be repeatedly trained and the sample weight will be updated until the number of wrong evaluations reaches the threshold or the number of weak evaluation learners has reached the maximum number previously set. Combining these weak evaluation learners forms a strong evaluation learner, whose formula is:

[0091] F (x) =∑ M m＝1 α m F m (x);

[0092] Among them, F (x) is a strong evaluation learner; finally, a CFRP hole quality evaluation model is formed;

[0093] After the learning phase is completed, the formed model is packaged and the hole making quality is evaluated online;

[0094] 9. After completing the learning phase, the formed model is packaged and the hole making quality is evaluated online.

[0095] The power analyzer is used to monitor the processing power of the hole-making equipment. A time window is intercepted through the start processing signal and the end processing signal of the hole-making equipment, so that only the power information in the window can be processed and calculated. The processing power data is passed to the established hole-making quality evaluation model to output the quality evaluation result of the current drilling.

[0096] The power analyzer is used to monitor the processing power of the hole-making equipment. A time window is intercepted through the start processing signal and the end processing signal of the hole-making equipment, so that only the power information in the window can be processed and calculated. The processing power data is passed to the established hole-making quality evaluation model to output the quality evaluation result of the current drilling.

[0097] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process can be rearranged without departing from the scope of protection of the present disclosure. The attached method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy described.

[0098] In the above detailed description, various features are grouped together in a single embodiment to simplify the disclosure. This method of disclosure should not be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly stated in each claim. On the contrary, as reflected in the appended claims, the invention is in a state of having less than all the features of the disclosed individual embodiments. Therefore, the appended claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

[0099] Those skilled in the art will also appreciate that the various illustrative logic blocks, modules, circuits, and algorithmic steps described in conjunction with the embodiments herein can all be implemented as electronic hardware, computer software, or a combination thereof. In order to clearly illustrate the interchangeability between hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above around their functions. Whether such functions are implemented as hardware or software depends on specific applications and the design constraints imposed on the entire system. A skilled person can implement the described functions in an alternative manner for each specific application, but such implementation decisions should not be interpreted as departing from the scope of protection of the present disclosure.

[0100] The steps of the method or algorithm described in conjunction with the embodiments herein may be directly embodied as hardware, a software module executed by a processor, or a combination thereof. The software module may be located in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a mobile disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is connected to the processor so that the processor can read information from the storage medium and can write information to the storage medium. Of course, the storage medium may also be an integral part of the processor. The processor and the storage medium may be located in an ASIC. The ASIC may be located in a user terminal. Of course, the processor and the storage medium may also be present in a user terminal as discrete components.

[0101] For software implementation, the techniques described in this application can be implemented with modules (e.g., procedures, functions, etc.) that perform the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented within the processor or outside the processor. In the latter case, it is coupled to the processor in a communication manner via various means, which are well known in the art.

[0102] The above description includes examples of one or more embodiments. Of course, it is not possible to describe all possible combinations of components or methods for the purpose of describing the above embodiments, but it should be recognized by those skilled in the art that the various embodiments may be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, the word is encompassed in a manner similar to the term "including," as explained by "including" used as a transitional word in the claims. In addition, any term "or" used in the specification of the claims is intended to mean "non-exclusive or."

Claims

1. An online evaluation method for CFRP hole making process quality, characterized in that: include: The power analyzer is used to collect the original processing power data of the hole-making equipment, wherein the power data is the spindle power data of the hole-making equipment; and the hole quality data of the CFRP material under the corresponding power data is collected by an electron microscope to form a sample database; after cleaning the power data in the sample database, the characteristics of the power data and the hole quality data are recombined to form a power quality data set; The constructed model is trained using the reorganized data in the power quality data set to generate a hole making quality evaluation model; The steps of generating the hole making quality evaluation model include: Randomly extract n groups of training data from the sample data set in the power quality data set, and assign the same weight to each group of data; According to the characteristics of the training data, the corresponding C4.5 tree model is established, the weak evaluation learner is constructed, and the number of cycles required by the model is specified to generate M C4.5 weak evaluation learners H m ; Calculate the weight of the weak evaluation learner. The AdaBoost algorithm sets a weight value α for each weak evaluation learner and calculates the weight value based on the evaluation error rate of each weak evaluation learner in the training data. Adjust the weight distribution of the training samples. According to the weights of each weak evaluation learner obtained in the above steps, the weights of the training sample data of the next cycle are updated and adjusted accordingly, reducing the weights of correctly classified samples and increasing the weights of incorrectly classified samples. Construct a strong evaluation learner. After the weights are adjusted, the Adaboost algorithm enters the next round of iterative process. In this process, the training sample data will be repeatedly trained and the sample weights will be updated until the number of wrong evaluations reaches the threshold or the number of weak evaluation learners has reached the maximum number previously set. These weak evaluation learners are combined according to the new weight values ​​to form a strong evaluation learner. Finally, the CFRP hole making quality evaluation model is formed; After the learning phase is completed, the formed model is packaged and the hole making quality is evaluated online; The obtained hole-making quality evaluation model is used to perform online evaluation on the real-time hole-making power data to obtain the quality of the current hole.

2. The online evaluation method for CFRP hole making process quality according to claim 1, characterized in that: The process of cleaning power data includes the following steps: By comparing the data in the sample database, duplicate items are deleted; The average value, root square amplitude, standard deviation, peak value and time domain peak-to-valley ratio of the power data during the entire processing process were extracted, and the characteristic data were normalized.

3. The online evaluation method for CFRP hole making process quality according to claim 1, characterized in that: The collected hole quality data includes image data of the hole entrance, hole exit and hole wall.

4. The online evaluation method for CFRP hole making process quality according to claim 3, characterized in that: Also included is a hole quality identifier, wherein the hole quality identifier includes a standard database, a feature extraction unit and a quality calibration unit; The standard database has preset hole standard feature metadata, including standard feature metadata of the hole entrance, exit and hole wall; The feature extraction unit is suitable for extracting features from hole quality data; The quality calibration unit identifies the extracted features based on the hole standard feature metadata and judges the hole quality data; If the hole quality data is judged to be qualified, its label is defined as 1, and if the hole quality data is judged to be unqualified, its label is defined as 0; The corresponding power characteristic data and hole quality labels are reorganized to form a power quality data set.

5. The online evaluation method for CFRP hole making process quality according to claim 2, characterized in that: The process of cleaning power data also includes the identification of outliers. The steps are as follows: Construct a corresponding hole-making equipment model according to the parameters of the hole-making equipment; Under the same working conditions of the hole-making device and the hole-making device model, a power parameter a of the hole-making device model and a power parameter b of the hole-making device are obtained to form a power parameter a data set and a power parameter b data set; Set a threshold and compare the corresponding data in the power parameter a data set with the power parameter b data set; The power parameters b exceeding the threshold are taken out, and the remaining power parameters b are output and cleaned.

6. The online evaluation method for CFRP hole making process quality according to claim 1, characterized in that: A power analyzer is used to monitor the processing power of the hole-making equipment. A time window is intercepted through the start processing signal and the end processing signal of the hole-making equipment, so that only the power information in the window is processed and calculated. The processing power data is transmitted to the established hole-making quality evaluation model to output the quality evaluation result of the current drilling.

Images