The content of the invention
The present invention provides a kind of hydro plant with reservoir non-flood period combined optimization power generation dispatching side for above-mentioned technical problem
Method.
The present invention provides a kind of hydro plant with reservoir non-flood period combined optimization power generation dispatching method, which is suitable for water
The non-flood period optimization power generation dispatching in storehouse power station, which refers to that Flood Season of Reservoir terminates to next flood season to arrive, and next flood
Reservoir reservoir level need to be reduced to flood season limit level before phase arrives, it is characterised in that include the following steps:
The first step:The period of the Runoff Forecast can be predicted with regimen condition to the reservoir incoming and reservoir of entire non-flood period
With selected as ten days;
Second step:According to power station arrangement form, the power station calculated under different reservoir reservoir levels, under different generating flows is optimal
Output storehouse, for a certain reservoir level, a certain generating flow, can according to flow-tailwater level curve of hydro plant with reservoir into
Row inquiry obtains the corresponding tailwater level of the generating flow, according to reservoir level, tailwater level and power station arrangement form, can must take office
The loss of flood peak of each unit under one unit assignment of traffic, so as to calculate the unit output under any assignment of traffic, to not cocurrent flow
Unit output under amount distribution carries out optimizing, calculates the optimal unit output under the reservoir level, the generating flow and optimal unit
Assignment of traffic is allocated all reservoir levels, all references flow, is calculated under different reservoir reservoir levels, different power generations
Power station optimum output storehouse under flow;
3rd step:Terminate first day in flood season, obtain the actual reservoir level of reservoir, the reservoir level after entire non-flood period finishing scheduling is
Flood season limit level, according to the actual reservoir level, the Dynamic Programming scheduling model of flood season limit level structure reservoir operation, the dynamic rule
The state variable for drawing scheduling model is the reservoir level of reservoir, and discrete, the Dynamic Programming is carried out to reservoir level according to scheduling requirement
The dispatching cycle of scheduling model be entire non-flood period, each scheduling slot be non-flood period each ten days, the Dynamic Programming scheduling model
Object function for power station optimal power generation amount, the water power obtained for any reservoir level, any letdown flow according to second step
Stand optimal output storehouse calculates power station and contributes, and input quantity is the prediction incoming situation of reservoir, and the constraints of foundation includes reservoir
Reservoir level constraint, the units limits in power station, the carrying capacity of pipe constraint in power station, reservoir are constrained with water, the Reservoir
For restriction of water level between dead water level and emerging sharp water level, the units limits in the power station are power station minimum load and maximum
Between output, the carrying capacity of pipe in power station is constrained to the minimum start flow in power station and completely between hair flow, the reservoir is used
Water constraint includes irrigation, urban water etc., and when processing is constrained with water, reservoir letdown flow should preferentially meet reservoir water will
It asks, hydropower station Optimized Operation is carried out on the basis of satisfaction is required with water;
4th step:The scheduling in the first ten days of non-flood period is performed according to the reservoir operation process obtained in the 3rd step, due to Dynamic Programming
Scheduling slot understand that the schedules traffic in the ten days is definite value, then is scheduled at first day of the first ten days according to the definite value, root
According to current reservoir level, schedules traffic, according to power station optimum output storehouse is obtained in second step, the flow point of different units is obtained
Match somebody with somebody, the power generation dispatching of first day is performed according to the assignment of traffic;
5th step:After first day the first ten days finishing scheduling, obtain the actual incoming situation of first day, and with the prediction in the first ten days
Incoming is compared, and the scheduling result in second day the first ten days is performed according to comparative result, and the comparative result is:If first day real
Border incoming is less than the prediction incoming in the first ten days, then generates electricity according to the Scheduling Flow magnitude in the ten days, if first day actual incoming is more than the
The prediction incoming in one ten days and when packing hair flow less than the corresponding water power of reservoir level after first day, according to the reality of first day
Border incoming generates electricity, if the corresponding water power of reservoir level packs hair flow after first day actual incoming is more than first day
When, then it packs hair flow according to the corresponding water power of reservoir level after described first day and generates electricity;Successively, when second day adjusts
After degree, the actual incoming of second day is obtained, according to above-mentioned comparative approach, compares the pre- of second day actual incoming and the first ten days
Incoming is surveyed, and schedules traffic and the scheduling scheme of the 3rd day are selected by comparative result, and so on, the tune until completing for the first ten days
Degree;
6th step after the first ten days finishing scheduling, obtains current reservoir level, according to the Dynamic Programming scheduling model weight in the 3rd step
New to establish scheduling model, the scheduling model and the initial water level of scheduling slot unlike the scheduling model in the 3rd are the
Reservoir reservoir level after one ten days finishing scheduling, scheduling end of term water level flood season limit level not yet, dispatching cycle is compared with the scheduling of the 3rd step
Cycle reduces a ten days, remaining constraints, object function are constant, performs Optimized Operation;
7th step:According to the Optimized Operation in the 6th step as a result, performing the operation plan in first day the second ten days of non-flood period, press afterwards
According in the 5th step dispatching method perform the second ten days by day dispatch, until second ten days finishing scheduling;
8th step:Second ten days finishing scheduling receive, perform non-flood period successively according to the dispatching method of the 6th step, the 7th step and own
In ten days and all ten days by day operation plan, until entire non-flood period terminates, reservoir reservoir level reaches flood season limit level.
Preferably, the power station optimum output in the second step under different reservoir reservoir levels, under different generating flows
During the calculating in storehouse, the difference reservoir reservoir levels are level of dead water between emerging sharp water level, and the discrete precision of water level is 0.5m, institute
It states different generating flows and starts shooting flow between completely sending out flow for minimum, the flow accuracy is 0.5m3/s.
Preferably, the Runoff Forecast can be predicted according to gray theory, wavelet theory etc., according to water during prediction
Storehouse actual ten days incoming situation prediction over the years.
Advantages of the present invention is as follows:
1st, calculated automatically using computer, realize the automatic realization that data storage calculates, liberated labour significantly, improved effect
Rate;
2nd, according to predicted flow rate, the science and reasonability of scheduling are added;
3rd, due to the error of prediction incoming, the present invention re-starts scheduling for the actual reservoir level after the scheduling of each ten days, adds
Deterministic reservoir level factor, adds the precision of scheduling, reduces error;
4th, for the operation plan in per ten days, according to the actual incoming situation of upper one day, the operation plan of next day of amendment, scheme is held
Row more tallies with the actual situation with more operability, scheduling.
Specific embodiment:Below in conjunction with specification specific embodiment, describe in detail to the present invention.
The present invention provides a kind of hydro plant with reservoir non-flood period combined optimization power generation dispatching method, which is suitable for water
The non-flood period optimization power generation dispatching in storehouse power station, which refers to that Flood Season of Reservoir terminates to next flood season to arrive, and next flood
Reservoir reservoir level need to be reduced to flood season limit level before phase arrives, it is characterised in that include the following steps:
The first step:The period of the Runoff Forecast can be predicted with regimen condition to the reservoir incoming and reservoir of entire non-flood period
With selected as ten days;
Second step:According to power station arrangement form, the power station calculated under different reservoir reservoir levels, under different generating flows is optimal
Output storehouse, for a certain reservoir level, a certain generating flow, can according to flow-tailwater level curve of hydro plant with reservoir into
Row inquiry obtains the corresponding tailwater level of the generating flow, according to reservoir level, tailwater level and power station arrangement form, can must take office
The loss of flood peak of each unit under one unit assignment of traffic, so as to calculate the unit output under any assignment of traffic, to not cocurrent flow
Unit output under amount distribution carries out optimizing, calculates the optimal unit output under the reservoir level, the generating flow and optimal unit
Assignment of traffic is allocated all reservoir levels, all references flow, is calculated under different reservoir reservoir levels, different power generations
Power station optimum output storehouse under flow;
3rd step:Terminate first day in flood season, obtain the actual reservoir level of reservoir, the reservoir level after entire non-flood period finishing scheduling is
Flood season limit level, according to the actual reservoir level, the Dynamic Programming scheduling model of flood season limit level structure reservoir operation, the dynamic rule
The state variable for drawing scheduling model is the reservoir level of reservoir, and discrete, the Dynamic Programming is carried out to reservoir level according to scheduling requirement
The dispatching cycle of scheduling model be entire non-flood period, each scheduling slot be non-flood period each ten days, the Dynamic Programming scheduling model
Object function for power station optimal power generation amount, the water power obtained for any reservoir level, any letdown flow according to second step
Stand optimal output storehouse calculates power station and contributes, and input quantity is the prediction incoming situation of reservoir, and the constraints of foundation includes reservoir
Reservoir level constraint, the units limits in power station, the carrying capacity of pipe constraint in power station, reservoir are constrained with water, the Reservoir
For restriction of water level between dead water level and emerging sharp water level, the units limits in the power station are power station minimum load and maximum
Between output, the carrying capacity of pipe in power station is constrained to the minimum start flow in power station and completely between hair flow, the reservoir is used
Water constraint includes irrigation, urban water etc., and when processing is constrained with water, reservoir letdown flow should preferentially meet reservoir water will
It asks, hydropower station Optimized Operation is carried out on the basis of satisfaction is required with water;
4th step:The scheduling in the first ten days of non-flood period is performed according to the reservoir operation process obtained in the 3rd step, due to Dynamic Programming
Scheduling slot understand that the schedules traffic in the ten days is definite value, then is scheduled at first day of the first ten days according to the definite value, root
According to current reservoir level, schedules traffic, according to power station optimum output storehouse is obtained in second step, the flow point of different units is obtained
Match somebody with somebody, the power generation dispatching of first day is performed according to the assignment of traffic;
5th step:After first day the first ten days finishing scheduling, obtain the actual incoming situation of first day, and with the prediction in the first ten days
Incoming is compared, and the scheduling result in second day the first ten days is performed according to comparative result, and the comparative result is:If first day real
Border incoming is less than the prediction incoming in the first ten days, then generates electricity according to the Scheduling Flow magnitude in the ten days, if first day actual incoming is more than the
The prediction incoming in one ten days and when packing hair flow less than the corresponding water power of reservoir level after first day, according to the reality of first day
Border incoming generates electricity, if the corresponding water power of reservoir level packs hair flow after first day actual incoming is more than first day
When, then it packs hair flow according to the corresponding water power of reservoir level after described first day and generates electricity;Successively, when second day adjusts
After degree, the actual incoming of second day is obtained, according to above-mentioned comparative approach, compares the pre- of second day actual incoming and the first ten days
Incoming is surveyed, and schedules traffic and the scheduling scheme of the 3rd day are selected by comparative result, and so on, the tune until completing for the first ten days
Degree;
6th step after the first ten days finishing scheduling, obtains current reservoir level, according to the Dynamic Programming scheduling model weight in the 3rd step
New to establish scheduling model, the scheduling model and the initial water level of scheduling slot unlike the scheduling model in the 3rd are the
Reservoir reservoir level after one ten days finishing scheduling, scheduling end of term water level flood season limit level not yet, dispatching cycle is compared with the scheduling of the 3rd step
Cycle reduces a ten days, remaining constraints, object function are constant, performs Optimized Operation;
7th step:According to the Optimized Operation in the 6th step as a result, performing the operation plan in first day the second ten days of non-flood period, press afterwards
According in the 5th step dispatching method perform the second ten days by day dispatch, until second ten days finishing scheduling;
8th step:Second ten days finishing scheduling receive, perform non-flood period successively according to the dispatching method of the 6th step, the 7th step and own
In ten days and all ten days by day operation plan, until entire non-flood period terminates, reservoir reservoir level reaches flood season limit level.
Preferably, the power station optimum output in the second step under different reservoir reservoir levels, under different generating flows
During the calculating in storehouse, the difference reservoir reservoir levels are level of dead water between emerging sharp water level, and the discrete precision of water level is 0.5m, institute
It states different generating flows and starts shooting flow between completely sending out flow for minimum, the flow accuracy is 0.5m3/s.
Preferably, the Runoff Forecast can be predicted according to gray theory, wavelet theory etc., according to water during prediction
Storehouse actual ten days incoming situation prediction over the years.
The principle of the present invention is described below:
The operability of Dynamic Programming:The problem of being applicable in Dynamic Programming must is fulfilled for principle of optimality and markov property.
1. principle of optimality(Optimal substructure):Principle of optimality can be illustrated so:One optimization strategy has
Such property, no matter past state and decision-making, for the state that the decision-making of front is formed, remaining all decision-makings must
Optimal policy must be formed.In brief, the substrategy of an optimization strategy is always optimal.One problem, which meets, optimizes original
Reason is also known as it with optimal substructure.
2. markov property by each stage according to certain sequential arrangement it is good after, for some given stage condition, it
The state in each stage can not directly affect its following decision-making in the past, and can only pass through this current state.In other words, often
A state, which is all that one of history in the past is complete, to be summarized.Here it is without rear tropism, also known as markov property.
It is analyzed for the combined dispatching of hydro plant with reservoir, when schedule periods are known segment, i.e., it dispatches initial time
Reservoir level it is known that scheduling the end of term water level it is known that i.e. the schedule periods be deterministic scheduling problem, carried out in entire schedule periods
Optimizing, meets above two applicable elements, and Dynamic Programming computational methods can be adapted for hydro plant with reservoir integrated distribution model
It solves.
Due to the uncertainty of incoming, cause with actual runoff not being inconsistent according to the scheduling model that set runoff is formulated, most
The uncertainty of excellent scheduling scheme nor actual optimum, i.e. runoff, causing the solution of scheduling model, there is also certain errors.On
It states error and is input parameter, is i.e. the error of runoff causes.Assuming that runoff is deterministic actual incoming, then above-mentioned scheduling scheme
Optimal scheduling scheme is should be, error analysis and improvement model foundation is carried out to it below:
Assuming that actual incoming is Qins, it is assumed that incoming Qinj, then carrying out stream error is:
FQ=Qins-Qinj
Error should be the vector of time series, the minimum period of runoff acquisition should be accurate to, to avoid the accumulation of above-mentioned error.
The gross generation E under scheduling scheme is then thought there is also corresponding error, and error is equal to:
FE=Es-Ej=f(FQ)
Above-mentioned error function should meet following relations under the scheduling model of this scheduling scheme:
f(0)=0
f(a)>f(b)
Wherein, a is runoff error amount 1, and b is runoff error amount 2, and a>b;Above-mentioned functional relation is represented by:I.e. when runoff misses
When difference is 0, above-mentioned scheduling model should be solved to optimal function solution;And error is bigger, corresponding dynamic programming model function is asked
It is bigger to solve error.
Since runoff has cumulative bad, i.e. cumulative errors can increase with the time, be solved always to be negative with Runoff Forecast
It releases and is described as follows:
Runoff obtains the period and is subject to the actual flow measurement of reservoir, generally with day calculation interval;And consider time factor and Runoff Forecast
Uncertainty, often calculated in actual reservoir operation with ten days or the moon, with day carry out Time segments division calculating, both greatly
Influence velocities solved, also due to the positive negative error of Runoff Forecast cause its it is actual have little significance, therefore, using ten days for schedule periods progress
During calculating, if without considering the influence of footpath stream error, deviation accumulation to ten days end is eliminated, i.e., is eliminated in per ten days end, meeting
Cause the accumulative of footpath stream error, so as to influence scheduling result, therefore, Ying Yi carries out error concealment for unit.
Consider the positive negative error of runoff, i.e., extra incoming causes footpath stream error for just, it is smaller come stream error be negative, general is actual
Incoming establishes the error digestion model in schedule periods compared with calculating incoming, the elimination and optimal case for error
Formulation plays an important roll;And terminate in a schedule periods, the eradicating efficacy of error is embodied on water level, i.e., when actually next in ten days
When stream is more than Runoff Forecast, water level rises;When actual incoming is less than Runoff Forecast in ten days, water level decreasing, on this basis,
Carry out computation model adjustment, introduce actual water level, that is, consider one ten days degree footpath stream error, above-mentioned footpath stream error will be next
It is eliminated in the scheduling of step.
The present invention program constructs two kinds of error concealment models, and carrying out daily dispatch scheduling according to actual incoming daily is missed
The first time of difference eliminates, and terminates in the calculating phase, says that footpath stream error is dissolved into water level, carry out follow-up error concealment.It is specific
Tonality model is as follows(Using ten days as schedule periods, revision of option is scheduled with day):
11st, runoff is predicted according to each ten days, establishes the reservoir operation of entire schedule periods and power station electricity generating plan, and in the first ten days first
It performs above-mentioned scheduling scheme;
12nd, fed back according to first day actual incoming, compared with ten days prediction runoff, perform the scheduling scheme amendment of second day;
13rd, fed back according to second day actual incoming, compared with ten days prediction runoff, perform the scheduling scheme amendment of the 3rd day;
…………
1N, basis are compared with ten days prediction runoff when the actual incoming in last two days of ten days, perform the scheduling of last day in this ten days
Revision of option;
21st, according to current reservoir level, n-1 ten days forecast dispatching phase(N is the total ten days number of schedule periods)Prediction incoming, formulate remaining scheduling
The reservoir operation of phase and power station electricity generating plan, and perform above-mentioned scheduling scheme within first day in the ten days;
…………
And so on, day amendment and ten days degree revision of option are realized, so as to fulfill the correction model of Runoff Forecast scheme.
Content described in this specification embodiment is only enumerating to the way of realization of the present invention, protection model of the invention
It encloses and is not construed as being only limitted to the concrete form that embodiment is stated, protection scope of the present invention also includes people in the art
Member according to the present invention design it is conceivable that equivalent technologies mean.