Electricity-gas-hydrogen system distribution robust optimization method considering hydrogen energy automobile

Abstract

The invention relates to an electricity-gas-hydrogen system distribution robust optimization method considering a hydrogen energy automobile, and belongs to the field of comprehensive energy systems. According to the method, a hydrogen energy automobile hydrogen charging-discharging demand response model based on consumer psychology is established, a utility function is introduced to quantify the consumer satisfaction degree, and the response willingness of participation in scheduling is reflected through the response capacity of hydrogen charging-discharging of a hydrogen energy automobile. A hydrogen supply chain model of hydrogen-doped gas inlet and long-tube trailer hydrogen transportation is established, and an operation mode that the hydrogen supply chain participates in system collaborative scheduling in consideration of hydrogen energy vehicle demand response is proposed. For wind power prediction error uncertainty, discrete scene information is included in a probability distance fuzzy set of wind power prediction error distribution to construct an event type distribution fuzzy set, a two-stage distribution robust optimization scheduling strategy based on the event type distribution fuzzy set is provided, and an event type affine decision rule is adopted for solving. According to the method, system operation economic benefits can be considered, and the capacity of the system for coping with wind power uncertainty is improved.

Description

Technical Field

[0001] The present invention relates to a method for distributionally robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle. Background Technique

[0002] The rapid growth of energy demand and environmental problems have prompted countries to actively develop new energy sources and promote the integrated development of multiple energy sources. It is of great significance to build an integrated energy system that couples the power grid and the natural gas grid, and to utilize the differential characteristics of multiple energies to achieve coupling and complementarity, and to realize the coordinated and economic operation of the power system and other energy systems. As a connection means between the electric and gas systems, the hydrogen energy subsystem (HES) can realize the production, conversion and storage of multiple energies. Building an electricity-gas-hydrogen integrated energy system (EGHIES) has strategic significance for supporting the consumption of a high proportion of intermittent new energy, realizing the low-cost long-distance transmission of hydrogen energy, and helping multiple energy industries to achieve deep decarbonization. Hydrogen blending in natural gas pipelines is an important supporting technology for the large-scale, long-distance and low-cost transmission of hydrogen energy. However, in existing demonstration projects, the use of hydrogen to replace natural gas in the natural gas pipeline network is still hindered by the upper limit of the hydrogen blending ratio in the pipeline, policies and regulations, and the safety of market users. The transportation of hydrogen by hydrogen trailers (HT) is formed by the coupling of the hydrogen energy system and the transportation road network, and is currently a more common and technically mature short-distance hydrogen transportation method. Existing research shows that the transportation of hydrogen by long tube trailers is conducive to the rational allocation of hydrogen energy resources, and the technology of blending hydrogen into gas can partially alleviate the problem of the excessively high construction cost of pure hydrogen pipelines at the present stage and improve the in-situ consumption rate of wind power. However, existing research only considers a single transportation method, ignoring the characteristics that the hydrogen supply chain composed of the two can coordinate and distribute hydrogen energy resources at different scales and distances, and does not incorporate the hydrogen supply chain into the EGHIES dispatching operation. It is difficult to give full play to the advantages of the flexibility of HT short-distance transfer and the realization of low-cost wind power-to-hydrogen consumption through pipeline hydrogen blending, and it is difficult to further improve the operating economy of EGHIES. Therefore, considering the combined action of hydrogen blending in natural gas pipelines and the transportation of hydrogen by long tube trailers to form a hydrogen supply chain and allowing it to participate in the collaborative dispatching operation mode of EGHIES can improve the wind power consumption rate, reduce the system load shedding amount, and effectively reduce the system operating cost.

[0003] In addition, hydrogen fuel cell vehicles (HFCVs) have the characteristics of high energy density and fast refueling speed, and have good development prospects. However, the existing research on the flexibility resource modeling of HFCVs is not yet perfect. Some studies only regard HFCVs as fixed hydrogen loads and do not consider the impact of large-scale access of HFCVs on system operation. The existing research on HFCVs participating in demand response only regards HFCVs as shared energy storage to participate in system scheduling, does not consider the difference between the two response processes of HFCV hydrogen refueling and reverse discharge, and fails to fully consider the psychological characteristics of HFCV users participating in scheduling. Regarding HFCVs only as shared energy storage to participate in scheduling cannot take into account both the flexibility resources of aggregated HFCVs and the different travel characteristics of HFCVs. When regarding HFCV users as participants in demand response for scheduling, the willingness of users to participate is strongly affected by price incentive factors. It is difficult to conform to the actual situation by ignoring the user psychology and energy consumption characteristics of HFCVs in the modeling. It can be seen that the existing research has not effectively established an accurate and effective HFCV demand response model.

[0004] In an EGHIES containing wind power generation, the uncertainty of wind power will have an important impact on the safe and economic operation of the system. The uncertainty of wind power mainly considers the fuzziness and randomness of wind power output. The research on wind power uncertainty is usually based on stochastic optimization (SO), robust optimization (RO), and distributionally robust optimization (DRO), etc. The DRO method combines the advantages of RO and SO. The existing research on the DRO method based on metric distance has been relatively extensive in the fields of multi-energy complementarity and optimal power flow. However, when the traditional Wasserstein distance DRO method uses limited and inaccurate historical data to construct a fuzzy set, it may lead to inaccurate distribution fuzzy sets and is difficult to ensure the safety and optimality of the scheduling results. However, the existing DRO research based on probability distance does not incorporate discrete scenario information into the construction of the fuzzy set, ignoring the discrete scenario information and making it difficult to accurately predict the wind power prediction error, resulting in the decision-making result being possibly too conservative. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for distributionally robust optimization of the electric-gas-hydrogen system considering hydrogen energy vehicles, establish a charging-discharging demand response model of HFCVs based on consumer psychology, and propose an operation mode for the hydrogen supply chain participating in the collaborative scheduling of EGHIES considering the demand response of HFCVs according to the characteristics of different transportation modes in the hydrogen supply chain. Finally, based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, an event-based distribution fuzzy set based on the Wasserstein distance is constructed, and an event-based affine decision rule is used for solution. This method can take into account the economic benefits of system operation and improve the system's ability to cope with the uncertainty of wind power.

[0006] To achieve the above object, the technical solution of the present invention is: A method for distributionally robust optimization of the electric-gas-hydrogen system considering hydrogen energy vehicles, including:

[0007] Establish a charging-discharging demand response model of hydrogen energy vehicles based on consumer psychology: Model the demand response of hydrogen fuel cell vehicles (HFCVs) based on price-based demand response. The psychological factors of HFCV users participating in the response are represented by relevant models of incentive signals and user response degrees. Establish a psychological characteristic model of HFCV owners, and design dynamic incentive prices to guide HFCV users to participate in scheduling, realizing peak shaving and valley filling on the load side;

[0008] Establish a hydrogen supply chain model for hydrogen injection into gas and hydrogen transportation by hydrogen tube trailers (HTs): According to the characteristics of different transportation modes in the hydrogen supply chain, while establishing a hydrogen injection into gas model in which hydrogen and natural gas are mixed into hydrogen-enriched natural gas at the hydrogen mixing node and then injected into the natural gas pipeline, through the scheduling of HTs, considering the time required for HTs to transport between different hydrogen energy subsystems (HESs), establish a transportation hydrogen model based on a time-space network, forming a hydrogen supply chain model for coordinating and distributing hydrogen energy at different scales and distances;

[0009] Based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, construct an event-based distribution fuzzy set based on the Wasserstein distance, propose a two-stage distributionally robust optimization scheduling strategy based on the event-based distribution fuzzy set, and use an event-based affine decision rule and strong duality theory for solution.

[0010] Further, the charging-discharging demand response model of hydrogen energy vehicles based on consumer psychology includes two parts: the travel characteristics of HFCVs and the response capacity model of hydrogen energy vehicles. Among them, the travel characteristics of HFCVs are as follows:

[0011]

[0012] In the formula: is the arrival time period of HFCVs; f D(·) is the probability density function of the arrival time of HFCV; μ i and σ i 2 respectively represent the mean and variance of the normal distribution of the arrival time of HFCV in the i-th area;

[0013] The probability density of the initial capacity characteristics after HFCV accesses the system is as follows:

[0014]

[0015] In the formula: is the arrival time of HFCV; f SOC (·) is a negatively skewed distribution, is the probability density function of the standard normal distribution, φ(·) is the cumulative probability distribution function of the standard normal distribution; α, β, and γ are negatively skewed distribution parameters;

[0016] The hydrogen energy vehicle response capacity model based on consumer psychology is obtained by analyzing the HFCV demand response model considering consumer psychology, taking the hydrogen filling price and discharge subsidy price as the incentive prices to guide users in the hydrogen filling stage and discharge stage, and designing dynamic incentive prices to guide HFCV users to participate in scheduling;

[0017] The non-linear incentive and response capacity model of the response area in the hydrogen filling-discharging process of HFCV is as follows:

[0018]

[0019] In the formula: C Cha (H price ) is the hydrogen filling response function of HFCV; C Dis (H price ) is the discharge response function of HFCV; κ is the user characteristic parameter representing the user's willingness, which is divided into active user type (i.e., κ < 0), normal user type (i.e., κ = 0), and inactive user type (i.e., κ > 0); C re_Cha is the maximum hydrogen filling response capacity of HFCV; C re_Dis is the maximum discharge response capacity; H price is the hydrogen filling price or discharge subsidy price; H shift is the corresponding change characteristic threshold; H shift_min is the minimum hydrogen filling price threshold; H shift_max is the maximum discharge price threshold; is the upper limit of the SOC of the hydrogen storage tank of HFCV; is the lower limit of the SOC of the hydrogen storage tank of HFCV to maintain travel demand; is the capacity of the hydrogen storage tank of HFCV;

[0020] Since the active user type is more suitable for ordinary users, the active response curve is selected to describe the hydrogen charging and discharging processes of HFCV, that is, κ < 0. Based on this, the designed dynamic incentive price scheme is as follows:

[0021]

[0022] The actual hydrogen charging power and discharging power of the response are as follows:

[0023]

[0024] In the formula: is the response capacity of the kth HFCV to charge the HES during the response period; is the response capacity of the kth HFCV to discharge the HES during the response period; is the actual hydrogen charging amount provided by the kth HFCV in the system; is the actual discharging amount of the kth HFCV to the power grid; and respectively represent the hydrogen charging efficiency and discharging efficiency of HFCV.

[0025] Furthermore, the calorific value calculation formula of the hydrogen - blended natural gas after mixing natural gas and hydrogen is as follows:

[0026] T mix = η in T H +(1 - η in )T gas

[0027] In the formula: η in is the hydrogen blending ratio; T H and T gas are the lower calorific values of hydrogen and natural gas respectively; T mix is the lower calorific value of the hydrogen - blended natural gas;

[0028] Among them, the hydrogen blending ratio constraint is as follows:

[0029]

[0030] In the formula: and are the hydrogen volume and natural gas volume injected into the natural gas pipeline at the same time in the t period respectively.

[0031] Furthermore, the constraints of the hydrogen transportation model for transportation based on the time - space network TSN are as follows:

[0032] 1) The time expression of HT transportation between different HESs is as follows:

[0033]

[0034] In the formula: and are the transportation time, the shortest transportation distance, and the average transportation speed of the k-th HT from the m-th HES to the n-th HES, respectively; represents the ceiling function; R and K are the sets of HESs and HTs, respectively; Ω p is the set of transportation paths between different HESs;

[0035] 2) TSN model constraints:

[0036]

[0037]

[0038] In the formula: is the position status variable, indicating whether the k-th vehicle is at the position of the m-th HES at time t. If the HT is at the m-th HES, then otherwise is the transportation status variable. When it is 1, it means that the k-th HT is transporting on the path m→n at time t; and represent the arrival and departure status variables, respectively. When they are 1, it means that the k-th HT arrives at or departs from the m-th HES at time t;

[0039]

[0040] In the formula: and are auxiliary boolean variables; is a boolean variable. When it is 1, it means that the k-th HT transfers from the m-th HES along the transportation path m→n at time t;

[0041]

[0042] In the formula: means that the k-th HT is located at the r-th HES before the start of the scheduling, means that the k-th HT must return to any one of the HESs at the end of the scheduling;

[0043] 3) TSN and HES coupling constraints: When the HT leaves the parking arc and enters the transfer arc, it means that the HT can transport hydrogen out of the HES at the starting point of the transfer arc and transport hydrogen into the HES at the ending point of the transfer arc; Constraints on the single transportation hydrogen volume of the HT are as follows:

[0044]

[0045] In the formula: is the hydrogen volume transported by the k-th HT from the m-th HES to the n-th HES at time t; and are the minimum and maximum values of the hydrogen quantity transported by HT per single transport respectively;

[0046] Ignoring the hydrogen quantity loss during the HT transportation, the hydrogen gas transported out of the m-th HES at time t through HT satisfies the following constraints:

[0047]

[0048] In the formula: K is the set of HT; is the hydrogen quantity transported out of the n-th HES at time, and transported into the m-th HES at time t; is the path set starting from m; is the path set ending at m.

[0049] Furthermore, based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, an event-based distribution fuzzy set based on the Wasserstein distance is constructed. The definition of the expected form of the Wasserstein distance is as follows:

[0050]

[0051] In the formula: represents the minimum expected value of the objective function under the joint distribution Q; is the empirical distribution and the true distribution the entire set of the joint distribution Q; ρ(·) is the basic distance of the Wasserstein distance; is the vector composed of the t-period random variables of the prediction error of the empirical distribution is the vector composed of the t-period random variables of the true distribution of the prediction error;

[0052] Based on the mixed Gaussian distribution, the historical prediction error of wind power is generated, and the k-means clustering method is used to obtain the typical discrete scenarios and scenario probabilities of the historical prediction error data of wind power; the event-based distribution fuzzy set of the wind power prediction error is constructed as follows:

[0053]

[0054] In the formula: is the feature space after dimensionality reduction; I e +1 is the dimension of the feature space; is the initial probability distribution; is the expected value function under the distribution ; S is the number of typical scenarios; is an auxiliary variable; is a scenario variable; p s is the probability of scenario s; is the support set of the fuzzy set under scenario s; and are the lower bound and upper bound of the wind power prediction error respectively; is the empirical distribution of the wind power prediction error under scenario s.

[0055] Furthermore, the method proposes an objective function for the day-ahead and intra-day two-stage distributionally robust optimal scheduling considering wind power uncertainty as follows:

[0056] 1) The objective function of the day-ahead scheduling cost in the two-stage distributionally robust:

[0057] c T x = C1 + C2 + C3 + C4

[0058] In the formula: c T x is the day-ahead scheduling cost; x is the day-ahead scheduling decision variable; c is the column vector of the corresponding cost coefficients for the day-ahead; C1 is the energy purchase cost; C2 is the hydrogen system scheduling cost; C3 is the day-ahead wind curtailment cost; C4 is the day-ahead load shedding cost;

[0059] 2) The energy purchase cost includes the power generation and gas purchase costs, and the expression is as follows:

[0060]

[0061] In the formula: N pg and N gas represent the numbers of thermal power units and natural gas sources in the system respectively; T is the total number of scheduling time periods; c PG is the power generation cost per unit power; c G,buy is the natural gas price for purchasing gas from the gas source; is the power generation power of the thermal power unit; is the natural gas power purchased from the gas source;

[0062] 3) The hydrogen system scheduling cost expression is as follows:

[0063] C2 = C HT + C HFCV

[0064] In the formula: C HFCV represents the scheduling cost of HFCV; C HT represents the HT transportation cost. The transportation cost includes two parts: fixed cost and variable cost. The variable cost is related to the transportation distance of HT and the amount of hydrogen transported. The HT transportation cost expression is as follows:

[0065]

[0066] In the formula: c HT,a is the fixed cost coefficient of HT; c HT,b is the variable cost coefficient of HT;

[0067] Among them, the scheduling cost expression of HFCV is as follows:

[0068]

[0069] In the formula: π HFCV is the satisfaction cost of HFCV; π HFCV,cha is the comprehensive energy system EGHIES revenue of electricity-gas-hydrogen corresponding to hydrogen filling of HFCV; π HFCV,dis represents the compensation cost paid by EGHIES to HFCV during the discharging of HFCV; α1 and α2 are the aversion coefficients when users deviate from the actual demand; is the maximum hydrogen filling response capacity of the kth HFCV at time t; represents the incentive price at time t within the ith HES; is the response capacity of the ith HFCV to hydrogen filling the HES during the response period at time t; is the response capacity of the ith HFCV to discharging the HES during the response period at time t;

[0070] 4) Day-ahead wind curtailment cost

[0071]

[0072] In the formula: N pw is the number of wind farms; c wind,cut is the wind curtailment penalty coefficient; is the wind curtailment power;

[0073] 5) Day-ahead load shedding cost

[0074]

[0075] In the formula: N HES is the number of HESs; c Hload,cut is the hydrogen load shedding penalty coefficient; is the hydrogen load shedding amount of the ith HES;

[0076] 6) Intra-day adjustment cost

[0077] d T y = D1 + D2 + D3

[0078] In the formula: d T y is based on the day-ahead scheduling decision variable x to cope with the wind power prediction error The intraday adjustment cost generated; y is the intraday decision variable; d is the column vector of intraday cost coefficients; D1 is the intraday purchase energy adjustment cost; D2 is the intraday curtailment adjustment cost; D3 is the intraday load shedding adjustment cost; the cost expressions of each part are as follows:

[0079]

[0080] In the formula: is the adjustment power of the thermal power unit; is the intraday gas purchase adjustment power; is the curtailment adjustment power; is the hydrogen load shedding adjustment power.

[0081] Furthermore, the constraints in the day-ahead scheduling of the two-stage distributionally robust optimization are:

[0082] 1) HES power balance

[0083]

[0084] In the formula: is the day-ahead predicted wind power output in the HES; is the curtailment power in the i-th HES; is the consumed electric power for power-to-gas to produce natural gas; is the consumed electric power for the electrolyzer to produce hydrogen; is the power injected into the power grid by the HES through the tie line; is the hydrogen production volume of the EL; is the hydrogen consumption volume of the hydrogen blending device; is the net hydrogen output volume of the HES through the HT; is the hydrogen load demand; is the actual hydrogen charging volume of the hydrogen energy storage; The actual hydrogen discharging volume of the hydrogen energy storage;

[0085] 2) Curtailment and load shedding constraints

[0086]

[0087] 3) HES tie line power constraints

[0088]

[0089] In the formula: is the interactive power between the HES and the power grid; is the lower limit of the power interaction between the HES and the power grid; P i con,max is the upper limit of the power interaction between the HES and the power grid; is the hydrogen-blended natural gas injected by the HES into the natural gas pipeline; is the lower limit of the interaction between HES and gas network flow; is the upper limit of the interaction between HES and gas network flow; T H is the lower calorific value of hydrogen; T gas is the lower calorific value of natural gas; T mix is the lower calorific value of hydrogen-blended natural gas; is the gas production power of power-to-gas;

[0090] 4) Power and gas network power flow constraints

[0091] The power network is modeled using the distflow power flow equation and relaxed using the second-order cone method. The natural gas network is modeled using the steady-state Weymouth power flow equation.

[0092] Furthermore, the intraday optimal scheduling in the two-stage distributionally robust approach makes real-time adjustments based on the day-ahead scheduling plan in the two-stage distributionally robust approach. The adjusted output of each device in the system during the day also satisfies the corresponding operation constraints and the system power balance constraint.

[0093] Furthermore, the expression of the overall scheduling decision for the day-ahead and intraday two-stage distributionally robust approach is as follows:

[0094]

[0095] Since the above objective function contains the sup upper bound function, it is actually a three-layer optimization problem of min-max-min. It is reconstructed and converted into a mixed-integer linear programming model for solution by using the scenario-based affine decision rule and the strong duality theory.

[0096] The present invention also provides a computer-readable storage medium, on which computer program instructions capable of being run by a processor are stored. When the processor runs the computer program instructions, the method steps as described in any of the above can be implemented.

[0097] Compared with the prior art, the present invention has the following beneficial effects:

[0098] The present invention models the HFCV as a price-based demand response, uses the hydrogen filling price and the discharge subsidy price as incentive signals, comprehensively considers the energy consumption comfort and energy consumption cost of HFCV users, enables the hydrogen filling and discharging behavior of the HFCV to be guided by price incentives, improves the orderly hydrogen filling and discharging of the HFCV, flattens the peak-valley curve of the system load power, and reduces the operating cost of the system. Secondly, the present invention establishes a hydrogen supply chain model that couples hydrogen injection into the gas and HT transportation, and proposes an operation mode in which the hydrogen supply chain considering the HFCV demand response participates in the collaborative scheduling of the EGHIES. It improves the wind power consumption rate, reduces the system load rejection amount, and effectively reduces the system operating cost. Finally, in order to improve the decision-making accuracy of the DRO model, based on the scenario tree theory, the discrete scenario information and the continuous probability distribution distance metric information are considered simultaneously, an event-based distribution fuzzy set based on the Wasserstein distance is constructed, a two-stage distributionally robust optimization scheduling strategy based on the event-based distribution fuzzy set is proposed, and the event-based affine decision rule is used for solution, which enhances the predictability of the intraday wind power prediction error, realizes a more accurate characterization of the fuzzy set, and effectively takes into account the economy and robustness of the system operation. Description of the Drawings

[0099] Figure 1 It is the internal structure diagram of the HES in the present invention;

[0100] Figure 2 It is the HFCV demand response model considering consumer psychology in the present invention;

[0101] Figure 3 It is the schematic diagram of the hydrogen filling response area and the discharge response area of the HFCV in the present invention;

[0102] Figure 4 It is the time-space transfer schematic diagram of the HT transportation model in the present invention;

[0103] Figure 5 It is the event-based distribution fuzzy set based on the Wasserstein distance in the present invention;

[0104] Figure 6 It is the network structure topology diagram in the following embodiments of the present invention;

[0105] Figure 7 It is the power operation plan in the HES optimized by Scheme 4 in the following embodiments of the present invention;

[0106] Figure 8 It is the gas / hydrogen operation plan in the HES optimized by Scheme 4 in the following embodiments of the present invention. Detailed Embodiments

[0107] The technical solutions of the present invention will be specifically described below with reference to the drawings.

[0108] The present invention provides a method for distributionally robust optimization of an electric-gas-hydrogen system considering hydrogen energy vehicles, including:

[0109] Establish a hydrogen refueling-discharging demand response model for hydrogen energy vehicles based on consumer psychology: Based on price-based demand response, model the demand response of hydrogen fuel cell vehicles (HFCVs). The psychological factors of HFCV users participating in the response are represented by relevant models of incentive signals and user response degrees. Establish a psychological characteristic model of HFCV owners, and design dynamic incentive prices to guide HFCV users to participate in scheduling, so as to achieve peak shaving and valley filling on the load side;

[0110] Establish a hydrogen supply chain model for hydrogen blending into gas and hydrogen transportation by hydrogen tube trailers (HTs): Aiming at the characteristics of different transportation methods in the hydrogen supply chain, while establishing a model for blending hydrogen and natural gas into hydrogen-blended natural gas at the hydrogen blending node and then injecting it into the natural gas pipeline, through the scheduling of HTs, considering the time required for HTs to transport between different hydrogen energy subsystems (HESs), establish a transportation hydrogen model based on a time-space network, and form a hydrogen supply chain model for coordinating the distribution of hydrogen energy at different scales and distances;

[0111] Based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, construct an event-based distribution fuzzy set based on the Wasserstein distance, propose a two-stage distributionally robust optimization scheduling strategy based on the event-based distribution fuzzy set, and use the event-based affine decision rule for solution.

[0112] The following are specific implementation examples of the present invention.

[0113] Select the EGHIES composed of the IEEE 30-node power network and the 20-node gas network in Belgium for simulation analysis. Nodes 7, 11, and 22 of the power network are respectively coupled with nodes 2, 9, and 17 of the gas network through HESs, and they are numbered as HES1, HES2, and HES3 in sequence. Its network topology is as Figure 6 shown. The installed capacity of the wind turbine is 60 MW, and the predicted output of each wind power is as Figure 7 shown. 1500 HFCVs are connected to each HES. The access time period and initial capacity of HFCVs are simulated using Monte Carlo sampling. The scheme design considering different states is shown in Table 1:

[0114] Table 1 Scheme settings for different states

[0115]

[0116] Among them, the scenario considering HTs includes 3 HTs, and the installed capacity of the thermal power unit is shown in Table 2.

[0117] Table 2 Parameters of Thermal Power Unit

[0118]

[0119] Scenario 1 is the basic scenario, and the remaining scenarios are compared with Scenario 1 for research.

[0120] For this embodiment, a distributionally robust optimization method of an electric-gas-hydrogen system considering hydrogen energy vehicles of the present invention is applied for system scheduling, and the optimized scheduling result of Scenario 4 is obtained as Figure 7 and Figure 8 . Table 3 shows the operating costs of each system under different scenario optimizations. The net revenue of HFCV in Table 3 is expressed as the discharge revenue of HFCV minus the hydrogen charging cost of HFCV.

[0121] Table 3 Cost Comparison under Different Scenarios

[0122]

[0123] From the results in Table 3, it can be seen that Scenario 2 considering hydrogen blending into gas and hydrogen transportation for HT in transportation can reduce the total system cost by 22.48% and the curtailment cost by 64.54% compared with Scenario 1. And based on Scenario 1, Scenario 3 introduces the hydrogen charging-discharging demand response of HFCV on the basis of considering the hydrogen supply chain, reducing the operating cost by 32.12%, the curtailment cost by 79.40%, and the hydrogen curtailment load cost by 81.69%.

[0124] Comparing Scenario 3 and Scenario 2, the total system cost of Scenario 3 is reduced by 32.12%, the energy purchase cost is reduced by 10.28%, and the curtailment cost is reduced by 81.69%. It shows that HFCV based on consumer psychology can participate in system scheduling, and the load curve of HFCV can be reasonably guided through incentive prices, thereby reducing the total operating cost of the system and effectively reducing the system curtailment volume.

[0125] Finally, different typical scenario numbers are set to 5, 10, and 15 respectively, and the influence of different typical scenario numbers on the scheduling results and calculation time is quantitatively analyzed. When the number of typical scenarios is 1, the event-based distribution fuzzy set degenerates into a traditional fuzzy set based on the Wasserstein distance. Table 4 shows the relationship between the number of typical scenarios and the scheduling cost under different sample numbers.

[0126] Table 4 Scheduling Cost Comparison under Different Sample Numbers and Numbers of Typical Scenarios

[0127]

[0128] As can be seen from Table 4, as the number of samples increases, the fuzzy set can contain more historical wind power data, can more accurately characterize the true distribution of wind power output, the fuzzy set range of the event-based DRO model shrinks, and the total operating cost in Table 4 decreases with the increase in the number of samples. When the number of samples is the same, when the number of selected typical scenarios is small, the difference between its sample distribution and the true distribution is large, it is difficult to accurately characterize the true distribution of wind power output, with large uncertainty, the fuzzy set range of the DRO model is large, resulting in higher costs for the cases with fewer scenario numbers at the same number of samples. As the number of typical scenarios increases, the introduction of typical scenarios corrects the sample distribution of the fuzzy set by introducing scenario probabilities during the model solution process, shrinks the fuzzy set range, and reduces the conservatism of decision-making. At the same number of samples, the cases with more typical scenarios correspond to lower costs. When the number of samples is 400, the operating cost of the case considering 15 typical scenarios is reduced by 4.31% compared with the traditional distributionally robust approach.

[0129] In summary, compared with the solution that does not consider the hydrogen supply chain and HFCV, the method of the present invention effectively improves the system operation economy. The discrete typical scenario information in the event-based fuzzy set corrects the sample distribution of the fuzzy set, thereby reducing the conservatism of decision-making. When the number of samples is 400, the operating cost is reduced by 4.31% compared with the traditional distributionally robust approach. Decision-makers can also adjust the number of typical scenarios and uncertainty parameters based on risk preferences, taking into account both the economy and robustness of system operation.

[0130] The present invention also provides a computer-readable storage medium, on which computer program instructions capable of being run by a processor are stored. When the processor runs the computer program instructions, the method steps as described in any one of the above can be implemented.

[0131] The above are the preferred embodiments of the present invention. All changes made according to the technical solution of the present invention, when the functions and effects produced do not exceed the scope of the technical solution of the present invention, fall within the protection scope of the present invention.

Claims

1. A method for distributionally robust optimization of an electric-gas-hydrogen system considering hydrogen energy vehicles, characterized in that, Including: Establish a hydrogen refueling-discharging demand response model for hydrogen energy vehicles based on consumer psychology: Model the demand response of hydrogen fuel cell vehicles (HFCVs) based on price-based demand response. The psychological factors of HFCV users participating in the response are represented by relevant models of incentive signals and user response levels. Establish a psychological characteristics model for HFCV owners and design dynamic incentive prices to guide HFCV users to participate in dispatching, achieving peak shaving and valley filling on the load side. Establish a hydrogen supply chain model for hydrogen injection into natural gas and hydrogen transportation by hydrogen tube trailers (HTs): In view of the characteristics of different transportation methods in the hydrogen supply chain, while establishing a hydrogen injection into natural gas model in which hydrogen and natural gas are mixed into hydrogen-enriched natural gas at the hydrogen mixing node and then injected into the natural gas pipeline, through the dispatching of HTs, considering the time required for HTs to transport between different hydrogen energy subsystems (HESs), establish a transportation hydrogen model based on the time-space network, forming a hydrogen supply chain model for coordinating the allocation of hydrogen energy at different scales and distances. Based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, construct an event-based distribution fuzzy set based on the Wasserstein distance, propose a two-stage distributionally robust optimization scheduling strategy based on the event-based distribution fuzzy set, and use methods including event-based affine decision rules and strong duality theory for solution.

2. A method for distribution robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that The hydrogen refueling-discharging demand response model for hydrogen energy vehicles based on consumer psychology includes two parts: the travel characteristics of HFCVs and the response capacity model of hydrogen energy vehicles. Among them, the travel characteristics of HFCVs are as follows: In the formula: is the arrival time period of HFCV; f D (·) is the probability density function of the arrival time period of HFCV; μ i and respectively represent the mean and variance of the normal distribution of the arrival time period of HFCV in the i-th area; The probability density of the initial capacity characteristics after HFCVs are connected to the system is as follows: In the formula: is the arrival time period of HFCV; f SOC (·) is a negatively skewed distribution, is the probability density function of the standard normal distribution, and φ(·) is the cumulative probability distribution function of the standard normal distribution; α, β, and γ are negatively skewed distribution parameters; The response capacity model of hydrogen energy vehicles based on consumer psychology is obtained by analyzing the HFCV demand response model considering consumer psychology, taking the hydrogen refueling price and the discharge subsidy price as the incentive prices to guide users in the hydrogen refueling stage and the discharge stage, and designing dynamic incentive prices to guide HFCV users to participate in dispatching. The non-linear incentive and response capacity model of the response area in the hydrogen refueling-discharging process of HFCVs is as follows: Where: C Cha (H price ) is the hydrogen charging response function of the HFCV; C Dis (H price ) is the discharge response function of the HFCV; κ is the user characteristic parameter representing the user's willingness, which is divided into an active user type where κ < 0, a normal user type where κ = 0, and an inactive user type where κ > 0; C re_Cha is the maximum hydrogen charging response capacity of the HFCV; C re_Dis is the maximum discharge response capacity; H price is the hydrogen charging price or discharge subsidy price; H shift is the corresponding change characteristic threshold; H shift_min is the minimum hydrogen charging price threshold; H shift_max is the maximum discharge price threshold; is the upper limit of the SOC of the hydrogen storage tank of the HFCV; is the lower limit of the SOC of the hydrogen storage tank of the HFCV to maintain travel demand; is the capacity of the hydrogen storage tank of the HFCV; Since the active user type is more suitable for ordinary users, select the active response curve to describe the two processes of HFCV hydrogen refueling and discharging, that is, κ < 0. Based on this, the designed dynamic incentive price scheme is as follows: The actual hydrogen refueling power and discharge power of the response are as follows: In the formula: is the response capacity of the k-th HFCV to hydrogen filling of the HES during the response period; is the response capacity of the k-th HFCV to discharging of the HES during the response period; is the actual hydrogen filling amount provided by the k-th HFCV of the system; is the actual discharging amount of the k-th HFCV to the power grid; and respectively represent the hydrogen filling efficiency and discharging efficiency of the HFCV.

3. A method for distributionally robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that The calorific value calculation formula of the hydrogen-enriched natural gas after mixing natural gas and hydrogen is as follows: T mix = η in T H +(1 - η in )T gas Where: η in is the hydrogen doping ratio; T H and T gas are the lower heating values of hydrogen and natural gas respectively; T mix is the lower heating value of the hydrogen-blended natural gas; Among them, the hydrogen blending ratio constraint is as follows: Wherein: and are the hydrogen gas volume and natural gas volume injected into the natural pipeline simultaneously in the t period, respectively.

4. A method for distributionally robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that The constraints of the transportation hydrogen model based on the time-space network (TSN) are as follows: 1) The time expression for HTs to transport between different HESs is as follows: Wherein: d m→n and are respectively the transportation time, the shortest transportation distance, and the average transportation speed of the k-th HT from the m-th HES to the n-th HES; represents the ceiling function; R and K are respectively the sets of HES and HT; Ω p is the set of transportation paths between different HES; 2) TSN model constraints: Wherein: is a position status variable, indicating whether the k-th vehicle is located at the position of the m-th HES in the t time period. If the HT is at the m-th HES, then otherwise is a transportation status variable. When it is 1, it means that the k-th HT is transporting on the path m→n in the t time period; and respectively represent the arrival and departure status variables. When it is 1, it means that the k-th HT arrives at or departs from the m-th HES at time t; Wherein: and are auxiliary Boolean variables; is a Boolean variable, and when it is 1, it means that the k-th HT transfers from the m-th HES along the transportation path m→n at time t; where: indicates that the k-th HT is located at the r-th HES before the start of scheduling; indicates that the k-th HT must return to any one of the HESs at the end of scheduling; 3) TSN and HES coupling constraints: When the HT leaves the parking arc and enters the transfer arc, it means that the HT can transport hydrogen out of the HES at the starting point of the transfer arc and transport hydrogen into the HES at the ending point of the transfer arc; constrain the single hydrogen transportation volume of the HT: Wherein: is the amount of hydrogen transported from the m-th HES to the n-th HES by the k-th HT during the t period; and are the minimum and maximum values of the hydrogen amount transported by the HT each time, respectively; Ignoring the hydrogen loss during the HT transportation process, the hydrogen transported out net by the m-th HES at time t through HT satisfies the following constraints: Where: K is the set of HT; is the amount of hydrogen transported out of the nth HES at time and transported into the mth HES at time t; is the set of paths starting from m; is the set of paths ending at m.

5. A method for distribution robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that, Based on the scenario tree theory, considering both discrete scenario information and continuous probability distribution distance metric information, construct an event-based distribution fuzzy set based on the Wasserstein distance. The definition of the expected form of the Wasserstein distance is as follows: In the formula: represents the minimum expected value of the objective function under the solution of the joint distribution Q; is the empirical distribution and the true distribution of the entire set of the joint distribution Q; ρ(·) is the basic distance of the Wasserstein distance; is the vector composed of the t-period random variables of the prediction error of the empirical distribution, is the vector composed of the t-period random variables of the true distribution of the prediction error; Generate the historical prediction error of wind power based on the mixture Gaussian distribution, and use the k-means clustering method to obtain the typical discrete scenarios and scenario probabilities of the historical prediction error data of wind power; construct the event-based distribution fuzzy set of the wind power prediction error as follows: In the formula: is the feature space after dimensionality reduction; I e +1 is the dimension of the feature space; is the initial probability distribution; is the distribution distribution is the expected value function under; S is the number of typical scenarios; is an auxiliary variable; is a scenario variable; p s is the probability of scenario s; is the support set of the fuzzy set under scenario s; and are respectively the lower bound and the upper bound of the wind power prediction error; is the empirical distribution of the wind power prediction error under scenario s.

6. A method for distribution robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that, For the method described above, propose an objective function for the day-ahead and intra-day two-stage distributionally robust optimal scheduling considering the uncertainty of wind power as follows: 1) The objective function of the day-ahead scheduling cost in the two-stage distributionally robust: c T x = C1 + C2 + C3 + C4 where: c T x is the day-ahead scheduling cost; x is the day-ahead scheduling decision variable; c is the cost coefficient column vector corresponding to the day-ahead; C1 is the energy purchase cost; C2 is the hydrogen system scheduling cost; C3 is the day-ahead wind curtailment cost; C4 is the day-ahead load shedding cost; 2) The energy purchase cost includes the power generation and gas purchase costs, and the expression is as follows: Where: N pg and N gas respectively represent the numbers of thermal power units and natural gas sources in the system; T is the total number of dispatching periods; c PG is the power generation cost per unit of power; c G,buy is the natural gas price for purchasing natural gas from the gas source; is the power generation power of the thermal power unit; is the natural gas power purchased from the gas source; 3) The expression of the hydrogen system scheduling cost is as follows: C2 = C HT + C HFCV Where: C HFCV represents the scheduling cost of HFCV; C HT represents the HT transportation cost. The transportation cost consists of two parts: fixed cost and variable cost. The variable cost is related to the transportation distance of HT and the amount of hydrogen transported. The expression of the HT transportation cost is as follows: where: c HT,a is the fixed cost coefficient of HT; c HT,b is the variable cost coefficient of HT; Among them, the expression of the scheduling cost of HFCV is as follows: Where: π HFCV is the satisfaction cost of HFCV; π HFCV,cha is the revenue of the electric-gas-hydrogen integrated energy system EGHIES corresponding to hydrogen filling of HFCV; π HFCV,dis represents the compensation cost paid by EGHIES to HFCV during the discharge of HFCV; α1 and α2 are the aversion coefficients when users deviate from the actual demand; is the maximum hydrogen filling response capacity of the kth HFCV at time t; represents the incentive price at time t within the ith HES; is the response capacity of the ith HFCV to hydrogen filling the HES during the response period at time t; is the response capacity of the ith HFCV to discharging the HES during the response period at time t; 4) The day-ahead wind curtailment cost Where: N pw is the number of wind farms; c wind,cut is the curtailment penalty coefficient; is the curtailment power; 5) The day-ahead load shedding cost Where: N HES is the number of HESs; c Hload,cut is the penalty coefficient for hydrogen cut-off load; is the hydrogen cut-off load of the i-th HES; 6) The intra-day adjustment cost d T y = D1 + D2 + D3 where: d T y is the intraday adjustment cost generated to cope with the wind power prediction error based on the day-ahead scheduling decision variable x ; y is the intraday decision variable; d is the column vector of the corresponding cost coefficients for the intraday; D1 is the intraday adjustment cost for energy purchase; D2 is the intraday adjustment cost for wind curtailment; D3 is the intraday adjustment cost for load shedding; the cost expressions for each part are as follows: Wherein: is the adjusted power of the thermal power unit; is the adjusted power of gas purchase within the day; is the adjusted power of abandoned wind; is the adjusted power of hydrogen load shedding.

7. A method for distribution robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that, The constraints in the day-ahead scheduling in the two-stage distributionally robust are: 1) HES power balance Wherein: is the predicted wind power output of the HES for the day-ahead; is the wind curtailment power of the i-th HES; is the consumed electric power for power-to-gas to produce natural gas; is the consumed electric power for the electrolyzer to produce hydrogen; is the power injected into the power grid by the HES through the tie line; is the hydrogen production amount of the EL; is the hydrogen consumption amount of the hydrogen blending into gas device; is the net hydrogen output of the HES through the HT; is the hydrogen load demand; is the actual hydrogen charging volume of the hydrogen energy storage; Actual hydrogen discharging volume of the hydrogen energy storage 2) Wind curtailment and load shedding constraints 3) HES tie-line power constraints In the formula: is the interaction power between the HES and the power grid; P i con,min is the lower limit of the power interaction between the HES and the power grid; P i con,max is the upper limit of the power interaction between the HES and the power grid; The HES is the hydrogen - mixed natural gas injected into the natural gas pipeline; The [lower limit] is the lower limit of the interaction between the HES and the gas network flow; The [upper limit] is the upper limit of the interaction between the HES and the gas network flow; T H is the lower heating value of hydrogen; T gas is the lower heating value of natural gas; T mix is the lower heating value of hydrogen-blended natural gas; is the gas production power of power-to-gas; 4) Power and gas network power flow constraints The power network is modeled by the distflow power flow equation and relaxed by the second-order cone method, and the natural gas network is modeled by the steady-state Weymouth power flow equation.

8. A method for distributionally robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that, The intra-day optimal scheduling in the two-stage distributionally robust makes real-time adjustments based on the day-ahead scheduling plan in the two-stage distributionally robust. The adjusted output of each device in the system during the intra-day also satisfies the corresponding operation constraints and the system power balance constraints.

9. A method for distributionally robust optimization of an electric-gas-hydrogen system considering a hydrogen energy vehicle, characterized in that, The expression of the overall scheduling decision of the day-ahead and intra-day two-stage distributionally robust is as follows: Since the above objective function contains the sup upper bound function, it is actually a three-layer optimization problem with min-max-min. It is reconstructed and transformed into a mixed-integer linear programming model for solution by using the event-based affine decision rule and the strong duality theory.

10. A computer-readable storage medium, on which computer program instructions that can be run by a processor are stored. When the processor runs the computer program instructions, the method steps described in any one of claims 1-9 can be implemented.

Images