Multi-objective energy combination optimization scheduling method and system based on model predictive control

Abstract

The present invention discloses a multi-objective energy combination optimization scheduling method based on model predictive control, including: establishing an electric, thermal and gas combination model; constructing the objective function and constraint conditions of the multi-energy combination model according to the electric, thermal and gas combination model; performing rolling prediction on the output scheduling return of distributed renewable energy based on a prediction algorithm; calculating the expectation and variance of the output scheduling return according to the output scheduling return at the terminal moment of the prediction time domain obtained by using the rolling prediction, solving the optimal scheduling of the multi-energy combination model based on a multi-objective genetic algorithm, and outputting the optimal scheduling reference value. The present invention can better plan the optimal energy combination scheduling scheme, while maximizing the benefits of various energy optimization scheduling, minimizing the risks generated by scheduling, and improving the consumption of renewable energy while ensuring safe scheduling.

Description

Technical Field

[0001] The present invention belongs to the technical field of integrated energy system optimization, and specifically relates to a multi-objective energy combination optimization scheduling method and system based on model predictive control. Background Art

[0002] Since the output of renewable energy is not stable, such as wind energy and light energy are intermittent, it is not easy to adjust the output of renewable energy to change according to specific loads. Therefore, an energy supply system that can flexibly adapt to intermittent renewable energy will become very important.

[0003] So far, with the vigorous development of multi-energy power generation, conversion and storage equipment, the coupling and interaction of multiple energy sources such as electricity, heat and gas have become increasingly close, and multi-energy combination optimization has also developed rapidly. However, the existing multi-energy combination optimization methods cannot respond to changes in user needs and wind and solar power combination energy efficiently and quickly, resulting in a large amount of energy waste. Summary of the invention

[0004] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a multi-objective energy combination optimization scheduling method and system based on model predictive control for the combination optimization scheduling between renewable energy sources such as light and wind and other forms of energy. The method can better plan the corresponding optimal energy combination optimization scheduling plan, while maximizing the benefits generated by the optimization scheduling of various energy sources, minimizing the risks generated by the scheduling, and improving the consumption of renewable energy while ensuring safe scheduling.

[0005] On the one hand, the present disclosure provides a technical solution for a multi-objective energy combination optimization scheduling method based on model predictive control:

[0006] A multi-objective energy combination optimization scheduling method based on model predictive control includes the following steps:

[0007] Establish an electric, thermal and gas combination model;

[0008] According to the electric, thermal and gas combination model, construct the objective function and constraint conditions of the multi-energy combination model;

[0009] Based on the prediction algorithm, a rolling prediction is made for the output dispatch return of distributed renewable energy. According to the output dispatch return at the terminal moment in the predicted time domain obtained by the rolling prediction, the expectation and variance of the output dispatch return are calculated, and the optimal dispatch of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal dispatch reference value is output.

[0010] A further technical solution, the objective function includes a benefit objective function and a risk objective function; the benefit objective function is the expected maximum value of the energy benefit of the multi-energy combination model at the terminal moment of the prediction time domain; the risk objective function is the minimum value of the variance of the energy benefit of the multi-energy combination model at the terminal moment of the prediction time domain.

[0011] A further technical solution, the constraints include energy combination benefit constraints, energy supply and demand balance constraints, wind turbine constraints, solar cogeneration system constraints, cogeneration unit constraints, electric boiler constraints, heat storage constraints, electricity storage constraints, and power-to-gas operation constraints.

[0012] A further technical solution is that the rolling prediction of the output dispatch return of distributed renewable energy based on the prediction algorithm means that according to the current energy output, the energy income at different times in the prediction time domain is solved based on the model predictive control algorithm, specifically:

[0013]

[0014] Among them, W(k) is the current k-time profit, R T (k) is the energy output investment return ratio at time k, U(k) is the energy output investment ratio of each energy at time k, c is the energy cost ratio, It is the income of energy participating in carbon emission trading.

[0015] A further technical solution is to solve the optimal scheduling of the multi-energy combination model based on a multi-objective genetic algorithm, which specifically includes the following steps:

[0016] For each moment in the prediction range, a group of energy combinations with different energy outputs are randomly generated according to the even distribution. Each energy combination is regarded as an individual. The parent population is selected according to the dominance relationship and crowding degree between individuals. The child population is obtained through selection, crossover and mutation operations.

[0017] The parent population is merged with the obtained child population, and a fast non-dominated sort is performed between individuals. According to the dominance relationship and crowding degree between individuals, multiple individuals are selected to form a new parent population.

[0018] If the number of generations of the new parent population is less than the set maximum number of generations, a new child population is generated again through the selection operation, crossover operation and mutation operation of the genetic algorithm, and the above merging and population selection operations are repeated; otherwise, the parent population is output as the Pareto frontier of the multi-objective optimization problem in the next prediction time domain.

[0019] According to a further technical solution, the output of the optimal scheduling reference value refers to outputting the optimal point of the Pareto front as the optimal scheduling reference value based on the Pareto front of the solved optimal scheduling problem.

[0020] Another aspect of the present disclosure provides a technical solution for a multi-objective energy combination optimization scheduling system based on model predictive control:

[0021] A multi-objective energy combination optimization scheduling system based on model predictive control, comprising:

[0022] Multi-energy combination model building module, used to build an electricity, heat and gas combination model;

[0023] An objective function and constraint condition construction module, used to construct the objective function and constraint conditions of the multi-energy combination model according to the electric, thermal and gas combination model;

[0024] The optimal scheduling solution module is used to make rolling predictions on the output scheduling returns of distributed renewable energy based on the prediction algorithm; according to the output scheduling returns at the terminal moment of the predicted time domain obtained by the rolling prediction, the expectation and variance of the output scheduling returns are calculated, and the optimal scheduling of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal scheduling reference value is output.

[0025] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0026] The method disclosed in the present invention regards the output of each energy source as an asset in a multi-energy combination. Different energy sources have different asset returns corresponding to their output. In the idea of ​​optimizing the multi-energy combination, the expectation and variance of the asset returns of different energy outputs are used as the objective function of the multi-objective energy combination model predictive control. Then, the constraints of different energy outputs are added, and a multi-objective genetic algorithm is used to solve the multi-energy optimization scheduling problem, and the Pareto frontier at the next prediction moment is obtained. The optimal point of the Pareto frontier is selected and applied to the system, and the operation is repeated for rolling optimization.

[0027] The multi-objective energy combination optimization scheduling method based on model predictive control disclosed in the present invention can better plan the corresponding optimal energy combination scheduling plan, maximize the benefits of various energy optimization scheduling while minimizing the risks generated by scheduling, and improve the consumption of renewable energy while ensuring safe scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0029] Figure 1This is a flow chart of the multi-objective energy combination optimization scheduling method described in Example 1 of the present invention;

[0030] Figure 2 This is a schematic diagram of the model predictive control method described in Example 1 of the present invention;

[0031] Figure 3 This is a flow chart of the multi-objective genetic algorithm described in Example 1 of the present invention. DETAILED DESCRIPTION

[0032] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0033] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used in the present disclosure have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0034] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0035] Example 1

[0036] Aiming at the problem that the existing energy combination optimization scheduling methods cannot respond efficiently and quickly to the changes of user demand and wind and solar power combination energy, resulting in a large amount of energy waste, the present invention proposes a multi-objective energy combination optimization scheduling method and system based on model predictive control, which regards different energy network output conditions as assets in investment, and different energy outputs correspond to different asset returns (i.e., returns). Considering the power grid, heating grid and natural gas grid of renewable energy power generation including wind power and solar power, each form of energy output has specific costs, returns and risks (volatility of returns), the idea of ​​portfolio investment is used, the expectation and variance of energy returns are considered, and the idea of ​​model predictive control is added. At the terminal of the prediction time domain, the expected return is maximized while minimizing the risk of energy return (return variance), and finally a multi-objective genetic algorithm is used to solve the Pareto frontier of the energy scheduling problem, and the optimal point of the Pareto frontier is applied to the scheduling system to obtain the optimal investment portfolio scheduling plan.

[0037] Among them, model predictive control (MPC) is a control method that approaches infinite-time optimal control by solving finite-time suboptimal control problems at each sampling moment. Model predictive control can be well applied to energy portfolio scheduling problems because MPC can handle the complex dynamics of the system well, perform energy portfolio scheduling under uncertainty, and optimize expected returns and risks within the forecast range. Multi-objective optimization is used because it can simultaneously consider risk minimization and return maximization, and provide a Pareto frontier that contains the best solution within the return and risk range.

[0038] This embodiment discloses a multi-objective energy combination optimization scheduling method based on model predictive control, such as Figure 1 As shown, the following steps are included:

[0039] Establish an electric, thermal and gas combination model;

[0040] According to the electric, thermal and gas combination model, construct the objective function and constraint conditions of the multi-energy combination model;

[0041] Based on the prediction algorithm, a rolling prediction is made for the output dispatch return of distributed renewable energy. According to the output dispatch return at the terminal moment in the predicted time domain obtained by the rolling prediction, the expectation and variance of the output dispatch return are calculated, and the optimal dispatch of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal dispatch reference value is output.

[0042] Specifically, first, an electric, thermal and gas combination model is established.

[0043] In this embodiment, the electric heat and gas combination includes an electric power subsystem, a thermal subsystem and a natural gas subsystem. In this embodiment, the electric power subsystem includes a wind turbine, a solar combined heat and power unit, a combined heat and power unit and an energy storage system; the thermal subsystem includes a solar combined heat and power unit, a combined heat and power unit, an electric boiler and a heat storage system; and the natural gas subsystem includes an electric-to-gas unit.

[0044] Secondly, according to the electricity, heat and gas combination model, the objective function and constraint conditions of the multi-energy combination model are constructed.

[0045] In this embodiment, the objective function corresponding to the combination model is constructed according to the electric, thermal and gas combination model. The objective function is the operating income and operating risk of the multi-energy combination model, which mainly includes the following decision-related variables: the power generation power of renewable energy, the output power of the thermal subsystem, the output power of the natural gas subsystem, the charging power and discharging power of the energy storage equipment, and the power converted between various energy sources. In fact, the present disclosure regards the output of each energy source in the multi-energy combination model as an asset in the multi-energy combination. Since the output of different energy sources corresponds to different output returns, different output returns are regarded as different asset returns. The operating income mentioned here refers to the output return of the multi-energy combination model, and the operating risk refers to the fluctuation of the output return of the multi-energy combination model.

[0046] First, the objective function of the energy portfolio model is set. The benefits and risks in the objective function are determined by the predicted transaction amount and cost. Therefore, the objective function of the energy portfolio model is the predicted values ​​of benefits and risks at the last moment of its prediction time domain, which are:

[0047]

[0048]

[0049] The benefit objective function is the expected maximum value of the energy benefit of the energy combination at the terminal time of the prediction time domain. In the above formula, E[W(k+m)] is the expected value of the energy benefit of the energy combination at the terminal time k+m of the prediction time domain, which is specifically expressed as:

[0050]

[0051] Among them, W(k) is the current k-time profit, R T (k) is the energy output investment return ratio at time k, U(k) is the energy output investment ratio of each energy at time k, c is the energy cost ratio, is the income of energy participating in carbon emission trading, i is a constant, indicating the i-th moment in the prediction time domain, and m is a constant, indicating the m-th moment in the prediction time domain.

[0052] The risk objective function is the minimum value of the variance of the energy benefits of the energy combination at the terminal moment of the forecast time domain. In the above formula, Var[E[W(k+m)]] is the variance of the energy benefits of the energy combination at the terminal moment k+m of the forecast time domain, representing the risk of investment, which is specifically expressed as:

[0053]

[0054] Among them, U T (k+i) is the energy investment output at time k+i, is the conditional covariance of the return on investment output of each energy source, is the energy output investment return ratio at time k+i, It is the return on investment of energy output at time k+j.

[0055] Furthermore, in order to ensure the safe and reliable operation of the combined power, heat and gas system, it is necessary to consider the constraints of the above objective function. After setting the objective function, the energy combination model is set with system operation constraints, and the constraints that meet the objective function include energy combination benefit constraints, energy supply and demand balance constraints, wind turbine constraints, solar cogeneration system constraints, cogeneration unit constraints, electric boiler constraints, heat storage constraints, electricity storage constraints, and power-to-gas operation constraints.

[0056] (1) The energy portfolio benefit constraint is:

[0057]

[0058] (2) The energy supply and demand balance constraints include power balance constraints, thermal balance constraints and gas grid balance constraints, specifically:

[0059] P WP (k)+P SP (k)+P CHP (k) = L E (k)+P BT (k)

[0060] Q SP (k)+Q CHP (k)+Q EB (k) = L Q (k)+Q HT (k)

[0061] V P2G (k)+V NG (k) = L V (k)

[0062] Among them, P WP (k) is the power output of the wind turbine at time k, P SP (k) is the power generation output of the solar thermal power cogeneration unit at time k, P CHP (k) is the power output of the cogeneration unit at time k, P BT (k) is the energy storage efficiency of the energy storage system at time k, Q SP (k) is the thermal output of the solar thermal power cogeneration unit at time k, Q CHP (k) is the heat output of the cogeneration unit at time k, Q EB (k) is the heat output of the electric boiler at time k, QHT (k) is the heat output of the heat storage system at time k, V P2G (k) is the natural gas supply from power-to-gas, V NG (k) is the natural gas purchased by the electric heating network system from the public at time k, L E (k), L Q (k), L V (k) is the load demand of electricity, heat and gas in the power, heat and gas network at time k.

[0063] (3) The wind turbine constraints include wind turbine output constraints and wind turbine climbing constraints. The wind turbine constraints are:

[0064] P WP,min (k)≤P WP (k)≤P WP,max (k)

[0065] K WP (k)ΔP WP,min (k)≤P WP (k)-P WP (k-1)≤K WP (k)ΔP WP,max (k)

[0066] Among them, P WP,min (k), P WP,max (k), ΔP WP,min (k), ΔP WP,max (k) are the upper and lower limits of wind turbine operating power and the upper and lower limits of ramp power respectively; K WP (k) is a 0-1 variable, marking the start / stop status of the wind turbine at time k.

[0067] (4) The solar cogeneration constraints include the solar cogeneration system output constraints, the solar cogeneration system ramp constraints, and the solar cogeneration system start-stop constraints. The solar cogeneration constraints are:

[0068] P SP,min (k)≤P SP (k)≤P SP,max (k)

[0069] K SP (k)ΔP SP,min (k)≤P SP (k)-P SP (k-1)≤K SP (k)ΔP SP,max (k)

[0070] P SP =H SP ·η SP

[0071] Q SP =H SP ·(1-η SP -η SPloss )

[0072] Where: P SP , Q SP are the output electrical power and thermal power of solar cogeneration respectively; η SP is the power generation efficiency of the solar thermal power cogeneration unit, η SPloss is the energy loss rate of the solar thermal power cogeneration unit, H SP is the energy output of the solar thermal power cogeneration unit; P SP,min (k), P SP,max (k), ΔP SP,min (k), ΔP SP,max (k) are the upper and lower limits of the operating power and the upper and lower limits of the ramp power of the solar thermal power cogeneration unit; K SP (k) is a 0-1 variable, marking the start and stop status of the solar thermal power cogeneration unit at time k.

[0073] (5) The cogeneration unit constraints include cogeneration unit output constraints, cogeneration unit ramp constraints, and cogeneration unit start and stop constraints. The cogeneration unit constraints are:

[0074] P CHP =V CHP ·H CHP ·η CHP

[0075] Q CHP =V CHP ·H CHP ·(1-η CHP -η CHPloss )

[0076] Where: P CHP , Q CHP are the output electrical power and thermal power of the cogeneration respectively; η CHP is the power generation efficiency of the cogeneration unit, η CHPloss is the energy loss rate of the cogeneration unit, V CHP is the natural gas consumption of the cogeneration unit, H CHP is the calorific value of natural gas.

[0077] P CHP,min (k)≤P CHP (k)≤P CHP,max (k)

[0078] K CHP (k)ΔP CHP,min(k)≤P CHP (k)-P CHP (k-1)≤K CHP (k)ΔP CHP,max (k)

[0079] Where: P CHP,min (k), P CHP,max (k), ΔP CHP,min (k), ΔP CHP,max (k) are the upper and lower limits of the operating power and the upper and lower limits of the ramping power of the cogeneration unit; K CHP (k) is a 0-1 variable, marking the start / stop status of the cogeneration unit at time k.

[0080] (6) The electric boiler constraints are:

[0081] Q EB (k) = η EB (1-μ loss ) EB (k)

[0082] Where: Q EB (k) is the thermal power of the electric boiler at time k, P EB (k) is the electric power consumed by the electric boiler at time k, η EB is the energy conversion power of the electric boiler, μ loss Indicates the heat loss efficiency of the electric boiler.

[0083] P EB,min (k)≤P EB (k)≤P EB,max (k)

[0084] K EB (k)ΔP EB,min (k)≤P EB (k)-P EB (k-1)≤K EB (k)ΔP EB,max (k)

[0085] Among them, P EB,min (k), P EB,max (k), ΔP EB,min (k), ΔP EB,max (k) are the upper and lower limits of the operating power and the upper and lower limits of the ramp power of the electric boiler; K EB (k) is a 0-1 variable, marking the start and stop status of the electric boiler at time k.

[0086] (7) The heat storage constraint is:

[0087] Q HTc,min (k)≤Q HTc(k)≤Q BTc,max (k)

[0088] Q HTdc,min (k)≤Q HTdc (k)≤Q HTdc,max (k)

[0089] Q SBT,min (k)≤Q SBT (k)≤Q SBT,max (k)

[0090] Where: Q HTc,min (k), Q HTc,max (k), Q HTdc,min (k), Q HTdc,max (k) are the lower and upper limits of the thermal storage charging and discharging power, respectively, Q SBT,max (k), Q SBT,min (k) represent the upper and lower limits of the capacity of the heat storage system respectively.

[0091] (8) The power storage constraint is:

[0092] P BTc,min (k)≤P BTc (k)≤P BTc,max (k)

[0093] P BTdc,min (k)≤P BTdc (k)≤P BTdc,max (k)

[0094] S BT,min (k)≤S BT (k)≤S BT,max (k)

[0095] Where: P BTc,min (k), P BTc,max (k), P BTdc,min (k), P BTdc,max (k) are the lower and upper limits of the storage charging and discharging power, respectively, S BT,max (k), S BT,min (k) represent the upper and lower limits of the capacity of the power storage system respectively.

[0096] (9) The power-to-gas operation constraints are:

[0097]

[0098] Where: Q P2G is the natural gas flow rate of power-to-gas, P P2G is the electrical power consumed by the power-to-gas unit, μ P2G is the energy conversion efficiency of the power-to-gas unit, H HVIt is the low calorific value of natural gas.

[0099] P P2G,min (k)≤P P2G (k)≤P P2G,max (k)

[0100] K P2G (k)ΔP P2G,min (k)≤P P2G (k)-P P2G (k-1)≤K P2G (k)ΔP P2G,max (k)

[0101] Among them, P P2G,min (k), P P2G,max (k), ΔP P2G,min (k), ΔP P2G,max (k) are the upper and lower limits of the operating power and the upper and lower limits of the ramping power of the power-to-gas unit; K P2G (k) is a 0-1 variable, marking the start / stop status of the electric-to-gas unit at time k.

[0102] Finally, a rolling forecast is made for the output dispatch return of distributed renewable energy based on the prediction algorithm. According to the output dispatch return at the terminal moment in the predicted time domain obtained by the rolling forecast, the expectation and variance of the output dispatch return are calculated, and the optimal dispatch of the energy combination model is solved based on the multi-objective genetic algorithm, and the optimal dispatch reference value is output.

[0103] like Figure 2 As shown, the output of distributed renewable energy is obtained, and a rolling forecast of the output of distributed renewable energy and the output return (i.e., income) generated by scheduling is made based on the prediction model. In this embodiment, according to the current energy output, the energy income at different times in the prediction time domain is solved based on the model prediction control algorithm, specifically:

[0104]

[0105] According to the output dispatch return at the terminal moment of the predicted time domain obtained by rolling forecast, that is, the benefit at the terminal moment of the predicted time domain, the expectation and variance of the benefit are calculated, and the optimal dispatch of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal dispatch reference value is output.

[0106] like Figure 3 As shown, the optimal scheduling of the multi-energy investment portfolio model based on the multi-objective genetic algorithm specifically includes the following steps:

[0107] For each moment in the prediction range, a group of energy combinations with different energy outputs are randomly generated according to the even distribution. Each energy combination is regarded as an individual. The parent population is selected according to the dominance relationship and crowding degree between individuals. The child population is obtained through selection, crossover and mutation operations.

[0108] The parent population is merged with the obtained child population, and a fast non-dominated sort is performed between individuals. According to the dominance relationship and crowding degree between individuals, multiple individuals are selected to form a new parent population.

[0109] If the number of generations of the new parent population is less than the set maximum number of generations, a new child population is generated again through the selection operation, crossover operation and mutation operation of the genetic algorithm, and the above merging and population selection operations are repeated; otherwise, the parent population is output as the Pareto frontier of the multi-objective optimization problem in the next prediction domain.

[0110] In the above genetic algorithm, the crossover operation refers to randomly exchanging genes at two randomly selected positions in two individuals; and the mutation operation refers to exchanging two randomly selected genes in one individual.

[0111] Specifically, the genetic algorithm adopts a non-dominated sorting genetic algorithm, and for each moment in the prediction range, a group of energy combinations of energy output conditions are randomly generated according to an even distribution.

[0112] First, at each moment, the randomly generated multiple energy combinations are regarded as a population, and each energy combination is regarded as an individual in the population, and the randomly generated population is non-dominated sorted. For example, at time k, k1, k2, ...k n A total of n energy combinations with different output conditions constitute a population.

[0113] The non-dominated sorting refers to stratifying the individuals in the population according to the dominance relationship of the individuals in the population. The specific steps are as follows:

[0114] The population of size n is stratified by the non-dominated sorting algorithm. The specific steps are as follows:

[0115] (1) Let a = 1;

[0116] (2) For all b = 1, 2, ... n, and a ≠ b, according to the above definition, compare individual k a and individual k b the relationship of domination and non-domination between them;

[0117] (3) If there is no individual k b Better than k a , then k a Labeled as nondominant individuals;

[0118] (4) Let a=a+1, go to step (2), and continue marking the dominance and non-domination relationships of other individuals, and so on, until all non-dominated individuals are found.

[0119] The set of non-dominated individuals obtained through the above steps is the first non-dominated layer of the population. Then, ignore these marked non-dominated individuals, that is, these individuals will not be compared in the next round, and follow the above steps (1) to (4) to obtain the second non-dominated layer. And so on, until the entire population is stratified.

[0120] The computational complexity is further reduced by using the non-dominated sorting algorithm of NSGA-II mentioned above.

[0121] According to the dominance relationship and crowding distance between individuals, a parent population containing N individuals is selected. N is a constant and can be set according to the specific situation. A parent population is generated at each prediction moment. At the same time, the parent population is subjected to selection, crossover and mutation operations to obtain the first child population at each moment.

[0122] The parent population is selected according to the dominance relationship and crowding distance between individuals. Specifically, individuals with a small number of non-dominated layers (i.e., the first non-dominated layer) are first selected. If the number of selected individuals is less than N, individuals with a large crowding distance in the next dominance layer (i.e., the second non-dominated layer) are selected, and so on, until N individuals are selected as the parent population.

[0123] The calculation formula of the congestion degree is:

[0124]

[0125] Among them, d j is the crowding distance of the jth individual in the population, j+1 and j-1 are the two individuals adjacent to individual j on both sides of the Pareto front where j is located, and f k j+1 and f k j-1 are the kth objective function values ​​of the j+1th and j-1th individuals respectively.

[0126] Then, starting from the second generation, the parent population at the current moment and the first child population are merged, and then the fast non-dominated sort is performed again, and at the same time, the crowding distance of the individuals in each non-dominated layer is calculated. Since the population has been merged, the number of individuals in the population at this time is twice as much as before, so it is necessary to select suitable individuals to form a new parent population based on the non-dominated relationship and the crowding distance of the individuals. At this time, the number of individuals selected to form the new parent population is the same as the number of individuals in the previous generation population, and the specific selection process is to first select individuals with a small number of non-dominated layers (i.e., the first non-dominated layer). If the number of non-dominated layers crosses the boundary, the individuals with a large crowding distance in the non-dominated layer (i.e., the second non-dominated layer) are preferentially selected.

[0127] The non-dominated layer number crossing the boundary means that when the populations are merged, there may be a non-dominated layer that is on the boundary of the number of selected individuals.

[0128] A new offspring population is generated through the basic operations of the genetic algorithm, and this cycle is repeated until the population generation number of the new parent population is not less than the set maximum generation number, then the parent population is output. At this time, the output parent population is the Pareto frontier of the multi-objective optimization problem in the next prediction time domain, that is, the solution set of the energy optimization scheduling obtained by optimization.

[0129] In this embodiment, the outputting of the optimal scheduling reference value means that based on the Pareto front of the solved optimal scheduling problem, the optimal point of the Pareto front is output as the optimal scheduling reference value.

[0130] Among them, both the benefits and the risks brought by the benefits are taken into consideration at the same time, and the optimal point of the Pareto frontier that meets the current needs is selected from the Pareto frontier according to the actual situation as the optimal resource scheduling combination output.

[0131] Example 2

[0132] The purpose of this embodiment is to provide a multi-objective energy combination optimization scheduling system based on model predictive control, including:

[0133] Multi-energy combination model building module, used to build an electricity, heat and gas combination model;

[0134] An objective function and constraint condition construction module, used to construct the objective function and constraint conditions of the multi-energy combination model according to the electric, thermal and gas combination model;

[0135] The optimal scheduling solution module is used to perform rolling prediction of distributed renewable energy scheduling results based on the prediction algorithm; according to the results of the predicted time domain terminal time obtained by using the rolling prediction, the expectation and variance of the prediction result are calculated, and the optimal scheduling of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal scheduling reference value is output.

[0136] The steps involved in the above embodiment 2 correspond to those in method embodiment 1. For the specific implementation method, please refer to the relevant description part of embodiment 1.

[0137] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0138] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0139] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A multi-objective energy combination optimization scheduling method based on model predictive control, characterized in that: The following steps are involved: Establish an electric, heat and gas combination model; According to the electric, heat and gas combination model, the objective function and constraint conditions of the multi-energy combination model are constructed; wherein the construction process includes: The objective function of the multi-energy combination model is set. The objective function includes a benefit objective function and a risk objective function. The objective function is the predicted value of benefit and risk at the last moment of its prediction time domain, which are respectively: That is, the benefit objective function is the expected maximum value of the energy benefit of the multi-energy combination model at the terminal time of the prediction time domain; in the above formula, E[W(k+m)] is the expected energy benefit of the energy combination at the terminal time k+m of the prediction time domain, which is expressed as: Among them, W(k) is the current k-time profit, R T (k) is the energy output investment return ratio at time k, U(k) is the energy output investment ratio of each energy at time k, c is the energy cost ratio, is the income of energy participating in carbon emission trading, i is a constant, indicating the i-th moment in the prediction time domain, and m is a constant, indicating the m-th moment in the prediction time domain; That is, the risk objective function is the minimum value of the variance of the energy benefits of the multi-energy combination model at the terminal moment of the forecast time domain; in the above formula, Var[E[W(k+m)]] is the variance of the energy benefits of the energy combination at the terminal moment k+m of the forecast time domain, representing the risk of investment, which is expressed as: Among them, U T (k+i) is the energy investment output at time k+i, is the conditional covariance of the return on investment output of each energy source, is the energy output investment return ratio at time k+i, is the return on investment of energy output at time k+j; Based on the prediction algorithm, a rolling prediction is made for the output dispatch return of distributed renewable energy. According to the output dispatch return at the terminal moment in the predicted time domain obtained by the rolling prediction, the expectation and variance of the output dispatch return are calculated, and the optimal dispatch of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal dispatch reference value is output.

2. The multi-objective energy combination optimization scheduling method based on model predictive control according to claim 1 is characterized in that: The constraints include energy combination benefit constraints, energy supply and demand balance constraints, wind turbine constraints, solar cogeneration system constraints, cogeneration unit constraints, electric boiler constraints, heat storage constraints, electricity storage constraints, and power-to-gas operation constraints.

3. The multi-objective energy combination optimization scheduling method based on model predictive control according to claim 2 is characterized in that: The energy supply and demand balance constraints include power balance constraints, thermal balance constraints and gas network balance constraints, specifically: P WP (k)+P SP (k)+P CHP (k)=L E (k)+P BT (k) Q SP (k)+Q CHP (k)+Q EB (k)=L Q (k)+Q HT (k) V P2G (k)+V NG (k)=L V (k) Among them, P WP (k) is the power output of the wind turbine at time k, P SP (k) is the power generation output of the solar thermal power cogeneration unit at time k, P CHP (k) is the power output of the cogeneration unit at time k, P BT (k) is the energy storage efficiency of the energy storage system at time k, Q SP (k) is the thermal output of the solar thermal power cogeneration unit at time k, Q CHP (k) is the heat output of the cogeneration unit at time k, Q EB (k) is the heat output of the electric boiler at time k, Q HT (k) is the heat output of the heat storage system at time k, V P2G (k) is the natural gas supply from power-to-gas, V NG (k) is the natural gas purchased by the electric heating network system from the public at time k, L E (k), L Q (k), L V (k) is the load demand of electricity, heat and gas in the power, heat and gas network at time k.

4. The multi-objective energy combination optimization scheduling method based on model predictive control according to claim 1 is characterized in that: The rolling prediction of the output dispatch return of distributed renewable energy based on the prediction algorithm refers to solving the energy income situation at different times in the prediction time domain based on the model predictive control algorithm according to the current energy output situation, specifically: Among them, W(k) is the current k-time profit, R T (k) is the energy output investment return ratio at time k, U(k) is the energy output investment ratio of each energy at time k, c is the energy cost ratio, It is the income of energy participating in carbon emission trading.

5. The multi-objective energy combination optimization scheduling method based on model predictive control according to claim 1 is characterized in that: The method for solving the optimal scheduling of the multi-energy combination model based on the multi-objective genetic algorithm specifically includes the following steps: For each moment in the prediction range, a group of energy combinations with different energy outputs are randomly generated according to the even distribution. Each energy combination is regarded as an individual. The parent population is selected according to the dominance relationship and crowding degree between individuals. The child population is obtained through selection, crossover and mutation operations. The parent population is merged with the obtained child population, and a fast non-dominated sort is performed between individuals. According to the dominance relationship and crowding degree between individuals, multiple individuals are selected to form a new parent population. If the number of generations of the new parent population is less than the set maximum number of generations, a new child population is generated again through the selection operation, crossover operation and mutation operation of the genetic algorithm, and the above merging and population selection operations are repeated; otherwise, the parent population is output as the Pareto frontier of the multi-objective optimization problem in the next prediction domain.

6. The multi-objective energy combination optimization scheduling method based on model predictive control according to claim 1 is characterized in that: Outputting the optimal scheduling reference value means outputting the optimal point of the Pareto front as the optimal scheduling reference value based on the Pareto front of the solved optimal scheduling problem.

7. A multi-objective energy combination optimization scheduling system based on model predictive control, characterized in that: include: Multi-energy combination model building module, used to build an electricity, heat and gas combination model; The objective function and constraint condition construction module is used to construct the objective function and constraint conditions of the multi-energy combination model according to the electric, heat and gas combination model; wherein the construction process includes: The objective function of the multi-energy combination model is set. The objective function includes a benefit objective function and a risk objective function. The objective function is the predicted value of benefit and risk at the last moment of its prediction time domain, which are respectively: That is, the benefit objective function is the expected maximum value of the energy benefit of the multi-energy combination model at the terminal time of the prediction time domain; in the above formula, E[W(k+m)] is the expected energy benefit of the energy combination at the terminal time k+m of the prediction time domain, which is expressed as: Among them, W(k) is the current k-time profit, R T (k) is the energy output investment return ratio at time k, U(k) is the energy output investment ratio of each energy at time k, c is the energy cost ratio, is the income of energy participating in carbon emission trading, i is a constant, indicating the i-th moment in the prediction time domain, and m is a constant, indicating the m-th moment in the prediction time domain; That is, the risk objective function is the minimum value of the variance of the energy benefits of the multi-energy combination model at the terminal moment of the forecast time domain; in the above formula, Var[E[W(k+m)]] is the variance of the energy benefits of the energy combination at the terminal moment k+m of the forecast time domain, representing the risk of investment, which is expressed as: Among them, U T (k+i) is the energy investment output at time k+i, is the conditional covariance of the return on investment output of each energy source, is the energy output investment return ratio at time k+i, is the return on investment of energy output at time k+j; The optimal scheduling solution module is used to make rolling predictions on the output scheduling returns of distributed renewable energy based on the prediction algorithm; according to the output scheduling returns at the terminal moment of the predicted time domain obtained by the rolling prediction, the expectation and variance of the output scheduling returns are calculated, and the optimal scheduling of the multi-energy combination model is solved based on the multi-objective genetic algorithm, and the optimal scheduling reference value is output.

8. The multi-objective energy combination optimization scheduling system based on model predictive control according to claim 7 is characterized in that: The constraints include energy combination benefit constraints, energy supply and demand balance constraints, wind turbine constraints, solar cogeneration system constraints, cogeneration unit constraints, electric boiler constraints, heat storage constraints, electricity storage constraints, and power-to-gas operation constraints.

Images