The production printer market is expected to increase due to the demand in high speed printing. In electrophotographic (EP) process, the printing speed is restricted by the toner cooling time. During continuous printing, if the printing speed is too high, following paper is printed before the toner on previous paper is solidified, which results in the adhesion of the toner on previous paper to the next paper. Therefore, it is important to reduce the toner cooling time to improve the printing speed. From the viewpoint of printing quality, although toner temperature is important in EP process, paper temperature is not. Therefore, to improve the printing speed without worsening printing quality, this report focuses on suppressing paper temperature rise after fusing, which means a method to predict paper temperature after fusing is required.
As one of effective approaches to predict paper temperature after fusing, numerical calculation based on physical model has been deployed. Constructing a physical model, however, has some problems such as complexity and uncertainty in particular parameter such as the thermal resistance. Printers have three dimensional and complex structure. The printing process itself is also a complex sequence to be appropriately modeled. It is also difficult to measure the thermal resistance and other properties due to the high speed motion of papers inside the machine. Therefore, as an alternative approach, machine learning can be used to correlate known inputs to the output temperature without relying on these uncertain parameters.
The objective of this research is to propose a new method for predicting and controlling the paper media temperature after fusing in electrophotographic process by machine learning.
Sensors are attached to the printer for the research. By using these sensors, temperature when printing is obtained. By using these data, machine learning is carried out to predict paper temperature after fusing.
This report constructs a new method to predict paper temperature after fusing. Our study demonstrates machine learning is an effective method for predicting and controlling the paper temperature after fusing.