Optimized scheduling method and device for multi-uncertainty integrated energy system

Abstract

The invention discloses an optimal scheduling method and device for a multi-uncertainty integrated energy system, and the method comprises the steps: building a double-layer optimal scheduling model of the integrated energy system, enabling an upper-layer model to be built with the energy purchasing cost as a target, and enabling a lower-layer model to be built with the energy selling income as a target; converting the lower-layer model and the constraint condition thereof into a constraint condition of an upper-layer model to obtain a single-layer optimization scheduling model of the integrated energy system; and constructing an opportunity pursuit model and a risk avoidance model based on an information gap decision, carrying out weight distribution on uncertain factors obtained by calculation of the opportunity pursuit model or the risk avoidance model, taking the uncertain factors as input parameters of a single-layer optimization scheduling model, and solving the single-layer optimization scheduling model to obtain a scheduling scheme. By adopting the technical scheme, the scheduling scheme obtained by solving is prevented from falling into local optimum, and the scheduling scheme which effectively reduces the operation cost or ensures the operation stability and meets the actual requirements is obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated energy system scheduling in a power grid system, and in particular to an optimization scheduling method and device for an integrated energy system with multiple uncertainties. Background Art

[0002] As the energy structure shifts toward multi-energy complementarity, integrated energy systems (IES) must simultaneously participate in multiple energy markets, including electricity, gas, heat, and cooling, and coordinate resources such as sources, loads, and storage to improve economic efficiency and reliability. However, existing technologies face technical bottlenecks in addressing the multiple uncertainties in IES optimal scheduling.

[0003] Existing technologies mostly employ deterministic load models and a single stochastic optimization framework. Deterministic load models struggle to capture the dynamic and adjustable characteristics of flexible loads (such as the curtailment / shifting ratio depending on electricity prices and fluctuations in user behavior). Single stochastic optimization frameworks rely on a large number of discrete scenarios and require pre-defined probability distributions, making it difficult to balance computational efficiency and robustness. Furthermore, stochastic optimization, due to the large number of scenarios, incurs high computational costs, while robust optimization requires conservatively setting fluctuation ranges, which can lead to overly pessimistic scheduling plans and result in revenue losses.

[0004] To address the above issues, a two-layer model architecture was proposed. However, in its application, the nonlinear complementary constraints of the lower-layer model are difficult to solve directly, and it often relies on heuristic algorithms (such as genetic algorithms). This leads to slow convergence and easy trapping in local optimality. The resulting scheduling solution does not meet the actual needs and expectations of all participants. For example, the operating cost cannot be reduced to the lowest range, or the operation stability is lacking.

[0005] In addition, the application model of information gap decision theory (IGDT) in the existing technology is mainly constructed for a single uncertain factor, failing to fully consider the relationship between market price signals and flexible load responses. The resulting scheduling scheme also does not meet actual needs and the expectations of all participants. Summary of the Invention

[0006] Purpose of the invention: The present invention provides an optimization scheduling method and device for an integrated energy system with multiple uncertainties, aiming to solve the problem in the prior art that it is difficult to obtain a scheduling solution that effectively reduces operating costs or ensures operational stability and meets actual needs for an integrated energy system with multiple uncertainties.

[0007] Technical solution: The present invention provides an optimization scheduling method for a multi-uncertainty integrated energy system, comprising: establishing a two-layer optimization scheduling model for the integrated energy system; the two-layer optimization scheduling model comprises an upper model and a lower model, wherein the upper model is established based on the energy purchase quantity and energy purchase price of the integrated energy system, with the energy purchase cost as the target, and the lower model is established based on the transfer load and load reduction, as well as the energy sales quantity and energy sales price of the integrated energy system, with the energy sales revenue as the target; the integrated energy system comprises traditional energy, as well as wind power and photovoltaic power; the lower model and its constraints are converted into the constraints of the upper model The information gap decision-making process is to build an opportunity pursuit model and a risk aversion model based on information gap decision-making. The uncertainties calculated by the opportunity pursuit model or the risk aversion model are used as input parameters of the single-layer optimization scheduling model after weight allocation. The single-layer optimization scheduling model is solved to obtain the energy purchase quantity of the integrated energy system and the output power parameters of traditional energy in the integrated energy system, and then they are executed.

[0008] Specifically, the upper-level model is a function established with the goal of minimizing the energy purchase cost, which is calculated based on the electricity purchase quantity, electricity purchase price, gas purchase quantity, and gas purchase price of the integrated energy system. The upper-level model is as follows: minF=∑ T t=1 λ BE P t BE +∑ T t=1 λ BG P t BG , Where F represents the energy purchase cost of the integrated energy system, T represents the total number of time periods, and λ BE and λ BG Represent the unit price of electricity and gas respectively, P t BE and P t BG They represent the electricity purchase power and gas purchase power during period t respectively.

[0009] Specifically, the lower-level model is a function established with the goal of maximizing energy sales revenue, based on the incentives corresponding to the electric load power and thermal load power transferred by the integrated energy system in each time period, the compensation corresponding to the reduced electric load power and thermal load power, and the unit price of electricity, heat, gas, and cooling, as well as the corresponding load power. The lower-level model is as follows: maxR=∑ T t=1 (δ e P t CL +δ h H t CL +ξ e P t SL +ξ h H t SL +λ SE P t SE +ρ E P t load +ρ H H t load +ρ G G t load +ρ C C t load ), Among them, R represents the energy sales revenue, δ e and δ h Represents the unit compensation coefficient for reducing electrical load and reducing thermal load, P t CL and H t CL They represent the reduced electric load and heat load during period t, ξ e and ξ h They represent the unit excitation coefficients of the transferred electrical load and the transferred thermal load, P t SL and H t SL They represent the electrical load and thermal load transferred during period t, respectively, SE represents the unit price of electricity sold to the energy market, P t SE represents the electricity sales power in period t, ρ E , ρ H , ρ G and ρ C They represent the unit price of electricity, heat, gas and cooling sold to the load side respectively, tload 、H t load , G t load and C t load They represent the electric load power, thermal load power, gas load power and cooling load power in period t respectively.

[0010] Specifically, the upper-level model also includes electric power balance constraints for electricity consumption and power generation balance, thermal power balance constraints for heat consumption and heat generation balance, gas power balance constraints for gas consumption and gas discharge balance, cooling power balance constraints for cooling and refrigeration balance, and output power ramp-up constraints for traditional energy units; the lower-level model also includes upper and lower limit constraints for transferred and reduced electric load power and thermal load power.

[0011] Specifically, the Lagrangian function corresponding to the lower-level model is established, and based on the complementary relaxation condition of KKT, the Lagrangian function corresponding to the lower-level model is converted into the nonlinear constraint condition of the upper-level model, and the nonlinear constraint condition is converted into the linear constraint condition of the upper-level model through the big M method.

[0012] Specifically, the linear constraints of the upper model are as follows: 0≤P t CL -P t CL,min γ≤ε1M1,0≤τ1≤(1-ε1)M1; 0≤P t CL,max γ-P t CL ≤ε2M2,0≤τ2≤(1-ε2)M2; 0≤P t SL -P t SL,min φ≤ε3M3,0≤τ3≤(1-ε3)M3; 0≤P t SL,max φ-P t SL ≤ε4M4,0≤τ4≤(1-ε4)M4; 0≤H t CL -H t CL,min β≤ε5M5,0≤τ5≤(1-ε5)M5; 0≤H t CL,max β-H t CL≤ε6M6,0≤τ6≤(1-ε6)M6; 0≤H t SL -H t SL,min σ≤ε7M7,0≤τ7≤(1-ε7)M7; 0≤H t SL,max σ-H t SL ≤ε8M8,0≤τ8≤(1-ε8)M8; Among them, P t CL,min and P t CL,max They represent the lower limit and upper limit of load power reduction in period t, P t SL,min and P t SL,max They represent the lower and upper limits of the load power transferred during period t, respectively. t CL,min and H t CL,max They represent the lower and upper limits of heat load reduction during period t, H t SL,min and H t SL,max represent the lower and upper limits of the transferred thermal load power in period t, respectively; γ, φ, β, and σ represent the state variables of the corresponding reduced or transferred electrical load or thermal load, ranging from 0 to 1; ε1 to ε8 represent the corresponding Boolean variables; τ1 to τ8 represent the Lagrange multipliers of the corresponding constraints; and M1 to M8 represent the corresponding boundary range definition parameters.

[0013] Specifically, the attribute values of the uncertain factors in the integrated energy system are normalized; the information entropy of each uncertain factor is calculated according to the information entropy formula, and the weight of the corresponding uncertain factor is determined by the value of the information entropy; the weight of each uncertain factor is calculated using the following formula: w i =(1-E i ) / (n-∑ n i=1 E i ), Among them, w i represents the weight of the i-th uncertainty factor, E i represents the information entropy of the i-th uncertain factor, and n represents the total number of uncertain factors; The information entropy of uncertain factors is calculated using the following formula: E i =-ln(T) -1 ∑T t=1 p it ln(p it ), Among them, p it It represents the proportion of the attribute value of the i-th uncertainty factor in period t to the attribute values of all periods.

[0014] Specifically, the opportunity pursuit model is used to utilize the fluctuation of uncertain factors to maximize the benefits of the integrated energy system; the risk aversion model is used to meet the needs of various types of loads when uncertain factors fluctuate.

[0015] Specifically, the opportunity pursuit model is as follows: minα, minf≤f0(1-δ c ), P t i ∈U(α i ,P t i,pre ); Where α represents the total uncertainty, α i represents the uncertainty of the i-th uncertain factor, f represents the energy purchase cost, f0 represents the preset value of the energy purchase cost, δ c represents the chance deviation coefficient, P t i represents the power parameter of the i-th uncertainty factor during period t, P t i,pre represents the predicted power of the ith uncertainty factor in period t, and the set U(α i ,P t i,pre ) as the input parameter of the single-layer optimization scheduling model; The risk aversion model is as follows: maxα, maxf≤f0(1+δ R ), P t i ∈U(α i ,P t i,pre ); Among them, δ R represents the robust bias coefficient.

[0016] The present invention also provides an optimization scheduling device for a multi-uncertainty integrated energy system, comprising: a model building unit, a model conversion unit, a weight allocation unit and a power parameter calculation unit, wherein: the model building unit is used to establish a two-layer optimization scheduling model for the integrated energy system; the two-layer optimization scheduling model comprises an upper model and a lower model, wherein the upper model is based on the energy purchase quantity and energy purchase price of the integrated energy system, and is established with the energy purchase cost as a target, and the lower model is based on the transfer load and load reduction, as well as the energy sales quantity and energy sales price of the integrated energy system, and is established with the energy sales income as a target; the integrated energy system includes traditional energy, as well as wind power and photovoltaic power; the model conversion unit is used to convert the lower model and its approximate value into a fixed value. The constraint conditions are converted into constraint conditions of the upper-level model and incorporated into the calculation of the upper-level model to obtain a single-level optimization scheduling model of the integrated energy system; the weight allocation unit is used to perform weight allocation on the uncertainty of wind power output power and photovoltaic output power in the integrated energy system, as well as the uncertainty of various types of loads; the power parameter calculation unit is used to construct an opportunity pursuit model and a risk aversion model based on information gap decision-making, and the uncertainty factors calculated by the opportunity pursuit model or the risk aversion model are used as input parameters of the single-level optimization scheduling model after weight allocation, and the single-level optimization scheduling model is solved to obtain the energy purchase quantity and energy purchase price of the integrated energy system, as well as the output power parameters of traditional energy in the integrated energy system, and then execute them.

[0017] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: converting the lower-level model and its constraints into the constraints of the upper-level model to avoid the scheduling scheme obtained from the solution from falling into the local optimum; introducing the uncertain factors of wind power and photovoltaic output and load demand, and assigning corresponding weights, combined with the information gap decision model, on the basis of fully considering the needs of all participants, solving a scheduling scheme that effectively reduces operating costs or ensures operational stability and meets actual needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flow chart of the optimization scheduling method for the integrated energy system provided by the present invention; Figure 2 This is a structural diagram of the integrated energy system for combined heat, electricity and air cooling provided by the present invention. DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0020] See Figure 1 , which is a flow chart of the optimization scheduling method for the integrated energy system provided by the present invention.

[0021] In an embodiment of the present invention, a two-layer optimization scheduling model for an integrated energy system is established; the two-layer optimization scheduling model includes an upper-layer model and a lower-layer model.

[0022] In an embodiment of the present invention, the upper-level model is established based on the energy purchase quantity and price of the integrated energy system, with the energy purchase cost as the target, and the lower-level model is established based on the transfer load and load reduction, as well as the energy sales quantity and price of the integrated energy system, with the energy sales revenue as the target.

[0023] In an embodiment of the present invention, the integrated energy system includes traditional energy, wind power and photovoltaic power.

[0024] See Figure 2 , which is a structural diagram of the integrated energy system for combined heat, electricity and air cooling provided by the present invention.

[0025] In the specific implementation, wind power, photovoltaics, gas turbines, and electric energy storage equipment inside the integrated energy system supply power to electric loads and electric refrigerators. Electric refrigerators and cold energy storage equipment supply energy to cold loads. Gas energy storage meets the natural gas needs of gas loads, gas turbines, and gas boilers. Gas boilers, waste heat recovery devices, and thermal energy storage equipment provide heat to thermal loads.

[0026] In practice, traditional energy sources in an integrated energy system include gas turbines and boilers, while new energy sources include wind power and photovoltaics. The energy generated by the integrated energy system is sold to the energy market, generating revenue, and to the load side, generating revenue. Furthermore, the system can also receive compensation or incentives from the demand response market based on the amount of electrical and thermal loads reduced or shifted. Furthermore, when the energy generated by the integrated energy system itself cannot meet the load demand, it must purchase energy from the energy market, incurring a cost. Clearly, the load side expects to meet its power demands at all times, while the integrated energy system expects to meet the load power demands while maximizing revenue and minimizing expenditure.

[0027] In an embodiment of the present invention, the upper model is a function established with the goal of minimizing the energy purchase cost, which is calculated based on the electricity purchase quantity, electricity purchase price, gas purchase quantity and gas purchase price of the integrated energy system.

[0028] In the embodiment of the present invention, the upper layer model is as follows: minF=∑ T t=1 λ BE P t BE +∑ T t=1 λ BG P t BG, Where F represents the energy purchase cost of the integrated energy system, T represents the total number of time periods, and λ BE and λ BG Represent the unit price of electricity and gas respectively, P t BE and P t BG They represent the electricity purchase power and gas purchase power during period t respectively.

[0029] In an embodiment of the present invention, the upper model also includes electric power balance constraints for balancing electricity consumption and power generation, thermal power balance constraints for balancing heat consumption and heat generation, gas power balance constraints for balancing gas consumption and gas discharge, cooling power balance constraints for balancing cooling consumption and refrigeration, and output power ramping constraints for traditional energy units.

[0030] In a specific implementation, the electric power balance constraints are as follows: P t load +P t SE +P t SL +P t ER +P t ESch = P t WT +P t PV +P t GT +P t BE +P t CL +P t ESdis , Among them, the left side of the equation P t load 、P t SE 、P t SL 、P t ER and P t ESch They represent the electric load power, IES electricity sales power, transferable electric load power, electric refrigerator power consumption and energy storage device charging power in period t respectively. The right side of the equation is P t WT 、P t PV 、P t GT 、P t BE 、P tCL and P t ESdis They represent the power generation power of wind power generation equipment, photovoltaic power generation equipment, gas turbine power generation power, IES power purchase power, curtailable load power and energy storage device discharge power in period t respectively.

[0031] In the specific implementation, the thermal power balance constraints are as follows: H t load +H t SL +P t HSch = H t GB +H t CL +H t UH +P t HSdis , Among them, the left side of the equation H t load 、H t SL and P t HSch They represent the heat load power, transferable heat load power and heat storage power of the energy storage device during period t respectively. The right side of the equation H t GB 、H t CL 、H t UH and P t HSdis They represent the thermal power of the gas boiler, the power of the heat load that can be reduced, the heating power of the waste heat recovery device, and the heat release power of the energy storage device during period t respectively.

[0032] In the specific implementation, the gas power balance constraints are as follows: G t load +G t GT +G t GB +P t GSch = P t BG +P t GSdis , Among them, the left side of the equation G t load , G t GT , G t GB and P tGSch They represent the gas load power, gas turbine natural gas consumption, gas boiler natural gas consumption, and energy storage device gas storage power in period t respectively. The right side of the equation is P t BG and P t GSdis They represent the IES gas purchasing power and energy storage device gas discharging power in period t respectively.

[0033] In the specific implementation, the cooling power balance constraints are as follows: C t load +P t CSch = C t ER +P t CSdis , Among them, the left side of the equation C t load and P t CSch They represent the cooling load power and the cooling power of the energy storage device during period t, respectively. The C on the right side of the equation t ER and P t CSdis They represent the cooling power of the electric refrigerator and the cooling power of the energy storage device during period t respectively.

[0034] In specific implementation, the output power ramp constraints of traditional energy units are as follows: P down GT ≤P t GT -P t-1 GT ≤P up GT , H down GB ≤H t GB -H t-1 GB ≤H up GB , Among them, P down GT and P up GT Represent the downward and upward ramp rates of the gas turbine, H down GB and H up GB Respectively represent the downward and upward climbing rates of the gas boiler.

[0035] In specific implementation, the upper-level model also includes upper and lower limit constraints on the output power of traditional energy units, namely the upper and lower limits of gas turbine power generation power, the upper and lower limits of waste heat recovery device heating power, the upper and lower limits of gas boiler power generation power, and the upper and lower limits of refrigeration unit cooling power.

[0036] In an embodiment of the present invention, the lower-level model is a function established with the goal of maximizing the energy sales revenue, based on the incentives corresponding to the electric load power and thermal load power transferred by the integrated energy system in each time period, the compensation corresponding to the reduced electric load power and thermal load power, and the unit price of electricity, heat, gas, and cooling, as well as the corresponding load power.

[0037] In the embodiment of the present invention, the lower layer model is as follows: maxR=∑ T t=1 (δ e P t CL +δ h H t CL +ξ e P t SL +ξ h H t SL +λ SE P t SE +ρ E P t load +ρ H H t load +ρ G G t load +ρ C C t load ), Among them, R represents the energy sales revenue, δ e and δ h Represents the unit compensation coefficient for reducing electrical load and reducing thermal load, P t CL and H t CL They represent the reduced electric load and heat load during period t, ξ e and ξ h They represent the unit excitation coefficients of the transferred electrical load and the transferred thermal load, P t SL and H t SL They represent the electrical load and thermal load transferred during period t, respectively, SErepresents the unit price of electricity sold to the energy market, P t SE represents the electricity sales power in period t, ρ E , ρ H , ρ G and ρ C They represent the unit price of electricity, heat, gas and cooling sold to the load side respectively, t load 、H t load , G t load and C t load They represent the electric load power, thermal load power, gas load power and cooling load power in period t respectively.

[0038] In this embodiment of the present invention, the lower-level model further includes upper and lower constraints on the power of the electrical and thermal loads that can be transferred or reduced. Specifically, the constraints include the upper and lower limits for the electrical load that can be reduced, the upper and lower limits for the electrical load that can be transferred, the upper and lower limits for the thermal load that can be reduced, and the upper and lower limits for the thermal load that can be transferred during time period t.

[0039] In a specific implementation, the transferred electric or thermal load means that the electric or thermal power originally used in a certain period (such as peak period) is changed to another period (such as off-peak period); the reduced electric or thermal load means that the used electric or thermal load is reduced.

[0040] In an embodiment of the present invention, the lower-level model and its constraints are converted into constraints of an upper-level model and incorporated into the calculation of the upper-level model to obtain a single-level optimization scheduling model for an integrated energy system.

[0041] In an embodiment of the present invention, a Lagrangian function corresponding to the lower-level model is established, and based on the complementary relaxation condition of KKT, the Lagrangian function corresponding to the lower-level model is converted into a nonlinear constraint condition of the upper-level model, and the nonlinear constraint condition is converted into a linear constraint condition of the upper-level model through the big M method.

[0042] In the specific implementation, there is a coupling relationship between the upper-level model and the lower-level model. By establishing the Lagrangian function of the lower-level model and based on the KKT complementary relaxation conditions (Karush-Kuhn-Tucker Conditions) of the lower-level model, the lower-level model can be converted into the constraint conditions of the upper-level model, that is, the two-level model is converted into a single-level model.

[0043] In practice, the complementary slack condition is a nonlinear expression that cannot be accurately solved using classical algorithms. The transformed nonlinear single-layer model is linearized using the Big M method (the Big M method converts nonlinear expressions such as the complementary slack condition into linear constraints, making the problem solvable using a MILP solver. In this invention, the Big M method is used to linearize the lower-level problem of the two-layer model, thereby integrating it into the upper-level model, ultimately forming a single-layer optimization problem). By introducing Big M and Boolean variables, the nonlinear constraints can be converted into linear constraints.

[0044] In the embodiment of the present invention, the linear constraints of the upper-level model obtained by converting the lower-level model are as follows: 0≤P t CL -P t CL,min γ≤ε1M1,0≤τ1≤(1-ε1)M1; 0≤P t CL,max γ-P t CL ≤ε2M2,0≤τ2≤(1-ε2)M2; 0≤P t SL -P t SL,min φ≤ε3M3,0≤τ3≤(1-ε3)M3; 0≤P t SL,max φ-P t SL ≤ε4M4,0≤τ4≤(1-ε4)M4; 0≤H t CL -H t CL,min β≤ε5M5,0≤τ5≤(1-ε5)M5; 0≤H t CL,max β-H t CL ≤ε6M6,0≤τ6≤(1-ε6)M6; 0≤H t SL -H t SL,min σ≤ε7M7,0≤τ7≤(1-ε7)M7; 0≤H t SL,max σ-H t SL ≤ε8M8,0≤τ8≤(1-ε8)M8; Among them, Pt CL,min and P t CL,max They represent the lower limit and upper limit of load power reduction in period t, P t SL,min and P t SL,max They represent the lower and upper limits of the load power transferred during period t, respectively. t CL,min and H t CL,max They represent the lower and upper limits of heat load reduction during period t, H t SL,min and H t SL,max where γ, φ, β, and σ represent the state variables of the corresponding reduced or transferred electrical or thermal loads, respectively, and range from 0 to 1. ε1 to ε8 represent the corresponding Boolean variables. τ1 to τ8 represent the Lagrange multipliers of the corresponding constraints. M1 to M8 represent the corresponding boundary range parameters. M1 to M8 are sufficiently large positive numbers used to define the boundary range of the constraints.

[0045] In practice, a two-level optimization model establishes a master-slave relationship between the upper and lower models. The upper model is the dominant party, responsible for setting the global optimization objective (e.g., minimizing cost). The lower model is the subordinate party, responsible for performing local optimization (e.g., maximizing revenue) based on the upper-level decision. The lower model is converted into a constraint for the upper model, as the upper-level decision must consider the optimal response of the lower model. Specifically, the upper-level model's decision affects the input parameters of the lower model. The lower model optimizes itself based on the upper-level decision and feeds the results back to the upper model. The upper model must use the optimal response of the lower model as a constraint to ensure its own decision is globally optimal. Including the lower model as a constraint in the upper model's calculations accurately reflects the impact of the upper-level decision on the lower layer and the feedback from the lower layer to the upper layer. If the upper-level decision fails to reflect the dominant role of the lower layer, the results may not meet the upper model's global optimization objective, resulting in a local optimum rather than a global optimum.

[0046] In the embodiment of the present invention, weights are allocated to the uncertainties of wind power output power and photovoltaic output power in the integrated energy system, as well as the uncertainties of power demands of various types of loads.

[0047] In practice, wind power and photovoltaic power output in integrated energy systems is subject to volatility and uncertainty. Therefore, as an uncertain factor, the power demand curves for various load types (electricity, heat, gas, and cooling) are not exactly the same every day and are subject to volatility and uncertainty. Therefore, these factors also constitute uncertainties. Wind power and photovoltaic power output represent energy output, while various load types represent energy consumption. Incorporating both output and consumption as uncertainties in the calculation fully accounts for energy output volatility, ensuring a stable energy supply. It also optimizes the output power curves of traditional energy sources, maximizing cost reductions, maximizing returns, and meeting the needs of all stakeholders.

[0048] In the embodiment of the present invention, the attribute values of the uncertain factors in the integrated energy system are normalized.

[0049] In a specific implementation, for a given set of n uncertain factors with T features, where the i-th uncertain factor is v i ={v i1, v i2 ,…v it …v iT}, where v it It represents the attribute value of the i-th uncertainty factor in time period t (usually expressed as power).

[0050] The normalized calculation formula is as follows: v it '=(v it -min(v i )) / (max(v i )-min(v i )), Among them, v it ' indicates v it Normalized value.

[0051] In the specific implementation, the normalized values are used to calculate the proportion of the attribute value of the i-th uncertainty factor in the t period to the attribute values of all periods. The formula is as follows: p it = v it ' / ∑ T t=1 v it ', Among them, p it It represents the proportion of the attribute value of the i-th uncertainty factor in period t to the attribute values of all periods.

[0052] In the embodiment of the present invention, the information entropy of each uncertain factor is calculated according to the information entropy formula, which is as follows: E i =-ln(T) -1 ∑ Tt=1 p it ln(p it ), Among them, E i Represents the information entropy of the i-th uncertain factor.

[0053] In the embodiment of the present invention, the weight of the corresponding uncertainty factor is determined by the value of information entropy, and the formula is as follows: w i =(1-E i ) / (n-∑ n i=1 E i ), Among them, w i represents the weight of the i-th uncertain factor, n represents the total number of uncertain factors, ∑ n i=1 w i =1.

[0054] In the specific implementation, the uncertainty of the uncertainty factor is α i =w i α, α represents the total uncertainty, α i represents the uncertainty of the i-th uncertain factor. The total uncertainty represents the combined impact of all uncertainties in the system. In the operation and scheduling of integrated energy systems, the volatility of wind power and photovoltaic output, as well as the random fluctuations of electricity, heating, gas, and cooling loads, may lead to the possibility of deviations between actual and predicted values.

[0055] In practice, weighting uncertain factors is one of the key improvements of the present invention. The weights are determined using the information entropy values of the uncertain factors. According to the formula for calculating information entropy values in this invention, the information entropy value reflects the magnitude of fluctuations in the attribute value (power) of the uncertain factor. Higher information entropy values indicate greater fluctuations in the attribute value, indicating greater uncertainty and the need for greater redundancy, thus assigning a higher weight. Therefore, through a reasonable weighting approach, the impact of fluctuations in uncertain factors on the operation of the integrated energy system can be fully considered, ensuring stable system operation and meeting the needs of all parties involved.

[0056] In an embodiment of the present invention, an opportunity pursuit model and a risk aversion model are constructed based on information gap decision-making. The uncertainty factors calculated by the opportunity pursuit model or the risk aversion model are used as input parameters of a single-layer optimization scheduling model after weight allocation. The single-layer optimization scheduling model is solved to obtain the energy purchase quantity of the integrated energy system and the output power parameters of traditional energy in the integrated energy system, and then execute them.

[0057] In this embodiment of the present invention, different decision makers have different risk preferences, which manifest as opportunity seeking and risk aversion. Therefore, the Information Gap Decision Theory (IGDT) is introduced to construct a scheduling optimization model that considers both opportunity seeking and risk aversion.

[0058] In practice, the opportunity-seeking model (opportunistic IGDT model) maximizes the use of favorable fluctuations in uncertainty factors while ensuring stable system operation, reducing costs and improving system economic returns. For example, it flexibly adjusts power generation plans to capitalize on opportunities when actual power output exceeds forecasts to generate additional revenue. The opportunistic IGDT model seeks the minimum uncertainty required to maintain dispatch costs within expectations when actual output and load fluctuate within the uncertainty range.

[0059] In the embodiment of the present invention, the opportunity pursuit model is as follows: minα, minf≤f0(1-δ c ), P t i ∈U(α i ,P t i,pre ); Where α represents the total uncertainty, α i represents the uncertainty of the i-th uncertain factor, f represents the energy purchase cost (the output of the single-layer optimization scheduling model), f0 represents the preset value of the energy purchase cost (the determined output of the single-layer optimization scheduling model), δ c represents the chance deviation coefficient, P t i represents the power parameter of the i-th uncertainty factor during period t, P t i,pre represents the predicted power of the ith uncertainty factor in period t, and the set U(α i ,P t i,pre ) is used as the input parameter of the single-layer optimization scheduling model (uncertainty has been given a weight). By solving the single-layer model, the optimal scheduling scheme of the system under the favorable fluctuation of uncertain factors is obtained. The scheme can make full use of the favorable fluctuation of uncertain factors and reduce the scheduling cost of the system under the premise of ensuring the stable operation of the system.

[0060] In practice, the opportunity deviation coefficient reflects the tolerance for system returns that may be lower than expected. Larger values indicate a greater willingness to accept greater return fluctuations in pursuit of higher returns, while smaller values indicate a preference for conservative stability, hoping that system returns remain as close to expected values as possible. The opportunity deviation coefficient typically ranges from [0, 1].

[0061] In practice, the risk-averse model (robust IGDT model) ensures that the system can meet the power demands of various load types and maintain stable operation even in the face of uncertainty, avoiding the additional costs and operational risks caused by uncertainty. The risk-averse model seeks the maximum uncertainty within which dispatch costs remain within expectations when actual output and load vary within the uncertainty range.

[0062] In the embodiment of the present invention, the risk aversion model is as follows: maxα, maxf≤f0(1+δ R ), P t i ∈U(α i ,P t i,pre ); Among them, δ R Represents the robust deviation coefficient, and the uncertainty factor set U(α i ,P t i,pre ) is used as the input parameter of the single-layer optimization scheduling model (uncertainty has been assigned a weight). By solving the single-layer model, the optimal scheduling scheme for the system under adverse fluctuations of uncertain factors is obtained. This scheme can ensure that the system can still operate stably in an uncertain environment and solve the scheduling cost in this case.

[0063] In practice, the robust deviation coefficient reflects the tolerance for system costs that may be higher than expected. Larger values indicate a willingness to accept higher costs in exchange for greater stability, while smaller values indicate a preference for lower costs. The robust deviation coefficient typically ranges from [0 to 1].

[0064] In specific implementation, by solving the single-layer optimization scheduling model, the operating parameters of the integrated energy system, namely the scheduling plan, can be obtained, which mainly includes the purchase quantity of different types of energy (electricity and gas) in multiple time periods, and the output power parameters of traditional energy in multiple time periods, namely the output power curve, and then executed.

[0065] The present invention also provides an optimization scheduling device for a multi-uncertainty integrated energy system, comprising: a model building unit, a model conversion unit, a weight allocation unit and a power parameter calculation unit, wherein: the model building unit is used to establish a two-layer optimization scheduling model for the integrated energy system; the two-layer optimization scheduling model comprises an upper model and a lower model, wherein the upper model is based on the energy purchase quantity and energy purchase price of the integrated energy system, and is established with the energy purchase cost as a target, and the lower model is based on the transfer load and load reduction, as well as the energy sales quantity and energy sales price of the integrated energy system, and is established with the energy sales income as a target; the integrated energy system includes traditional energy, as well as wind power and photovoltaic power; the model conversion unit is used to convert the lower model and its approximate value into a fixed value. The constraint conditions are converted into constraint conditions of the upper-level model and incorporated into the calculation of the upper-level model to obtain a single-level optimization scheduling model of the integrated energy system; the weight allocation unit is used to perform weight allocation on the uncertainty of wind power output power and photovoltaic output power in the integrated energy system, as well as the uncertainty of various types of loads; the power parameter calculation unit is used to construct an opportunity pursuit model and a risk aversion model based on information gap decision-making, and the uncertainty factors calculated by the opportunity pursuit model or the risk aversion model are used as input parameters of the single-level optimization scheduling model after weight allocation, and the single-level optimization scheduling model is solved to obtain the energy purchase quantity and energy purchase price of the integrated energy system, as well as the output power parameters of traditional energy in the integrated energy system, and then execute them.

[0066] In a specific implementation, the optimization scheduling device for a multi-uncertainty integrated energy system provided by the present invention, wherein the execution unit for executing functions, steps or methods, the functions, steps or methods executed by it can refer to the optimization scheduling method for a multi-uncertainty integrated energy system provided by the present invention.

Claims

1. An optimization scheduling method for a multi-uncertainty integrated energy system, characterized in that: include: Establishing a two-tiered optimization scheduling model for an integrated energy system; the two-tiered optimization scheduling model includes an upper-tier model and a lower-tier model, wherein the upper-tier model is established based on the energy purchase quantity and price of the integrated energy system, with energy purchase cost as a target, and the lower-tier model is established based on the load transfer and load reduction, as well as the energy sales quantity and price of the integrated energy system, with energy sales revenue as a target; the integrated energy system includes traditional energy, as well as wind power and photovoltaic power; The lower-level model and its constraints are converted into constraints of the upper-level model and incorporated into the calculation of the upper-level model to obtain a single-level optimization scheduling model for the integrated energy system; Weight allocation is performed on the uncertainties of wind power output and photovoltaic output power in the integrated energy system, as well as the uncertainties of power demands of various types of loads; Based on information gap decision-making, an opportunity pursuit model and a risk aversion model are constructed. The uncertain factors calculated by the opportunity pursuit model or the risk aversion model are used as input parameters of the single-layer optimization scheduling model after weight allocation. The single-layer optimization scheduling model is solved to obtain the energy purchase quantity of the integrated energy system and the output power parameters of traditional energy in the integrated energy system, and then they are executed.

2. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 1, characterized in that: The upper-level model is based on the energy purchase quantity and price of the integrated energy system, with energy purchase cost as the target, and includes: The upper-level model is a function established with the goal of minimizing the energy purchase cost, which is calculated based on the electricity purchase quantity, electricity purchase price, gas purchase quantity, and gas purchase price of the integrated energy system. The upper-level model is as follows: minF=∑ T t=1 l BE P t BE +∑ T t=1 l BG P t BG , Where F represents the energy purchase cost of the integrated energy system, T represents the total number of time periods, and λ BE and λ BG Represent the unit price of electricity and gas respectively, P t BE and P t BG They represent the electricity purchase power and gas purchase power during period t respectively.

3. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 2, characterized in that: The lower-level model is based on the shifted and reduced loads of the integrated energy system, as well as the energy sales quantity and price, with energy sales revenue as the target, and includes: The lower-level model is a function established with the goal of maximizing energy sales revenue, based on the incentives corresponding to the electric load power and thermal load power transferred by the integrated energy system in each time period, the compensation corresponding to the reduced electric load power and thermal load power, and the unit price of electricity, heat, gas, and cooling, as well as the corresponding load power. The lower-level model is as follows: maxR=∑ T t=1 (d e P t CL +d h H t CL +ξ e P t SL +ξ h H t SL +λ SE P t SE +r E P t load +r H H t load +r G G t load +r C C t load ), Among them, R represents the energy sales revenue, δ e and δ h Represents the unit compensation coefficient for reducing electrical load and reducing thermal load, P t CL and H t CL They represent the reduced electric load and heat load during period t, ξ e and ξ h They represent the unit excitation coefficients of the transferred electrical load and the transferred thermal load, P t SL and H t SL They represent the electrical load and thermal load transferred during period t, respectively, SE represents the unit price of electricity sold to the energy market, P t SE represents the electricity sales power in period t, ρ E , ρ H , ρ G and ρ C They represent the unit price of electricity, heat, gas and cooling sold to the load side respectively, t load 、H t load , G t load and C t load They represent the electric load power, thermal load power, gas load power and cooling load power in period t respectively.

4. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 3, characterized in that: The upper-level model also includes electric power balance constraints for balancing electricity consumption and power generation, thermal power balance constraints for balancing heat consumption and heat generation, gas power balance constraints for balancing gas consumption and gas discharge, cooling power balance constraints for balancing cooling consumption and cooling, and output power ramping constraints for traditional energy units. The lower layer model also includes upper and lower limit constraints on the transferred and reduced electric load power and thermal load power.

5. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 4, characterized in that: The converting the lower layer model and its constraints into the constraints of the upper layer model includes: The Lagrangian function corresponding to the lower-level model is established. Based on the complementary relaxation condition of KKT, the Lagrangian function corresponding to the lower-level model is converted into the nonlinear constraint condition of the upper-level model. The nonlinear constraint condition is then converted into the linear constraint condition of the upper-level model through the big M method.

6. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 5, characterized in that: The linear constraints of the upper model are as follows: 0≤P t CL -P t CL,min γ≤ε1M1,0≤τ1≤(1-ε1)M1; 0≤P t CL,max γ-P t CL ≤ε2M2,0≤τ2≤(1-ε2)M2; 0≤P t SL -P t SL,min φ≤ε3M3,0≤τ3≤(1-ε3)M3; 0≤P t SL,max φ-P t SL ≤ε4M4,0≤τ4≤(1-ε4)M4; 0≤H t CL -H t CL,min β≤ε5M5,0≤τ5≤(1-ε5)M5; 0≤H t CL,max β-H t CL ≤ε6M6,0≤τ6≤(1-ε6)M6; 0≤H t SL -H t SL,min σ≤ε7M7,0≤τ7≤(1-ε7)M7; 0≤H t SL,max σ-H t SL ≤ε8M8,0≤τ8≤(1-ε8)M8; Among them, P t CL,min and P t CL,max They represent the lower limit and upper limit of load power reduction in period t, P t SL,min and P t SL,max They represent the lower and upper limits of the load power transferred during period t, respectively. t CL,min and H t CL,max They represent the lower and upper limits of heat load reduction during period t, H t SL,min and H t SL,max represent the lower and upper limits of the transferred thermal load power in period t, respectively; γ, φ, β, and σ represent the state variables of the corresponding reduced or transferred electrical load or thermal load, ranging from 0 to 1; ε1 to ε8 represent the corresponding Boolean variables; τ1 to τ8 represent the Lagrange multipliers of the corresponding constraints; and M1 to M8 represent the corresponding boundary range definition parameters.

7. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 6, characterized in that: The weight distribution of the uncertainties of wind power output power and photovoltaic output power in the integrated energy system, as well as the uncertainties of power demands of various types of loads, includes: Normalize the attribute values of uncertain factors in the integrated energy system; The information entropy of each uncertain factor is calculated according to the information entropy formula, and the weight of the corresponding uncertain factor is determined by the value of the information entropy. The weight of each uncertain factor is calculated using the following formula: w i =(1-E i ) / (n-∑ n i=1 HAVE BEEN i ), Among them, w i represents the weight of the i-th uncertainty factor, E i represents the information entropy of the i-th uncertain factor, and n represents the total number of uncertain factors; The information entropy of uncertain factors is calculated using the following formula: HAVE BEEN i =-ln(T) -1 ∑ T t=1 pp it ln(p it ), Among them, p it It represents the proportion of the attribute value of the i-th uncertainty factor in period t to the attribute values of all periods.

8. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 7, characterized in that: The opportunity pursuit model is used to utilize the fluctuation of uncertain factors to maximize the benefits of the integrated energy system; the risk aversion model is used to meet the needs of various types of loads when uncertain factors fluctuate.

9. The optimization scheduling method for a multi-uncertainty integrated energy system according to claim 8, characterized in that: The opportunity pursuit model is as follows: minα, minf≤f0(1-δ c ), P t i ∈U(α i ,P t i,pre ); Where α represents the total uncertainty, α i represents the uncertainty of the i-th uncertain factor, f represents the energy purchase cost, f0 represents the preset value of the energy purchase cost, δ c represents the chance deviation coefficient, P t i represents the power parameter of the i-th uncertainty factor during period t, P t i,pre represents the predicted power of the ith uncertainty factor in period t, and the set U(α i ,P t i,pre ) as the input parameter of the single-layer optimization scheduling model; The risk aversion model is as follows: maxα, maxf≤f0(1+δ R ), P t i ∈U(α i ,P t i,pre ); Among them, δ R represents the robust bias coefficient.

10. An optimization scheduling device for a multi-uncertainty integrated energy system, characterized in that: include: Model building unit, model conversion unit, weight distribution unit and power parameter calculation unit, where: The model establishment unit is configured to establish a two-layer optimization scheduling model for an integrated energy system; the two-layer optimization scheduling model comprises an upper-layer model and a lower-layer model, wherein the upper-layer model is established based on the energy purchase quantity and price of the integrated energy system, with energy purchase cost as a target, and the lower-layer model is established based on the load transfer and load reduction, as well as the energy sales quantity and price of the integrated energy system, with energy sales revenue as a target; the integrated energy system comprises traditional energy, as well as wind power and photovoltaic power; The model conversion unit is used to convert the lower-level model and its constraints into the constraints of the upper-level model, and incorporate them into the calculation of the upper-level model to obtain a single-level optimization scheduling model for the integrated energy system; The weight allocation unit is used to weight the uncertainty of wind power output power and photovoltaic output power in the integrated energy system, as well as the uncertainty of each type of load; The power parameter calculation unit is used to construct an opportunity pursuit model and a risk aversion model based on information gap decision-making, and use the uncertainty factors calculated by the opportunity pursuit model or the risk aversion model as input parameters of the single-layer optimization scheduling model after weight distribution, solve the single-layer optimization scheduling model, obtain the energy purchase quantity and energy purchase price of the integrated energy system, and the output power parameters of traditional energy in the integrated energy system, and execute them.

Images