JP4882735B2 - Cutting method of continuous cast steel pieces - Google Patents

Description

  The present invention relates to a method for cutting continuous cast steel pieces, and more particularly to a method for calculating a correction coefficient used when cutting continuously cast steel pieces.

  In the continuous casting equipment, the molten steel transported by the ladle is poured into the tundish from the bottom of the ladle, and is temporarily stored here and led to a mold as a stable flow. The molten steel is cooled in the mold and the surroundings solidify to form a shell, which is pulled out while being oscillated by oscillation, then cooled with cooling water from the spray zone and corrected horizontally by a curved roller apron. Strands. This strand is fed to a torch car by a primary pinch roll, and is melted into a slab of a predetermined length.

  The slab length to be cut in such a manufacturing process is determined from a nominal unit weight (hereinafter referred to as a nominal unit weight) and a correction coefficient for correcting the slab length in accordance with the requested weight in the subsequent process. Here, the nominal unit weight is the weight per unit length of the strand determined by using the specific gravity determined from the cross-sectional area of the mold and the composition of the molten steel. For example, the cross-sectional area of the strand is immediately after being drawn from the mold and cooled. Since it is different from that after being cut, the correction unit is used to correct the unit weight of the strand immediately before cutting.

  This correction coefficient has been determined by a continuous casting operator based on experience so far, or a table of correction coefficients has been provided and operations have been performed accordingly. However, there is a problem that the weight of the steel slab varies with respect to the target because there is variation in the specification of the correction coefficient by the operator. Another problem is that it is necessary to review the correction coefficient table when a change occurs in conditions that occur in the process.

Therefore, there is a technique disclosed in Patent Document 1. In this technique, the contribution ratio of casting condition information to the correction coefficient is obtained from casting condition information regarding a plurality of cut slabs and the weight of the plurality of slabs, and the contribution ratio and castings regarding the slab to be cut are calculated. The correction coefficient relating to the slab to be cut is obtained from the insertion condition information, and the length of the slab having a required weight is obtained using the correction coefficient.
JP-A-6-114519

  However, in the technique disclosed in Patent Document 1, since the contribution ratio suitable for them is obtained for all the slabs that have been cut so far, for example, the slab that is to be cut is In the case of casting condition information where the data is too small or the conditions are greatly different, there is a problem that the billet weight varies from the target.

  The present invention has been made to solve the above-described problem, and an object of the present invention is to provide a continuous cast steel slab cutting method that determines a correction coefficient that does not vary the steel slab weight with respect to a target.

  The invention according to claim 1 is a novel method for cutting a continuous cast steel slab targeted in a continuous casting manufacturing process, and determines a correction coefficient that affects the slab weight of the continuous cast steel slab. A similarity calculation step for calculating a similarity between a performance input variable representing an operation condition stored in a past performance operation database and a new input variable of the continuous cast steel slab, and a predetermined high degree of similarity A weight calculation process for selecting a number of data and calculating a weighting coefficient for the selected data, a model creation process for creating a model using the selected data, and a correction for calculating the correction coefficient from the created model And a coefficient calculation step. A method for cutting a continuous cast steel piece, comprising: a coefficient calculation step.

  The invention according to claim 2 is the method for cutting continuous cast steel pieces according to claim 1, wherein the input variables include at least a casting speed, a specific water amount, a steel piece surface speed, and a secondary cooling zone from a mold. Any one of the steel piece passage times up to and including a continuous casting cast steel piece cutting method.

Furthermore, the invention which concerns on Claim 3 WHEREIN: In the cutting method of the continuous cast cast steel piece of any one of Claim 1 or Claim 2,
In the data selection, a new input variable of a new continuous cast steel slab to be processed and an actual input variable stored in the actual operation database are respectively statistics of the actual input variable stored in the actual operation database. Normalize,
The normalized norm of the difference between the new input variable and the actual input variable is calculated, and a predetermined number of actual operation data having a small calculated norm are selected as data having high similarity. This is a method for cutting continuous cast steel pieces.

  In the present invention, for each billet for which the correction coefficient is to be determined, an optimal regression equation is created for each billet, and the correction factor is determined based on the created regression equation. Appropriate accuracy is improved and the yield can be improved. In addition, since the regression equation is created for each steel piece, even if a change in conditions occurs in the process, it is possible to respond quickly.

  In the present invention, in determining the correction coefficient, an approximate model capable of directly calculating the correction coefficient is created, and the correction coefficient is calculated from the created approximate model. In the present invention, when creating this approximate model, past steel slabs similar to the steel slab for which the correction coefficient is calculated are collected, and an appropriate model is constructed from the collected performance data. There are many possible methods for constructing this model, but the present invention provides one of them, and this method collects the vicinity of the operating condition of the billet for which the correction coefficient is specified, A model established in the vicinity is constructed each time.

  The best mode for carrying out the present invention will be described below with reference to the drawings and formulas. FIG. 1 is a flowchart showing a correction coefficient determination method according to the present invention.

First, in S1 and S2, new and actual casting conditions are input from a host computer or a case database. Here, input variables and billet weight are defined as shown below. Note that the following are examples of input variables, and other variables may be used.
X _m ^n: input variables
Y ⁿ : Billet weight where
n: Number of data When n = 0, data of new billet to be estimated When n = 1 to N, actual data of past billet m: Subscript representing each input variable. For example, the following types There is.

m = 1: Casting speed, m = 1: Specific water amount m = 3: Steel slab surface temperature, m = 4: Steel slab transit time from mold to secondary cooling zone m = 5: Correction coefficient, m = 6: Date Next, in S3, the minimum value _{Xm, min} and the maximum value _{Xm, max} of each input variable are calculated. In S4, the input variable is normalized from the following equation (1).

  Then, in S5, similarity calculation between past performance data and new billet data to be estimated is performed. In this calculation, for example, the calculation is performed as 2-norm as in the following equation (2), but other distance measures may be used.

Furthermore, the data are rearranged in descending order of similarity, and the weighting coefficient W ^k shown in the following equation (3) is calculated for the data up to k-th (for example, 400th) (S6).

  Next, in S7, a linear multiple regression model for obtaining the target steel weight Y as shown in the following equation (4) is created by the weighted least square method using the selected data with high similarity.

  Finally, in S8, equation (4) is transformed into equation (5) to obtain a correction coefficient for the steel piece to be estimated.

  Examples according to the present invention will be described below. The present embodiment is for showing the effectiveness of the invention, and shows an example. Therefore, the present invention is not limited to these examples.

  In this example, the unsteady part (the part where an event such as pan change occurred) was handled as the correction coefficient designation steel piece. Factors used were operating conditions such as casting speed, specific water amount, steel slab surface temperature, and slab transit time from the mold to the secondary cooling zone. Based on the extracted similar data, a regression equation is created by the least square method, a correction factor is calculated based on the created regression equation, and the appropriateness accuracy of the billet weight when this correction factor is used is evaluated by simulation did.

  FIG. 2 is a diagram showing a comparative example of the method proposed in the present invention and the conventional method. The added weight, which indicates the appropriateness of the target billet weight as a percentage, is plotted on the horizontal axis and the number of billets on the vertical axis. As shown in FIG. It can be seen that the variation can be reduced.

3 is a flowchart illustrating a correction coefficient determination method according to the present invention. It is a figure which shows the comparative example of the method proposed by this invention, and the conventional method.

Claims (3)

A novel continuous cast steel slab cutting method targeted in the continuous casting manufacturing process,
In determining the correction coefficient that affects the slab weight of the continuous cast steel slab,
A similarity calculation step for calculating a similarity between a performance input variable representing an operation condition stored in a past performance operation database and a new input variable of the continuous cast steel slab,
A weighting calculation step of selecting a predetermined number of data having a high calculated similarity and calculating a weighting coefficient for the selected data;
A model creation process for creating a model using the selected data;
A correction coefficient calculation step of calculating the correction coefficient from the created model,
A method for cutting continuous cast steel pieces characterized by comprising: In the cutting method of the continuous casting cast steel piece of Claim 1,
The input variable includes at least any one of a casting speed, a specific water amount, a steel slab surface speed, and a steel slab passing time from a mold to a secondary cooling zone. In the cutting method of the continuous casting cast steel piece of any one of Claim 1 or Claim 2,
In the data selection, a new input variable of a new continuous cast steel slab to be processed and an actual input variable stored in the actual operation database are respectively statistics of the actual input variable stored in the actual operation database. Normalize,
The normalized norm of the difference between the new input variable and the actual input variable is calculated, and a predetermined number of actual operation data having a small calculated norm are selected as data having high similarity. Cutting method of continuous cast steel pieces.

Images