CN117574581A - Temperature-sensitive scroll spring thermal stress relaxation prediction method and system - Google Patents
Temperature-sensitive scroll spring thermal stress relaxation prediction method and system Download PDFInfo
- Publication number
- CN117574581A CN117574581A CN202410049991.6A CN202410049991A CN117574581A CN 117574581 A CN117574581 A CN 117574581A CN 202410049991 A CN202410049991 A CN 202410049991A CN 117574581 A CN117574581 A CN 117574581A
- Authority
- CN
- China
- Prior art keywords
- model
- neural network
- data
- memory neural
- long
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000008646 thermal stress Effects 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 26
- 230000015654 memory Effects 0.000 claims abstract description 64
- 238000012549 training Methods 0.000 claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 claims abstract description 38
- 238000003062 neural network model Methods 0.000 claims abstract description 24
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000002360 preparation method Methods 0.000 claims abstract description 10
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 238000011156 evaluation Methods 0.000 claims abstract description 7
- 238000012800 visualization Methods 0.000 claims abstract description 7
- 230000006399 behavior Effects 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 8
- 235000015220 hamburgers Nutrition 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000006403 short-term memory Effects 0.000 claims description 3
- 230000007787 long-term memory Effects 0.000 description 2
- 238000011089 mechanical engineering Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000009662 stress testing Methods 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/10—Noise analysis or noise optimisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Computer Hardware Design (AREA)
- Mathematical Analysis (AREA)
- Medical Informatics (AREA)
- Computational Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses a temperature-sensitive scroll spring thermal stress relaxation prediction method and a temperature-sensitive scroll spring thermal stress relaxation prediction system, wherein the method comprises the following steps: step 1: data preparation and preprocessing; step 2: extracting key features from the standardized data; step 3: preparing long-short-term memory neural network training data; step 4: defining a temperature sensitive long-short-time memory neural network unit; step 5: defining a self-attention mechanism; step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism; step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data; step 8: model evaluation and prediction, evaluating a model on a test set different from the training set; step 9: result analysis and visualization.
Description
Technical Field
The present invention relates to the field of material science and mechanical engineering, and more particularly to a method and system for predicting thermal stress relaxation of a spiral spring under different temperature conditions using a temperature-sensitive long and short time memory network (long and short time memory neural network) and a self-attention mechanism.
Background
Spiral springs are key components for wide-ranging applications in various machines and structures, whose performance at different temperatures is critical. Currently, predictions of thermal stress relaxation for wrap springs rely primarily on traditional empirical formulas and physical models, which tend to be inflexible and have limited accuracy in handling complex environments. Therefore, it is important to develop an efficient and accurate prediction method.
Disclosure of Invention
In order to achieve the above object, the present invention provides a temperature-sensitive method for predicting thermal stress relaxation of a spiral spring, which is characterized in that the method comprises:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data;
step 8: model evaluation and prediction, evaluating a model on a test set different from the training set;
step 9: result analysis and visualization.
In a preferred embodiment, the data preparation and preprocessing specifically comprises:
the data are loaded from the experimental data of the spiral spring under various temperature conditions, and the data are normalized by using maximum and minimum value normalization, so that the uniformity and stability of model input are ensured.
In a preferred embodiment, the key features include a time step and a temperature, wherein the time step is defined to construct time series data, which is used to train the LSTM network.
In a preferred embodiment, preparing long-term memory neural network training data specifically includes:
according to the defined time step, long-short memory neural network training data are prepared from normalized data, and a series of past n steps of data and the target data of the next step are created from time sequence data;
a time window is set for slicing the dataset and creating sequences for model training, each sequence having a target value of the thermal stress value after the last time step of the sequence for training the model predicted thermal stress.
In a preferred embodiment, defining the temperature-sensitive long-time memory neural network unit specifically includes:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
In a preferred embodiment, defining the self-attention mechanism specifically includes:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of the model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
In a preferred embodiment, constructing the overall long short-term memory neural network model specifically includes:
the temperature sensitive long-short time memory neural network unit is combined with the standard LSTM framework, so that the model can dynamically adjust the state of the model according to time step and temperature information in input data, the self-attention mechanism defined in the step 5 is integrated, attention and processing capacity of the model to key moments in a sequence are enhanced, and finally a full-connection layer is arranged on the model for generating final prediction output.
The invention provides a temperature-sensitive spiral spring thermal stress relaxation prediction system, which is characterized by comprising: a processor and a memory, wherein the memory has stored therein processor-executable instructions that, when executed by the processor, cause the processor to:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data;
step 8: model evaluation and prediction, evaluating a model on a test set different from the training set;
step 9: result analysis and visualization.
In a preferred embodiment, defining the temperature-sensitive long-time memory neural network unit specifically includes:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
In a preferred embodiment, defining the self-attention mechanism specifically includes:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of the model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
Compared with the prior art, the method has the advantages that the accuracy and the reliability of the thermal stress relaxation prediction of the scroll spring under different temperature conditions are improved, and the method has better adaptability and flexibility due to the characteristic of data driving, and is suitable for practical industrial production environments, particularly the application in the fields of mechanical engineering, material science and the like.
Drawings
FIG. 1 is a method flow diagram of one embodiment of the present invention.
FIG. 2 is a graph comparing actual thermal stress with predicted thermal stress at 25℃according to the present invention.
FIG. 3 is a graph comparing actual thermal stress with predicted thermal stress at 150℃according to the present invention.
FIG. 4 is a graph comparing actual thermal stress with predicted thermal stress at 250℃according to the present invention.
FIG. 5 is a bar graph of model predictive accuracy at different temperatures in accordance with the present invention.
Description of the embodiments
The following detailed description of embodiments of the invention is, therefore, to be taken in conjunction with the accompanying drawings, and it is to be understood that the scope of the invention is not limited to the specific embodiments.
Throughout the specification and claims, unless explicitly stated otherwise, the term "comprise" or variations thereof such as "comprises" or "comprising", etc. will be understood to include the stated element or component without excluding other elements or components.
Example 1
As shown in fig. 1, the method of the present invention comprises the steps of:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data; the model is evaluated on a test set different from the training set. The thermal stress relaxation of the wrap spring at 25 ℃ was predicted using a model.
Step 8: model evaluation and prediction, evaluating a model on a test set different from the training set; the model predicted thermal stress is analyzed and compared with the actual measured value. A graph is used to visualize the comparison between the actual and predicted values.
Step 9: result analysis and visualization.
In a preferred embodiment, the data preparation and preprocessing specifically comprises:
the data are loaded from the experimental data of the spiral spring under various temperature conditions, and the data are normalized by using maximum and minimum value normalization, so that the uniformity and stability of model input are ensured.
In a preferred embodiment, the key features include a time step and a temperature, wherein the time step is defined to construct time series data, which is used to train the LSTM network.
In a preferred embodiment, preparing long-term memory neural network training data specifically includes:
according to the defined time step, long-short memory neural network training data are prepared from normalized data, and a series of past n steps of data and the target data of the next step are created from time sequence data;
a time window (e.g., 3 time steps) is set, one for slicing the data set and creating sequences for model training, each sequence targeting a thermal stress value after the last time step of the sequence for training the model predicted thermal stress.
Where n is, for example, 2, 3, 4, 5, n may of course take on even larger values, e.g. 10, 11, 12, 13, depending on the situation and need, and in other examples n may have larger values, e.g. any integer value from 20-100, without being particularly limited thereto, and the same will be understood for the following n.
In a preferred embodiment, defining the temperature-sensitive long-time memory neural network unit specifically includes:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
In a preferred embodiment, defining the self-attention mechanism specifically includes:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of the model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
In a preferred embodiment, constructing the overall long short-term memory neural network model specifically includes:
the temperature sensitive long-short time memory neural network unit is combined with the standard LSTM framework, so that the model can dynamically adjust the state of the model according to time step and temperature information in input data, the self-attention mechanism defined in the step 5 is integrated, attention and processing capacity of the model to key moments in a sequence are enhanced, and finally a full-connection layer is arranged on the model for generating final prediction output.
Example 2
The invention provides a temperature-sensitive spiral spring thermal stress relaxation prediction system, which is characterized by comprising: a processor and a memory, wherein the memory has stored therein processor-executable instructions that, when executed by the processor, cause the processor to:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data;
step 8: model evaluation and prediction, evaluating a model on a test set different from the training set;
step 9: result analysis and visualization.
In a preferred embodiment, defining the temperature-sensitive long-time memory neural network unit specifically includes:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
In a preferred embodiment, defining the self-attention mechanism specifically includes:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of the model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
Example 3
The embodiment of the invention provides a scroll spring thermal stress relaxation prediction method based on a temperature sensitive long-short-term memory neural network (LSTM) and a self-attention mechanism. In order to verify the effectiveness of the proposed method, the following experiments were performed. Experimental data were collected from thermal stress testing of spiral springs at different temperature conditions (150 ℃, 250 ℃, 300 ℃ and 25 ℃), model was trained with data at 150 ℃, 250 ℃, 300 ℃ and data at 25 ℃ was used for testing, the prediction results of the model were as follows:
FIG. 2 is a graph comparing actual thermal stress with predicted thermal stress at 25℃according to the present invention. It can be seen that the proposed prediction method can accurately capture the change trend of thermal stress along with time, and the prediction method is well matched with an actual measurement value.
FIG. 3 is a graph comparing actual thermal stress with predicted thermal stress at 150℃according to the present invention. The prediction curve closely follows the actual thermal stress variation despite the temperature rise, showing the sensitivity and adaptability of the model to thermal stress variation at high temperatures.
FIG. 4 is a graph comparing actual thermal stress with predicted thermal stress at 250℃according to the present invention. As shown in FIG. 4, the model also shows higher prediction accuracy, and the prediction capability of the model on the thermal stress relaxation behavior of the spiral spring at different temperatures is further verified.
Finally, fig. 5 presents a comparison of model prediction accuracy at different temperatures. The prediction performance of the model under various temperature conditions can be intuitively seen by calculating the average absolute error (MAE) between the predicted value and the actual value of the model, which shows that the method has good prediction precision under various temperature conditions.
In summary, the prediction method combining the temperature sensitive LSTM and the self-attention mechanism provided by the invention can effectively predict the thermal stress relaxation behavior of the scroll spring under different temperature conditions, and has good practicability and wide application prospect.
Claims (10)
1. A method for predicting thermal stress relaxation of a temperature sensitive wrap spring, the method comprising:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data;
step 8: model evaluation and prediction, evaluating a model on a test set different from the training set;
step 9: result analysis and visualization.
2. The method according to claim 1, wherein the data preparation and preprocessing specifically comprises:
the data are loaded from the experimental data of the spiral spring under various temperature conditions, and the data are normalized by using maximum and minimum value normalization, so that the uniformity and stability of model input are ensured.
3. The method of claim 2, wherein the key features include a time step and a temperature, wherein the time step is defined to construct time series data, the time series data being used to train the LSTM network.
4. The method of claim 3, wherein preparing long-short term memory neural network training data specifically comprises:
according to the defined time step, long-short memory neural network training data are prepared from normalized data, and a series of past n steps of data and the target data of the next step are created from time sequence data;
a time window is set for slicing the dataset and creating sequences for model training, each sequence having a target value of thermal stress value after the last time step of the sequence for training the model predicted thermal stress.
5. The method of claim 4, wherein defining the temperature-sensitive long-short-term memory neural network element specifically comprises:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
6. The method of claim 5, wherein defining the self-attention mechanism specifically comprises:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of a model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
7. The method of claim 6, wherein constructing the overall long short term memory neural network model specifically comprises:
the temperature sensitive long-short time memory neural network unit is combined with the standard LSTM framework, so that the model can dynamically adjust the state of the model according to time step and temperature information in input data, the self-attention mechanism defined in the step 5 is integrated, attention and processing capacity of the model to key moments in a sequence are enhanced, and finally a full-connection layer is arranged on the model for generating final prediction output.
8. A temperature sensitive wrap spring thermal stress relaxation prediction system, said system comprising: a processor and a memory, wherein the memory has stored therein processor-executable instructions that, when executed by the processor, cause the processor to:
step 1: data preparation and preprocessing;
step 2: extracting key features from the standardized data;
step 3: preparing long-short-term memory neural network training data;
step 4: defining a temperature sensitive long-short-time memory neural network unit;
step 5: defining a self-attention mechanism;
step 6: constructing an overall long-short-time memory neural network model, wherein the overall long-short-time memory neural network model comprises a self-defined long-short-time memory neural network unit and a self-attention mechanism;
step 7: training a model, namely training a long-time memory neural network model by using the preprocessed data;
step 8: model evaluation and prediction, evaluating a model on a test set different from the training set;
step 9: result analysis and visualization.
9. The system of claim 8, wherein defining the temperature sensitive long and short time memory neural network element specifically comprises:
designing a custom LSTM unit based on a Burgers viscoelastic model, wherein the LSTM unit is used for dynamically adjusting the internal state of the LSTM unit according to time step and temperature information in input data;
introducing a relaxation time tau and a creep factor alpha, wherein the relaxation time tau and the creep factor alpha are used for simulating the material behavior;
the state output beta of the LSTM unit is adjusted according to the state adjustment formula,
wherein, the state adjustment formula is:
，
where t represents the time step from the beginning of the sequence.
10. The system of claim 9, wherein defining the self-attention mechanism specifically comprises:
implementing a self-attention module for identifying and assigning higher weights to the most important parts of the sequence data;
setting the size of a hidden layer, constructing a linear layer based on the size of the hidden layer for generating attention scores, and simultaneously improving the stability of a model by using layer standardization;
the self-attention mechanism generates attention weights through a linear layer and applies these weights to calculate a weighted average of the sequences;
the self-attention mechanism is calculated by the following formula:
，
q is a query vector representing the current input element;
a key vector representing a sequence element to be compared with the current element;
v, value vector, representing actual sequence element, its correspondent weight is determined by matching degree between inquiry and key;
the dimension of the key vector is used to scale the dot product to prevent it from becoming too large.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410049991.6A CN117574581B (en) | 2024-01-15 | 2024-01-15 | Temperature-sensitive scroll spring thermal stress relaxation prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410049991.6A CN117574581B (en) | 2024-01-15 | 2024-01-15 | Temperature-sensitive scroll spring thermal stress relaxation prediction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117574581A true CN117574581A (en) | 2024-02-20 |
| CN117574581B CN117574581B (en) | 2024-03-22 |
Family
ID=89864541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410049991.6A Active CN117574581B (en) | 2024-01-15 | 2024-01-15 | Temperature-sensitive scroll spring thermal stress relaxation prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117574581B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102507114A (en) * | 2011-10-19 | 2012-06-20 | 天津大学 | Method for predicting relaxation life of spiral spring on basis of stress relaxation theory of temperature accelerating material |
| CN105910883A (en) * | 2016-04-11 | 2016-08-31 | 天津大学 | Predication method of stress relaxation life of torsion spring |
| DE102021000099A1 (en) * | 2021-01-12 | 2021-03-18 | Thomas Schmidt | Method for determining mechanical stresses in high-temperature pipelines, taking relaxation effects into account |
| CN116991981A (en) * | 2023-09-27 | 2023-11-03 | 深圳市特区建工科工集团盛腾科技有限公司 | Method and system for processing simulation data of prestress component by combining finite element analysis |
-
2024
- 2024-01-15 CN CN202410049991.6A patent/CN117574581B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102507114A (en) * | 2011-10-19 | 2012-06-20 | 天津大学 | Method for predicting relaxation life of spiral spring on basis of stress relaxation theory of temperature accelerating material |
| CN105910883A (en) * | 2016-04-11 | 2016-08-31 | 天津大学 | Predication method of stress relaxation life of torsion spring |
| DE102021000099A1 (en) * | 2021-01-12 | 2021-03-18 | Thomas Schmidt | Method for determining mechanical stresses in high-temperature pipelines, taking relaxation effects into account |
| CN116991981A (en) * | 2023-09-27 | 2023-11-03 | 深圳市特区建工科工集团盛腾科技有限公司 | Method and system for processing simulation data of prestress component by combining finite element analysis |
Non-Patent Citations (1)
| Title |
|---|
| 魏芳荣: "螺旋压缩弹簧应力松弛的动态研究及服役寿命预测", 中国优秀硕士学位论文全文数据库 工程科技II辑, no. 4, 15 April 2009 (2009-04-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117574581B (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3575892B1 (en) | Model parameter value estimation device and estimation method, program, recording medium with program recorded thereto, and model parameter value estimation system | |
| Kong et al. | Two-phase degradation process model with abrupt jump at change point governed by Wiener process | |
| EP3191797B1 (en) | Gas turbine sensor failure detection utilizing a sparse coding methodology | |
| Si et al. | Nonlinear degradation modeling and prognostics: A Box-Cox transformation perspective | |
| Cheung et al. | Simulation of fault impacts for vapor compression systems by inverse modeling. Part I: Component modeling and validation | |
| CN107710089B (en) | Plant equipment diagnosis device and plant equipment diagnosis method | |
| JP2013257251A (en) | Anomaly detection method, program, and system | |
| CN116432091B (en) | Equipment fault diagnosis method based on small sample, construction method and device of model | |
| JP2010277577A (en) | Real-time scheduling of linear model for control and estimation | |
| CN110348055B (en) | Method for obtaining and optimizing material parameters of Chaboche viscoplasticity constitutive model | |
| Zhang et al. | A novel approach for analog circuit fault prognostics based on improved RVM | |
| Sun et al. | Accelerated degradation process analysis based on the nonlinear Wiener process with covariates and random effects | |
| US6678627B1 (en) | Computer program and method for determination of electronic circuit card product life capability under exposure to vibration | |
| Zhuang et al. | Remaining useful life prediction of a mechanism considering wear correlation of multiple joints | |
| Singh | An additive Perks–Weibull model with Bathtub-shaped hazard rate function | |
| CN117574581B (en) | Temperature-sensitive scroll spring thermal stress relaxation prediction method and system | |
| CN107924182A (en) | The control method of monitoring arrangement and monitoring arrangement | |
| Chen et al. | Training material models using gradient descent algorithms | |
| Yousef et al. | Estimation of Gumbel parameters under ranked set sampling | |
| CN117113102B (en) | Electronic component life prediction method | |
| Yun et al. | An efficient method for estimating the parameter global reliability sensitivity analysis by innovative single-loop process and embedded Kriging model | |
| US10845792B2 (en) | Failure diagnosis apparatus, monitoring apparatus, failure diagnosis method and recording medium | |
| Chao et al. | Calibration and uncertainty quantification of gas turbine performance models | |
| Zhang et al. | A set of novel correlation tests for nonlinear system variables | |
| CN114236272B (en) | Intelligent detection system of electronic product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |