I think the main function of MPPT is to find the max output power
Yes.
for us to let the power efficiency become the MAX.
That depends on the definition of "efficiency" (for example a gasoline engine may not have max efficiency at max power).
However,i am still not quite understand MPPT. Is MPPT a schematic or an algorithm?
"Maximum Power Point Tracking" isn't even an algorithm, it is simply a principle, a description of "a system that will adjust certain parameters to make sure it will operate at maximum available power depending on current conditions".
An algorithm/schematic are ways to implement this in a particular context.
For example, choosing the optimum gear on a bicycle is MPPT. In this case the implementation consists of an algorithm in your head, force sensors in your tendons, plus a mechanical drive train with multiple gears.
We usually see a point of MPPT ,which is the top point of the output power curve.
Yes, but keep in mind the curve itself depends on variations in what powers the generator (sunlight, wind speed, etc) so for these variable sources your MPPT will need to be able to react to changes in conditions.
what should we modify to let the output power become and keep the value which is the same as MPPT?
In the case of a solar panel, one possible algorithm (there are others) is Perturb & Observe: draw a little more current, then a little less, and measure the voltage, then compute the power in both cases. Then increase or decrease the current depending on which value gave more power, and repeat.
If your power source is a windmill, or anything else, some of these algorithms may still work, or others may be better.
