A multi-stage stochastic optimization scheduling method and system for a photovoltaic-thermal power system

Abstract

The application discloses a kind of multi-stage random optimization scheduling method and system of photovoltaic-thermal power system, comprising: step one: obtaining unit parameter and new energy output prediction curve;Step two: based on prediction curve, generate at least five kinds of scene, construct the scene tree structure of multi-stage;Step three: on the basis of scene tree structure, with multi-stage random programming model as framework, construct the multi-stage random optimization scheduling model of photovoltaic-thermal power system;Step four: stage 3 is optimized using load aggregation, photovoltaic-thermal unit aggregation two methods, and construct acceleration optimization model;Step five: solve acceleration optimization comprehensive model.The system includes data acquisition module, scene tree construction module, model construction module, model optimization module and model solution module.The application can make up the slow speed of multi-stage random optimization scheduling model solution of photovoltaic-thermal power system, and it has certain significance for large-scale medium and long term power system planning.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power system planning and operation, and particularly relates to a multi-stage random optimization scheduling method and system for a power system containing a photo-thermal power generation. BACKGROUND

[0002] Photo-thermal power generation technology is a new renewable energy generation technology in recent years, which has the advantages of zero carbon emission, excellent controllable characteristics, high energy utilization rate, etc. Large-scale grid connection of photo-thermal power generation can play its role in peak load shifting, effectively reduce carbon emissions, promote new energy consumption, and reduce the operation cost of the power system. According to the different heat collection technologies, photo-thermal power stations can be divided into four types: trough type, tower type, dish type and Fresnel type.

[0003] However, due to the uncertainty of light resources, there are certain difficulties in large-scale grid connection of photo-thermal power generation. In this regard, researchers have proposed a multi-stage random planning model to solve the uncertainty problem of photo-thermal power generation by using a multi-scenario algorithm. The present application takes a three-stage random planning model as a framework to construct a multi-stage random optimization scheduling model for a power system containing photo-thermal power generation. However, due to the large amount of calculation of the multi-scenario algorithm, the long time span involved, the large number of variables, and the complex constraints, the solving time is long and the solving efficiency is low. Load aggregation and photo-thermal unit aggregation as two kinds of accelerated optimization algorithms can reduce the solving scale, combine the state variables of photo-thermal units, and thus improve the solving efficiency. At present, there is no modeling of multi-stage random optimization scheduling of power systems combined with load aggregation and photo-thermal unit aggregation. By combining these two kinds of accelerated optimization algorithms for modeling, the model solving efficiency is improved, which is of great significance for the operation planning analysis of the power system containing photo-thermal power generation. SUMMARY

[0004] The purpose of the present application is to provide a multi-stage random optimization scheduling method and system for a power system containing photo-thermal power generation, to make up for the slow solving speed caused by the multi-stage random optimization scheduling model for a power system containing photo-thermal power generation involving a large number of variables and complex constraints.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] A multi-stage random optimization scheduling method for a power system containing photo-thermal power generation, comprising the following steps:

[0007] Step 1: Obtain the parameters of thermal power units and photo-thermal units in the power system, as well as the predicted curve of new energy output;

[0008] Step 2: Based on the predicted curve of new energy output, at least five scenarios are generated using a scenario generation method to construct a multi-stage scenario tree structure to describe the uncertainty of photo-thermal power generation and wind power generation;

[0009] Step three: on the basis of the scenario tree structure of step two, a multi-stage stochastic programming model is used to construct a three-stage stochastic optimization scheduling model of the power system containing the solar-thermal power generation, combined with the parameters of the thermal power unit and the solar-thermal unit obtained in step one, wherein stage 1 is to perform day-ahead scheduling on the first day, based on the predicted value of the new energy output, the day-ahead start-stop plan of the unit is solved; stage 2 is to use the principle of economic allocation to allocate the unit output and standby of the first day plan based on the scenario value of the new energy output; stage 3 is to perform forward operation on the second day to the seventh day, based on the principle of economic allocation, considering the possible operation conditions in the future few days, the unit output in the future few days and the heat storage amount of the solar-thermal power station are arranged;

[0010] Step four: the optimization methods of load aggregation and solar-thermal unit aggregation are used to optimize stage 3 involving a long time period, and a comprehensive acceleration optimization model is constructed;

[0011] Step five: the comprehensive acceleration optimization model is solved to obtain the day-ahead scheduling plan of the unit and the solving time.

[0012] Further improvement of the present application is that in step two, the scenario generation method adopts the scenario construction based on Weibull distribution.

[0013] Further improvement of the present application is that in step three, on the basis of the scenario tree structure of step two, a multi-stage stochastic programming model is used to construct a three-stage stochastic optimization scheduling model of the power system containing the solar-thermal power generation, and the objective function of the scheduling model is as follows:

[0014] minimize Cost Sys =Cost DA +Cost RT +Cost LA (1)

[0015]

[0016] Wherein t represents the time sequence number of stage 1 and stage 2 from 1 to N T ; k represents the time sequence number of stage 3 from 1 to N K ; g represents the unit sequence number from 1 to N G ; SU represents the start-stop cost calculation matrix of the thermal power unit; 0-1 variable x represents the start-stop state of the thermal power unit; C G represents the operation cost calculation matrix of the thermal power unit; P represents the unit output; C Ru , C Rd respectively represent the deployment cost of the upward / downward rotation standby of the thermal power unit; R Ru , R Rdrespectively represent upward / downward spinning reserve output in operation; s represents from 1 to N s stage 2 scenario number; h represents from 1 to N H stage 3 scenario number; π represents the probability of occurrence of each scenario; with "~" represents the corresponding two-stage variable; with "^" represents the corresponding three-stage variable; δ Voll represents the penalty cost of load rejection; L Cur represents the size of load rejection of the system; i represents from 1 to N I load node number;

[0017] The objective function is to minimize the expected operation cost Cost Sys ; the first part is the day-ahead generation plan deployment cost Cost DA , including the start-stop cost of thermal power units, coal consumption cost, upward / downward spinning reserve plan cost; the second part is the expected real-time allocation cost Cost RT , including the reserve deployment cost of thermal power units and the penalty cost of load rejection; the third part is the expected operation cost Cost LA of the next few days;

[0018] The related constraints of stage 1 are as follows:

[0019] The related constraints of thermal power units are as follows:

[0020]

[0021] wherein, UR g , RD g respectively represent the ramp rate of the unit; Δt represents the unit ramp time; 0-1 variables m, n respectively represent the start-stop action of the unit; respectively represent the minimum start-stop time of the unit; respectively represent the maximum and minimum output limit of the unit;

[0022] Equations (5), (6) represent the ramp constraints of the unit; equations (7), (8) represent the minimum start-stop time constraints of the unit; equations (9), (10) represent the 0-1 variable constraints of the start-stop of the unit; equation (11) represents the upper and lower limit constraints of the output of the unit;

[0023] The related constraints of wind power units are as follows:

[0024]

[0025] wherein, w represents from 1 to N w wind farm number; represents the output of wind farm w at period t; represents the wind curtailment of wind farm w at period t; represents the predicted value of wind power resources in period t;

[0026] Formula (12) represents the output constraint of the wind turbine;

[0027] The relevant constraints for CSP units are as follows:

[0028]

[0029] ES c,min ≤ES c,t ≤ES c,max (17)

[0030] ES c,ini =ES c,0 (18)

[0031]

[0032] Among them, c represents from 1 to N C Serial number of the CSP unit; represents the predicted value of solar energy absorbed by CSP unit c during period t; Energy provided to the electric heating system; To normalize the energy loss in each link to a unified expression of the concentrated solar collector system; The heat input into the heat storage system by the concentrating solar collector system; Release heat from the thermal storage system into the power generation system for power generation; are the maximum input and output heat allowed by the heat storage system; ES c,t represents the heat storage capacity of CSP unit c during period t; η in ,η out are the heat input efficiency and heat output efficiency of the heat storage system respectively; ES c,min , ES c,max Respectively represent the upper and lower limits of heat storage system; ES c,ini , ES c,end represent the initial heat storage and final heat storage of the heat storage system respectively; They are all 0-1 variables, 0 represents shutdown and 1 represents startup; Respectively represent the start and stop status of the steam turbine; It is the initial start-up and shutdown state of the steam turbine; Represents the start and stop action of the steam turbine; Respectively represent the minimum start and stop time of the steam turbine; is the output of turbine c of the power generation system at time t, which includes two parts: the power delivered to the electric heating system and the power supplied to the system α c , β crespectively, are the coefficients related to the thermal-electric conversion efficiency of the steam turbine; represents the efficiency of the power conversion into heat in the working fluid of the electric heating system; represents the heat delivered by the electric heating system into the working fluid; respectively, represent the reserve power of the power generation system; represents the efficiency of the heat conversion into electric energy of the power generation system;

[0033] Equation (13) represents the heat balance constraint of the solar-thermal power generation system; equations (14), (15) represent the upper and lower bound constraints of the heat flow of the thermal storage system; equation (16) represents the dynamic constraint that the heat storage amount of the thermal storage system should satisfy the time variation; equation (17) represents the upper and lower bound constraint of the heat storage amount of the thermal storage system; equations (18), (19) represent the constraints on the initial heat storage amount and the final heat storage amount of the thermal storage system; equations (20)-(23) represent the 0-1 variable constraints that the steam turbine needs to satisfy; equations (24), (25) represent the minimum start-stop time constraints that the steam turbine needs to satisfy during operation; equation (26) can be understood as: is the heat flowing into the unit, represents the heat consumed when the unit is started, and the difference between the two is the heat actually used by the steam turbine for power generation, multiplied by the efficiency coefficient c is the actual output of the unit; equation (27) represents that the output of the solar-thermal unit can be divided into two parts: is the tidal flow input into the system, is the power supplied to the electric heating system, is the power flowing into the electric heating system is converted into heat energy in the heat-conducting working fluid with an efficiency ; equation (28) represents the upper and lower bound constraints of the output of the solar-thermal unit, as well as the upper and lower bound constraints of the power of the electric heating system; equation (29) represents that when the steam turbine is in the power generation state, the upward reserve provided by the power generation system does not exceed and the available heat release amount interval of the thermal storage system; equation (30) represents that the upward reserve provided by the electric heating system does not exceed its expected output, and similarly, the downward reserve provided by the power generation system and the electric heating system is constrained by equations (31), (32);

[0034] The relevant constraints for the system are as follows:

[0035]

[0036] wherein i represents the load node sequence number from 1 to N I ; D i,t represents the load of node i at time t; Gi(t) represents the size of the load shedding at node i at time period t; GSTrepresents the transfer factor matrix of thermal power units' output to lines; GSWrepresents the transfer factor matrix of wind power units' output to lines; GSCrepresents the transfer factor matrix of solar-thermal units' output to lines; GSWrepresents the transfer factor matrix of node load to lines; F l max represents the maximum value of the allowed power flow in the forward and reverse directions of the line;

[0037] Equation (33) represents that the output and load of the system are always balanced; equation (34) represents that the power flow in the forward and reverse directions of each branch cannot exceed the upper limit of the allowed power flow of the branch; equations (35) and (36) represent that the reserve provided by the thermal power unit, the solar-thermal unit and the electric heating system needs to meet the minimum upward / downward reserve requirement of the system;

[0038] The relevant constraints of stage 2 are as follows:

[0039] The relevant constraints of the thermal power unit are as follows:

[0040]

[0041] Equations (37) and (38) represent the ramping constraints that the output of the thermal power unit in stage 2 needs to meet; equation (39) represents the upper and lower limit constraints of the unit output; equation (40) represents that the output of the thermal power unit in stage 2 is equal to the sum of the output in stage 1 and the upward / downward spinning reserve put into operation; equation (41) represents that the output of the upward / downward spinning reserve put into operation in stage 2 does not exceed the spinning reserve capacity deployed in stage 1;

[0042] The relevant constraints of the wind power unit are as follows:

[0043]

[0044] wherein, is the wind power resource value corresponding to scenario s; equation (42) has the same meaning as equation (12) in stage 1, which will not be described here;

[0045] The relevant constraints of the solar-thermal unit are as follows:

[0046]

[0047] wherein, is the solar-thermal resource value corresponding to scenario s;

[0048] The operation state of the power generation system is represented by formula (53) determined in stage 1; formula (54), (56) represent that the output distribution of the unit is equal to the sum of the planned output before the day and the real-time deployed rotating standby; formula (55), (57) represent that the real-time standby distribution should be limited in the feasible range, that is, the standby output of the power generation system and the electric heater system put into operation does not exceed the planned deployed standby obtained in stage 1; other constraints are the same as stage 1, only the expression form of stage 2 variables is replaced;

[0049] The relevant constraints of the system are as follows:

[0050]

[0051] The relevant constraints of stage 3 are as follows:

[0052] The relevant constraints of the thermal power unit are as follows:

[0053]

[0054] The relevant constraints of the wind power unit are as follows:

[0055]

[0056] The relevant constraints of the solar thermal power station are as follows:

[0057]

[0058]

[0059] The relevant constraints of the system are as follows:

[0060]

[0061] In summary, the objective functions (1)-(4) and the constraints (5)-(85) constitute a multi-stage stochastic optimization scheduling model of a power system containing a solar thermal power station.

[0062] Further improvement of the application is that in step four, the stage 3 model established in step three is modified by using a load aggregation optimization algorithm, and the stage 3 model modified by using the load aggregation method is as follows:

[0063] The objective function of stage 3 is changed to:

[0064]

[0065] Wherein, v represents the time sequence number of the load aggregation, and the range is 1 to N V ;Y v represents the length of the time sequence section v of the LSTC curve;

[0066] The constraints related to thermal power units are as follows:

[0067]

[0068] When the operating state of a thermal power unit changes, the power is limited to The power of the same time sequence segment of the aggregated load curve is limited to the same value, which is difficult to accurately describe the above situation, so the weighted average formula (87) is used to approximate the power of the time sequence segment where the start-stop action occurs to In the formula, Y ave represents the average value of the length of each time sequence segment, which is calculated by formula (88); Y ave is the minimum start-up time constraint, and the minimum shutdown time constraint is determined by formula (91) to (94); the minimum start-up time constraint (91) is taken as an example, where the parameter takes the minimum value that satisfies formula (93), which means the minimum number of time sequence load segments required for the unit to maintain the on-state to meet the minimum start-up time constraint, and the parameter is determined in the same way;

[0069] The constraints related to wind power units are as follows:

[0070]

[0071] The constraints related to solar thermal power stations are as follows:

[0072]

[0073]

[0074] Among them, formula (110)-(113) is modified by analogy with the minimum start-up and shutdown time constraint of thermal power units, which will not be repeated here;

[0075] The constraints related to the system are as follows:

[0076]

[0077] After the original time sequence load curve is aggregated, the renewable energy resource scenario output curve should also be changed accordingly to meet the calculation needs. Take the wind power resource scenario curve as an example to illustrate: the original curve is segmented according to the length of each time sequence segment Y v , and the time sequence segment v corresponds to Y vThe value corresponding to each time is considered to be equally likely to occur in this time segment, and the value corresponding to each time is summed up and averaged as the value of the time segment v, i.e. The value corresponding to each time is considered to be equally likely to occur in this time segment, and the value corresponding to each time is summed up and averaged as the value of the time segment v, i.e. The photothermal resource data is also subjected to similar segmentation processing.

[0078] Further improvement of the present application is that in step four, the stage 3 model established in step three is subjected to accelerated optimization algorithm of photothermal unit aggregation, and the model solving is subjected to accelerated optimization; specifically, the operation state variable of the photothermal unit in the original stage 3 model And the unit start-stop action variable All adopt integer variable, the value is 0, 1, 2, …, N C , N C Is the number of units of this type, indicating the number of units in the running state, starting action and stopping action in class c; based on the original stage 3 model, the thermal power and wind power unit constraints remain unchanged, the photothermal unit is aggregated, and the related constraints of the photovoltaic power station are modified as follows:

[0079]

[0080] Among them, for the aggregated unit class c, the parameters of the heat storage link such as heat storage capacity, minimum and maximum power are the sum of the parameters of the units in the same class.

[0081] Further improvement of the present application is that in step four, the model of stage 3 is modified using two kinds of accelerated optimization algorithms, and the objective functions (1)-(3), (86), stage 1 constraints (5)-(36), stage 2 constraints (37)-(59), stage 3 constraints (87)-(98), (117)-(134) jointly constitute a multi-stage random optimization scheduling accelerated optimization comprehensive model of a photothermal power generation power system.

[0082] A multi-stage random optimization scheduling system of a photothermal power generation power system, comprising:

[0083] A data acquisition module acquires the parameters of thermal power units and photothermal units in a power system, and a predicted curve of new energy output;

[0084] A scenario tree construction module generates at least five scenarios using a scenario generation method based on the predicted curve of new energy output, and constructs a multi-stage scenario tree structure to describe the uncertainty of photothermal power generation and wind power generation;

[0085] A model construction module, on the basis of the scene tree structure, adopts a multi-stage stochastic programming model, combines the parameters of the thermal power unit and the solar-thermal unit, and constructs a three-stage stochastic optimization scheduling model of the solar-thermal power generation system, wherein stage 1 is to perform day-ahead scheduling on the first day, based on the predicted value of the new energy output, to solve the day-ahead start-stop plan of the unit; stage 2 is to use the principle of economic allocation to allocate the unit output and standby of the first day plan based on the scenario value of the new energy output; and stage 3 is to perform forward operation on the second day to the seventh day, and based on the principle of economic allocation, the possible operation conditions of the future days are considered to arrange the unit output of the future days and the heat storage amount of the solar-thermal power station heat storage system.

[0086] A model optimization module uses two acceleration optimization methods of load aggregation and solar-thermal unit aggregation to optimize the stage 3 involving a long time period, and constructs an acceleration optimization comprehensive model.

[0087] A model solution module solves the acceleration optimization comprehensive model to obtain the day-ahead scheduling plan of the unit and the solution time.

[0088] A computer readable storage medium stores a computer program, and the computer program, when executed by a processor, implements the steps of the multi-stage stochastic optimization scheduling method of the solar-thermal power generation system.

[0089] Compared with the prior art, the present application has at least the following beneficial technical effects:

[0090] The multi-stage stochastic optimization scheduling method and system of the solar-thermal power generation system provided by the present application are based on the multi-stage stochastic programming scheduling model of the solar-thermal power generation system, fully consider the operation characteristics of the solar-thermal units in the same solar-thermal park, aggregate the solar-thermal units with the same or similar parameters in the same park into a category, simplify the variables and constraints at the category level, and further combine the same or similar load levels to reduce the time scale. The present application reduces the number of variables, simplifies the constraint conditions, reduces the solution scale, greatly improves the solution speed of the model, maintains the time sequence characteristics of the model as much as possible, can accurately calculate the total cost, and formulates the day-ahead operation plan of various units, thereby providing an efficient solution tool for the long-term time sequence production simulation of the solar-thermal power generation system. BRIEF DESCRIPTION OF DRAWINGS

[0091] Figure 1 The overall flowchart of the present application.

[0092] Figure 2 The three-stage scene tree structure diagram.

[0093] Figure 3Single-scenario wind power resource curve.

[0094] Figure 4 Single-scenario solar thermal resource curve.

[0095] Figure 5 Load curve after load aggregation when the number of load states is 10.

[0096] Figure 6 Solution error of load aggregation of different number of load states.

[0097] Figure 7 Solution error of accelerated optimization integrated model of different number of load states.

[0098] Figure 8 Day-ahead unit output planning corresponding to the exact solution.

[0099] Figure 9 Day-ahead output and heat storage plan of the solar thermal power station corresponding to the exact solution.

[0100] Figure 10 Day-ahead unit output planning solved by the accelerated optimization integrated model.

[0101] Figure 11 Day-ahead output and heat storage plan of the solar thermal power station solved by the accelerated optimization integrated model.

[0102] Figure 12 Solution error of the accelerated optimization integrated model corresponding to different number of load states.

[0103] Figure 13 The structure block diagram of a kind of multi-stage stochastic optimization scheduling system of photothermal power system of the present application. DETAILED DESCRIPTION

[0104] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0105] As Figure 1 shown, the present application provides a kind of multi-stage random optimization scheduling method of photothermal power system, comprising the following steps:

[0106] Step one: obtain the parameters of thermal power units, photo-thermal units in the power system, and the predicted curve of new energy output. The present application takes the power system containing thermal power generation, photo-thermal power generation, and wind power generation as an example for introduction, and the predicted curve of new energy output mainly includes the predicted curves of wind power resources and photo-thermal resources. In actual application, other types of power generation forms such as photovoltaic power generation and hydroelectric power generation can be included, and the method used can be extended for use.

[0107] The example is based on the improved 168h scheduling plan of IEEE-RTS79 system, which includes a total of 26 thermal power units with installed capacity ranging from 12MW to 400MW, 5 wind farms, and 10 photo-thermal units. The predicted value curve of new energy is from the statistical data of a certain province in northwest China.

[0108] Step two: based on the original renewable energy output prediction curve, a certain number of scenes are generated using scene generation methods (such as constructing scenes based on Weibull distribution), and a multi-stage scene tree structure is constructed to describe the uncertainty of new energy output such as photo-thermal power generation.

[0109] The present application takes the three-stage scene tree result as an example for introduction, and the three-stage scene tree structure diagram is as shown in Figure 2

[0110] Step three: based on the scene tree structure of step two, a multi-stage stochastic programming model is used to construct a multi-stage stochastic optimization scheduling model of the photo-thermal power system. The present application takes the three-stage stochastic programming model as an example for description, and in actual application, the corresponding total number of stages can be selected as needed. Stage 1 is to perform day-ahead scheduling for the first day, based on the predicted curve of new energy output, to solve the day-ahead start-stop plan of the unit; stage 2 is to use the principle of economic allocation to allocate the unit output and standby of the first day plan based on the scene value of new energy output; stage 3 is to perform forward operation from the second day to the seventh day, and based on the principle of economic allocation, the possible operation conditions of the next few days are considered to arrange the unit output of the next few days and the heat storage amount of the photo-thermal power station. If the time period is longer, stages 4, 5, etc. can be added for modeling according to needs;

[0111] Step four: for stage 3 involving a longer time period, two acceleration optimization methods of load aggregation and photo-thermal unit aggregation are used for optimization to construct an acceleration optimization comprehensive model and improve the model solving efficiency.

[0112] ​Load aggregation refers to that on the basis of preserving the time sequence characteristics of the original time sequence load curve as much as possible, a certain clustering method (such as k-means clustering) is adopted, based on the specified load state number, the load values with the same or similar load level on the original time sequence load curve are combined into the same load level, then the actual load values on the original time sequence load curve are replaced by the clustered load levels, and a plurality of time periods with similar load levels with a length of 1 are combined into a time period with a length of several, thus the load aggregation is realized.

[0113] Compared with the initial model in stage 3, the method using load aggregation combines the same or similar load levels, and combines the time periods with the same load level after combination into a new large time period, thereby greatly reducing the calculation time scale of the stage 3 model, and at the same time, the corresponding processing is supplemented for the new energy scene output, and the corresponding modification is made for the constraints, thereby realizing the improvement of the solving efficiency.

[0114] Photothermal unit aggregation refers to that according to the basic assumption of unit aggregation research, it is considered that after the photothermal unit aggregation, the unit parameters in the same class are consistent, the unit power at the same time in the same class in the same state is the same, and the operation of the entire class is represented only by the number of units in the class in each state, and the units in the class are not distinguished in detail.

[0115] Compared with the model established in the initial stage 3, the photothermal unit aggregation modeling combines the running state variables and the start-stop action variables from 0-1 variables into integer variables, increases part of the continuous variables, and modifies the values of part of the parameters, realizes the approximate description of the operation process of a class of photothermal units, reduces the number of variables and constraints in the stage 3 model, and achieves the purpose of improving the solving efficiency.

[0116] In this step, only stage 3 is modified by using the accelerated optimization algorithm, and the constraints of stages 1 and 2 are not modified, because the constraints of stages 1 and 2 only involve the operation plan of the first day, and the variables and constraints contained are relatively few, which will not occupy too much solving time, while stage 3 involves a long subsequent time period and is a mixed integer linear programming, and the solving rate is relatively low, therefore, the modification of stage 3 by using the accelerated optimization algorithm can improve the calculation rate, and the operation plan of the first day solved will be put into use, if the aggregation method is used in stages 1 and 2, the accuracy of the solving result may be affected to a certain extent, while the solving result of stage 3 is only used to estimate the total operation cost and adjust the operation plan of the first day, and is not put into use, therefore, the accelerated optimization algorithm can be used for modification to improve the solving efficiency. If the model constructed in the actual use of this patent is more than three stages, stages 1 and 2 should also be accurately solved, and the accelerated optimization algorithm should be used in the subsequent time period with a long time scale, so as to improve the model solving speed on the premise of ensuring the accuracy of the result.

[0117] Step five: solve the accelerated optimization integrated model to obtain the unit day-ahead scheduling plan and the solving time, and verify the solving efficiency and accuracy of the accelerated optimization integrated model.

[0118] The following standards are used to evaluate the acceleration effect of the aggregation method. The results obtained by the aggregation method are denoted by a superscript “*”, and otherwise, the results are obtained without using the aggregation method.

[0119] (1) System generation cost error:

[0120]

[0121] (2) Generation power error:

[0122]

[0123] (3) Number of on-off states error:

[0124]

[0125] (4) Acceleration ratio:

[0126] The ratio of the solving time of the original model to the solving time of the model using the accelerated optimization algorithm is defined as the acceleration ratio, which is used to evaluate the acceleration effect of the accelerated optimization algorithm.

[0127]

[0128] First, in order to verify the advantages of the present application in solving accuracy and speed, while reducing the amount of calculation required for verification as much as possible and improving the verification efficiency, a single-scenario model is introduced. The single-scenario model refers to extracting a certain wind power and solar-thermal resource scenario from the multi-scenario model in step three to solve the three-stage stochastic programming model, and the remaining parameters are the same. The data of this single scenario are calculated to verify the acceleration effect and accuracy of the accelerated optimization algorithm, and to reduce the verification scale and improve the verification efficiency. The values of the wind power and solar-thermal resource of a single scenario are shown in Figure 3 , Figure 4 .

[0129] In order to compare the advantages of the present application, a simplified model is introduced for comparison. One of the reasons for the long solving time of the stochastic optimization scheduling model of the solar-thermal power system is that stage 3 involves a long time period, a large number of variables, and the three-stage itself is a mixed integer linear programming operation. A large number of 0-1 variables greatly reduce the solving efficiency, and the stage 3 scheduling plan itself is only used to estimate the overall cost and does not enter operation. Therefore, a simplified model is introduced to simplify the original constraints to constraints without 0-1 variables, in order to improve the solving efficiency and reduce the running time. Therefore, the solving results of this model are used as a set of comparison results.

[0130] A single-scenario model is used, and the solution of the stochastic optimization scheduling model containing the CSP power system in step 3 is used as the exact solution. The results obtained by solving the simplified model, only load aggregation, only CSP unit aggregation, and the accelerated optimization comprehensive model proposed in this invention are compared. The results are shown in Table 1. Figure 5 This is the result obtained after processing the time series load curve when the number of load states is set to 10.

[0131] Table 1 Comparison of verification results of single scenario examples

[0132]

[0133] Based on the comparison of the above results, we can find that: 1) Since the simplified model does not consider the 0-1 variable constraint in stage 3, the running time is greatly reduced, but the results obtained have larger errors compared with the other three methods; 2) The accelerated optimization comprehensive model uses a combination of two accelerated optimization algorithms, resulting in the largest speedup ratio, that is, the best acceleration effect. However, compared with the other two results, the error is also slightly larger, but the accuracy is still within an acceptable range, which is equivalent to slightly sacrificing accuracy in exchange for a significant improvement in solution efficiency.

[0134] Further study on the effect of load state number on load aggregation effect, the load state number is 7-13, the calculation results of load aggregation and acceleration optimization comprehensive model are as follows: Figure 6 、 Figure 7 shown.

[0135] from Figure 6 、 Figure 7 It can be found that for load aggregation, when the number of load states is small, the solution time is short, but the error is large; and as the number of load states gradually increases, the error of the solution result gradually decreases, the accuracy gradually improves, but the corresponding solution time will also increase.

[0136] By comparing the above results, the acceleration effect and solution accuracy of the accelerated optimization comprehensive model that simultaneously adopts load aggregation and CSP unit aggregation are demonstrated. It is applied to the random optimization scheduling model of the power system containing CSP proposed in step 3, and compared with the solution result without adopting the accelerated optimization algorithm as the exact solution. Figure 8 、 Figure 9 In order to accurately solve the corresponding unit output planning, output and heat storage plan of the solar thermal power station, Figure 10 、 Figure 11 Table 2 shows the results of the accelerated optimization comprehensive model for solving the unit output planning, output and heat storage plan of the solar thermal power station at the same time. Figure 12Solving error of the acceleration optimization comprehensive model corresponding to different load state numbers.

[0137] Table 2 solving result of the acceleration optimization comprehensive model

[0138]

[0139] From the above data, it can be seen that the acceleration optimization comprehensive model can shorten the solving time to one fifth to one tenth of the original solving method while maintaining high accuracy, and the acceleration effect is excellent. This good acceleration advantage will be more significant in solving longer and larger scale joint operation scheduling models, and has certain significance for efficient long-term planning of large-scale power systems.

[0140] As shown in Figure 13 The application provides a multi-stage random optimization scheduling system of a photothermal power system, which comprises:

[0141] A data acquisition module acquires parameters of thermal power units and photothermal units in the power system and a predicted curve of new energy output;

[0142] A scenario tree construction module generates at least five scenarios based on the predicted curve of new energy output using a scenario generation method, and constructs a multi-stage scenario tree structure to describe the uncertainty of photothermal power generation and wind power generation;

[0143] A model construction module, on the basis of the scenario tree structure, adopts a multi-stage random programming model, combines the parameters of the thermal power units and the photothermal units, and constructs a three-stage random optimization scheduling model of the photothermal power system, wherein stage 1 is day-ahead scheduling for the first day, the day-ahead start-stop plan of the units is solved based on the predicted value of the new energy output; stage 2 is real-time allocation of the unit output and standby of the first day based on the scenario value of the new energy output using the principle of economic allocation; and stage 3 is prospective operation for the second day to the seventh day, the unit output of the future few days and the heat storage amount of the photothermal power station heat storage system are arranged based on the principle of economic allocation and considering the possible operation conditions of the future few days;

[0144] A model optimization module adopts two acceleration optimization methods of load aggregation and photothermal unit aggregation to optimize stage 3 involving a long time period, and constructs an acceleration optimization comprehensive model;

[0145] A model solving module solves the acceleration optimization comprehensive model to obtain a unit day-ahead scheduling plan and a solving time.

[0146] The application provides a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer program realizes steps of the method for multi-stage random optimization scheduling of the solar-thermal power generation system when being executed by a processor.

[0147] Those skilled in the art will appreciate that embodiments of the application can be supplied as methods, systems, or computer program products. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk memory, CD-ROMs, optical storage media, etc.) having computer usable program code embodied therein.

[0148] The application is described with reference to flowcharts and / or block diagrams of methods, systems and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, a special purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions, which are executed via the processor of the computer or other programmable data processing apparatus, generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 a system to perform the functions specified in the flowcharts and / or block diagrams.

[0149] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufacture product including an instruction device, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 a system to perform the functions specified in the flowcharts and / or block diagrams.

[0150] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 a system to perform the functions specified in the flowcharts and / or block diagrams.

[0151] It should be pointed out that the above examples are only used to illustrate the technical solutions of the present application, and the protection scope of the present application is not limited thereto, and any skilled person in the art can make equivalent replacements or changes according to the technical solutions and the inventive concept of the present application within the scope disclosed by the present application, which shall all fall into the protection scope of the present application.

Claims

1. A method for multi-stage stochastic optimization scheduling of a solar-thermal power system, characterized in that, The method comprises the following steps: Step 1: obtaining parameters of thermal power units and solar-thermal units in a power system and a predicted curve of new energy output; Step 2: generating at least five scenes based on the predicted curve of new energy output by using a scene generation method, and constructing a multi-stage scene tree structure to describe the uncertainty of solar-thermal power generation and wind power generation; Step 3: based on the scene tree structure in Step 2, a three-stage stochastic optimization scheduling model of the power system containing solar-thermal power generation is constructed by using a multi-stage stochastic programming model and combining the parameters of the thermal power units and the solar-thermal units obtained in Step 1, wherein stage 1 is day-ahead scheduling for the first day, the day-ahead start-stop plan of the units is solved based on the predicted value of new energy output; stage 2 is real-time distribution of the unit output and standby of the first day plan based on the scene value of new energy output by using the principle of economic distribution; stage 3 is prospective operation from the second day to the seventh day, the unit output of the future days and the heat storage amount of the solar-thermal power station heat storage system are arranged by considering the possible operation conditions of the future days based on the principle of economic distribution; the objective function of the scheduling model is as follows: minimize Cost Sys = Cost DA + Cost RT + Cost LA (1) where t represents from 1 to N T Stage 1, Stage 2 time sequence number; k represents from 1 to N K Stage 3 time sequence number; g represents from 1 to N G Unit sequence number; SU represents the start-stop cost calculation matrix of the thermal power unit; 0-1 variable x represents the start-stop state of the thermal power unit; C G represents the operation cost calculation matrix of the thermal power unit; P represents the output of the thermal power unit; C Ru , C Rd respectively represent the deployment cost of the thermal power unit upward / downward rotating standby; R Ru , R Rd respectively represent the upward / downward rotating standby output put into operation; s represents from 1 to N s Stage 2 scenario sequence number; h represents from 1 to N H Stage 3 scenario sequence number; π represents the probability of occurrence of each scenario; with "~" represents the corresponding two-stage variable; with "^" represents the corresponding three-stage variable; δ Voll represents the penalty cost of abandoned load; L Cur represents the size of the abandoned load of the system; i represents from 1 to N I Load node sequence number; The objective function is to minimize the expected operation cost Cost Sys ; the first part is the day-ahead generation schedule deployment cost Cost DA , including the start-up / shut-down cost of thermal power units, coal consumption cost, and up / down spinning reserve schedule cost; the second part is the expected real-time allocation cost Cost RT , including the reserve deployment cost of thermal power units and the penalty cost of load shedding; and the third part is the expected operation cost Cost LA of the next few days; Step 4: the optimization method of load aggregation and solar-thermal unit aggregation is used to optimize stage 3 involving a long time period, and a comprehensive acceleration optimization model is constructed; Step 5: the comprehensive acceleration optimization model is solved to obtain the day-ahead scheduling plan of the units and the solving time.

2. The method of claim 1, wherein, In Step 2, the scene generation method adopts Weibull distribution to construct scenes.

3. A multi-stage stochastic optimization scheduling system for a solar-thermal power generation power system, characterized in that, The method comprises the following steps: A data acquisition module is configured to obtain parameters of thermal power units and solar-thermal units in a power system and a predicted curve of new energy output; A scene tree construction module is configured to generate at least five scenes based on the predicted curve of new energy output by using a scene generation method, and construct a multi-stage scene tree structure to describe the uncertainty of solar-thermal power generation and wind power generation; A model construction module is configured to construct a three-stage stochastic optimization scheduling model of the power system containing solar-thermal power generation based on the scene tree structure by using a multi-stage stochastic programming model and combining the parameters of the thermal power units and the solar-thermal units, wherein stage 1 is day-ahead scheduling for the first day, the day-ahead start-stop plan of the units is solved based on the predicted value of new energy output; stage 2 is real-time distribution of the unit output and standby of the first day plan based on the scene value of new energy output by using the principle of economic distribution; stage 3 is prospective operation from the second day to the seventh day, the unit output of the future days and the heat storage amount of the solar-thermal power station heat storage system are arranged by considering the possible operation conditions of the future days based on the principle of economic distribution; the objective function of the scheduling model is as follows: minimize Cost Sys = Cost DA + Cost RT + Cost LA (1) where t represents from 1 to N T Stage 1, Stage 2 time sequence number; k represents from 1 to N K Stage 3 time sequence number; g represents from 1 to N G Unit sequence number; SU represents the start-stop cost calculation matrix of the thermal power unit; 0-1 variable x represents the start-stop state of the thermal power unit; C G represents the operation cost calculation matrix of the thermal power unit; P represents the output of the thermal power unit; C Ru , C Rd respectively represent the deployment cost of the upward / downward rotating standby of the thermal power unit; R Ru , R Rd respectively represent the upward / downward rotating standby output in operation; s represents from 1 to N s Stage 2 scenario sequence number; h represents from 1 to N H Stage 3 scenario sequence number; π represents the probability of occurrence of each scenario; with "~" represents the corresponding two-stage variable; with "^" represents the corresponding three-stage variable; δ Voll represents the penalty cost of abandoned load; L Cur represents the size of the abandoned load of the system; i represents from 1 to N I Load node sequence number; The objective function is to minimize the expected operation cost Cost Sys ; the first part is the day-ahead generation schedule deployment cost Cost DA , including the start-up / shut-down cost of thermal power units, coal consumption cost, and up / down spinning reserve schedule cost; the second part is the expected real-time allocation cost Cost RT , including the reserve deployment cost of thermal power units and the penalty cost of load shedding; and the third part is the expected operation cost Cost LA of the next few days; A model optimization module is configured to use the optimization method of load aggregation and solar-thermal unit aggregation to optimize stage 3 involving a long time period, and construct a comprehensive acceleration optimization model; A model solving module is configured to solve the comprehensive acceleration optimization model to obtain the day-ahead scheduling plan of the units and the solving time.

4. The multi-stage stochastic optimization scheduling system for a concentrated solar power system of claim 3, wherein, In the scene tree construction module, the scene generation method adopts Weibull distribution to construct scenes.

5. A computer readable storage medium, characterized in that, The computer readable storage medium stores a computer program, which, when executed by the processor, implements the steps of the multi-stage stochastic optimization scheduling method of the solar-thermal power generation system according to claim 1 or 2.

Images