WO2022226952A1 - 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 - Google Patents
耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 Download PDFInfo
- Publication number
- WO2022226952A1 WO2022226952A1 PCT/CN2021/091245 CN2021091245W WO2022226952A1 WO 2022226952 A1 WO2022226952 A1 WO 2022226952A1 CN 2021091245 W CN2021091245 W CN 2021091245W WO 2022226952 A1 WO2022226952 A1 WO 2022226952A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- hydropower station
- water level
- period
- feasible
- hydropower
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Definitions
- the invention belongs to the field of hydropower generation scheduling, and in particular relates to a dimension reduction method for large-scale hydropower station group optimization scheduling by coupling feasible region identification and random sampling.
- Feasible region identification can effectively reduce the size of the decision-making search range without changing the optimization mechanism of the optimization algorithm. Using constraints to determine the feasible region can redefine the decision-making range. On the other hand, how to avoid invalid calculation is also an important dimensionality reduction idea. Random sampling can not only reduce the decision combination of calculation, but also reduce the state combination. Both of the above-mentioned ideas can effectively reduce the calculation and storage amount of the optimal dispatching of the hydropower system.
- the present invention takes the hydropower system in the southwest region as the background, relies on the National Natural Science Foundation of China (52079014), and conducts research on the dimensional disaster problem of large-scale hydropower system dispatching.
- the technical problem to be solved by the present invention is to provide a dimensionality reduction method for large-scale hydropower station group optimization and dispatch by coupling feasible region identification and random sampling.
- the feasible region is used to reduce the search range of the results, and the random sampling optimization method is adopted to ensure the reliability of the results. It avoids traversing all solutions in the search range, which greatly improves the computational efficiency and solves the problem of dimensionality disaster in the optimization and scheduling of large-scale hydropower station groups to a certain extent.
- the present invention discloses a method for optimizing dispatching dimension reduction of large-scale hydropower station group by coupling feasible region identification and random sampling, and completing the deterministic optimization dispatching dimension reduction of hydropower station group according to the following steps 1-12:
- Step 1 Based on the efficient utilization of water energy, establish the optimal dispatching model of the hydropower station group, as follows:
- E represents the power generation of the hydropower system
- ⁇ t' represents the number of hours in the time period t
- i represents the serial number of the hydropower station
- K i represents the output coefficient of the hydropower station i
- Ni ,t represents the output of the hydropower station in the time period t
- q i,t represents the discharge flow of the hydropower station in the time period t
- hi represents the average output net head of the hydropower station in the time period t
- q i,t represents the power generation flow of the hydropower station i in the time period t
- Ni ,t represents the hydropower station
- I represents the number of hydropower stations.
- Step 2 Initialize calculation parameters, including hydropower station operation parameters, water level control constraints, discharge flow constraints, power station output constraints, and water balance constraints;
- V i,t+1 V i,t +3600(Q i,t -R i,t )* ⁇ t'(5)
- Z i,0 represents the initial water level of hydropower station i
- Z i,T represents the final water level of hydropower station i.
- Z i,S and Z i,Z represent the minimum and maximum water level of hydropower station i respectively
- Z i,t represent the water level of hydropower station i in time period t
- Ri ,min and R i,max are the outgoing water level of hydropower station i respectively
- R i,t represents the discharge flow of hydropower station i in time period t
- q i,min and q i,max are the upper and lower limits of power generation flow of hydropower station i respectively
- q i,t represents the power generation of hydropower station i in time period t flow
- d i,t is the abandoned water flow of hydropower station i in time period t
- Ni ,min and Ni ,max are the lower limit and upper limit of output of hydropower station i respectively
- Ni ,t represents the
- Step 3 Set the reliability indicators a, b, and the accuracy ⁇ 1 and ⁇ 5 of the optimal solution and the fifth optimal solution of the sampling method;
- Step 4 Determine the parameter sequence ⁇ a 1 , a 2 , a 3 . . . , a M ⁇ , and the sampling number sequence ⁇ S 1 , S 2 , S 3 . ,a 3 ...,a M ⁇ and the number of elements of ⁇ S 1 ,S 2 ,S 3 ...,S M ⁇ ;
- Step 6 Determine the feasible domain range of the hydropower station at the end of the period t, couple the above-mentioned various hydropower station scheduling and operation constraints, and determine the feasible domain range, including the following steps:
- q i,t represent the upper limit and lower limit of the leakage flow, respectively;
- q i,pmax , q i,pmin respectively represent the equivalent leakage flow corresponding to the upper limit and lower limit of the output.
- the outflow runoff is inversely proportional to the storage capacity at the end of the period, that is, the minimum discharge flow can be used to obtain the upper limit of the storage capacity at the end of the period, and the maximum discharge flow can be used to obtain the lower limit of the storage capacity at the end of the period. value.
- the discharge flow constraint is first transformed into a storage capacity constraint. After the transformation, it is compared with the original storage capacity constraint, and the intersection is taken as the upper and lower limits of the feasible range.
- Z i,t respectively represent the upper and lower limits of the feasible area of reservoir water level; H i,normal , H i,dead represent the normal high water level and dead water level of hydropower station i respectively; A represents various runoff conditions of the i-th hydropower station during time period t Array of water levels below.
- Step 7 Carry out probability sampling within the feasible water level range of the hydropower station to calculate the output during the facing period: in the calculation, the probability that the water level of each reservoir is in the satisfactory solution interval is a%. Then every time a solution is randomly drawn, the probability that it is not in the satisfactory solution interval is (1-a%), and s candidate water levels are continuously drawn in the feasible region determined in the previous step, and the probability that these water levels are not in the top a% is ( 1-a%) s when:
- Step 11 Obtain the optimal solution sequence ⁇ f1 ⁇ and the fifth optimal solution ⁇ f5 ⁇ for hydropower station scheduling, and obtain the optimal solution and the fifth optimal solution sequence for each sampling combination ⁇ f 1,1 , f 1,2 , f 1 ,3 ,... ⁇ , ⁇ f 5,1 , f 5,2 , f 5,3 ,... ⁇ , to solve the change rate d1 and d5 of the sequence, the specific formula is as follows:
- the method proposed by the present invention can use the feasible region of a large-scale hydropower system dispatch model to reduce the scope, and adopts the random sampling optimization method to avoid the problem under the premise of ensuring the reliability of the results. Traversing all the solutions within the search range greatly improves the computational efficiency and solves the problem of dimensionality disaster in the optimal scheduling of hydropower stations to a certain extent.
- the invention can significantly reduce the scope of inferior decision-making, reduce unnecessary calculation and storage costs, solve the problem of dimensionality disaster to a certain extent, and realize efficient processing of large-scale hydropower system optimization scheduling.
- Figure 1 is a schematic diagram of the upper and lower limits of the water level of the hydropower station
- FIG. 2 is a schematic diagram of a feasible domain identification method
- FIG. 3 is a schematic diagram of the distribution probability of the optimal scheduling result.
- the present invention relates to a dimension reduction method for optimal dispatching of large-scale hydropower station groups by coupling feasible region identification and random sampling.
- the present invention will be further described below with reference to the accompanying drawings and examples.
- the present invention mainly solves two problems: firstly, how to determine the feasible solution range of the dispatching model and reduce it according to the dispatching requirements; secondly It is to randomly sample within the range of feasible solutions to avoid a large number of invalid or inefficient calculations.
- the solutions to the two problems are described below.
- the present invention aims at efficient utilization of water energy in a super-large-scale hydropower system under the condition of deterministic incoming water, and constructs a maximum power generation model, the details are as follows:
- V i,t+1 V i,t +3600(Q i,t -R i,t )* ⁇ t'(22)
- Z i,0 represents the initial water level of hydropower station i
- Z i,T represents the final water level of hydropower station i.
- Z i,S and Z i,Z represent the minimum and maximum water level of hydropower station i respectively
- Z i,t represent the water level of hydropower station i in time period t
- Ri ,min and R i,max are the outgoing water level of hydropower station i respectively
- R i,t represents the discharge flow of hydropower station i in time period t
- q i,min and q i,max are the upper and lower limits of power generation flow of hydropower station i respectively
- q i,t represents the power generation of hydropower station i in time period t flow
- d i,t is the abandoned water flow of hydropower station i in time period t
- Ni ,min and Ni ,max are the lower limit and upper limit of output of hydropower station i respectively
- Ni ,t represents the
- a solution that satisfies all constraints of the mathematical model is called a feasible solution.
- the set composed of all feasible solutions is called the feasible region, that is, the feasible region is only related to the constraints of the problem, and has nothing to do with the objective function of the problem and the solution method. Therefore, the feasible region for the maximum power generation model constructed by the present invention is a set of solutions that satisfy all the above constraints.
- the outflow runoff is inversely proportional to the storage capacity at the end of the period, that is, the minimum discharge flow can be used to obtain the upper limit of the storage capacity at the end of the period, and the maximum discharge flow can be used to obtain the lower limit of the storage capacity at the end of the period. value.
- the discharge flow constraint is first transformed into a storage capacity constraint. After the transformation, it is compared with the original storage capacity constraint, and the intersection is taken as the upper and lower limits of the feasible range.
- the water level at the end of the period is used as the decision variable, so the obtained water level storage capacity range is transformed into the water level range through the water level-storage capacity relationship.
- the artificial simulated runoff conforms to the theoretical frequency curve of monthly runoff. It can be considered that the given runoff process includes all possible runoff processes.
- single-reservoir DP scheduling is performed on the reservoir to obtain the water level change process of the reservoir for many years, and the upper and lower limits of the water level are obtained.
- multi-year water level adjustment process it can be seen that under the condition that the initial water level and the final water level are fixed throughout the year, no matter how the runoff conditions change, a part of the water level will never be reached. As shown in Figure 1: the water level between the upper water level and the normal high water level, the water level between the lower water level and the dead water level will not be encountered during the scheduling process, so these ranges can be excluded before the scheduling calculation to further reduce the feasible area scope.
- a three-dimensional matrix A is formed by assembling the month-end water level processes of all hydropower stations.
- the three dimensions are: runoff, month, and hydropower, and then upper and lower water levels are set for each reservoir in each period according to A.
- Z i,t respectively represent the upper and lower limits of the feasible area of reservoir water level; H i,normal , H i,dead represent the normal high water level and dead water level of hydropower station i respectively; A represents various runoff conditions of the i-th hydropower station during time period t Array of water levels below.
- the feasible areas are determined in order according to the size of the period from small to large.
- period 1 the feasible region at the end of the period is obtained according to the initial water level combined with the constraints of the first period and the amount of water inflow.
- this method samples water levels from feasible areas to obtain initial conditions for the next epoch.
- the determination of the feasible region and the sampling can be carried out in sequence, and the scope of the feasible region can be dynamically determined.
- it can better delete the water level range that does not meet the constraints, and the feasible region range is greatly reduced. Therefore, it can be seen that by sequentially calculating the feasible area of each period in each period in this way, the computational efficiency of the HDRS can be greatly improved.
- the principle of the FRI method is shown in Figure 2.
- the invention uses RS optimization on the basis of random sampling and increases the result evaluation, and dynamically adjusts the sampling number according to the search results.
- acceptability and reliability parameters are introduced into RS optimization.
- Selected water level states are iteratively evaluated to evaluate the improvement of the target value, which will be used as a convergence criterion throughout the search process.
- the water level subsection sampling weight is given, and the random sampling is evolved into a probability sampling.
- n_N represents the number of output N
- n_all represents the number of all output results
- a(N) represents the probability of the output of N.
- A(N1) represents the probability of occurrence of an outcome less than N1.
- Fig. 3 it can be seen that the number of results is positively correlated with the quality of the results, and the closer to the optimal solution, the more results are included in the interval.
- the probability of each decision appearing is the same, and the probability of obtaining the optimal solution is too small, but the probability of obtaining an approximate optimal solution in the vicinity of the optimal solution is significantly improved.
- the parameter a is set, and it is considered that when the solutions are sorted according to the pros and cons, the solutions in the first a% interval are all satisfactory solutions.
- the samples in each interval have a certain probability to be drawn. When completely random, the probability of each sample being drawn is:
- n satisfactory solutions represent the number of solutions in the satisfactory solution interval.
- the probability of random sampling is only related to the number of samples. When different samples are given different probabilities through objective reasons, when sampling is based on the sample probability, random sampling changes into probability sampling. When the number of sampling times is sufficient, the distribution of random sampling results is relatively even. , which approximates the overall result distribution.
- the distribution of results of probability sampling is related to the probability value of the results. When the results with better performance have a greater probability, the distribution of sampling results shows that the number of better results increases and the number of poor results decreases. Therefore, the use of probability sampling can improve the reliability of sampling when assigning a larger probability value to the sample with better results.
- the goal is to obtain results in the satisfactory solution interval, and each solution has a probability of a% in the satisfactory solution interval. Then every time a solution is randomly selected, the probability that it is not in the satisfactory solution interval is (1-a%), and s candidate solutions are continuously extracted, and the probability that they are not in the top a% is (1-a%) s , due to sampling
- the goal is to extract a solution that lies in the satisfactory solution interval, then when:
- Step 1 Set parameters. Including parameters a, b, ⁇ 1 and ⁇ 5.
- Step 2 Determine the parameter sequence ⁇ a 1 , a 2 , a 3 ..., a M ⁇ and ⁇ S 1 , S 2 , S 3 ..., S M ⁇ ;
- Step 3 Sampling according to the sequence ⁇ S 1 , S 2 , S 3 . m usually takes 20-50;
- Step 4 Solve the change rates d1 and d5 of the solution sequence, and judge the accuracy. If the accuracy judgment is satisfied, the optimal solution will be output. If not, then go to Step 5.
- Step 5 Increase the number of samples, add the solution to the existing solution set, and return to Step 4.
- Step 1 Establish a deterministic optimal dispatch model for a super-large-scale hydropower system, with the goal of maximizing power generation;
- Step 2 Read basic data and initialize calculation parameters, including hydropower station operation parameters, water level control constraints, discharge flow constraints, power station output constraints, and water balance constraints;
- Step 3 Set the parameters a, b, m, l, ⁇ 1 and ⁇ 5, the water level matrix A, and the weight distribution w of each interval;
- Step 4 Determine the parameter sequence ⁇ a 1 , a 2 , a 3 ..., a M ⁇ and ⁇ S 1 , S 2 , S 3 ..., S M ⁇ ;
- Step 6 Calculate the time period t according to the feasible region identification principle, and determine the feasible region range at the end of the time period;
- Step 7 Carry out probability sampling within the feasible domain, and extract a final water level to calculate the output of this period;
- Step 11 Obtain the optimal solution sequence ⁇ f1 ⁇ and the fifth optimal solution ⁇ f5 ⁇ , and solve the change rate d1 and d5 of the sequence;
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Feedback Control In General (AREA)
Abstract
一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法,以发电量最大为目标建立确定性来水情况下超大规模水电系统调度模型,利用可行域识别方法对水位、流量、出力各种限制约束进行等效转换处理,获取模型可行解范围,并对可行解范围进行缩减;采用随机抽样算法求解模型,引入可接受性和可靠性参数,依据多年的调度结果赋予水位分段抽样权重,将随机抽样进化为概率抽样,通过迭代优化计算得到超大规模水电系统最优水位与出力过程。本发明能充分削减较劣决策范围、大幅减少不必要水库运行水位状态计算,与离散微分动态规划和逐步优化方法相比,在解质量相近的条件下,分别减少计算时间98.73%和96.86%,显著缓解了水电优化调度的维数灾难题。
Description
本发明属于水力发电调度领域,特别涉及一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法。
近二十年来,我国水电实现了跨越式的发展,截止2020年底全国水电总装机达到3.7亿kW,单一区域电网水电超过1亿kW、单一省级电网超过7500万kW,单一流域梯级达到2000万kW,区域和省级电网统调水电站数量超过100甚至200座,如此庞大的水电系统,传统以动态规划、解析式规划为主的优化调度方法的计算和存储规模需求成指数级增长,“维数灾”问题尤为突出,是我国系统高比重水电电网、特大流域梯级水电站群发电调度面临的首要难题,亟需高效实用的求解方法。
随着参与计算的水电站数目增加,单时段的状态组合与决策组合都呈指数增长,导致系统的计算量大幅增长,所以减少单时段的状态组合与决策组合能够有助于解决维数灾问题。离散微分动态规划与逐步优化算法相对于经典动态规划,都减少了每个阶段的状态组合,实际应用中也证明了这些方法能有效地提升计算效率。然而,它们依旧存在计算量和存储量随水电站数目增加而呈指数增长的维数问题,所以针对超大规模水电站群优化调度问题亟需探索新的求解方法与技术。一方面,减小搜索范围对于降维是重要的,可行域识别能在不改变优化算法寻优机制的前提下有效地减少决策搜索范围的大小,利用约束条件来确定可行域能重新定义决策范围的边界,减少计算的决策组合;另一方面,如何规避无效计算也是重要的降维思路,随机抽样不仅能减少计算的决策组合,也能减少状态组合。前述两种思路都可以有效减少水电系统优化调度的计算和存储量。
为此,本发明以西南地区水电系统为背景,依托国家自然科学基金(52079014),针对大规模水电系统调度的维数灾问题开展研究,利用可行域识别和随机抽样技术,提出了具有较强实用价值和推广应用价值的大规模水电站群优化调度降维方法。
发明内容
本发明要解决的技术问题是提供一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法,利用可行域减少结果搜索范围,采用随机抽样优化方法在保证结果可靠度的前提下避免遍历搜索范围内的所有解,极大地提高了计算效率,在一定程度上解决了大规模水电站群优化调度中的维数灾问题。
本发明的技术方案为:本发明揭示了一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法,按照下述步骤1-12完成水电站群确定性优化调度降维:
步骤1.以水能高效利用为准则,建立水电站群优化调度模型,具体如下:
式中:该式中E代表水电系统的发电量;Δt’代表时段t内的小时数;i代表水电站序号;K
i代表水电站i的出力系数;N
i,t,代表水电站在时段t的出力;q
i,t代表水电站在时段t的下泄流量;h
i,t代表水电站在时段t的平均出力净水头;q
i,t代表水电站i在时段t的发电流量;N
i,t代表水电站i在时段t的出力;I代表水电站的个数。
步骤2.初始化计算参数,包括水电站运行参数、水位控制约束、下泄流量约束、电站出力约束、水量平衡约束;
N
i,min≤N
i,t≤N
i,max (4)
V
i,t+1=V
i,t+3600(Q
i,t-R
i,t)*Δt′ (5)
式中:Z
i,0代表水电站i的初始水位,Z
i,T代表水电站i的末水位。Z
i,S与Z
i,Z分别代表水电站i的水位最小值与最大值;Z
i,t代表水电站i在时段t的水位;R
i,min与R
i,max分别为水电站i的出库流量下限与上限;R
i,t代表水电站i在时段t的下泄流量;q
i,min与q
i,max分别为水电站i的发电流量上下限;q
i,t代表水电站i在时段t的发电流量;d
i,t为水电站i在时段t的弃水流量;N
i,min与N
i,max分别为水电站i的出力下限与上限;N
i,t代表水电站i在时段t的出力;V
i,t为水电站i在时段t的初始库容;Q
i,t为水电站i在时段t的入库径流;Q
i,t'为水电站i在时段t的区间流量;∑R
t代表上游水电站在时段t的下泄流量之和;当水电 站i=1时,入库径流为区间径流,当水电站i>1时,入库径流为区间径流与上游水电站下泄流量之和。
步骤3.设置可靠度指标a、b,以及抽样方法最优解和第五优解的精度ε1与ε5;
步骤4.确定参数序列{a
1,a
2,a
3…,a
M},与抽样数目序列{S
1,S
2,S
3…,S
M};其中M是数组{a
1,a
2,a
3…,a
M}和{S
1,S
2,S
3…,S
M}的元素个数;
步骤5.设置n=1,k=1,t=1,取S=S
k;其中n是抽样样本的序号,k是数组{a
1,a
2,a
3…,a
M}和{S
1,S
2,S
3…,S
M}序列里的序号,S是抽样的数目,S
k代表{S
1,S
2,S
3…,S
M}序列里第k个元素;
步骤6.确定水电站在时段t末的可行域范围,耦合上述多种水电站调度运行约束,确定可行域范围,包含以下几步:
(1)约束条件转换:
对于出力约束,利用耗水率μ
i将其转化为下泄流量约束,如下所示:
对于下泄流量约束,使用q
i,min与q
i,max寻求可行域,比较两组下泄流量约束,并将它们的交集作为定义可行区域的一组下泄流量约束如下:
(3)可行域范围确定
据水量平衡方程可知,在其他变量确定的前提下,出库径流与时段末库容呈反比,即可以使用最小下泄流量求取时段末库容上限值,使用最大下泄流量求取时段末库容下限值。如下式所示,利用水量平衡,在已知的入库流量与时段初库容的条件下,将下泄流量约束先转变为库容约束。在转变后与原有的库容约束进行比较,取交集作为可行范围的上下限。
根据上式得到的库容范围,通过水位~库容关系曲线将其转变为等效的水位范围。
(4)采用下式进一步缩减可行域范围
步骤7.在水电站的水位可行域范围内进行概率抽样,以计算面临时段的出力:在计算中,每个库水位位于满意解区间的概率为a%。则每随机抽出一个解,它不处于满意解区间的概率为(1-a%),在上一步骤确定的可行域内连续抽出s个候选水位,这些水位都不是处于前a%的概率为(1-a%)
s,当:
(1-a%)
s≤b% (12)
则抽出S个库水位,为了保证含有满意解,则:
当抽出的样本数目大于S
min时,则有(1-b%)的概率抽取到满意的水位。
步骤8.令t=t+1;判断t=T是否成立,若成立,则取全年末水位为时段末水位进行出力计算,并求出各时段出力之和为样本n-s的抽样结果,设置t=1,进入步骤9;若不成立,则返回步骤6继续计算;
步骤9.判断n=S是否成立,若成立,求取S个库水位样本中的最优解与第五优解,设置n=1,进入步骤10,若不成立,n=n+1,返回步骤6继续计算;
步骤10.判断k=M是否成立,若成立,进入步骤11,若不成立,k=k+1,返回步骤6继续计算;
步骤11.获得水电站调度最优解序列{f1}与第五优解{f5},获得每个抽样组合的最优解与第五优解序列{f
1,1,f
1,2,f
1,3,…},{f
5,1,f
5,2,f
5,3,…},求解序列的变化率d1与d5,具体公式如下:
d1<ε1 (16)
d5<ε5 (17)
步骤12.进行精度判断,如果满足精度判断则输出最优解,若不满足则计算增加抽样数目,设置S=ΔS,k=M+1,返回步骤6继续计算。
对比现有技术,实施本发明的技术方案,可实现如下有益效果:本发明提出的方法可利用大规模水电系统调度模型可行域缩减范围,采用随机抽样优化方法在保证结果可靠度的前提下避免遍历搜索范围内的所有解,极大地提高了计算效率,在一定程度上解决了水电站群优化调度中的维数灾问题。本发明能够显著缩减较劣决策范围、减少不必要计算和存储成本,在一定程度上解决维数灾问题,实现大规模水电系统优化调度的高效处理。
图1是水电站水位上下限示意图;
图2是可行域识别方法示意图;
图3是优化调度结果分布概率示意图。
本发明涉及一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法,以下结合附图和实例对本发明作进一步的描述。
超大规模的水电站群调度是极具挑战的系统优化问题,为了实现系统的高效求解,本发明主要解决两个方面的问题:一是如何根据调度需求确定调度模型的可行解范围并进行缩减;二是在可行解范围内进行随机抽样,以规避大量无效或低效计算。以下分别阐述两方面问题的解决方案。
(a)水电站群优化调度模型可行解范围缩减
本发明以确定性来水情况下超大规模水电系统水能高效利用为目标,构建发电量最大模型,具体如下:
模型约束如下:
①水位控制约束
②下泄流量约束
②出力约束
N
i,min≤N
i,t≤N
i,max (21)
③水量平衡约束
V
i,t+1=V
i,t+3600(Q
i,t-R
i,t)*Δt′ (22)
式中:Z
i,0代表水电站i的初始水位,Z
i,T代表水电站i的末水位。Z
i,S与Z
i,Z分别代表水电站i的水位最小值与最大值;Z
i,t代表水电站i在时段t的水位;R
i,min与R
i,max分别为水电站i的出库流量下限与上限;R
i,t代表水电站i在时段t的下泄流量;q
i,min与q
i,max分别为水电站i的发电流量上下限;q
i,t代表水电站i在时段t的发电流量;d
i,t为水电站i在时段t的弃水流量;N
i,min与N
i,max分别为水电站i的出力下限与上限;N
i,t代表水电站i在时段t的出力;V
i,t为水电站i在时段t的初始库容;Q
i,t为水电站i在时段t的入库径流;Q
i,t'为水电站i在时段t的区间流量;∑R
t代表上游水电站在时段t的下泄流量之和;当水电站i=1时,入库径流为区间径流,当水电站i>1时,入库径流为区间径流与上游水电站下泄流量之和。
满足数学模型所有约束的解决方案称为可行解决方案。由所有可行解组成的集合称为可行区域,即可行域只与问题的约束有关,与问题的目标函数以及求解方式等都无直接关系。因此,对于本发明所构建的发电量最大模型的可行区域即为满足上述所有约束的一组解决方案。
(1)模型约束转换
在进行模型可行解范围缩减之前,需要先找出各约束所对应的决策范围,即对约束进行等价转换。对于出力约束,利用耗水率μ
i转化为总排水约束,如下所示:
对于下泄流量约束,已知R>q,所以使用q
i,min与q
i,max寻求可行域,比较两组下泄流量约束,并将它们的交集作为定义可行区域的一组下泄流量约束如下:
(2)可行集范围确定
据水量平衡方程可知,在其他变量确定的前提下,出库径流与时段末库容呈反比,即可以使用最小下泄流量求取时段末库容上限值,使用最大下泄流量求取时段末库容下限值。如下式所示,利用水量平衡,在已知的入库流量与时段初库容的条件下,将下泄流量约束先转变为库容约束。在转变后与原有的库容约束进行比较,取交集作为可行范围的上下限。
本发明中以时段末水位作为决策变量,故将获得的水位库容范围通过水位-库容关系转变为水位范围。
此外,利用已知的多年径流过程以及人工模拟的方式获得足够长的径流过程,人工模拟径流符合每月径流的理论频率曲线。可以认为所给出的径流过程包含了所有可能出现的径流过程。在确定性径流来水的前提下,对水库进行单库DP调度获得水库多年的水位变化过程并取水位的上下限。通过多年调度水位过程可知,在全年的初水位与末水位固定的条件下,无论径流条件如何变化,一部分的水位永远不会达到。如图1所示:水位上限与正常高水位之间的水位,水位下限与死水位之间的水位不会在调度过程中遇到,故可以在调度计算之前将这些范围排除,进一步缩减可行区域范围。
(3)可行集范围缩减
将所有水电站的月末水位过程集合形成三维矩阵A。三个维度分别是:径流、月份与水电站,然后根据A为每个时段中的每个水库设置了水位的上限和下限。将满足约束条件的水位范围与这些水位限制进行比较,并将重叠区域作为期间i的可行区域,删除不满足约束条件的任何水位范围以及矩阵A中的水位范围未覆盖的任何范围,具体计算公式如下:
在计算过程中,按照时段大小从小到大依次确定可行区域。在时期1中,根据初始水位并结合第一时期的约束条件和入水量来获得时段末的可行域。结合RS优化方法,该方法从可行区域采样水位,以获得下一个时期的初始条件。这样可以实现确定可行域与抽样依次进行,动态确定可行域范围。相较于利用前一阶段的可行域边界来确定该阶段的可行域,能更好的删除不满足约束的水位范围,可行域范围被大大缩小了。因此,可知通过以这种方式依次计算每个时期中每个时段的可行区,可以大大提高HDRS的计算效率。FRI方法原理如图2所示。
(b)水电站群确定性调度模型求解
对于大规模水电系统调度问题,难以将所有可能结果都计算出来进行比较。本发明在随机抽样的基础上使用RS优化并增加了结果评定,根据搜索结果动态调整采样数目。为确保从所有候选解决方案中选择的样本具有合理的可靠性水平,将可接受性和可靠性参数引入到RS优化中。迭代评估选定的水位状态,以评估目标值的提高,该目标值将用作整个搜索过程的收敛标准。为了进一步提升样本的可靠度,依据多年的调度结果,赋予水位分段抽样权重,将随机抽样进化为概率抽样。
(1)满意解序列与可靠度指标
为了了解水库调度的解分布并辅助解释随机抽样优化方法的可靠度指标a%,b%,选择动态规划计算的小湾某时段决策与状态组合得到的所有解进行分析,具体分布概率如下:
a(N)=n_N/n_all (29)
式中:n_N代表出力N出现的个数;n_all代表所有出力结果的个数;a(N)代表出力为N的结果出现的概率。A(N1)代表小于N1的结果出现的概率。
根据附图3可知,结果的数目与结果优劣呈正相关关系,越接近最优解的区间内包含的结果数目越多。在随机抽样的情况下各个决策出现的概率相同,抽取得到最优解的可能性太小,但是抽取得到最优解附近范围的近似最优解的概率则明显提升很多。
对于大规模的水库调度问题,决策搜索范围过大无法直接进行求解,相对于找出最优解,找出近似解更为简单。设置参数a,认为当解按照优劣排序时,处于前a%区间的解都是满意的解。在随机抽样过程中,每个区间的样本都有一定的概率被抽到,当完全随机时,每个样本被抽到的概率为:
a′%=1/n_all (31)
对应的,位于满意解被抽到的概率为:
a%=n
满意解/n_all (32)
式中,n
满意解代表满意解区间的解数目。
随机抽样概率只与样本数目有关,当通过客观原因赋予不同样本不同的概率,依据样本概率抽样时,随机抽样就变化成了概率抽样,在抽样次数足够多的情况下,随机抽样结果分布较为平均,与总体结果分布近似。概率抽样的结果分布与结果的概率值相关,当表现较优的结果拥有更大的概率时,抽样结果的分布则表现为较优结果数目增加,较劣结果数目减少。所以当赋予给较优结果的样本更大的概率值时,使用概率抽样能提升抽样的可靠性。
(2)随机抽样
在计算中,目标是获得满意解区间的结果,每个解位于满意解区间的概率为a%。则每随机抽出一个解,它不处于满意解区间的概率为(1-a%),连续抽出s个候选解,它们都不是处于前a%的概率为(1-a%)
s,由于抽样目标是抽出一个位于满意解区间的解,那么当:
(1-a%)
s≤b% (33)
则抽出S个样本,为了保证含有满意解,则:
即,只要抽出的样本数目大于S
min,则就有(1-b%)的概率抽取到满意解。
由于实际计算中,参数的选取需要根据经验决定,所以为了保证能抽取到满意解,还需要对得到的解进行评估。评估时,我们认为,如果继续增大抽样数目对于改进优化的效果不够明显,则认为已经找到了抽样的最优解,若继续增大抽样数目能够明显改善优化结果,则认为还需继续增加抽样数目,则进入迭代,直到找到满意解为止。
为了完成评估,改变a值,形成一个序列{a
1,a
2,a
3…,a
M},对应的抽样数目序列为{S
1,S
2,S
3…,S
M}。按照这些样本数目进行抽样,获得每个抽样组合的最优解与第五优解序列{f1,1,f1,2,f1,3,…},{f5,1,f5,2,f5,3,…}。对解序列进行求导:
d1<ε1 (37)
d5<ε5 (38)
由f1与f5随a的变化情况评估所得到的解是否令人满意,设置精度ε1与ε5。当精度公式可以满足,且f1与f5较为接近时,表明获得的满意解密度较大,且阶段健壮性较强,f1为满意解。当两个精度公式不能满足时,不能判定此时获得的最优解为满意解,需要继续扩大抽样数目。
将增加的样本解加入以有的解集合中,判断新的解集合是都满足精度要求。
(3)计算步骤
Step 1:设置参数。包括参数a,b,ε1与ε5。
Step 2:确定参数序列{a
1,a
2,a
3…,a
M}与{S
1,S
2,S
3…,S
M};
Step 3:依据序列{S
1,S
2,S
3…,S
M}第M-l至M个元素进行抽样,并计算获得最优解序列{f1}与第五优解{f5}。m通常取20-50;
Step 4:求解解序列的变化率d1与d5,并进行精度判断,如果满足精度判断则输出最优解,若不满足则进入Step 5。
Step 5:增加抽样数目,并将解加入已有的解集中,返回Step 4。
(c)总体求解方法步骤
结合上述关键问题的解决思路,一次完整的水电站群确定性优化调度降维过程可通过如下步骤表述:
步骤1.建立超大规模水电系统确定性优化调度模型,该模型以发电量最大为目标;
步骤2.读取基础数据,初始化计算参数,包括水电站运行参数、水位控制约束、下泄流量约束、电站出力约束、水量平衡约束;
步骤3.设置参数a,b,m,l,ε1与ε5,水位矩阵A,各区间权重分布w;
步骤4.确定参数序列{a
1,a
2,a
3…,a
M}与{S
1,S
2,S
3…,S
M};
步骤5.设置n=1,k=1,t=1,取S=S
k;其中n是抽样样本的序号,k是数组{a
1,a
2,a
3…,a
M} 和{S
1,S
2,S
3…,S
M}序列里的序号,S是抽样的数目,S
k代表{S
1,S
2,S
3…,S
M}序列里第k个元素;
步骤6.根据可行域识别原理对时段t进行计算,确定时段末的可行域范围;
步骤7.在可行域范围内进行概率抽样,抽出一个末水位进行该时段的出力计算;
步骤8.令t=t+1;判断t=T是否成立,若成立,则取全年末水位为时段末水位进行出力计算,并求出各时段出力之和为样本n的抽样结果,设置t=1,进入步骤9;若不成立,则返回步骤6继续计算;
步骤9.判断n=S是否成立,若成立,求取S个样本中的最优解与第五优解,设置n=1,进入步骤10,若不成立,n=n+1,返回步骤6继续计算;
步骤10.判断k=M是否成立,若成立,进入步骤11,若不成立,k=k+1,返回步骤6继续计算;
步骤11.获得最优解序列{f1}与第五优解{f5},求解序列的变化率d1与d5;
步骤12.进行精度判断,如果满足精度判断则输出最优解,若不满足则计算增加抽样数目,设置S=ΔS,k=M+1,返回步骤6继续计算。
现以澜沧江干流12个水电站、金沙江8个水电站和南盘江1个水电站,共21个水电站为研究对象,选择该系统53年的实际径流过程,即1954-1959、1961-1999和2008-2015,进行计算,以验证本发明方法的效率。表1中列出了详细的发电量和计算时间数据,表2列出了各水电站发电量计算结果,表3列出了所有年份的最佳结果和最差结果。
根据表1、表2,与离散微分动态规划(DDDP)和逐步优化算法(POA)相比,本发明提出的方法计算的年发电量分别提高了4.37%和8.74%,计算时间分别减少了98.73%和96.86%。而且,在多次计算中发现,发电量的变化很小,说明解的质量具有较好的稳定性。平均计算耗时明显小于DDDP和POA的计算时间。结果说明,本发明提出的方法能在更短的时间内获得更好的水电站优化调度结果。
表1 21个水电站多年计算结果对比
表2各水电站发电量对比
Claims (1)
- 一种耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法,其特征在于包括以下步骤:步骤1.以水能高效利用为准则,建立水电站群优化调度模型,具体如下:式中:E代表所有水电站的发电量之和;Δt’代表时段t内的小时数;i代表水电站序号;K i代表水电站i的出力系数;h i,t代表水电站在时段t的平均出力净水头;q i,t代表水电站i在时段t的发电流量;N i,t代表水电站i在时段t的出力;I代表水电站的个数;步骤2.初始化计算参数,包括水电站运行参数、水位控制约束、下泄流量约束、电站出力约束、水量平衡约束;N i,min≤N i,t≤N i,max (4)V i,t+1=V i,t+3600(Q i,t-R i,t)*Δt′ (5)式中:Z i,0代表水电站i的初始水位,Z i,T代表水电站i的末水位;Z i,S与Z i,Z分别代表水电站i的水位最小值与最大值;Z i,t代表水电站i在时段t的水位;R i,min与R i,max分别为水电站i的出库流量下限与上限;R i,t代表水电站i在时段t的下泄流量;q i,min与q i,max分别为水电站i的发电流量上下限;q i,t代表水电站i在时段t的发电流量;d i,t为水电站i在时段t的弃水流量;N i,min与N i,max分别为水电站i的出力下限与上限;N i,t代表水电站i在时段t的出力;V i,t为水电站i在时段t的初始库容;Q i,t为水电站i在时段t的入库径流;Q i,t'为水电站i在时段t的区间流量;∑R t代表上游水电站在时段t的下泄流量之和;当水电站i=1时,入库径 流为区间径流,当水电站i>1时,入库径流为区间径流与上游水电站下泄流量之和;步骤3.设置可靠度指标a、b,以及抽样方法最优解和第五优解的精度ε1与ε5;步骤4.确定参数序列{a 1,a 2,a 3…,a M},与抽样数目序列{S 1,S 2,S 3…,S M};其中M是数组{a 1,a 2,a 3…,a M}和{S 1,S 2,S 3…,S M}的元素个数;步骤5.设置n=1,k=1,t=1,取S=S k;其中n是抽样样本的序号,k是数组{a 1,a 2,a 3…,a M}和{S 1,S 2,S 3…,S M}序列里的序号,S是抽样的数目,S k代表{S 1,S 2,S 3…,S M}序列里第k个元素;步骤6.确定水电站在时段t末的可行域范围,耦合上述多种水电站调度运行约束,确定可行域范围,包含以下几步:(1)约束条件转换:对于出力约束,利用耗水率μ i将其转化为下泄流量约束,如下所示:对于下泄流量约束,使用q i,min与q i,max寻求可行域,比较两组下泄流量约束,并将它们的交集作为定义可行区域的一组下泄流量约束如下:(2)可行域范围确定据水量平衡方程知,在其他变量确定的前提下,出库径流与时段末库容呈反比,即使用最小下泄流量求取时段末库容上限值,使用最大下泄流量求取时段末库容下限值;如下式所示,利用水量平衡,在已知的入库流量与时段初库容的条件下,将下泄流量约束先转变为库容约束;在转变后与原有的库容约束进行比较,取交集作为可行范围的上下限;根据上式得到的库容范围,通过水位~库容关系曲线将其转变为等效的水位范围;(4)采用下式进一步缩减可行域范围步骤7.在水电站的水位可行域范围内进行概率抽样,以计算面临时段的出力:在计算中,每个库水位位于满意解区间的概率为a%;则每随机抽出一个解,它不处于满意解区间的概率为(1-a%),在上一步骤确定的可行域内连续抽出s个候选水位,这些水位都不是处于前a%的概率为(1-a%) s,当:(1-a%) s≤b% (12)则抽出S个库水位,为了保证含有满意解,则:当抽出的样本数目大于S min时,则有(1-b%)的概率抽取到满意的水位;步骤8.令t=t+1;判断t=T是否成立,若成立,则取全年末水位为时段末水位进行出力计算,并求出各时段出力之和为样本n的抽样结果,设置t=1,进入步骤9;若不成立,则返回步骤6继续计算;步骤9.判断n=S是否成立,若成立,求取S个库水位样本中的最优解与第五优解,设置n=1,进入步骤10,若不成立,n=n+1,返回步骤6继续计算;步骤10.判断k=M是否成立,若成立,进入步骤11,若不成立,k=k+1,返回步骤6继续计算;步骤11.获得水电站调度最优解序列{f1}与第五优解{f5},获得每个抽样组合的最优解与第五优解序列{f 1,1,f 1,2,f 1,3,…},{f 5,1,f 5,2,f 5,3,…},求解序列的变化率d1与d5,具体公式如下:d1<ε1 (16)d5<ε5 (17)步骤12.进行精度判断,如果满足精度判断则输出最优解,若不满足则计算增加抽样数目,设置S=ΔS,k=M+1,返回步骤6继续计算。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2021/091245 WO2022226952A1 (zh) | 2021-04-30 | 2021-04-30 | 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2021/091245 WO2022226952A1 (zh) | 2021-04-30 | 2021-04-30 | 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022226952A1 true WO2022226952A1 (zh) | 2022-11-03 |
Family
ID=83846730
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2021/091245 Ceased WO2022226952A1 (zh) | 2021-04-30 | 2021-04-30 | 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 |
Country Status (1)
| Country | Link |
|---|---|
| WO (1) | WO2022226952A1 (zh) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115879674A (zh) * | 2023-02-21 | 2023-03-31 | 珠江水利委员会珠江水利科学研究院 | 一种基于双层逐步试算的梯级水库调度模型修正方法 |
| CN116187595A (zh) * | 2023-04-27 | 2023-05-30 | 北京玻色量子科技有限公司 | 多目标多任务路径调度效率优化方法、装置、介质及设备 |
| CN117674293A (zh) * | 2023-12-07 | 2024-03-08 | 华能西藏雅鲁藏布江水电开发投资有限公司 | 一种梯级水电站的长期发电优化调度方法及装置 |
| CN118378857A (zh) * | 2024-06-21 | 2024-07-23 | 武汉华飞智能电气科技有限公司 | 一种梯级水电站设备故障应急调度方法及系统 |
| CN118657352A (zh) * | 2024-08-19 | 2024-09-17 | 安徽金海迪尔信息技术有限责任公司 | 基于神经网络技术的水利调度方法及系统 |
| CN119151045A (zh) * | 2024-08-22 | 2024-12-17 | 中国长江电力股份有限公司 | 一种基于可行域搜索映射的水电站群优化调度降维方法 |
| CN119765509A (zh) * | 2025-03-05 | 2025-04-04 | 云南华电金沙江中游水电开发有限公司 | 一种防洪蓄水的水库中长期发电调度决策快速生成方法 |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103745023A (zh) * | 2013-11-22 | 2014-04-23 | 华中科技大学 | 水电站发电出力方案制作与最优负荷分配耦合建模方法 |
| CN108520317A (zh) * | 2018-03-27 | 2018-09-11 | 华中科技大学 | 一种水电系统联合优化调度的拉丁方动态规划降维方法 |
| CN111353649A (zh) * | 2020-03-06 | 2020-06-30 | 大连理工大学 | 一种基于kl展开的大规模水电站群优化调度降维方法 |
| CN112633578A (zh) * | 2020-12-24 | 2021-04-09 | 国电电力发展股份有限公司和禹水电开发公司 | 一种引调水工程影响下梯级水库群优化调度方法 |
-
2021
- 2021-04-30 WO PCT/CN2021/091245 patent/WO2022226952A1/zh not_active Ceased
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103745023A (zh) * | 2013-11-22 | 2014-04-23 | 华中科技大学 | 水电站发电出力方案制作与最优负荷分配耦合建模方法 |
| CN108520317A (zh) * | 2018-03-27 | 2018-09-11 | 华中科技大学 | 一种水电系统联合优化调度的拉丁方动态规划降维方法 |
| CN111353649A (zh) * | 2020-03-06 | 2020-06-30 | 大连理工大学 | 一种基于kl展开的大规模水电站群优化调度降维方法 |
| CN112633578A (zh) * | 2020-12-24 | 2021-04-09 | 国电电力发展股份有限公司和禹水电开发公司 | 一种引调水工程影响下梯级水库群优化调度方法 |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115879674A (zh) * | 2023-02-21 | 2023-03-31 | 珠江水利委员会珠江水利科学研究院 | 一种基于双层逐步试算的梯级水库调度模型修正方法 |
| CN115879674B (zh) * | 2023-02-21 | 2023-08-08 | 珠江水利委员会珠江水利科学研究院 | 一种基于双层逐步试算的梯级水库调度模型修正方法 |
| CN116187595A (zh) * | 2023-04-27 | 2023-05-30 | 北京玻色量子科技有限公司 | 多目标多任务路径调度效率优化方法、装置、介质及设备 |
| CN117674293A (zh) * | 2023-12-07 | 2024-03-08 | 华能西藏雅鲁藏布江水电开发投资有限公司 | 一种梯级水电站的长期发电优化调度方法及装置 |
| CN118378857A (zh) * | 2024-06-21 | 2024-07-23 | 武汉华飞智能电气科技有限公司 | 一种梯级水电站设备故障应急调度方法及系统 |
| CN118657352A (zh) * | 2024-08-19 | 2024-09-17 | 安徽金海迪尔信息技术有限责任公司 | 基于神经网络技术的水利调度方法及系统 |
| CN119151045A (zh) * | 2024-08-22 | 2024-12-17 | 中国长江电力股份有限公司 | 一种基于可行域搜索映射的水电站群优化调度降维方法 |
| CN119765509A (zh) * | 2025-03-05 | 2025-04-04 | 云南华电金沙江中游水电开发有限公司 | 一种防洪蓄水的水库中长期发电调度决策快速生成方法 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2022226952A1 (zh) | 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 | |
| CN108109076B (zh) | 一种考虑径流预报的梯级水电站群发电调度弃水风险分析方法 | |
| CN113312735B (zh) | 一种城市供水管网dma分区方法 | |
| WO2021142900A1 (zh) | 一种基于指标联动分析的电力系统调峰调度多目标优化方法 | |
| CN105243438A (zh) | 一种考虑径流不确定性的多年调节水库优化调度方法 | |
| CN107274302B (zh) | 一种跨省互联水电站群长期联合调峰调度方法 | |
| CN115238993A (zh) | 基于nsga-ⅱ算法的水库群多目标优化调度方案决策系统 | |
| CN116050628A (zh) | 一种基于数字孪生的考虑预报不确定性的水库防洪优化调度方法 | |
| WO2019119413A1 (zh) | 一种耦合相对目标接近度和边际分析原理的梯级水电站多目标优化调度方法 | |
| CN104268653B (zh) | 基于集束径流预报的梯级水库优化调度方法 | |
| CN113128757B (zh) | 考虑跨流域调水影响下的水库分期汛限水位优化设计方法 | |
| CN111079086B (zh) | 一种基于多元联合分布的水资源系统多重风险评估方法 | |
| CN107730054A (zh) | 一种基于支持向量回归的燃气负荷组合预测方法 | |
| CN113128782A (zh) | 耦合可行域识别和随机抽样的大规模水电站群优化调度降维方法 | |
| CN106599562B (zh) | 基于概率加权fdc法的河流生态需水量计算方法 | |
| CN106682810A (zh) | 巨型水电站动态投产下跨流域梯级水电站群长期运行方法 | |
| CN107742000B (zh) | 锅炉燃烧含氧量建模方法 | |
| CN111080157A (zh) | 一种梯级水电站排磷量的调度方法和系统 | |
| CN115222105A (zh) | 考虑风险和效益博弈均衡的梯级电站调度优化方法及系统 | |
| CN116703259A (zh) | 一种适用于流域梯级水电站群的水能转化关系构建方法 | |
| WO2025091699A1 (zh) | 一种考虑检修和生态调度的梯级水电站中长期优化调度模型 | |
| CN115456205A (zh) | 一种大规模多目标机组检修优化方法 | |
| CN106127336A (zh) | 一种基于多目标飞蛾算法的小型水电站优化调度方法 | |
| CN103336998A (zh) | 一种风电场功率波动平抑目标值的优化计算方法 | |
| CN115619120A (zh) | 一种面向恢复天然水文情势的水库调度方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21938430 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 21938430 Country of ref document: EP Kind code of ref document: A1 |





























