CN112633578B

CN112633578B - Cascade reservoir group optimal scheduling method under influence of diversion and water diversion engineering

Info

Publication number: CN112633578B
Application number: CN202011550856.8A
Authority: CN
Inventors: 李福威; 丁文昌; 雷晓辉; 张云辉; 王凤利; 相立锋; 王超; 廖卫红; 张志刚; 陆涛
Original assignee: Guodian Electric Power Development Co Ltd And Yu Hydropower Development Co ltd; China Institute of Water Resources and Hydropower Research
Current assignee: Guodian Electric Power Development Co Ltd And Yu Hydropower Development Co ltd; China Institute of Water Resources and Hydropower Research
Priority date: 2020-12-24
Filing date: 2020-12-24
Publication date: 2024-06-21
Anticipated expiration: 2040-12-24
Also published as: CN112633578A

Abstract

The invention discloses a cascade reservoir group optimal scheduling method under the influence of diversion and water diversion engineering, which comprises the following steps: step 1: establishing an objective function: the method comprises the steps of maximum step generating capacity, minimum water discarding amount, minimum comprehensive risk rate and minimum water diversion amount; step 2: consider the constraint: the method comprises reservoir water quantity balance constraint, flow balance constraint, drainage flow constraint, water storage level constraint and output constraint; step 3: solving the established cascade reservoir group water quantity joint optimization scheduling model by adopting a dynamic programming algorithm and a discrete differential dynamic programming algorithm; step 4: a scheduling graph-based generalized optimization model, a dynamic variable sampling space-based improved genetic algorithm and a non-dominant ranking genetic algorithm are selected as scheduling rules. The invention can improve the joint operation level of the cascade reservoir group and can be widely applied to the field of optimal dispatching of the cascade reservoir.

Description

Cascade reservoir group optimal scheduling method under influence of diversion and water diversion engineering

Technical Field

The invention relates to a cascade reservoir optimization theory, in particular to a cascade reservoir group optimization scheduling method under the influence of diversion and water diversion engineering.

Background

In a plurality of water mixing projects started in China, the original design standard of a downstream cascade reservoir in some projects is low, the design full-discharge flow rate of a downstream water power station is lower than the full-discharge flow rate of an upstream water power station, the flow rate of an upstream region and a downstream region is added, the downstream water power station must discard water when the upstream water power station is full-discharge, huge waste of water energy is caused, the comprehensive benefit exertion of the cascade power station in a river basin is limited, and the requirement of economic form development is hardly met. Therefore, how to master the time of full-load power generation of the reservoir and fully utilize the water energy resources is a difficult problem faced by the river basin cascade hydropower station.

When the cascade hydropower station is optimally scheduled, the upstream and downstream water diversion conditions of the reservoir are also considered, the original single-reservoir scheduling rule designed under the natural condition is not applicable any more, and the rule adapting to cascade reservoir group joint scheduling under the new condition should be optimized again.

Disclosure of Invention

The invention aims to overcome the defects of the background technology, and provides the cascade reservoir group optimizing and scheduling method under the influence of the diversion project, so that the cascade optimizing problem of the cascade hydropower station group of the river basin under the diversion condition of the diversion project of the river basin can be effectively solved, the multi-objective requirements of flood control, diversion, power generation, irrigation, cultivation, travel and the like of the river basin are coordinated, and the water resources of the river basin can be efficiently utilized.

The invention provides a cascade reservoir group optimal scheduling method under the influence of diversion and water diversion engineering, which comprises the following steps: step 1: establishing an objective function: the method comprises the steps of maximum step generating capacity, minimum water discarding amount, minimum comprehensive risk rate and minimum water diversion amount; step 2: consider the constraint: the method comprises reservoir water quantity balance constraint, flow balance constraint, drainage flow constraint, water storage level constraint and output constraint; step 3: solving the established cascade reservoir group water quantity joint optimization scheduling model by adopting a dynamic programming algorithm and a discrete differential dynamic programming algorithm; step 4: a scheduling graph-based generalized optimization model, a dynamic variable sampling space-based improved genetic algorithm and a non-dominant ranking genetic algorithm are selected as scheduling rules.

In the above technical solution, the specific calculation process of the objective function in step 1 is as follows: step 1.1: maximum step power generation amount: In the formula (1): e is the total power generation amount of the cascade reservoir in the whole dispatching period; t and T are respectively a scheduling period sequence number and a total number of the scheduling period in the scheduling period, and Deltat is a unit period for carrying out scheduling in the step reservoir; m and M are respectively the serial numbers of reservoirs and the total number of reservoirs; n (m, t) is the average output of the mth period of the mth reservoir; step 1.2: minimum water discard amount: /(I) In the formula (2): q _a is the total water disposal amount of the cascade reservoirs in the whole dispatching period, T and T are the sequence numbers of the dispatching period and the total number of the dispatching period in the dispatching period, M and M are the sequence numbers of the reservoirs and the total number of the reservoirs respectively, and Q _a (M and T) is the total water disposal amount of the mth period of the mth reservoir; step 1.3: minimum overall risk rate: /(I) In the formula (3): p is a comprehensive risk; b _n is an nth reservoir dispatching risk index, omega _n is an nth reservoir dispatching risk index weight, and the reservoir dispatching risk index consists of a water shortage rate, a water supply destruction depth, a cascade power generation insufficient risk rate and a power output insufficient risk rate; step 1.4: minimum water diversion amount: the minimum water diversion amount can be converted into the minimum comprehensive water shortage amount in the reservoir optimal scheduling process, and the water diversion amount is/is In the formula (4): q _s is the comprehensive water shortage of each department including the industrial living water shortage, the agricultural water shortage and the ecological water shortage in the dispatching period, T and T are the serial numbers of the dispatching period and the total number of the dispatching period in the dispatching period, M and M are the serial numbers of reservoirs and the total number of the reservoirs respectively, and Q _s (M and T) is the comprehensive water shortage of the mth period of the mth reservoir.

In the above technical solution, the specific calculation process of the constraint condition in the step 2 is as follows: step 2.1: water balance constraint of reservoir: the constraints of the model are specifically as follows: v (m, t+1) =v (m, t) +q _in(m,t)×Δt-Q_out (m, t) ×Δt (5) in formula (5): t is the number of a dispatching time period in a dispatching period, and Deltat is a unit time period for carrying out dispatching on the cascade reservoir; m is the number of each reservoir; v (m, t) and V (m, t+1) are respectively the initial reservoir capacity value of the mth reservoir in the t period and the end reservoir capacity value of the mth reservoir in the t period; Q _in (m, t) is the warehousing flow rate of the mth reservoir in the t-th period, and Q _out (m, t) is the ex-warehouse flow rate of the mth reservoir in the t-th period; Step 2.2: flow balance constraint: q _in(m+1,t)＝Q_out (m, t) +q (m, t) (6) in formula (6): q _in (m+1, t) is the warehousing flow rate of the (m+1) th reservoir in the t-th period, and Q _out (m, t) is the ex-warehouse flow rate of the (m) th reservoir in the t-th period; q (m, t) is the interval inflow of the m and m+1 reservoirs at the t-th period; step 2.3: lower leakage flow constraint: q _out,min(m,t)≤Q_out(m,t)≤Q_out,max (m, t) (7) in formula (7): q _out (m, t) is the discharge flow of the mth reservoir at the t-th period; q _out,min (m, t) is the lower limit constraint of the discharge flow of the mth reservoir in the t-th period, and Q _out,max (m, t) is the upper limit constraint of the discharge flow of the mth reservoir in the t-th period; Step 2.4: water storage level constraint: z _min(m,t)≤Z(m,t)≤Z_max (m, t) (8) in formula (8): the water level of the m-th reservoir in the t-th period of Z (m, t), Z _min (m, t) being the lower limit constraint of the water level of the m-th reservoir in the t-th period of Z _max (m, t) being the upper limit constraint of the water level of the m-th reservoir in the t-th period of Z; Wherein Z _min (m, t) is the dead water level of the mth reservoir in the t-th period, and Z _max (m, t) is determined by the downstream flood control task of the mth reservoir in the t-th period and the self-safety requirement of the reservoir; step 2.5: force constraint: n _min(m,t)≤N(m,t)≤N_max (m, t) (9) in formula (9): n (m, t) represents the outlet force of the mth reservoir in the t period, N _min (m, t) represents the lower outlet force allowable limit of the mth reservoir in the t period, N _max (m, t) represents the upper outlet force allowable limit of the mth reservoir in the t period, The allowable output upper and lower limit values are influenced by rated output, expected output and maximum output of the unit.

In the above technical solution, the specific calculation process of the dynamic programming algorithm 3.1 in the step 3 is as follows: step 3.1.1: solving a long-term power generation scheduling model of a reservoir, which comprises the following steps: step 3.1.1.1: dividing the phases and determining phase variables: for the reservoir with the regulation performance, dividing the dispatching period into T stages according to the period of month or ten days, and representing the stage variable by T, wherein t=1 to T, T is the facing period, and t+1 to T are the remaining periods; step 3.1.1.2: determining a state variable: selecting the water storage capacity or water level of a reservoir as a state variable, recording Vt-1 as the initial water storage capacity of the reservoir in the t period, and recording Vt as the final water storage capacity of the reservoir in the t period; step 3.1.1.3: determining decision variables: taking down the leakage flow Qt as a decision variable; step 3.1.1.4: determining a state transition equation: i.e., the water balance equation vt=vt—1+ (It-Qt) Δt; step 3.1.1.5: determining a stage index: the output Nt (Vt-1, qt) of the system in each stage represents the output of the t stage when the initial state of the period is Vt-1 and the decision variable of the period is Qt; step 3.1.1.6: determining an optimal value function: ft (Vt-1) represents the sum of the optimal forces from the T-th stage initial volume to Vt-1; the inverse time sequential recurrence equation for dynamic programming is then obtained as:

in the formula (10): /(I) Representing the water storage value when the initial discrete point of the t period is taken as m1,/>The water storage value at the end of the t period, namely when the initial discrete point of the t+1 period is taken as M2, and m1=0, 1, …, M, m2=0, 1, …, M is the water storage discrete point; qt represents the average leak-down flow amount in the t-th period; t=1, 2, …, T, representing a period number; /(I)Represents the initial water storage amount from the t-th period as/>Starting from the sum of the optimal outputs to the T-th period; /(I)Represents the initial water storage amount from the t+1st period as/>Starting from the sum of the optimal outputs to the T-th period; /(I)Indicating that the water storage capacity state at the beginning of the period is/>The output value when the average leakage flow is Qt; omega t is a decision variable and represents a decision set for the average drainage flow Qt to meet various constraints of the reservoir of the hydropower station when the water storage capacity Vt-1 is given; step 3.1.2: if It is known that the allowable minimum water storage capacity of a certain reservoir in each period of the dispatching period is Vt, min, and the allowable maximum water storage capacity is Vt, max, and the Z-up-V curve, the Z-down-Q curve, the initial reservoir capacity V0 and the runoff It in each period are known, the specific solving steps are as follows: step 3.1.2.1: dividing the adjusting period into T time periods, and dispersing the available reservoir capacity, namely Vt, min-Vt and max, in each time period into M water storage quantity state points; step 3.1.2.2: let t=t and obtain the water storage boundary values V0 and VT; calculating the average reservoir water level Z level of the time period according to the water storage value Vm1T-1 corresponding to the initial discrete point M1 of the time period and the time period end water storage value VT from the upper to the lower curves, wherein m1=1, 2, … and M; the average drainage flow QT in a period can be obtained by a water balance equation, and the output NT and the power generation reference flow QT in the period can be obtained by the iterative calculation of the output; step 3.1.2.3: applying the above recursive formula, since f×t+1 (VT) =0, the method comprises the steps of

Step 3.1.2.4: let t=t-1, m1=1, 2, …, M, repeat step 3.1.2.2 for a period of initial discrete point M1, m2=1, 2, …, M, and calculate the maximum remaining benefit corresponding to period of initial discrete point M1 from a recurrence equation, wherein the period of end water storage value at this time is Vtm2And the corresponding time period end optimal water storage point/>And storing the related information until t=1, and finishing the reverse timing recurrence calculation; step 3.1.2.5: from the initial water storage capacity V0 of the scheduling period and the corresponding end optimal water storage capacity V1 (V0) of the 1 st period, the time period optimal output N1 and the corresponding average drainage flow Q1 are calculated, the time period optimal output N1 is replaced to t=t, and the sequential recursive calculation is finished.

In the above technical solution, the specific calculation process of the discrete differential dynamic programming algorithm 3.2 in the step 3 is as follows: step 3.2.1: selecting an initial state sequence and a decision sequence: firstly, judging a near-optimal decision sequence according to general experience and analysis, and obtaining an initial state sequence corresponding to the initial decision sequence, wherein for the optimal dispatching problem of a reservoir, the initial state sequence is an initial dispatching line; step 3.2.2: selecting an increment to form a gallery: each of the initial state sequences is shifted by a small range, which is called delta, to form a strip-shaped 'corridor'; step 3.2.3: and (3) optimizing by using a conventional dynamic programming method in the gallery range: optimizing in a small signing band range by using a conventional dynamic programming recurrence method to obtain a new improved state sequence and a decision sequence, namely a new dispatching line which is closer to the optimal; step 3.2.4: iterating until convergence: a new iteration is carried out on the basis, namely a new increment delta is continuously changed near the obtained state, and the optimization is carried out again; repeating the loop in this way, and iterating successively until the optimal decision sequence and the optimal state sequence are approximated; when the iteration is stopped, the relative precision of the target is preset according to the precision requirement, and when the objective function value obtained by the two iterations meets the precision requirement, the iteration can be stopped.

In the above technical solution, in the step 3.2.3, the specific steps of optimizing by using a conventional dynamic programming recurrence method in a small range of signing a belt shape are as follows: 3.2.3.1: an adaptive gallery: in the early evolution process, selecting a wide corridor to quickly find an optimal solution; in the later evolution process, the gallery width is gradually reduced so as to improve the calculation accuracy; delta _i ^k＝α^kδ_i ^k-1 (12) in formula (12): delta _i ^k is the gallery width of the kth generation of the ith power station, alpha ^k is the self-adaptive coefficient, and the gallery width is dynamically changed along with the evolution algebra and the evolution result; 3.2.3.2: offset gallery technique: the eccentric corridor technology obtains the water level change trend of the corresponding hydropower station through the operation process of the first two generations of power stations, then generates an eccentric corridor with the upper and lower sides asymmetric according to the change trend, reduces the corridor width in the traversed space, enlarges the corridor width in the non-optimized space, ensures the effectiveness of the generated corridor,

In the formula (13): /(I)Upper and lower gallery boundaries and operating water levels respectively for the jth period of the ith power station in the kth generation,/>And a and b are unknown variables for the gallery width of the kth generation of the ith power station.

In the above technical solution, in the step 4, a specific process based on the generalization of the scheduling diagram is as follows: and (3) adopting a time and water level combined inflection point type mixed generalization mode, generalizing each line into inflection points represented by a group of boxes and line segments between two points, and optimizing a scheduling diagram by optimizing time or flow and water level variables of the inflection points.

In the above technical solution, in the step 4, the specific steps of the improved genetic algorithm based on the dynamic variable sampling space are as follows: the genetic algorithm is improved by combining the concept of dynamic variable space and the characteristics of the dispatching diagram optimization problem, the dispatching diagram is defined as a group of combinations of optimization variables in a model in a generalization mode of a mixed dispatching diagram, according to the two-dimensional spatial relation of the variables, t and z coordinates of inflection points of dispatching lines in the dispatching diagram can be considered to be recorded by the variables, n dispatching lines to be optimized are generated by combining the variables, the corresponding upper and lower relations between the dispatching lines must be met, and in the initial population generation and evolution process, the two-dimensional spatial relation between the dispatching lines of each dispatching diagram individual must be met to form a feasible solution, so that the information interaction mechanism between the dispatching diagram variables of the framework in the improved genetic algorithm is realized.

In the above technical solution, in the step 4, the specific steps based on the non-dominant ranking genetic algorithm are as follows: an elite strategy and a crowding degree calculation strategy are introduced into the non-dominant ranking genetic algorithm, and the elite strategy can keep the optimal solution obtained in the searching process; the congestion degree calculation strategy can improve the efficiency of the optimization algorithm, reduce the complexity of calculation and is simple to realize; there are various optimized scheduling modes, which are divided into: a long-term power generation schedule of about one year is formulated by taking the month or the ten days as a period; making a medium-term power generation schedule with a day as a period of time; setting short-term power generation schedule of 1 day to several days in the future by taking 15min,30min or 1h as a period; the real-time scheduling is to monitor the condition of the power stations for carrying out power grid load down in real time by taking 15 minutes, 30 minutes or 1 hour as a period of time, so as to carry out economical allocation in the plant and correct the plan.

In the above technical solution, in the non-dominant ranking genetic algorithm based on step 4, the time duration of several days in the short-term power generation schedule is within 5-7 days.

The cascade reservoir group optimal scheduling method under the influence of diversion and water diversion engineering has the following beneficial effects:

1) Comprehensively considering water diversion and dispatching projects at the upstream and downstream of a reservoir and a river basin cascade hydropower station, based on the warehouse-in flow forecast of different time scales, building a river basin cascade combined optimal dispatching mathematical model according to certain optimal dispatching criteria and constraint conditions, solving by using an optimization technology, searching for an optimal reservoir dispatching process and a water diversion dispatching scheme, and maximizing the total benefit of different targets such as power generation, flood control, irrigation, water supply and the like of each cascade in the whole analysis period.

2) Based on the incoming water forecasting models with different scales, the incoming water forecasting model is coupled with the reservoir dispatching model, meanwhile, the diversion and regulation water engineering is also used as a node of the model, the influence of diversion and regulation water quantity on the total water quantity of the river basin is considered, and the joint dispatching of the cascade hydropower station group under the diversion condition is realized.

3) In the short-term power generation plan making, namely week plan making and day plan making, in each power station, an optimization model is mainly optimized for a reservoir dispatching process, and a nonlinear programming method is adopted for solving. In the long-term optimizing and dispatching of the long-term cascade reservoir group, the method is mainly optimized for a reservoir dispatching diagram. The optimized objective function mainly comprises: the step generating capacity is maximum, the water discarding amount is minimum, the power generation guarantee rate is maximum, and the like. The multi-objective and dimension disaster problem faced in the step optimization is solved.

4) And supplementing diversion engineering in the reservoir group joint scheduling model and the optimized scheduling model, and adding the minimum diversion amount into the objective function and adding diversion amount constraint into the constraint condition. The cascade reservoir optimal scheduling model under the condition of taking the diversion into consideration is further perfected.

Drawings

FIG. 1 is a schematic diagram of the overall flow of a cascade reservoir group optimization scheduling method under the influence of a diversion and water transfer project;

FIG. 2 is a reverse time sequence recursive grid diagram of a dynamic planning algorithm in the cascade reservoir group optimal scheduling method under the influence of water diversion project;

FIG. 3 is a sequential recurrence and inverse timing recurrence calculation flow chart of a dynamic planning algorithm in a step reservoir group optimization scheduling method under the influence of diversion and water diversion engineering;

FIG. 4 is a diagram of the information interaction mechanism of an improved genetic algorithm based on a dynamic variable sampling space in the cascade reservoir group optimal scheduling method under the influence of diversion and water diversion engineering;

FIG. 5 is a combination chart of power generation dispatching modes based on a non-dominant ranking genetic algorithm in the cascade reservoir group optimization dispatching method under the influence of diversion and water diversion engineering.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, which should not be construed as limiting the invention.

The technical scheme adopted by the invention is as follows: the optimal dispatching method for the cascade reservoir group under the influence of the diversion and water transfer project, as shown in fig. 1, comprises the following steps:

Step 1: establishing an objective function: the method comprises the steps of maximum generated energy, minimum water discarding amount, minimum comprehensive risk rate and minimum water diversion amount, and the objective function is as follows:

Step 1.1: the step generating capacity is the largest:

In the formula (1): e is the total power generation amount of the cascade reservoir in the whole dispatching period; t and T are respectively a scheduling period sequence number and a total number of the scheduling period in the scheduling period, and Deltat is a unit period for carrying out scheduling in the step reservoir; m and M are respectively the serial numbers of reservoirs and the total number of reservoirs; n (m, t) is the average output of the mth period of the mth reservoir.

Step 1.2: the water discarding amount is the smallest:

In the formula (2): q _a is the total water disposal amount of the cascade reservoirs in the whole dispatching period, T and T are the sequence numbers of the dispatching period and the total number of the dispatching period in the dispatching period, M and M are the sequence numbers of the reservoirs and the total number of the reservoirs, and Q _a (M and T) is the total water disposal amount of the mth period of the mth reservoir.

Step 1.3: the comprehensive risk rate is the smallest:

In the formula (3): p is a comprehensive risk; b _n is the nth reservoir dispatching risk index, omega _n is the nth reservoir dispatching risk index weight, and the reservoir dispatching risk index consists of a water shortage rate, a water supply destruction depth, a cascade power generation quantity shortage risk rate, a power output shortage risk rate and the like.

Step 1.4: the water diversion amount is the smallest:

the minimum water diversion amount can be converted into the minimum comprehensive water shortage amount in the reservoir optimal scheduling process.

In the formula (4): q _s is the comprehensive water shortage of each department including the industrial living water shortage, the agricultural water shortage and the ecological water shortage in the dispatching period, T and T are the serial numbers of the dispatching period and the total number of the dispatching period in the dispatching period, M and M are the serial numbers of reservoirs and the total number of the reservoirs respectively, and Q _s (M and T) is the comprehensive water shortage of the mth period of the mth reservoir.

Step 2: consider the constraint: the method comprises reservoir water quantity balance constraint, flow balance constraint, drainage flow constraint, water storage level constraint and output constraint, wherein the specific constraint conditions are as follows:

Step 2.1: water balance constraint of reservoir:

the water balance of the reservoir is the most important constraint in a water resource system, and describes a water balance relation among inflow, water storage and water supply of the reservoir, and the constraint conditions of the model are specifically as follows:

V(m,t+1)＝V(m,t)+Q_in(m,t)×Δt-Q_out(m,t)×Δt (5)

In formula (5): t is the number of a dispatching time period in a dispatching period, and Deltat is a unit time period for carrying out dispatching on the cascade reservoir; m is the number of each reservoir; v (m, t) and V (m, t+1) are respectively the initial reservoir capacity value of the mth reservoir in the t period and the end reservoir capacity value of the mth reservoir in the t period; q _in (m, t) is the flow rate of the storage of the mth reservoir in the t-th period, and Q _out (m, t) is the flow rate of the delivery of the mth reservoir in the t-th period.

Step 2.2: flow balance constraint:

Q_in(m+1,t)＝Q_out(m,t)+q(m,t) (6)

In formula (6): q _in (m+1, t) is the warehousing flow rate of the (m+1) th reservoir in the t-th period, and Q _out (m, t) is the ex-warehouse flow rate of the (m) th reservoir in the t-th period; q (m, t) is the interval inflow of the m and m+1 reservoirs at the t-th period.

Step 2.3: lower leakage flow constraint:

Q_out,min(m,t)≤Q_out(m,t)≤Q_out,max(m,t) (7)

In the formula (7): q _out (m, t) is the discharge flow of the mth reservoir at the t-th period; q _out,min (m, t) is the lower limit constraint of the discharge flow rate of the mth reservoir at the t-th period, and Q _out,max (m, t) is the upper limit constraint of the discharge flow rate of the mth reservoir at the t-th period.

Step 2.4: water storage level constraint:

Z_min(m,t)≤Z(m,t)≤Z_max(m,t) (8)

In formula (8): the water level of the m-th reservoir in the t-th period of Z (m, t), Z _min (m, t) being the lower limit constraint of the water level of the m-th reservoir in the t-th period of Z _max (m, t) being the upper limit constraint of the water level of the m-th reservoir in the t-th period of Z; wherein Z _min (m, t) is the dead water level of the m reservoir in the t period, and Z _max (m, t) is determined by the downstream flood control task of the m reservoir in the t period and the self-safety requirement of the reservoir.

Step 2.5: force constraint:

N_min(m,t)≤N(m,t)≤N_max(m,t) (9)

In the formula (9): n (m, t) represents the output of the mth reservoir in the t period, N _min (m, t) represents the lower allowable output limit of the mth reservoir in the t period, N _max (m, t) represents the upper allowable output limit of the mth reservoir in the t period, wherein the upper and lower allowable output limits are influenced by the rated output, expected output and maximum output of the unit.

The invention comprehensively considers the water diversion project at the upstream and downstream of the reservoir and the river basin cascade hydropower station, based on the warehouse-in flow forecast of different time scales, builds a river basin cascade joint optimization scheduling mathematical model according to certain optimization scheduling criteria and constraint conditions, solves by using the optimization technology, searches for the optimal reservoir scheduling process and water diversion scheduling scheme, and ensures that the total benefit of different targets such as power generation, flood control, irrigation, water supply and the like of each cascade is the largest in the whole analysis period.

The invention is based on the incoming water forecasting model with different scales, the incoming water forecasting model is coupled with the reservoir dispatching model, meanwhile, the diversion and regulation project is also used as a node of the model, the influence of diversion and regulation quantity on the total water quantity of the river basin is considered, and the joint dispatching of the cascade hydropower station group under the diversion condition is realized.

In the invention, in the short-term power generation plan making, namely the week plan making and the day plan making in each power station, an optimization model is mainly optimized for the reservoir dispatching process. The long-term regular reservoir dispatching is mainly carried out according to the dispatching diagram, so that the long-term optimization dispatching of the long-term cascade reservoir group is mainly carried out according to the dispatching diagram. The constraint conditions include: reservoir water balance, reservoir water storage, reservoir water level, reservoir drainage and the like, and the optimization objective function mainly comprises: the step generating capacity is maximum, the water discarding amount is minimum, the power generation guarantee rate is maximum, and the like. The key point is to solve the multi-objective and dimension disaster problem faced in the step optimization.

Based on the reservoir group joint scheduling model and the optimized scheduling model, water diversion engineering is supplemented, the minimum water diversion amount is added in the objective function, and water diversion amount constraint is added in the constraint condition. The cascade reservoir optimal scheduling model under the condition of diversion is further perfected and considered.

Step 3: the invention adopts Dynamic Programming (DP) and Discrete Differential Dynamic Programming (DDDP) to solve and establish a cascade reservoir group water quantity joint optimization scheduling model:

3.1: dynamic programming algorithm (DP):

In a plurality of solving methods of the reservoir optimal scheduling model, dynamic planning is widely applied by effectively processing the problem of the stage property and the nonlinearity. However, when solving the related problem with dynamic programming, the following three conditions need to be satisfied:

① In a multi-stage decision process, the evolution characteristics of a stage can be described by the change of a stage variable, and the effect of state transition or change depends on the change of the stage decision variable, and the condition that the initial state of the next stage is the final state of the previous stage is satisfied.

② No aftereffect is satisfied, i.e. the past state is independent of future decisions and only related to the state currently in existence.

③ The piecewise optimal decision is subject to the overall process optimal decision.

Step 3.1.1: the reservoir long-term power generation scheduling model is solved by utilizing a dynamic programming method, and the method comprises the following steps:

Step 3.1.1.1: dividing the phases and determining phase variables: for reservoirs with regulation performance, the scheduling period (typically one year) is divided into T stages by month or ten-day period, which is regarded as a multi-stage decision problem, and T represents a stage variable (t=1 to T). T is the facing period and t+1 to T are the remaining periods.

Step 3.1.1.2: determining a state variable: the water storage capacity (or water level) of the reservoir is generally selected as a state variable, vt-1 is the water storage capacity of the reservoir at the beginning of the t period, and Vt is the water storage capacity of the reservoir at the end of the t period. Because the final water storage capacity of each period is the initial water storage capacity of the next period, the principle of no aftereffect is satisfied.

Step 3.1.1.3: determining decision variables: the bleed flow Qt is typically taken as a decision variable.

Step 3.1.1.4: determining a state transition equation: i.e. the water balance equation vt=vt-1+ (It-Qt) Δt.

Step 3.1.1.5: determining a stage index: i.e., the output Nt (Vt-1, qt) of the system at each stage, which represents the output of the t-th stage when the initial state of the period is Vt-1 and the decision variable of the period is Qt.

Step 3.1.1.6: determining an optimal value function: i.e., ft (Vt-1), represents the sum of the optimal output (power generation) from the T-th stage initial capacity to the T-th stage starting from Vt-1.

The inverse time sequential recurrence equation for dynamic programming is then obtained as:

In the formula (10): representing the water storage value when the initial discrete point of the t period is taken as m1,/> Representing the water storage value when the discrete point at the end of the t period (t+1 period is at the beginning) is taken as M2, wherein m1=0, 1, …, M, m2=0, 1, …, M is the number of the water storage discrete points; qt represents the average leak-down flow amount in the t-th period; t=1, 2, …, T, representing a period number; /(I)Represents the initial water storage amount from the t-th period as/>Starting from the sum of the optimal output (power generation) to the T-th period. /(I)Represents the initial water storage amount from the t+1st period as/>Starting from the sum of the optimal output (power generation) to the T-th period. /(I)Indicating that the water storage capacity state at the beginning of the period is/>The output value when the average leak flow rate is Qt. Omega _t is a decision variable representing a decision set in which the average bleed-down flow Qt meets the constraints of the hydroelectric reservoir when the reservoir capacity Vt-1 is given.

When solving the model by using the above formula, the water storage amount needs to be firstly discretized. For reservoirs with regulation, the reservoir level is generally varied between flood control limit and dead water levels, while the flood control period is generally between normal water level and dead water level, so the available water storage capacity can be divided into M-1 grids with a step DeltaV, for a total of M points, as shown in FIG. 2.

After the model is built, a specific solution is adopted, and the solution of the dynamic programming model mainly comprises two steps (an inverse push method):

① And (5) performing reverse time sequence recursive calculation: according to an inverse time sequence recursion equation, starting from the last stage (stage T), recursing to the first stage (stage 1) from time intervals forward, and solving an inverse time sequence process with the maximum sum of the generated energy of the power station in each time interval in the whole scheduling period under the condition of meeting related constraint conditions, namely solving an optimal value sigma Nt;

② Sequential recursive calculation: and solving an optimal strategy { Qt } corresponding to the optimal value sigma Nt and corresponding each optimal state point value { Vi }.

Step 3.1.2: if It is known that the allowable minimum water storage capacity of a certain reservoir in each period of the dispatching period is Vt, min, and the allowable maximum water storage capacity is Vt, max, and the Z-up-V curve, the Z-down-Q curve, the initial reservoir capacity V0 and the runoff It in each period are known, the specific solving steps are as follows:

Step 3.1.2.1: dividing the regulating period into T time periods, and dispersing the available water storage capacity (Vt, min-Vt, max) in each time period into M water storage capacity state points;

Step 3.1.2.2: let t=t and obtain the water storage boundary values V0 and VT. Calculating the average reservoir water level Z level of the time period according to the water storage value Vm1T-1 corresponding to the initial discrete point M1 (m1=1, 2, …, M) of the time period and the last water storage value VT of the time period by checking the curves from Z to V; the average drainage flow QT in a period can be obtained by a water balance equation, and the output NT and the power generation reference flow QT in the period can be obtained by the iterative calculation of the output;

Step 3.1.2.3: applying the above recursive formula, since f×t+1 (VT) =0, the method comprises the steps of

Step 3.1.2.4: let t=t-1, repeat step 2.2 for the initial discrete point M1 (m1=1, 2, …, M) at the end of the period of time with water storage value Vtm2 (m2=1, 2, …, M), and calculate the maximum remaining benefit corresponding to the initial discrete point M1 by the recurrence equationAnd the corresponding time period end optimal water storage point/>And storing the related information until t=1, and finishing the reverse timing recurrence calculation;

step 3.1.2.5: from the initial water storage capacity V0 of the scheduling period and the corresponding end optimal water storage capacity V1 (V0) of the 1 st period, the time period optimal output N1 and the corresponding average drainage flow Q1 are calculated, the time period optimal output N1 is replaced to t=t, and the sequential recursive calculation is finished.

The above recursive calculation process (reverse time sequence+forward time sequence) can be represented by fig. 3, where m1 in fig. 3 represents the period initial storage capacity discrete point index value, and m2 represents the period end storage capacity discrete point index value.

The dynamic programming method is a global searching method, which converts the original problem into a series of sub-problems with similar structures and relatively simple, and then carries out combination traversal optimizing on all the sub-problems.

3.2: Discrete differential dynamic programming algorithm:

To solve the "dimension disaster" problem of conventional dynamic planning, larson et al in 1968 proposed a discrete differential dynamic planning algorithm (DDDP). The basic principle of the algorithm is that an initial state sequence is determined by experience or a conventional scheduling method; determining the increment of each stage according to the constraint condition of the limitation, and changing the initial solution up and down by a small range according to the corresponding increment value to form a solution gallery; and then optimizing by using conventional dynamic programming solution in the gallery. Meanwhile, a state encryption strategy is added in the actual application, namely, in the iterative process, the increment can be gradually reduced from large to small; the increment can be selected only at one side of the initial solution according to the actual situation; the number and size of the increments at each stage and on both the upper and lower sides may also be different. The method can solve the problem of one dimension or even multiple dimensions. It should be further noted that the DDDP result is not necessarily the optimal solution, and typically requires trial calculation by selecting different initial solutions and decision sequences.

DDDP is a cyclic DP, which starts from a feasible initial scheduling line, optimizes by using DP in the neighborhood of the initial scheduling line, searches an optimal track in the field, and then takes the optimal track as the initial track, and repeats the operation until reaching convergence accuracy. Because DDDP optimizes in the given dispatching line field, the discrete point number is less in each iterative calculation, thereby greatly reducing the required storage space and calculation time consumption, and being suitable for the joint optimization dispatching problem of large-scale reservoir groups. The main disadvantage of DDDP is that it requires an initial decision trajectory, on the one hand, it is difficult to give a reasonable initial trajectory, and on the other hand, the optimal solution obtained from the initial trajectory cannot be guaranteed to be a global optimal solution. The common solution is to set a plurality of initial test tracks for calculation and to take the optimal solution.

The discrete differential dynamic programming method comprises the following steps:

Step 3.2.1: selecting an initial state sequence and a decision sequence:

Firstly, a decision sequence which is as close to optimal as possible is determined according to common experience and analysis or by other simple methods, and an initial state sequence corresponding to the initial decision sequence can be obtained. For the optimal dispatching problem of the reservoir, the initial state sequence is the initial dispatching line.

Step 3.2.2: selecting an increment to form a gallery:

Each of the sequences of initial states is shifted by a small range, called delta (denoted delta), forming a ribbon-like "corridor".

Step 3.2.3: and (3) optimizing by using a conventional dynamic programming method in the gallery range:

And optimizing in the small band-shaped range by using a conventional dynamic programming recurrence method, and obtaining a new improved state sequence and a decision sequence, namely a new dispatching line which is closer to the optimal. Since DDDP is preferred over the gallery range of the initial solution, it is also referred to as "gallery method".

Because the step reservoir optimal scheduling solving steps are complex and tedious, the calculation time is often not guaranteed when the library group is large in scale. Considering that the calculated amount increases exponentially with the increase of the discrete accuracy, the accuracy is too low to meet the corresponding requirement. Therefore, the research works adopt a DDDP algorithm to optimize the gradient water quantity scheduling process of the target power station, and adopt self-adaptive gallery and partial gallery technologies at the same time, so that the problem solving speed is increased at the same time under the condition of ensuring the calculation accuracy.

3.2.3.1: An adaptive gallery:

In the optimizing process, the calculation amount is too large if the corridor is too wide, and the evolution speed is too slow if the corridor is too narrow, so that the adaptive corridor which changes along with the evolution result and the evolution algebra is adopted. In the early evolution process, selecting a wide corridor to quickly find an optimal solution; in the later evolution process, the gallery width is gradually reduced so as to improve the calculation accuracy, thereby realizing win-win of the calculation amount and the calculation accuracy.

Wherein,For the gallery width of the kth generation of the ith power station, alpha ^k is an adaptive coefficient, and the gallery width is dynamically changed along with the evolution algebra and the evolution result.

3.2.3.2: Offset gallery technique:

the common gallery technology adopts symmetrical galleries with equal widths up and down on two sides of the previous water level process, then performs optimization in the corresponding galleries to obtain a new optimization process without exceeding the boundary of the galleries, and when the next gallery is determined, the symmetrical width galleries are taken on two sides of the new optimization process, so that one gallery overlaps with the previous gallery, and the repeated optimization in the area is invalid, thereby directly affecting the optimization speed of the subsystem.

In order to improve the optimization efficiency of the subsystem, an asymmetric gallery biasing technology is provided. According to the method, through the operation process of the first two generations of power stations, the water level change trend of the corresponding hydropower station is obtained, then asymmetric inclined galleries on the upper side and the lower side are generated according to the change trend, the width of the galleries in the traversed space is reduced, the width of the galleries in the non-optimized space is enlarged, the effectiveness of the generated galleries is ensured, and therefore the optimizing efficiency of the subsystem is improved.

In the formula (13), the amino acid sequence of the compound,Upper and lower gallery boundaries and operating water levels respectively for the jth period of the ith power station in the kth generation,/>And a and b are unknown variables for the gallery width of the kth generation of the ith power station.

Step 3.2.4: iterating until convergence:

A new iteration is performed on the basis of the above, i.e. a new increment delta is continued around the obtained state and is performed again. The method is repeatedly circulated and iterated successively until the optimal decision sequence and the optimal state sequence are approximated. To determine when to stop the iteration, the relative accuracy of the targets should be predefined according to accuracy requirements. And stopping iteration when the objective function value obtained by the two iterations meets the precision requirement.

Step 4: selecting a scheduling graph generalized optimization model, a dynamic variable sampling space-based improved genetic algorithm and a non-dominant sorting genetic algorithm as scheduling rules:

4.1: generalizing a scheduling diagram:

And an inflection point type mixing generalization mode of time and water level combination is adopted. Each line is generalized to a set of inflection points represented by a box and a line segment between the two points, and the scheduling graph is optimized by optimizing time and water level (or flow and water level) variables of the inflection points. The generalization mode can greatly reduce the number of decision variables and reduce the optimization scale, but simultaneously the dimension of the decision variables is increased, and a one-dimensional real variable group of a pure water level is converted into a two-dimensional shaping and real mixed variable group of a time and water level combined turning point type, so that the feasible region of the optimization problem becomes narrower, and the searching difficulty of an optimal solution is improved. The characteristics of few variables and direct application of the optimization result without correction are the main advantages of the mixed generalization mode, and lay a foundation for improving the efficiency of the optimization model and searching the global optimal solution. Therefore, the optimization efficiency and effect of the scheduling diagram can be improved as long as the problem of difficulty in searching the optimal solution is solved.

4.2: Improved genetic algorithm based on dynamic variable sampling space:

The dynamic variable sampling space (Dynamic Variablies SAMPLING SPACE, DVSS for short) is a general method for processing constraint optimization problems constructed on the basis of the GA algorithm concept. The DVSS finds a decision variable ordering mode by analyzing the relation among decision variables, so that the sampling space of the current variable can be corrected according to the information of the sampled variable in the sampling process, and the dynamic change of the sampling space of the variable is realized, thereby ensuring that the solutions generated in the optimization process are all feasible solutions or are as close to a feasible domain as possible. On the other hand, DVSS introduces biological concepts of bases, genes and genomes, classifies and groups the variables according to the relation among decision variables, and builds information interaction platforms among different levels and groups of variables, thereby building a dynamic correction effective mechanism of a decision variable sampling space.

The genetic algorithm is improved by combining the concept of dynamic variable space and the characteristics of the dispatching diagram optimization problem. Taking a power generation schedule diagram as an example, a generalized mode of a hybrid schedule diagram is used, the schedule diagram is defined as a group of time variable and water level variable combinations in a model, and according to the two-dimensional spatial relationship of each variable, each variable can be considered to record t and z coordinates of inflection points of each schedule line in the schedule diagram. Therefore, when the variables are combined to generate n scheduling lines to be optimized, the corresponding upper and lower relationships among the scheduling lines are required to be met, and the two-dimensional space relationship among the scheduling lines of each scheduling diagram individual in the initial population generation and evolution process is required to be met, so that a feasible solution can be realized. Based on the analysis, the information interaction mechanism among the framework dispatching diagram variables in the improved genetic algorithm can effectively control the sampling space of the variables in a ray searching mode as shown in fig. 3, and the solution in the initial population generation and evolution process is ensured to be a feasible solution.

The optimization of flood control scheduling adopts the same mode to construct a variable information interaction mechanism, except that the optimization variables are flow and water level.

4.3: Based on non-dominant ranking genetic algorithm (NSGA-II)

The GA algorithm is the basis of the optimization model of the dispatching diagram, and the GA algorithm is used as a mature heuristic algorithm and comprises various different improvement or evolution algorithms. In the study, NSGA-II is adopted as a basic algorithm for optimizing a scheduling model. The NSGA-II algorithm was developed from the Non-dominant ranking genetic algorithm NSGA (Non-dominated Sorting Genetic Algorithms). Compared with NSGA algorithm, the NSGA-II algorithm introduces elite strategy and congestion degree calculation strategy. The elite strategy can keep the optimal solution obtained in the searching process; the congestion degree calculation strategy can improve the efficiency of the optimization algorithm, reduce the calculation complexity and is simpler to realize. Overall, the NSGA-II algorithm is a very mature and efficient multi-objective optimization algorithm.

The optimal scheduling modes are various and are divided into mainly: a long-term power generation schedule of about one year is formulated by taking the month or the ten days as a period; making a medium-term power generation schedule with a day as a period of time; setting short-term power generation schedule of 1 day to several days (more than 5 days) in the future by taking 15min,30min or 1h as a period; the real-time scheduling is to monitor the condition of the power stations for carrying out power grid load down in real time by taking 15 minutes, 30 minutes or 1 hour as a period of time, so as to carry out economical allocation in the plant and correct the plan. The power generation scheduling pattern classification is as shown in fig. 4.

The invention relates to a principle and a method for optimizing a dispatching function of a reservoir group multi-level guarantee rate, which take Huijiang river basin multi-level reservoir joint dispatching as a research case, construct a segmented dispatching function based on the guarantee rate, and prove that the invention is scientific, reasonable and feasible through the simulation operation of the function.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

What is not described in detail in this specification is prior art known to those skilled in the art.

Claims

1. The utility model provides a diversion and water transfer engineering influences step reservoir crowd optimal scheduling method which characterized in that: the method comprises the following steps:

step 1: establishing an objective function: the method comprises the steps of maximum step generating capacity, minimum water discarding amount, minimum comprehensive risk rate and minimum water diversion amount;

step 1.3: minimum overall risk rate:

In the formula (3): p is a comprehensive risk; b _n is an nth reservoir dispatching risk index, omega _n is an nth reservoir dispatching risk index weight, and the reservoir dispatching risk index consists of a water shortage rate, a water supply destruction depth, a cascade power generation insufficient risk rate and a power output insufficient risk rate;

Step 2: consider the constraint: the method comprises reservoir water quantity balance constraint, flow balance constraint, drainage flow constraint, water storage level constraint and output constraint;

Step 3: solving the established cascade reservoir group water quantity joint optimization scheduling model by adopting a dynamic programming algorithm and a discrete differential dynamic programming algorithm;

the specific calculation process of the dynamic programming algorithm 3.1 in the step 3 is as follows:

step 3.1.1: solving a long-term power generation scheduling model of a reservoir, which comprises the following steps:

step 3.1.1.1: dividing the phases and determining phase variables: for the reservoir with the regulation performance, dividing the dispatching period into T stages according to the period of month or ten days, and representing the stage variable by T, wherein t=1 to T, T is the facing period, and t+1 to T are the remaining periods;

Step 3.1.1.2: determining a state variable: selecting the water storage capacity or water level of a reservoir as a state variable, recording Vt-1 as the initial water storage capacity of the reservoir in the t period, and recording Vt as the final water storage capacity of the reservoir in the t period;

Step 3.1.1.3: determining decision variables: taking down the leakage flow Qt as a decision variable;

step 3.1.1.4: determining a state transition equation: i.e., the water balance equation vt=vt—1+ (It-Qt) Δt;

Step 3.1.1.5: determining a stage index: the output Nt (Vt-1, qt) of the system in each stage represents the output of the t stage when the initial state of the period is Vt-1 and the decision variable of the period is Qt;

step 3.1.1.6: determining an optimal value function: ft (Vt-1) represents the sum of the optimal forces from the T-th stage initial volume to Vt-1;

In the formula (10): The water storage value when the initial discrete point of the t period is taken as M1, V _t ^m2 represents the water storage value when the initial discrete point of the t period, namely t+1 period, is taken as M2, and m1=0, 1, …, M, m2=0, 1, …, M and M are the water storage discrete points; qt represents the average leak-down flow amount in the t-th period; t=1, 2, …, T, representing a period number; /(I) Represents the initial water storage amount from the t-th period as/>Starting from the sum of the optimal outputs to the T-th period; f+1 (V _t ^m2) represents the sum of the optimal forces from the time period t+1 when the initial water storage is V _t ^m2 to the time period T; /(I)Indicating that the water storage capacity state at the beginning of the period is/>The output value when the average leakage flow is Qt; omega t is a decision variable and represents a decision set for the average drainage flow Qt to meet various constraints of the reservoir of the hydropower station when the water storage capacity Vt-1 is given;

step 3.1.2.1: dividing the adjusting period into T time periods, and dispersing the available reservoir capacity, namely Vt, min-Vt and max, in each time period into M water storage quantity state points;

Step 3.1.2.2: let t=t and obtain the water storage boundary values V0 and VT; calculating the average reservoir water level Z level of the time period according to the water storage value Vm1T-1 corresponding to the initial discrete point M1 of the time period and the time period end water storage value VT from the upper to the lower curves, wherein m1=1, 2, … and M; the average drainage flow QT in a period can be obtained by a water balance equation, and the output NT and the power generation reference flow QT in the period can be obtained by the iterative calculation of the output;

Step 3.1.2.3: the reverse-timing recurrence equation is applied, because f×t+1 (VT) =0

Step 3.1.2.4: let t=t-1, m1=1, 2, …, M, repeat step 3.1.2.2 for a period of initial discrete point M1, m2=1, 2, …, M, and calculate the maximum remaining benefit corresponding to period of initial discrete point M1 from a recurrence equation, wherein the period of end water storage value at this time is Vtm2And the corresponding time period end optimal water storage point/>And storing the related information until t=1, and finishing the reverse timing recurrence calculation;

Step 3.1.2.5: from the initial water storage capacity V0 of the scheduling period and the corresponding optimal water storage capacity V1 (V0) at the end of the 1 st period, the optimal output power N1 of the period and the corresponding average drainage flow Q1 of the period can be obtained, the time is replaced to t=t, and sequential recursive calculation is finished;

Step 4: selecting a scheduling graph-based generalized optimization model, a dynamic variable sampling space-based improved genetic algorithm and a non-dominant sorting genetic algorithm as scheduling rules;

In the step 4, the specific steps of the improved genetic algorithm based on the dynamic variable sampling space are as follows:

The genetic algorithm is improved by combining the concept of dynamic variable space and the characteristics of the dispatching diagram optimization problem, the dispatching diagram is defined as a group of combinations of optimization variables in a model in a generalization mode of a mixed dispatching diagram, according to the two-dimensional spatial relation of the variables, t and z coordinates of inflection points of dispatching lines in the dispatching diagram can be considered to be recorded by the variables, n dispatching lines to be optimized are generated by combining the variables, the corresponding upper and lower relations between the dispatching lines must be met, and in the initial population generation and evolution process, the two-dimensional spatial relation between the dispatching lines of each dispatching diagram individual must be met to form a feasible solution, so that the information interaction mechanism between the dispatching diagram variables of the framework in the improved genetic algorithm is realized.

2. The method for optimizing and scheduling the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 1, which is characterized in that: the specific calculation process of the objective function in the step 1 is as follows:

Step 1.1: maximum step power generation amount:

In the formula (1): e is the total power generation amount of the cascade reservoir in the whole dispatching period; t and T are respectively a scheduling period sequence number and a total number of scheduling periods in a scheduling period, and delta T is a unit period for carrying out scheduling in the step reservoir; m and M are respectively the serial numbers of reservoirs and the total number of reservoirs; n (m, t) is the average output of the mth period of the mth reservoir;

step 1.2: minimum water discard amount:

In the formula (2): q _a is the total water disposal amount of the cascade reservoirs in the whole dispatching period, T and T are the sequence numbers of the dispatching period and the total number of the dispatching period in the dispatching period, M and M are the sequence numbers of the reservoirs and the total number of the reservoirs respectively, and Q _a (M and T) is the total water disposal amount of the mth period of the mth reservoir;

Step 1.4: minimum water diversion amount:

The minimum water diversion amount can be converted into the minimum comprehensive water shortage amount in the reservoir optimal scheduling process,

3. The method for optimizing and scheduling the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 2, which is characterized in that: the specific calculation process of the constraint condition in the step 2 is as follows:

Step 2.1: water balance constraint of reservoir:

The constraints of the model are specifically as follows:

V(m,t+1)＝V(m,t)+Q_in(m,t)×Δt-Q_out(m,t)×Δt (5)

In formula (5): t is the number of a dispatching time period in a dispatching period, and delta t is a unit time period for carrying out dispatching on the cascade reservoir; m is the number of each reservoir; v (m, t) and V (m, t+1) are respectively the initial reservoir capacity value of the mth reservoir in the t period and the end reservoir capacity value of the mth reservoir in the t period; q _in (m, t) is the warehousing flow rate of the mth reservoir in the t-th period, and Q _out (m, t) is the ex-warehouse flow rate of the mth reservoir in the t-th period;

Step 2.2: flow balance constraint:

Q_in(m+1,t)＝Q_out(m,t)+q(m,t) (6)

In formula (6): q _in (m+1, t) is the warehousing flow rate of the (m+1) th reservoir in the t-th period, and Q _out (m, t) is the ex-warehouse flow rate of the (m) th reservoir in the t-th period; q (m, t) is the interval inflow of the m and m+1 reservoirs at the t-th period;

step 2.3: lower leakage flow constraint:

Q_out,min(m,t)≤Q_out(m,t)≤Q_out,max(m,t) (7)

In the formula (7): q _out (m, t) is the discharge flow of the mth reservoir at the t-th period; q _out,min (m, t) is the lower limit constraint of the discharge flow of the mth reservoir in the t-th period, and Q _out,max (m, t) is the upper limit constraint of the discharge flow of the mth reservoir in the t-th period;

Step 2.4: water storage level constraint:

Z_min(m,t)≤Z(m,t)≤Z_max(m,t) (8)

In formula (8): the water level of the m-th reservoir in the t-th period of Z (m, t), Z _min (m, t) being the lower limit constraint of the water level of the m-th reservoir in the t-th period of Z _max (m, t) being the upper limit constraint of the water level of the m-th reservoir in the t-th period of Z; wherein Z _min (m, t) is the dead water level of the mth reservoir in the t-th period, and Z _max (m, t) is determined by the downstream flood control task of the mth reservoir in the t-th period and the self-safety requirement of the reservoir;

Step 2.5: force constraint:

N_min(m,t)≤N(m,t)≤N_max(m,t) (9)

4. The method for optimizing and scheduling the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 1, which is characterized in that: the specific calculation process of the discrete differential dynamic programming algorithm 3.2 in the step 3 is as follows:

Step 3.2.1: selecting an initial state sequence and a decision sequence:

firstly, judging a near-optimal decision sequence according to general experience and analysis, and obtaining an initial state sequence corresponding to the decision sequence, wherein for the optimal dispatching problem of a reservoir, the initial state sequence is an initial dispatching line;

step 3.2.2: selecting an increment to form a gallery:

Each of the initial state sequences is shifted by a small range, which is called delta, to form a strip-shaped 'corridor';

Optimizing in a small signing band range by using a conventional dynamic programming recurrence method to obtain a new improved state sequence and a decision sequence, namely a new dispatching line which is closer to the optimal;

step 3.2.4: iterating until convergence:

A new iteration is carried out on the basis of the steps 3.2.2 and 3.2.3, namely a new increment delta is continuously changed near the obtained state, and the optimization is carried out again; repeating the loop in this way, and iterating successively until the optimal decision sequence and the optimal state sequence are approximated; when the iteration is stopped, the relative precision of the target is preset according to the precision requirement, and when the objective function value obtained by the two iterations meets the precision requirement, the iteration can be stopped.

5. The optimal scheduling method for the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 4, which is characterized by comprising the following steps: in the step 3.2.3, the specific steps of optimizing by using a conventional dynamic programming recurrence method in a small range of signing the belt shape are as follows:

3.2.3.1: an adaptive gallery:

In the early evolution process, selecting a wide corridor to quickly find an optimal solution; in the later evolution process, the gallery width is gradually reduced so as to improve the calculation accuracy;

in the formula (12): the gallery width of the kth generation of the ith power station is the self-adaptive coefficient, and alpha k is dynamically changed along with the evolution algebra and the evolution result;

3.2.3.2: offset gallery technique:

The eccentric corridor technology obtains the water level change trend of the corresponding hydropower station through the operation process of the first two generations of power stations, then generates an eccentric corridor with the upper and lower sides asymmetric according to the change trend, reduces the corridor width in the traversed space, enlarges the corridor width in the non-optimized space, ensures the effectiveness of the generated corridor,

In the formula (13): The upper and lower gallery boundaries and the running water level of the jth period of the ith power station in the kth generation are respectively, And a and b are unknown variables for the gallery width of the kth generation of the ith power station.

6. The method for optimizing and scheduling the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 5, which is characterized in that: in the step 4, the specific process based on the generalization of the scheduling diagram is as follows:

And (3) adopting a time and water level combined inflection point type mixed generalization mode, generalizing each line into inflection points represented by a group of boxes and line segments between two points, and optimizing a scheduling diagram by optimizing time or flow and water level variables of the inflection points.

7. The method for optimizing and scheduling the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 1, which is characterized in that: in the step 4, the specific steps based on the non-dominant ranking genetic algorithm are as follows:

an elite strategy and a crowding degree calculation strategy are introduced into the non-dominant ranking genetic algorithm, and the elite strategy can keep the optimal solution obtained in the searching process; the congestion degree calculation strategy can improve the efficiency of the optimization algorithm, reduce the complexity of calculation and is simple to realize;

There are various optimized scheduling modes, which are divided into: a long-term power generation schedule of about one year is formulated by taking the month or the ten days as a period; making a medium-term power generation schedule with a day as a period of time; setting short-term power generation schedule of 1 day to several days in the future by taking 15min,30min or 1h as a period; the real-time scheduling is to monitor the condition of the power stations for carrying out power grid load down in real time by taking 15 minutes, 30 minutes or 1 hour as a period of time, so as to carry out economical allocation in the plant and correct the plan.

8. The optimal scheduling method for the cascade reservoir group under the influence of diversion and water diversion engineering according to claim 7, which is characterized by comprising the following steps: in the non-dominant ranking genetic algorithm based on step 4, the time duration of several days in the short-term power generation schedule is within 5-7 days.