Power supply system for optimizing a performance metric using artificial intelligence
According to an aspect, a power supply system includes a power stage, a power supply controller configured to control operations of the power stage, a metering circuit configured to sense measured conditions of the power stage, and a system performance controller configured to be coupled to the power supply controller and the metering circuit. The system performance controller is configured to set or adjust a control parameter for the power stage based on energy conversion efficiency of the power stage. The system performance controller includes an efficiency computation circuit configured to compute the energy conversion efficiency of the power stage based on the measured conditions, and a control manipulation module configured to modify the control parameter until the energy conversion efficiency achieves a threshold condition.
1. A power supply system comprising:
a power stage;
a power supply controller configured to control operations of the power stage;
a metering circuit configured to sense measured conditions of the power stage; and
a system performance controller configured to be coupled to the power supply controller and the metering circuit, the system performance controller configured to set or adjust a control parameter for the power stage based on energy conversion efficiency of the power stage, the system performance controller including:
an efficiency computation circuit configured to compute the energy conversion efficiency of the power stage based on the measured conditions; and
a control manipulation module configured to modify the control parameter until the energy conversion efficiency achieves a threshold condition, the control manipulation module configured to apply historical data to a neural network to predict a value for the control parameter.
2. The power supply system of claim 1 , wherein the system performance controller is configured to select a combination of control parameters that result in the energy conversion efficiency achieving the threshold condition.
3. The power supply system of claim 1 , wherein the system performance controller is configured to determine a first value of the control parameter that results in the energy conversion efficiency achieving the threshold condition for a first environment condition, and determine a second value of the control parameter that results in the energy conversion efficiency achieving the threshold condition for a second environment condition.
4. The power supply system of claim 3 , wherein the first environment condition is a first input voltage, and the second environment condition is a second input voltage, the second input voltage being different than the first input voltage.
5. The power supply system of claim 1 , wherein the control manipulation module is configured to iteratively modify the control parameter by selecting different values for the control parameter based on an artificial intelligence (AI) algorithm.
6. The power supply system of claim 1 , wherein the control manipulation module is configured to:
select a first value for the control parameter;
send the first value to the power supply controller;
compute the energy conversion efficiency of the power stage that operates according to the first value based on the measured conditions;
modify the first value to obtain a second value in response to the energy conversion efficiency not achieving the threshold condition;
send the second value to the power supply controller; and
compute the energy conversion efficiency of the power stage that operates according to the second value based on the measured conditions.
7. The power supply system of claim 1 , wherein the system performance controller is configured to update the control parameter in response to a triggering event.
8. The power supply system of claim 7 , wherein the triggering event includes detection of a change in one or more environment conditions of the power stage.
9. The power supply system of claim 7 , wherein the triggering event includes detection of the energy conversion efficiency not achieving the threshold condition.
10. The power supply system of claim 7 , wherein the triggering event includes an activation of the power stage.
11. A power supply system comprising:
a power stage configured to convert an input voltage to an output voltage;
a power supply controller storing a control parameter to control operations of the power stage;
a metering circuit configured to sense measured conditions of the power stage, the metering circuit including a voltage sensor configured to sense the input voltage of the power stage; and
a system performance controller connected to the power supply controller and the metering circuit, the system performance controller configured to periodically monitor an energy conversion efficiency of the power stage during operation of the power stage and periodically update the control parameter, the system performance controller including:
an efficiency computation circuit configured to compute the energy conversion efficiency of the power stage based on the measured conditions; and
a control manipulation module configured to modify the control parameter until the energy conversion efficiency achieves a threshold condition,
the system performance controller configured to provide the modified control parameter to the power supply controller,
the system performance controller configured to determine a change to the input voltage of the power stage and update the control parameter in response to the input voltage being determined as changed,
the control manipulation module configured to apply historical data to a neural network to predict a value for the control parameter.
12. The power supply system of claim 11 , wherein the measured conditions include the input voltage, input current, output voltage, and output current.
13. The power supply system of claim 11 , wherein the metering circuit includes one or more analog-to-digital converters configured to convert the measured conditions from an analog format to a digital format.
14. The power supply system of claim 11 , wherein the control parameter includes a zero-voltage switching (ZVS) control parameter, a switching frequency, a pulse width, a duty cycle, or a dead time.
15. The power supply system of claim 11 , wherein the power stage is a first power stage, and the power supply system includes a second power stage, the control parameter including a phase transition parameter relating to phase shedding or phase adding.
16. A method for controlling a control parameter of a power stage using energy conversion efficiency, the method comprising:
detecting a triggering event;
selecting a value for the control parameter based on the energy conversion efficiency, including:
receiving measured conditions of the power stage;
computing the energy conversion efficiency of the power stage based on the measured conditions;
modifying the control parameter until the energy conversion efficiency achieves a threshold condition, including applying historical data to a neural network to predict a value for the control parameter;
providing the selected value to a power supply controller to control the power stage with the selected value.
17. The method of claim 16 , wherein the neural network predicts a combination of control parameters that result in the energy conversion efficiency achieving the threshold condition.
18. The method of claim 16 , wherein the control parameter includes at least one of a zero-voltage switching (ZVS) control parameter, switching frequency, pulse width, duty cycle, or dead time.
19. The method of claim 16 , wherein the triggering event includes activation of the power stage, detection of a change in one or more environment conditions of the power stage, expiration of a timer, or detection of the energy conversion efficiency of the power stage not achieving the threshold condition.
20. A system performance controller for controlling a power stage of a power supply system, the system performance controller comprising:
a computation circuit configured to monitor a performance metric of a power stage by receiving measured conditions of the power stage and computing a performance metric based on the measure conditions; and
a control manipulation module configured to execute, in response to a triggering event, an artificial intelligence (AI) algorithm to identify a value for a control parameter of the power stage that results in the performance metric achieving a threshold condition, the AI algorithm including an applied learning algorithm that applies historical data to a neural network to predict a value for the control parameter.
21. The system performance controller of claim 20 , wherein the triggering event includes activation of the power stage, detection of a change in one or more environment conditions of the power stage, detection of the performance metric not achieving the threshold condition, or detection of an expiration of a timer.