Method for realizing coiling temperature prediction through machine learning algorithm

Abstract

The invention provides a method for realizing coiling temperature prediction through a machine learning algorithm, and relates to the technical field of computer software application. According to the technical scheme, excellent data of actual production is selected for machine learning, coiling temperature prediction is achieved, and the method has the beneficial effects that the control precision is improved, the production cost is reduced, the stability of the production process is enhanced, and remarkable economic benefits and social benefits are brought to enterprises. The method can be widely applied to scenes of all walks of life needing specific data prediction through a big data machine learning model training method.

Description

Technical Field

[0001] The present invention relates to the technical field of computer software application, and in particular to a method for realizing coiling temperature prediction using a machine learning algorithm. Background Art

[0002] Since there are many uncertain factors in the production process of hot-rolled strip, such as raw material composition, rolling speed, cooling conditions, etc., changes in these factors will affect the coiling temperature. Through coiling temperature prediction, the impact of these changes on the coiling temperature can be predicted in advance, so that corresponding measures can be taken to enhance the stability of the production process, eliminate the influence of the lag of the entire cooling control system, avoid frequent feedback control caused by excessive fluctuations in the final rolling temperature, and thus improve the uniformity of the mechanical properties of the strip. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present invention provides a method for realizing coiling temperature prediction using a machine learning algorithm. By predicting the coiling temperature, the control accuracy is improved, the production cost is reduced, and the stability of the production process is enhanced, which brings significant economic and social benefits to the enterprise and solves the problems existing in the background technology.

[0004] To achieve the above objectives, the present invention is implemented by the following technical solutions: A method for realizing coiling temperature prediction by a machine learning algorithm, specifically comprising the following steps:

[0005] Step 1: Select the type of training parameter variables;

[0006] Step 2: Organize the preparation of data items for prediction;

[0007] Step 3: Divide the training set and test set;

[0008] Step 4: Training model;

[0009] Step 5: Prediction model evaluation;

[0010] Step 6: New sample prediction.

[0011] Preferably, the first step is to select the type of training parameter variables, and select the type of variables that affect the coiling temperature through professional knowledge, which are used for the training parameter types of the model.

[0012] Preferably, the step 2 is to organize the preparation of data items for prediction, read production data, screen the data, and screen the data with a CT hit rate greater than 95%.

[0013] Preferably, the step three is to divide the data into a training set and a test set, and for the sample data, divide the data into data for model training and test data for model evaluation.

[0014] Preferably, the step four is to train the model, use machine learning to train the model using a training data set, analyze the importance, adjust the model, and retrain.

[0015] Preferably, the step five is prediction model evaluation, which involves evaluating the prediction results of the model, finding those with high evaluation scores, and storing them.

[0016] Preferably, the step six is ​​new sample prediction, and the data to be predicted is predicted using this model to obtain a prediction result.

[0017] The present invention provides a method for realizing coiling temperature prediction using a machine learning algorithm. It has the following beneficial effects:

[0018] 1. The present invention provides a method for realizing coiling temperature prediction by using a machine learning algorithm. The method selects excellent data from actual production for machine learning to realize coiling temperature prediction, improves control accuracy, reduces production costs, and enhances the stability of the production process, thus bringing significant economic and social benefits to enterprises. This method can be widely used in such scenarios in various industries that require specific data prediction through big data machine learning model training methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a training model diagram of the present invention;

[0020] Figure 2 It is a diagram of a decision tree of the training model of the present invention;

[0021] Figure 3 is a comparison diagram of the prediction results of the present invention;

[0022] Figure 4 is a hash map of the prediction results of the present invention;

[0023] Figure 5 It is the importance analysis diagram of the model of the present invention;

[0024] Figure 6 It is a prediction model diagram of the present invention;

[0025] Figure 7 is a prediction result data graph of the present invention;

[0026] Figure 8 It is a flow chart of establishing the mathematical model of the present invention. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] The embodiment of the present invention provides a method for realizing coiling temperature prediction using a machine learning algorithm, comprising the following steps:

[0029] Step 1: Select the type of training parameter variables;

[0030] Step 2: Organize the preparation of data items for prediction;

[0031] Step 3: Divide the training set and test set;

[0032] Step 4: Training model;

[0033] Step 5: Prediction model evaluation;

[0034] Step 6: New sample prediction.

[0035] Step 1 of the technical solution of the present invention is to select the type of training variables, and the main parameters are as follows:

[0036] (1) Raw material composition data that affects the coiling temperature, that is, the final chemical composition of the slab of each coil, C, Si, Mn...

[0037] e=[C,Si,Mn,P,S,Al,Alsol,Cu,Ni,Cr,V,Ti,Ca,W,Mg,O,N]

[0038] (2) Target setting values ​​for steel coils, mainly including final rolling temperature target, coiling temperature target,

[0039] Target thickness, FDTTARG, CTTMP, CTHK_AI M.

[0040]

[0041] (3) The rolling speed and cooling conditions that affect the coiling temperature, namely the finishing strip speed F_THRDSTRIPSPD, the cooling mode CTCCTRLMODE,

[0042] Roll box mode CBFLAG.

[0043]

[0044] Step 2 of the technical solution of the present invention is to organize the preparation of the data items for prediction, mainly performing the following work:

[0045] (1) Select excellent data with a CT hit rate greater than 95% from historical production data, and filter data conditions: CT hit rate greater than 95%

[0046] f=[VIEW QUACTC .FULLON>0.95]

[0047] (2) Data is processed through feature processing, feature selection, dimensionality reduction, etc.

[0048] Main data Each steel coil is represented by a unique code:

[0049] (1) Coil code: K = (1, 2, 3, ... n).

[0050] (2) The endpoint chemical composition data code e = (1, 2, 3, ... n) of the steel coil corresponding to the slab has been defined in step 1.

[0051] (3) Steel coil target setting value code: J, which has been defined in step 1.

[0052] (4) Parameter code for rolling speed and cooling conditions: m, which has been defined in step 1.

[0053] (5) Create data items for each steel coil:

[0054]

[0055] Step three of the technical solution of the present invention is to divide the training set and the test set.

[0056] Sample set D = {(a 1 ,b 1 ,c 1 ……),(a 2 ,b 2 ,c 2 ……),……,(a n ,b n ,c n …)}

[0057] Training set D' = {(a 1 ,b 1 ,c 1 ……),(a 2 ,b 2 ,c 2 ……),……,(a m ,b m ,c m …)}

[0058] Test set D-D' = {(a m+1 ,b m+1,c m+1 ……),(a m+2 ,b m+2 ,c m+2 ……),……,(a n ,b n ,c n …)}

[0059] Given a data set D containing n samples, randomly select a sample from it and put it into the sampling set D', and then put the sample back into the original data set D, so that the sample is still likely to be selected the next time it is sampled. In this way, after n random sampling operations, a sampling set D' containing n samples is obtained. Some samples in the original data set D will appear multiple times in D', while other samples will not appear. Assume that the probability of each sample being sampled is Then the probability of not being sampled is Then the probability that a sample is not sampled once in n samplings is Taking the limit we get:

[0060]

[0061] Step 4 of the technical solution of the present invention is to train the model, and the formula is as follows:

[0062]

[0063] M base learners {f 1 ,f 2 ,……,f M}.

[0064] Training model such as Figure 1 As shown, one of the training model decision trees is Figure 2 shown.

[0065] Step five of the technical solution of the present invention is prediction model evaluation.

[0066] The prediction results are compared as Figure 3 The prediction result is hashed as Figure 4 The model importance analysis is shown in Figure 5 As shown. Step 6 of the technical solution of the present invention is new sample prediction. The prediction model is as follows Figure 6 The prediction results are shown in Figure 7 shown.

[0067] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for realizing coiling temperature prediction using a machine learning algorithm, characterized in that: The specific steps include: Step 1: Select the type of training parameter variables; Step 2: Organize the preparation of data items for prediction; Step 3: Divide the training set and test set; Step 4: Training model; Step 5: Prediction model evaluation; Step 6: New sample prediction.

2. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The first step is to select the type of training parameter variables, and select the type of variables that affect the coiling temperature through professional knowledge, which are used as the training parameter types of the model.

3. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The second step is to organize the preparation of the data items to be predicted, read the production data, screen the data, and screen the data with a CT hit rate greater than 95%.

4. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The third step is to divide the training set and the test set, and divide the sample data into data for model training and test data for model evaluation.

5. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The fourth step is to train the model, use machine learning to train the model using the training data set, analyze the importance, adjust the model, and retrain.

6. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The step five is to evaluate the prediction model, which involves evaluating the prediction results of the model and finding those with high evaluation scores for storage.

7. The method for realizing coiling temperature prediction using a machine learning algorithm according to claim 1, characterized in that: The sixth step is new sample prediction, which uses the model to predict the data to be predicted and obtains the prediction result.

Images