CN101752903A - Time sequence progressive power dispatching method - Google Patents

Abstract

The invention relates to a time sequence progressive power dispatching method, belonging to the technical field of power dispatching automatization, comprising setting three-level dispatching time sequence steps of monthly power generation scheme, weekly power generation scheme, daily power generation scheme; a dispatching center configures a monthly power generation scheme mould of the next month according to the overhaul management information of each set and monthly load prediction information in the electric power plant of the next month and the planned electric quantity of the next month of each set to obtain the monthly power generation scheme of the next month; the dispatching center configures and solves a weekly power generation scheme mould of the next week according to the monthly power generation scheme of the current month to obtain the weekly power generation scheme of the next week; and the dispatching center configures and solves a daily power generation scheme mould according to the weekly power generation scheme of the current week to obtain the daily power generation scheme issued for power plant for carrying out within t+1 days; the practical generating capacity of each electric power plant is accounted at t+2 day, daily-weekly feedback is carried out at t+2 day, and daily-monthly feedback is carried out at the end of the current month. The method can guide safe and economic operation of electric system better.

Description

A kind of time sequence progressive power dispatching method

Technical field

The invention belongs to technical field of power dispatching automation, relate to sequential go forward one by one differentiation generation schedule decision-making technique, the actual motion pattern of coordination approach, each sequential link.

Background technology

Power scheduling is that electrical network adopts various optimal control technology, coordinates various generating resources, keeps the equilibrium of supply and demand, guarantees the main means of power grid security, economical operation.The essence of power scheduling work is that a class is in the certain hour yardstick, under sending out, transmitting electricity resource-constrained situation, industrial optimization problem from time and space angle overall arrangement electrical production, regulation goal relates to operation of power networks cost, electric network security/reliability index or the like, the scheduling variable is unit start and stop state state I and unit output P, is expressed as following form usually:

Regulation goal function: minZ=f (unit start and stop state state P, unit output I)

Power supply and demand balance constraints: h (P, I)=0

Unit, power plant's mechanical property constraints: g (P, I)≤0

Send out, transmission of electricity resource constraint: J (P, I)≤0

Along with the difference of scheduling time yardstick, be divided into mid-term (monthly) and short-term (week, day) scheduling problem.

Growth along with development of electric power industry and electricity needs, each province's generating and transmission system are woven into a complex electric network, scheduling method is to diversified development, the requirement of operation plan promptness improves constantly, and generating is produced and how power grid security coordinates progressively to become the research of power scheduling work and use difficult point.Mid-term (monthly), short-term (week, day) power scheduling work are that the realization electric power resource is distributed core link rationally, and its importance has obtained the common recognition of industrial quarters and academia.

There is following problem at least in existing dispatching technique: the power scheduling work of each sequential link at present generally is isolated carrying out, and most optimum theory and method research that concentrates on generation schedule.Though advanced Optimization Model has obtained tremendous development and extensive use, but the inclined to one side pure mathematicsization of these methods and theorizing, emphasize the optimization of single sequential link, lack whole prediction at aspects such as different timing coordinations and look back range optimization, therefore really do not obtain the overall process of electrical network and optimize effect, restricted the management development of whole operation plan lean, thereby made operation of power networks cost, electric network security/reliability index be difficult to obtain optimum efficiency.

Summary of the invention

The objective of the invention is for overcoming the weak point of prior art, a kind of time sequence progressive power dispatching method is proposed, be scheduling to target with lean, go forward one by one with sequential, from a plurality of aspects such as optimization aim, boundary condition, the ability to ward off risks realize from outward appearance to inner essence, from the superficial to the deep overall process scheduling, guarantee electrical network economy, safety, reliability service.

A kind of time sequence progressive power dispatching method that the present invention proposes is characterized in that, described power scheduling may further comprise the steps for the scheduling that realizes the moon, week, day three sequential links cooperates and coordination:

Step 1, the three-class power scheduling sequential link of monthly generation schedule, all generation schedules, day generation schedule is set; Wherein, the scheduling time scope of monthly generation schedule is the whole month, and the scheduling time scope of all generation schedules is 7～10 days; The scheduling time scope of day generation schedule is a plurality of periods in 1 day;

Step 2, control centre are according to the overhaul management information and the monthly load prediction information of each unit in next month power plant, and the plan electric weight of each unit next month, construct the monthly generation schedule model of next month, and find the solution the monthly generation schedule that obtains next month, each power plant in electrical network issues;

Each unit Zhou Fadian priority in next week is calculated at first according to current month monthly generation schedule by step 3, control centre; Each day start and stop state, the Zhou Fadian priority in next week of unit, overhaul management information and all load prediction information in next week in next week of unit that provides according to current month monthly generation schedule then, structure is also found the solution all generation schedule models in the next week of t+1 day to t+7～10 day, and t was for to work as the day before yesterday; Obtain all generation schedules in next week, issue to power plant;

Step 4, control centre be at first according to all generation schedules in current week, calculates each unit day generating priority of next day; Day generating priority, the overhaul management information of next day and the daily load prediction information of unit next day day part start and stop state that provides according to all generation schedules in current week, unit next day then, structure is also found the solution t+1 day generation schedule model, obtains under day generation schedule power plant for carrying out t+1 day;

After step 5, t+1 day power plant pressed t+1 day generation schedule and carry out, the control centre was at the real energy output of adding up each power plant t+2 day, carrying out t+2 day day-week feedback, the end of month of the current moon carry out the day-feed back by the moon.

Characteristics of the present invention and beneficial effect:

The present invention adopts three grades of timing coordination methods;

1) implement the global optimization of big time scale by monthly generation schedule, be used to arrange unit operation continuously in many days, reduce the start and stop cost, overall process is optimized the on-road efficiency in mid-term;

2) all generation schedules are the coordination links of monthly generation schedule and day generation schedule, for day generation schedule feasible basic generation schedule is provided; The electrical network situation analysis that instant all generation schedules provided slide and advance in 7～10 days for electrical network.Unit start and stop state, the electric weight schedule of further adjustment of the electric network information after utilization is upgraded and the monthly generation schedule of refinement; After monthly plan is changed, guarantee that by compensation adjustment in follow-up many days generating progress and start and stop state meet monthly initial optimization direction, for day decision-making link provides feasible basic generation schedule, avoid the interim start and stop peak regulation in the daily planning.

3) day generation schedule is last link of dispatching of power netwoks decision-making, directly instructs the generating production of next day, can accomplish the explication de texte of electrical network part, and the generating progress error that has taken place is made effective adjustment.Thereby realize that accurate safe is checked and the network loss management; Realize becoming more meticulous of power grid security and economy coordination.

Above characteristics make the inventive method have following advantage: through monthly global optimization, the three grades of sequential decision-making links of fine setting of adjusting targetedly in week, in a few days become more meticulous, the running status of whole electrical network from integral body fuzzy not clear progressively carry out the transition to clear accurately.Power grid risk is shared, is digested, is absorbed in each link, and from the monthly the whole network node line that slowly is retracted to day of risk risk among a small circle on a large scale, power grid risk constantly reduces in the actual motion.It is minimum that the adjustment amount again of day generation schedule and operating pressure can be reduced to, thereby guarantee safe space, transaction space and the scheduling space of electrical network.

Description of drawings

Fig. 1 finds the solution flow chart for a day generation schedule.

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, be described in further detail below in conjunction with accompanying drawing and embodiments of the present invention.

A kind of time sequence progressive power dispatching method that the present invention proposes is characterized in that, described power scheduling may further comprise the steps for the scheduling that realizes the moon, week, day three sequential links cooperates and coordination:

Step 1, the three-class power scheduling sequential link of monthly generation schedule, all generation schedules, day generation schedule is set; Wherein, the scheduling time scope of monthly generation schedule is the whole month, and the scheduling time scope of all generation schedules is 7～10 days (comprising common week and golden week festivals or holidays); The scheduling time scope of day generation schedule is a plurality of periods (as being divided into for 96 periods) in 1 day;

Step 2, control centre are according to the overhaul management information and the monthly load prediction information of each unit in next month power plant, and the plan electric weight of each unit next month, construct the monthly generation schedule model of next month, and find the solution the monthly generation schedule that obtains next month, each power plant in electrical network issues;

Described overhaul management information comprises unit maintenance information and overhaul of the equipments information.Unit maintenance information comprises servicing machine group ID, maintenance from date, maintenance Close Date; In monthly generation schedule model, the unit that is overhauled is in stopped status from overhauling from date till overhauling the Close Date.Overhaul of the equipments information comprises repair apparatus (equipment such as transformer, switch, circuit) ID, maintenance from date, maintenance Close Date; The control centre constructs the constraints of monthly generation schedule according to overhaul of the equipments information calculations electrical network section transmission capacity.

Described monthly load prediction information is electric load conditions of demand next month that (public use) obtains according to monthly load prediction software, the peak load that comprises each day of next month, the load electric weight of each Ri Feng, flat, paddy period, each Ri Feng, flat, each regional Load distribution factor of paddy period, promptly regional load accounts for the ratio of the whole network load.

The plan electric weight of described unit next month exceeds the quata or the adding up of vacancy energy output and the monthly electric weight contract signed next month in this month for each unit.Accumulate mode is if unit real energy output of this month has exceeded the plan electric weight, then to deduct from the monthly electric weight contract of next month; If unit real energy output of this month is then supplied in the monthly electric weight contract next month less than the plan electric weight.

Described monthly generation schedule model is according to a plurality of regulation goals and constraints, structure multi-target non-linear MIXED INTEGER Optimization Model, utilize ideal point approach degree method that above-mentioned four targets are converted into simple target again, use non-linear mixed integer programming at last and find the solution the simple target Optimization Model.

Described current monthly generation schedule is for determining current month unit each day start and stop state and Pinggu, peak period energy output; This monthly generation schedule is estimated roughly as electrical production, and each day electrical network start capacity can be provided, and satisfying each daily load electricity needs of peak period, but does not directly instruct electrical production; Unit each day start and stop that described monthly generation schedule provides are used for being provided with the initial condition of all generation schedule unit day parts.

Each unit Zhou Fadian priority in next week is calculated at first according to current month monthly generation schedule by step 3, control centre; Each day start and stop state, the Zhou Fadian priority in next week of unit, overhaul management information and all load prediction information in next week in next week of unit that provides according to current month monthly generation schedule then, structure is also found the solution all generation schedule models in the next week of t+1 day to t+7～10 day, and t was for to work as the day before yesterday; Obtain all generation schedules in next week, issue to power plant;

The overhaul management information in described next week comprises the unit maintenance information and the overhaul of the equipments information in next week.Unit maintenance information comprises servicing machine group ID, overhauls the initial period, the maintenance processing completion time used for them; In all generation schedule models, the unit that is overhauled is in stopped status from overhauling the initial period till overhauling processing completion time used for them.Overhaul of the equipments information comprises repair apparatus (equipment such as transformer, switch, circuit) ID, overhauls the initial period, the maintenance processing completion time used for them; The control centre constructs the constraints of all generation schedules according to overhaul of the equipments information calculations electrical network section, circuit transmission capacity.

Described all load prediction information is the electric load conditions of demand in the next week that (public use) obtains according to short-term load forecasting software, the load that comprises next all day parts, and each regional Load distribution factor of day part, promptly regional load accounts for the ratio of the whole network load.

Described all generation schedule models are according to regulation goal (operating cost-unit generation cost and unit start and stop cost sum minimize) and constraints, based on DC power flow structure single goal linear mixed-integer Optimization Model, be this Optimization Model decoupling zero unit combination submodel and economic dispatch submodel again, utilize mixed integer programming to find the solution unit combination submodel, utilize Non-Linear Programming to find the solution the economic dispatch submodel.

This in week generation schedule comprise determine each unit the start and stop state of (in promptly following 7～10 days) each period (half an hour or one hour are a period) in next week with exert oneself, provide day part electrical network start capacity and electrical network to exert oneself, to satisfy each daily load electricity needs and the day part load electric weight demand of peak period.

Wherein, the unit day part start and stop state determined of all generation schedules be used for day the generation schedule model be set to the unit actual motion state of next day.

Week, each unit of determining of generation schedule was in the day part in the next week two kinds of purposes of having exerted oneself:

1) exceed 10% unit for monthly electric weight completion rate and the average completion rate deviation of the whole network, day part is exerted oneself and is used to issue execution, day generation schedule and actual schedule in do not make change, to guarantee the contract electric weight schedule of this part unit.

Described monthly electric weight completion rate is the ratio that unit real energy output of this month and unit are answered the energy output plan.

The average completion rate of described the whole network for all units real energy output sums of this month and this month all units answer the ratio of energy output plan sum.

2) for all the other units, day part is exerted oneself and is intended for use in calculating the plan load factor or the plan utilization hourage in next week, obtains these units Zhou Fadian priority in next week.

The Zhou Fadian priority in next week of described each unit is the plan load factor in next week of unit or the plan utilization hourage in next week of unit (can select as required in the actual motion wherein any as generating priority).The plan load factor in next week of unit is but that unit is at the plan electric weight in next week and the ratio of next all unit maximum energy output; The plan load factor is a dimensionless number.The plan utilization hourage in next week of unit be unit at the plan electric weight in next week and the ratio of unit capacity, unit is hour.Utilize " electricity price " of Zhou Fadian priority in all generation schedule models for each unit constructing virtual; Plan load factor with next week is an example, and the electricity price of the unit that the plan load factor in next week is high is lower; Otherwise the electricity price of the unit that the plan load factor in next week is low is higher; In computation optimization according to etc. the principle of little gaining rate arrange the multiple electricity of the high unit of priority, the unit that priority is low generates electricity less.

Described unit the plan electric weight in next week by statistics from the beginning of the month to each unit power stretch amount or vacancy energy output till the day before yesterday, each day in next week of unit energy output of having arranged with current month monthly generation schedule adds up and obtains.

Step 4, control centre be at first according to all generation schedules in current week, calculates each unit day generating priority of next day; Day generating priority, the overhaul management information of next day and the daily load prediction information of unit next day day part start and stop state that provides according to all generation schedules in current week, unit next day then, structure is also found the solution t+1 day generation schedule model, obtains under day generation schedule power plant for carrying out t+1 day;

The overhaul management information of described next day comprises unit maintenance information and the overhaul of the equipments information of next day.Unit maintenance information comprises servicing machine group ID, overhauls the initial period, the maintenance processing completion time used for them; In day generation schedule model, the unit that is overhauled is in stopped status from overhauling the initial period till overhauling processing completion time used for them.Overhaul of the equipments information comprises repair apparatus (equipment such as transformer, switch, circuit) ID, overhauls the initial period, the maintenance processing completion time used for them; The control centre constructs the constraints of day generation schedule according to overhaul of the equipments information calculations electrical network section, circuit transmission capacity.

Described daily load prediction information is the electric load conditions of demand of next day that (public use) obtains according to short-term load forecasting software, the load that comprises next day day part, and each regional Load distribution factor of day part, promptly regional load accounts for the ratio of the whole network load.

The generation schedule model adopted two regulation goals (to purchase electric cost minimization in described day, transmission line margin of safety value maximization, wherein the latter aims to provide circuit transmission of electricity margin of safety, load uncertainty with reply electrical network microcosmic node level, prevent that the circuit trend is out-of-limit) sue for peace as single optimization aim, adopt electric network active, idle, voltage quantities based on AC power flow structure single goal nonlinear optimization model, utilize alternating current-direct current mixing iterative algorithm to find the solution then and obtain.

Described day generation schedule is to determine exerting oneself of each 96 period of unit of next day (15 minutes is a period), and the power plant is carried out down.

The day generating priority of described unit next day is the unit daily load rate, but promptly the unit determined of all generation schedules in current week next day energy output and next day unit maximum energy output ratio; Day generation schedule is " electricity price " of each unit constructing virtual according to this information.The electricity price of the unit that daily load rate is high is lower; Otherwise the electricity price of the unit that daily load rate is low is higher; In computation optimization according to etc. the principle of little gaining rate arrange the multiple electricity of the high unit of priority, the unit that priority is low generates electricity less.

After step 5, t+1 day power plant pressed t+1 day generation schedule and carry out, the control centre was at the real energy output of adding up each power plant t+2 day, carrying out t+2 day day-week feedback, the end of month of the current moon carry out the day-feed back by the moon;

Described day-week is fed back to: statistics from the beginning of the month to each unit power stretch amount or vacancy energy output till the day before yesterday, next all unit each day energy output of having determined with current month monthly generation schedule adds up, again add up all load factors or the Zhou Liyong hourage of unit, obtain the Zhou Fadian priority in next week of unit, forward step 3 to, be used for structure and find the solution next all generation schedules.

Described day-the moon is fed back to: current month each unit power stretch amount of tabulate statistics or vacancy energy output, the electric weight plan of having signed with next month adds up, as unit next month answer the energy output plan, forward step 2 to.

The current monthly generation schedule model of the structure of described step 2 and finding the solution obtains current monthly generation schedule and specifically describes as follows:

Be set as follows regulation goal:

1) day reliability index (LOLP) uniformity;

$\min f_1=\underset{1\≤t\≤T}{\max }\left\{{\mathrm{\ξ}}_t\right\}/\underset{1\≤t\≤T}{\min }\left\{{\mathrm{\ξ}}_t\right\}$

In the formula: t is this per medio of a sequence number; T is total fate of this moon; ξ _tMistake load probability LOLP for t day the whole network; This target function is by minimizing the relative different of each day reliability index, and balanced each day lost the load probability, thereby avoids the difference of the potential operation risk of each day electric power system excessive;

2) the average start and stop cost minimization of the whole network unit;

$\min f_2=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\underset{t=1}{\overset{T}{\mathrm{\Σ}}}|I_i^{t-1}-I_i^t|C_{\mathrm{si}}$

In the formula: i is the unit sequence number; N is the total platform number of unit; I _i ^tBe the running status (1 be operation, 0 be stoppage in transit/maintenance) of unit i in t day; T=0 is defined as the proxima luce (prox. luc) of this planning cycle; I _i ⁰Be defined as the pre-state of unit i; C _SiSingle start and stop cost for unit i;

3) power plant load rate uniformity;

$\min f_3^s=\frac{1}{K}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{\underset{1\≤t\≤T}{\max }\left\{{\mathrm{\ρ}}_k^{t\left(s\right)}\right\}}{\underset{1\≤t\≤T}{\min }\left\{{\mathrm{\ρ}}_k^{t\left(s\right)}\right\}},\∀s=P,F,V$

In the formula: ρ _k ^{T (s)}Be the load factor of the k of power plant in 5 periods of t day;

${\mathrm{\ρ}}_k^{t\left(s\right)}=\left(\underset{i\∈k}{\mathrm{\Σ}}{E}_i^{t\left(s\right)}\right){\left(h^s\underset{i\∈k}{\mathrm{\Σ}}{I}_i^t{C}_i^{\max }\right)}^{\text{-1}}$

Segment type when s is (P, F, V represent peak, flat, paddy respectively), E _i ^{T (s)}Be unit i t day s period energy output; h ^sIt is the hourage of s period in one day; The present invention is in order to solve " simultaneity " problem of electric weight plan and direction of energy safety, and the decomposition of monthly plan electric weight and safety are checked and refine to peak, flat, paddy section, with indirect consideration power balance, improve electric network security; K is power plant's sequence number; K is power plant's total number; C _i ^MaxBe the capacity of unit i, this target function is by minimizing the relative different of power plant's day part every day load factor, and the nargin that power plant is kept in balance at day part is with the load random fluctuation in the reply daily planning;

4) the zone clean electric weight uniformity of injecting of each day;

$\min f_4^s=\frac{1}{M}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\underset{1\≤t\≤T}{\max }\left\{\right|F_m^{t\left(s\right)}\left|\right\}/\underset{1\≤t\≤T}{\min }\left\{\right|F_m^{t\left(s\right)}\left|\right\},\∀s=P,F,V$

In the formula: m is regional sequence number; M is regional number; F _m ^{T (s)}Be the clean electric weight (to be injected to positive direction) that injects of the t day s period in m area:

$F_m^{t\left(s\right)}=D^t{\mathrm{\β}}_m^t{\mathrm{\γ}}^{t\left(s\right)}-\underset{i\∈m}{\mathrm{\Σ}}{E}_i^{t\left(s\right)}$

In the formula: D ^tBe the whole network t day total load electric weight; β _m ^tAccount for the scale factor of the whole network t day total load electric weight at the load electric weight of t day for regional m; γ ^{T (s)}Account for the ratio of t day total load electric weight for the load electric weight of the whole network t day s period; E _i ^{T (s)}For this area unit i at t day s period energy output; The section of day part injects the electric weight fluctuation to this target function only by minimizing in a few days, for real time execution is reserved balanced transmission line capability nargin; This target is not considered the unfixed section of actual direction of tide;

The constraints of monthly generation schedule model:

1) unit plan electric weight constraint

$\left\{\begin{array}{c}\underset{t=1}{\overset{T}{\mathrm{\Σ}}}(E_i^{t\left(P\right)}+E_i^{t\left(F\right)}+E_i^{t\left(V\right)})=E_i\\ \∀i=\mathrm{1,2},...,N\end{array}\right.$

E in the following formula _iAnswer the energy output plan for unit i this month;

2) unit is specified the running status constraint

Above-mentioned constraint is only worked to following two class units: (1) arranges the unit of maintenance need specify maintenance outage; The unit that the electric weight contract has been finished can require to arrange to shut down according to power plant; (2) the slower unit of electric weight progress needs to specify start;

3) peak, load electric quantity balancing flat, the paddy section retrain

$\left\{\begin{array}{c}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{E}_i^{t\left(s\right)}=D^t{\mathrm{\γ}}^{t\left(s\right)}\\ \∀t=\mathrm{1,2},...,T\∀s=P,F,V\end{array}\right.$

4) the whole network start abundant intensity constraint

$\left\{\begin{array}{c}{L}_{\max }^t(1+\underset{\‾}{r})\≤\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{I}_i^t{C}_i^{\max }\≤L_{\max }^t(1+\stackrel{\‾}{r})\\ \∀i=\mathrm{1,2},...,N,\∀t=\mathrm{1,2},...,T\end{array}\right.$

In the following formula: L _Max ^tBe t day maximum load,

rBe respectively the predefined day maximum load period positive rotation percentage reserve upper limit, lower limit in generation schedule;

5) unit peak, flat, paddy period generating capacity constraint in a few days

$\left\{\begin{array}{c}{I}_i^t\underset{\‾}{E_i^{t\left(s\right)}}\≤I_i^t{E}_i^{t\left(s\right)}\≤I_i^t\stackrel{\‾}{E_i^{t\left(s\right)}}\\ \∀i=\mathrm{1,2},...,N\∀t=\mathrm{1,2},...,T\∀s=P,F,V\end{array}\right.$

In the following formula E _i ^{T (s)} ,

Be respectively the lower limit and the upper limit of unit i t day s period energy output;

6) area peak, flat, clean injection of paddy period electric weight constraint in a few days

$\left\{\begin{array}{c}\underset{\‾}{F_m^{t\left(s\right)}}\≤F_m^{t\left(s\right)}\≤\stackrel{\‾}{F_m^{t\left(s\right)}}\\ \∀t=\mathrm{1,2},...T\∀m=\mathrm{1,2},...M\∀s=P,F,V\end{array}\right.$

In the following formula F _m ^{T (s)} ,

Be the clean electric weight lower limit and the upper limit injected of m area t day s period;

7) unit continuous working period constraint

$\left\{\begin{array}{c}[X_i^{\mathrm{on}}(t-1)-T_i^{\mathrm{on}}](I_i^{t-1}-I_i^t)\≥0\\ [X_i^{\mathrm{off}}(t-1)-T_i^{\mathrm{off}}](I_i^t-I_i^{t-1})\≥0\\ \∀t=\mathrm{1,2},...,T\∀i=\mathrm{1,2},...,N\end{array}\right.$

In the following formula: X _i ^On(t) be that unit has continued the start fate at t during day; X _i ^Off(t) be that unit i has continued to shut down fate at t during day; X during t=0 _i ^On(0)/X _i ^Off(0) is defined as the pre-state I of unit i _i ⁰The fate that continues; T _i ^OffContinue to shut down fate for unit i is minimum; T _i ^OnContinue the start fate for unit i is minimum;

The present invention is based on ideal point approach degree method four regulation goals (sub-goal) are converted into single regulation goal; Each sub-goal f (x) is minimized form, and the object set vector representation that then comprises p sub-goal is:

$F\left(x\right)=\underset{x\∈R}{\min }{(f_1\left(x\right),f_2\left(x\right),...,f_p\left(x\right))}^T$

The ideal point of multi-objective optimization question comprises positive ideal point (lower limit that each sub-goal is the most extreme) and negative ideal point (the acceptable upper limit of each sub-goal); Set distributed area

Optimize each sub-goal respectively and obtain the most extreme lower limit f _j Determine the acceptable upper limit according to the operation of power networks rules

In order to eliminate the influence of the dimension and the order of magnitude, adopt the average statistical f of sub-goal to the result of decision _{J_M}Each sub-goal and distributed area are carried out standardization processing:

$\left\{\begin{array}{c}\underset{\‾}{f_j^{\′}}\left(x\right)=f_j\left(x\right)/f_{j\_M}\\ \stackrel{\‾}{f_j^{\′}}=\stackrel{\‾}{f_j}/f_{j\_M}\\ \underset{\‾}{f_j^{\′}}=\underset{\‾}{f_j}/f_{j\_M},\∀j=\mathrm{1,2},...,p\end{array}\right.$

Positive and negative ideal point F, negative ideal point

Expression formula be respectively:

Calculate sub-goal collection vector and positive ideal point Euclidean distance d ⁺, with negative ideal point Euclidean distance d ^-:

$\left\{\begin{array}{c}{d}^{+}=\sqrt{\underset{j=1}{\overset{p}{\mathrm{\Σ}}}{(f_j^{\′}\left(x\right)-\underset{\‾}{f_j^{\′}})}^2}\\ d^{-}=\sqrt{\underset{j=1}{\overset{p}{\mathrm{\Σ}}}{(f_j^{\′}\left(x\right)-\stackrel{\‾}{f_j^{\′}})}^2}\end{array}\right.$

The approach degree that draws current sub-goal collection vector and positive ideal point thus is:

$d=\frac{d^{+}}{d^{-}+d^{+}}$

The ideal point approach degree characterized current sub-goal collection vector press close to positive ideal point, away from the relative extent of negative ideal point, physical significance is clear, is convenient to practical operation, can be used as simple target and is used for computation optimization;

Adopt non-linear mixed integer programming algorithm to find the solution above-mentioned model, can obtain monthly generation schedule;

All generation schedule models in next week of the structure of step 3 are also found the solution all generation schedules that obtain next week and are specifically described as follows:

All generation schedule simulated target functions are as follows:

$\min C=\underset{i\∈G}{\mathrm{\Σ}}\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\underset{h=1}{\overset{H}{\mathrm{\Σ}}}{c}_i\left(P_i^{t\left(h\right)}\right)+\underset{i\∈G}{\mathrm{\Σ}}\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\underset{h=1}{\overset{H}{\mathrm{\Σ}}}|I_i^{t\left(h\right)}-I_i^{t(h-1)}|C_{\mathrm{si}}$

Wherein G is the generating set set, and t is the date sequence number, and T is total fate of week plan; 1 _i ^{T (P)}Be the start and stop state (=1: start ,=0: shut down) of unit i at t day peak period P; G is the unit set; c _SiSingle start and stop cost for unit i; H is each day hop count when total, and h is the period sequence number; I during h=0 _i ^{T (h)}Be defined as the most last period state of proxima luce (prox. luc); P _i ^{T (h)}For unit i exerts oneself c in the t day h period _i(P _i ^{T (h)}) purchase electric cost function for unit; For three public scheduling methods, it is relevant with the generating priority of unit that unit is purchased electric cost function; Under the energy-saving distribution pattern, it is relevant with the energy consumption ordering of unit that unit is purchased electric cost function;

All generation schedules adopt following constraints:

1) unit day start and stop state is adjusted direction constrain

$(I_i^{t\left(P\right)}-{\tilde{I}}_i^t)(L_{\max }^t-\widetilde{L_{\max }^t})\≥0,\∀i\∈G,\∀t\∈T_{\mathrm{re}\_\mathrm{UC}}$

$I_i^{t\left(P\right)}={\tilde{I}}_i^t,\∀i\∈G,\∀t\∈T_{m\_\mathrm{UC}}$

1) I in the formula _i ^{T (P)}Be the start and stop state of unit i at t daily load P peak period;

Be the start and stop state of unit i in the monthly plan in t day,

Be t daily load peak value in the monthly load data, L _Max ^tBe t daily load peak value in all load prediction data; 1) T in the formula _{Re_UC}Because load deviation is excessive, need rearrange the date of unit day start and stop state for this week; T _{M_UC}Be defined as and remove T this week _{Re_UC}Do not need the date set of unit recombinant outward; If in the result of decision of monthly plan, unit is reasonable at the start capacity of t day peak period, then continues to use the start and stop result of monthly plan;

2) unit electric weight this week schedule constraint

$\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\underset{h=1}{\overset{H}{\mathrm{\Σ}}}{I}_i^{t\left(P\right)}\underset{\‾}{E_i^t}\≤E_i\≤\underset{t=1}{\overset{T}{\mathrm{\Σ}}}\underset{h=1}{\overset{H}{\mathrm{\Σ}}}{I}_i^{t\left(P\right)}\stackrel{\‾}{E_i^t},\∀i\∈G$

E in the following formula _iFor planning electric weight this week of monthly plan formulation; In the following formula E _i ^t ,

Be respectively the lower limit and the upper limit of unit i t day energy output; Following formula can make the electric weight progress of monthly arrangement drop in the unit generating capacity this week scope, guarantees that the electric weight progress of monthly plan is obtaining succession week in the works and realizing;

Constraints 1) and 2) show, in all generation schedules, the unit start stop operation must be based on monthly start and stop state, and only start capacity vacancy or the redundant enforcement that load deviation is caused is adjusted, and will guarantee in the recombinant operation that each unit capacity variable quantity and load variations amount keep with increasing or subtracting together; 1) formula and 2) formula can avoid all generation schedules that the unit start and stop state of monthly generation schedule is made too much modification, can inherit the global optimization effect of monthly generation schedule to greatest extent;

3). meritorious equilibrium equation

$\underset{i\∈G}{\mathrm{\Σ}}{P}_i^{t\left(h\right)}=D^{t\left(h\right)},\∀h=\mathrm{1,2},...,H$

D wherein ^{T (h)}The whole network load for the t day h period;

4). the constraint of the whole network start abundant intensity

$L_{\max }^t(1+\underset{\‾}{r})\≤\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{I}_i^{t\left(P\right)}{C}_i^{\max }\≤L_{\max }^t(1+\stackrel{\‾}{r}),\∀t=\mathrm{1,2},...,T$

This constraints is introduced the zone of reasonableness of the information constrained electrical network of all load predictions peak period start capacity;

5) section tidal current constraint

$|\underset{i\∈m}{\mathrm{\Σ}}{P}_i^{t\left(h\right)}-D_m^{t\left(h\right)}|\≤\stackrel{\‾}{F_m^{t\left(h\right)}},m\∈M\∀h=\mathrm{1,2},...,H$

D wherein _m ^{T (h)}For regional m at the t load of day h period; Wherein M is the area set;

Trend safety margins for section m period h;

6) circuit trend constraint

$|\underset{i\∈G}{\mathrm{\Σ}}{G_{i\_l}}^{t\left(h\right)}{P}_i^{t\left(h\right)}+\underset{j\∈D}{\mathrm{\Σ}}{G_{j\_l}}^{t\left(h\right)}{D}_i^{t\left(h\right)}|\≤\stackrel{\‾}{F_l^{t\left(h\right)}},\∀l\∈L\∀h=\mathrm{1,2},...,H$

G wherein _{I_l} ^{T (h)}Be the generated output transfer distribution factor of unit t day h period i to circuit l,

Be the meritorious trend upper limit of circuit l in the t day h period; L is a line set; D is the set of load bus;

7) the adjustable limit value constraint of exerting oneself of unit

$\left\{\begin{array}{c}{I}_i^{t\left(h\right)}{C}_i^{\min }\≤P_i^{t\left(h\right)}\≤I_i^{t\left(h\right)}{C}_i^{\max }\\ P_i^{\left(t\right)(h-1)}-\underset{\‾}{v_i}\≤P_i^{t\left(h\right)}\≤P_i^{\left(t\right)(h-1)}+\stackrel{\‾}{v_i}\end{array}\right.\∀i\∈G^t,\∀h=\mathrm{1,2},...,H,\∀t=\mathrm{1,2},...,T$

Following formula has been considered unit capacity constraint and climbing capacity constraint; P during h=0 _i ^{T (h)}Be defined as the exerting oneself of pre-state of unit t day, promptly go up one day the most last period and exert oneself;

8) unit continuous working period constraint;

$\left\{\begin{array}{c}[X_i^{\mathrm{on}}\left(t(h-1)\right)-T_i^{\mathrm{on}}](I_i^{t(h-1)}-I_i^{t\left(h\right)})\≥0\\ [X_i^{\mathrm{off}}\left(t(h-1)\right)-T_i^{\mathrm{off}}](I_i^{t\left(h\right)}-I_i^{t(h-1)})\≥0\\ \∀t=\mathrm{1,2},...,T\∀i\∈G,\∀h=\mathrm{1,2},...,H\end{array}\right.$

In the following formula: X _i ^On(t (h)) is unit hop count when the t day h period continues start; X _i ^Off(t (h)) is unit i hop count when the t day h period continues to shut down; X during t=0 _i ^On(0), X _i ^Off(0) is defined as the pre-state I of unit i _i ⁰The time hop count that continues; T _i ^OffHop count when continuing to shut down for unit i is minimum; T _i ^OnHop count when continuing start for unit i is minimum;

9) the dynamic adjustments ability at the electrical network equilibrium of supply and demand retrains

9-1) the t day the whole network load period h that skyrockets:

$\underset{i}{\mathrm{\Σ}}\stackrel{\‾}{{\mathrm{\ΔP}}_i^{t\left(h\right),t(h+1)}}\≥\stackrel{\‾}{D^{t(h+1)}}-D^{t\left(h\right)}\∀i\∈G^{t\left(h\right)}$

In the following formula

Maximum from the h period to the h+1 period raises ability in t day for unit i;

$\stackrel{\&OverBar;}{{\mathrm{\&Delta;P}}_i^{t\left(h\right),t(h+1)}}=\left\{\begin{array}{cc}\stackrel{\&OverBar;}{v_i}& f{C}_i^{\max }\&GreaterEqual;P_i^{t\left(\right)}+\stackrel{\&OverBar;}{v_i}\\ C_i^{\max }-P_i^{t\left(h\right)}& \mathrm{if}{C}_i^{\max }<P_i^{t\left(h\right)}+\stackrel{\&OverBar;}{v_i}\end{array}\right.$

Wherein

Be the single period climbing capacity of unit i, P _i ^{T (h)}Be meritorious the exerting oneself of unit i in the t day h period; D ^{T (h)}The whole network load for the t day h period;

For the h period loads and carries out the interval upper limit; G ^{T (h)}In unit be divided into two classes: satisfy

Condition for rising rapid change unit, the rise of this class unit is exerted oneself and is not subjected to capacity limit, defines the set G of this class unit ^{T (h), U}Satisfy

Condition for rising gradual unit, this class unit is subjected to capacity limit, the scope that can raise is less, defines its set G ^{T (h), u}

9-1) physical significance of formula is: if the whole network unit began to raise with maximal rate from the h period, and should be to the maximum output that the h+1 period can provide greater than the load confidential interval upper limit;

9-2) period h falls in t day the whole network load suddenly:

$\underset{i}{\mathrm{\&Sigma;}}\underset{\&OverBar;}{{\mathrm{\&Delta;P}}_i^{t\left(h\right),t(h+1)}}\&GreaterEqual;D^{t\left(h\right)}-\underset{\&OverBar;}{D^{t(h+1)}}\&ForAll;i\&Element;G^{t\left(h\right)}$

Wherein Δ P _i ^{T (h), t (h+1)} Maximum downward modulation ability for unit i from the h period to the h+1 period;

$\underset{\&OverBar;}{{\mathrm{\&Delta;P}}_i^{t\left(h\right),t(h+1)}}=\left\{\begin{array}{cc}\underset{\&OverBar;}{v_i}& f{C}_i^{\min }\&le;P_i^{t\left(h\right)}-\underset{\&OverBar;}{v_i}\\ P_i^{t\left(h\right)}-C_i^{\min }& \mathrm{if}{C}_i^{\min }>P_i^{t\left(h\right)}-\underset{\&OverBar;}{v_i}\end{array}\right.$

v _i Be unit i downward modulation speed; Definition is satisfied

For decline rapid change unit, affiliated set G ^{T (h), D}Definition is satisfied

Unit for descending gradual unit, affiliated set G ^{T (h), d}

9-2) physical significance of formula is: if unit began to reduce with maximal rate from the h period, and should be to the maximum output that the h+1 period can provide greater than load confidential interval lower limit;

10) the dynamic adjustments ability at the section fail safe retrains

10-1) be subjected to the electric regional m load period h that skyrockets t day:

Wherein

Be the upper limit of regional m in the load confidential interval of period h+1;

Trend safety margins for section m period h+1; Area m is at the load of period h+1

S wherein _m ^{T (h+1)}For h+1 period m regional load accounts for the ratio that the whole network is loaded, i.e. the regional load factor; 10-1) physical significance of formula is: if unit began to raise with maximal rate from the h period in the regional m, should provide enough exerting oneself to the h+1 period, guarantee that section injection trend does not exceed safety margins;

10-2) send electric regional m ' load to fall period h suddenly t day:

Wherein D _m ^{T (h+1)} Lower limit for the load confidential interval of regional m ' period h+1; 10-2) physical significance of formula is: if the interior unit of regional m ' began with maximal rate downward modulation from the h period, should guarantee to exert oneself to the h+1 period drops to enough lowly, guarantees that section exports trend and do not exceed safety margins;

Wherein M is the area set; The unit in area is not only considered the dynamic constrained at the electrical network equilibrium of supply and demand under having; And the unit in possession district not only provides regulating power for the whole network load equilibrium of supply and demand to some extent, and provides the dynamic adjustments ability for section tidal current safety; The two is contradiction not, because the tendency of regional load and the whole network load tendency is basic identical, so the unit in m area also provides regulating power for the whole network equilibrium of supply and demand when providing regulating power for this locality load;

11) unit is specified the running status constraint

Above-mentioned constraint is only worked to following two class units: (1) arranges the unit of maintenance need specify maintenance outage; The unit that the electric weight contract has been finished can require to arrange to shut down according to power plant; (2) the slower unit of electric weight progress needs to specify start;

All generation schedule models can be decomposed into unit combination and two sub-Optimization Model of economic dispatch;

Unit makes up sub-Optimization Model:

$\left\{\begin{array}{c}\min C=\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}|I_i^{(t-1)\left(P\right)}-I_i^{t\left(P\right)}|C_{\mathrm{si}}\\ s.t1),2),4),12)\end{array}\right.$

The sub-Optimization Model of economic dispatch is:

$\left\{\begin{array}{c}\min C=\underset{i\&Element;G}{\mathrm{\&Sigma;}}\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}\underset{h=1}{\overset{H}{\mathrm{\&Sigma;}}}{c}_i\left(P_i^{t\left(h\right)}\right)\\ s.t3),5)~11)\end{array}\right.$

Adopt mixed integer programming to find the solution the unit group successively and make up sub-Optimization Model, adopt Non-Linear Programming to find the solution the sub-Optimization Model of economic dispatch, can obtain all generation schedules;

The day generation schedule model and find the solution the day generation schedule that obtains next day and specifically describe as follows of structure next day of step 4:

The target function of day generation schedule has been taken all factors into consideration electrical network economy and fail safe two class decision elements, be intended to seek economic benefit best with power grid security nargin ensure compromise, with the list period be example:

$\min C=\underset{i\&Element;G}{\mathrm{\&Sigma;}}{c}_i\left(P_{\mathrm{Gi}}\right)-\underset{l\&Element;\mathrm{AL}}{\mathrm{\&Sigma;}}{\mathrm{\&lambda;}}_l(S_{l,\max }-|S_l\left|\right)$

Wherein G is the unit set; Wherein, c _iFor unit i purchases electricity charge function; P _GiExert oneself for unit i is meritorious; A part is for purchasing the electricity charge with minimum before the model, and a back part is maximum for the transmission line margin of safety is worth; For three public scheduling methods, purchase electric cost c _i(P _Gi) relevant with the electric weight progress completion rate of unit; Under the energy-saving distribution pattern, it is relevant with the energy consumption ordering of unit to purchase electric cost; Under market environment, unit is purchased electric cost and is depended on the unit quotation; The present invention utilizes the virtual electricity price of generating priority structure unit; Wherein AL is the key transmission sets of lines, need be in conjunction with scheduling experience and operation of power networks situation by manually choosing; The weight coefficient λ relevant with the transmission line margin of safety _l〉=0, its economics implication is the value of circuit l unit margin of safety among the key transmission sets of lines AL, is the cost that the dispatcher is willing to mean secure payment; As all λ _lWhen all getting null value, this model is pursues the traditional economy scheduling model of purchasing electric cost minimization; Need satisfy following constraints:

1). the node power balance

$P_k-{\mathrm{jQ}}_k=\underset{j\&Element;k}{\mathrm{\&Sigma;}}{\stackrel{\&CenterDot;}{S}}_{\mathrm{kj}}^k+{\stackrel{\&CenterDot;}{S}}_{k0}\&ForAll;k\&Element;N$

2). the constraint of unit output limit value

P _Gi，min≤P _Gi≤P _Gi，max

3). the constraint of circuit trend

|S _i|≤S _l，max

Wherein N is a node set, P _kBe meritorious clean injection of node k; Q _kBe the idle injection of node k;

For sending the conjugation of power on the circuit between node k, j by the k node,

Be the node k conjugation of loss power over the ground; L is a line set, S _lBe the apparent power of circuit 1, S _{L, max}Be the steady limit of the heat of circuit l;

Ignore the variation of circuit trend idle component in the meritorious scheduling, based on the simplification of DC power flow, master mould is equivalent to following DC Model:

$\left\{\begin{array}{c}\min C=\underset{i\&Element;G}{\mathrm{\&Sigma;}}{c}_i\left(P_{\mathrm{Gi}}\right)+\underset{l\&Element;\mathrm{AL}}{\mathrm{\&Sigma;}}{\mathrm{\&lambda;}}_l\left(\underset{i\&Element;G}{\mathrm{\&Sigma;}}{G}_{i\_l}{P}_{\mathrm{Gi}}\right)\\ \underset{i\&Element;G}{\mathrm{\&Sigma;}}{P}_{\mathrm{Gi}}=D+P_{\mathrm{Loss}}\\ P_{\mathrm{Gi},\min }\&le;P_{\mathrm{Gi}}\&le;P_{\mathrm{Gi},\max }\&ForAll;i\&Element;G\\ F_l\&ap;F_l^P\&le;\sqrt{{\left|S_{l,\max }\right|}^2-{\left(F_l^Q\right)}^2}=F_{l,\max }\end{array}\right.$

The present invention has designed the computing mechanism of alternating current-direct current mixing iteration and has found the solution a day generation schedule model; At first utilize the linear optimization algorithm to find the solution above-mentioned DC Model, obtain initial feasible day generation schedule, enter iterative process then: based on up-to-date meritorious result of calculation, use the P-Q decomposition method to finish the AC power flow iteration one time, revise electric network swim; According to current system losses and circuit AC power flow, upgrade the meritorious balance constraints and the circuit trend constraints of DC Model, call the linear optimization algorithm and find the solution; Enter the next round iteration then, out-of-limit and trend convergence until the power grid wireless road; As shown in Figure 1;

The effect that the inventive method realizes illustrates as follows:

One, the start and stop cost of monthly generation schedule and traditional scheduler method contrast among the present invention

Be scheduling to example with 6 months unit start and stop of certain provincial power network, the start and stop cost variance of monthly generation schedule of the present invention and traditional scheduler method has been showed in table 1 contrast: the whole network unit startup-shutdown number of times that adopts monthly generation schedule of the present invention to calculate has reduced by 274 times than traditional scheduler method, saves 3,854 ten thousand yuan of start and stop costs;

Table 1

Power plant	Reduce number of times	Save the start and stop cost
			??1	??75	??868.15
??2	??37	??561.6
			??3	??32	??192
??4	??22	??380
			??5	??22	??380
??6	??21	??172.62
			??7	??18	??440.5
??8	??18	??127.8
			??9	??14	??560
??10	??9	??111
			??11	??6	??59.9

Power plant	Reduce number of times	Save the start and stop cost
			Amount to	??274	??3854

Two, the present invention's week generation schedule contrasts with the start and stop cost of tradition day generation schedule

Table 2

With 7 days unit start and stop of certain provincial power network decision-making is example, and the start and stop cost of all generation schedules of the present invention and tradition day generation schedule has been showed in table 2 contrast.Monthly generation schedule is not considered in the tradition daily planning, and only based on proxima luce (prox. luc) running status and the start and stop of workload demand decision-making on same day unit, randomness is bigger, causes in the follow up date start and stop cost high always; All generation schedules can be inherited the global optimization effect of monthly generation schedule, and consider minimizing of newly-increased start and stop cost; All generation schedules and a tradition day generation schedule can directly instruct the unit start and stop, and still total start and stop cost of traditional day generation schedule exceeds 12.3% than total start and stop cost of all generation schedules;

Three, the regional maximum of the present invention's week generation schedule and the calculating of traditional scheduler method can raise the contrast of exerting oneself

Table 3

41～43 period of the evening peak Monday equilibrium of supply and demand with somewhere in certain provincial power network is scheduling to example, listed respectively in the table 3 utilize the present invention week generation schedule and the regional maximum of calculating respectively of traditional scheduler method can raise and exert oneself; In the result of calculation of traditional scheduler method, the major part of unit adjusting space is used for the follow load variation during evening peak, therefore remains the rise ability and can't tackle upward fluctuating that regional load occurs in 41 and 42 periods, and section injects trend may be out-of-limit; In the result of calculation of the present invention's week generation schedule, unit has been reserved certain dynamic rise space before load skyrockets, and therefore the regional maximum of 41,42 periods can raise to exert oneself and significantly improve, and can guarantee the power grid security under the load fluctuation situation;

Four, the active power loss contrast of of the present invention day generation schedule and generation schedule calculating in traditional day

Table 4

With the IEEE30 meshed network is case study, contrast that day generation schedule of the present invention and tradition day generation schedule calculate 96 day period generation schedule.By table 4 as seen, the active power loss that day generation schedule of the present invention calculates is littler by 23.16% than the result of calculation of tradition day generation schedule, falls that to decrease effect obvious;

Five, the safe check result contrast of of the present invention day generation schedule and traditional day generation schedule

Table 5

The transmission capacity of setting circuit 31 is 18MVA, in tradition day generation schedule according to 16.2MW (capacity of trunk * 0.9) as meritorious transmission limit.The safe check result of of the present invention day generation schedule of comparative analysis and traditional day generation schedule.By table 5 as seen, a tradition day generation schedule can only adopt DC power flow to do safe check, though that the DC power flow result of circuit 31 does not have is out-of-limit, calculates the electrical network AC power flow of this moment, the result shows that the apparent power of circuit 31 has surpassed 18MVA, and is in fact also dangerous; And of the present invention day generation schedule adopts AC power flow to do safe check, and the circuit trend is distributed between 17.2～17.3MVA, satisfies the safety requirements of operation of power networks fully.

The above only is preferred embodiment of the present invention, not in order to restriction the present invention, all any modifications of being done within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.

Claims (5)

1. a time sequence progressive power dispatching method is characterized in that, described power scheduling may further comprise the steps for the scheduling that realizes the moon, week, day three sequential links cooperates and coordination:

Step 1, the three-class power scheduling sequential link of monthly generation schedule, all generation schedules, day generation schedule is set; Wherein, the scheduling time scope of monthly generation schedule is the whole month, and the scheduling time scope of all generation schedules is 7～10 days; The scheduling time scope of day generation schedule is a plurality of periods in 1 day;

Step 2, control centre are according to the overhaul management information and the monthly load prediction information of each unit in next month power plant, and the plan electric weight of each unit next month, construct the monthly generation schedule model of next month, and find the solution the monthly generation schedule that obtains next month, each power plant in electrical network issues;

Each unit Zhou Fadian priority in next week is calculated at first according to current month monthly generation schedule by step 3, control centre; Each day start and stop state, the Zhou Fadian priority in next week of unit, overhaul management information and all load prediction information in next week in next week of unit that provides according to current month monthly generation schedule then, structure is also found the solution all generation schedule models in the next week of t+1 day to t+7～10 day, and t was for to work as the day before yesterday; Obtain all generation schedules in next week, issue to power plant;

Step 4, control centre be at first according to all generation schedules in current week, calculates each unit day generating priority of next day; Day generating priority, the overhaul management information of next day and the daily load prediction information of unit next day day part start and stop state that provides according to all generation schedules in current week, unit next day then, structure is also found the solution t+1 day generation schedule model, obtains under day generation schedule power plant for carrying out t+1 day;

After step 5, t+1 day power plant pressed t+1 day generation schedule and carry out, the control centre was at the real energy output of adding up each power plant t+2 day, carrying out t+2 day day-week feedback, the end of month of the current moon carry out the day-feed back by the moon.

2. as claimed in claim 1, it is characterized in that, monthly generation schedule model in the described step 2 is according to a plurality of regulation goals and constraints, structure multi-target non-linear MIXED INTEGER Optimization Model, utilize ideal point approach degree method that above-mentioned four targets are converted into simple target again, use non-linear mixed integer programming at last and find the solution the simple target Optimization Model; Described current monthly generation schedule is for determining current month unit each day start and stop state and Pinggu, peak period energy output;

3. as claimed in claim 1, it is characterized in that, all generation schedule models in the described step 3 are according to regulation goal and constraints, based on DC power flow structure single goal linear mixed-integer Optimization Model, be this Optimization Model decoupling zero unit combination submodel and economic dispatch submodel again, utilize mixed integer programming to find the solution unit combination submodel, utilize Non-Linear Programming to find the solution the economic dispatch submodel; Generation schedule comprised and determined that each unit at the start and stop state of next each period in week with exert oneself, provides day part electrical network start capacity and electrical network to exert oneself in week this.

4. as claimed in claim 1, it is characterized in that, day generation schedule model in the described step 4 adopts purchases electric cost minimization, two regulation goals of transmission line margin of safety value maximization, summation is as single optimization aim, adopt electric network active, idle, voltage quantities based on AC power flow structure single goal nonlinear optimization model, utilize alternating current-direct current mixing iterative algorithm to find the solution then and obtain; The day generating priority of described unit next day is the unit daily load rate, but promptly the unit determined of all generation schedules in current week next day energy output and next day unit maximum energy output ratio.

5. as claimed in claim 1, it is characterized in that, day-week in the described step 5 is fed back to: statistics from the beginning of the month to each unit power stretch amount or vacancy energy output till the day before yesterday, next all unit each day energy output of having determined with current month monthly generation schedule adds up, again add up all load factors or the Zhou Liyong hourage of unit, obtain the Zhou Fadian priority in next week of unit, forward step 3 to, be used for structure and find the solution next all generation schedules; Described day-the moon is fed back to: current month each unit power stretch amount of tabulate statistics or vacancy energy output, the electric weight plan of having signed with next month adds up, as unit next month answer the energy output plan, forward step 2 to.

Images