Recurrent Neural Network Offset Current Generation for Display Brightness Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional backlight module technologies face challenges in accurately controlling light-emitting diodes' brightness due to process variations, leading to inconsistencies in preset parameters, which affect the contrast ratio of displays.
Innovation Solution
A method using a recurrent neural network to generate offset current values by establishing a current setting sequence, measuring current values, and adjusting driving values through negative feedback control, with the network comprising an input layer, hidden layer, and output layer, to correct brightness levels across regions of a display panel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If preset parameters are used to drive light-emitting diodes, then the control process is simple, but the brightness accuracy deteriorates due to process variations
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing offset current values in a lookup table before actual display operation. During calibration, the system measures actual current values and computes correction offsets that are stored for future use. This pre-computed correction data enables rapid compensation without real-time complex calculations, maintaining operational simplicity while improving brightness accuracy.
Solution Approach 2:
The patent implements feedback through a calibration process that measures actual current values drawn by light-emitting diodes and uses these measurements to generate corrective offset values. The measured current values feed back into the system to adjust the driving parameters, creating a closed-loop control mechanism that compensates for process variations and ensures accurate brightness control.
2Ease of manufacture
If conventional calibration methods are used, then the process is straightforward, but the time required for calibration increases
Solution Approach 1:
The patent reduces calibration time by performing preliminary calculations and storing offset values in a lookup table during the calibration phase. Once calibrated, the stored offset values enable rapid compensation without requiring repeated complex measurements or calculations during operation, significantly reducing the time needed for subsequent calibration or adjustment operations.
Solution Approach 2:
The patent applies partial action by focusing calibration efforts on measuring only the actual current values drawn by light-emitting diodes at specific dimming levels, rather than performing exhaustive calibration across all possible operating conditions. The recurrent neural network then generalizes from these partial measurements to predict offset values for all dimming levels, reducing calibration time while maintaining accuracy.
3Device complexity
If process variations are not corrected, then the device complexity remains low, but the contrast ratio performance deteriorates
Solution Approach 1:
The patent uses feedback to measure actual current values and generate corrective offset values that compensate for process variations affecting contrast ratio. By implementing a calibration process that measures real-world performance and feeds this information back into the system for correction, the patent improves contrast ratio reliability without requiring overly complex real-time adjustment mechanisms.
Solution Approach 2:
The patent applies preliminary action by pre-computing and storing offset current values that correct for process variations before they affect display performance. This advance preparation allows the system to maintain high contrast ratio performance through simple lookup operations rather than complex real-time calculations, balancing performance reliability with device complexity.
Data Source
AI summary
A method for generating offset current values includes: setting a current setting sequence which includes multiple current setting values; driving a light emitting unit and measuring a current value of the light emitting unit; establishing a recurrent neural network (RNN) including an input layer, a hidden layer and an output layer; and inputting the current value into the hidden layer, inputting the current setting values into the input layer sequentially, and obtaining offset values from the output layer sequentially. The offset values correspond to the current setting values respectively.