CN-121989375-A - Cutting control method, slicer, storage medium and control device

CN121989375ACN 121989375 ACN121989375 ACN 121989375ACN-121989375-A

Abstract

The invention provides a cutting control method, a slicing machine, a storage medium and a control device, wherein the cutting control method comprises the steps of obtaining actual cutting data of main shaft torque of the slicing machine when a target crystal bar is cut, inputting the actual cutting data into a torque decision model to obtain expected cutting torque of the main shaft for cutting the target crystal bar, wherein the torque decision model is obtained by training effective cutting data according to history of the main shaft torque, and adjusting the actual cutting torque of the main shaft for cutting the target crystal bar according to the expected cutting torque to adjust the difference value between the actual cutting torque and the expected cutting torque within a preset range. According to the technical scheme provided by the invention, the problem that the quality of the obtained wafer is difficult to control when a slicing machine is used for cutting the target crystal bar in the prior art can be solved.

Inventors

ZHANG AIXIN
LI LU
XING XU
ZHUANG XUSHENG

Assignees

高测(盐城)技术有限公司

Dates

Publication Date: 20260508
Application Date: 20241105

Claims (11)

1.A cutting control method, characterized by comprising: When cutting the target crystal bar, acquiring actual cutting data of the spindle torque of the slicer; inputting the actual cutting data into a torque decision model to obtain the expected cutting torque of the main shaft for cutting the target crystal bar, wherein the torque decision model is obtained by training according to the historical effective cutting data of the main shaft torque; And adjusting the actual cutting torque of the main shaft for cutting the target crystal bar according to the expected cutting torque so as to adjust the difference value between the actual cutting torque and the expected cutting torque to be within a preset range.
2. The method of claim 1, wherein the target ingot has k feed positions, the target ingot has a desired cutting value of E t , t = 1,2, k for the spindle torque at the t-th feed position, and the adjusting the actual cutting data according to the desired cutting torque comprises: Acquiring an actual cutting value x t and an expected cutting value E t of the torque of the main shaft when the target crystal bar is at a t-th feeding position; Calculating a difference E (t) =x t -E t between the actual cut value x t and the desired cut value E t when the target ingot is at the t-th feed position; And controlling e (t) within a preset range.
3. The cutting control method according to claim 2, wherein the controlling of e (t) within a preset range includes: Acquiring residual accumulation time T i of the target crystal bar when the target crystal bar is at the T feeding position; And controlling the feeding speed of the target crystal bar according to e (T) and T i so as to enable e (T) to be in the preset range.
4. The cutting control method according to claim 3, wherein the controlling the feed rate of the target ingot according to e (T) and T i comprises: Acquiring an actual feeding speed v t1 of the target crystal bar at a t feeding position in the cutting process; acquiring a speed control parameter u (t) of the target crystal bar, Adjusting the feeding speed v t1 of the target crystal bar at the t feeding position to v t2 ,v t2 ＝[1-u(t)]v t1 ; Wherein K P is a scaling factor, K i is a time adjustment parameter, T D is a differential time constant, Δe (T) =e (T) -e (T-1).
5. The cutting control method according to claim 1, wherein the training step of the torque decision model includes: acquiring historical effective cutting data from historical cutting data, wherein the historical effective cutting data comprises cutting data corresponding to sample cutting yields in the historical cutting data larger than a preset yield; and inputting the historical effective cutting data into a decision tree model for model training to obtain the torque decision model.
6. The method of claim 5, wherein inputting the historical effective cut data into a decision tree model for model training to obtain the torque decision model comprises: Dividing the historical effective cutting data into a training cutting number set and a testing cutting number set; Constructing a first decision tree, inputting at least part of the training cutting number set, a cutting position number set corresponding to at least part of the training cutting number set and a cutting parameter number set corresponding to at least part of the training cutting number set into the first decision tree, and training the first decision tree to obtain prediction data; Constructing a plurality of decision trees in series one by one, and training each decision tree according to the sample residual errors of the predicted data and the real data of the previous decision tree as a target; and obtaining the decision model according to the trained multiple decision trees connected in series.
7. The method of claim 6, wherein inputting the historical effective cut data into a decision tree model for model training to obtain the torque decision model, further comprising: The training minimum error MSE and the sample residual epsilon i are calculated, Epsilon i ＝f(x i )-y i ,i＝1,2…,m;f(x i ) is the predicted value of the ith sample, and y i is the true target value of the ith sample; wherein MSE <0.4, and/or, ε i <0.6。
8. The cutting control method according to claim 1, wherein the inputting the actual cutting data into the decision model to obtain desired cutting data comprises: the strong learning expression h t (x) is established as follows: Where t represents the round, j represents the number of leaf nodes, R tj represents the region corresponding to each leaf node, I (x ε R tj ) is an indicator function, and c ij is the best fit value.
9. A microtome, characterized in that the microtome is adapted to perform the cutting control method of any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored program, wherein the program when run performs the cutting control method of any one of claims 1 to 8.
11. A control apparatus, characterized by comprising: the acquisition module is used for acquiring actual cutting data of the spindle torque of the slicing machine when the target crystal bar is cut; The fitting module is used for inputting the actual cutting data into a torque decision model to obtain the expected cutting torque of the main shaft for cutting the target crystal bar, wherein the torque decision model is trained according to the historical effective cutting data of the main shaft torque; And the cutting control module is used for adjusting the actual cutting torque of the main shaft for cutting the target crystal bar according to the expected cutting torque so as to adjust the difference value between the actual cutting torque and the expected cutting torque to be within a preset range.

Description

Cutting control method, slicer, storage medium and control device Technical Field The invention relates to the technical field of photovoltaic cutting, in particular to a cutting control method, a slicing machine, a storage medium and a control device. Background At present, the consolidated diamond abrasive wire cutting technology has become the mainstream technology of photovoltaic silicon wafer cutting, and has the advantages of narrow kerf, good dicing quality, high cutting efficiency and the like. However, when a silicon rod is cut by using a diamond wire, the yield of silicon wafers obtained under different conditions often varies greatly, resulting in difficulty in controlling the quality of the cut silicon wafers. Disclosure of Invention The invention mainly aims to provide a cutting control method, a slicing machine, a storage medium and a control device, which are used for solving the problem that the quality of a wafer obtained by cutting a target crystal bar by using the slicing machine is difficult to control in the prior art. In order to achieve the above object, according to one aspect of the present invention, there is provided a cutting control method comprising: When cutting the target crystal bar, acquiring actual cutting data of the spindle torque of the slicer; Inputting actual cutting data into a torque decision model to obtain the expected cutting torque of a main shaft cutting target crystal bar, wherein the torque decision model is obtained by training according to the historical effective cutting data of the main shaft torque; And adjusting the actual cutting torque of the main shaft cutting target crystal bar according to the expected cutting torque so as to adjust the difference value between the actual cutting torque and the expected cutting torque to be within a preset range. Further, the target ingot has k feed positions, the target ingot has a desired cutting value of E t, t=1, 2, at the t-th feed position, k, and the adjusting of the actual cutting data according to the desired cutting torque comprises: Acquiring an actual cutting value x t and an expected cutting value E t of the spindle torque of the target crystal bar when the target crystal bar is at the t feeding position; Calculating a difference E (t) =x t-Et between an actual cut value x t and an expected cut value E t of the target ingot when at the t-th feed position; And controlling e (t) within a preset range. Further, controlling e (t) within a preset range includes: Acquiring residual accumulation time T i of the target crystal bar when the target crystal bar is at the T feeding position; And controlling the feeding speed of the target crystal bar according to e (T) and T i so as to enable e (T) to be in a preset range. Further, controlling the feed rate of the target ingot according to e (T) and T i, including: Acquiring an actual feeding speed v t1 of the target crystal bar at a t feeding position in the cutting process; Acquiring a speed control parameter u (t) of the target crystal bar, Adjusting the feed speed v t1 of the target ingot at the t-th feed position to v t2,vt2＝[1-u(t)]vt1; wherein K P is a proportionality coefficient, K i is a time adjustment parameter, T i is a time for accumulating residual errors of the target ingot when the target ingot is at the T-th feeding position, T D is a differential time constant, Δe (T) =e (T) -e (T-1). Further, the training step of the torque decision model includes: acquiring historical effective cutting data from the historical cutting data, wherein the historical effective cutting data comprises cutting data corresponding to sample cutting yields in the historical cutting data greater than a preset yield; And inputting the historical effective cutting data into a decision tree model for model training to obtain a torque decision model. Further, inputting the historical effective cut data into a decision tree model for model training to obtain a torque decision model, comprising: dividing the historical effective cutting data into a training cutting number set and a testing cutting number set; Constructing a first decision tree, inputting at least part of a training cutting number set, a cutting position number set corresponding to at least part of the training cutting number set and a cutting parameter number set corresponding to at least part of the training cutting number set into the first decision tree, and training the first decision tree to obtain prediction data; Constructing a plurality of decision trees in series one by one, and training each decision tree according to the sample residual errors of the predicted data and the real data of the previous decision tree as a target; And obtaining a decision model according to the trained multiple decision trees connected in series. Further, the method for inputting the historical effective cutting data into the decision tree model for model training to obtain a torque decision model further comprises the fol