Comprehensive energy system optimization method based on hierarchical uncertainty

Abstract

The invention discloses an integrated energy system optimization method based on hierarchical uncertainty, and the method comprises the following steps: taking an error between historical power generation data and historical prediction power generation data of an integrated energy system as a historical prediction error, and constructing a prediction error fuzzy set according to the historical prediction error, constructing a distribution robust optimization model based on the prediction error fuzzy set; obtaining the day-ahead one-hour power prediction data of the integrated energy system, and inputting the day-ahead one-hour power prediction data into the distributed robust optimization model to obtain a day-ahead scheduling scheme of the integrated energy system; the method comprises the following steps: constructing an intra-day random scene set, constructing a random optimization model based on the intra-day random scene set, acquiring intra-day power prediction data based on a day-ahead scheduling scheme by taking 15 minutes as a period, inputting the intra-day power prediction data into the random optimization model to obtain an intra-day scheduling scheme, and repeating the steps until optimization scheduling of all periods in the day is completed. And obtaining a final scheduling scheme of the integrated energy system.

Description

Technical Field

[0001] The present invention relates to the field of integrated energy system optimization, and in particular to an integrated energy system optimization method based on hierarchical uncertainty. Background Art

[0002] Integrated energy systems have attracted much attention in recent years due to their high efficiency, reliability and environmental protection. This system can integrate multiple energy forms, such as electricity, heat, hydrogen and cold energy, and can achieve efficient use of energy through optimized scheduling and energy management. However, with the access of large-scale renewable energy, the intermittent and volatile nature of wind power and photovoltaics has brought great challenges to the reliable operation of integrated energy systems.

[0003] Existing technologies often use robust optimization, stochastic optimization and distributed robust optimization to deal with the uncertainty of new energy. Robust optimization uses uncertainty sets to describe the fluctuation range of random variables, with the worst-case optimization as the core, and often obtains conservative solutions. Stochastic optimization usually assumes that random variables precisely obey a certain probability distribution, and generates a large number of random scenarios to combat the interference of uncertain factors in the system, often obtaining optimistic solutions, and a large number of discrete scenarios also make the calculation efficiency low. Distributed robust optimization combines the characteristics of robust optimization and stochastic optimization well, can strike a balance between economy and robustness, and constructs fuzzy sets in a data-driven way to describe random variables.

[0004] However, most of the existing distributed robust optimization methods are based on a single time scale, that is, they only make day-ahead scheduling decisions, without taking into account the inherent characteristics of wind power generation and photovoltaic power generation that the errors increase as the time scale becomes longer. This can easily lead to a large deviation between the scheduling plan and the actual situation, thus falling into a local optimum. Summary of the invention

[0005] The present invention proposes an integrated energy system optimization method based on hierarchical uncertainty, which solves the problem that most existing optimization methods only make day-ahead scheduling decisions and do not take into account the characteristics that the errors of wind power generation and photovoltaic power generation increase as the time scale becomes longer.

[0006] In order to solve the above technical problems, the present invention provides a comprehensive energy system optimization method based on hierarchical uncertainty, comprising the following steps:

[0007] Step S1: taking the error between the historical power generation data and the historical predicted power generation data of the integrated energy system as the historical prediction error, constructing a prediction error fuzzy set according to the historical prediction error, and constructing a distributed robust optimization model based on the prediction error fuzzy set;

[0008] Step S2: obtaining the power forecast data of the integrated energy system at the one-hour level before the day, inputting the power forecast data of the one-hour level before the day into the distributed robust optimization model, and obtaining the day-ahead scheduling plan of the integrated energy system;

[0009] Step S3: Construct a set of intra-day random scenarios, and construct a random optimization model based on the intra-day random scenario set. With a period of 15 minutes, obtain intra-day power forecast data based on the day-ahead scheduling plan, input the intra-day power forecast data into the random optimization model to obtain the intra-day scheduling plan, repeat the above steps until the optimization scheduling of all cycles within the day is completed, and obtain the final scheduling plan of the integrated energy system.

[0010] Preferably, constructing a prediction error fuzzy set according to the historical prediction error in step S1 comprises the following steps:

[0011] Step S11: obtaining an empirical distribution of prediction errors of the integrated energy system according to the historical prediction errors, analyzing the empirical distribution to obtain characteristics of the prediction errors, obtaining predicted power generation of the integrated energy system, and constructing a true distribution of prediction errors of the integrated energy system according to the predicted power generation and the characteristics of the prediction errors;

[0012] Step S12: clustering the historical prediction errors to obtain several typical scenarios of prediction errors and the scenario probabilities corresponding to each typical scenario, and constructing a prediction error support set:

[0013] Step S13: Calculate the difference between the empirical distribution and the true distribution, set the parameters of the fuzzy set according to the difference, and construct the prediction error fuzzy set according to the prediction error support set and the parameters of the fuzzy set.

[0014] Preferably, the expression of the prediction error support set in step S12 is:

[0015]

[0016] In the formula, s is the prediction error support set of scenario s; is a random vector of real distribution, is a random variable that is actually distributed in time period t; T is the total number of time periods; R is the set of real numbers; are the minimum and maximum values ​​of historical forecast errors respectively; is the sth scenario of historical forecast error.

[0017] Preferably, in step S13, the difference between the empirical distribution and the true distribution is calculated by using the Wasserstein probability distance, and the expression of the Wasserstein probability distance is:

[0018]

[0019] In the formula, For the empirical distribution and the true distribution The Wasserstein probability distance between them; inf(·) is the infimum function; For the empirical distribution and the true distribution The entire set of joint distributions Q of; E(·) is the expected function; ρ(·) is the L2 norm, which represents the basic distance of Wasserstein probability distance; is a random vector from the empirical distribution, is a random variable with empirical distribution in period t.

[0020] Preferably, the expression of the prediction error fuzzy set in step S13 is:

[0021]

[0022] In the formula, is the true value of the error of wind power and photovoltaic power generation of the integrated energy system randomly generated from the real distribution; is an auxiliary random variable; E(·) is the expected function; The true distribution A collection of; S is the total number of typical scenes; is a variable representing a typical scenario; p s is the scenario probability of the typical scenario s; θ is the conservative constant of the fuzzy set, which is set according to the difference between the empirical distribution and the true distribution.

[0023] Preferably, the expression of the distributed robust optimization model is:

[0024]

[0025] In the formula, F da is the objective function of the distributed robust optimization model; x d To determine the decision variables in the robust optimization model; is an uncertain random variable generated according to the true distribution; sup(·) is the supremum function; E(·) is the expectation function; is the real distribution; is a fuzzy set.

[0026] Preferably, the objective function of the distributed robust optimization model is expressed as:

[0027]

[0028]

[0029] In the formula, C buy C is the energy purchase cost; om is the operation and maintenance cost; C cut is the cost of power abandonment; R CO2 is the carbon tax revenue; T is the total number of time periods; p gas are electricity and gas prices respectively; are the power of electricity and gas purchased in period t respectively; R is the total number of equipment in the integrated energy system; is the operation and maintenance cost of device r; is the operating power of device r during period t; are the penalty costs for abandoning wind and solar power respectively; c CO2 is the carbon tax price, δ re , δ q , δ g are the carbon emission coefficients of renewable energy, main gas grid, and main power grid respectively; are wind power and photovoltaic power in period t respectively; is the amount of natural gas consumed by the gas turbine during period t; is the amount of natural gas consumed by the gas boiler during period t.

[0030] Preferably, the constraints of the distributed robust optimization model include equipment operation constraints, power balance constraints, power abandonment constraints and energy purchase constraints;

[0031] The equipment operation constraints include gas turbine constraints, gas boiler constraints, electric refrigerator constraints, absorption refrigerator constraints, electrolyzer constraints, hydrogen fuel cell constraints and energy storage equipment constraints;

[0032] The expression of the gas turbine constraint is:

[0033]

[0034] In the above formula, P t GT , are the power generation power and heating power of the gas turbine during period t respectively; is the amount of natural gas consumed by the gas turbine during period t; are the power generation efficiency and heating efficiency of the gas turbine respectively; are the upper and lower limits of the gas turbine ramp rate, respectively; are the upper and lower limits of the gas turbine power generation respectively; is the on / off state of the gas turbine during period t, with a value of 1 indicating on and a value of 0 indicating off;

[0035] The expression of the gas boiler constraint is:

[0036]

[0037] In the formula, is the amount of natural gas consumed by the gas boiler during period t; is the heating power of the gas boiler during period t; is the heating efficiency of the gas boiler; They are the upper and lower limits of the heating power of the gas boiler respectively;

[0038] The expression of the electric refrigerator constraint is:

[0039]

[0040] In the formula, P is the cooling power of the electric refrigerator during period t; t EC is the power consumption of the electric refrigerator during period t; is the cooling efficiency of the electric refrigerator; They are the upper and lower limits of the refrigeration power of the electric refrigerator respectively;

[0041] The expression of the absorption refrigerator constraint is:

[0042]

[0043] In the formula, is the cooling power of the absorption chiller during period t; is the heat consumption power of the absorption refrigerator during period t; is the refrigeration efficiency of the absorption chiller; are the upper and lower limits of the refrigeration power of the absorption chiller

[0044] The expression of the electrolyzer constraint is:

[0045]

[0046] In the formula, is the hydrogen production power of the electrolyzer during period t; P t EL is the power consumption of the electrolytic cell during period t; is the hydrogen production efficiency of the electrolyzer; They are the upper and lower limits of the power consumption of the electrolytic cell respectively;

[0047] The expression of the hydrogen fuel cell constraint is:

[0048]

[0049] Where P t HFC is the power generation of the hydrogen fuel cell during period t; is the hydrogen consumption power of the hydrogen fuel cell during period t; The power generation efficiency of hydrogen fuel cells; They are the upper and lower limits of hydrogen power consumption of hydrogen fuel cells respectively;

[0050] The expression of the energy storage device constraint is:

[0051]

[0052] In the formula, X represents energy storage equipment, including batteries BES and hydrogen storage tanks HES; E represents energy, including electricity P and hydrogen H; is the energy stored in energy storage device X during period t; are the power of charging and discharging energy E of energy storage device X during period t respectively; are the charging and discharging efficiencies of energy storage device X respectively; are the upper and lower limits of the energy stored in the energy storage device X; They are the charging and discharging states of the energy storage device X in period t, with a value of 1 indicating charging / discharging and a value of 0 indicating no charging / discharging; They are the upper and lower limits of the charging power of energy storage device X respectively; are the upper and lower limits of the energy storage device X’s discharge power respectively; Δt represents the scheduling period interval; represents the energy stored in energy storage device X at the end of the scheduling period; represents the energy stored in energy storage device X at the beginning of the scheduling period;

[0053] The power balance constraints include electric power balance, thermal power balance, hydrogen power balance, cooling power balance and natural gas power balance;

[0054] The expression of the electric power balance is:

[0055]

[0056] The expression of thermal power balance is:

[0057]

[0058] The expression of the hydrogen power balance is:

[0059]

[0060] The expression of the cooling power balance is:

[0061]

[0062] The expression of the natural gas power balance is:

[0063]

[0064] In the above formula, are the charging and discharging powers of the battery during period t respectively; are the power of charging and discharging hydrogen in the hydrogen storage tank during period t; P t load , are the electric, heating and cooling load powers in period t respectively;

[0065] The expression of the power curtailment constraint is:

[0066]

[0067] In the formula, are the wind power and solar power abandonment in period t respectively; P t WT , P t PV are wind power and photovoltaic power in period t respectively;

[0068] The energy purchase constraints include electricity purchase constraints and gas purchase constraints;

[0069] The expression of the power purchase constraint is:

[0070]

[0071] The expression of the gas purchase constraint is:

[0072]

[0073] In the above formula, They are the upper limits of electricity and gas purchase power respectively.

[0074] Preferably, in step S3, an initial scene is obtained by Latin hypercube sampling, and a K-means clustering algorithm is used to perform scene reduction on the initial scene to obtain several typical scenes, and the several typical scenes are used as a random scene set.

[0075] Preferably, the expression of the stochastic optimization model is:

[0076]

[0077] F id =C buy +C om +C adj +C cut -RCO2 ;

[0078]

[0079] In the formula, x i is the decision variable of the stochastic optimization model; ξ i is an uncertain random variable generated according to a set of random scenarios; C adj Adjust costs for equipment power; is the power adjustment cost of device r; is the power adjustment amount of device r during period t.

[0080] The benefits of the present invention include at least:

[0081] 1. Use historical forecast error data to construct fuzzy sets, so that the distributed robust optimization model can quantify and handle the uncertainty of power generation in the integrated energy system, thereby improving the robustness of system operation; day-ahead optimal scheduling allows the system to plan in advance and can consider uncertainties on a longer time scale, thereby formulating a more economical and effective day-ahead scheduling plan;

[0082] 2. The intraday stochastic optimization model can be adjusted based on the real-time updated power forecast data, enabling the system to quickly respond to short-term fluctuations in renewable energy generation. The 15-minute update frequency provides a finer time scale, which helps capture rapid changes in renewable energy generation and achieve more accurate scheduling.

[0083] 3. By combining the day-ahead distributed robust optimization and the intraday stochastic optimization model, the optimization decisions of the integrated energy system are stratified, which can balance economy and robustness at different time scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;

[0085] Figure 2 This is an operation framework diagram of the integrated energy system according to an embodiment of the present invention;

[0086] Figure 3 Schematic diagram of total cost change under different Wasserstein sphere radii in an embodiment of the present invention. DETAILED DESCRIPTION

[0087] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0088] like Figure 1 As shown, an embodiment of the present invention provides a comprehensive energy system optimization method based on hierarchical uncertainty, comprising the following steps:

[0089] Step S1: Establish an integrated energy system operation framework.

[0090] Specifically, the operating framework of the integrated energy system of the embodiment of the present invention is as follows: Figure 2 As shown in the figure, the main energy flows in the integrated energy system are electricity, heat, hydrogen, cold and natural gas. Among them, wind turbines, photovoltaic generators, main power grids and main gas grids are used as energy production and input links; electrolyzers, hydrogen fuel cells, gas turbines, gas boilers, electric refrigerators and absorption refrigerators are used as intermediate links in energy conversion; energy storage equipment includes batteries and hydrogen storage tanks; and electricity, heat and cold loads are used as energy consumption links.

[0091] Step S2: Input the power forecast data of various loads and wind power and photovoltaic power at the 1h level before the day, use the data-driven Wasserstein fuzzy set to characterize the uncertainty of new energy errors at a long time scale before the day, and solve the optimal scheduling plan before the day with a time interval of 1h through the distributed robust optimization model, including the following steps:

[0092] Step S21: Taking the error between the historical power generation data and the historical predicted power generation data of the integrated energy system as the historical prediction error, and obtaining the empirical distribution of the prediction error of the integrated energy system based on the historical prediction error By analyzing the empirical distribution, we can get the characteristics of the prediction error and obtain the predicted power generation of the integrated energy system. Based on the predicted power generation and the characteristics of the prediction error, we can construct the true distribution of the prediction error of the integrated energy system.

[0093] Step S22: cluster the historical prediction errors to obtain several typical prediction error scenarios and the scenario probabilities corresponding to each typical scenario, thereby constructing a prediction error support set, where the expression of the prediction error support set is:

[0094]

[0095] In the formula, s is the prediction error support set of scenario s; is the random variable of the t period with the true distribution of the new energy error The vector is composed of, in the embodiment of the present invention, t is 1h; T is the dimension of the random variable, that is, the total number of time periods, in the embodiment of the present invention, T is 24; is the auxiliary random variable The vector composed of; R is the set of real numbers; are the minimum and maximum values ​​of historical forecast errors respectively; is the sth scenario of historical forecast error.

[0096] Step S23: Calculate the difference between the empirical distribution and the true distribution through the Wasserstein probability distance, set the parameter θ of the fuzzy set according to the difference, and then construct the Wasserstein fuzzy set of the prediction error according to the support set of the prediction error and the parameters of the fuzzy set.

[0097] Specifically, the expression of Wasserstein probability distance is:

[0098]

[0099] In the formula, For the empirical distribution and the true distribution The Wasserstein probability distance between them; inf(·) is the infimum function; For the empirical distribution and the true distribution The entire set of joint distributions Q of; E(·) is the expected function; ρ(·) is the L2 norm, which represents the basic distance of Wasserstein probability distance; is a random variable in period t with the empirical distribution of the new energy error Vector composed of.

[0100] The expression of Wasserstein fuzzy set is:

[0101]

[0102] In the formula, is the true value of the error of wind power and photovoltaic power generation of the integrated energy system randomly generated from the real distribution; is an auxiliary random variable; E(·) is the expected function; The true distribution A collection of; S is the total number of typical scenes; is a variable representing a typical scenario; p s is the scenario probability of the typical scenario s; θ is a constant that controls the conservatism of the Wasserstein fuzzy set.

[0103] Wasserstein fuzzy sets include the probability space with Wasserstein distance as the metric and empirical distribution as the All probability distributions in the sphere with θ as the radius and θ as the center are the true distributions of the new energy error. As the sample size of historical data increases, the empirical distribution will gradually approach the true distribution That is, data-driven features.

[0104] Step S24: construct a distributionally robust optimization model based on the Wasserstein fuzzy set of the prediction error, and set the constraints of the distributionally robust optimization model.

[0105] Specifically, the objective function of the upper layer distributed robust optimization model is the total cost F of the integrated energy system. da Minimum is the optimization goal. The objective function includes four parts: energy purchase cost C buy , Operation and maintenance cost C om , power abandonment cost C cut and carbon tax revenue R CO2 , whose expression is:

[0106]

[0107]

[0108] In the formula, p gas are electricity and gas prices respectively; are the power of electricity and gas purchased in period t respectively; R is the total number of equipment in the integrated energy system; is the operation and maintenance cost of equipment r, and r∈{GT,GB,EC,AC,EL,HFC,BES,HES}, GT is gas turbine, GB is gas boiler, EC is electric refrigerator, AC is absorption refrigerator, EL is electrolyzer, HFC is hydrogen fuel cell, BES is battery, and HES is hydrogen storage tank; is the operating power of device r during period t; are the penalty costs for abandoning wind and solar power respectively; c CO2 is the carbon tax price, δ re , δ q , δ g are the carbon emission coefficients of renewable energy, main gas grid, and main power grid respectively; are wind power and photovoltaic power in period t respectively; is the amount of natural gas consumed by the gas turbine during period t; is the amount of natural gas consumed by the gas boiler during period t.

[0109] The constraints of the distributed robust optimization model include equipment operation constraints, power balance constraints, power abandonment constraints and energy purchase constraints.

[0110] The equipment operation constraints include gas turbine constraints, gas boiler constraints, electric refrigerator constraints, absorption refrigerator constraints, electrolyzer constraints, hydrogen fuel cell constraints and energy storage equipment constraints.

[0111] (1) The expression of gas turbine constraint is:

[0112]

[0113] In the above formula, P t GT , are the power generation power and heating power of the gas turbine during period t respectively; is the amount of natural gas consumed by the gas turbine during period t; are the power generation efficiency and heating efficiency of the gas turbine respectively; are the upper and lower limits of the gas turbine ramp rate, respectively; are the upper and lower limits of the gas turbine power generation respectively; It is the on / off state of the gas turbine during period t. A value of 1 indicates on and a value of 0 indicates off.

[0114] (2) The expression of gas boiler constraint is:

[0115]

[0116] In the formula, is the amount of natural gas consumed by the gas boiler during period t; is the heating power of the gas boiler during period t; is the heating efficiency of the gas boiler; They are the upper and lower limits of the heating power of the gas boiler respectively.

[0117] (3) The expression of the electric refrigerator constraint is:

[0118]

[0119] In the formula, P is the cooling power of the electric refrigerator during period t; t EC is the power consumption of the electric refrigerator during period t; is the cooling efficiency of the electric refrigerator; They are the upper and lower limits of the cooling power of the electric refrigerator respectively.

[0120] (4) The expression of the absorption refrigerator constraint is:

[0121]

[0122] In the formula, is the cooling power of the absorption chiller during period t; is the heat consumption power of the absorption refrigerator during period t; is the refrigeration efficiency of the absorption chiller; They are the upper and lower limits of the refrigeration power of the absorption chiller respectively.

[0123] (5) The expression of the electrolytic cell constraint is:

[0124]

[0125] In the formula, is the hydrogen production power of the electrolyzer during period t; P t EL is the power consumption of the electrolytic cell during period t; is the hydrogen production efficiency of the electrolyzer; They are the upper and lower limits of the power consumption of the electrolytic cell respectively.

[0126] (6) The expression of hydrogen fuel cell constraint is:

[0127]

[0128] Where P t HFC is the power generation of the hydrogen fuel cell during period t; is the hydrogen consumption power of the hydrogen fuel cell during period t; The power generation efficiency of hydrogen fuel cells; They are respectively the upper and lower limits of hydrogen consumption power of hydrogen fuel cells.

[0129] (7) The expression of energy storage device constraint is:

[0130]

[0131] In the formula, X represents energy storage equipment, including batteries BES and hydrogen storage tanks HES; E represents energy, including electricity P and hydrogen H; is the energy stored in energy storage device X during period t; are the power of charging and discharging energy E of energy storage device X during period t respectively; are the charging and discharging efficiencies of energy storage device X respectively; are the upper and lower limits of the energy stored in the energy storage device X; They are the charging and discharging states of the energy storage device X in period t, with a value of 1 indicating charging / discharging and a value of 0 indicating no charging / discharging; They are the upper and lower limits of the charging power of energy storage device X respectively; are the upper and lower limits of the energy storage device X’s discharge power respectively; Δt represents the scheduling period interval; represents the energy stored in energy storage device X at the end of the scheduling period; It represents the energy stored in energy storage device X at the beginning of the scheduling period.

[0132] The power balance constraints include electric power balance, thermal power balance, hydrogen power balance, cooling power balance and natural gas power balance.

[0133] The expression of electric power balance is:

[0134]

[0135] The expression of thermal power balance is:

[0136]

[0137] The expression for hydrogen power balance is:

[0138]

[0139] The expression of cold power balance is:

[0140]

[0141] The expression for natural gas power balance is:

[0142]

[0143] In the above formula, are the charging and discharging powers of the battery during period t respectively; are the power of charging and discharging hydrogen in the hydrogen storage tank during period t; P t load , They are the electric, heating and cooling load powers in period t respectively.

[0144] The expression of curtailment constraint is:

[0145]

[0146] In the formula, are the wind power and solar power abandonment in period t respectively; P t WT , P t PV are the wind power and photovoltaic power in period t respectively.

[0147] Energy purchase constraints include electricity purchase constraints and gas purchase constraints;

[0148] The expression of power purchase constraint is:

[0149]

[0150] The expression of gas purchase constraint is:

[0151]

[0152] In the above formula, They are the upper limits of electricity and gas purchase power respectively.

[0153] In summary, the upper layer distributed robust optimization model can be summarized as follows:

[0154]

[0155] In the formula, x d The decision variables in the distributed optimization model include continuous variables representing the operating power of each device and binary variables representing the start and stop status of the device; include and and are random variables of new energy uncertainty generated according to the real distribution of wind power and photovoltaic power generation; sup(·) is the supremum function; E(·) is the expectation function; P is the real distribution; is a Wasserstein fuzzy set.

[0156] Step S3: Update the power forecast data of various loads, wind power and photovoltaic power at the 15-minute level within the day according to the results of the day-ahead optimization scheduling. Use Latin hypercube sampling and K-means clustering to generate a random scenario set to characterize the uncertainty of new energy errors at a short time scale within the day. Use the random optimization model to adjust the power based on the day-ahead optimization results. Roll the calculation every 15 minutes to solve the final scheduling plan within the day, including the following steps:

[0157] Step S31: The initial scene is obtained by Latin hypercube sampling, and the K-means clustering algorithm is used to reduce the initial scene to obtain a typical scene, and the typical scene is used as a random scene set. The Latin hypercube sampling assumes that the wind power and photovoltaic errors satisfy the normal distribution, that is, the following form:

[0158]

[0159] In the formula, μ WT , σ WT are the expected value and standard deviation of wind power error respectively; μ PV , σ PV are the expected and standard deviation of the photovoltaic error, respectively.

[0160] The lower stochastic optimization model takes the total cost F of the integrated energy system as id The optimization goal is to minimize the power adjustment of each device as much as possible. The objective function adds the device power adjustment cost C on the basis of the upper layer distributed robust optimization model. adj ,Right now

[0161] Fid =C buy +C om +C adj +C cut -R CO2 ;

[0162]

[0163] In the formula, is the power adjustment cost of device r; is the power adjustment of device r during period t, including ΔP t GT , ΔP t HFC ,

[0164] The constraints of the lower-level stochastic optimization model include equipment operation constraints, power balance constraints, power abandonment constraints, and energy purchase constraints. The results of solving the upper layer distributed robust optimization model should also be followed.

[0165] In summary, the underlying stochastic optimization model can be summarized as follows:

[0166]

[0167] In the formula, x i is the decision variable of the random optimization model at the lower level, including the continuous variable representing the operating power of each device and the binary variable representing the start and stop status of the device; i include and and are the uncertain random variables of wind power and photovoltaic power generation generated according to a set of random scenarios.

[0168] So far, a hierarchical uncertainty optimization model of the integrated energy system has been established. The RSOME toolbox can be used in the MATLAB environment to build the optimization model and the Gurobi commercial solver can be called to solve it to obtain the optimal scheduling plan.

[0169] In order to verify the advantages of the hierarchical uncertainty optimization method for the integrated energy system of the embodiment of the present invention, the following four scenarios are set:

[0170] Scenario 1: Perform upper-level distributed robust optimization and directly use the upper-level optimization result as the final scheduling plan.

[0171] Scenario 2: Perform upper-layer distributed robust optimization and coordinate with lower-layer rolling random optimization to obtain the final scheduling plan, which is the technical solution of the present invention.

[0172] Scenario 3: Perform upper-layer robust optimization and coordinate with lower-layer rolling random optimization to obtain the final scheduling plan.

[0173] Scenario 4: Perform upper-layer random optimization and coordinate with lower-layer rolling random optimization to obtain the final scheduling plan.

[0174] In scenarios 1 and 2, the parameter θ in the Wasserstein fuzzy set of the upper layer's robust optimization is set to 10MW, and the number of historical data samples input by the support set is 1000. In the four scenarios, the lower layer's random scenario set assumes that the wind power error satisfies the normal distribution with an expectation of 0 and a standard deviation of 3%; the photovoltaic error satisfies the normal distribution with an expectation of 0 and a standard deviation of 2%; the number of initial scenarios generated by Latin hypercube sampling is set to 1000, and the number of typical scenarios generated by K-means clustering is set to 5.

[0175] The optimization scheduling results under different scenarios are shown in Table 1, where the unit of scheduling cost is 10 3 Yuan.

[0176] Table 1 Scheduling costs under different scenarios

[0177]

[0178] Comparing scenario 1 with scenarios 2, 3, and 4, it can be seen that the hierarchical optimization method based on multiple time scales has better economic efficiency than the optimization method based on a single time scale. Although compared with scenario 1, the optimization scheduling under multiple time scales will increase the operation and maintenance costs and power adjustment costs of the equipment, by adjusting flexible resources such as gas turbines and gas boilers during the intraday stage, the energy purchase cost in scenarios 2, 3, and 4 is reduced, and the carbon tax revenue is increased. At the same time, as the time scale is shortened, the scheduling and control of energy storage equipment is more refined, and its advantage of peak shaving and valley filling also makes the cost of power abandonment lower.

[0179] The optimization dispatch results of scenario 2, scenario 3 and scenario 4 are compared and analyzed. The upper layer of scenario 4 uses the random optimization method, which leads to the blind pursuit of economy in the optimization results, making the energy purchase cost as low as possible and the lowest total cost. However, the optimistic decision makes a large amount of wind power and photovoltaic resources wasted, and the cost of power abandonment is the highest, which is not conducive to the dispatch and operation of the integrated energy system with a high proportion of renewable energy. The upper layer of scenario 3 uses the robust optimization method. Although it can solve the problem of over-optimism in random optimization, it tends to purchase electricity from the main power grid and gas from the main gas grid to counter the uncertainty of new energy. The dispatch decision is too conservative, and the pursuit of system robustness makes the total cost the highest. The upper layer of scenario 2 uses the distributed robust optimization method, and its total cost is between scenario 3 and scenario 4. It can well balance the economy and robustness of system dispatch. On the one hand, it does not rely too much on the main power grid and the main gas grid to smooth the error of new energy. On the other hand, it avoids overly optimistic dispatch decisions, which reduces the phenomenon of wind and solar abandonment, and the cost of power abandonment is the lowest. It can effectively adapt to the consumption of high-penetration renewable energy.

[0180] The number of samples of historical data of new energy error input in upper layer distributed robust optimization in scenario 2 and upper layer random optimization in scenario 4 was changed respectively, and the number of samples was set from 100 to 2000 for simulation calculation. The optimization results are shown in Table 2, where the unit of total cost is 10 3 Yuan.

[0181] Table 2 Total cost under different sample numbers

[0182]

[0183] From the results in Table 2, it can be seen that when the upper layer in Scenario 2 uses data-driven distributed robust optimization, the total scheduling cost obtained by the final optimization decreases as the number of input samples increases, while the random optimization in Scenario 4 does not have this feature. This shows that the technical solution of the present invention can make full use of a large amount of historical data, thereby further improving the economy of system scheduling.

[0184] In order to verify the influence of the parameter θ of the Wasserstein fuzzy set in the upper-level distributed robust optimization model on the system scheduling cost, the value of the Wasserstein ball radius θ in scenario 2 is changed to perform simulation calculations. The calculation results are shown in Figure 3 As shown. Figure 3The results show that when the Wasserstein ball radius θ is 0, the Wasserstein fuzzy set only contains empirical distribution, the upper layer of the robust optimization model of scenario 2 degenerates into the same random optimization model as scenario 4, and the scheduling tends to be economical; when the Wasserstein ball radius θ gradually increases, the optimization result of scenario 2 will approach the result of the upper layer of scenario 3 for robust optimization, and the scheduling tends to be robust. The increase in the Wasserstein ball radius θ means that the uncertainty of new energy increases, so the scheduling decision maker can flexibly choose the value of the Wasserstein ball radius θ and make a trade-off between optimistic and risky decisions.

[0185] Embodiment 2

[0186] An embodiment of the present invention provides an integrated energy system optimization system based on hierarchical uncertainty, including an integrated energy system construction module, an upper layer distributed robust optimization module and a lower layer rolling random optimization module.

[0187] The integrated energy system building module is used to establish the operating framework of the integrated energy system.

[0188] The upper distributed robust optimization module is used to input the power forecast data of various loads and wind power and photovoltaic power at the 1h level before the day, and use the data-driven Wasserstein fuzzy set to characterize the uncertainty of new energy errors on a long time scale before the day. The distributed robust optimization model is used to solve the optimal scheduling plan before the day with a time interval of 1h.

[0189] The lower-level rolling random optimization module is used to update the power forecast data of various loads, wind power, and photovoltaic power at the 15-minute level within the day. Latin hypercube sampling and K-means clustering are used to generate a set of random scenarios to characterize the uncertainty of new energy errors on a short time scale within the day. The power is adjusted based on the day-ahead optimization results through the random optimization model, and the rolling calculation is performed every 15 minutes to obtain the final scheduling plan for the day.

[0190] The present invention combines multi-time scale optimization with uncertainty optimization, and solves the shortcoming of traditional uncertainty optimization that is prone to fall into local optimality through layered collaboration, so that the final scheduling scheme has better economic efficiency. By using data-driven Wasserstein fuzzy sets to characterize the uncertainty of new energy in the upper layer of distributed robust optimization, the final scheduling scheme obtained by the lower layer rolling random optimization takes into account both economy and robustness, and can effectively adapt to the impact of high proportion of wind power and photovoltaic uncertainty in the integrated energy system.

[0191] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. Only the preferred embodiments of the present invention are expressed. The description is more specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. As long as there is no contradiction in the combination of these technical features, they should be considered as the scope recorded in this specification.

[0192] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these modifications and improvements all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A comprehensive energy system optimization method based on hierarchical uncertainty, characterized in that: The following steps are involved: Step S1: taking the error between the historical power generation data and the historical predicted power generation data of the integrated energy system as the historical prediction error, constructing a prediction error fuzzy set according to the historical prediction error, and constructing a distributed robust optimization model based on the prediction error fuzzy set; Step S2: obtaining the power forecast data of the integrated energy system at the one-hour level before the day, inputting the power forecast data of the one-hour level before the day into the distributed robust optimization model, and obtaining the day-ahead scheduling plan of the integrated energy system; Step S3: Construct a set of intra-day random scenarios, and construct a random optimization model based on the intra-day random scenario set. With a period of 15 minutes, obtain intra-day power forecast data based on the day-ahead scheduling plan, input the intra-day power forecast data into the random optimization model to obtain the intra-day scheduling plan, repeat the above steps until the optimization scheduling of all cycles within the day is completed, and obtain the final scheduling plan of the integrated energy system.

2. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 1, characterized in that: Constructing a prediction error fuzzy set according to the historical prediction error in step S1 includes the following steps: Step S11: obtaining an empirical distribution of prediction errors of the integrated energy system according to the historical prediction errors, analyzing the empirical distribution to obtain characteristics of the prediction errors, obtaining predicted power generation of the integrated energy system, and constructing a true distribution of prediction errors of the integrated energy system according to the predicted power generation and the characteristics of the prediction errors; Step S12: clustering the historical prediction errors to obtain several typical scenarios of prediction errors and scenario probabilities corresponding to each typical scenario, and constructing a prediction error support set; Step S13: Calculate the difference between the empirical distribution and the true distribution, set the parameters of the fuzzy set according to the difference, and construct the prediction error fuzzy set according to the prediction error support set and the parameters of the fuzzy set.

3. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 2, characterized in that: The expression of the prediction error support set in step S12 is: In the formula, s is the prediction error support set of scenario s; is a random vector of real distribution, is a random variable that is actually distributed in time period t; T is the total number of time periods; R is the set of real numbers; are the minimum and maximum values ​​of historical forecast errors respectively; is the sth scenario of historical forecast error.

4. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 3 is characterized in that: In step S13, the difference between the empirical distribution and the true distribution is calculated by the Wasserstein probability distance, and the expression of the Wasserstein probability distance is: In the formula, For the empirical distribution and the true distribution The Wasserstein probability distance between them; inf(·) is the infimum function; For the empirical distribution and the true distribution The entire set of joint distributions Q of; E(·) is the expected function; ρ(·) is the L2 norm, which represents the basic distance of Wasserstein probability distance; is a random vector from the empirical distribution, is a random variable with empirical distribution in period t.

5. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 4, characterized in that: The expression of the prediction error fuzzy set in step S13 is: In the formula, is the true value of the error of wind power and photovoltaic power generation of the integrated energy system randomly generated from the real distribution; is an auxiliary random variable; E(·) is the expected function; The true distribution A collection of; S is the total number of typical scenarios; is a variable representing a typical scenario; p s is the scenario probability of the typical scenario s; θ is the conservative constant of the fuzzy set, which is set according to the difference between the empirical distribution and the true distribution.

6. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 1, characterized in that: The expression of the distributed robust optimization model is: In the formula, F da is the objective function of the distributed robust optimization model; x d To determine the decision variables in the robust optimization model; is an uncertain random variable generated according to the true distribution; sup(·) is the supremum function; E(·) is the expectation function; is the real distribution; is a fuzzy set.

7. The method for optimizing a comprehensive energy system based on hierarchical uncertainty according to claim 1, characterized in that: The objective function of the distributed robust optimization model is expressed as: In the formula, C buy C is the energy purchase cost; om C is the operation and maintenance cost; cut is the cost of power abandonment; R CO2 is the carbon tax revenue; T is the total number of time periods; p gas are electricity and gas prices respectively; are the power of electricity and gas purchased in period t respectively; R is the total number of devices in the integrated energy system; is the operation and maintenance cost of device r; is the operating power of device r during period t; are the penalty costs for abandoning wind and solar power respectively; c CO2 is the carbon tax price, δ re , δ q , δ g are the carbon emission coefficients of renewable energy, main gas grid, and main power grid respectively; are wind power and photovoltaic power in period t respectively; is the amount of natural gas consumed by the gas turbine during period t; is the amount of natural gas consumed by the gas boiler during period t.

8. The method for optimizing a comprehensive energy system based on hierarchical uncertainty according to claim 1, characterized in that: The constraints of the distributed robust optimization model include equipment operation constraints, power balance constraints, power abandonment constraints and energy purchase constraints; The equipment operation constraints include gas turbine constraints, gas boiler constraints, electric refrigerator constraints, absorption refrigerator constraints, electrolyzer constraints, hydrogen fuel cell constraints and energy storage equipment constraints; The expression of the gas turbine constraint is: In the above formula, P t GT , are the power generation power and heating power of the gas turbine during period t respectively; is the amount of natural gas consumed by the gas turbine during period t; are the power generation efficiency and heating efficiency of the gas turbine respectively; are the upper and lower limits of the gas turbine ramp rate, respectively; are the upper and lower limits of the gas turbine power generation respectively; is the on / off state of the gas turbine during period t, with a value of 1 indicating on and a value of 0 indicating off; The expression of the gas boiler constraint is: In the formula, is the amount of natural gas consumed by the gas boiler during period t; is the heating power of the gas boiler during period t; is the heating efficiency of the gas boiler; They are the upper and lower limits of the heating power of the gas boiler respectively; The expression of the electric refrigerator constraint is: In the formula, is the cooling power of the electric refrigerator during period t; is the power consumption of the electric refrigerator during period t; is the cooling efficiency of the electric refrigerator; They are the upper and lower limits of the refrigeration power of the electric refrigerator respectively; The expression of the absorption refrigerator constraint is: In the formula, is the cooling power of the absorption chiller during period t; is the heat consumption power of the absorption refrigerator during period t; is the refrigeration efficiency of the absorption chiller; are the upper and lower limits of the refrigeration power of the absorption chiller The expression of the electrolyzer constraint is: In the formula, is the hydrogen production power of the electrolyzer during period t; P t EL is the power consumption of the electrolytic cell during period t; is the hydrogen production efficiency of the electrolyzer; They are the upper and lower limits of the power consumption of the electrolytic cell respectively; The expression of the hydrogen fuel cell constraint is: Where P t HFC is the power generation of the hydrogen fuel cell during period t; is the hydrogen consumption power of the hydrogen fuel cell during period t; The power generation efficiency of hydrogen fuel cells; They are the upper and lower limits of hydrogen power consumption of hydrogen fuel cells respectively; The expression of the energy storage device constraint is: In the formula, X represents energy storage equipment, including batteries BES and hydrogen storage tanks HES; E represents energy, including electricity P and hydrogen H; is the energy stored in energy storage device X during period t; are the power of charging and discharging energy E of energy storage device X during period t respectively; are the charging and discharging efficiencies of energy storage device X respectively; are the upper and lower limits of the energy stored in the energy storage device X; They are the charging and discharging states of the energy storage device X in period t, with a value of 1 indicating charging / discharging and a value of 0 indicating no charging / discharging; They are the upper and lower limits of the charging power of energy storage device X respectively; are the upper and lower limits of the energy storage device X’s discharge power respectively; Δt represents the scheduling period interval; represents the energy stored in energy storage device X at the end of the scheduling period; represents the energy stored in energy storage device X at the beginning of the scheduling period; The power balance constraints include electric power balance, thermal power balance, hydrogen power balance, cold power balance and natural gas power balance; The expression of the electric power balance is: The expression of thermal power balance is: The expression of the hydrogen power balance is: The expression of the cooling power balance is: The expression of the natural gas power balance is: In the above formula, are the charging and discharging powers of the battery during period t respectively; are the power of charging and discharging hydrogen in the hydrogen storage tank during period t; P t load , are the electric, heating and cooling load powers in period t respectively; The expression of the power curtailment constraint is: In the formula, are the wind power and solar power abandonment in period t respectively; P t WT , P t PV are wind power and photovoltaic power in period t respectively; The energy purchase constraints include electricity purchase constraints and gas purchase constraints; The expression of the power purchase constraint is: The expression of the gas purchase constraint is: In the above formula, They are the upper limits of electricity and gas purchase power respectively.

9. The method for optimizing an integrated energy system based on hierarchical uncertainty according to claim 1, characterized in that: In step S3, an initial scene is obtained by Latin hypercube sampling, and a K-means clustering algorithm is used to reduce the initial scene to obtain several typical scenes, and the several typical scenes are used as a random scene set.

10. The method for optimizing a comprehensive energy system based on hierarchical uncertainty according to claim 7 is characterized in that: The expression of the stochastic optimization model is: F id =C buy +C om +C adj +C cut -R CO2 ; In the formula, x i is the decision variable of the stochastic optimization model; ξ i is an uncertain random variable generated according to a set of random scenarios; C adj Adjust costs for equipment power; is the power adjustment cost of device r; is the power adjustment amount of device r during period t.

Images