Cold rolling oiling device oiling analysis scoring method based on support vector regression algorithm
The oil coating scoring model is constructed through the support vector regression algorithm, which solves the problem of inaccurate oil coating quality scoring in the prior art, and realizes high-precision oil coating scoring prediction, which is suitable for complex nonlinear data processing.
Patent Information
- Application Number
- CN202311002618.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-10
- Publication Date
- 2025-02-18
AI Technical Summary
It is difficult for the prior art to accurately score the oil coating quality of cold-rolled strip oil coating machine devices. Traditional methods rely on empirical judgment or simple rules and cannot effectively deal with complex nonlinear factors and interactions.
Using a method based on the support vector regression algorithm, a model that can predict oil coating scores with high precision is constructed through the steps of data preprocessing, model establishment and training, model evaluation and score prediction.
It realizes high-precision oil-coating score prediction, has strong modeling and prediction capabilities, can process complex nonlinear data, avoid overfitting and underfitting, and has a wide range of applications.
Smart Images
Figure CN119477017A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a scoring analysis technology for a cold-rolled strip oiling machine device, and more specifically, to an oiling analysis and scoring method for a cold-rolled strip oiling machine device based on a support vector regression algorithm. Background Art
[0002] The cold rolled strip oiler is a device that applies oil to the surface of the strip after the cold rolling process. The importance of oiling quality to the strip is mainly reflected in the following aspects:
[0003] 1) Protect the surface of the strip: A thin film can be formed by oiling, which can effectively prevent the strip surface from being oxidized and adsorbed by air and other substances in the processing process before being sent to the next process.
[0004] 2) Efficient conventional processing: Applying a suitable oiling agent can improve the processing efficiency of the subsequent process and reduce the processing time.
[0005] 3) Improve the quality of finished products: Very fine oiling agents can avoid scratches during the cutting process that affect the appearance quality.
[0006] The oiling score can be used to indirectly judge the quality of strip steel at the factory. Since the oiled product is only a small link in the strip steel manufacturing process, the overall quality level of strip steel production cannot be reflected from the perspective of the oiling step alone. However, the advantages and disadvantages of similar products can be evaluated through certain basis and standards.
[0007] The oiling score is related to the quality of the strip and the surface defects. The better the oiling quality, the fewer surface defects the product may have, and the higher the oiling score. Therefore, in the entire process of strip production, the oiling score is one of the indicators that indirectly reflects the quality level of strip production, and is used together with other indicators to evaluate the quality of strip products.
[0008] There are complex nonlinear factors and interactions in the process of oil coating machines, and traditional oil coating scoring methods usually require empirical judgment or set scoring criteria based on some simple rules, which cannot give true and accurate results. Summary of the invention
[0009] In view of the defects existing in the prior art, the purpose of the present invention is to provide a cold rolling oiling device oiling analysis scoring method based on a support vector regression algorithm, which can fit the optimal prediction according to the actual data set and has strong modeling and prediction capabilities.
[0010] To achieve the above object, the present invention adopts the following technical solution:
[0011] A cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm comprises the following steps:
[0012] S1, data preprocessing;
[0013] S2, build the model and train it;
[0014] S3, model evaluation;
[0015] S4. Use the model to score the oil application.
[0016] Preferably, the data preprocessing in step S1 specifically includes the following steps:
[0017] S11. Collect data;
[0018] S12, data cleaning;
[0019] S13, extracting labels;
[0020] S14, selecting features;
[0021] S15, perform normalization processing.
[0022] Preferably, the step S11 of collecting data further comprises:
[0023] The data including machine status, operating parameters and machine performance information are obtained from the production process of the cold rolling oiling device.
[0024] Preferably, the data specifically include inverter output current, inverter fault information, actual speed of metering pump, temperature of metering pump, oil tank level, flow meter flow, unit strip speed, oiling set value, and high voltage generator status.
[0025] Preferably, the step S12 of data cleaning further comprises:
[0026] Clean invalid or erroneous data and handle missing values and outliers.
[0027] Preferably, the step S13 of extracting tags further comprises:
[0028] The score of the cold rolling oiling device is extracted from the data set as the label to be predicted.
[0029] Preferably, the step S14 of selecting features further includes:
[0030] The inverter output current, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling set value and the high-voltage generator state are selected as input features.
[0031] Preferably, the normalization process in step S15 further includes:
[0032] The inverter output current, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling setting value and the high-voltage generator state are numerically normalized so that the values are between 0 and 1. The specific calculation formula is:
[0033]
[0034] Where I is the original current value, I max and I min are the maximum and minimum current values respectively.
[0035] Preferably, the step S2 of establishing a model and performing training specifically includes the following steps:
[0036] S21, select model;
[0037] S22, dividing data;
[0038] S23, training model;
[0039] S24. Model adjustment.
[0040] Preferably, the step S21 of selecting a model further comprises:
[0041] The support vector regression model was selected as the machine learning model.
[0042] Preferably, the step S22 of dividing the data further comprises:
[0043] The normalized data is randomly divided into training data set and test data set.
[0044] Preferably, the step S23 of training the model further comprises:
[0045] The training data set is used to train the support vector regression model.
[0046] Preferably, the calculation formula of the support vector regression model is as follows:
[0047]
[0048] Where x is the input variable vector, x i is the sample vector in the training data set, ai is the coefficient of the support vector, K(x,xi) is the kernel function, and b is the bias top;
[0049] The kernel function is selected from a linear kernel function, a polynomial kernel function or a Gaussian kernel function.
[0050] Preferably, the model adjustment in step S24 further includes:
[0051] Observe the training effect of the support vector regression model and adjust and optimize parameters as needed.
[0052] Preferably, the step S3 model evaluation specifically includes the following steps:
[0053] S31, calculating the prediction error;
[0054] S32, calculating the coefficient of determination;
[0055] S33, iterative optimization.
[0056] Preferably, the step S31 of calculating the prediction error further comprises:
[0057] The trained support vector regression model is applied to the test data set, the prediction ability of the support vector regression model is evaluated, and the error between the predicted score and the actual score is calculated.
[0058] Preferably, the step S32 of calculating the determination coefficient further comprises:
[0059] Calculate the coefficient of determination for the test set data.
[0060] Preferably, the iterative optimization in step S33 further comprises:
[0061] If the performance of the support vector regression model is unacceptable or there is room for further optimization, it is necessary to readjust the parameters of the support vector regression model or select another model, and iterate repeatedly until satisfaction is achieved.
[0062] Preferably, the step S4 of using the model to perform oiling scoring specifically includes the following steps:
[0063] S41, data input;
[0064] S42, predicting a new score;
[0065] S43: Inverse normalization processing.
[0066] Preferably, the data input in step S41 further includes:
[0067] The inverter output current of the new oiler device, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling set value and the high-voltage generator status are input into the optimized support vector regression model.
[0068] Preferably, the step S42 of predicting a new score further comprises:
[0069] The support vector regression model gives a score for predicting the new oiler device based on the input parameters.
[0070] Preferably, the inverse normalization process in step S43 further includes:
[0071] The score value of the oiler device predicted in step S42 is inversely normalized to obtain the true score of the oiler device; the calculation formula for inverse normalization is as follows:
[0072] score=score norm (score max -score min )+score min
[0073] In the formula, score norm is the predicted score value, score max and score min are the maximum and minimum score values respectively.
[0074] The present invention provides a cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm, which has the following beneficial effects:
[0075] 1) High-precision prediction: SVR is a very powerful prediction model because it can accurately determine the function based on the distribution of the sample data set, thereby producing highly accurate prediction results.
[0076] 2) Strong generalization ability: SVR has very good generalization ability and can handle high-dimensional and complex data sets while avoiding overfitting and underfitting.
[0077] 3) Handling outliers: In actual situations, there may be some outliers or noise data in the data set due to various reasons. The SVR algorithm can effectively handle these outliers and still maintain a high prediction accuracy.
[0078] 4) Wide range of applications: The SVR algorithm can be applied to various types of data sets, and the corresponding parameters can be adjusted as needed to adapt to different prediction requirements and data set characteristics. BRIEF DESCRIPTION OF THE DRAWINGS
[0079] Figure 1 It is a schematic flow chart of the oiling analysis and scoring method of the cold rolling oiling device of the present invention. DETAILED DESCRIPTION
[0080] In order to better understand the above technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0081] Support vector regression (SVR) is a type of algorithm for regression analysis based on support vector machines. Different from the traditional linear regression model, SVR uses kernel functions to map data to a high-dimensional space and constructs a hyperplane in this space to predict the value of continuous variables.
[0082] The overall idea of SVR is to transform the training samples into a high-dimensional feature space and find an optimal hyperplane to minimize the error between all training samples and the hyperplane while avoiding overfitting. Specifically, the objective function to be optimized is:
[0083]
[0084] Where C is the regularization coefficient, which is used to control the complexity of the model; i and It represents the difference between the actual value and the model predicted value and the true value, and approximates the nonlinear function by setting different kernel functions.
[0085] Combination Figure 1 As shown, the present invention provides a cold rolling oiling device oiling analysis and scoring method based on a support vector regression algorithm, comprising the following steps:
[0086] S1. Data preprocessing: It is necessary to extract the corresponding label value (score of the oiling machine device) from the existing data set for model training. Here, 9 key parameters (inverter output current, inverter fault information, metering pump actual speed, metering pump temperature, oil tank level, pipe clamp flowmeter flow, unit strip speed, oiling set value, high voltage generator status) are selected as input variables and normalized. Specifically, the following steps are included:
[0087] S11. Collect data: obtain data including machine status, operating parameters, and machine performance information from the production process of the cold rolling oiling device.
[0088] S12. Data cleaning: Clear invalid or erroneous data, process missing values and outliers, and ensure data integrity and accuracy.
[0089] S13. Extract labels: extract the scores of the cold rolling oiling device from the data set as the target (label) to be predicted.
[0090] S14. Select features: select the inverter output current, inverter fault information, actual speed of the metering pump, temperature of the metering pump, oil tank level, flow meter flow, unit strip speed, oiling set value and high voltage generator status as input features.
[0091] S15. Perform normalization processing: perform numerical normalization processing on the inverter output current, inverter fault information, actual speed of the metering pump, temperature of the metering pump, oil tank level, flow meter flow, unit strip speed, oiling set value and high-voltage generator status so that the values are between 0 and 1 to eliminate the dimensional influence between the data.
[0092] Taking the inverter output current as an example, assuming that its value range is 0-10A, it can be normalized to between 0-1. The specific calculation formula is:
[0093]
[0094] Where I is the original current value, I max and I min are the maximum and minimum current values respectively.
[0095] S2. Build a model and train it: Build an SVR model on the preprocessed data. SVR is a support vector regression model whose goal is to find an optimal solution within the maximum margin and limited error range. Specifically, it is necessary to find the support vector from the training data and obtain the coefficient and bias term of each support vector to build the SVR model. The formula of the SVR model is as follows:
[0096]
[0097] Where x is the input variable vector, x i is the sample vector in the training data set, ai is the coefficient of the support vector, K(x,xi) is the kernel function, and b is the bias top;
[0098] The kernel function uses different types of kernel functions such as linear kernel function, polynomial kernel function or Gaussian kernel function to learn the model.
[0099] The specific steps include:
[0100] S21. Select model: Select the support vector regression model (SVR) as the machine learning model.
[0101] S22. Divide data: randomly divide the normalized data into a training data set and a test data set.
[0102] S23, training model: the training data set is used to train the support vector regression model, during which the relevant parameters of the support vector regression model may need to be adjusted according to the specific situation.
[0103] S24. Model adjustment: Observe the training effect of the support vector regression model and adjust and optimize parameters as needed, such as adjusting regularization parameters and kernel function parameters.
[0104] S3, model evaluation: Use the test data set to verify the support vector regression model. In the support vector regression model verification, it is necessary to calculate the prediction error, determination coefficient and other indicators of the support vector regression model to evaluate the prediction performance of the support vector regression model. For specific indicators, please refer to the regression problem evaluation indicators in statistical learning. Specifically, the following steps are included:
[0105] S31. Calculate prediction error: Apply the trained support vector regression model on the test data set, evaluate the prediction ability of the support vector regression model, and calculate the error between the predicted score and the actual score.
[0106] S32. Calculate the coefficient of determination: Calculate the coefficient of determination (R^2) of the test set data, which is a commonly used indicator for evaluating the prediction effect of the support vector regression model.
[0107] S33, iterative optimization: If the performance of the support vector regression model is unacceptable or there is room for further optimization, it is necessary to readjust the support vector regression model parameters or select other models and iterate repeatedly until you are satisfied.
[0108] S4. Use the model to score the oiling: Input the process parameters of the new oiling machine, i.e., the inverter output current, inverter fault information, actual speed of the metering pump, temperature of the metering pump, tank level, pipe clamp flowmeter flow, unit strip speed, oiling setting value and high-voltage generator state into the trained support vector regression model. The support vector regression model can calculate in the feature space and output the corresponding oiling score. Perform inverse normalization on the predicted score value to obtain the original score result. The calculation formula for inverse normalization is:
[0109] score=score norm (score max -score min )+score min
[0110] In the formula, score norm is the predicted score value, score max and score min are the maximum and minimum score values respectively.
[0111] The specific steps include:
[0112] S41. Data input: Input the 9 key parameters of the new oiling machine device into the optimized support vector regression model.
[0113] S42. Predict new score: The support vector regression model will give a predicted oiler device score based on the input parameters.
[0114] S43, inverse normalization processing: the predicted score value between 0 and 1 is inversely normalized to obtain the actual oiler device score.
[0115] Because traditional oiling scoring methods usually require empirical judgment or set scoring criteria based on some simple rules. Using SVR for scoring, it can adaptively adjust parameters and fit the optimal prediction model according to the actual data set. In the process of the oiling machine, there are complex nonlinear factors and interactions. The support vector regression model has strong modeling and prediction capabilities when dealing with these nonlinear features. It can also obtain high-precision prediction results, has good generalization ability and outlier processing capabilities, and has a wide range of applications.
[0116] Example
[0117] This embodiment provides a cold rolling oiling device oiling analysis and scoring method based on a support vector regression algorithm, and the specific steps are as follows:
[0118] S1. Data preprocessing
[0119] The following data about the oiling machine device is available on site:
[0120] Inverter output current: 5A
[0121] Inverter fault information: 0
[0122] Actual speed of metering pump: 75rpm
[0123] Metering pump temperature: 40℃
[0124] Fuel tank level: 60L
[0125] Pipe clamp flow meter flow rate: 8m 3 / h
[0126] Strip speed of the unit: 23m / min
[0127] Oiling setting value: 2.5μm
[0128] High voltage generator status: 1
[0129] Oiler Rating: 85
[0130] After normalizing these data, the following results are obtained:
[0131] Inverter output current: 0.5
[0132] Inverter fault information: 0
[0133] Actual speed of metering pump: 0.75
[0134] Metering pump temperature: 0.4
[0135] Fuel tank level: 0.6
[0136] Pipe clamp flow meter flow rate: 0.8
[0137] Strip speed of the unit: 0.23
[0138] Oiling setting value: 0.25
[0139] High voltage generator status: 1
[0140] Oiler rating: 0.85
[0141] S2, Modeling and Training
[0142] A support vector regression model is trained using 80% of the feature data (inverter output current, inverter fault information, etc.), and C and epsilon are set to default values.
[0143] S3. Model evaluation
[0144] Assume that the R^2 value of the support vector regression model on the test dataset is 0.92.
[0145] S4. Rating prediction
[0146] There is a new oiler device on site, and its characteristic data is normalized as follows:
[0147] Inverter output current: 0.6
[0148] Inverter fault information: 0
[0149] Actual speed of metering pump: 0.7
[0150] Metering pump temperature: 0.5
[0151] Fuel tank level: 0.65
[0152] Pipe clamp flow meter flow rate: 0.85
[0153] Strip speed of the unit: 0.3
[0154] Oiling setting value: 0.3
[0155] High voltage generator status: 1
[0156] This data was fed into a support vector regression model for prediction, and the predicted score was 0.87. After inverse normalization, the actual score prediction for the new oiler unit was 87 points.
[0157] Those skilled in the art should recognize that the above embodiments are only used to illustrate the present invention, and are not intended to limit the present invention. As long as they are within the spirit of the present invention, any changes or modifications to the above embodiments will fall within the scope of the claims of the present invention.
Claims
1. A cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm, characterized in that: The following steps are involved: S1, data preprocessing; S2, build the model and train it; S3, model evaluation; S4. Use the model to score the oil application.
2. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 1 is characterized in that: The data preprocessing step S1 specifically includes the following steps: S11. Collect data; S12, data cleaning; S13, extracting labels; S14, selecting features; S15, perform normalization processing.
3. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 2 is characterized in that: The step S11 of collecting data further comprises: The data including machine status, operating parameters and machine performance information are obtained from the production process of the cold rolling oiling device.
4. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 3 is characterized in that: The data specifically include inverter output current, inverter fault information, actual speed of metering pump, temperature of metering pump, oil tank level, flow meter flow, unit strip speed, oiling set value, and high-voltage generator status.
5. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 3 is characterized in that: The step S12 data cleaning further includes: Clean invalid or erroneous data and handle missing values and outliers.
6. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 5 is characterized in that: The step S13 of extracting tags further comprises: The score of the cold rolling oiling device is extracted from the data set as the label to be predicted.
7. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 6 is characterized in that: The step S14 of selecting features further includes: The inverter output current, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling set value and the high-voltage generator state are selected as input features.
8. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 7 is characterized in that: The normalization process in step S15 further includes: The inverter output current, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling setting value and the high-voltage generator state are numerically normalized so that the values are between 0 and 1. The specific calculation formula is: Where I is the original current value, I max and I min are the maximum and minimum current values respectively.
9. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 8 is characterized in that: The step S2 of establishing a model and performing training specifically includes the following steps: S21, select model; S22, dividing data; S23, training model; S24. Model adjustment.
10. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 9 is characterized in that: The step S21 of selecting a model further comprises: The Support Vector Regression model was selected as the machine learning model.
11. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 10 is characterized in that: The step S22 of dividing the data further comprises: The normalized data is randomly divided into training data set and test data set.
12. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 11 is characterized in that: The step S23 of training the model further comprises: The training data set is used to train the support vector regression model.
13. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 12 is characterized in that: The calculation formula of the support vector regression model is as follows: Where x is the input variable vector, x i is the sample vector in the training data set, ai is the coefficient of the support vector, K(x,xi) is the kernel function, and b is the bias top; The kernel function is selected from a linear kernel function, a polynomial kernel function or a Gaussian kernel function.
14. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 12 is characterized in that: The step S24 of model adjustment further includes: Observe the training effect of the support vector regression model and adjust and optimize parameters as needed.
15. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 14 is characterized in that: The step S3 model evaluation specifically includes the following steps: S31, calculating the prediction error; S32, calculating the coefficient of determination; S33, iterative optimization.
16. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 15 is characterized in that: The step S31 of calculating the prediction error further comprises: The trained support vector regression model is applied to the test data set, the prediction ability of the support vector regression model is evaluated, and the error between the predicted score and the actual score is calculated.
17. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 16 is characterized in that: The step S32 of calculating the coefficient of determination further comprises: Calculate the coefficient of determination for the test set data.
18. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 17 is characterized in that: The iterative optimization step S33 further includes: If the performance of the support vector regression model is unacceptable or there is room for further optimization, it is necessary to readjust the parameters of the support vector regression model or select another model, and iterate repeatedly until satisfaction is achieved.
19. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 18 is characterized in that: The step S4 uses the model to perform oiling scoring, which specifically includes the following steps: S41, data input; S42, predicting new scores; S43: Inverse normalization processing.
20. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 19, characterized in that: The step S41 data input further includes: The inverter output current of the new oiler device, the inverter fault information, the actual speed of the metering pump, the temperature of the metering pump, the oil tank level, the flow meter flow, the unit strip speed, the oiling set value and the high-voltage generator status are input into the optimized support vector regression model.
21. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 20 is characterized in that: The step S42 of predicting a new score further comprises: The support vector regression model gives a score for predicting the new oiler device based on the input parameters.
22. The cold rolling oiling device oiling analysis and scoring method based on support vector regression algorithm according to claim 21 is characterized in that: The step S43 of inverse normalization further includes: The score value of the oiler device predicted in step S42 is inversely normalized to obtain the true score of the oiler device; the calculation formula for inverse normalization is as follows: score=score norm (score max -score min )+score min In the formula, score norm is the predicted score value, score max and score min are the maximum and minimum score values respectively.
Citation Information
Cited By
Automotive electronic control method, system, medium and product with embedded neural network
CN122703990A