Background technology
No matter that wind energy or sun power are due to the restriction by equipment cost, so be all that the power that exports on same equipment is The more the better controlling target in output power, but the power that equipment exports in the certain situation of equipment and some external environment conditions has individual maximal value, the target controlling the controller of output power is exactly allow the power exported, this maximal value of infinite approach.Also maximal power tracing device (MPPT) is.
How to judge whether present output power is peak power, if not the Left or right that this power so exported now is at maximum power point, namely controling parameters should become still to diminish greatly and just can get back to maximum power point, maximal power tracing device (MPPT) groundwork that Here it is.Wherein most crucial part is the method judging relation between present output power and peak power, several conventional method is have now been formed in the process of new forms of energy development, each method cuts both ways, mainly require in reaction velocity according to Practical Project, controller hardware cost, the method for accuracy rating etc. various aspects balanced compromise.
Introduce the method for now conventional several searching maximum power points below:
1, hill climbing
Also belonging to disturbance observation, is also now the most frequently used maximal power tracing (MPPT) control method.Its first sample detecting present output power Pi, then control voltage U is changed, and then the Pi+1 that samples out, if output power becomes large i.e. Pi+1>Pi so illustrate that control voltage adjustment is in the right direction, output power, close to maximum power point direction, so continue this direction of peace and becomes large control voltage.If Pi+1<Pi to illustrate tune up control voltage U after output power diminished, the adjustment anisotropy of control voltage, like this will reverse regulable control voltage, control voltage U is diminished, makes output power close to maximum power point.
All control voltage change direction and output power converter logic relation as follows:
When control voltage U becomes large, if calculate output power by sampling also become large, so control voltage change direction is correct, and continuation is become large by control voltage.
When control voltage U becomes large, if calculating output by sampling is that power diminishes, so illustrate that control voltage change direction is wrong, next step control voltage of change direction will diminish by control voltage.
When control voltage U diminishes, if calculating output power by sampling is become large, so control voltage change direction is correct, and control voltage diminishes continuing.
When control voltage U diminishes, if calculating output by sampling is that power diminishes, so illustrate that control voltage change direction is wrong, control voltage will change next step control voltage of direction will become large.
Quantity of state in said process is turned to:
Control voltage change is set to 1 greatly, diminish and be set to 0, power becomes and is set to greatly 1. and diminishes and be set to 0, if (last time becomes and continues greatly to become greatly in the direction of control voltage continuation change last time, diminished last time and continue to diminish) be 1, if changing change direction (last time, change diminished greatly specifically, and diminished last time current change greatly) is 0.
So needing the change direction of control voltage of adjustment actual is this logical equation below:
The continuation change direction of change=control voltage that current change (XOR) output power of control voltage is current
Hill climbing is not high to hardware requirement, realizes, but there is erroneous judgement, the shortcoming that reaction velocity is slow than being easier to.
2, conductance increment method
Suppose that output power is the function of control voltage and can leads,
Known when power is maximum value dP/dU=0
The left side that voltage U drops on maximum power point control voltage is controlled, so control voltage should be increased until dP/dU=0 as dP/dU>0
The right side that voltage U drops on maximum power point control voltage is controlled, so control voltage should be reduced until dP/dU=0 as dP/dU<0
Although conductance increment method judges that precision is high but requires high to hardware requirement height especially sensor computing module, all costs are high to be of little use in Practical Project.
No matter existing searching peak power point methods is various derivative method or various interference method, is if certain any power power of being greater than at adjacent 2 substantially, and judges that this point is maximum power point.Main is all under desirable state, and as shown in Figure 4, method is substantially all if Pi<Pi+1, continues to compare P
i+1and P
i+2if, P
i+1>P
i+2, P
i+1<P
i+2continue again to compare P
i+3.In nonideal situation, the limit of local can be mistaken for the limit of the overall situation.
In real engineering, powertrace is just overall in statistical significance obeys parabolic distribution, and actual on the sampled point of local Discrete is irregular, as shown in Figure 4 rough.The maximum power point that hill climbing interference method is found out can only ensure that the power of this point is greater than the power of adjacent two points, the maximum power point found out by hill climbing and interference method, is probably the maximum power point of local.
Embodiment
Below in conjunction with drawings and Examples, the present invention is described in further detail.
The present invention is a kind of new peak power finding method, i.e. state transition method, first it be not searching maximum power point is regarded as a continuous print sampling compare flow process, but the individual discrete state of n is decomposed in peak power searching, maximum power value in the value of new sampling and present state compares, then determine the state of next step redirect, and next step should sampled value.The definition of n discrete state is determined according to K value.
So after the value determining peak power neighbouring sample scope K, namely ensure that the maximum power point found out must be maximum left and right K sampled point scope.Just can determine the constitutional diagram under this K value.The implication of constitutional diagram is at+K, is that within the scope of 2K, where maximal value is likely in, can determines next step value needing sampling according to the position of maximal value in-K scope.Just the judgement that all needs do can be compared after confirming K value like this and confirm with the work of next step needs.The new power samples value be identified only needs to compare with the maximal value in confirmed K interval, according to comparative result redirect in various state, jumps to the searching that done state has been peak power.
Method of the present invention is highly suitable for the sampling in digital circuit, and the maximum power point that computer program makes a decision is found.Be exactly hill climbing or interference method as K=1, in order to avoid erroneous judgement can increase K value, but also can increase calculated amount, so need in engineering according to each needs doing a balance and compromise between reduction False Rate and the calculated amount of increase.
The present invention is that a kind of anti-erroneous judgement state transition obtains peak power point methods, comprises following step:
Step one: determine security interval value k
Proper security interval value k is got as required with the balance of calculated amount, because the determination of k value means, if control output power by voltage U, the control voltage U of the peak power so determined by this method
maximum power pointthe power that controls of k sampling voltage on the left side namely than U
maximum power pointthe power that k large U controls is certainly than U
maximum power pointthe power controlled is little, and U is compared on same the right
maximum power pointthe power that little control voltage controls is also little than peak power, ensure that at U
maximum power pointin a left and right k voltage range of voltage, the power that it exports is maximum.In principle, the scope of the larger confirmation of k is wider, but can increase calculated amount and detection time.This will do a compromise according to actual demands of engineering, and in the present invention, the value of k is 2, and namely about guarantee maximum power point, the power of 2 sampling spot surveys is certain to the power being less than maximum power point.
Step 2: determine various state according to security interval k and be numbered:
According to the method for definition status, the general define method as k=n is first described, should has (n+1) as n=0
2individual state define method be state 0 in the middle of end-state i.e. maximum power point, the sampled power that left and right respectively has n to be less than it.Then, successively from the left side, all from 0, each increase by 1, until n, like this along with the increase counted in left and right, is defined as 1 to (n+1) respectively
2-1 each state, state is (n+1) altogether
2individual.
As k=1 and be usually said hill climbing, if first determine end-state k=1, control voltage U
maximum power pointthe power P controlled
maximal valuethe right and left next-door neighbour two power to ensure P
maximal value-1<P
maximal value, P
maximal value+1<P
maximal valueso put power P
maximal valueit is exactly maximum power point.Several state is divided into carry out status number it like this, if 3 some power meets maximum end-state when so it is exactly k=1 of intermediate point power, a power points is only had if be defined as state 0. only sampling should be carried out, namely maximum power point is that himself itself is defined as state 1. maximum power point and has a sampled power to be defined as state 2 in the left side on the right, the right has a power points left side not to be defined as state 3, if it is also end-state that left and right respectively has one so just to get back to state 0.
Same principle respectively has the state of 2 power points to be end-state 0 in middle left and right when k=2 maximum power point, initial state be 1 only have a power points about all do not have power points, the left side is for counting 0, and the right is defined as state 2 respectively by the state of 1 to 2, state 3.Count and increase to 1 in the right, the left side, by 0 to 1 to 2, is defined as state 4, and when counting on the right of state 5 state 6. increases again is 2, counting and be increased to 1 to 2 by 0 in the left side, is defined as state 7 state 8 and gets back to state 0.
K=3 can be determined, k=4 according to this method ... various states during k=n under various condition.So just determine various state according to the size of k.
Step 3: the position determining next step sampled point of sampling.
According to the state of present maximum power point, next state from from present state to definition carries out redirect with good conditionsi, the condition jumping to next state is if the power of the sampled point increased newly by next condition is less than existing maximum power point so enter NextState, state (n+1)
2if the next state of-1 and be greater than new state that maximum power point so enters by as described below for end-state 0..
When security interval is n, according to step 2 definition, end-state 0 state is the power points respectively having about maximum power point n to be less than it.J the right is had to have i the point being less than maximum power point if present state is the maximum power point left side, according to definition, it should be in state i-th * (n+1)+j+1 state, so i* (n+1) according to its next state of definition if so its next state of+j+2 j<n should do new sampling in the position of leftmost sampled point-1, namely distance peak power left side j+1 sampling, if j=n so according to definition its NextState be increase a sampled point at rightmost.The position of next sampled point is the position sampling of rightmost sampled point+1, namely apart from the position sampling of i+1 on the right of maximum power point.
The position of next sampled point is so just determined according to the definition of step 2.If that is defining next state is that on the left side increases a point, so sample to subtract with regard on the left side one the new power of station acquisition, if defining next power points is increase a point so just the adds one up-to-date power of station acquisition at rightmost on the right, the state transition gathered for next step after the power of the power points of NextState is ready.
Step 4: according to the state after conditional decision redirect, obtain maximum power point.
After new power points has been sampled, compare with the power of maximum power point, if the power that its value is less than maximum power point enters step 2 definition NextState automatically, if so it gets the hang of and is divided into following several situation the power points of the new sampling peak power that is greater than preexist:
A is greater than existing maximum power point in Far Left sampling and new sampled point power:
If 1 (j+i+1) <n condition is set up ,+1 state that redirect enters (j+i+1) (n+1).And last samples point becomes new up-to-date maximum power point.
The establishment if 2 (j+i+1) <n condition is false, redirect enters n (n+1)+1 state.And last samples point becomes new up-to-date maximum power point.
For redirect A simplicity of explanation once, if the new power gathered is less than existing maximum power point so normally enter next state in definition, peak power remains unchanged, if at the existing maximum power point of the power ratio of Far Left collection, so it will become new maximum power point greatly, like this, the position of maximum power point just becomes leftmost point, at this moment each the existing power points number on the right of it and the last state determining it, because only need respectively there is the individual power points being less than it of n around, in (2) if in their part being greater than n of counting out all be left in the basket, so the situation of all n of being greater than all jumps to maximum power point at Far Left in (2), there is n power points state on the right.And (1) has j+i+1 the state put on the right of maximum power point on the left side ,+1 is little so be considered to now general point to add a point because of the point that relatively more front peak power is up-to-date more.
If B is greater than existing maximum power point in rightmost sampling and new sampled point power: redirect enters n+1 state.And last samples point becomes new up-to-date maximum power point.
Explanation for redirect B is not quite alike, if because be that the right increase by 1 is actually because there has been n sampled point on the left side according to the next state of certain state of definition of step 2, the point on the left side has met defining of peak power dotted state 0, if the new rightmost power gathered is less than existing maximum power point so normally enter next state in definition, peak power remains unchanged, if at the existing maximum power point of the power ratio of rightmost collection, so it will become new maximum power point greatly, like this, the position of maximum power point just becomes rightmost point, at this moment each existing power points number on its left side is greater than n certainly so be all jump to maximum power point at rightmost in this case, there is the state of n power on the left side.
Embodiment:
The anti-erroneous judgement state transition of one of the present invention obtains peak power point methods, comprises following step:
Step one: determine security interval value K
First determine the interval range k of discrete extreme value point, the implication of k is, the size that the value within the scope of maximal value effect k is all less than this maximal value k determines the possibility that local extreme points is mistaken for overall limit.
Step 2: determine various state according to security interval k and be numbered:
Define each (n+1)
2individual state, as shown in Figure 1;
Step 3: determine next step sampled point of sampling position and sampling compare with existing peak power, decision redirect after state.
End-state is state 0, and maximum power point respectively has n power points in middle left and right, and initial state is state 1, only has a power points, so it is thought maximum power point.
At this moment, the next state of state 1 is state 2, and namely on its left side, namely control voltage reduces 1 △ U (control voltage minimum step value) and gathers a power P on the maximum power point left side
2if P
2<P
temporary transient maximal value3 on the left sides that so get the hang of continue sampling and control voltage reduces by 2 △ U.If P
2>P
temporary transient maximal valueso this point just becomes up-to-date maximum power point, and its right has two points to enter 2 (n+1)+1 state.
Have j the right to have i the point being less than maximum power point if present state is the maximum power point left side, according to definition, it should be in state i-th * (n+1)+j+1 state.
If the up-to-date sampled point power of A is less than maximum power point and so jumps to i* (n+1)+j+2 state.
If the up-to-date sampled point power of B is greater than maximum power point and so divides three kinds of situations:
If 1 j=n redirect enters n+1 state.
If 2 (j+i+1) <n condition is set up ,+1 state that redirect enters (j+i+1) (n+1).And last samples point becomes new up-to-date maximum power point.
The establishment if 3 (j+i+1) <n condition is false, redirect enters n (n+1)+1 state.And last samples point becomes new up-to-date maximum power point.
Finally jump to state 0, complete maximum power point and calculate, or jump to left and right boundary point and complete peak power in certain boundary conditions and calculate.Namely sampled point arrives boundary condition, has power maximum point, when peak power the right and left or the condition that can not meet K=n under boundary condition, be exactly briefly maximum power point on one side or the distance of the frontier point on both sides be less than n.
As shown in Figure 2, Fig. 2 is cond redirect;
If n=1 is exactly hill climbing, illustrate now that a kind of situation is as n=2 again.How to calculate.
Step 1 during N=2: first define various states during n=2.Fig. 3 is all states as K=2.
Step 2 during N=2: next step sampling location of each state can be defined according to said process.Following sampling location can be drawn according to figure tri-with about the definition of state:
If present status is that so next step should 1 unit sampling on the maximum power point left side for state 1.
If present status is that so next step should 2 unit samplings on the maximum power point left side for state 2.
If present status is that so next step should 1 unit sampling on the right of maximum power point for state 3.
If present status is that so next step should 1 unit sampling on the maximum power point left side for state 4.
If present status is that so next step should 2 unit samplings on the maximum power point left side for state 5.
If present status is that so next step should 2 unit samplings on the right of maximum power point for state 6.
If present status is that so next step should 1 unit sampling on the maximum power point left side for state 7.
If present status is that so next step should 2 unit samplings on the maximum power point left side for state 8.
Step 3 during N=2: new sampled power relatively after redirect state, obtain maximum power point:
Adopt previous step 3 method sampling after and existing peak power relatively after.
If if new power is less than existing peak power and jumps to state 2, if be greater than existing peak power to jump to 4. when state 1
If if new power is less than existing peak power and jumps to state 3, if be greater than existing peak power to jump to 7. when state 2
If if new power is less than existing peak power and jumps to state 4, if be greater than existing peak power to jump to 3. when state 3
If if new power is less than existing peak power and jumps to state 5, if be greater than existing peak power to jump to 7. when state 4
If if new power is less than existing peak power and jumps to state 6, if be greater than existing peak power to jump to 7. when state 5
If if new power is less than existing peak power and jumps to state 7, if be greater than existing peak power to jump to 3. when state 6
If if new power is less than existing peak power and jumps to state 8, if be greater than existing peak power to jump to 7. when state 7
If if new power is less than existing peak power and jumps to state 0, if be greater than existing peak power to jump to 7. when state 8
If jump to state 0 namely terminate peak power calculating, about meeting when defining K=2, respectively there are 2 points being less than maximum power point.
Finally illustrate a bit clear when program calculation in this way, can the size of control n value very easily, each time be more all the comparison of new sampled point and existing maximum point, only do and once compare, except maximum power point, remaining magnitude relationship a little need not be considered, so also greatly simplifies more complicated.