CN102904249B - Security constraint-based real-time generation planning method - Google Patents

Abstract

The invention discloses a security constraint-based real-time generation planning method. On the basis of a result of a day power generation plan, a turbine generator output plan of each time period in future 30 minutes is a capacity plane of a unit at each stage within 30 minutes in future is complied in an optimized way by combining the ultra-short period load prediction within 30 minutes and taking the security constraint of a power grid into consideration. By the method, the turbine generator output is moderately adjusted according to actual situations by fully utilizing the target and the characteristics of a real-time generation plan on the basis of carrying on the global optimum of the day power generation plan; the power limit in an actual system can be effectively eliminated or retarded; and the safe, excellent and economic operation of a power system is ensured.

Description

A kind of real-time generation schedule method based on security constraint

Technical field

The invention belongs to power system dispatching planning field, be specifically related to a kind of real-time generation schedule method based on security constraint.

Background technology

Real-time generation schedule function is according to ultra-short term, consider the various constraintss such as unit operation constraint, Network Security Constraints, optimize the unit output plan of the establishment day part of following 30 minutes, period interval 5 minutes, it is distribution load between operating unit, does not carry out start and stop adjustment to unit.

Real-time generation schedule needs and in a few days generation schedule matches, and so-called in a few days generation schedule refers to the unit plan optimizing following 1h to following 13h.Compared with real-time generation schedule, the time range of in a few days generation schedule consideration is longer, meets better, global optimization better effects if to contract Constraint, but the accuracy of its load prediction is not as real-time generator plan.By comparing the pluses and minuses of the two, real-time generation schedule makes full use of the accuracy of real time ultra-short term load prediction, and in guarantee with in a few days generation schedule is without under the prerequisite of relatively large deviation, appropriateness adjusts unit output, the more cost effective operation of the system that realizes.

In actual schedule is run, the limit of transmission cross-section and the transmission of electricity such as branch road, transformer element retrains the key factor causing generation schedule optimization not restrain often, because the unappeasable situation of the Network Security Constraints really had, often there is the out-of-limit operation in short-term of some transmission cross-section in the such as peak of power consumption period, out-of-limit in order to eliminate, the unit that generation schedule can employ sensitivity less significantly adjusts, the stable operation of impact scheduling; And if when the unit participating in adjustment does not have power calibration out-of-limit, can cause without feasible solution.For this situation, propose a kind of strategy loosening Network Security Constraints herein, ensure the safe and stable operation of electric power system.

Summary of the invention

For the deficiencies in the prior art, the invention provides a kind of real-time generation schedule method based on security constraint, it combines the advantage of the global optimization of in a few days generation schedule and accuracy two aspect of real time ultra-short term load prediction, realizes the economical operation of electric power system.

A kind of real-time generation schedule method based on security constraint provided by the invention, its improvements are, the time of setting is divided into the time period, carry out optimal load flow calculating successively to each time period; Carry out optimal load flow calculating to each time period to comprise the steps:

(1) according to optimizing the unit output data of N-1 time point and the ultra-short term data of N number of time point that calculate, dynamic communication Load flow calculation is carried out;

(2) according to the dynamic communication calculation of tidal current that step (1) obtains, calculate itself and the power deviation of the in a few days generation schedule of corresponding time point, determine target function and the constraints of optimal load flow;

(3) the dynamic communication calculation of tidal current that obtains of determining step (1), if result of calculation does not have out-of-limit, then carries out optimal load flow calculating according to described target function and constraints, obtains the generation schedule optimum results of N number of period; If result of calculation is out-of-limit, in controlled unit, then select participation optimize the adjustable generating set of calculating and revise out-of-limit limit value, optimal load flow calculating is carried out again according to described target function and constraints, if without feasible solution, loosen out-of-limit limit value until optimal load flow convergence, obtain the generation schedule optimum results of N number of period.

Wherein, when N equals 1, the unit output data of N-1 time point get state estimation result.

Wherein, step (1) ultra-short term data comprise active power data and reactive power data.

Wherein, during step (1) dynamic communication Load flow calculation, the imbalance power that load variations produces is by adjustable generator shared, and the share of bearing of generator is determined according to its frequency response characteristic coefficient.

Wherein, in step (2) day, the time interval of generation schedule is set to 15 minutes, then N number of time point ± within 7.5 minutes, in interval, search in a few days generation schedule result, calculate the deviation of AC power flow unit output and in a few days generation schedule:

$\mathrm{\μ}=\left(\underset{i=1}{\overset{m}{\mathrm{\Σ}}}\right|P_{\mathrm{Gi},N}-P_{\mathrm{Gi},\mathrm{rN}}\left|\right)/P_{D,N}---\left(1\right)$

In formula, m represents total total m platform unit, P _{gi, N}represent the optimization active power of N number of period i-th unit of real-time generation schedule, P _{gi, rN}represent the optimization active power of in a few days N number of period i-th unit of generation schedule, P _{d, N}represent the ultra-short term system loading prediction of N number of period.

Wherein, step (2) determines that the step of the target function of optimal load flow comprises:

If μ ₀for real-time in a few days generation scheduling error threshold, as μ>=μ ₀, select target function:

$\min f=\underset{i=1}{\overset{m}{\mathrm{\Σ}}}|P_{\mathrm{Gi},N}-P_{\mathrm{Gi},\mathrm{rN}}|---\left(2\right)$

Otherwise μ < μ ₀, select target function:

$\min f=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}c\left(P_{\mathrm{Gi},N}\right)---\left(3\right)$

In formula, c (P _{gi, N}) be the generating expense of N number of period i-th unit, real-time generation schedule does not do Unit Combination, does not have switching cost in target function.

Wherein, step (2) determines that the constraints of optimal load flow is:

$P_{\mathrm{Gk},N}-P_{\mathrm{Lk},N}=U_k\underset{j\&Element;k}{\mathrm{\&Sigma;}}{U}_j(G_{\mathrm{kj}}\cos {\mathrm{\&theta;}}_{\mathrm{kj}}+B_{\mathrm{kj}}\sin {\mathrm{\&theta;}}_{\mathrm{kj}})---\left(4\right)$

$Q_{\mathrm{Gk},N}-Q_{\mathrm{Lk},N}=U_k\underset{j\&Element;k}{\mathrm{\&Sigma;}}{U}_j(G_{\mathrm{kj}}\sin {\mathrm{\&theta;}}_{\mathrm{kj}}+B_{\mathrm{kj}}\cos {\mathrm{\&theta;}}_{\mathrm{kj}})---\left(5\right)$

$P_{\mathrm{Gi}}^{\min }\&le;P_{\mathrm{Gi},N}\&le;P_{\mathrm{Gi}}^{\max }---\left(6\right)$

$r_{\mathrm{Gi}}^{\min }\&le;P_{\mathrm{Gi},N}-P_{\mathrm{Gi},N-1}\&le;r_{\mathrm{Gi}}^{\max }---\left(7\right)$

$P_{l,N}\&le;P_l^{\max }---\left(8\right)$

In formula, P _{gk, N}for the optimization of the N number of period node k unit of real-time generation schedule is gained merit, P _{lk, N}for the N number of period node k load prediction active power of real-time generation schedule, Q _{gk, N}for the N number of period node k unit prediction reactive power of real-time generation schedule, Q _{lk, N}for the N number of period node k load prediction reactive power of real-time generation schedule, θ _kjfor the phase angle difference of node k and node j, U _kfor node k voltage magnitude, U _jfor node j voltage magnitude, for the meritorious upper limit of exerting oneself of unit, for the meritorious lower limit of exerting oneself of unit, for the unit ramp loss upper limit, for unit ramp loss lower limit, P _l,Nfor the active power of branch road or section l, for the firm power limit value of branch road or section l.

Wherein, in step (3), if result of calculation is out-of-limit, for certain out-of-limit l, calculate the adjustment index of the adjustable generating set i of its correspondence, it is:

α _l，i＝ω ₁s _li+ω ₂(b _i/b _base)+ω ₃(P _w/P _base) （9）

In formula, s _lifor the meritorious injection of unit i is to the sensitivity of l, b _ifor the inverse of a cost coefficient of unit i, P _wfor unit i is to next period variable capacity (being the heap(ed) capacity that generating set can be adjusted), b _base, P _basefor the base value by a cost coefficient Reciprocals sums variable capacity standardization, ω ₁, ω ₂, ω ₃be respectively the weight coefficient of sensitivity, unit cost, variable capacity.

Wherein, in step (3), if result of calculation is out-of-limit, the method revising out-of-limit limit value is:

The out-of-limit order of severity by sorting from big to small, the numerical value using equal and opposite quantities in pairs method to estimate out-of-limit branch road or section successively can be adjusted to, that is:

For out-of-limit l, after calculating unit adjustment index, upper adjustment unit and lower adjustment unit sort according to corresponding adjustment index is descending, the maximum unit of each selection two carries out pairing and calculates, no longer selected in lower adjustment by the unit selected in upper adjustment, no longer selected in upper adjustment by the unit selected in lower adjustment, when the difference of the adjustment index of selected lower adjustment and upper adjustment unit is less than given threshold value α ₀, the unit optimum selection of out-of-limit l is complete.

After unit optimum selection, the unit participating in calculating is done one's best adjustment, finally obtains the maximum correction amount that each is out-of-limit.For certain out-of-limit l, if its power is P _{l, N}, limit value is maximum correction amount is Δ P _{l, N}.Consider the error of sensitivity linearization approximate, certain conservative estimation is carried out to the recoverable amount of section, be reduced to (1-β) Δ P _{l, N}, wherein, β is limit value coefficient of relaxation, and initial value is set to 5%.

When time, in optimal load flow calculates, the limit value of out-of-limit l is still got otherwise, when time, in optimal load flow calculates, the limit value of out-of-limit l is revised as P _{l, N}-(1-β) Δ P _{l, N}.

In practical power systems, also exist out-of-limit between correlation, the factor such as cause new section out-of-limit, still may cause to optimize calculating without feasible solution.In without feasible solution, each increase β value 10%, revises out-of-limit limit value, restarts to optimize to calculate until convergence.

Compared with the prior art, beneficial effect of the present invention is:

1, present invention incorporates the advantage of the global optimization of in a few days generation schedule and accuracy two aspect of real time ultra-short term load prediction, realize the economical operation of electric power system.

2, the present invention is when consideration Network Security Constraints, has considered sensitivity, exert oneself cost and variable capacity, has obtained rational out-of-limit correcting value of next period, ensures convergence, meets the engine request that practical power systems safety and steady runs.

3, electric power system is nonlinear complication system, and optimal load flow of the present invention meets real time execution situation more than linear programming relax; In addition, computational speed of the present invention, because calculation interval is few, can reaches level level second completely, also meet actual motion demand.

Accompanying drawing explanation

Fig. 1 is the flow chart of the real-time generation schedule method based on security constraint provided by the invention;

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.

A kind of real-time generation schedule method considering security constraint that the present invention adopts, calculate the unit output plan of following 5 to 30 minutes day parts, period interval 5 minutes, the known number of calculation interval is altogether 6.Because the time span of in a few days generation schedule consideration is long, global optimum can be realized better, so real-time generation schedule of the present invention does not optimize calculating 6 periods simultaneously, and adopt from 1 to 6 time period and carry out the method for optimal load flow calculating successively, realize the Real-time Economic Dispatch in a few days generation schedule basis.Its flow chart as shown in Figure 1, calculates N(=1 ~ 6) the concrete enforcement of the generation schedule of individual period is achieved through the following technical solutions:

Step 1: according to optimizing the unit output data of N-1 the time point calculated (if N=1, then get state estimation result) and the ultra-short term data (load prediction comprises meritorious power data and reactive power data) of N number of time point, carry out dynamic communication Load flow calculation.The ultra-short term that the present embodiment proposes refers within 30min.

Wherein, AC power flow adopts dynamic power flow to calculate, and the imbalance power that load variations produces is by adjustable generator shared, and the share of bearing of generator is distributed according to its frequency response characteristic coefficient.

Step 2: step 1 obtains AC power flow unit output result, calculates itself and the power deviation of the in a few days generation schedule of corresponding time point, determines target function and the constraints of optimal load flow.

Wherein: in a few days the time interval of generation schedule is set to 15 minutes, then N number of time point ± within 7.5 minutes, in interval, search in a few days generation schedule result, calculate the deviation of AC power flow unit output and in a few days generation schedule:

$\mathrm{\&mu;}=\left(\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}\right|P_{\mathrm{Gi},N}-P_{\mathrm{Gi},\mathrm{rN}}\left|\right)/P_{D,N}---\left(1\right)$

In formula, m represents total total m platform unit, P _{gi, N}represent the optimization active power of N number of period i-th unit of real-time generation schedule, P _{gi, rN}represent the optimization active power of in a few days N number of period i-th unit of generation schedule, P _{d, N}represent the ultra-short term system loading prediction of N number of period.

If μ ₀for real-time in a few days generation scheduling error threshold, as μ>=μ ₀, select target function:

$\min f=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}|P_{\mathrm{Gi},N}-P_{\mathrm{Gi},\mathrm{rN}}|---\left(2\right)$

Otherwise μ < μ ₀, select target function:

$\min f=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}c\left(P_{\mathrm{Gi},N}\right)---\left(3\right)$

In formula, c (P _{gi, N}) be the generating expense of N number of period i-th unit, real-time generation schedule does not do Unit Combination, does not have switching cost in target function.

Constraints is:

$P_{\mathrm{Gk},N}-P_{\mathrm{Lk},N}=U_k\underset{j\&Element;k}{\mathrm{\&Sigma;}}{U}_j(G_{\mathrm{kj}}\cos {\mathrm{\&theta;}}_{\mathrm{kj}}+B_{\mathrm{kj}}\sin {\mathrm{\&theta;}}_{\mathrm{kj}})---\left(4\right)$

$Q_{\mathrm{Gk},N}-Q_{\mathrm{Lk},N}=U_k\underset{j\&Element;k}{\mathrm{\&Sigma;}}{U}_j(G_{\mathrm{kj}}\sin {\mathrm{\&theta;}}_{\mathrm{kj}}+B_{\mathrm{kj}}\cos {\mathrm{\&theta;}}_{\mathrm{kj}})---\left(5\right)$

$P_{\mathrm{Gi}}^{\min }\&le;P_{\mathrm{Gi},N}\&le;P_{\mathrm{Gi}}^{\max }---\left(6\right)$

$r_{\mathrm{Gi}}^{\min }\&le;P_{\mathrm{Gi},N}-P_{\mathrm{Gi},N-1}\&le;r_{\mathrm{Gi}}^{\max }---\left(7\right)$

$P_{l,N}\&le;P_l^{\max }---\left(8\right)$

In formula, P _{gk, N}represent that the optimization of the N number of period node k unit of real-time generation schedule is gained merit, P _{lk, N}represent the N number of period node k load prediction active power of real-time generation schedule, Q _{gk, N}represent the N number of period node k unit prediction reactive power of real-time generation schedule, Q _{lk, N}represent the N number of period node k load prediction reactive power of real-time generation schedule, θ _kjrepresent the phase angle difference of node k and node j, U _kfor node k voltage magnitude, U _jfor node j voltage magnitude, represent the meritorious upper limit of exerting oneself of unit, represent the meritorious lower limit of exerting oneself of unit, represent the unit ramp loss upper limit, represent unit ramp loss lower limit, P _{l, N}represent the active power of branch road or section l, represent the firm power limit value of branch road or section l.

Step 3: AC power flow result step 1 obtained is divided into two kinds of situations: if a. result of calculation does not exist out-of-limit situation, then directly carry out optimal load flow calculating, because power flow solutions is without feasible solution during security constraint, therefore can convergence be ensured, obtain the generation schedule optimum results of N number of period; If b. there is out-of-limit situation in result of calculation, in controlled unit, then select participation optimize the adjustable unit of calculating and revise out-of-limit limit value, carry out optimal load flow calculating, if without feasible solution, continue to loosen out-of-limit limit value until optimal load flow convergence, obtain the generation schedule optimum results of N number of period.

Wherein:

1. under situation b, for certain out-of-limit l, the adjustment index calculating the adjustable unit i of its correspondence is:

α _l，i＝ω ₁s _li+ω ₂(b _i/b _base)+ω ₃(P _w/P _base) （9）

In formula, s _lifor the meritorious injection of unit i is to the sensitivity of l, b _ifor the inverse of a cost coefficient of unit i, P _wfor unit i is to next period variable capacity, b _base, P _basefor the base value by a cost coefficient Reciprocals sums variable capacity standardization, ω ₁, ω ₂, ω ₃it is the weight coefficient of sensitivity, unit cost, variable capacity.The present embodiment determines according to adjustment index the generating set correcting out-of-limit l in adjustable generating set, and namely when adjustment index exceedes set point (according to user's request setting), corresponding generator is the generator for correcting out-of-limit l.

2. under situation b, the method revising out-of-limit limit value is as follows: the out-of-limit order of severity by sorting from big to small, the reasonable value using equal and opposite quantities in pairs method to estimate out-of-limit branch road or section successively can be adjusted to, that is: for out-of-limit l, after calculating unit adjustment index, upper adjustment unit and lower adjustment unit sort according to corresponding adjustment index is descending, the maximum unit of each selection two carries out pairing and calculates, no longer selected in lower adjustment by the unit selected in upper adjustment, no longer selected in upper adjustment by the unit selected in lower adjustment, when the difference of the adjustment index of selected lower adjustment and upper adjustment unit is less than given threshold value α ₀, the unit optimum selection of out-of-limit l is complete.

After unit optimum selection, the unit participating in calculating is done one's best adjustment, finally obtains the maximum correction amount that each is out-of-limit.For certain out-of-limit l, if its power is P _{l, N}, limit value is maximum correction amount is Δ P _{l, N}.Consider the error of sensitivity linearization approximate, certain conservative estimation is carried out to the recoverable amount of section, be reduced to (1-β) Δ P _{l, N}, wherein, β is limit value coefficient of relaxation, and initial value is set to 5%.

When time, in optimal load flow calculates, the limit value of out-of-limit l is still got otherwise, when time, in optimal load flow calculates, the limit value of out-of-limit l is revised as P _{l, N}-(1-β) Δ P _{l, N}.

In practical power systems, also exist out-of-limit between correlation, the factor such as cause new section out-of-limit, still may cause to optimize calculating without feasible solution.In without feasible solution, each increase β value 10%, revises out-of-limit limit value, restarts to optimize to calculate until convergence.

Finally should be noted that: above embodiment is only in order to illustrate that technical scheme of the present invention is not intended to limit, although with reference to above-described embodiment to invention has been detailed description, those of ordinary skill in the field are to be understood that: still can modify to the specific embodiment of the present invention or equivalent replacement, and not departing from any amendment of spirit and scope of the invention or equivalent replacement, it all should be encompassed in the middle of right of the present invention.

Claims (9)

1. based on a real-time generation schedule method for security constraint, it is characterized in that, the time of setting is divided into the time period, successively optimal load flow calculating is carried out to each time period; Carry out optimal load flow calculating to each time period to comprise the steps:

(1) according to optimizing the unit output data of N-1 time point and the ultra-short term data of N number of time point that calculate, dynamic communication Load flow calculation is carried out;

(2) according to the dynamic communication calculation of tidal current that step (1) obtains, the power deviation of the unit output in calculating dynamic communication calculation of tidal current and the in a few days generation schedule of corresponding time point, determines target function and the constraints of optimal load flow;

(3) the dynamic communication calculation of tidal current that obtains of determining step (1), if result of calculation does not have out-of-limit, then carries out optimal load flow calculating according to described target function and constraints, obtains the generation schedule optimum results of N number of period; If result of calculation is out-of-limit, in controlled unit, then select participation optimize the adjustable generating set of calculating and revise out-of-limit limit value, optimal load flow calculating is carried out again according to described target function and constraints, if without feasible solution, loosen out-of-limit limit value until optimal load flow convergence, obtain the generation schedule optimum results of N number of period.

2. real-time generation schedule method as claimed in claim 1, it is characterized in that, when N equals 1, the unit output data of N-1 time point get state estimation result.

3. real-time generation schedule method as claimed in claim 1, it is characterized in that, step (1) ultra-short term data comprise active power data and reactive power data.

4. real-time generation schedule method as claimed in claim 1, it is characterized in that, during step (1) dynamic communication Load flow calculation, the imbalance power that load variations produces is by adjustable generator shared, and the share of bearing of generator is determined according to its frequency response characteristic coefficient.

5. real-time generation schedule method as claimed in claim 1, it is characterized in that, in step (2) day, the time interval of generation schedule is set to 15 minutes, then N number of time point ± within 7.5 minutes, in interval, search in a few days generation schedule result, calculate the deviation of AC power flow unit output and in a few days generation schedule:

In formula, m represents total total m platform unit, P _{gi, N}represent the optimization active power of N number of period i-th unit of real-time generation schedule, P _{gi, rN}represent the optimization active power of in a few days N number of period i-th unit of generation schedule, P _d,Nrepresent the ultra-short term system loading prediction of N number of period.

6. real-time generation schedule method as claimed in claim 1, it is characterized in that, step (2) determines that the step of the target function of optimal load flow comprises:

If μ ₀for real-time in a few days generation scheduling error threshold, as μ>=μ ₀, select target function:

Otherwise μ < μ ₀, select target function:

In formula: m represents total total m platform unit, P _{gi, N}represent the optimization active power of N number of period i-th unit of real-time generation schedule, P _{gi, rN}represent the optimization active power of in a few days N number of period i-th unit of generation schedule, f is target function, and minf represents the minimization of object function, c (P _{gi, N}) be the generating expense of N number of period i-th unit, real-time generation schedule does not do Unit Combination.

7. real-time generation schedule method as claimed in claim 1, it is characterized in that, step (2) determines that the constraints of optimal load flow is:

In formula, P _{gk, N}for the optimization of the N number of period node k unit of real-time generation schedule is gained merit, P _{lk, N}for the N number of period node k load prediction active power of real-time generation schedule, Q _{gk, N}for the N number of period node k unit prediction reactive power of real-time generation schedule, Q _{lk, N}for the N number of period node k load prediction reactive power of real-time generation schedule, θ _kjfor the phase angle difference of node k and node j, U _kfor node k voltage magnitude, U _jfor node j voltage magnitude, for the meritorious upper limit of exerting oneself of unit, for the meritorious lower limit of exerting oneself of unit, for the unit ramp loss upper limit, for unit ramp loss lower limit, P _l,Nfor the active power of branch road or section l, for the firm power limit value of branch road or section l.

8. real-time generation schedule method as claimed in claim 1, it is characterized in that, in step (3), if result of calculation is out-of-limit, for certain out-of-limit l, calculate the adjustment index of the adjustable generating set i of its correspondence, it is:

α _l,i＝ω ₁s _li+ω ₂(b _i/b _base)+ω ₃(P _w/P _base) (9)

In formula, s _lifor the meritorious injection of unit i is to the sensitivity of l, b _ifor the inverse of a cost coefficient of unit i, P _wfor unit i is to next period variable capacity, b _base, P _basefor the base value by a cost coefficient Reciprocals sums variable capacity standardization, ω ₁, ω ₂, ω ₃be respectively the weight coefficient of sensitivity, unit cost, variable capacity.

9. real-time generation schedule method as claimed in claim 1, it is characterized in that, in step (3), if result of calculation is out-of-limit, the method revising out-of-limit limit value is:

The out-of-limit order of severity by sorting from big to small, the numerical value using equal and opposite quantities in pairs method to estimate out-of-limit branch road or section successively can be adjusted to.