Buck converter is a power converter which drops high input voltage into a low output voltage in high efficiency. With this characteristic, it has been used in a great number of applications. Optimized the maximum load to increase the buck converter's efficiency at the cost of light load efficiency is a general way used in a traditional buck converter because it has a higher impact on power consumption. We propose a novel way of designing the two-phase buck converter with light load efficiency improvement in this thesis.
The purposed two-phase buck converter uses RC delay to control switch frequency. Different frequency will affect the buck converter in output value and efficiency. RC delay includes two parts; part one connect with phase one, part two connect with phase two. After the test, when resister's value of part one is 100kΩ, and the capacitor's value is 50 pF, the resister's value of path two is 40kΩ, and the capacitors' value is 50 pF, the buck converter can reach maximum efficiency.
The inspiration of the neural network is derived from the biological brain, neural is similar with the human neural, and the synaptic weights can treat as the connection between two nodes. Reservoir computing can be seen as an extension of the neural network since it is a framework for computation. Echo State Network(ESN) is one of the major types of reservoir computing, and it is a recurrent neural network. Compared with a neural network, it only trains output weights, which can save a lot of time but keep the accuracy of the training at the same time.
The efficiency of the two-phase buck converter and power loss for each phase in the control scheme were measured. The input voltage set to be 30V, with the switch frequency change from 40's to 100's, the output voltages change from 9.2V to 6V, the output current range is 18 mA to 30 mA. The efficiency ranges are 94% to 98%. The teaching target set for the ESN is the output voltage of the two-phase buck converter. The ESN will read data from two-phase buck converter's simulation, including input voltage, the frequency of the switches and based on that to compute the output voltage. / Master of Science / Buck converter is a power converter which drops high input voltage into a low output voltage in high efficiency. With this characteristic, it has been used in a great number of applications. Most of the buck converter optimized the maximum load to increase the efficiency, however, it will also increase the power consumption of the buck converter. For this reason, we propose a novel way of designing the two-phase buck converter optimize with Echo State Network(ESN). The inspiration of neural network is derived from the biological brain, similar with a human brain, the neural network also have self-learning ability. Reservoir computing is one kind of neural network, it can save more time on computing data and increase the efficiency at the same time. Compare with normal two-phase buck converter, the purposed two-phase buck converter optimize with ESN can increase the efficiency and also decrease the running time.
| Identifer | oai:union.ndltd.org:VTETD/oai:vtechworks.lib.vt.edu:10919/87439 |
| Date | 04 February 2019 |
| Creators | Cheng, Shuang |
| Contributors | Electrical Engineering, Yi, Yang, Ha, Dong S., Jia, Xiaoting |
| Publisher | Virginia Tech |
| Source Sets | Virginia Tech Theses and Dissertation |
| Detected Language | English |
| Type | Thesis |
| Format | ETD, application/pdf |
| Rights | In Copyright, http://rightsstatements.org/vocab/InC/1.0/ |
Page generated in 0.0025 seconds