Medium-term dispatching method for river basin hydro-wind-solar system considering short-term load features of multi-receiving-end power grid

Abstract

The present invention relates to the field of power system dispatching. Disclosed is a mid-term dispatching method for a river basin hydro-wind-solar system considering short-term load features of a multi-receiving-end power grid. A peak-flat-valley load segmentation model and a variable step size search strategy are used to integrate a daily average load, a daily load rate, a daily peak-valley difference rate, a peak period load rate and a valley period load rate, and a peak occurrence time index and a valley occurrence time index, so as to accurately describe differential load features and a peak regulation demand of a multi-receiving-end power grid, thereby reconstructing a receiving-end load demand process. A multi-objective medium- and short-term nested dispatching model for a river basin hydro-wind-solar system having the maximum power generation capacity and the minimum power transmission deviation is provided, and model optimization solution is performed by means of mixed integer linear programming to determine a medium- and short-term power capacity and a power transmission process plan. By means of instance analysis of a hydro-wind-solar hybrid system in an extra-large river basin, the results show that the present invention can significantly reduce the deviation between a power transmission plan and a receiving end demand and reduce power curtailment while ensuring the total power generation capacity of the river basin.

Description

Mid-term scheduling method of watershed water-wind-solar system considering short-term load characteristics of multi-receiving-end power grid TECHNICAL FIELD

[0001] The present application belongs to the field of power system dispatching, and relates to a mid-term scheduling method of watershed water-wind-solar system considering short-term load characteristics of multi-receiving-end power grid. BACKGROUND

[0002] The reverse distribution characteristics of energy resources and load demand in China determine that large-scale cross-provincial and cross-regional power transmission is a key way to promote the large-scale consumption of water-wind-solar clean power in the southwest, northwest and north China. However, due to the large difference in electricity demand and the different load characteristics of the receiving-end provinces (cities), the strong peak shaving demand puts forward high requirements for the cross-provincial and cross-regional DC power receiving process. It is particularly important to accurately describe the load characteristics of the multi-receiving-end in the power transmission plan arrangement. In addition, the rapid increase in the scale and proportion of wind and solar installations in clean energy bases further increases the difficulty of power transmission plan arrangement, which has become a bottleneck problem to be solved in the actual dispatching and operation of high-proportion new energy power grids and watershed clean energy bases.

[0003] To accurately describe the load characteristics of the power grid, the following indicators are commonly used: daily load rate reflecting the overall load change, daily peak-valley difference rate reflecting the peak regulation capacity of the power grid, daily maximum load utilization time reflecting the time utilization efficiency, maximum and minimum load occurrence time reflecting the peak and valley nodes (Wan Q, Yu Y. Power load pattern recognition algorithm based on characteristic index dimension reduction and improved entropy weight method[J]. Energy Reports, 2020, 6: 797-806.), peak period, flat period, and valley period load rate reflecting the load change of peak, flat, and valley periods (Si C, Xu S, Wan C, et al. Electric load clustering in smart grid: Methodologies, applications, and future trends[J]. Journal of Modern Power Systems and Clean Energy, 2021, 9(2): 237-252.). These indicators generally start from the overall characteristics of the load, and lack detailed description of local load changes and timing flexibility requirements. Therefore, some studies propose load reconstruction methods, mainly dividing the load curve in the dispatching period into time periods (Yang H, Gao Y, Ma Y, et al. Optimal modification of peak-valley period under multiple time-of-use schemes based on dynamic load point method considering reliability[J]. IEEE Transactions on Power Systems, 2021, 37(5): 3889-3901.), and then constructing equivalent load curves according to the load characteristics in each period. However, there are still problems such as the reconstructed curve cannot avoid the isolated points of the original load, and the equivalent characteristics of non-adjacent periods are difficult to describe, leading to frequent fluctuations in water, wind, and solar power transmission plans and single-period mutations, which brings great adjustment pressure to water and electricity and is not conducive to the adjustment of the dispatching plan. Therefore, how to consider the complex load demand of the receiving end power grid in detail is very important for the coordinated consumption of water, wind, and solar power generation.

[0004] At present, the research on the complementary operation of water and wind power generation at home and abroad mainly focuses on the business problems such as power generation planning, scheduling rule extraction, complementary capacity analysis, risk and benefit quantification, joint peak shaving, etc., and focuses on long-term power compensation optimization, short-term flexible response, and the evaluation of the effect of complementary operation. For the continuous multi-day extreme large and small mode scenarios of wind and light new energy, a few studies have begun to focus on it, such as the identification and extraction of new energy continuous extreme output scenarios (Jiang Haiyang, Du Ershun, Ma Jiahao, et al. New power system planning method considering long-period supply-demand imbalance risk[J / OL]. Proceedings of the CSEE. https: / / doi.org / 10.13334 / j.0258-8013.pcsee.230440), preventive and emergency dispatch methods under extreme weather (Junjie R, Ming Z, Zhi Z, et al. Coordination of preventive and emergency dispatch in renewable energy integrated power systems under extreme weather[J]. IET Renewable Power Generation, 2024, 18(7): 1164-1176.), risk analysis and optimal dispatching method under extreme scenarios (Cai X, Qin Z, Hou Y. Improving wind power utilisation under stormy weather condition by risk-limiting unit commitment[J]. IET Renewable Power Generation, 2018, 12(15): 1778-1785.), which provide effective ideas for dealing with multi-day extreme power generation of new energy. However, how to consider the different time scales of the water and power coupling characteristics of the multi-energy complementary system, the mutual constraint effect of water level and power dispatching boundary, the complex hydraulic connection of "one station and multiple plants" cascade hydropower stations and the fine modeling and calculation of plant output, and the different power demand of the receiving end under the "one plant and multiple transmission" mode are new problems and new needs of large river basin water, wind and light systems, which require modeling and solving methods that adapt to them to determine more efficient and practical power transmission plans.

[0005] To solve the above problems, the application provides a medium-term scheduling method of a watershed water-wind-solar system considering short-term load characteristics of multiple receiving-end power grids, and application tests are carried out relying on a certain watershed water-wind-solar complementary system. The results show that the application can significantly reduce the deviation of power transmission planning and receiving-end demand and reduce power curtailment under the condition of ensuring the total power generation of the watershed. The verification results show that, under the condition of basically unchanged power generation, the deviation of the calculation scheme of the application from the original method is reduced by 88.6% in the dry season and 69.9% in the flood season, which presents good practicability.

[0006] SUMMARY

[0007] The technical problem to be solved by the application is to provide a medium-term scheduling problem of a watershed water-wind-solar system considering short-term load characteristics of multiple receiving-end power grids. By accurately describing the differential load characteristics and peak shaving demand of multiple receiving-end power grids, the receiving-end power demand process is reconstructed, and a multi-objective medium-short-term nested scheduling model with minimum power transmission deviation and maximum power generation is created to determine the efficient multi-day power transmission amount and daily power transmission process plan of the water-wind-solar system.

[0008] The technical scheme of the application is as follows:

[0009] The medium-term scheduling method of a watershed water-wind-solar system considering short-term load characteristics of multiple receiving-end power grids comprises the following steps:

[0010] (1) According to the peak-valley characteristics of the load of each province, a time period division model is established based on the criterion of minimum load variance in the same type of time period set.

[0011] Objective function:

[0012] Constraint condition:

[0013] In the formula, L t is the load of the t period, MW; g represents the time period type, g=1 represents the valley period, g=2 represents the flat period, and g=3 represents the peak period; represents the time period set contained by the gth time period type; I g represents the number of time periods contained by the gth time period type; represents the i th element in the time period set .

[0014] (2) A variable step search strategy is established based on the numpy module in the python programming language to solve the time period division model in step (1). The solving idea is to find two load dividing points Y1 and Y2, where Y1<Y2; when L t <Y1, the period t is a valley period; when Y1≤L t ≤Y2, the period t is a flat period; and when L tY2, the time period t is a peak time period; the specific steps are as follows:

[0015] Step 1: sort the loads of each time period of the original load curve from small to large to obtain an increasing load sequence l1, l2, …, l m ,…,l M , obtain the average value of l m and l m+1 , and obtain a variable step search set {y1, y2, …, y m ,…,y M-1}, wherein y m =(l m +l m+1 ) / 2;

[0016] Step 2: define k1 and k2 as the sequence numbers of each element in the search set, let k1=1, k2=k1+1, obtain each type of time period set and obtain the corresponding objective function value let

[0017] Step 3: search upwards until the highest load interval, let k1=k1+1 or k2=k2+1, until k1=M-2, k2=M-1, obtain the corresponding objective function value if , then let

[0018] Step 4: obtain the optimal objective function value and the corresponding set of each type of time period

[0019] (3) Calculate the load characteristic index, and establish a load reconstruction model;

[0020] The objective function is as follows:

[0021] In the formula: CI u represents the characteristic index of the original load demand curve; RCI u represents the characteristic index of the reconstructed load demand curve; w u represents the weight coefficient of the characteristic index; and DR represents the gap between the characteristic index of the reconstructed load demand curve and the original load curve.

[0022] The constraint condition is as follows:

[0023] 1) The characteristic index of the original load curve CI1=L ave (4) CI2=L ave / L max(5) CI3=(L max -L min ) / L max (6) CI4=L ave,peak / L ave (7) CI5=L ave,low / L ave (8) CI6=T max (9) CI7=T min (10)

[0024] In the formula: CI1, CI2, CI3, CI4, CI5, CI6, CI7 represent the original load curve daily average load, daily load rate, daily peak-valley difference rate, peak period load rate, valley period load rate, peak time, valley time, respectively; ave , L max , L min respectively represent the average value, maximum value, minimum value of daily load; ave,peak , L ave,low respectively represent the average value of peak period and the average value of valley period; max , T min respectively represent the peak time and the valley time;

[0025] 2) Reconstructed load curve characteristic index

[0026] In the formula: RCI1, RCI2, RCI3, RCI4, RCI5, RCI6, RCI7 represent the daily average load, daily load rate, daily peak-valley difference rate, peak period load rate, valley period load rate, peak time, valley time of the reconstructed load curve, respectively; respectively represent the average value, maximum value, minimum value of the reconstructed load curve; respectively represent the average value of peak period and the average value of valley period of the reconstructed load curve; respectively represent the peak time and the valley time of the reconstructed load curve;

[0027] 3) Peak, flat, and valley periods of the reconstructed load curve

[0028] In the formula: respectively represent the values of the valley period, flat period, and peak period of the reconstructed load curve;

[0029] (4) Use the python-numpy package to build an equal step search method to solve the load reconstruction model in step (3), as follows:

[0030] Step1: Find the maximum value L from the load sequence max and the minimum value L min , determine the search step size sw, get the equal step search set {r1, r2, …, r a , …, r A}, where r a = L min + sw*(a-1), a is the element sequence number in the search set, A is the number of elements in the search set;

[0031] Step2: Define b1, b2, b3 as the sequence number of each element in the search set, let b1=1, b2=b1+1, b3=b2+1, get the corresponding reconstructed load curve, calculate the load characteristic index and target value Let

[0032] Step3: Let b1=b1+1 or b2=b2+1 or b3=b3+1, until b1=A-2, b2=A-1, b3=A, find the corresponding target function value If , let find the optimal target function value corresponding to the reconstructed load curve;

[0033] Step4: Get the optimal target function value DR and its corresponding reconstructed load curve;

[0034] (5) Considering the left and right banks, upstream and downstream connections of cascade hydropower stations, and combining short-term scheduling constraints, a short-term nested scheduling model for the water, wind and light system of the river basin is constructed:

[0035] The maximum power generation objective function is as follows:

[0036] In the formula: E is the maximum power generation objective function, MWh; represents the output of hydropower station n power plant j on the dth day, MW; and P n,j,d respectively represent the output of wind power plant and photovoltaic power plant on the dth day, MW; P

[0037] The minimum power transmission deviation objective function is as follows:

[0038] ED = EDmin (EDmax - EDmin) (15) where ED is the minimum target function of power transmission deviation, MW 2 ; denotes the power transmission plan of power plant n, j, d, t to province s, MW; ξ s denotes the load demand of province s, using the load reconstruction curve obtained in Section 2; η s,n,j,d denotes the scaling coefficient of the receiving end load demand; s and S represent the province serial number and the total number of provinces, respectively; t and T represent the time period serial number and the total number of time periods, respectively;

[0039] 2) Constraint conditions

[0040] In addition to considering medium-term scheduling and short-term scheduling, the medium-short-term nested power constraint is also considered, i.e., the daily power obtained by medium-term scheduling should be equal to the sum of power generation in each time period within the day, as shown in equation (15);

[0041] (6) On the Gurobi solver platform, the medium-short-term nested scheduling model of the river basin water, wind and light system is converted into a mixed integer linear programming model using a Python program to achieve efficient solution, as follows:

[0042] Step 1: The multi-objective model of the maximum power generation target function and the minimum power transmission deviation target function is converted into a single objective model using the constraint conversion method, and the specific method is to convert the maximum power generation target function into the constraint condition of the minimum power transmission deviation target, as shown in equation (16);

[0043] where E set denotes the total power generation requirement of the river basin water, wind and light system within the scheduling period, MWh; E set The initial value of E

[0044] Step 2: For the one-dimensional nonlinear function relationship of water level and reservoir capacity, piecewise linearization is used for processing; for the two-dimensional nonlinear function relationship of the water power generation function, the parallelogram two-dimensional interpolation method is used for linearization processing;

[0045] Step 3: A set of non-inferior solution set (Pareto frontier) of the multi-objective model is obtained by gradually relaxing the power generation constraint.

[0046] Step 4: The comprehensive benefit index is calculated, and the decision scheme is selected from the solution set, and the index calculation formula is shown in equation (17);

[0047] Compared with the prior art, the present application has the beneficial effects that: the reconstructed curve obtained by using the load demand reconstruction method based on peak and valley period division can accurately reflect the characteristics and change trend of the load of the receiving end power grid; the medium and short-term nested scheduling model of water, wind and light can significantly reduce the deviation between the power transmission plan and the demand of the receiving end under the condition of ensuring the total power generation of the basin, effectively alleviate the adjustment pressure of the receiving end province, and significantly reduce the curtailment of electricity and improve the consumption capacity of clean energy. BRIEF DESCRIPTION OF DRAWINGS

[0048] Fig. 1 is a period division principle diagram;

[0049] Fig. 2 is a medium and short-term nested mode schematic diagram;

[0050] Fig. 3 is a parallelogram meshing schematic diagram;

[0051] Fig. 4 is a power generation and power transmission deviation relationship curve (dry season);

[0052] Fig. 5 is a power generation and power transmission deviation relationship curve (flood season). DETAILED DESCRIPTION

[0053] The specific embodiment of the present application will be further described below in combination with the drawings and technical solutions.

[0054] The medium-term scheduling method of the water, wind and light system of the basin considering the short-term load characteristics of the multi-receiving end power grid is as follows:

[0055] (1) According to the peak and valley characteristics of the load of each provincial power grid, a period division model is established with the minimum load variance in the same type period set as the criterion.

[0056] Objective function

[0057] Constraint condition

[0058] In the formula: L t is the load of the t period, MW; g represents the period type, g=1 represents the valley period, g=2 represents the flat period, and g=3 represents the peak period; represents the period set contained by the gth period type; I g represents the number of periods contained by the gth period type; represents the i th element in the period set .

[0059] (2) A variable step search strategy is established based on the numpy module in the python programming language to solve the period division model in (1) to obtain the period division result. The solution idea is to find two load dividing points Y1 and Y2, wherein Y1<Y2. When L tWhen Y1, the time period t is a valley period; when Y1≤L t When Y2, the time period t is a flat period; when L t When Y2, the time period t is a peak period. The principle is shown in Fig. 1, and the specific steps are as follows:

[0060] Step 1: Sort the loads at each time point from small to large to obtain an increasing sequence l1, l2, …, l m , …, l M , find the average value of l m and l m+1 , and obtain a variable step search set {y1, y2, …, y m , …, y M-1}, wherein y m = (l m +l m+1 ) / 2.

[0061] Step 2: Define k1 and k2 as the sequence numbers of each element in the search set, let k1=1, k2=k1+1, Y1=y k1 , Y2=y k2 , obtain each type of time period set and find the corresponding objective function value Var 1,2 , let Var=Var 1,2 .

[0062] Step 3: Search upwards until the highest load interval, let k1=k1+1 or k2=k2+1, until k1=M-2, k2=M-1, find the corresponding objective function value If , then

[0063] Step 4: Obtain the optimal objective function value and the corresponding set of each type of time period

[0064] (3) A reconstruction model is established according to the criterion that the difference between the reconstructed load demand process and the original load process is minimum, the maximum and minimum load occurrence time is a constraint condition, and the peak, flat and valley load values of the reconstructed curve are decision variables. The objective function is shown in formula (20).

[0065] In the formula: RCI u and CI u represent the characteristic indexes of the reconstructed load curve and the original load curve respectively; w u represents the weight coefficient of the characteristic index.

[0066] The constraint conditions are as follows:

[0067] 1) Original curve characteristic index CI1=L ave (21) CI2=L ave / L max (22) CI3=(L max -L min ) / L max (23) CI4=L ave,peak / L ave (24) CI5=L ave,low / L ave (25) CI6=T max (26) CI7=T min (27)

[0068] In the formula: CI1, CI2, CI3, CI4, CI5, CI6, CI7 respectively represent the original load curve daily average load, daily load rate, daily peak-valley difference rate, peak period load rate, valley period load rate, peak occurrence time, valley occurrence time; L ave , L max , L min respectively represent the average value, maximum value, minimum value of daily load; L ave,peak , L ave,low respectively represent the average value of peak period and the average value of valley period; T max , T min respectively represent the peak occurrence time and the valley occurrence time.

[0069] 2) Reconstruction load curve characteristic index

[0070] In the formula: RCI1, RCI2, RCI3, RCI4, RCI5, RCI6, RCI7 respectively represent the daily average load, daily load rate, daily peak-valley difference rate, peak period load rate, valley period load rate, peak occurrence time, valley occurrence time of the reconstructed load curve; respectively represent the average value, maximum value, minimum value of the reconstructed load curve; respectively represent the average value of peak period and the average value of valley period of the reconstructed load curve; respectively represent the peak occurrence time and the valley occurrence time of the reconstructed load curve.

[0071] 3) Peak, flat and valley periods of the reconstructed load curve

[0072] In the formula: respectively represent the values of the valley period, flat period, peak period of the reconstructed load curve.

[0073] (4) Construct an equal step search method using the python-numpy package to solve the reconstructed load model in (3), as follows:

[0074] Step 1: Find the maximum value L from the load sequence max and the minimum value L min , determine the search step sw, get the equal step search set {r1, r2, …, r a , …, r A}, where r a = L min + sw*(a-1), a is the element sequence number in the search set, A is the number of elements in the search set.

[0075] Step 2: Define b1, b2, b3 as the sequence number of each element in the search set, let b1=1, b2=b1+1, b3=b2+1, get the corresponding reconstructed load curve, calculate the load characteristic index and target value Let

[0076] Step 3: Let b1=b1+1 or b2=b2+1 or b3=b3+1, until b1=A-2, b2=A-1, b3=A, find the corresponding target function value If , let find the optimal reconstructed load curve corresponding to the optimal target function value;

[0077] Step 4: Get the optimal target function value DR and its corresponding reconstructed load curve.

[0078] (5) Considering the left and right banks, upstream and downstream connections of cascade hydropower stations, and combining short-term scheduling constraints, a short-term nested scheduling model for the river basin water, wind and light system is constructed. The idea of medium-term and short-term nesting is to take the daily power generation obtained by medium-term scheduling as the control condition of short-term scheduling, and finely decompose it to each time period within the day to determine the timing output process of hydropower to smooth the wind and light fluctuations as much as possible and meet the complex load demand of the receiving end, as shown in Figure 2. The model is as follows:

[0079] 1) Objective function

[0080] where: E is the maximum target function of power generation, MWh; P P n,j,dPn j d t represents the total power of water, wind and light of power plant n and power plant j on the dth day, MW;△t represents the number of hours per day, h; n and N represent the serial number of water power station and the total number of water power stations respectively; j represents the serial number of power plant, j=1 represents the left bank power plant, and j=2 represents the right bank power plant; d and D represent the serial number of days and the total number of days respectively.

[0081] In the formula: ED is the minimum target function of power transmission deviation, MW 2 ; Pn j d t represents the power transmission plan of power plant n and power plant j on the dth day t to the province s, MW;ξ s (t) represents the load demand of the province s, and the load reconstruction curve obtained in the second section is adopted;η s,n,j,d scales the load demand of the receiving end. s and S represent the serial number of the province and the total number of provinces respectively; t and T represent the serial number of the period and the total number of periods respectively.

[0082] 2) Constraint condition

[0083] In addition to considering the traditional medium-term scheduling and short-term scheduling, the present application also considers the medium-short-term nested power constraint, that is, the daily power obtained by medium-term scheduling should be equal to the sum of power generation in each period within a day, as shown in formula (32).

[0084] (6) On the Gurobi solver platform, a Python program is used to convert the medium-short-term nested scheduling model of the river basin water, wind and light system into a mixed integer linear programming model, so that efficient solving is realized, and the specific process is as follows:

[0085] Step 1: the constraint conversion method is used to convert the multi-objective model into a single-objective model, and the specific method is to convert the power generation target into a constraint condition of the minimum model of power transmission deviation, as shown in formula (33).

[0086] In the formula: E set represents the total power generation requirement of the river basin water, wind and light within the scheduling period, MWh. The initial value is the maximum value of power generation when the power transmission deviation target is not considered, that is, the optimal value of the target function of the daily-scale power generation maximum model.

[0087] Step 2: for the one-dimensional nonlinear function relationship of water level and reservoir capacity, the piecewise linearization is used for processing; for the two-dimensional nonlinear function relationship of water power generation function, the parallelogram two-dimensional interpolation method is used for linearization processing, and the power generation flow Q and the water level Z are discretized, as shown in figure 3 (taking 2*2 grid as an example), Q a,b and Z b are sampling points, PP a,b =f PZQ (Z b ,Q a,b), A and B are the number of rows and columns of the grid respectively, a and b are sequence indexes, a e {1, 2, …, A+1}, b e {1, 2, …, B+1}.

[0088] Step 3: Obtain a non-inferior solution set (Pareto front) of a group of multi-objective models by gradually relaxing the power generation constraint.

[0089] Step 4: Calculate the comprehensive benefit index, and select a decision scheme from the solution set, the index calculation formula is shown in formula (34).

[0090] The invention is verified by taking a water-wind-sunlight complementary base in a downstream basin as an example, four libraries of WDD, BHT, XLD and XJB are selected for the cascade hydropower stations, and the total installed capacity reaches 46460MW, at present, 10487MW of new energy is connected to the cascade hydropower, and clean electric energy is transmitted to the southeast load center through the ultra-high voltage direct current. The invention takes four hydropower stations and the wind power and photovoltaic bundled with each hydropower station for simulation calculation object, and verifies the effectiveness of the method. Since the continuous extreme output scene of new energy appears in a ten-day period, therefore, a ten-day period is selected as a dispatching cycle. The invention selects two typical periods of dry season and flood season in a year as examples for analysis, and the interval flow, initial and final water level and load process of the hydropower station are referenced to the actual operation data.

[0091] In order to fully reflect the peak-valley characteristics of the receiving end load, the invention selects the hourly load sequence of the day with the maximum peak-valley difference in the dispatching cycle for load reconstruction. Table 1 and Table 2 compare the characteristic indexes of the reconstructed load curve and the original load curve in the dry season and the flood season respectively, from the tables, it can be seen that the difference of each characteristic index is small. Among them, the daily load rate and the daily peak-valley difference rate are completely equal, indicating that the reconstructed load curve can accurately reflect the imbalance and change range of the daily load; the difference of the daily average load is within 1%, which reflects that the daily load level of the reconstructed load curve is equivalent to that of the original load curve; the difference of the peak period load rate and the valley period load rate is within 6.5%, which shows that the reconstructed load curve fully reflects the change of the load peak period and the valley period. The correlation coefficients of the reconstructed load curve and the original load curve are all above 90%, which reflects the overall consistency of the two.

[0092] Table 1 Comparison of load characteristic indexes (dry season)

[0093] Table 2 Comparison of load characteristic indexes (flood season)

[0094] The method converts the multi-objective model into a single-objective model by constraint conversion method, and obtains a set of non-inferior solution set by gradually relaxing the constraint conditions. Figures 4 and 5 respectively show the relationship curves of power generation and power transmission deviation in dry season and flood season. As can be seen from the figures, when the power generation is greater than a certain value, there is an obvious contradictory relationship between the power generation and the power transmission deviation, the greater the power generation, the greater the power transmission deviation; on the contrary, the smaller the power generation, the smaller the power transmission deviation. When the power generation is less than the value, the power transmission deviation reaches the minimum value and no longer changes. Pursuing the maximum power generation alone will cause the deviation of the power transmission process and the demand of the receiving end to be too large, increase the adjustment pressure of the receiving end power grid, and is not conducive to clean energy consumption; pursuing the minimum power transmission deviation alone will lose the power generation benefit. Therefore, the scheduling scheme needs to be selected by comprehensively considering the demands of both the power generation side and the power receiving side.

[0095] The power generation maximum scheme (scheme 1) which gives priority to the power generation target, the decision scheme (scheme 2), and the minimum power transmission deviation scheme (scheme 3) which gives priority to the power transmission deviation target are selected for comparative analysis.

[0096] From the perspective of power generation, in the dry season example, the total power generation of the three schemes is 4610.87 GWh, 4610.00 GWh, and 4608.50 GWh respectively, and the power generation of schemes 2 and 3 is reduced by 0.02% and 0.05% respectively compared with scheme 1; in the flood season example, the total power generation of the three schemes is 7135.93 GWh, 7115.00 GWh, and 7071.39 GWh respectively, and the power generation of schemes 2 and 3 is reduced by 0.29% and 0.91% respectively compared with scheme 1, and the change range of power generation is small.

[0097] From the perspective of power transmission deviation, in the dry season example, the power transmission deviation of the three schemes is 11.52 GW 2 , 1.31 GW 2 , and 0 respectively, and the power transmission deviation of schemes 2 and 3 is reduced by 88.6% and 100% respectively compared with scheme 1; in the flood season example, the power transmission deviation of the three schemes is 100.86 GW 2 , 30.36 GW 2 , and 0 respectively, and the power transmission deviation of schemes 2 and 3 is reduced by 69.9% and 100% respectively compared with scheme 1, and the reduction range of power transmission deviation is large.

Claims

1.A mid-term scheduling method for a watershed water-wind-solar system considering short-term load characteristics of a multi-receiving-end power grid, characterized in that, Comprising the following steps: (1) According to the peak-valley characteristics of each provincial load, the load variance in the same type of time period set is established as the criterion, and the time period division model is established; Objective function: Constraints: wherein: L t L is the load in MW for time period t; g indicates the time period type, g = 1 for valley period, g = 2 for flat period, g = 3 for peak period; represents a set of time slots included in the gth time slot type; I g represents the number of time slots included in the gth time slot type; representing a set of time periods The i-th element in the array; (2) Based on the numpy module in the python programming language, a variable step search strategy is established to solve the time period division model in step (1). The solving idea is to find two load dividing points Y1 and Y2, where Y1 t <Y1, the time period t is a valley period; when Y1≤L t ≤Y2, the time period t is a flat period; when L t >Y2, the time period t is a peak period; the specific steps are as follows: Step 1: sort the loads of each time period of the original load curve from small to large to obtain an increasing load sequence l1, l2, …, l m ,…,l M , find the average value of l m and l m+1 , and obtain a variable step search set {y1, y2, …, y m ,…,y M-1}, where y m =(l m +l m+1 ) / 2; Step 2: define k1, k2 as the serial number of each element in the search set, let k1 = 1, k2 = k1 + 1, obtaining a set of time periods of each type and the corresponding objective function value is found Let Step 3: Search upwards until the highest load interval, let k1 = k1 + 1 or k2 = k2 + 1, until k1 = M-2, k2 = M-1, find the corresponding objective function value If Let Step 4: obtain the optimal objective function value and the corresponding set of each type of time period (3) Calculate the load characteristic index, and establish the load reconstruction model; The objective function is as follows: wherein: CI u represents a characteristic index of the original load demand curve; RCI u represents a characteristic index of the reconstructed load demand curve; w u represents a weight coefficient of the characteristic index; DR represents the gap between the characteristic index of the reconstructed load demand curve and the original load curve characteristic index; The constraint conditions are as follows: 1) The characteristic index of the original load curve CI1 = L ave (4) CI2= L ave / L max (5) CI3 = (L max - L min ) / L max (6) CI4 = L ave,peak / L ave (7) CI5 = L ave,low / L ave (8) CI6 = T max (9) CI7 = T min (10) In the formula: CI1, CI2, CI3, CI4, CI5, CI6, CI7 respectively represent the original load curve daily average load, daily load rate, daily peak valley difference rate, peak period load rate, valley period load rate, peak time, valley time; L ave , L max , L min respectively represent the average value, maximum value, minimum value of daily load; L ave,peak , L ave,low respectively represent the average value of the peak period and the average value of the valley period; T max , T min respectively represent the peak time and the valley time; 2) reconfiguration load curve characteristic index In the formula: RCI1, RCI2, RCI3, RCI4, RCI5, RCI6, RCI7 respectively represent the daily average load, daily load rate, daily peak-valley difference rate, peak period load rate, valley period load rate, peak occurrence time, valley occurrence time of the reconstructed load curve; respectively represent the average, maximum, and minimum values of the reconstructed load curve; Respectively represent the peak the average of the period and the average of the valley period; Respectively represent the peak time and valley time of the reconstructed load curve; 3) Peak-trough periods of reconstituted load curves In the formulae: Respectively represent the value of the reconstructed load curve valley period, flat period and peak period; (4) Use the python-numpy package to build an equal step search method to solve the load reconstruction model in step (3), as follows: Step 1: Find the maximum value L from the load sequence max and the minimum value L min , determine the search step sw, get the equal step search set {r1, r2, …, r a , …, r A}, where r a = L min + sw*(a-1), a is the element sequence number in the search set, and A is the element number in the search set; Step 2: define bl, b2, b3 as the serial number of each element in the search set, let bl = 1, b2 = bl + 1, b3 = b2 + 1, The corresponding reconstructed load curve is obtained, and a load characteristic index and a target value are calculated Let Step 3: Let b1 = b1 + 1 or b2 = b2 + 1 or b3 = b3 + 1 until b1 = A - 2, b2 = A - 1, b3 = A, and find the corresponding objective function value If Let Find the optimal objective function value corresponding to the reconstructed load curve; Step4: Get the optimal objective function value DR and its corresponding reconstructed load curve; (5) Considering the left and right banks, upstream and downstream connections of cascade hydropower stations, and combining short-term scheduling constraints, a short-term nested scheduling model of the basin water, wind and light system is constructed: The maximum power generation objective function is as follows: E = maximum target function of power generation, MWh Pnjd represents the output of the nth power plant j on day d, MW; The outputs (MW) of wind power plants and photovoltaic power plants bundled with hydropower station n and power plant j for power transmission on day d are respectively. n,j,d Let represent the total output of water, wind and solar power bundled and transmitted by hydropower station n and power plant j on day d, in MW; △t represents the number of hours per day, in h; n and N represent the serial number of the hydropower station and the total number of hydropower stations, respectively; j represents the serial number of the power plant, j=1 represents the left bank power plant, j=2 represents the right bank power plant; d, D represent the day number and total days respectively; The power transmission deviation minimum objective function is as follows: In the formula: ED is the objective function for minimizing power transmission deviation, MW 2 ; denotes the power delivery plan of power plant n to province s at time t on day d, MW; ξ s (t) denotes the load demand of province s, using the load reconstruction curve obtained in Section 2; η s,n,j,d denotes the scaling coefficient of the receiving end load demand; s, S represent the province serial number and the total number of provinces respectively; t, T represent the time interval serial number and the total number of time intervals respectively; 2) Constraint conditions In addition to considering the medium-term scheduling and short-term scheduling, the medium-short-term nested power constraints are also considered, i.e. the daily power obtained by the medium-term scheduling should be equal to the sum of the power generated in each time period within the day, as shown in formula (15); (6) On the Gurobi solver platform, use Python program to convert the short-term nested scheduling model of the basin water, wind and light system into a mixed integer linear programming model, and realize efficient solution, as follows: Step 1: The multi-objective model of the maximum power generation objective function and the minimum power transmission deviation objective function is converted into a single objective model by using the constraint conversion method. Specifically, the maximum power generation objective function is converted into a constraint condition of the minimum power transmission deviation objective function, as shown in equation (16); In the formula, E set E represents the total power generation requirement of the river basin in the dispatching period, MWh; E set The initial value of E is the maximum value of the power generation without considering the power transmission deviation target, that is, the optimal value of the objective function of the daily-scale power generation maximum model. Step2: For the one-dimensional nonlinear function relationship of water level and reservoir capacity, piecewise linearization is adopted for processing; for the two-dimensional nonlinear function relationship of water power generation function, parallelogram two-dimensional interpolation method is adopted for linearization processing; Step3: Obtain a set of non-inferior solution set of multi-objective model by gradually relaxing the power generation constraint; Step 4: Calculate the comprehensive benefit index, select the decision scheme from the solution set, and the index calculation formula is shown in equation (17);

Images