RIES multi-objective optimization planning method considering source load prediction uncertainty

Abstract

The invention provides an RIES multi-objective optimization planning method considering the uncertainty of source load prediction, and the method comprises the steps: building an RIES basic structure, generating an uncertainty scene set of a source load prediction error through a Latin hypercube sampling method, carrying out the reduction of the uncertainty scene set through a clustering technology, and carrying out the reduction of the uncertainty scene set. According to the method, total cost, carbon dioxide emission and system reliability are taken as objective functions, an RIES multi-objective optimization planning model considering source load prediction uncertainty is constructed, a multi-objective optimization function is solved by adopting a non-dominated sorting genetic algorithm NSGA-III, a Pareto leading set is obtained, and through a strategy of combining a fuzzy entropy theory and TOPSIS, a Pareto leading set is obtained. And selecting the optimal solution of the Pareto leading edge set to obtain a multi-target optimization planning result of each typical scene. The method can comprehensively consider the influence of uncertain source load prediction on planning, improves the reliability of the system and the utilization rate of renewable energy sources, reduces the investment and operation and maintenance cost of the system, and reduces the emission of carbon dioxide at the same time.

Description

Technical Field

[0001] The present invention relates to the technical field of integrated energy system optimization, and in particular to a RIES multi-objective optimization planning method considering source-load prediction uncertainty. Background Art

[0002] With the gradual depletion of fossil energy and the increasing global environmental pollution, the energy situation will face unprecedented challenges. Regional integrated energy system (RIES) is an effective way to alleviate the current contradiction between energy supply and demand and reduce pollutant emissions, and therefore has attracted much attention. RIES integrates multiple energy forms such as electricity, natural gas, heat energy, and cold energy, and can achieve coordinated planning and unified scheduling of various energy sources. It is an effective model for promoting efficient cascade utilization of energy and emission reduction. Therefore, RIES has become a research hotspot in the field of global energy systems.

[0003] Under the RIES architecture, a large number of energy conversion devices are increasingly coupled in the supply, transmission, and demand links. However, different energy sources are coupled to each other through a large number of conversion devices, and different heterogeneous energy sources are converted and distributed through coupling devices, which brings great challenges to the coordinated planning and energy management of RIES. The introduction of wind, light, and storage into RIES further exacerbates the multiple uncertain challenges faced by RIES in the planning and operation process, increases the difficulty of planning and operation modeling, and deterministic planning technology is no longer applicable.

[0004] In the in-depth study of RIES, the source-load uncertainty modeling of RIES occupies a pivotal position. Given the inherent complexity and high uncertainty of the energy demand and supply system, it is crucial to accurately and reasonably construct uncertainty models for all aspects. It can not only effectively guide and optimize the planning and design of RIES, but also significantly improve the operating efficiency of RIES and further ensure the safe and stable supply of energy. Therefore, accurate modeling of source-load uncertainty in RIES has become an important topic of common concern and exploration in academia and industry.

[0005] RIES source-load uncertainty modeling is a complex and important field. In order to effectively deal with these uncertainties, researchers have adopted a variety of modeling methods. Commonly used methods include probability theory, fuzzy logic, interval analysis, random processes, and scenario methods. The probability theory method describes uncertainty by modeling the statistical characteristics of energy demand and supply; however, the probability theory method has the disadvantages of large computational complexity and high data processing complexity. The fuzzy logic method models source-load uncertainty through fuzzy sets and fuzzy rules, but it requires expert knowledge, has a certain degree of subjectivity, and cannot provide accurate quantitative results. Interval analysis represents uncertainty in the energy system through interval numbers, but this method is too rough and cannot quantify detailed information on the actual distribution of variables; random processes can model the dynamic behavior of complex systems, but it requires processing of a large number of random variables and equations, and also requires the support of a large amount of data. Some literatures have proposed scenario-based uncertainty modeling methods that can integrate different modeling techniques and information, and comprehensively consider various possibilities of uncertainty by constructing multiple scenarios, thereby providing more comprehensive analysis results. Compared with the full probability model method, it can effectively reduce the computational complexity of the model. Some literatures set 8 scenarios based on cooling, heating and electricity loads when optimizing the comprehensive energy scheduling in uncertain areas. However, the above studies consider renewable energy production or load uncertainty for scenario division separately, and rarely consider both at the same time. In addition, the impact of scenario division methods on optimization planning results is rarely discussed.

[0006] To achieve the expected advantages of RIES, reasonable early planning and later optimization operation are very important. Its essence is a complex problem involving multiple objectives, such as economy, environment, reliability, etc. At present, the main multi-objective optimization methods include weighted summation method, ε-constraint method, Pareto optimization, hierarchical optimization method, etc. The advantages of weighted summation method are simple and easy to understand and easy to implement. Its disadvantages are that the selection of weights is too subjective. When there is an obvious conflict between objectives, it cannot achieve a balance between objectives. For non-convex problems, it can only find a local optimal solution instead of a global optimal solution. The ε-constraint method can effectively handle multiple objectives and focus on the optimization of the main objectives at the same time. It is suitable for different types of objective functions. Its disadvantages are that it needs to reasonably set the constraint threshold, which has a large amount of calculation and is difficult to find a suitable constraint range. The hierarchical optimization method can give priority to important objectives and ensure the optimization of the main objectives. The disadvantage is that the relationship between objectives at different levels is not clear, resulting in an incoherent optimization process. The optimization of low-level objectives is restricted by high-level objectives, which affects the final result. It cannot fully utilize the relationship between objectives, resulting in the omission of potential solutions. Pareto optimization can provide a set of Pareto optimal solutions, showing the trade-offs between different objectives. It does not require setting weights in advance and is highly objective. NAGAII Non-dominated Sorting Genetic Algorithm II is a Pareto frontier optimization method based on genetic algorithm, which is specifically used to solve multi-objective optimization problems. It has the advantages of fast running speed, good solution convergence, and low computational complexity. It has been widely used in energy system optimization research. Some literature uses the improved NSGA-II method to effectively solve the Pareto frontier of multi-objective optimization problems of integrated energy systems. There are also literatures that use NSGA-II to construct a comprehensive evaluation index for distributed multi-energy complementary energy systems with the goals of economic, environmental protection, and energy efficiency optimization. However, when dealing with multidimensional problems, NSGA-II has weak convergence ability, resulting in uneven distribution of solutions and increased computational complexity. NSGA-III is an improved algorithm of NSGA-II. It introduces the reference point method specifically for high-dimensional multi-objective optimization. It can effectively find uniformly distributed solutions on the Pareto frontier, guide the search and improve the distribution of solutions, ensure diversity, have better convergence on high-dimensional problems, and can more accurately approach the true Pareto frontier. Especially when dealing with larger populations, NSGA-III is more efficient and can quickly adapt to different numbers of objectives. Summary of the invention

[0007] The purpose of the present invention is to provide a RIES multi-objective optimization planning method that takes into account the uncertainty of source-load prediction, which can comprehensively consider the impact of source-load prediction uncertainty on planning, improve system reliability and renewable energy utilization, and reduce system investment and operation and maintenance costs, while reducing carbon dioxide emissions.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is: a RIES multi-objective optimization planning method considering the uncertainty of source load prediction, characterized in that it includes the following steps: Step 1: Establish the RIES energy conversion model; Step 2: Based on the established RIES energy conversion model, use the historical supply and demand data to predict the source load data and calculate the prediction error. Based on the obtained prediction error, establish an uncertainty scenario set and classify and eliminate it; Step 3: Establish a multi-objective optimization function with total system cost, carbon dioxide emissions, and purchased energy as targets; Step 4: Use NSGA-III to solve the multi-objective function to obtain the Pareto solution set, and use the strategy combining fuzzy entropy theory with TOPSIS to select the Pareto solution set to obtain the best multi-objective optimization planning result.

[0009] Preferably, the RIES in step 1 includes a variety of energy supply and energy conversion equipment, including photovoltaic PV, wind power WT, micro gas turbine MT, gas boiler GB, electric refrigerator EC, absorption refrigeration AC, electric boiler EB, carbon capture device CCS, combined cooling, heat and power CCHP, power to gas P2G, electric storage device ESS, heat storage device HSS, cold storage device CSS and gas storage device GSS.

[0010] Preferably, the energy conversion and input and output energy balance relationship of electricity, heat, cold and gas in the RIES energy conversion model in step 1 is as follows: Where L is the load matrix; C is the coupling matrix; P is the multi-energy supply matrix; are the electricity, cooling and heating, and gas load demands respectively; θ is the proportion of electricity allocated to electrical equipment, θ and σ are the proportions of natural gas consumed by CCHP and GB respectively, θ, σ∈[0,1]; They are CCHP, EB, GB, EC, AC, and P2G multi-energy conversion efficiency. The charge and discharge power of multi-energy storage, positive indicates discharge, negative indicates charge; K G is the calorific value of natural gas; It is the energy supply of the system in RIES.

[0011] Preferably, in step 2, electricity, cooling, heating, and gas load demands and prediction errors are used as sampling sources, and a Latin hypercube sampling method is used to simulate uncertainty scenarios, and a large number of possible source-load uncertainty scenarios are generated through multiple random samplings.

[0012] Preferably, generating a large number of possible source-load uncertainty scenarios by multiple random samplings comprises the following steps: Step 201: Define the sampling space and divide it into intervals. Assume that the source load prediction error is an m-dimensional random vector X = [X1, X2, ... X m ], where each X i The sampling space is [c i ,b i ], each dimension X i Divided into N probability intervals, the width of each interval is shown as follows: Step 202: Generate a random order and select points in the interval. Randomly select a number in each interval of each dimension to generate a random order from 1 to N. For each dimension i and the jth number in the order, calculate the corresponding interval [c i +(j-1)Δ i ,c i +jΔ i ), and then the nth sampling value is obtained by inverse transformation, and then the N sampling values ​​of each dimension are arranged into a column to form an N×m initial sampling matrix A0. The element a in the matrix A0 mn The expression is as follows: in is the inverse cumulative distribution function of the m-dimensional variable; s mn is a random variable uniformly distributed in the range [0,1]; Step 203: Disturbance sorting, rearranging each column in the matrix A0 to form a final sampling matrix A of the uncertainty of the source-load prediction error.

[0013] Preferably, clustering and reducing the original sampled source load prediction uncertainty scenario set in step 2 includes the following steps: Step 211: Initialization, randomly selecting D scenes as initial cluster centers; Step 212: Reordering, select K scenes from the total scene set A and reorder them to obtain a sorting matrix A1. The reordering calculation process is shown in the following formula: A1=A{(K ω -K1·cosθ,K ω +K1·cosθ),1:m} Where ω∈[0,1], K ω =K·ω,θ∈[0,90]; Step 213: Distance calculation, for each scene s in A1 i , calculate its relationship with all cluster centers cj The distance is as follows: Among them, s ik It's scenes i The kth scene, c jk is the cluster center c j and the kth scene; Step 214: Update and iterate, recalculate the center of each cluster, as shown in the following formula: In the above formula, C j is the set of all scenes in cluster j, |C j | is the number of scenes in cluster j, and it is assigned to the nearest cluster center until the cluster center no longer changes significantly; Step 215: scene selection, select a representative scene from each cluster and take the cluster center as the representative, so as to obtain the final scene reduction result.

[0014] Preferably, the multi-objective optimization function in step 3 includes economic objectives, environmental objectives, and reliability objectives; the economic objectives are composed of initial investment costs, annual operation and maintenance costs, and purchased energy costs, and various costs are calculated as shown in the following formula: minC TOT =C IIC +C OM +C COE where c IIC,L is the equivalent annual value coefficient of the planning object, calculated as follows: L is the capacity of the planning object, ω L is the unit capacity investment cost of the corresponding equipment; Where r is the discount rate, which is set to 7%. T is the planned life of the equipment. The annual operation and maintenance cost is calculated as follows: Where i is the device, c is om is the maintenance cost, E e The cost of purchased energy is calculated based on the capacity of the equipment as follows: where c e , c g are the time-of-use price of purchased energy, P buy , G buy They are purchased electricity and natural gas; The total carbon dioxide emissions of the environmental target are composed of carbon emissions generated by CCS within RIES and indirect carbon emissions generated by purchasing electricity from the power grid. The smaller the value, the higher the environmental target, as shown in the following formula: Where C co2 is the total carbon dioxide emissions, and are the natural gas consumed by CCS and the purchased electricity, U c,g and U c,e is the carbon dioxide emission factor; The less the total purchased power of the reliability target RIES is, the higher its reliability is. The calculation formula of the reliability index is as follows: Among them, B EN It represents the total amount of purchased energy of RIES. The smaller its value is, the higher the reliability of the system is.

[0015] Preferably, the constraints of the multi-objective optimization function in step 3 include energy balance constraints and purchased energy constraints; the energy balance constraints are based on the constraints of electricity, heating, and cooling as equation constraints, and the energy balance constraints satisfy the energy balance constraints between the electric energy supply equipment model, the cold energy supply equipment model, the heat energy supply equipment model, the natural gas supply equipment model, the energy storage model, and the carbon dioxide production and consumption model, and also satisfy the output constraints of each device and the CO2 balance constraint; the equipment output constraint is shown in the following formula: The CO2 balance constraint is as follows: The purchased energy constraint is as follows: Where P buy,max and G buy,max Represent the maximum amount of purchased electricity and natural gas respectively.

[0016] Preferably, the step 4 of using NSGA-III to solve the multi-objective function to obtain the Pareto solution set includes the following steps: Step 401: Initialization, generating a set of initial population P, the population size is N; Step 402: Evolution operation, select some individuals in the population for crossover operation to generate a new offspring population Q. Generate a new solution by exchanging some genetic information of the parent individuals, then perform mutation operation on the individuals in the offspring population to increase the diversity of the population, and then merge the parent population P and the offspring population Q to form a new population R with a size of 2N; Step 403: Non-dominated sorting and selection. Perform non-dominated sorting on the population R and divide the individuals in the population into multiple non-dominated layers. NSGA-III introduces a reference point mechanism to maintain the diversity of the population. First, generate a set of evenly distributed reference points in the target space. Then, associate the individuals in each non-dominated layer with the reference points, select the non-dominated individuals close to the reference points to enter the next generation population, and add the individuals to the new population S in turn until the size of S reaches N. Step 404: Iteration and termination, repeat the above evolution operation until the preset number of iterations or convergence condition is reached, and then extract all non-dominated solutions from the final population S, which is the Pareto optimal solution set of the multi-objective optimization planning.

[0017] Preferably, the strategy of combining fuzzy entropy theory with TOPSIS to select the Pareto solution set in step 4 includes the following steps: Step 411: Initialization, assuming that there is a set of m decision solutions, each solution has n attributes, and each attribute is normalized; Step 412: Determine the weight, determine the weight ω of each attribute i , so that Step 413: Calculate the fuzzy entropy. For each attribute i, calculate the fuzzy entropy according to its fuzzy membership function μ ij (x), the fuzzy entropy H is calculated as follows i , Step 414: Calculate the weight coefficient according to the fuzzy entropy H i Calculate the weight coefficient K of each attribute i , as shown below: Fuzzy entropy H i The larger the K i The smaller it is, the higher the fuzziness of attribute i is, and the smaller the impact on the decision result is; Step 415: Calculate the weighted normalized score. For each decision solution j, calculate its weighted normalized score λ j , as shown below: Among them, x ij is the normalized value of scheme j on attribute i; Step 416: Calculate the similarity of the solutions, and calculate the distance between each solution and the positive and negative optimal solutions, as shown in the following formula: in and are the weighted scores of the positive and negative optimal solutions on attribute i respectively; Step 417: Calculate the similarity index and sort and select, calculate the similarity index ζ of each solution j , as shown below: The schemes are sorted according to the similarity index, and the scheme with the highest similarity index is selected as the optimal decision scheme. TOPSIS sorts the decision schemes by calculating the similarity index, and uses fuzzy entropy to provide a fuzziness measure for each attribute to help determine the weight and evaluate the rationality of the decision.

[0018] The beneficial effects of the present invention are:

[0019] This scheme comprehensively considers the impact of uncertainty in source-load forecasting on planning, aiming to improve system reliability, renewable energy utilization, and reduce system investment and operation and maintenance costs, while reducing carbon dioxide emissions. Through in-depth theoretical analysis and actual simulation verification, the following conclusions are drawn:

[0020] (1) The planning model proposed in this paper not only realizes the planning of energy production equipment, energy conversion equipment, and energy storage equipment in RIES, but also plans carbon capture and storage (CCS) and power-to-gas (P2G) equipment. In addition to conventional investment and operating costs, the model also studies the reliability and carbon emissions of the system. The simulation results significantly show that in typical scenarios 9 and 10, when the installed capacity of renewable energy increases by 30MW, the total cost can be reduced by 660,000 yuan, which verifies the effectiveness of the model in improving energy efficiency and cost control.

[0021] (2) By comparing the three different planning schemes, it is found that the annual total cost of Case 3 is significantly lower than that of Case 1 and Case 2, reducing by 1.46 million yuan and 940,000 yuan respectively. This result strongly proves that the method of considering the uncertainty of source-load forecasting error can further optimize the planning and operation costs of RIES and help reduce carbon dioxide emissions, thereby improving the overall reliability of the system. This finding is of great significance for improving the practicality of comprehensive planning and operation schemes for regional integrated energy systems.

[0022] (3) The current scheme has not considered the conversion relationship between hydrogen production, hydrogen storage system and energy storage power station system and RIES. Future research can include the integration of hydrogen production and hydrogen storage system and system operation into planning goals, in order to achieve more efficient use of energy and further improve the reliability and robustness of the RIES system. By continuously improving the planning model, it is expected to promote the development of regional integrated energy systems in a more intelligent, efficient and environmentally friendly direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1 It is a structural schematic diagram of the regional integrated energy system RIES of the present invention.

[0025] Figure 2 It is a schematic diagram of the uncertainty scene generation process and reduction results of the present invention.

[0026] Figure 3 This is a RIES planning framework diagram of the present invention that takes into account the uncertainty of source load prediction errors.

[0027] Figure 4 It is a flow chart for solving the multi-objective optimization planning model of the present invention.

[0028] Figure 5 It is a typical daily load and daily wind power and photovoltaic output curve of the present invention.

[0029] Figure 6 This is a schematic diagram of the RIES energy supply and demand balance operation scheduling results in the transition season of the present invention.

[0030] Figure 7 This is a schematic diagram of the winter RIES energy supply and demand balance operation scheduling results of the present invention.

[0031] Figure 8 This is a schematic diagram of the summer RIES energy supply and demand balance operation scheduling results of the present invention.

[0032] Fig. 9 It is a schematic diagram of the optimal planning results of different schemes of the present invention. DETAILED DESCRIPTION

[0033] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0034] The present invention discloses a RIES multi-objective optimization planning method considering the uncertainty of source load prediction. Figure 1 As shown, the embodiment includes the following steps: Step 1: Establish the RIES energy conversion model; Step 2: Based on the established RIES energy conversion model, use the historical supply and demand data to predict the source load data and calculate the prediction error. Based on the obtained prediction error, establish an uncertainty scenario set and classify and eliminate it; Step 3: Establish a multi-objective optimization function with total system cost, carbon dioxide emissions, and purchased energy as targets; Step 4: Use NSGA-III to solve the multi-objective function to obtain the Pareto solution set, and use the strategy combining fuzzy entropy theory with TOPSIS to select the Pareto solution set to obtain the best multi-objective optimization planning result.

[0035] RIES is a complex multi-energy coupling system that integrates power supply equipment, cooling equipment, heating equipment, electricity, cold / heat energy storage equipment, and renewable energy output equipment, and can meet the diverse energy needs of users at the same time. In order to make RIES give full play to the maximum benefits of the equipment in the system, it is necessary to analyze the operating mechanism of each device in the system and establish a corresponding mathematical model to analyze the conversion and flow path of energy in the system. The characteristics of multi-energy integration make the optimization planning objectives more diversified. Different energy links are different at the equipment level and system level. It is necessary to clarify the factors affecting energy utilization efficiency in production operation and the mechanism of action of related factors, and on this basis, propose a RIES structure and model suitable for multi-objective optimization planning.

[0036] The RIES in step 1 covers a variety of energy supply and energy conversion equipment, including photovoltaic PV, wind power WT, micro gas turbine MT, gas boiler GB, electric refrigerator EC, absorption refrigeration AC, electric boiler EB, carbon capture device CCS, combined cooling, heat and power CCHP, power to gas P2G, and adds energy storage devices such as electric storage device ESS, heat storage device HSS, cold storage device CSS, gas storage device GSS. RIES load mainly includes electricity, heat, cold and gas. The structure of the RIES of the present invention is as follows Figure 1 shown.

[0037] Figure 1 It includes the relevant conversion equipment of RIES from the source side to the load side through the intermediate multi-energy coupling. The source side is supplied with electricity and natural gas by the power grid and gas grid, and the load side is consumed by electricity, cold, heat, and gas users. The P2G equipment in the intermediate coupling equipment absorbs CO2 provided by CCS and consumes electricity to convert it into natural gas. CCS captures CO2 gas produced by CCHP, MT, and GB. MT consumes natural gas to generate heat and electricity. GB consumes gas to generate heat. EB converts electricity into heat. AC absorbs heat energy and evaporates the refrigerant by heating. Then it absorbs environmental heat and evaporates in the evaporator to achieve the cooling effect. EB consumes electricity to generate cold energy. CCHP converts natural gas into cold, heat, and electricity to supply the load demand in RIES. Multi-energy coupling equipment cannot completely convert from one energy source to another in the mutual conversion. There is a problem of conversion efficiency, so conversion efficiency needs to be considered in modeling.

[0038] For the RIES energy conversion model in step 1, the conversion between multiple energy forms such as electricity, heat, cold, and gas and the input-output energy balance relationship are a core issue. This balance relationship can usually be described by one or more mathematical expressions, reflecting the conversion efficiency, input-output quantity, and interaction between different energy forms in the system. According to the general characteristics of the RIES energy conversion model, the energy conversion and input-output energy balance relationship of electricity, heat, cold, and gas is given as shown in formula (1): L=CP+P NSS (1)

[0039] Where L is the load matrix; C is the coupling matrix; P is the multi-energy supply matrix; P NSS is the charging and discharging power matrix of each energy storage device. The multi-energy supply matrix P is as shown in formula (2).

[0040] In the formula Respectively represent the total electricity, cold, heat, and gas energy at time t, P e EN (t),P e PV (t),P e WT (t),P e CCHP (t) represents the electric energy generated by EN, PV, WT, and CCHP at time t, P c CCHP (t),P c EC (t),P c AC (t) represents the cooling energy generated by CCHP, EC, and AC at time t, P h CCHP (t),P h EB (t),P h GN (t) represents the heat energy generated by CCHP, EB, and GB at time t, represents the natural gas produced by P2G at time t, The power is purchased from natural gas.

[0041] According to the conversion relationship of each unit in RIES, substituting equation (2) into equation (1), the energy conversion and input and output energy balance relationship of electricity, heat, cold and gas can be written as shown in equation (3):

[0042] In the formula are the electricity, cooling and heating, and gas load demands respectively; θ is the proportion of electricity allocated to electrical equipment, θ and σ are the proportions of natural gas consumed by CCHP and GB respectively, θ, σ∈[0,1]; They are CCHP, EB, GB, EC, AC, P2G multi-energy conversion efficiency, P e ESS (t),P h HSS (t),P c CSS (t),P g GSS (t) The charge and discharge power of the multi-energy storage, where positive indicates discharge and negative indicates charge; K G is the calorific value of natural gas; It is the energy supply of the system in RIES.

[0043] For the RIES energy conversion model in step 1, multiple energy supplies are mainly provided by different devices. There is a certain amount of energy loss in the conversion between multiple energy sources. It is not a 100% conversion from one energy to another. The conversion efficiency needs to be considered in the modeling. The following is an explanation from four aspects: electricity, cold, heat, and gas.

[0044] (1) Power supply equipment model

[0045] Depend on Figure 1 It can be seen that the power supply equipment includes photovoltaic power generation, wind power generation, MT, CCHP, etc. The mathematical models of each power supply equipment are shown in equations (4)-(7).

[0046] The power of photovoltaic power generation system is easily affected by external weather changes and has strong randomness. The photovoltaic output is related to light intensity and temperature. The photovoltaic power generation power is calculated as follows: P e PV (t) = η PV AI(t) (4)

[0047] Where η PV is the conversion efficiency of photovoltaic power generation, A is the area of ​​the photovoltaic array, and I(t) is the light intensity.

[0048] The output power of wind power generation WT is closely related to the wind speed v, as shown in formula (5). P e WT (t) = P(t)·C P =0.5ρ·A T ·V 3 (t)·C P (5)

[0049] In the formula, C P is the power coefficient of the wind turbine impeller, which is taken as 0.593; P WT (t) is the wind power actually converted by the wind turbine; P(t) is the wind power; A T is the area swept by the impeller; ρ is the air density. P The relationship between the tip speed ratio and the pitch angle is as follows:

[0050] Wherein, the coefficients are: c1 = 0.5176, c2 = 116, c3 = 0.4, c4 = 5, c5 = 21, c6 = 0.0068. λ is the tip speed ratio, β is the pitch angle.

[0051] MT converts the input natural gas into electrical energy and heat energy output. CCHP uses natural gas as the main fuel and converts it into electrical energy, heat energy, and cold energy output. Its conversion power model can be described by formula (7).

[0052] P e MT (t) is the electrical power generated by MT; is the thermal power, For power generation efficiency, is the waste heat loss rate, K g The natural calorific value, For the consumed natural gas.

[0053] (2) Cold energy supply equipment model

[0054] Cold energy is mainly provided by EC, AC, and CCHP equipment. EC converts electrical energy into cold energy, AC converts heat energy into cold energy, and the waste heat discharged after CCHP generates electricity is used to provide cooling to users through waste heat recovery equipment. The model is shown in formula (8):

[0055] Where P c EC (t), P c AC (t), P c CCHP (t) are the cooling power generated by EC, AC, and CCHP, respectively, P e EC (t) is the power consumption of EC, P h AC (t) is the thermal power consumed by AC, P g CCHP (t) is the natural gas consumed by CCHP, are the conversion efficiencies, respectively.

[0056] (3) Heat supply equipment model

[0057] Thermal energy is mainly provided by MT, CCHP, GB, EB and other equipment. The mathematical model of each energy supply equipment is shown in formula (9):

[0058] In the formula, is the thermal power generated by MT, CCHP, GB, and EB, is the conversion efficiency of each device, P e MT is the electrical power output by MT, P e EB (t) is the electrical power consumed by EB, and is the natural gas power consumed by CCHP and GB.

[0059] (4) Natural gas supply equipment model

[0060] When there is excess electricity in the RIES system, P2G consumes electricity and uses the carbon dioxide collected by CCS to convert it into natural gas. This embodiment does not consider other added gases. The model is as shown in formula (10):

[0061] In the formula, is the gas production power of P2G, P e P2G (t) power consumption, For conversion efficiency.

[0062] For the RIES energy conversion model in step 1, energy storage adopts a unified model, as shown in equations (11)-(16): γ nch (t)+γ ndis (t)≤1 (11)

[0063] Where γ nch (t),γ ndis (t) Provide charging and discharging marks, and represents the initial and final capacity, T n nss (t+1) and is the capacity at time t+1 and time t, Indicates the charging and discharging efficiency, Indicates the charging and discharging power.

[0064] For the RIES energy conversion model in step 1, CCS captures the carbon dioxide produced by CCHP, MT and GB, and P2G consumes the CO2 captured by CCS. The mathematical model of CCS is shown in formula (17):

[0065] In the above formula, represents the CO2 collected by CCS at time t, are the CO2 emission rates of CCHP, MT and GB respectively. The electric power consumed by CCS is as shown in formula (18):

[0066] In the formula, They are the electric power consumed by CCS and the energy conversion efficiency respectively.

[0067] P2G first generates hydrogen and oxygen by electrolyzing water, and then reacts hydrogen with carbon dioxide to generate methane and water in the methanogenization device. Its consumption of carbon dioxide is shown in formula (19):

[0068] In the above formula Indicates the P2G output natural gas volume and conversion efficiency.

[0069] To ensure the safe and stable operation of the regional integrated energy system RIES, while meeting the needs of all loads in RIES and maintaining sufficient spare capacity. Since wind power generation and photovoltaic power generation are affected by natural conditions and other external factors, there are errors between the predicted results and the actual power generation; there are also certain errors between the load forecast and the actual demand. When the error exceeds a certain range, these errors must be taken into account. Therefore, it is necessary to consider the uncertainty of source and load forecasts in the RIES planning stage.

[0070] Uncertainty in source-load forecasting is the main cause of volatility and uncertainty between energy supply and demand, and has multiple impacts on RIES power balance and dispatching operations: (1) Difficulty in balancing supply and demand: Volatility between energy supply and demand makes it difficult for the system to maintain a balance between supply and demand. (2) Increased operating costs: Uncertainty in source-load forecasting errors will increase the operating costs of the system. (3) Reduced energy efficiency: Uncertainty will affect the energy efficiency of the system. In the case of an imbalance between supply and demand, energy waste or energy shortage will occur, reducing the effective utilization of energy. (4) System planning and design challenges: Uncertainty increases the difficulty of system planning and design. When designing and planning a regional integrated energy system, it is necessary to consider the mutual influence between different energy forms and the impact of uncertainty on system performance.

[0071] The uncertainty of source load prediction error is affected by many factors, such as wind speed, light intensity, load demand, etc. Existing research methods mainly use probability density function fitting of specific distribution to describe the uncertainty of source load prediction error. In fact, the probability density distribution of RIES source load uncertainty set is asymmetric, and it is difficult to describe its uncertainty using a specific probability function, which affects the early planning and capacity configuration of the RIES system.

[0072] To simulate uncertainty scenarios, this embodiment uses electricity, cooling, heating, and gas load demands and prediction errors as sampling sources, and uses the Latin hypercube sampling method. Through multiple random samplings, a large number of possible source-load uncertainty scenarios are generated. The Latin hypercube sampling method can evenly select sampling points and ensure that the sampling points in each row and column are distributed as evenly as possible, thereby improving the efficiency and accuracy of sampling. The specific implementation steps are as follows:

[0073] (1) Define the sampling space and divide it into intervals

[0074] Assume that the source load prediction error is an m-dimensional random vector X = [X1, X2, ... X m ], where each X i The sampling space is [c i ,b i ]. Each dimension X i It is divided into N probability intervals, and the width of each interval is shown in (20).

[0075] (2) Generate a random order and select a point within the interval

[0076] Randomly select a number in each interval of each dimension to generate a random permutation from 1 to N. For each dimension i and the jth number in the sort, calculate the corresponding interval [c i +(j-1)Δ i ,c i +jΔ i ). Then the nth sampling value is obtained by inverse transformation, and the N sampling values ​​of each dimension are arranged into a column to form an N×m initial sampling matrix A0. The element a in the matrix A0 mn The expression is shown in formula (21).

[0077] in is the inverse cumulative distribution function of the m-dimensional variable; s mn is a random variable uniformly distributed in the range [0,1].

[0078] (3) Perturbation sorting

[0079] In order to maintain the independence of the samples, each column in the matrix A0 is rearranged to form the final sampling matrix A of the source-load prediction error uncertainty, as shown in Figure 2 (a) shown.

[0080] In order to reduce the amount of calculation, the original sampled source load prediction uncertainty scenario set is clustered and reduced. The specific implementation steps are as follows:

[0081] Step 1: Initialization

[0082] Randomly select D scenes as initial cluster centers.

[0083] Step 2: Reorder

[0084] Select K scenes from the total scene set A and re-rank them to obtain the ranking matrix A1. The re-ranking calculation process is shown in formula (22), and the ranking result is shown in Figure 2 (b) as shown. A1=A{(K ω -K1·cosθ,K ω +K1·cosθ),1:m} (22) Where ω∈[0,1], K ω =K·ω,θ∈[0,90].

[0085] Step 3: Distance Calculation

[0086] For each scenario s in A1 i , calculate its relationship with all cluster centers c j The distance is shown in formula (23).

[0087] Among them, s ik It's scenes i The kth scene, c jk is the cluster center c j and the kth scene.

[0088] Step 4: Update and Iterate

[0089] Recalculate the center of each cluster as shown in equation (24):

[0090] In the above formula, C j is the set of all scenes in cluster j, |C j | is the number of scenes in cluster j. And assign it to the nearest cluster center until the cluster center no longer changes significantly.

[0091] Step 5: Scene Selection

[0092] Select a representative scene from each cluster and take the cluster center as the representative. The final scene reduction result is obtained, such as Figure 2 (c) as shown.

[0093] Step 3 establishes the basic framework of the model considering the uncertainty of source-load prediction; establishes a multi-objective optimization function with economy, environment and reliability as the goals, uses the minimum total cost to characterize the best economic goal, uses the minimum carbon emission to characterize the environmental goal, and uses the minimum purchased energy to characterize the reliability goal; and gives the specific solution process of the model and the methods used.

[0094] According to the characteristics and optimization objectives of RIES, the basic framework of the multi-objective optimization planning model is constructed as follows: Figure 3 As shown. A multi-objective optimization function is established with economy, environment and reliability as the goals, with the minimum economic input as the economic goal; the minimum carbon dioxide emission as the environmental goal; and the minimum external energy purchase amount of the system as the reliability goal. The specific model is as follows:

[0095] (1) Economic objectives

[0096] The economic goal is composed of the initial investment cost, annual operation and maintenance cost, and external energy cost. The various cost calculations are expressed as follows: min C TOT =C IIC +C OM +C COE (25)

[0097] where c IIC,L is the equivalent annual value coefficient of the planning object, calculated as in formula (27), E L is the capacity of the planning object, ω L is the unit capacity investment cost of the corresponding equipment.

[0098] Where r is the discount rate, which is set to 7% in this embodiment, and T is the planned life of the equipment. The annual operation and maintenance cost is calculated as shown in formula (28):

[0099] Where i is the device, c is om is the maintenance cost, E e The capacity of the corresponding equipment. The cost of purchased energy is calculated as shown in formula (29):

[0100] where c e , c g are the time-of-use price of purchased energy, P buy , Gbuy They are purchased electricity and natural gas respectively.

[0101] (2) Environmental objectives

[0102] The total carbon dioxide emissions are composed of the carbon emissions generated by CCS in the RIES system and the indirect carbon emissions generated by purchasing electricity from the power grid. The smaller the value, the higher the environmental target, as shown in formula (30).

[0103] Where C co2 is the total carbon dioxide emissions, and are the natural gas consumed by CCS and the purchased electricity, U c,g and U c,e is the carbon dioxide emission factor.

[0104] (3) Reliability

[0105] The less the total external power purchased by RIES, the higher its reliability. The calculation formula of reliability index is shown in (31).

[0106] Among them, B EN It represents the total amount of purchased energy of RIES. The smaller its value is, the higher the reliability of the system is.

[0107] In order to ensure that the optimization results meet the various constraints of the actual planning, it is necessary to further consider the value range of the optimization variables and the constraint relationship between the variables. Constraints mainly include equality constraints and inequality constraints. Equality constraints are used to ensure the balance of supply and demand configuration, while inequality constraints are used to limit the feasible solution space of the optimization plan.

[0108] Energy balance is based on the constraints of electricity, heating and cooling as equation constraints. The electricity demand is met by photovoltaic PV, wind power WT, gas turbine GT, grid EN, and electric storage ES. The heat load is shared by CCHP, absorption chiller AC, electric boiler EB and heat storage HSS. The cooling load can be met by AC and EC. The energy supply strategy is to determine the demand for heat energy based on cooling energy, and the remaining shortfall is supplied by electricity.

[0109] In addition to the energy balance constraints of equations (4)-(19) above, the following output and balance constraints of each device must also be met.

[0110] (1) Equipment output constraints:

[0111] (2) CO2 balance constraints:

[0112] When internal electricity and natural gas cannot meet the load demand, they need to be purchased from the external network, and the constraints are as shown in formula (36):

[0113] Where P buy,max and G buy,max Represent the maximum amount of purchased electricity and natural gas respectively.

[0114] For the proposed multi-objective optimization planning model, this embodiment uses the NSGA-III algorithm to solve it. The solution process is as follows: Figure 4 As shown in the figure, it is implemented by the following two parts: (1) NSGA-III is used to solve the multi-objective optimization model function and obtain the three-objective Pareto solution set; (2) the final planning result is obtained by combining fuzzy entropy theory with TOPSIS.

[0115] 1) NSGA-III solves the multi-objective optimization model process

[0116] (1) Initialization

[0117] Generate an initial population P with a population size of N.

[0118] (2) Evolutionary Operation

[0119] Select some individuals in the population for crossover operation to generate a new offspring population Q. New solutions are generated by exchanging some genetic information of the parent individuals. Then perform mutation operation on the individuals in the offspring population to increase the diversity of the population. Then merge the parent population P and the offspring population Q to form a new population R with a size of 2N.

[0120] (3) Non-dominated sorting and selection

[0121] The population R is non-dominated and sorted, and the individuals in the population are divided into multiple non-dominated layers. NSGA-III introduces a reference point mechanism to maintain the diversity of the population. First, a set of uniformly distributed reference points is generated in the target space. Then, the individuals in each non-dominated layer are associated with the reference points, and the non-dominated individuals close to the reference points are selected to enter the next generation of the population. Individuals are added to the new population S in sequence until the size of S reaches N.

[0122] (4) Iteration and Termination

[0123] Repeat the above evolutionary operation until the preset number of iterations or convergence condition is reached. Then extract all non-dominated solutions from the final population S, which is the Pareto optimal solution set for multi-objective optimization planning.

[0124] 2) Strategy selection

[0125] There is no globally unique optimal solution for multi-objective optimization problems. The solution result is an optimal solution set consisting of multiple groups of non-dominated Pareto optimal solutions. After solving the Pareto frontier, this embodiment uses a method based on the combination of fuzzy entropy theory and TOPSIS to construct a compromise solution selection strategy. The combination of the two can select the best combined optimal solution from the Pareto optimal solution set. The specific implementation process is as follows:

[0126] (1) Initialization

[0127] Assume that there is a set of m decision options, each option has n attributes, and each attribute is normalized.

[0128] (2) Determine the weight

[0129] Determine the weight ω of each attribute i , so that

[0130] (3) Calculate fuzzy entropy

[0131] For each attribute i, according to its fuzzy membership function μ ij (x), and the fuzzy entropy H is calculated as in formula (38) i .

[0132] (4) Calculate the weight coefficient

[0133] According to the fuzzy entropy H i Calculate the weight coefficient K of each attribute i , as shown in formula (39)

[0134] Fuzzy entropy H i The larger the K i The smaller it is, the higher the fuzziness of attribute i is and the smaller its impact on the decision result is.

[0135] (5) Calculate the weighted normalized score

[0136] For each decision option j, calculate its weighted normalized score λ j , as shown in formula (40)

[0137] Among them, x ij is the normalized value of scheme j on attribute i.

[0138] (6) Calculate the similarity of the solutions

[0139] Calculate the distance between each solution and the positive and negative optimal solutions, as shown in formula (41)

[0140] in and are the weighted scores of the positive and negative optimal solutions on attribute i, respectively.

[0141] (7) Calculate similarity index and sort and select

[0142] Calculate the similarity index ζ for each solution j , as shown in formula (42):

[0143] The schemes are sorted according to the similarity index, and the scheme with the highest similarity index is selected as the optimal decision scheme. TOPSIS sorts the decision schemes by calculating the similarity index, and uses fuzzy entropy to provide the fuzziness measurement of each attribute to help determine the weight and evaluate the rationality of the decision. The combination of the two can more comprehensively consider the fuzziness and uncertainty between attributes, thereby obtaining the best three-objective combination decision scheme.

[0144] This example uses a comprehensive energy system in a city in northwest China as a case study. The system structure is as follows: Figure 1 The load power and wind and solar output curves on a typical day are shown in Figure 5 As shown in the figure, a year is 365 days long, divided into 182 days of transition season, 94 days of summer, and 89 days of winter. Among them, the decision variables are PV, WT, MT, CCS, P2G, ESS, GSS, HSS, GB, CCHP, and the optimization step size is designed to be 10kW. The overall energy conservation principle is to use electric cooling and electric heating as the bottom line, so EC, EB, and CSS are no longer optimized as decision variables.

[0145] The performance parameters of energy storage equipment and the economic parameters of equipment are shown in Table 1-2, and the price of purchased energy is shown in Table 3. The relevant parameter settings of NSGA-III are shown in Table 4. Table 1 Energy storage equipment parameter settings Table 2 Equipment economic parameters Table 3 Electrical price list name Electricity￥ / kWh Gas￥ / m3 Time Peak period 0.9 0.48 8:00-12:0019:00-21:00 Normal period 0.6 0.48 13:00-19:0021:00-24:00 Low point 0.3 0.48 24:00-7:00 Table 4 Parameter settings of NSGA-III

[0146] Based on the planning model proposed in this embodiment, the Pareto frontier is obtained through NSGA-III iteration, and the fuzzy entropy theory and TOPSIS method are used to compromise and select the solution of the Pareto frontier set. This strategy takes the best economy, carbon dioxide emissions, and reliability as the goals to obtain the optimal planning scheme, mainly considering the investment and fixed maintenance costs in the planning stage, as well as the energy purchase and operation and maintenance costs in the operation stage.

[0147] Based on the scenario reduction results, the planning results of RIES energy equipment under typical scenarios are listed in Table 5. The composition of RIES planning cost is shown in Table 6. Table 5 Multi-objective optimization planning results under typical scenarios Typical PV WT EB EC EA CCS GSS HSS ESS CCHP GB MT 1 200 330 300 400 300 200 300 250 300 300 250 300 2 220 310 250 320 300 180 300 230 280 270 250 300 3 250 300 280 340 280 200 260 250 280 300 230 280 4 300 320 250 350 300 200 280 220 270 280 250 300 5 260 300 260 330 270 160 270 240 250 280 240 270 6 200 300 200 300 260 180 200 220 200 200 230 250 7 250 280 290 370 290 170 250 230 260 250 250 300 8 300 320 270 360 250 180 240 250 290 280 250 260 9 320 300 300 320 300 200 250 200 250 290 240 290 10 300 290 300 300 280 200 300 250 300 300 250 300 Table 6 Total cost of RIES system in typical scenarios Based on the planning results in Table 5 and the total system cost of typical scenarios in Table 6, the final planning results are shown in Table 7: Table 7 RIES planning results

[0148] According to the planning results in Table 7, the energy supply and demand balance operation and dispatch results of RIES in the transition season are as follows: Figure 6 , Figure 7 , Figure 8 shown.

[0149] Depend on Figure 6 (a) It can be seen that RIES gives priority to wind power to meet its electricity demand. The shortage is first met by MT, and then the electricity demand is met by external purchase. As time goes by, photovoltaic power generation begins to generate electricity, and the output of MT decreases with the increase of renewable energy power output. For example, at 13:00, the power output of MT is reduced to 0. Figure 6 (b) It can be seen that the heat load demand in the transition season is first supplied by MT, the insufficient part is supplied by GB, and the remaining heat energy is stored by the heat storage system. When the heat load demand increases at 10-11 o'clock, the heat energy provided by MT and GB cannot meet the system, and the heat storage system begins to release heat at this time; Figure 6 (c) It can be seen that the cooling load is first supplied by AC, and the shortfall is provided by EC; the gas load demand is as follows Figure 6 As shown in (d), it can be seen that the gas load demand is mainly met by P2G and purchased natural gas.

[0150] Depend on Figure 7It can be seen that in winter, the demand for heating load increases, while the demand for cooling load decreases. The demand for electricity and natural gas also increases. The reason is that in order to meet the increased demand for heating load, more electricity and natural gas are often needed to provide heat. Specifically, the increase in electricity demand is because the heating system requires more electricity to maintain a suitable indoor temperature. The increase in natural gas demand is because many heating systems use natural gas to generate heat.

[0151] The operation and scheduling results for a typical summer day are as follows: Figure 8 As shown in the figure, the cooling demand increased, which was mainly met by EC and CSS, while the shortage was made up by EC. The heating demand decreased, and the demand for natural gas and electricity increased compared with the transition season.

[0152] The balance of supply and demand of regional integrated energy and multiple energy sources is affected by many factors. On the one hand, operators can achieve more effective energy management and load regulation through collaboration with the user side. Providing flexible energy supply solutions, while users adjust their own load curves through demand response measures, the cooperation of the two can better cope with fluctuations in the energy market and promote the balance of energy supply and demand. On the other hand, multi-type demand response measures on the user side can complement the load. When the demand for electricity is high, natural gas heating can be increased to reduce the pressure on electricity demand. Users can also achieve time shifting of loads by adjusting the energy consumption time. Using electrically driven equipment during low peaks in electricity demand and other energy sources during peak demand can smooth the load curve and avoid the imbalance between supply and demand caused by peak demand.

[0153] In order to verify the effectiveness of the model proposed in this embodiment, three typical case simulation analyses are conducted to study the planning effects under different strategy combinations. The specific cases are described as follows.

[0154] Case 1: Applying the single-objective optimization planning method without considering the impact of source load prediction uncertainty.

[0155] Case 2: Applying the single-objective optimization planning method, considering the impact of source-side prediction uncertainty.

[0156] Case 3: Apply multi-objective optimization planning method to consider the impact of source load forecast uncertainty.

[0157] Fig. 9 The optimal planning results of different schemes are shown in Table 8. The optimization results of the three schemes are shown in Table 8. The cost calculation results of different schemes are shown in Table 9. The multi-objective optimization output parameters are shown in Table 10. Table 8 Comparison of planning results of different schemes Table 9 Cost calculation results of different schemes Table 10 Multi-objective optimization output parameters based on different schemes

[0158] It can be found that among the three schemes, the planned capacity of renewable energy in Scheme 3 is the smallest, and the annual total cost of Scheme 3 is the lowest, which is 1.46 million yuan and 940,000 yuan less than that of Scheme 1 and Scheme 2, respectively. In addition, the annual total cost of Case 3 is lower than that of Case 1 and Case 2, indicating that the method of considering the uncertainty of source-load forecasting error can further reduce the planning and operation costs of RIES. As shown in Table 8, the carbon dioxide emissions in Scheme 3 are the smallest, which are reduced by 16.9% and 9.5% compared with Schemes 1 and 2, respectively, and the reliability is improved by 13% and 5%, respectively. The main reasons are as follows: (1) By more accurately considering the source-load forecasting error, the required spare capacity of the system can be better evaluated. The over-configured spare capacity to cope with forecasting uncertainty can be reduced, thereby reducing pre-operation investment and subsequent operating costs. Considering the uncertainty of forecasting error helps to better dispatch available multi-energy resources, so that energy production and consumption can be closer to actual demand, thereby reducing the phenomenon of wind and solar power abandonment. (2) Better understanding and coping with forecasting errors can reduce energy market price fluctuations caused by supply and demand imbalances. Stable demand forecasts can reduce the high costs caused by unstable market prices. It can also reduce failures caused by system overload, ensure system stability and reliability, and reduce operation and maintenance costs. (3) In the system planning stage, considering the uncertainty of prediction errors can more accurately predict the required infrastructure investment, thereby avoiding unnecessary overinvestment and reducing capital expenditure. More flexible and efficient planning strategies can be formulated to improve resource allocation efficiency and return on investment. (4) The planning model that considers the uncertainty of prediction errors enables the energy system to better respond to sudden changes in demand or supply, improve the system's resilience, and reduce the additional costs caused by emergencies.

[0159] In conclusion, by considering the uncertainty of source-load forecast error in RIES planning, the operating efficiency of RIES can be significantly improved, resource waste and cost fluctuations can be reduced, thereby reducing costs and carbon dioxide emissions in planning and operation and increasing the system reliability of RIES.

[0160] It should be noted that the parts not described in detail in the above embodiments are all prior art.

[0161] The above description is only a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention should be covered within the protection scope of the present invention.

Claims

1. A RIES multi-objective optimization planning method considering the uncertainty of source load prediction, characterized by: The following steps are involved: Step 1: Establish the RIES energy conversion model; Step 2: Based on the established RIES energy conversion model, use the historical supply and demand data to predict the source load data and calculate the prediction error. Based on the obtained prediction error, establish an uncertainty scenario set and classify and eliminate it; Step 3: Establish a multi-objective optimization function with total system cost, carbon dioxide emissions, and purchased energy as targets; Step 4: Use NSGA-III to solve the multi-objective function to obtain the Pareto solution set, and use the strategy combining fuzzy entropy theory with TOPSIS to select the Pareto solution set to obtain the best multi-objective optimization planning result.

2. A RIES multi-objective optimization planning method considering source-load prediction uncertainty according to claim 1, characterized in that: The RIES in step 1 includes a variety of energy supply and energy conversion equipment, including photovoltaic PV, wind power WT, micro gas turbine MT, gas boiler GB, electric refrigerator EC, absorption refrigeration AC, electric boiler EB, carbon capture device CCS, combined cooling, heat and power CCHP, power to gas P2G, electric storage device ESS, heat storage device HSS, cold storage device CSS and gas storage device GSS.

3. The RIES multi-objective optimization planning method considering the uncertainty of source-load prediction according to claim 2 is characterized by: The energy conversion and input and output energy balance relationship of electricity, heat, cold and gas in the RIES energy conversion model in step 1 is shown in the following formula: Where L is the load matrix; C is the coupling matrix; P is the multi-energy supply matrix; are the electricity, cooling, heating and gas load demands respectively; θ is the proportion of electricity allocated to electrical equipment, σ is the proportion of natural gas consumed by CCHP and GB, They are CCHP, EB, GB, EC, AC, and P2G multi-energy conversion efficiency. The charge and discharge power of multi-energy storage, positive indicates discharge, negative indicates charge; K G is the calorific value of natural gas; It is the energy supply of the system in RIES.

4. The RIES multi-objective optimization planning method considering the uncertainty of source-load prediction according to claim 1 is characterized by: In step 2, electricity, cooling, heating, and gas load demands and prediction errors are used as sampling sources, and the Latin hypercube sampling method is used to simulate uncertainty scenarios. A large number of possible source-load uncertainty scenarios are generated through multiple random samplings.

5. The RIES multi-objective optimization planning method considering the uncertainty of source-load prediction according to claim 4 is characterized by: The method of generating a large number of possible source-load uncertainty scenarios by multiple random samplings includes the following steps: Step 201: Define the sampling space and divide it into intervals. Assume that the source load prediction error is an m-dimensional random vector X = [X1, X2, ... X m ], where each X i The sampling space is [c i ,b i ], each dimension X i Divided into N probability intervals, the width of each interval is shown as follows: Step 202: Generate a random order and select points in the interval. Randomly select a number in each interval of each dimension to generate a random order from 1 to N. For each dimension i and the jth number in the order, calculate the corresponding interval [c i +(j-1)Δ i ,c i +jΔ i ), and then the nth sampling value is obtained by inverse transformation, and then the N sampling values ​​of each dimension are arranged into a column to form an N×m initial sampling matrix A0. The element a in the matrix A0 mn The expression is as follows: in is the inverse cumulative distribution function of the m-dimensional variable; s mn is a random variable uniformly distributed in the range [0,1]; Step 203: Disturbance sorting, rearranging each column in the matrix A0 to form a final sampling matrix A of the uncertainty of the source-load prediction error.

6. A RIES multi-objective optimization planning method considering source-load prediction uncertainty according to claim 5, characterized in that: The clustering and reduction of the original sampled source load prediction uncertainty scenario set in step 2 includes the following steps: Step 211: Initialization, randomly selecting D scenes as initial cluster centers; Step 212: Reordering, select K scenes from the total scene set A and reorder them to obtain a sorting matrix A1. The reordering calculation process is shown in the following formula: A1=A{(K ω -K1·cosθ,K ω +K1·cosθ),1:m} Where ω∈[0,1], K ω =K·ω,θ∈[0,90]; Step 213: Distance calculation, for each scene s in A1 i , calculate its relationship with all cluster centers c j The distance is as follows: Among them, s ik It's scenes i The kth scene, c jk is the cluster center c j and the kth scene; Step 214: Update and iterate, recalculate the center of each cluster, as shown in the following formula: In the above formula, C j is the set of all scenes in cluster j, |C j | is the number of scenes in cluster j, and it is assigned to the nearest cluster center until the cluster center no longer changes significantly; Step 215: scene selection, select a representative scene from each cluster and take the cluster center as the representative, so as to obtain the final scene reduction result.

7. The RIES multi-objective optimization planning method considering the uncertainty of source-load prediction according to claim 1 is characterized by: The multi-objective optimization function in step 3 includes economic objectives, environmental objectives, and reliability objectives; The economic target is composed of initial investment cost, annual operation and maintenance cost and purchased energy cost. The various costs are calculated as shown in the following formula: minC TOT =C IIC +C OM +C COE where c IIC,L is the equivalent annual value coefficient of the planning object, calculated as follows: L is the capacity of the planning object, ω L is the unit capacity investment cost of the corresponding equipment; Where r is the discount rate, which is set to 7%. T is the planned life of the equipment. The annual operation and maintenance cost is calculated as follows: Where i is the device, c is om is the maintenance cost, E e The cost of purchased energy is calculated based on the capacity of the equipment as follows: where c e , c g are the time-of-use price of purchased energy, P buy , G buy They are purchased electricity and natural gas; The total amount of carbon dioxide emissions of the environmental target is composed of carbon emissions generated by CCS within RIES and indirect carbon emissions generated by purchasing electricity from the power grid. The smaller the value, the higher the environmental target, as shown in the following formula: Where C co2 is the total carbon dioxide emissions, and are the natural gas consumed by CCS and the purchased electricity, U c,g and U c,e is the carbon dioxide emission factor; The less the total purchased power of the RIES reliability target is, the higher its reliability is. The calculation formula of the reliability index is as follows: Among them, B EN It represents the total amount of purchased energy of RIES. The smaller its value is, the higher the reliability of the system is.

8. A RIES multi-objective optimization planning method considering source-load prediction uncertainty according to claim 7, characterized in that: The constraints of the multi-objective optimization function in step 3 include energy balance constraints and purchased energy constraints; The energy balance constraint is an equation constraint that uses the constraints of electricity, heating, and cooling as constraints. The energy balance constraint satisfies the energy balance constraints among the electric energy supply equipment model, the cold energy supply equipment model, the heat energy supply equipment model, the natural gas supply equipment model, the energy storage model, and the carbon dioxide production and consumption model, and also satisfies the output constraints of each device and the CO2 balance constraint; The equipment output constraint is as follows: The CO2 balance constraint is as follows: The purchased energy constraint is as follows: Where P buy,max and G buy,max Represent the maximum amount of purchased electricity and natural gas respectively.

9. The RIES multi-objective optimization planning method considering the uncertainty of source-load prediction according to claim 1 is characterized by: The step 4 of using NSGA-III to solve the multi-objective function to obtain the Pareto solution set includes the following steps: Step 401: Initialization, generating a set of initial populations P, the population size is N; Step 402: Evolution operation, select some individuals in the population for crossover operation to generate a new offspring population Q. Generate a new solution by exchanging some genetic information of the parent individuals, then perform mutation operation on the individuals in the offspring population to increase the diversity of the population, and then merge the parent population P and the offspring population Q to form a new population R with a size of 2N; Step 403: Non-dominated sorting and selection. Perform non-dominated sorting on the population R and divide the individuals in the population into multiple non-dominated layers. NSGA-III introduces a reference point mechanism to maintain the diversity of the population. First, generate a set of evenly distributed reference points in the target space. Then, associate the individuals in each non-dominated layer with the reference points, select the non-dominated individuals close to the reference points to enter the next generation population, and add the individuals to the new population S in turn until the size of S reaches N. Step 404: Iteration and termination, repeat the above evolution operation until the preset number of iterations or convergence condition is reached, and then extract all non-dominated solutions from the final population S, which is the Pareto optimal solution set of the multi-objective optimization planning.

10. The RIES multi-objective optimization planning method considering source-load prediction uncertainty according to claim 1 is characterized by: The strategy of combining fuzzy entropy theory with TOPSIS in step 4 to select the Pareto solution set includes the following steps: Step 411: Initialization, assuming that there is a set of m decision solutions, each solution has n attributes, and each attribute is normalized; Step 412: Determine the weight, determine the weight ω of each attribute i , so that Step 413: Calculate the fuzzy entropy. For each attribute i, calculate the fuzzy entropy according to its fuzzy membership function μ ij (x), the fuzzy entropy H is calculated as follows i , Step 414: Calculate the weight coefficient according to the fuzzy entropy H i Calculate the weight coefficient K of each attribute i , as shown below: Fuzzy entropy H i The larger the K i The smaller it is, the higher the fuzziness of attribute i is, and the smaller the impact on the decision result is; Step 415: Calculate the weighted normalized score. For each decision solution j, calculate its weighted normalized score λ j , as shown below: Among them, x ij is the normalized value of scheme j on attribute i; Step 416: Calculate the similarity of the solutions, and calculate the distance between each solution and the positive and negative optimal solutions, as shown in the following formula: in and are the weighted scores of the positive and negative optimal solutions on attribute i respectively; Step 417: Calculate the similarity index and sort and select, calculate the similarity index ζ of each solution j , as shown below: The schemes are sorted according to the similarity index, and the scheme with the highest similarity index is selected as the optimal decision scheme. TOPSIS sorts the decision schemes by calculating the similarity index, and uses fuzzy entropy to provide a fuzziness measure for each attribute to help determine the weight and evaluate the rationality of the decision.

Images