CN-122008551-A - Printing adjustment method and system based on deep learning

Abstract

The application relates to the technical field of 3D printing, in particular to an adjustment printing method and system based on deep learning, comprising the steps of collecting printing data and obtaining a first data set; the method comprises the steps of preprocessing an acquired first data set to acquire a second data set, inputting the second data set into a trained parameter adjustment model to perform parameter adjustment to acquire a third data set, sending the third data set to printing equipment to adjust printing behaviors of the printing equipment, predicting ideal quantities of printing parameters by adopting the parameter adjustment model, reversely mapping and changing the parameters according to the predicted quantities, enabling a printing result to be closer to actual requirements, avoiding deviation of the printing result due to errors of manual adjustment by adopting a parameter adjustment model adjustment mode, further, penalty marking the printing parameters by adopting a cost sensitive learning model, and adjusting the printing result in a self-learning mode when the deviation of the printing result is overlarge.

Inventors

• LIU JIURU
• LIU JIA

Assignees

• 西北工业大学

Dates

Publication Date

20260512

Application Date

20260114

Claims (10)

1. 1. The printing adjustment method based on deep learning is characterized by comprising the following steps: Acquiring printing data to obtain a first data set; Preprocessing the acquired first data set to acquire a second data set; Inputting the second data set into a trained parameter adjustment model for parameter adjustment to obtain a third data set; The third data set is sent to the printing device, and the printing behavior of the printing device is adjusted.
2. 2. The method for printing adjustment based on deep learning according to claim 1, wherein the performing print data acquisition, obtaining the first data set specifically comprises: Acquiring printing parameters and printing quality data input by a user; The print parameters and print quality data are integrated into a first data set.
3. 3. The method for printing adjustment based on deep learning according to claim 2, wherein the step of preprocessing the acquired first data set to acquire the second data set comprises the following steps: And carrying out standardization processing on the first data set, and eliminating the influence of dimension.
4. 4. The deep learning based adjustment printing method of claim 3, wherein the step of inputting the second data set into the trained parameter adjustment model to perform parameter adjustment and obtain the third data set comprises: normalizing the real-time printing parameters in the parameter adjustment model, and predicting the parameter adjustment quantity corresponding to the printing quality; and reversely mapping the parameter adjustment quantity corresponding to the predicted printing quality to obtain the predicted and adjusted printing parameter.
5. 5. The deep learning based adjustment printing method of claim 4, wherein the training step of the parameter adjustment model is: performing learning parameter mapping on the historical acquired printing data set, and defining a parameter model; training a deep learning neural network for the parameter model and optimizing the training; And updating the training parameters by using a gradient descent method until the trained model parameters are acquired.
6. 6. The printing system based on deep learning is characterized by comprising a data acquisition module, a preprocessing module, a printing parameter adjustment module and a printing behavior adjustment module; the data acquisition module is used for acquiring printing data and acquiring a first data set; The preprocessing module is used for preprocessing the acquired first data set to acquire a second data set; The printing parameter adjustment module is used for inputting the second data set into the trained parameter adjustment model to perform parameter adjustment and obtaining a third data set; and the printing behavior adjustment module is used for transmitting the third data set to the printing equipment and adjusting the printing behavior of the printing equipment.
7. 7. The deep learning based adjustment printing system of claim 6, wherein the data acquisition module comprises a data acquisition unit and a data integration unit; The data acquisition unit is used for acquiring printing parameters and printing quality data input by a user; the data integration unit is used for integrating the printing parameters and the printing quality data into a first data set.
8. 8. The deep learning based adjustment printing system of claim 7, wherein the printing parameter adjustment module comprises a parameter adjustment amount prediction unit and a printing parameter reverse mapping unit; the parameter adjustment quantity prediction unit is used for normalizing the real-time printing parameters in the parameter adjustment model and predicting the parameter adjustment quantity corresponding to the printing quality; and the printing parameter reverse mapping unit is used for standardizing the real-time printing parameters in the parameter adjustment model and predicting the parameter adjustment quantity corresponding to the printing quality.
9. 9. An electronic device, comprising: The apparatus comprises one or more processors, memory, and one or more computer programs, wherein the one or more computer programs are stored in the memory, the one or more computer programs comprising instructions, which when executed by the apparatus, cause the apparatus to perform the steps of the deep learning based adjustment printing method of any of claims 1 to 5.
10. 10. A computer-readable storage medium, in which a computer program is stored which, when run on a computer, causes the computer to perform the steps of the deep learning based adjustment printing method of any one of claims 1 to 5.

Description

Printing adjustment method and system based on deep learning Technical Field The application relates to the technical field of 3D printing, in particular to an adjustment printing method and system based on deep learning. Background With the rapid development of digital manufacturing technology, the adjustment of printing technology (such as 3D printing, industrial printing, etc.) plays an increasingly important role in the fields of precision machining, customized production, etc. In order to ensure high precision and consistency of the printed product, real-time adjustment in the printing process becomes a key link. However, the conventional adjustment method relies on static parameters or manual experience, which is difficult to cope with complex dynamic changes, and it is highly desirable to introduce intelligent prediction and automation control mechanisms to improve the overall performance. In recent years, deep learning has shown great potential in the fields of image recognition, predictive analysis and the like, and provides a new idea for intelligent adjustment of a printing process, but the prior art has not fully integrated the advanced means. At present, the printing process is mainly based on the following technologies of controlling printing quality through a standardized process based on a fixed printing mode of preset process parameters, adopting sensors (such as position and temperature sensors) for real-time monitoring and performing fine adjustment based on simple feedback rules (such as PID control), and relying on operator observation and manual intervention in the printing process to empirically adjust parameters such as printing speed, material flow and the like. However, the above prior art has significant drawbacks: The existing method can not accurately predict the final result before or during the printing, so that the printing deviation can only be found after the fact, and the resource waste and the disqualification of the finished product are caused. The adjustment process is mostly completed manually by operators, is influenced by factors such as subjective experience, fatigue and the like, is easy to cause judgment errors or operation delay, and reduces the reliability and efficiency of printing. Disclosure of Invention In order to overcome the problems in the prior art, the application provides a method and a system for adjusting printing based on deep learning, which adopts the following technical scheme: in a first aspect, the application provides an adjustment printing method based on deep learning, comprising the steps of collecting printing data and obtaining a first data set; Preprocessing the acquired first data set to acquire a second data set; Inputting the second data set into a trained parameter adjustment model for parameter adjustment to obtain a third data set; The third data set is sent to the printing device, and the printing behavior of the printing device is adjusted. Further, the acquiring the print data, the acquiring the first data set specifically includes: Acquiring printing parameters and printing quality data input by a user; The print parameters and print quality data are integrated into a first data set. Further, the preprocessing is performed on the obtained first data set to obtain a second data set, and the steps specifically include: And carrying out standardization processing on the first data set, and eliminating the influence of dimension. Further, the step of inputting the second data set into the trained parameter adjustment model to perform parameter adjustment to obtain a third data set includes: normalizing the real-time printing parameters in the parameter adjustment model, and predicting the parameter adjustment quantity corresponding to the printing quality; and reversely mapping the parameter adjustment quantity corresponding to the predicted printing quality to obtain the predicted and adjusted printing parameter. Further, the training step of the parameter adjustment model is as follows: performing learning parameter mapping on the historical acquired printing data set, and defining a parameter model; training a deep learning neural network for the parameter model and optimizing the training; And updating the training parameters by using a gradient descent method until the trained model parameters are acquired. Further, the method further comprises the step of introducing the adjusted printing parameters into a cost sensitive learning model to perform secondary learning, specifically: introducing a cost matrix for marking the adjusted printing parameters, and marking different cost penalty forces for different material data; and presetting a higher punishment coefficient corresponding to the material in the key area according to the importance of the processing complexity by the cost matrix and the processing difficulty of the material. Scanning the printing model to obtain actual model parameters of the printing model; comparing