Island microgrid source-load-storage co-integration optimization method containing frequency deviation constraint

Abstract

The invention provides an island micro-grid source-load-storage co-integration optimization method containing frequency deviation constraint, and aims at minimizing the operation cost of an island micro-grid and maximizing the consumption of new energy. A source-load-storage coordinated operation model, a source-load-storage coordinated operation model containing frequency deviation, and a source-load-storage co-integration optimization operation model containing frequency deviation stability constraint are established. According to the coordination or co-integration optimization model containing the frequency deviation, the frequency deviation allowable range serves as an available resource in optimization scheduling, compared with a coordination operation model, the feasible solution range is enlarged, higher economical efficiency is achieved, the system operation cost is reduced, and the new energy consumption rate is increased; due to the constraint of the source-load-storage co-integration relationship, the co-integration model better ensures the frequency safety during the operation of the island microgrid than the coordination model, and the optimization extreme value is improved on the premise of ensuring the normal operation.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrids, and in particular to an island microgrid source-load-storage cointegration optimization method with frequency offset constraints. Background Art

[0002] Microgrid, as a small power system composed of distributed power sources, energy storage devices, loads, etc., is currently the most effective organizational form of distributed power sources, which can maximize the utilization of distributed power sources. Coordinated control of sources, loads and storage can improve the flexibility and reliability of microgrid operation and reduce wind and solar power abandonment. The essence of source, load and storage coordination is to adjust the power of each source to maintain a real-time power balance between all power generation and all power consumption (including energy storage).

[0003] Whether it is a microgrid island operation or a large system network operation, the real-time power balance is the energy conservation of a series of power curves on the power supply side and the power consumption side at any time. Specifically, it is manifested as the traditional generator output curve, the new energy output curve, the load power curve, the energy storage output curve and other curves, and the linear combination of the time series is zero. The equation constraint. These curves are generally non-stationary time series in the mathematical sense, but if their linear combination is zero, these curves can be regarded as having a cointegration relationship. According to the cointegration theory, if the linear combination of non-stationary variables is stationary, there is a cointegration relationship between the variables. Therefore, the real-time power balance on the time series was found to be a special cointegration relationship in this study.

[0004] Cointegration theory has been applied to power systems to a certain extent. The first is related research on load forecasting in power systems, such as using cointegration tests to find economic factors that have a long-term equilibrium relationship with electricity, using Granger causality test methods to determine economic factors that are helpful for electricity forecasting, and reducing the number of input variables in the forecasting model; the second is used for the cointegration operation of sources, loads and storage in power systems and related indicators, which focuses on the study of the characterization and adjustment of the source-load relationship using cointegration theory. This type of research is still in its infancy. For example, cointegration theory is used to characterize related indicators of cointegration operation, such as cointegration index, new energy-load similarity index and curve volatility index; using cointegration index as a constraint or target, a multi-objective optimization model is established to reduce the frequent adjustment of thermal power units and improve the utilization rate of new energy. Cointegration indicators and methods such as multi-power curve cointegration are applied to offshore oilfield platform systems in oilfield production areas and the supply and demand matching of regional integrated energy systems, but no relevant literature has been cited, and their application is slightly flawed.

[0005] The application and discovery of cointegration theory and cointegration relationship in real-time power balance are the first. For isolated microgrids, real-time power balance is difficult due to the random fluctuations of new energy and load. That is, there may be a power imbalance between the total power generation power and the total load power at each moment. At present, there is no deterministic description of power imbalance. Uncertain variables are often used to describe the uncertainty of source and load, and robust optimization, interval optimization or random optimization are used to deal with optimization problems containing uncertain variables. However, its essence is to express the mathematical characteristics of variables, and there is no physical representation of the objective state of the new power balance containing power imbalance. In the operation of the power system, power balance reflects the physical characteristic of energy conservation, and the so-called power imbalance seems to be contrary to the basic principle of energy conservation. In fact, power imbalance is a primary frequency modulation function relationship between frequency offset and system frequency modulation coefficient. Therefore, power balance containing power imbalance is a real-time energy conservation at the cost of frequency deviation.

[0006] The existing optimization scheduling and planning models do not describe the situation where power imbalance causes frequency offset. The real-time power imbalance and frequency offset conditions should not be ignored. Summary of the invention

[0007] In order to solve the above technical problems, the present invention proposes an isolated microgrid source-load-storage cointegration optimization method with frequency offset constraints, adds the active power cointegration balance equation equation constraints containing frequency offset into the source-load-storage coordinated optimization model, and uses the allowable range of frequency offset as an available resource in the optimization scheduling, thereby relaxing the constraints, increasing the range of feasible solutions, improving the optimization extreme value, reducing the system operating cost, and increasing the new energy consumption rate.

[0008] To achieve the above object, the technical solution adopted by the present invention is:

[0009] The source-load-storage cointegration optimization method of an isolated microgrid with frequency offset constraints is characterized by comprising the following steps:

[0010] S1: Model building

[0011] Taking the minimization of the operating cost of the isolated microgrid and the maximization of the new energy consumption as multiple objectives, Model 1, Model 2 and Model 3 are constructed:

[0012] S11, Model 1: Source-load-storage coordinated operation model including thermal units, new energy units, demand-side response and batteries:

[0013] S111. Set the objective function:

[0014] S1111, Minimum operating cost:

[0015] The operating cost includes the thermal unit operating cost C G(P dg,t ) and C MT (P mt,t ), demand side response cost C D (ΔP d,t ) Cost of curtailing wind and solar power C W (ΔP w,t ), battery energy storage operating cost C stor (P stor,t ),

[0016] F1=min(C G (P dg,t )+C MT (P mt,t )+

[0017] C D (ΔP d,t )+C W (ΔP w,t )+C stor (P stor,t ))

[0018]

[0019]

[0020] Where: U t and U t-1 Respectively represent the start and stop status of the thermal engine unit at time t and t-1; a, b, c represent the fuel cost parameters of the thermal engine unit; α, β, τ are the start and stop cost parameters of the thermal engine unit; δ is the downtime of the thermal engine unit; P dg,t is the output of the thermal engine unit in period t; C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η mt is the power generation efficiency of the gas turbine; ρ is the unit compensation cost of the demand response load; ΔP d,t is the load participating in demand response in period t; ω is the penalty coefficient for wind and solar power abandonment; ΔP w,t is the wind and solar power abandoned in period t; λ is the energy storage cost coefficient; P stor,t is the charging and discharging power of the energy storage during period t. Positive indicates charging, and negative indicates discharging.

[0021] S1112, the largest consumption of new energy

[0022]

[0023] Where: P w,t is the actual output of new energy obtained by optimizing during period t,

[0024] S112. Set constraints:

[0025] S1121, Traditional Active Power Balance

[0026]

[0027] S1122, Energy Storage Output Constraints

[0028] Energy storage output takes into account charging and discharging constraints and energy storage capacity constraints

[0029]

[0030] Where: The upper limit of energy storage charging and discharging power; X ch , Y dis It is a 0-1 variable, 0 represents energy storage not working, and 1 represents energy storage charging and discharging respectively; E es,t is the energy storage capacity at time t; E Bat is the rated capacity of energy storage; SOC min , SOC max is the minimum and maximum value of the energy storage charge state,

[0031] S1123, demand-side response load constraints

[0032]

[0033] Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period,

[0034] S12, Model 2: Based on Model 1, a source-load-storage coordinated operation model with frequency offset is constructed:

[0035] The active power balance constraint for frequency offset is given by

[0036]

[0037] Where: t=1,2,…,T; ΔP t is the power imbalance; f0 is the rated frequency, which is 50Hz; k is the frequency modulation coefficient,

[0038] The objective function and other constraints of Model 2 are consistent with those of Model 1.

[0039] S13, Model 3: Based on Model 2, a source-load-storage cointegration operation model with frequency shift stability constraints is constructed:

[0040] The constraints of the active power cointegration balance equation with frequency offset are as follows:

[0041]

[0042] Where: ADF.test(kΔf t ) is the frequency shift function kΔf t The ADF stationarity test of , if the test constraint is met, the cointegration relationship between the source, load and storage power time series is established.

[0043] The objective function and other constraints of Model 3 are consistent with those of Model 1;

[0044] S2: Case parameters

[0045] The energy storage rated capacity, energy storage equipment parameters, and demand response parameters are set respectively, and a dispatch cycle and a dispatch period are set. Through the above settings, the unit compensation cost of the demand response load, the wind and solar power abandonment penalty coefficient, and the energy storage cost coefficient are calculated;

[0046] S3: Simulation results and analysis

[0047] S31, Pareto frontier and ideal solutions of each model

[0048] The Pareto frontiers of Model 1, Model 2, and Model 3 are obtained by augmented ε constraint method. The number of segments g = 5 is selected, and the number of solution sets is 6, which are solution sets 1 to 6 from left to right.

[0049] Solution set 1 and solution set 6 are the minimum operating cost and the maximum new energy consumption when optimized separately, respectively.

[0050] As the amount of new energy consumed in the solution set continues to increase, its operating cost continues to increase accordingly, which means that there is no optimal solution set that can ensure the maximum amount of new energy consumed while minimizing the operating cost.

[0051] Therefore, the Pareto frontier represents the set of possible solutions with a sacrificial relationship between the two goals of operating cost and new energy consumption. In practice, in order to obtain the ideal solution in the Pareto frontier, the decision maker is often selected according to his subjective will. For example, if the decision maker focuses on economy, he will select solution set 1 with the smallest operating cost. If he hopes to reduce wind and solar power abandonment, he will select solution set 6 with the largest new energy consumption.

[0052] The game theory-based combined weighting and TOPSIS comprehensive evaluation method are used to obtain the ideal solution in the Pareto frontier and obtain the optimal trade-off between system operation cost and new energy consumption.

[0053] Thus, the comprehensive scores of each solution set in Model 1, Model 2 and Model 3 are obtained.

[0054] From left to right, the models are the new energy consumption, operation cost and comprehensive score corresponding to solution sets 1 to 6, which correspond to the solution sets in the Pareto frontier.

[0055] The higher the TOPSIS comprehensive evaluation score, the better the solution set. Model 1, Model 2, and Model 3 all have the highest comprehensive scores in Solution Set 2. Therefore, the ideal solutions of the three models are all Solution Set 2.

[0056] Model 1, Model 2 and Model 3 obtain the ideal solution of optimal system operation cost and new energy consumption from the Pareto frontier.

[0057] The operation cost and new energy consumption of isolated microgrids are analyzed: Model 1 has the largest total operation cost and the least new energy consumption in coordinated operation; Model 2 has the smallest total operation cost and the largest new energy consumption in coordinated operation with frequency offset; Model 3 is in between the two in cointegration operation.

[0058] When the equality constraint is expanded to a constraint containing frequency offset, the constraint range is relaxed, thus obtaining a better target value.

[0059] S32. Analysis of power imbalance and frequency offset of each model

[0060] The coordinated operation of model 2 and the cointegrated operation of model 3 both use the frequency offset allowable range as the frequency modulation resource, so there are power imbalance and frequency offset. The coordinated operation of model 1 does not use the frequency offset allowable range as the frequency modulation resource, and its power imbalance and frequency offset are both 0.

[0061] Model 2 coordinated operation only considers the allowable range of microgrid frequency deviation as ±0.5Hz. Since it fully utilizes the allowable range of frequency deviation as frequency modulation resources, it leads to multiple critical values ​​of frequency deviation. Compared with the other two operation models, although it is economical, it cannot guarantee the safety of system frequency.

[0062] In the cointegration operation of model 3, the cointegration theory is used to characterize the balance relationship between the non-stationary time series of the source-load-storage multi-power curves. Considering the correlation between the multi-power curves, on the basis of the upper and lower limit constraints of the frequency offset, the cointegration relationship between the total power supply and the total load is established by constraining the stability of the power imbalance.

[0063] The total power supply and total load variables are kept in long-term balance, so the power imbalance and frequency offset will not show an upward or downward trend. If the variable value at a certain moment deviates from its stable trend, then a corrective change will occur at the next moment to readjust the variable back to its stable state. Therefore, the stability constraint ensures that the variables maintain a certain stable equilibrium relationship in the long term, even if there is a deviation in the short term.

[0064] In the cointegration operation of model 3, the frequency offset always fluctuates around 0 Hz and does not touch the boundary of frequency offset, ensuring the frequency safety during the operation of the isolated microgrid. After adding the stability constraint, the cointegration relationship is established. Its essence is to prevent the frequency offset from being at a critical level. Therefore, the constraints will be tightened, the frequency safety level will be improved, and the target value will be reduced, thus improving the optimization extreme value while ensuring normal operation.

[0065] As a preferred technical solution of the present invention: in step S1, the isolated island microgrid includes energy units, batteries, demand-side responses and loads, the energy units include thermal engine units and new energy units, the batteries are used for energy storage, the demand-side response is a microgrid operation control center, the load is used to control the flow of electric energy, and the microgrid control center is used to control and receive information flow.

[0066] As a preferred technical solution of the present invention: the thermal engine unit includes a diesel engine unit and a gas turbine, wherein:

[0067] The cost function of the diesel generator in the diesel unit is:

[0068]

[0069] Where: U t and U t-1 Respectively represent the start and stop status of the diesel generator set at time t and t-1; a, b, c represent the fuel cost parameters of the diesel generator set; α, β, τ are the start and stop cost parameters of the diesel generator set; δ is the downtime of the diesel generator set; P dg,t is the output of the unit in period t;

[0070] Gas Turbine Cost Function [2] for:

[0071]

[0072] Where: C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η m t is the power generation efficiency of the gas turbine.

[0073] As a preferred technical solution of the present invention: the new energy generator set includes a wind turbine generator set and a photovoltaic generator set, wherein:

[0074] The specific relationship between the output power and wind speed of the wind turbine in the wind turbine is:

[0075]

[0076] Where: v, v ci 、vco 、v r are the actual wind speed, rated wind speed, cut-in wind speed, and cut-out wind speed at the hub height of the wind turbine respectively; P r is the rated output of the fan,

[0077] The distributed photovoltaic output power in the photovoltaic unit is:

[0078]

[0079] Where: P STC G is the rated power of photovoltaic output under standard conditions; AC is the irradiation intensity at the working point; k is the temperature power coefficient; T c is the battery temperature at the operating point; G STC is the solar irradiance; T t is the temperature of the battery under rated conditions.

[0080] As the preferred technical solution of the present invention: from the perspective of load operation scheduling, user loads are divided into rigid loads and flexible loads. Rigid loads cannot be regulated by the power system, and flexible loads are power loads that can adjust their power consumption behavior within the control range according to the power system demand and market signals. The flexible loads that can be called in each time period and scheduling cycle are limited.

[0081]

[0082] Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin They are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period.

[0083] As a preferred technical solution of the present invention: the function of the battery state of charge is:

[0084]

[0085] Where: E BESS,t E is the energy storage charge and discharge amount in period t, a positive value indicates charging amount, and a negative value indicates discharging amount; Bat is the rated capacity of the energy storage.

[0086] As a preferred technical solution of the present invention: in step S2, the scheduling cycle is 24 hours and the scheduling period is 0.5 hours.

[0087] Compared with the prior art, the present invention has the following beneficial effects:

[0088] The present invention adds the real-time power balance including frequency offset as an equality constraint in the source-load-storage coordinated optimization model, and takes the frequency offset allowable range as an available resource in the optimization scheduling, thereby relaxing the constraint conditions, increasing the feasible solution range, and improving the optimization extreme value, thereby reducing the system operation cost and improving the new energy consumption rate.

[0089] In order to solve the frequency safety problem caused by the relaxation of constraints, the stability constraint of the frequency offset function is added. Its essence is to prevent the frequency offset from being at a critical level. Therefore, the constraints are slightly tightened, the frequency safety level is appropriately improved, and the target value is slightly reduced, thereby improving the optimization extreme value while ensuring normal operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 is a flow chart of the method of the present invention;

[0091] Figure 2 It is a schematic diagram of the isolated island microgrid structure in the present invention;

[0092] Figure 3 It is the power curve of each unit of the non-stationary time series characteristics;

[0093] Figure 4 It is a table of the value range of the objective function;

[0094] Figure 5 is the parameter diagram of the energy storage device;

[0095] Figure 6 is the parameter diagram of demand response

[0096] Figure 7 is the Pareto front diagram;

[0097] Figure 8 It is the comprehensive score graph of the Pareto solution set of models 1-3;

[0098] Fig. 9 It is a table showing the operating cost and new energy consumption of each model;

[0099] Fig.10 This is the power curve diagram of coordinated operation of model 1;

[0100] Fig.11 This is the power curve diagram of coordinated operation of model 2;

[0101] Fig.12 This is the cointegration operation power curve of Model 3;

[0102] Fig.13 is the frequency shift line plot for model 2;

[0103] Fig.14 is the line graph of power imbalance of model 2;

[0104] Fig.15 is the frequency shift line plot for Model 3;

[0105] Fig.16 It is a line graph of the power imbalance of Model 3. DETAILED DESCRIPTION

[0106] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:

[0107] like Figure 1-16 As shown, the present invention proposes an island microgrid source-load-storage cointegration optimization method with frequency offset constraints, comprising the following steps:

[0108] S1: Model building

[0109] Taking the minimization of the operating cost of the isolated microgrid and the maximization of the new energy consumption as multiple objectives, Model 1, Model 2 and Model 3 are constructed:

[0110] S11, Model 1: Source-load-storage coordinated operation model including thermal units, new energy units, demand-side response and batteries:

[0111] S111. Set the objective function:

[0112] S1111, Minimum operating cost:

[0113] The operating cost includes the thermal unit operating cost C G (P dg,t ) and C MT (P mt,t ), demand side response cost C D (ΔP d,t ) Cost of curtailing wind and solar power C W (ΔP w,t ), battery energy storage operating cost C stor (P stor,t ),

[0114] F1=min(C G (P dg,t )+C MT (P mt,t )+

[0115] C D (ΔP d,t )+C W (ΔP w,t )+C stor (P stor,t ))

[0116]

[0117]

[0118] Where: U t and U t-1 Respectively represent the start and stop status of the thermal engine unit at time t and t-1; a, b, c represent the fuel cost parameters of the thermal engine unit; α, β, τ are the start and stop cost parameters of the thermal engine unit; δ is the downtime of the thermal engine unit; P dg,t is the output of the thermal engine unit in period t; C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η mt is the power generation efficiency of the gas turbine; ρ is the unit compensation cost of the demand response load; ΔP d,t is the load participating in demand response in period t; ω is the penalty coefficient for wind and solar power abandonment; ΔP w,t is the wind and solar power abandoned in period t; λ is the energy storage cost coefficient; P stor,t is the charging and discharging power of the energy storage during period t. Positive indicates charging, and negative indicates discharging.

[0119] S1112, the largest consumption of new energy

[0120]

[0121] Where: P w,t is the actual output of new energy obtained by optimizing during period t,

[0122] S112. Set constraints:

[0123] S1121, Traditional Active Power Balance

[0124]

[0125] S1122, Energy Storage Output Constraints

[0126] Energy storage output takes into account charging and discharging constraints and energy storage capacity constraints

[0127]

[0128] Where: The upper limit of energy storage charging and discharging power; X ch , Y dis It is a 0-1 variable, 0 represents energy storage not working, and 1 represents energy storage charging and discharging respectively; E es,t is the energy storage capacity at time t; E Bat is the rated capacity of energy storage; SOC min , SOC max is the minimum and maximum value of the energy storage charge state,

[0129] S1123, demand-side response load constraints

[0130]

[0131] Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period,

[0132] S12, Model 2: Based on Model 1, a source-load-storage coordinated operation model with frequency offset is constructed:

[0133] The active power balance constraint for frequency offset is given by

[0134]

[0135] Where: t=1,2,...,T; ΔP t is the power imbalance; f0 is the rated frequency, which is 50Hz; k is the frequency modulation coefficient,

[0136] The objective function and other constraints of Model 2 are consistent with those of Model 1.

[0137] S13, Model 3: Based on Model 2, a source-load-storage cointegration operation model with frequency shift stability constraints is constructed:

[0138] The constraints of the active power cointegration balance equation with frequency offset are as follows:

[0139]

[0140] Where: ADF.test(kΔf t ) is the frequency shift function kΔf t The ADF stationarity test of , if the test constraint is met, the cointegration relationship between the source, load and storage power time series is established.

[0141] The objective function and other constraints of Model 3 are consistent with those of Model 1;

[0142] S2: Case parameters

[0143] The energy storage rated capacity, energy storage equipment parameters, and demand response parameters are set respectively, and a dispatch cycle and a dispatch period are set. Through the above settings, the unit compensation cost of the demand response load, the wind and solar power abandonment penalty coefficient, and the energy storage cost coefficient are calculated;

[0144] S3: Simulation results and analysis

[0145] S31, Pareto frontier and ideal solutions of each model

[0146] The Pareto frontiers of Model 1, Model 2, and Model 3 are obtained by augmented ε constraint method. The number of segments g = 5 is selected, and the number of solution sets is 6, which are solution sets 1 to 6 from left to right.

[0147] Solution set 1 and solution set 6 are the minimum operating cost and the maximum new energy consumption when optimized separately, respectively.

[0148] As the amount of new energy consumed in the solution set continues to increase, its operating cost continues to increase accordingly, which means that there is no optimal solution set that can ensure the maximum amount of new energy consumed while minimizing the operating cost.

[0149] Therefore, the Pareto frontier represents the set of possible solutions with a sacrificial relationship between the two goals of operating cost and new energy consumption. In practice, in order to obtain the ideal solution in the Pareto frontier, the decision maker is often selected according to his subjective will. For example, if the decision maker focuses on economy, he will select solution set 1 with the smallest operating cost. If he hopes to reduce wind and solar power abandonment, he will select solution set 6 with the largest new energy consumption.

[0150] The game theory-based combined weighting and TOPSIS comprehensive evaluation method are used to obtain the ideal solution in the Pareto frontier and obtain the optimal trade-off between system operation cost and new energy consumption.

[0151] Thus, the comprehensive scores of each solution set in Model 1, Model 2 and Model 3 are obtained.

[0152] From left to right, the models are the new energy consumption, operation cost and comprehensive score corresponding to solution sets 1 to 6, which correspond to the solution sets in the Pareto frontier.

[0153] The higher the TOPSIS comprehensive evaluation score, the better the solution set. Model 1, Model 2, and Model 3 all have the highest comprehensive scores in Solution Set 2. Therefore, the ideal solutions of the three models are all Solution Set 2.

[0154] Model 1, Model 2 and Model 3 obtain the ideal solution of optimal system operation cost and new energy consumption from the Pareto frontier.

[0155] The operation cost and new energy consumption of isolated microgrids are analyzed: Model 1 has the largest total operation cost and the least new energy consumption in coordinated operation; Model 2 has the smallest total operation cost and the largest new energy consumption in coordinated operation with frequency offset; Model 3 is in between the two in cointegration operation.

[0156] When the equality constraint is expanded to a constraint containing frequency offset, the constraint range is relaxed, thus obtaining a better target value.

[0157] S32. Analysis of power imbalance and frequency offset of each model

[0158] The coordinated operation of model 2 and the cointegrated operation of model 3 both use the frequency offset allowable range as the frequency modulation resource, so there are power imbalance and frequency offset. The coordinated operation of model 1 does not use the frequency offset allowable range as the frequency modulation resource, and its power imbalance and frequency offset are both 0.

[0159] Model 2 coordinated operation only considers the allowable range of microgrid frequency deviation as ±0.5Hz. Since it fully utilizes the allowable range of frequency deviation as frequency modulation resources, it leads to multiple critical values ​​of frequency deviation. Compared with the other two operation models, although it is economical, it cannot guarantee the safety of system frequency.

[0160] In the cointegration operation of model 3, the cointegration theory is used to characterize the balance relationship between the non-stationary time series of the source-load-storage multi-power curves. Considering the correlation between the multi-power curves, on the basis of the upper and lower limit constraints of the frequency offset, the cointegration relationship between the total power supply and the total load is established by constraining the stability of the power imbalance.

[0161] The total power supply and total load variables are kept in long-term balance, so the power imbalance and frequency offset will not show an upward or downward trend. If the variable value at a certain moment deviates from its stable trend, then a corrective change will occur at the next moment to readjust the variable back to its stable state. Therefore, the stability constraint ensures that the variables maintain a certain stable equilibrium relationship in the long term, even if there is a deviation in the short term.

[0162] In the cointegration operation of model 3, the frequency offset always fluctuates around 0 Hz and does not touch the boundary of frequency offset, ensuring the frequency safety during the operation of the isolated microgrid. After adding the stability constraint, the cointegration relationship is established. Its essence is to prevent the frequency offset from being at a critical level. Therefore, the constraints will be tightened, the frequency safety level will be improved, and the target value will be reduced, thus improving the optimization extreme value while ensuring normal operation.

[0163] The isolated island microgrid described in the present invention includes energy units, batteries, demand-side responses and loads. The energy units include thermal engine units and new energy units. The batteries are used for energy storage. The demand-side response is a microgrid operation control center. The load is used to control the flow of electric energy. The microgrid control center is used to control and receive information flow.

[0164] The thermal engine unit includes a diesel engine unit and a gas turbine, wherein:

[0165] The cost function of the diesel generator in the diesel unit is:

[0166]

[0167] Where: U t and U t-1 Respectively represent the start and stop status of the diesel generator set at time t and t-1; a, b, c represent the fuel cost parameters of the diesel generator set; α, β, τ are the start and stop cost parameters of the diesel generator set; δ is the downtime of the diesel generator set; P dg,t is the output of the unit in period t;

[0168] Gas Turbine Cost Function [2] for:

[0169]

[0170] Where: C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η m t is the power generation efficiency of the gas turbine.

[0171] As a preferred technical solution of the present invention: the new energy generator set includes a wind turbine generator set and a photovoltaic generator set, wherein:

[0172] The specific relationship between the output power and wind speed of the wind turbine in the wind turbine is:

[0173]

[0174] Where: v, v ci 、v co 、v r are the actual wind speed, rated wind speed, cut-in wind speed, and cut-out wind speed at the hub height of the wind turbine respectively; P r is the rated output of the fan,

[0175] The distributed photovoltaic output power in the photovoltaic unit is:

[0176]

[0177] Where: P STC G is the rated power of photovoltaic output under standard conditions; AC is the irradiation intensity at the working point; k is the temperature power coefficient; T c is the battery temperature at the operating point; G STC is the solar irradiance; T t is the temperature of the battery under rated conditions.

[0178] As the preferred technical solution of the present invention: from the perspective of load operation scheduling, user loads are divided into rigid loads and flexible loads. Rigid loads cannot be regulated by the power system, and flexible loads are power loads that can adjust their power consumption behavior within the control range according to the power system demand and market signals. The flexible loads that can be called in each time period and scheduling cycle are limited.

[0179]

[0180] Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin They are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period.

[0181] As a preferred technical solution of the present invention: the function of the battery state of charge is:

[0182]

[0183] Where: E BESS,t E is the energy storage charge and discharge amount in period t, a positive value indicates charging amount, and a negative value indicates discharging amount; Bat is the rated capacity of the energy storage.

[0184] Source-load-storage cointegration relationship of isolated island microgrid in the present invention

[0185] 1. Cointegration relationship:

[0186] Cointegration theory is an analytical theory for non-stationary time series, mainly used to study the correlation between various indicators or parameters. If two or more non-stationary time series have a linear combination that is stationary, they have a cointegration relationship.

[0187] According to the definition of cointegration relationship, when a non-stationary single-order integrated time series X 1t With X 2t If μ t If X is a stationary series, then 1t With X 2t Cointegration.

[0188] X 1t =α+βX 2t +μ t (1)

[0189] Where: α, β are constant terms; μ t is the residual sequence.

[0190] 2. Special cointegration relationship of active power balance

[0191] All units in the isolated microgrid have their own power curves, and the output shows typical non-stationary time series characteristics, such as the output of thermal engine units (diesel units, gas turbines), the output of new energy (wind turbines, photovoltaic units), the power of user loads and the power of battery energy storage, such as Figure 3 shown.

[0192] In the isolated microgrid, the total power supply power and the total load power are equal at each moment in the time series, that is, the real-time balance of active power is an inherent characteristic that must be guaranteed in the operation of the power system, as shown in formula (2). The linear combination of the power curves of each unit is shown in formula (3). The left side of the equation is the total power supply power, which is composed of the output of the thermal engine unit and the output of new energy; the right side of the equation is the total load power, which is composed of the user load power and the energy storage power.

[0193]

[0194] P dg,t +P mt,t +P w,t =P loa d ,t +P stor,t (3)

[0195] Where: t=1,2,...,T; P w,t Contribute to new energy; dg,t Output for diesel generator set; P mt,t Output for gas turbine; P load,t is the user load power; P stor,t It is the energy storage power, which is positive when charging and negative when discharging.

[0196] Formula (2) is both a real-time power balance and a special cointegration relationship of time series. Compared with the cointegration relationship formula (1), formula (2) is a constant term α = 0, β = 1, and the residual sequence μ t =0 special relationship.

[0197] According to the cointegration theory, combined with Figure 3 As with formula (3), the power curves of each unit are non-stationary time series and there are linear combinations. Therefore, if the residual sequence μ t If is a stationary sequence, the active power balance is a special cointegration relationship with a residual sequence of zero, that is, there is a cointegration relationship between the total power curve and the total load power curve. The stationarity of the residual sequence μt is proved as follows:

[0198] If the time series {X t}(t=1,2,…T) satisfies the following three conditions at the same time, called X t is a stationary time series.

[0199] Condition 1: E(X t)=φ; that is, X t The mean value of does not change over time.

[0200] Condition 2: Var(X t )=σ 2 ; That is, X t The variance does not change over time.

[0201] Condition 3: Cov(X t ,X t+s )=γ s ; That is, X t With X t+s The covariance between them depends only on the interval length s.

[0202] For active power balance equation (2), its residual sequence μ t =0, its mean, variance and covariance are as follows.

[0203]

[0204]

[0205] So the residual sequence μ t is a stationary sequence, so the active power balance is the residual sequence μ t =0 special cointegration relationship.

[0206] 3. Potential cointegration relationship of active power balance with frequency offset

[0207] For isolated microgrids, there is no support from the large power grid, and due to their own capacity limitations and the uncertainty of renewable energy output, there are certain problems in real-time power balance.

[10] That is, there may be a power imbalance between the total power generation and the total load power at each moment. Considering the real-time existence of power imbalance, the active power balance (Equation (2)) is rewritten as the power imbalance (ΔP t ) is constrained by the active power balance equation, as shown in equation (4).

[0208]

[0209] Where: t=1,2,...,T; ΔP t is the power imbalance.

[0210] The existing power system dispatching does not have a deterministic description of power imbalance, and often uses uncertain variables to describe the uncertainty of source and load. Robust optimization, interval optimization and stochastic optimization are used to deal with optimization problems with uncertain variables.

[12] In the actual operation of the power system, the power imbalance at each moment can be expressed as a primary frequency modulation function including frequency offset. Therefore, the power imbalance in the isolated island microgrid is written as a function of the system frequency offset. The active power balance including frequency offset is shown in formula (5), and the allowable range of system frequency offset is shown in formula (6).

[0211]

[0212] 49.5≤f0+Δf t ≤50.5 (6)

[0213] In the formula: t=1,2,...,T; f0 is the rated frequency, which is 50Hz; k is the frequency modulation coefficient.

[0214] Compared with the cointegration relationship (2), formula (5) is the constant term α = 0, β = 1, and the residual sequence μ t ≠0, where the residual sequence μ t The physical connotation of can be regarded as a frequency offset function, as shown in formula (7).

[0215] μ t =ΔP t = kΔf t (7)

[0216] Where: t = 1, 2, ..., T.

[0217] From the definition of cointegration relationship, we know that if the frequency shift function kΔf t If it is a stationary time series, the cointegration relationship holds, so the active power balance with frequency offset can be regarded as a potential cointegration relationship.

[0218] 4. Cointegration relationship of active power balance with frequency offset

[0219] For the active power balance equation (5) containing frequency offset, since the allowable range of system frequency offset has key upper and lower limit constraints, as shown in equation (6), the frequency offset function is expected to be stable in time series.

[0220] In order to test whether the sequence between variables is cointegrated, Engle and Granger proposed a two-step test method in 1987, namely the EG test. The core idea is to use the ADF test method to test the stationarity of the residual sequence. Therefore, it is considered to add ADF constraints to make the frequency offset function satisfy the stationarity. Then, the cointegration relationship of active power containing frequency offset is established, as shown in formula (8).

[0221]

[0222] Among them, ADF.test(kΔf t) is the frequency shift function kΔf t The ADF stationarity test of is carried out. If the test constraint is met, the cointegration relationship between the source, load and storage power time series is established.

[0223] In steps S2 and S3, the present invention adopts the augmented ε constraint method to solve the multi-objective model. The basic idea is to select the main objective function to construct a single objective problem, and establish other objective functions as secondary objective functions. At the same time, the ε parameter is introduced to put the secondary objective function into the constraint condition, and the Pareto frontier of the original multi-objective optimization problem is obtained by gradually increasing the ε value.

[0224] In order to obtain a well-distributed Pareto frontier using the augmented ε constraint method, it is necessary to calculate the value range of each objective function, such as Figure 4 shown.

[0225] Z 11 , Z 22 The operating cost and the new energy consumption have the smallest negative value, Z 12 To obtain the minimum negative value of new energy consumption under the constraint of minimum operating cost, Z 21 The minimum operating cost is obtained under the constraint of the minimum negative value of new energy consumption.

[0226] Taking the minimum operating cost as the main objective function, the minimum negative value of new energy consumption is put into the constraint condition, and combined with the ε parameter and slack variable, the following is obtained:

[0227]

[0228] st-F2(x)+s=ε (10)

[0229] ε=lb+(k*r) / g (11)

[0230] Where: is the coefficient, the value range is 10 -6 ~10 -3 ; s is a non-negative slack variable; r is the difference between the upper and lower limits of the secondary objective function; lb is the minimum value of the secondary objective function; g is the number of segments; k = 1, 2…, g.

[0231] The present invention also obtains an ideal solution from the Pareto frontier by adopting a combined weighting method based on game theory and a TOPSIS comprehensive evaluation method, wherein the combined weighting includes an objective weighting method CRITIC and a subjective weighting analytic hierarchy process (AHP).

[0232] According to the idea of ​​game theory, the goal is to minimize the difference between the index combination weight and the original weight, and seek the optimal linear combination coefficient.

[20] Assume that the weights of CRITIC and AHP are ω1 and ω2 respectively, the combined weight is ω′, and the optimal linear combination coefficients are α′ and β′ respectively.

[0233] Its objective function is as follows:

[0234] min(||ω′-ω1||2+ω′-ω2||2) (12)

[0235] According to the principle of differentiation, when the above formula takes the minimum value, it can be transformed into the solution of the following matrix equation.

[0236]

[0237] By normalizing α′ and β′, we can get the corresponding optimal linear combination coefficients.

[0238] ω′=α′ω1+β′ω2 (14)

[0239] TOPSIS is called "top-up ranking method", which is essentially a comprehensive evaluation method that uses distance as an evaluation criterion. The basic steps of the TOPSIS method can be found in the literature.

[0240] In step S3, when performing simulation analysis and result comparison, taking an isolated island microgrid as an example, the microgrid structure is as follows: Figure 2 The rated capacity of the energy storage is 1.6 (MWh), and the parameters of the energy storage equipment are shown in Figure 5 , demand response parameters see Figure 6 .

[0241] The power curve of wind power and photovoltaic power is as follows: Fig.10 , 11 , 12 new energy power curves. Assume that a dispatch cycle is 24 hours and a dispatch period is 0.5 hours. The unit compensation cost of demand response load and the penalty coefficient for wind and solar power abandonment are 180 yuan / (MWh), and the energy storage cost coefficient is 45 yuan / (MWh).

[0242] Simulation analysis and result comparison:

[0243] Taking the minimization of the operating cost of the isolated microgrid and the maximization of the consumption of new energy as multiple objectives, the traditional source-load-storage coordination, source-load-storage coordination with frequency offset, and source-load-storage cointegration optimization operation models with frequency offset stability constraints are constructed:

[0244] Model 1: Source-load-storage coordinated operation including thermal units, new energy units, demand-side response and battery energy storage.

[0245] Model 2: Based on Model 1, source-load-storage coordinated operation with frequency offset.

[0246] Model 3: Based on Model 2, source-load-storage cointegration operation with frequency shift stationarity constraint is performed.

[0247] Pareto front:

[0248] The Pareto frontiers of Model 1 to Model 3 are obtained by augmented ε constraint method, as shown in Figure 7 shown.

[0249] In the figure, if the number of segments g=5 is selected, the number of solution sets is 6, from left to right, they are solution sets 1 to 6. Solution set 1 and solution set 6 are the minimum operating cost and the maximum new energy consumption when optimized separately. As the amount of new energy consumption in the solution set continues to increase, its operating cost continues to increase accordingly, which means that there is no optimal solution set that can ensure the maximum amount of new energy consumption while minimizing the operating cost. Therefore, Figure 7 The middle Pareto frontier represents the possible solution set with a sacrificial relationship between the two goals of operating cost and new energy consumption. In practice, in order to obtain the ideal solution in the Pareto frontier, the decision maker is often selected according to his subjective will. For example, if the decision maker focuses on economy, he will select solution set 1 with the minimum operating cost. If he hopes to reduce wind and solar power abandonment, he will select solution set 6 with the maximum new energy consumption.

[0250] The ideal solution in the Pareto frontier is obtained by using the combined weighting based on game theory and the TOPSIS comprehensive evaluation method, and the optimal trade-off between system operation cost and new energy consumption is obtained. The comprehensive scores of each solution set in Model 1 to Model 3 are as follows: Figure 8 shown.

[0251] exist Figure 8 In the figure, the models are respectively the new energy consumption, operation cost and comprehensive score corresponding to solution sets 1 to 6 from left to right, and Figure 7 The higher the TOPSIS comprehensive evaluation score, the better the solution set. Models 1 to 3 have the highest comprehensive scores in solution set 2, so the ideal solutions of each model are all solution set 2.

[0252] Ideal solutions and analysis of each model:

[0253] 1. Target values ​​and analysis of ideal solutions for each model

[0254] Models 1 to 3 obtain the ideal solution of optimal system operation cost and new energy consumption from the Pareto frontier as follows: Fig. 9 .

[0255] Fig. 9As can be seen from the table, for the island microgrid operation cost and new energy consumption target, model 1 has the largest total operation cost and the least new energy consumption; model 2 has the smallest total operation cost and the largest new energy consumption with frequency offset; model 3 is in between the two with cointegration. After the equality constraint is expanded to a constraint with frequency offset, the constraint range is relaxed, so a better target value is obtained.

[0256] 2. Power curves of each model

[0257] The power curves of each model under the ideal solution of isolated island microgrid are as follows: Figure 10-12 shown.

[0258] Depend on Figures 10 to 12 It can be seen that both coordinated operation and cointegrated operation combine the two dispatching resources of load and energy storage with diesel units and gas turbines to adapt to the fluctuation of new energy. New energy can be effectively absorbed through peak shaving and valley filling on the demand side, charging and discharging of energy storage, and regulation of thermal engine units.

[0259] In the 0-4 period, the output of new energy is large while the load demand is small. The battery system stores the excess electricity for backup. At the same time, the flexible load, as a responsive load, flexibly participates in the demand-side response and adjusts the system's energy consumption. A large amount of abandoned wind power is absorbed, which alleviates the mismatch between the output of new energy and the load demand to a certain extent. In the 16-20 period, the load demand increases while the output of new energy is small. The battery system releases the stored electricity, the flexible load transfers the use period of some electrical equipment, the thermal engine unit responds quickly, and provides stable power output to make up for the insufficient output of new energy.

[0260] exist Fig.11 and Fig.12 In the cointegration operation of model 2, the three types of dispatching resources, load, energy storage and frequency deviation allowable range, are combined with thermal units to adapt to the fluctuation of new energy. Compared with the source-load-storage coordinated optimization model of model 1, the cointegration operation of model 2 and model 3 expands the traditional real-time balance equation constraint of active power into the real-time balance equation constraint of the primary frequency modulation function and the frequency deviation inequality constraint. The feasible solution range is increased, the operation cost of the isolated grid is reduced, and the utilization rate of new energy is improved.

[0261] Power imbalance and frequency offset of each model

[0262] Both the coordinated operation of model 2 and the cointegrated operation of model 3 use the frequency offset allowable range as the frequency modulation resource, so there are power imbalance and frequency offset, such as Figures 13-16 In the coordinated operation of model 1, the allowed range of frequency offset is not used as a frequency modulation resource, and its power imbalance and frequency offset are both 0.

[0263] The coordinated operation of model 2 only considers the allowable range of microgrid frequency deviation as ±0.5Hz. In order to reduce the operating cost of the system and increase the absorption of new energy, it fully utilizes the allowable range of frequency deviation as a frequency modulation resource, resulting in multiple critical values ​​of frequency deviation (such as Fig.13 Compared with the other two operation modes, although it is more economical, it cannot guarantee the safety of system frequency.

[0264] In the cointegration operation of Model 3, the cointegration theory is used to characterize the equilibrium relationship between the non-stationary time series of the source-load-storage multi-power curves. Considering the correlation between the multiple power curves, on the basis of the upper and lower limit constraints of the frequency offset, the cointegration relationship between the total power supply and the total load (including energy storage) is established by constraining the stability of the power imbalance. The two variables of total power supply and total load maintain a long-term balance, so the power imbalance and frequency offset will not show an upward or downward trend. If the variable value at a certain moment deviates from its stable trend (such as Fig.15 21 periods), then a corrective change will occur at the next moment to adjust the variable back to its stable state. Therefore, the stability constraint ensures that the variables maintain a certain stable equilibrium relationship in the long run, even if there is a deviation in the short run. Fig.15 It can be seen that the frequency offset in the cointegration operation of model 3 always fluctuates around 0Hz, without touching the boundary of frequency offset, ensuring the frequency security of the isolated microgrid during operation. After adding the stability constraint, the cointegration relationship is established. Its essence is to prevent the frequency offset from being at a critical level. Therefore, the constraint is slightly tightened, the frequency safety level is appropriately improved, and the target value is slightly reduced, improving the optimization extreme value under the premise of ensuring normal operation.

[0265] in conclusion

[0266] When the power system is running, the power deviation caused by load and power supply fluctuations is called power imbalance. The power-frequency characteristics of the generator speed regulator and load make the system power reach a new balance, which is the result of the primary frequency regulation of the system. This real-time power balance containing the primary frequency regulation characteristics, namely the frequency deviation function, objectively reflects the operating status of the system.

[0267] In the source-load-storage coordinated optimization model, the real-time power balance with frequency offset is added as an equality constraint, and the allowable range of frequency offset is used as an available resource in the optimization scheduling, which relaxes the constraint conditions, increases the range of feasible solutions, and improves the optimization extreme value, thereby reducing the system operation cost and improving the new energy consumption rate.

[0268] In order to solve the frequency safety problem caused by the relaxation of constraints, the stability constraint of the frequency offset function is added to expand the special cointegration relationship of the original power real-time balance with zero residual sequence into a general cointegration relationship containing the frequency offset function. Then, the cointegration theory is applied to expand the source-load-storage coordination model into the source-load-storage cointegration model, which provides a theoretical basis.

[0269] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A source-load-storage cointegration optimization method for isolated microgrids with frequency offset constraints, characterized by: The steps include: S1: Model building Taking the minimization of the operating cost of the isolated microgrid and the maximization of the consumption of new energy as multiple objectives, Model 1, Model 2 and Model 3 are constructed: S11, Model 1: Source-load-storage coordinated operation model including thermal engine units, new energy units, demand-side response and batteries: S111. Set the objective function: S1111, Minimum operating cost: The operating cost includes the operating cost of the thermal unit C G (P dg,t ) and C MT (P mt,t ), demand side response cost C D (ΔP d,t ) Cost of curtailing wind and solar power C W (ΔP w,t ), battery energy storage operating cost C stor (P stor,t ). F1=min(C G (P dg,t )+C MT (P mt,t )+ C D (ΔP d,t )+C W (ΔP w,t )+C stor (P stor,t )) Where: U t and U t-1 Respectively represent the start and stop status of the thermal engine unit at time t and t-1; a, b, c represent the fuel cost parameters of the thermal engine unit; α, β, τ are the start and stop cost parameters of the thermal engine unit; δ is the downtime of the thermal engine unit; P dg,t is the output of the thermal engine unit in period t; C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η mt is the power generation efficiency of the gas turbine; ρ is the unit compensation cost of the demand response load; ΔP d,t is the load participating in demand response in period t; ω is the penalty coefficient for wind and solar power abandonment; ΔP w,t is the abandoned wind and solar power in period t; λ is the energy storage cost coefficient; P stor,t is the charging and discharging power of the energy storage during period t. Positive indicates charging, and negative indicates discharging. S1112, the largest consumption of new energy Where: P w,t is the actual output of new energy obtained by optimizing during period t, S112. Set constraints: S1121, Traditional Active Power Balance S1122, Energy Storage Output Constraints Energy storage output takes into account charging and discharging constraints and energy storage capacity constraints Where: The upper limit of energy storage charging and discharging power; X ch , Y dis It is a 0-1 variable, 0 represents energy storage not working, and 1 represents energy storage charging and discharging respectively; E es,t is the energy storage capacity at time t; E Bat is the rated capacity of energy storage; SOC min , SOC max It is the minimum and maximum value of energy storage charge state. S1123, demand-side response load constraints Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period, S12, Model 2: Based on Model 1, a source-load-storage coordinated operation model with frequency offset is constructed: The active power balance constraint for frequency offset is given by Where: t=1,2,...,T; ΔP t is the power imbalance; f0 is the rated frequency, which is 50Hz; k is the frequency modulation coefficient. The objective function and constraints of Model 2 are consistent with those of Model 1. S13, Model 3: Based on Model 2, a source-load-storage cointegration operation model with frequency shift stability constraints is constructed: The constraints of the active power cointegration balance equation with frequency offset are as follows: Where: ADF.test(kΔf t ) is the frequency shift function kΔf t The ADF stationarity test of is carried out. If the test constraint is met, the cointegration relationship between the source, load and storage power time series is established. The objective function and constraints of Model 3 are consistent with those of Model 1; S2: Case parameters The energy storage rated capacity, energy storage equipment parameters, and demand response parameters are set respectively, and a dispatch cycle and a dispatch period are set. Through the above settings, the unit compensation cost of the demand response load, the wind and solar power abandonment penalty coefficient, and the energy storage cost coefficient are calculated; S3: Simulation results and analysis S31, Pareto frontier and ideal solutions of each model The Pareto frontiers of Model 1, Model 2, and Model 3 are obtained by augmented ε constraint method. The number of segments g = 5 is selected, and the number of solution sets is 6, which are solution sets 1 to 6 from left to right. Solution set 1 and solution set 6 are the minimum operating cost and the maximum new energy consumption when optimized separately, respectively. As the amount of new energy consumed in the solution set continues to increase, its operating cost continues to increase accordingly, which means that there is no optimal solution set that can ensure the maximum amount of new energy consumed while minimizing the operating cost. Therefore, the Pareto frontier represents the set of possible solutions with a sacrificial relationship between the two goals of operating cost and new energy consumption. In practice, in order to obtain the ideal solution in the Pareto frontier, the decision maker is often selected according to his subjective will. For example, if the decision maker focuses on economy, he will select solution set 1 with the smallest operating cost. If he hopes to reduce wind and solar power abandonment, he will select solution set 6 with the largest new energy consumption. The game theory-based combined weighting and TOPSIS comprehensive evaluation method are used to obtain the ideal solution in the Pareto frontier and obtain the optimal balance between system operation cost and new energy consumption. Thus, the comprehensive scores of each solution set in Model 1, Model 2 and Model 3 are obtained. From left to right, the models are the new energy consumption, operation cost and comprehensive score corresponding to solution sets 1 to 6, which correspond to the solution sets in the Pareto frontier. The higher the TOPSIS comprehensive evaluation score, the better the solution set. Model 1, Model 2, and Model 3 all have the highest comprehensive scores in Solution Set 2. Therefore, the ideal solutions of the three models are all Solution Set 2. Model 1, Model 2 and Model 3 obtain the ideal solution of system operation cost and new energy consumption from the Pareto frontier, and analyze the operation cost and new energy consumption of isolated island microgrid: Model 1 has the largest total operation cost and the least new energy consumption in coordinated operation; Model 2 has the smallest total operation cost and the largest new energy consumption in coordinated operation with frequency offset; Model 3 operates in the middle of the two in cointegration. After the equality constraint is expanded to a constraint containing frequency offset, the constraint range is relaxed, thus obtaining a better target value; S32, analysis of power imbalance and frequency offset of each model The coordinated operation of model 2 and the cointegrated operation of model 3 both use the frequency offset allowable range as the frequency modulation resource, so there are power imbalance and frequency offset. The coordinated operation of model 1 does not use the frequency offset allowable range as the frequency modulation resource, and its power imbalance and frequency offset are both 0. Model 2 coordinated operation only considers the allowable range of microgrid frequency deviation as ±0.5Hz. Since it fully utilizes the allowable range of frequency deviation as frequency modulation resources, it leads to multiple critical values ​​of frequency deviation. Compared with the other two operation models, although it is economical, it cannot guarantee the safety of system frequency. In the cointegration operation of model 3, the cointegration theory is used to characterize the balance relationship between the non-stationary time series of the source-load-storage multi-power curves. Considering the correlation between the multi-power curves, on the basis of the upper and lower limit constraints of the frequency offset, the cointegration relationship between the total power supply and the total load is established by constraining the stability of the power imbalance. The total power supply and total load variables are kept in long-term balance, so the power imbalance and frequency offset will not show an upward or downward trend. If the variable value at a certain moment deviates from its stable trend, then a corrective change will occur at the next moment to readjust the variable back to its stable state. Therefore, the stability constraint ensures that the variables maintain a certain stable equilibrium relationship in the long term, even if there is a deviation in the short term. In the cointegration operation of model 3, the frequency offset always fluctuates around 0 Hz and does not touch the boundary of frequency offset, ensuring the frequency safety during the operation of the isolated microgrid. After adding the stability constraint, the cointegration relationship is established. Its essence is to prevent the frequency offset from being at a critical level. Therefore, the constraints will be tightened, the frequency safety level will be improved, and the target value will be reduced, thus improving the optimization extreme value while ensuring normal operation.

2. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 1, characterized in that: In step S1, the isolated island microgrid includes energy units, batteries, demand-side responses and loads. The energy units include thermal units and new energy units. The batteries are used for energy storage. The demand-side responses are microgrid operation control centers. The loads are used to control the flow of electric energy. The microgrid control center is used to control and receive information flows.

3. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 2, characterized in that: The thermal engine unit includes a diesel engine unit and a gas turbine, wherein: The cost function of the diesel generator in the diesel unit is: Where: U t and U t-1 Respectively represent the start and stop status of the diesel generator set at time t and t-1; a, b, c represent the fuel cost parameters of the diesel generator set; α, β, τ are the start and stop cost parameters of the diesel generator set; δ is the downtime of the diesel generator set; P dg,t is the output of the unit in period t; The gas turbine cost function is: Where: C is the unit price of fuel gas; Δt is the duration of the scheduling unit period; L HV is the lower calorific value of natural gas; P mt,t is the gas turbine output during period t; η mt The efficiency of gas turbine power generation.

4. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 2, characterized in that: The new energy units include wind turbines and photovoltaic units, wherein: The specific relationship between the output power and wind speed of the wind turbine in the wind turbine is: Where: v, v ci 、v co 、v r are the actual wind speed, rated wind speed, cut-in wind speed, and cut-out wind speed at the hub height of the wind turbine respectively; P r is the rated output of the fan, The distributed photovoltaic output power in the photovoltaic unit is: Where: P STC G is the rated power of photovoltaic output under standard conditions; AC is the irradiation intensity at the working point; k is the temperature power coefficient; T c is the battery temperature at the operating point; G STC is the solar irradiance; T t is the temperature of the battery under rated conditions.

5. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 2, characterized in that: From the perspective of load operation and scheduling, user loads are divided into rigid loads and flexible loads. Rigid loads cannot be regulated by the power system, while flexible loads are power loads that can adjust their power consumption behavior within the control range according to the power system demand and market signals. The flexible loads that can be called upon in each time period and scheduling cycle are limited. Where: P dmax , P dmin are the upper and lower limits of the demand side response load respectively; S dmax , S dmin They are the upper and lower limits of the total capacity of the demand-side response load within the dispatch period.

6. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 2, characterized in that: The function of battery state of charge is: Where: E BESS,t E is the energy storage charge and discharge amount in period t, a positive value indicates charging amount, and a negative value indicates discharging amount; Bat is the rated capacity of the energy storage.

7. The isolated island microgrid source-load-storage cointegration optimization method with frequency offset constraint according to claim 1, characterized in that: In step S2, the scheduling period is 24 hours and the scheduling period is 0.5 hours.

Images