A method for processing information during oxidation of aircraft parts
The characteristics of aircraft parts are extracted through the two-stage particle size network model, and the unstable rate and oxidation efficiency reference value are calculated, which solves the problems of low efficiency and inconsistent results of existing oxidation monitoring technology, and accurately judges and controls the oxidation status of aircraft parts. Improves the safety and reliability of the aviation manufacturing industry.
Patent Information
- Application Number
- CN202411273001.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-12
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-12
Smart Images

Figure CN118779757B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of aircraft parts oxidation data processing method, in particular to an information processing method in the aircraft parts oxidation process. Background Art
[0002] Oxidation treatment of aircraft parts is a critical link in the aviation manufacturing industry, which is directly related to the performance and safety of the aircraft. The oxidation process usually involves exposing the parts to high temperature and oxygen environment to form a protective oxide layer. This oxide layer can prevent further corrosion and wear, thereby extending the service life of the parts.
[0003] However, existing oxidation monitoring technology mainly relies on manual detection and empirical judgment, which has several limitations. First, manual detection is inefficient and cannot achieve large-scale rapid detection, which makes it difficult to meet the needs of modern aviation manufacturing for high efficiency and mass production. Second, reliance on empirical judgment may lead to subjectivity and inconsistency in test results, affecting the stability and reliability of oxidation quality. In addition, manual detection is difficult to capture subtle changes in the oxidation process, which may lead to neglect of early oxidation anomalies and miss the best time for adjustment and intervention.
[0004] With the development of aviation material science and manufacturing technology, the requirements for the oxidation quality of aircraft parts are getting higher and higher. Traditional oxidation monitoring methods have been unable to meet the needs of high precision and high reliability. For example, in the manufacturing process of certain high-performance aircraft parts, the thickness, uniformity and chemical composition of the oxide layer need to be strictly controlled to ensure that the parts can work normally in extreme environments. However, existing oxidation monitoring technologies often cannot provide sufficient data support and analysis capabilities, making it difficult to achieve these high-precision requirements.
[0005] In addition, existing oxidation monitoring technologies also have deficiencies in data processing and analysis. A large amount of oxidation process data has not been effectively utilized and analyzed, resulting in an inability to deeply understand the inherent laws and influencing factors of the oxidation process. This limits the optimization and improvement of the oxidation process and also affects the prevention and control of oxidation anomalies. Summary of the invention
[0006] In view of the shortcomings of the prior art, the present invention proposes a method for processing information during the oxidation process of aircraft parts. The method processes and analyzes the data during the oxidation process of aircraft parts, realizes accurate judgment and effective control of the oxidation state of parts, and helps to improve the safety and reliability of aircraft. The implementation of the present invention will bring revolutionary technological progress to the aviation manufacturing industry, and has important practical value and broad application prospects.
[0007] The technical solution of the present invention is as follows:
[0008] A method for processing information during oxidation of aircraft parts, comprising:
[0009] Obtain characteristic data of aircraft parts during oxidation process;
[0010] A two-stage granular network model is constructed based on autoencoders and stacked autoencoders, and coarse-grained features and fine-grained features are gradually extracted through the two-stage granular network.
[0011] Based on the extracted coarse-grained features and fine-grained features, the instability rate of aircraft parts during the oxidation process is calculated. According to the instability rate, the parts are divided into reference parts with stable oxidation state and abnormal parts using the 3σ criterion.
[0012] For the reference part, calculate its oxidation efficiency reference value; for the abnormal part, calculate its oxidation efficiency reference value and calculate the ratio with the oxidation efficiency reference value of the reference part;
[0013] Wherein, the calculation formula of the oxidation efficiency reference value is:
[0014] ;
[0015] in, It is the baseline value of oxidation efficiency; is a constant term, representing the basal level of oxidative efficiency; Representative coarse-grained features; Representative Fine-grained features; is the corresponding coarse-grained feature The coefficient of Corresponding to fine-grained features The coefficient of It is a coarse-grained feature and fine-grained features The interaction coefficient between ; is the corresponding coarse-grained feature The square term coefficient of represents the nonlinear effect; Corresponding to fine-grained features The square term coefficient of represents the nonlinear effect; is the total number of coarse-grained features; is the total number of fine-grained features; is the error term, representing random variation that cannot be explained by the model;
[0016] Based on the ratio results, the risk level of abnormal parts is determined; the difference of abnormal parts within the same risk level is calculated to obtain the difference of oxidation efficiency benchmark values of these parts; the control sequence is determined based on the difference of oxidation efficiency benchmark values to realize the division of risk levels of abnormal parts and the control sequence of the same risk level.
[0017] Furthermore, the characteristic data includes temperature, humidity, and oxide layer thickness.
[0018] Furthermore, the autoencoder is used to extract coarse-grained features, and the extracted coarse-grained features include: the morphology of the overall oxide layer and the distribution of the overall oxide layer;
[0019] The input layer of the autoencoder receives characteristic data of aircraft parts during the oxidation process, including temperature (T), humidity (H), and oxide layer thickness (d), and the input dimension is n;
[0020] Hidden layer 1 of the autoencoder: contains m1 neurons, uses the ReLU activation function, and the calculation formula is:
[0021] ;
[0022] in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; Represents the input vector, including temperature T, humidity H, oxide layer thickness d, etc., =[T,H,d,]; is the weight vector, =(m1,n); is the bias vector;
[0023] Hidden layer 2 of the autoencoder: contains m2 neurons, uses the ReLU activation function, and the calculation formula is:
[0024] ;
[0025] in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; is the weight matrix, =(m2,n); is the bias vector;
[0026] The output layer of the autoencoder: The output dimension is the same as the input dimension, and attempts to reconstruct the input data. The calculation formula is:
[0027] ;
[0028] Represents the output reconstruction data; When weight matrix; Represents the coarse-grained features of the output of hidden layer 2 of the autoencoder; is the bias vector;
[0029] Using MSE as the loss function, the calculation formula is:
[0030]
[0031] Represents the loss function value of the autoencoder; represents the number of samples; Indicates The real input data of samples; Represents the model for Output data of samples; Indicates The square of the error between the true input data and the reconstructed data for samples.
[0032] Furthermore, the stacked autoencoder is used to extract fine-grained features; the extracted fine-grained features include: the microstructure of the oxide layer, the difference in the local oxidation degree;
[0033] Among them, the stacked autoencoder is composed of multiple autoencoders stacked in sequence, the output of each autoencoder is used as the input of the next autoencoder, and then the stacked autoencoder is trained to extract more detailed fine-grained features.
[0034] Furthermore, the calculation method of the instability rate of the aircraft parts during the oxidation process is:
[0035] ;
[0036] in, Indicates The instability rate of each feature; is the number of features, For the The weight of the feature, and Respectively The standard deviation and mean of each feature.
[0037] Furthermore, the method for dividing the reference parts with stable oxidation state and the abnormal parts includes:
[0038] Set the threshold to ;
[0039] The instability rate of the parts Compared with the threshold, if ≤ , then the part is classified as a reference part with stable oxidation state; if > , then the part is classified as an abnormal part.
[0040] Furthermore, the calculation formula of the oxidation efficiency reference value ratio of the reference part is: ;in, It is the ratio of the oxidation efficiency reference value of the abnormal part to the oxidation efficiency reference value of the reference part; It is the oxidation efficiency benchmark value of abnormal parts; It is the reference value of oxidation efficiency of the reference part.
[0041] Further, the risk level of the abnormal part is determined by comparing the ratio result with a set maximum threshold and a minimum threshold to determine the risk level of the abnormal part;
[0042] Among them, the risk level is determined as follows:
[0043] When the ratio is less than the minimum threshold, the abnormal part is classified as a low-risk level;
[0044] When the ratio is between the maximum threshold and the minimum threshold, the abnormal part is classified as a medium risk level;
[0045] When the ratio is greater than the maximum threshold, the abnormal parts are classified as high risk levels.
[0046] Furthermore, the calculation formula of the difference is:
[0047] ;
[0048] in, Indicates the degree of difference; For the The oxidation efficiency benchmark value of abnormal parts, is the average value of the oxidation efficiency benchmark within the risk level;
[0049] Using the difference value, abnormal parts within the same risk level are quickly sorted to determine the control sequence.
[0050] The beneficial effects of the present invention are as follows: by extracting coarse-grained features and fine-grained features through a dual-stage granular network model, and combining the calculation of the instability rate and the application of the 3σ criterion, the oxidation state of aircraft parts can be accurately determined, and reference parts and abnormal parts can be distinguished.
[0051] Based on the oxidation efficiency ratio between abnormal parts and reference parts, the risk level of abnormal parts can be accurately determined, and the difference calculation of abnormal parts within the same risk level can help to more finely manage and control the oxidation of parts.
[0052] By determining the control sequence based on the difference in oxidation efficiency benchmark values, it is possible to divide the risk levels of abnormal parts and the control sequences of the same risk level, so as to take more targeted measures, such as maintenance and replacement, to effectively control the oxidation of aircraft parts and improve the safety and reliability of aircraft.
[0053] This method provides a richer and more accurate basis for aircraft maintenance and management, helping to optimize maintenance plans, reduce maintenance costs, and improve aircraft operating efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Figure 1 It is the overall flow chart of the present invention;
[0055] Figure 2 is a flowchart of the specific steps of S200;
[0056] Figure 3 is a flowchart of the specific steps of S300;
[0057] Figure 4 is a flowchart of the specific steps of S400;
[0058] Figure 5 It is a flowchart of the specific steps of S500. DETAILED DESCRIPTION
[0059] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0060] See also Figure 1 , which shows an information processing method in an oxidation process of aircraft parts provided by an embodiment of the present invention, comprising:
[0061] S100: Obtain various characteristic data of aircraft parts during oxidation process;
[0062] S200: A two-stage granular network model is constructed based on autoencoders and stacked autoencoders, and coarse-grained features and fine-grained features are gradually extracted through the two-stage granular network;
[0063] S300: Based on the extracted coarse-grained features and fine-grained features, the instability rate of aircraft parts during the oxidation process is calculated. According to the instability rate, the parts are divided into reference parts with stable oxidation state and abnormal parts using the 3σ criterion.
[0064] S400: For a reference part, calculating its oxidation efficiency reference value; for an abnormal part, calculating its oxidation efficiency reference value and performing a ratio calculation with the oxidation efficiency reference value of the reference part;
[0065] S500: Determine the risk level of abnormal parts based on the ratio result; calculate the difference of abnormal parts within the same risk level to obtain the difference of oxidation efficiency benchmark values of these parts; determine the control sequence based on the difference of oxidation efficiency benchmark values to achieve the division of risk levels of abnormal parts and control sequences of the same risk level.
[0066] The specific contents are as follows:
[0067] The purpose of S100 is to obtain various characteristic data of aircraft parts during the oxidation process, including temperature, humidity, oxide layer thickness, etc.
[0068] Sensors are set at key locations of aircraft parts, such as thermocouples or thermistor sensors for temperature, capacitive or resistive humidity sensors for humidity, and ultrasonic or optical measurement sensors for oxide layer thickness, to accurately obtain data such as temperature, humidity, and oxide layer thickness. At the same time, the frequency of data collection is determined according to the characteristics and requirements of the oxidation process to ensure that changes in the oxidation process can be captured in a timely manner. For example, a temperature sensor is used to measure the temperature on the surface or inside of an aircraft part in real time, and the temperature value and change trend are recorded. The humidity information of the environment around the aircraft part is obtained through a humidity sensor, and the humidity value is recorded. The thickness of the oxide layer is measured using non-destructive testing technology or other appropriate methods, and the thickness value is recorded.
[0069] S200 mainly constructs a two-stage granular network model, and gradually extracts the coarse-grained features and fine-grained features of aircraft parts through this model.
[0070] In some possible implementations, such as Figure 2 As shown, S200 specifically includes:
[0071] S210: Construct a two-stage granular network model. The two-stage granular network model consists of an autoencoder and a stacked autoencoder. The autoencoder is used to extract coarse-grained features. The stacked autoencoder is a stack of multiple autoencoders, which is used to further extract fine-grained features based on coarse-grained features.
[0072] When designing an autoencoder, we first determine the number of layers and the number of neurons in each layer, select the activation function, and then train the autoencoder so that it can learn the main features of the input data.
[0073] In this embodiment, the input layer of the autoencoder receives characteristic data of aircraft parts during the oxidation process, including temperature (T), humidity (H), and oxide layer thickness (d), and the input dimension is n.
[0074] Hidden layer 1 of the autoencoder: contains m1 neurons, uses the ReLU activation function, and the calculation formula is:
[0075] ;
[0076] in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; Represents the input vector, including temperature (T), humidity (H), oxide layer thickness (d), etc., x=[T,H,d,...]; is the weight vector, = (m1, n), used to perform linear transformation on input features; is the bias vector.
[0077] Hidden layer 2 of the autoencoder: contains m2 neurons, uses the ReLU activation function, and the calculation formula is:
[0078] ;
[0079] in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; is the weight matrix, =(m2,n); is the bias vector.
[0080] The output layer of the autoencoder: The output dimension is the same as the input dimension, and attempts to reconstruct the input data. The calculation formula is:
[0081] ;
[0082] Reconstructed data representing the output, i.e., a prediction or reconstruction of the input data; is the weight matrix used to adjust the hidden layer The output of is linearly transformed; Represents the coarse-grained features of the output of hidden layer 2 of the autoencoder; is a bias vector used to adjust the output result.
[0083] When training the autoencoder, MSE is used as the loss function, and the calculation formula is:
[0084]
[0085] Represents the loss function value of the autoencoder; represents the number of samples; Indicates The real input data of samples; Represents the model for The output data of samples, i.e., the reconstructed data; Indicates The square of the error between the true input data and the reconstructed data for samples.
[0086] When designing a stacked autoencoder, multiple autoencoders are stacked in sequence, and the output of each autoencoder is used as the input of the next autoencoder. The stacked autoencoder is then trained to extract more detailed fine-grained features.
[0087] In this embodiment, the first autoencoder structure is similar to the above autoencoder, but the input is the coarse-grained features output by the hidden layer 2 of the autoencoder. The second autoencoder takes the output of the first autoencoder as input to further extract fine-grained features. When training the stacked autoencoder, MSE is also used as the loss function.
[0088] S220: Input various characteristic data of aircraft parts during the oxidation process into the two-stage granular network model.
[0089] S230: Coarse-grained feature extraction. The autoencoder processes the input data and extracts the coarse-grained features of the aircraft parts. The extracted coarse-grained features include: the morphology of the overall oxide layer, such as the shape and flatness of the oxide layer; the distribution of the overall oxide layer, such as the coverage and uniformity of the oxide layer on the surface of the part.
[0090] S240: Fine-grained feature extraction. Use stacked autoencoders to further extract more detailed fine-grained features from coarse-grained features. Through the step-by-step processing of the stacked autoencoder, the fine-grained features of the aircraft parts are finally obtained. Fine-grained features include: the microstructure of the oxide layer, such as the crystal structure and porosity of the oxide layer; the difference in the degree of local oxidation, such as the change in the degree of oxidation in different parts of the parts. Fine-grained features can more accurately describe the oxidation state of the parts, including information such as the microstructure of the oxide layer and the difference in the degree of local oxidation. Fine-grained features provide a richer and more accurate basis for subsequent analysis and decision-making, which helps to more accurately judge the oxidation state of parts, distinguish between baseline parts and abnormal parts, and perform more refined risk level classification and control of abnormal parts.
[0091] S300 is about how to utilize coarse-grained and fine-grained features extracted from a two-stage granular network model to evaluate the stability of aircraft parts during oxidation and perform anomaly detection based on this.
[0092] In some possible embodiments, such as Figure 3 As shown, S300 specifically includes:
[0093] S310: For the extracted coarse-grained and fine-grained features, calculate their averages over a period of time. The average calculation formula is: , represents the number of samples; Indicates The real input data of samples; calculate the standard deviation of each feature to reflect the volatility of the feature. The standard deviation calculation formula is: At the same time, based on the importance or relevance of the features, the weights of the coarse-grained features and the fine-grained features are determined.
[0094] S320: Calculate the instability rate of aircraft parts during oxidation.
[0095] The instability rate calculation formula is: ,in, Indicates The instability rate of each feature; is the number of features, For the The weight of the feature, and Respectively The standard deviation and mean of each feature.
[0096] S330: Use the 3σ criterion to classify the part status. First, according to the 3σ criterion, set the threshold value as . The instability rate of the parts is compared with the threshold. If ≤ , then the part is classified as a reference part with a stable oxidation state. > , then the part is classified as an abnormal part.
[0097] S400 is about how to evaluate and analyze the oxidation efficiency of the reference parts and abnormal parts by calculating the oxidation efficiency reference values of the reference parts and the abnormal parts, and comparing the oxidation efficiency reference value of the reference parts with the oxidation efficiency reference value of the abnormal parts.
[0098] In some possible embodiments, such as Figure 4 As shown, S400 specifically includes:
[0099] S410: Calculate the oxidation efficiency baseline values of the baseline parts and the abnormal parts, and the calculation formula is:
[0100] ;
[0101] in, It is the baseline value of oxidation efficiency; is a constant term, representing the basal level of oxidative efficiency; Representative coarse-grained features; Representative Fine-grained features; is the corresponding coarse-grained feature The coefficient of Corresponding to fine-grained features The coefficient of It is a coarse-grained feature and fine-grained features The interaction coefficient between ; is the corresponding coarse-grained feature The square term coefficient of represents the nonlinear effect; Corresponding to fine-grained features The square term coefficient of represents the nonlinear effect; is the total number of coarse-grained features; is the total number of fine-grained features; is the error term, representing the random variation that cannot be explained by the model.
[0102] S420: Calculate the ratio of the oxidation efficiency reference value of the abnormal part to the oxidation efficiency reference value of the reference part.
[0103] The ratio calculation formula is:
[0104] ;
[0105] in, It is the ratio of the oxidation efficiency reference value of the abnormal part to the oxidation efficiency reference value of the reference part; It is the oxidation efficiency benchmark value of abnormal parts; It is the reference value of oxidation efficiency of the reference part.
[0106] S500 is a process about how to determine the risk level of abnormal parts based on the oxidation efficiency ratio between abnormal parts and reference parts, and calculate the difference within the risk level and divide the control sequence.
[0107] In some possible embodiments, such as Figure 5 As shown, S500 specifically includes:
[0108] S510: Determine the risk level of the abnormal part by comparing the ratio result calculated in S400 with the set maximum threshold and minimum threshold.
[0109] The risk levels are determined as follows:
[0110] When the ratio is less than the minimum threshold, the abnormal part is classified as a low-risk level;
[0111] When the ratio is between the maximum threshold and the minimum threshold, the abnormal part is classified as a medium risk level;
[0112] When the ratio is greater than the maximum threshold, the abnormal parts are classified as high risk levels.
[0113] S520: For abnormal parts within the same risk level, calculate the difference between their oxidation efficiency baseline values.
[0114] The calculation formula of difference is:
[0115] ;
[0116] in, Indicates the degree of difference; For the The oxidation efficiency benchmark value of abnormal parts, It is the average value of the oxidation efficiency benchmark within this risk level.
[0117] S530: Use the difference value to quickly sort the abnormal parts within the same risk level and determine the control sequence. At the same time, manage and control the oxidation of aircraft parts based on the risk level and control sequence.
[0118] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.
[0119] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.
[0120] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.
[0121] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features of the present invention.
[0122] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. A method for processing information during oxidation of aircraft parts, characterized in that: include: Obtain characteristic data of aircraft parts during oxidation process; A two-stage granular network model is constructed based on autoencoders and stacked autoencoders, and coarse-grained features and fine-grained features are gradually extracted through the two-stage granular network. Based on the extracted coarse-grained features and fine-grained features, the instability rate of aircraft parts during oxidation is calculated. According to the instability rate, the parts are divided into reference parts with stable oxidation state and abnormal parts using the 3σ criterion. For the reference part, calculate its oxidation efficiency reference value; for the abnormal part, calculate its oxidation efficiency reference value and calculate the ratio with the oxidation efficiency reference value of the reference part; Wherein, the calculation formula of the oxidation efficiency reference value is: ; in, It is the baseline value of oxidation efficiency; is a constant term, representing the basal level of oxidative efficiency; Representative coarse-grained features; Representative Fine-grained features; is the corresponding coarse-grained feature The coefficient of Corresponding to fine-grained features The coefficient of It is a coarse-grained feature and fine-grained features The interaction coefficient between ; is the corresponding coarse-grained feature The square term coefficient of represents the nonlinear effect; Corresponding to fine-grained features The square term coefficient of represents the nonlinear effect; is the total number of coarse-grained features; is the total number of fine-grained features; is the error term, representing random variation that cannot be explained by the model; Based on the ratio results, the risk level of abnormal parts is determined; the difference of abnormal parts within the same risk level is calculated to obtain the difference of oxidation efficiency benchmark values of these parts; the control sequence is determined based on the difference of oxidation efficiency benchmark values to realize the division of risk levels of abnormal parts and the control sequence of the same risk level.
2. The information processing method in the oxidation process of aircraft parts according to claim 1 is characterized in that: The characteristic data include temperature, humidity, and oxide layer thickness.
3. The information processing method in the oxidation process of aircraft parts according to claim 1 is characterized in that: The autoencoder is used to extract coarse-grained features, and the extracted coarse-grained features include: the morphology of the overall oxide layer and the distribution of the overall oxide layer; The input layer of the autoencoder receives characteristic data of aircraft parts during the oxidation process, including temperature (T), humidity (H), and oxide layer thickness (d), and the input dimension is n; Hidden layer 1 of the autoencoder: contains m1 neurons, uses the ReLU activation function, and the calculation formula is: ; in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; Represents the input vector, including temperature T, humidity H, oxide layer thickness d, etc., =[T,H,d,]; is the weight vector, =(m1,n); is the bias vector; Hidden layer 2 of the autoencoder: contains m2 neurons, uses the ReLU activation function, and the calculation formula is: ; in, It is the coarse-grained feature output by the hidden layer 1 of the autoencoder; is the weight matrix, =(m2,n); is the bias vector; The output layer of the autoencoder: The output dimension is the same as the input dimension, and attempts to reconstruct the input data. The calculation formula is: ; Represents the output reconstruction data; When weight matrix; Represents the coarse-grained features of the output of hidden layer 2 of the autoencoder; is the bias vector; Using MSE as the loss function, the calculation formula is: ; Represents the loss function value of the autoencoder; represents the number of samples; Indicates The real input data of samples; Represents the model for Output data of samples; Indicates The square of the error between the true input data and the reconstructed data for samples.
4. The information processing method in the oxidation process of aircraft parts according to claim 1 is characterized in that: The stacked autoencoder is used to extract fine-grained features; The fine-grained features extracted include: the microstructure of the oxide layer, differences in the local oxidation degree; Among them, the stacked autoencoder is composed of multiple autoencoders stacked in sequence, the output of each autoencoder is used as the input of the next autoencoder, and then the stacked autoencoder is trained to extract more detailed fine-grained features.
5. The information processing method in the oxidation process of aircraft parts according to claim 1 is characterized in that: The calculation method of the instability rate of the aircraft parts during the oxidation process is: ; in, Indicates The instability rate of each feature; is the number of features, For the The weight of the feature, and Respectively The standard deviation and mean of each feature.
6. The information processing method in the oxidation process of aircraft parts according to claim 5 is characterized in that: The method for dividing the reference parts with stable oxidation state and the abnormal parts comprises: Set the threshold to ; The instability rate of the parts Compared with the threshold, if ≤ , then the part is classified as a reference part with stable oxidation state; if > , then the part is classified as an abnormal part.
7. The information processing method in the oxidation process of aircraft parts according to claim 1 is characterized in that: The calculation formula for the oxidation efficiency benchmark ratio of the benchmark part is: ;in, It is the ratio of the oxidation efficiency reference value of the abnormal part to the oxidation efficiency reference value of the reference part; It is the oxidation efficiency benchmark value of abnormal parts; It is the reference value of oxidation performance of the reference part.
8. The information processing method in the oxidation process of aircraft parts according to claim 7 is characterized in that: The risk level of the abnormal part is determined by comparing the ratio result with the set maximum threshold and minimum threshold to determine the risk level of the abnormal part; Among them, the risk level is determined as follows: When the ratio is less than the minimum threshold, the abnormal part is classified as a low-risk level; When the ratio is between the maximum threshold and the minimum threshold, the abnormal part is classified as a medium risk level; When the ratio is greater than the maximum threshold, the abnormal parts are classified as high risk levels.
9. The information processing method in the oxidation process of aircraft parts according to claim 8 is characterized in that: The calculation formula of the difference is: ; in, Indicates the degree of difference; For the The oxidation efficiency benchmark value of abnormal parts, is the average value of the oxidation efficiency benchmark within the risk level; Using the difference value, abnormal parts within the same risk level are quickly sorted to determine the control sequence.
