A two-stage stochastic programming method for integrated energy microgrid based on multi-scenario technology

Abstract

The present invention relates to a two-stage random planning method for an integrated energy microgrid based on multi-scenario technology, which comprises the following steps: Step 1: Establishing a time-segment probability distribution model of uncertain factors; Step 2: Time correlation model and parameter optimization based on a hybrid Copula function; Step 3: Establishing a first-stage random planning model for an integrated energy microgrid based on a typical scenario; Step 4: Establishing a second-stage random planning model for an integrated energy microgrid based on a non-typical scenario. The present invention fully considers the uncertainty of photovoltaic, wind power, price and load, is more in line with actual statistical laws, and reduces the impact of uncertainty on the planning of integrated energy microgrids.

Description

Technical Field

[0001] The present invention belongs to the field of integrated energy system operation optimization, and specifically relates to a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology. Background Art

[0002] Renewable energy generation is renewable and pollution-free, which is of great significance to alleviating energy shortage and environmental pollution. As an important part of distributed power generation, distributed photovoltaic and wind power generation have been widely used in distribution networks and microgrids, making important contributions to improving the utilization rate of green energy and reducing user costs.

[0003] As a new integrated system that can integrate multiple energy sources such as cold, heat, electricity, and gas, and realize coordinated planning, optimized operation, and complementary assistance among different energy sources, the integrated energy system has received more and more attention and research. However, wind power, photovoltaic power generation, various types of loads, and prices in the integrated energy system are uncertain and change in real time. Some studies have not considered the time correlation of the operation of the integrated energy system. Research that takes this feature into consideration has certain limitations. Summary of the invention

[0004] The purpose of the present invention is to provide a two-stage stochastic planning method for an integrated energy microgrid based on multi-scenario technology, which fully considers the uncertainty of photovoltaic, wind power, price and load, is more in line with actual statistical laws, and reduces the impact of uncertainty on the planning of integrated energy microgrids.

[0005] The present invention adopts the following technical solution:

[0006] A two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology comprises the following steps:

[0007] Step 1: Establish a time-division probability distribution model of uncertain factors;

[0008] Step 2: Time correlation model and parameter optimization based on hybrid Copula function;

[0009] Step 3: Establish the first-stage stochastic programming model of the integrated energy microgrid based on typical scenarios;

[0010] Step 4: Establish the second-stage stochastic programming model of the integrated energy microgrid based on the non-typical scenario.

[0011] Furthermore, in step (1), a day is divided into d time periods T1, T2, ..., T d , the time-period probability distribution model of uncertain factors is:

[0012]

[0013] Where h is the bandwidth and n is the number of samples.

[0014] Furthermore, the kernel function in the time-division probability distribution model of the uncertain factors is a Gaussian kernel function.

[0015] Furthermore, the expression of the mixed Copula function is:

[0016] C(Z n ,Z m )=α1C1+α2C2+...α d C d

[0017] Among them, C d is the type of the dth Copula function, α d The weight of the dth Copula function.

[0018] Furthermore, the types of the Copula function include Normal-Copula function, t-Copula function, Frank-Copula function, Clayton-Copula function and Gumbel Copula function.

[0019] Furthermore, we ensure that T1, T2, ..., T d The time correlation of the probability distribution model in different time periods is the best, and the optimal calculation formula for the type and weight of the Copula function is:

[0020]

[0021] Where N is the sampling scale, F(ΔP t,i ) is the probability distribution function of the actual data, ΔP t,i =P t,i -P t-1,i , P t,i is the actual prediction error, F′(ΔP t,i ′) is the probability distribution function for generating data, ΔP t,i ′=P t,i ′-P t-1,i ′,P t,i ′ is the generated prediction error; when the value I is the smallest, it indicates that the parameters α1, α2, ..., α d Best.

[0022] Furthermore, according to the time-division probability distribution model and the mixed Copula function, the d probability distribution models are randomly sampled to obtain scenes by the sampling method; after the scenes are clustered by K-means clustering, S typical scenes O are obtained. s, s=1,2,...,S., the probability corresponding to each typical scene is ξ s , the rest of the scenes are atypical scenes O′ s′ , s′=1,2,...,NS., the probability corresponding to each atypical scene is ξ′ s′ .

[0023] Furthermore, the first-stage stochastic programming model of the integrated energy microgrid described in step 3 takes the total cost as the objective function, and the input of the model is a typical scenario, and the formula is:

[0024] minC=C inv +C grid +C fuel +C heat -C sold

[0025] Among them, C is the total cost, C inv is the investment cost, C grid is the cost of purchasing electricity from the power grid, C fuel is the cost of purchasing natural gas, C heat The cost of purchasing heat from the heating grid, C sold It is the income from selling electricity to the power grid, and the unit is Yuan.

[0026] Furthermore, the investment cost is the funds for purchasing various equipment, and the calculation formula is:

[0027]

[0028] Among them, γ is the capital recovery rate, c mt 、c gb 、c es,P 、c es,E 、c tst 、c ac 、c ec are the power of micro gas turbine, gas boiler, energy storage battery, capacity of energy storage battery, unit price of heat storage device, absorption chiller and air conditioner, respectively. mt,N,s , P gb,N,s , P es,N,s 、E es,N,s , H tst,N,s , Q ac,N,s , Q ec,N,s They are respectively the configured micro gas turbine power, gas boiler power, energy storage battery power, energy storage battery capacity, heat accumulator capacity, absorption chiller power and air conditioning power;

[0029] The calculation formula for the cost of purchasing electricity from the power grid is:

[0030]

[0031] The calculation formula for the cost of purchasing natural gas is:

[0032]

[0033] The calculation formula for the cost of purchasing heat energy from the heating network is:

[0034]

[0035] The calculation formula for the revenue from selling electricity to the grid is:

[0036]

[0037] Among them, ξ s represents the probability of the sth scene occurring, the subscript t represents the time, and the subscript s represents the sth scene; p grid,t,s is the electricity price of the power grid, yuan / kW; P grid,t,s is the purchased power, kW; Δt is the time interval, hours; p ng,t,s is the natural gas price, yuan / cubic meter; G gt,t,s is the calorific value of the gas from the micro gas turbine, kWh; G gb,t,s is the calorific value of gas in gas boiler, kWh; H ng is the calorific value of natural gas, kWh / cubic meter; p heat,t,s is the unit price of purchasing heat energy, RMB / kW; H heat,t,s is the power of purchased heat energy, kW; p sold,t,s is the electricity price, RMB / kW; P sold,t,s is the electricity sales power, kW.

[0038] Furthermore, the input of the second-stage stochastic programming model of the integrated energy microgrid described in step 4 is an atypical scenario, and the formula is:

[0039]

[0040] Where C is the total cost of the first-stage stochastic programming model, ξ′ s′ is the probability of the s′th atypical scenario occurring, κ l , κ w , κ pv are the penalty costs of load shedding, wind power abandonment and photovoltaic power generation, respectively, l,t,s′ 、x w,t,s′ 、x pv,t,s′ They are load shedding, wind power abandonment and photovoltaic power generation respectively.

[0041] The beneficial effects of the present invention are as follows: by establishing a prediction error time correlation model based on the covariance matrix, the present invention fully considers the uncertainty of photovoltaic, wind power, price and load in the optimization operation model of the integrated energy microgrid with multiple energy sources such as cooling, heating and electricity, and reduces the impact of the prediction error on the operation of the integrated energy microgrid. It has a certain inspiration for the study of the probability distribution statistics of random factors and the scenario generation method. The operation results take into account the uncertainty of the prediction data, which is more realistic. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 The present invention is a block diagram of the scenario-based integrated energy microgrid stochastic planning method.

[0043] Figure 2 Schematic diagram of the integrated energy microgrid in the present invention. DETAILED DESCRIPTION

[0044] The present invention will now be described in further detail with reference to the accompanying drawings.

[0045] According to the comprehensive energy microgrid stochastic programming method of the present invention, a specific process of comprehensive energy system optimization is given.

[0046] Figure 1 The block diagram of the scenario-based integrated energy microgrid stochastic planning method implemented in the present invention. As shown in the figure, the method includes the following steps:

[0047] Step 1: Establish a time-phased probability distribution model for uncertain factors.

[0048] The non-parametric kernel density estimation is used to establish the probability distribution model of uncertain factors in different time periods. The uncertain factors described in the present invention include: photovoltaic, wind power, electric load, cooling load, heating load and electricity purchase price.

[0049] Kernel density estimation is a type of nonparametric estimation. It is not based on the basic assumption that the data follows a distribution. It is a method that does not require any prior knowledge and studies the distribution characteristics of data completely from data samples. Assume x1, x2, ..., x i For n sample points (photovoltaic power) of independent and identical distribution F, let its probability density function be f, and the kernel density estimate is:

[0050]

[0051] Wherein, h is the bandwidth, n is the number of samples, and K(u) is the kernel function. The present invention uses the Gaussian kernel function,

[0052] In the present invention, a day is divided into d time periods T1, T2, ..., T d, the probability distribution model of uncertain factors in different periods is:

[0053]

[0054] Taking photovoltaic power generation as an example, the power of photovoltaic power generation in each time period corresponds to a probability distribution model, and there is a time correlation between each probability distribution model.

[0055] Step 2: Time correlation model and parameter optimization based on hybrid Copula function.

[0056] The hybrid Copula function is used to take into account the time correlation of d probability distribution models in the probability distribution model of different time periods of uncertain factors. Taking into account the time correlation can improve the fitting accuracy of the discrete scene to the original scene. The expression of the hybrid Copula function is:

[0057] C(Z n ,Z m )=α1C1+α2C2+...α d C d (3)

[0058] In the formula, C d is the type of the dth Copula function, α d The weight of the dth Copula function. d The value range is 0 to 1.

[0059] Currently, there are many Copula functions. We selected five representative functions, namely Normal-Copula function, t-Copula function, Frank-Copula function, Clayton-Copula function, and Gumbel Copula function. Each function has its own characteristics and is suitable for probability distribution models with different correlation characteristics. However, the characteristics of a single function are not enough to describe complex correlation characteristics. Therefore, we use a mixed Copula function.

[0060] The mixed Copula function in formula (3) contains multiple different types of Copula functions and weights. It is necessary to solve the best Copula function type and corresponding weight in the mixed Copula function to ensure that T1, T2, ..., T d The time correlation of the probability distribution model in different time periods is the best, and the optimal calculation formula of the Copula function type and weight is:

[0061]

[0062] Where N is the sampling scale, F(ΔP t,i) is the probability distribution function of the actual data, ΔP t,i =P t,i -P t-1,i , P t,i is the actual prediction error, F′(ΔP t,i ′) is the probability distribution function for generating data, ΔP t,i ′=P t,i ′-P t-1,i ′,P t,i ′ is the generated prediction error. When the value I of formula (4) is the smallest, it means that the generated scene best meets the random characteristics of photovoltaic power generation. The parameters α1, α2, ..., α d Best.

[0063] After the time-segment probability distribution model and the mixed Copula function are established, the d probability distribution models are randomly sampled by sampling to obtain the scene. In the present invention, the scene refers to the curve formed by the power of a certain device at different times in a day. PV For example, d probability distribution models are sampled at the same time, and the sampling results are P PV,1 ,P PV,2 ,...,P PV,d , the power curve formed by connecting d sampling results is a scene.

[0064] Similarly, assuming that d probability distribution models of photovoltaic power, wind power, electric load, cooling load, heating load and electricity purchase price are sampled at the same time, a scenario O can be obtained. s , s=1,2,...,N.,This scenario is a combination of scenarios with multiple uncertain factors, and its expression is:

[0065] O s :(P PV,s ,P W,s ,P L,s ,C L,s ,H L,s ,p grid,s ) (5)

[0066] Among them, P PV,s ,P W,s ,P L,s ,C L,s ,H L,s ,p grid,s They are the scenarios of photovoltaic, wind power, electric load, cooling load, heating load, and electricity purchase price.

[0067] In statistics, the more sampling times, the more obvious the statistical laws of the sampling results are. Therefore, multiple sampling is required to obtain a large number of scenes. These scenes need to be calculated in the stochastic programming model described in the present invention. However, the more the number, the greater the amount of calculation. Therefore, a large number of scenes are clustered by K-means clustering to obtain S typical scenes. s , s=1,2,...,S., the probability corresponding to each typical scene is ξ s , the rest of the scenes are atypical scenes O′ s′ , s′=1,2,...,NS., the probability corresponding to each atypical scene is ξ′ s′ .

[0068] Step 3: Establish the first-stage stochastic programming model of the integrated energy microgrid based on typical scenarios.

[0069] The input of the first stage planning model is the typical scenario in step 2. The purpose of the stochastic planning model is mainly to determine the size or capacity of various types of equipment in the integrated energy microgrid. The schematic diagram of the integrated energy microgrid system is shown in Figure 2 As shown in the figure, the main equipment included are photovoltaic, wind power, energy storage battery, air conditioner, micro gas turbine, absorption chiller, gas boiler and heat storage. The internal load demand is electricity, cooling and heating demand, and the external energy supply is the power grid, gas grid and heat grid.

[0070] In the present invention, the stochastic programming model of the first stage of the integrated energy microgrid is mainly a function with total cost as the target.

[0071] The goal of integrated energy microgrid planning is to minimize the total system cost, which includes the system investment cost and system operation cost. The system operation cost includes the cost of purchasing electricity from the grid, the cost of purchasing natural gas, the cost of purchasing heat energy from the heating grid, and the income from selling electricity to the grid.

[0072] The total cost of planning a comprehensive energy microgrid is calculated as follows:

[0073] minC=C inv +C grid +C fuel +C heat -C sold (6)

[0074] The investment cost is the funds for purchasing various types of equipment, and the calculation formula is:

[0075]

[0076] Among them, γ is the capital recovery rate, c mt 、c gb 、c es,P 、c es,E、c tst 、c ac 、c ec are the power of micro gas turbine, gas boiler, energy storage battery, capacity of energy storage battery, unit price of heat storage device, absorption chiller and air conditioner, respectively. mt,N,s , P gb,N,s , P es,N,s 、E es,N,s , H tst,N,s , Q ac,N,s , Q ec,N,s They are respectively the configured micro gas turbine power, gas boiler power, energy storage battery power, energy storage battery capacity, heat accumulator capacity, absorption refrigerator power and air conditioning power.

[0077] The calculation formula for the cost of purchasing electricity from the power grid is:

[0078]

[0079] The calculation formula for the cost of purchasing natural gas is:

[0080]

[0081] The calculation formula for the cost of purchasing heat energy from the heating network is:

[0082]

[0083] The calculation formula for the revenue from selling electricity to the grid is:

[0084]

[0085] Where C is the total cost, C inv is the investment cost, C grid is the cost of purchasing electricity from the power grid, C fuel is the cost of purchasing natural gas, C heat The cost of purchasing heat from the heating grid, C sold is the income from selling electricity to the power grid, in Yuan. s represents the probability of the sth scene occurring, the subscript t represents the time, and the subscript s represents the sth scene. grid,t,s is the electricity price of the power grid (yuan / kW), P grid,t,s is the purchased power (kW), Δt is the time interval (hours), p ng,t,s is the natural gas price (yuan / cubic meter), G gt,t,s is the calorific value of the gas from the micro gas turbine (kWh), G gb,t,s is the calorific value of gas in the gas boiler (kWh), H ng is the calorific value of natural gas (kWh / cubic meter). heat,t,s is the unit price of heat energy (yuan / kW), Hheat,t,s is the power of purchased heat energy (kW), p sold,t,s is the electricity price (yuan / kW), P sold,t,s is the electricity sales power (kW).

[0086] Step 4: Establish the second-stage stochastic programming model of the integrated energy microgrid based on the non-typical scenario.

[0087] The input of the second stage planning model is the atypical scenario in step 2. The formula of the second stage stochastic planning model is:

[0088]

[0089] Where C is the total cost of the first-stage stochastic programming model, ξ′ s′ is the probability of the s′th atypical scenario occurring, κ l , κ w , κ pv are the penalty costs of load shedding, wind power abandonment and photovoltaic power generation, respectively, l,t,s′ 、x w,t,s′ 、x pv,t,s′ They are load shedding, wind power abandonment and photovoltaic power generation respectively.

[0090] Step 5: Solution method of two-stage stochastic programming model for integrated energy microgrid

[0091] The present invention adopts a genetic algorithm to solve the two-stage random programming model of the integrated energy microgrid. First, the scenario generated in step 2 is substituted into the calculation model; secondly, an initial population is randomly generated, and parameters such as population size, crossover rate, mutation rate and maximum number of iterations are set. The fitness of all individuals in the population, that is, the objective function value of each individual, is calculated, and then the previous generation of population is selected, crossed and mutated to generate a new generation of population, and individuals with higher fitness function values ​​in the new generation of population are replaced and saved; finally, until the maximum number of iterations is reached or the objective function value changes to meet the set conditions, the optimal solution is output and the optimal solution and its corresponding probability under each scenario are substituted into the expected value model to obtain the optimal planning result.

Claims

1. A two-stage stochastic programming method for integrated energy microgrid based on multi-scenario technology, characterized in that: It includes the following steps: Step 1: Establish a time-division probability distribution model of uncertain factors; Divide a day into d time periods T1, T2, …, T d , the time-period probability distribution model of uncertain factors is: Where h is the bandwidth and n is the number of samples; Step 2: Time correlation model and parameter optimization based on hybrid Copula function; The expression of the mixed Copula function is: C(Z n ,Z m )=α1C1+α2C2+…+α d C d Among them, C d is the type of the dth Copula function, α d is the weight of the dth Copula function; Step 3: Establish the first-stage stochastic programming model of the integrated energy microgrid based on typical scenarios; The first-stage stochastic programming model of the integrated energy microgrid takes the total cost as the objective function, and the input of the model is a typical scenario, and the formula is: minC=C inv +C grid +C fuel +C heat -C sold Among them, C is the total cost, C inv is the investment cost, C grid is the cost of purchasing electricity from the power grid, C fuel is the cost of purchasing natural gas, C heat The cost of purchasing heat from the heating grid, C sold is the revenue from selling electricity to the power grid, in RMB; Step 4: Establish the second-stage stochastic programming model of the integrated energy microgrid based on atypical scenarios; The input of the second stage stochastic programming model of the integrated energy microgrid is an atypical scenario, and the formula is: Where C is the total cost of the first-stage stochastic programming model, ξ′ s′ is the probability of the s′th atypical scenario occurring, κ l , κ w , κ pv are the penalty costs of load shedding, wind power abandonment and photovoltaic power generation, respectively, l,t,s′ 、x w,t,s′ 、x pv,t,s′ They are load shedding, wind power abandonment and photovoltaic power generation.

2. According to claim 1, a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology is characterized in that: The kernel function in the time-division probability distribution model of the uncertain factors is a Gaussian kernel function.

3. According to claim 1, a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology is characterized in that: The types of the Copula function include normal-Copula function, t-Copula function, Frank-Copula function, Clayton-Copula function and Gumbel Copula function.

4. According to claim 3, a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology is characterized in that: Ensure that T1, T2, …, T d The time correlation of the probability distribution model in different time periods is the best, and the optimal calculation formula for the type and weight of the Copula function is: Where N is the sampling scale, F(ΔP t,i ) is the probability distribution function of the actual data, ΔP t,i =P t,i -P t-1,i , P t,i is the actual prediction error, F′(ΔP′ t,i ) is the probability distribution function for generating data, ΔP′ t,i =P′ t,i -P′ t-1,i is the generated prediction error; when the value I is the smallest, it indicates that the parameters α1, α2, …, α d Best.

5. According to claim 4, a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology is characterized in that: According to the time-division probability distribution model and the mixed Copula function, the d probability distribution models are randomly sampled to obtain the scenarios by the sampling method; the scenarios are clustered by K-means clustering to obtain S typical scenarios O s , s=1,2,…,S, the probability corresponding to each typical scene is ξ s , the rest of the scenes are atypical scenes O′ s′ , s′=1,2,…,NS, the probability corresponding to each atypical scene is ξ′ s′ .

6. According to claim 1, a two-stage stochastic programming method for an integrated energy microgrid based on multi-scenario technology is characterized in that: The investment cost is the funds for purchasing various types of equipment, and the calculation formula is: Among them, γ is the capital recovery rate, c mt 、c gb 、c es,P 、c es,E 、c tst 、c ac 、c ec are the power of micro gas turbine, gas boiler, energy storage battery, capacity of energy storage battery, unit price of heat storage device, absorption chiller and air conditioner, respectively. mt,N,s , P gb,N,s , P es,N,s 、E es,N,s , H tst,N,s , Q ac,N,s , Q ec,N,s They are respectively the configured micro gas turbine power, gas boiler power, energy storage battery power, energy storage battery capacity, heat accumulator capacity, absorption chiller power and air conditioning power; The calculation formula of the power grid purchase cost is: The calculation formula for the cost of purchasing natural gas is: The calculation formula for the cost of purchasing heat energy from the heating network is: The calculation formula for the revenue from selling electricity to the grid is: Among them, ξ s represents the probability of the sth scene occurring, the subscript t represents the time, and the subscript s represents the sth scene; p grid,t,s is the electricity price of the power grid, yuan / kW; P grid,t,s is the purchased power, kW; Δt is the time interval, hours; p ng,t,s is the natural gas price, yuan / cubic meter; G gt,t,s is the calorific value of the gas from the micro gas turbine, kWh; G gb,t,s is the calorific value of gas in gas boiler, kWh; H ng is the calorific value of natural gas, kWh / cubic meter; p heat,t,s is the unit price of purchasing heat energy, RMB / kW; H heat,t,s is the power of purchased heat energy, kW; p sold,t,s is the electricity price, RMB / kW; P sold,t,s is the electricity sales power, kW.

Images