CN116541997A - Hydrogen-containing hybrid energy supply infrastructure planning method - Google Patents

Abstract

The invention discloses a planning method of hydrogen-containing hybrid energy supply infrastructure, which relates to the field of hydrogen energy, and comprises the following steps: step 1, modeling three networks of a power grid, a hydrogen network and a traffic network in a coupling way: constructing a coupling network physical scene, establishing an optimization target, constructing a traffic network constraint, constructing a hydrogen network constraint, constructing a power grid constraint and constructing a coupling constraint; step 2, constructing a decision-dependent distribution robust optimization model: hydrogen demand decision-making relies on fuzzy set construction, and a distributed robust optimization framework construction; step 3, reconstructing and solving a distributed robust optimization problem: decision-dependent scene probability construction, linearization of norm items and opportunistic constraint reconstruction; and 4, performing example simulation analysis on the three-network coupling system. The invention effectively solves the problem that the hydrogen fuel automobile is not popular, and the chicken or egg is in advance between the hydrogen automobile and the hydrogenation station, realizes reasonable planning of hydrogen infrastructure, reduces investment and operation cost, improves new energy consumption of a power grid, and effectively avoids energy waste in the early development stage of the hydrogen automobile.

Description

Hydrogen-containing hybrid energy supply infrastructure planning method

Technical Field

The invention relates to the field of hydrogen energy, in particular to a planning method for a hydrogen-containing hybrid energy supply infrastructure.

Background

In recent years, hydrogen has received increasing attention as a clean zero carbon fuel. Hydrogen fuelled automobiles are considered new vehicles that replace traditional fuelled automobiles in the future. In the aspects of driving mileage and refueling time, the hydrogen fuel automobile is obviously superior to an electric automobile, has the characteristics of high energy density, zero emission, quick refueling and the like, and has wide application prospect. However, hydrogen-fueled automobiles have not yet been fully popularized due to the immature hydrogen-fueled automobile technology and the low hydrogen-fueling station occupancy. How to accelerate the popularization of hydrogen-fuelled automobiles without hydrogen infrastructure and how to attract investment in hydrogen stations without hydrogen-fuelled automobiles on roads is known as the "pre-chicken or pre-egg" problem of cars and stations. However, the hydrogen fuelled automotive market must have at least some hydrogen stations put into operation before the car is sold to the public. The profitability of the hydro-station is largely dependent on the utilization rate, which is unavoidable in the early stages of market development. How best to coordinate the development of infrastructure and automotive sales is critical to the successful transition from a traditional energy vehicle to a hydrogen-fuelled vehicle. Then, the hydrogen filling infrastructure is planned and built before the hydrogen fuel automobile is popularized, and the hydrogen filling station is reasonably planned while the social and economic benefits are ensured when the exact data information of the hydrogen filling requirement of the future hydrogen fuel automobile is not available. In addition, aiming at the coupling synergistic effect of the hydrogen network, the power grid and the traffic network, whether the introduction of the hydrogen infrastructure can improve the utilization rate of renewable energy sources and relieve traffic jams or not. No effective solution to the above-mentioned problems has been proposed yet.

In previous inventions, the problem of joint planning of power and hydrogen infrastructure, taking into account source load uncertainty, has been widely studied. Most of these are addressed by robust optimization methods, where the hydrogen demand is set to exogenous uncertainty, or by stochastic optimization methods, where the exact probability distribution is known. However, in the case that the hydrogen fuel automobile at the current stage is not popular, the limitation of hydrogen station planning is not considered, and the obtained planning result may cause the situations of low energy utilization rate, over planning or insufficient planning scale. In fact, the larger the investment scale of the hydrogen infrastructure, the more consumers will be attracted to purchase hydrogen fuel cars, i.e. the development of the hydrogen fuel car trade market will depend on the planning decisions of the hydrogen plant. Conversely, hydrogen demand uncertainty of hydrogen fuelled vehicles also needs to be considered in the hydrogen plant planning problem. Thus, this constitutes exactly the decision-dependent relationship between hydrogen demand and hydrogen plant planning decisions.

Based on the above discussion, the present invention considers that decision-dependent uncertainty of hydrogen vehicle hydrogenation requirements needs to be considered in the hydrogen infrastructure planning problem in the early stages of hydrogen fuel vehicle development. In real life, many random factors are affected by decision selection, known as decision-dependent uncertainty (DDU). For example, production decisions are affected by investment information, component reliability is affected by maintenance decisions, transportation requirements are affected by road expansion decisions, and so on. Depending on the specific decision mechanism in the application scenario, decision-dependent uncertainties can be broadly divided into two categories: 1) Decisions affect the implementation of uncertainty factors, which often model optimization problems as robust or random optimizations. 2) The decision affects the probability distribution of the uncertainty factor and is achieved independent of the decision. The former is well studied in grid scenarios, often utilizing optimality conditions and projection theory when solving optimization problems. The latter is typically modeled as a Distributed Robust Optimization (DRO) problem, applied to pre-disaster planning, system or road maintenance decision making, etc. However, according to previous research experience, the optimization decision of the robust optimization model is relatively conservative, and random optimization needs to obtain accurate probability distribution of uncertain parameters in advance. In practical problems we may only be able to obtain partial hydrogen demand data for hydrogen fuelled automobiles, which happens to fit the modeling characteristics of the distributed robust optimization problem.

In distributed robust optimization with decision-dependent uncertainty, the key to modeling is to define fuzzy sets, which are widely used including fuzzy sets based on moment and on the neisseria metric. For the moment-based decision-dependent distribution robust optimization problem, the mean and variance of the uncertainty parameters depend on the decision, but such fuzzy sets do not guarantee any convergence of the unknown distribution to the true distribution. For the decision-dependent distribution robust optimization problem based on the Neisseria metric, empirical distribution is constructed by using samples, probability weights are not distributed, and the confidence upper limit is quantized by the Neisseria radius, so that better sample external performance can be realized. Meanwhile, the fuzzy set based on the Neisseria metric can control the conservation degree of the solution through the radius, so that the operation flexibility of the power system is ensured. And the fuzzy set based on the Neisseria metric has a potential polyhedral structure, and is more suitable for the algorithm development in a linear programming or linear cone dual framework. In the prior art, the decision-dependent uncertainty research based on the gas metric is less, only the situation that the decision variable and the uncertainty variable are both binary variables is considered, which does not accord with the characteristics of the hydrogen infrastructure planning problem. And the decision-dependent distribution robust optimization theory based on the fuzzy set of the gas metric is not applied to the energy network.

In summary, to accelerate the low-carbon construction of urban traffic, how to reasonably plan a hybrid energy supply infrastructure capable of simultaneously charging and hydrogenating is a blank in the current hydrogen infrastructure planning field under the condition that hydrogen fuel automobiles are not popular at present. Meanwhile, the cooperative effect of the three-network coupling network is fully utilized, redundant renewable energy sources of the power grid are fully consumed through site selection and scale planning of the infrastructure, and traffic flow of the traffic network is redistributed. Furthermore, as to the modeling method of the robust optimization problem considering uncertainty, it is known from the above discussion that decision-dependent distributed robust optimization with low degree of decision conservation has a significant advantage in the hydrogen infrastructure planning problem, and has almost never been studied in the energy network scenario.

Accordingly, those skilled in the art are working to develop a hydrogen-containing Hybrid Energy Supply Infrastructure (HESI) planning method, including site and scale planning of fast fill stations (FCS), hydro stations (HRS), electrical transfer units (P2G), and hydrogen pipelines. When hydrogen fuel automobiles (HFCV) are not popular, the method can effectively reduce conservation of planning decisions, and simultaneously well adapt to the increase of the hydrogenation demands of the hydrogen fuel automobiles in the future. In addition, the method can effectively reduce the Photovoltaic (PV) light rejection rate and simultaneously reduce the traffic jam cost. Has important significance for promoting the popularization of hydrogen fuel automobiles.

Disclosure of Invention

In view of the above-mentioned drawbacks of the prior art, the present invention aims to solve the technical problems that in the early stage of development of a hydrogen-fueled vehicle, the development of the hydrogen-fueled vehicle is mutually restricted with the planning of a hydrogen infrastructure, the utilization rate of the hydrogen infrastructure is low, and the insufficient data of hydrogen charging requirements cause the planning difficulty, that is, the problem of 'prior chickens or prior eggs' between the vehicle and a station, and meanwhile, most of the existing planning technologies of hydrogen-containing hybrid energy supply infrastructure consider only the benefit of energy sources, and do not fully utilize the synergistic effect of the three-network coupling network of a power network, a hydrogen network and a traffic network, so as to realize the joint optimization of the three networks.

To achieve the above object, the present invention provides a hydrogen-containing hybrid energy supply infrastructure planning method including the steps of:

step 1, modeling three networks, namely a power grid, a hydrogen network and a traffic network;

step 2, constructing a decision-dependent distribution robust optimization model;

step 3, reconstructing and solving a distributed robust optimization problem;

step 4, performing example simulation analysis on the three-network coupling system;

the step 1 further includes:

step 1.1, constructing a physical scene of a coupling network and establishing an optimization target;

step 1.2, constructing a traffic network constraint;

step 1.3, constructing hydrogen net constraint;

Step 1.4, constructing power grid constraint;

step 1.5, constructing coupling constraint;

the step 2 further includes:

step 2.1, constructing a hydrogen demand decision-dependent fuzzy set;

step 2.2, distributing robust optimization framework construction;

the step 3 further includes:

step 3.1, constructing decision-dependent scene probability;

step 3.2, linearizing a norm term;

and 3.3, opportunistic constraint reconstruction.

Further, the step 1.1 further includes: in the hydrogen-containing hybrid energy supply infrastructure planning, the hydrogen adding station, the charging station, the electric conversion device and the hydrogen pipeline are planned according to the charging and hydrogen adding requirements and the economic benefits of the node, and the investment decisions are respectivelyAnd w _l Are binary decision variables; assuming that the hydrogen charge of each hydrogen fuelled automobile is fixed, then the hydrogen load is a discrete variable; the hydrogen net is coupled with the power grid through an electric gas conversion device P2G, renewable energy sources of the power grid are fully utilized, the hydrogen produced by the water electrolysis device is compressed through a compressor and stored in a hydrogen storage tank (HS), and the hydrogen is supplied to a hydrogen fuel automobile (HFCV) through a hydrogenation station (HRS) after energy conversion and storage; when the PV output is surplus, the part which cannot be stored by the energy storage device (ES) reduces the light rejection rate through P2G hydrogen production; hydrogen loading that is not met for P2G when PV output is insufficient;

The optimization objective is to minimize the sum of the HESI annual investment cost and the annual average operating cost as follows:

wherein c ^hy 、c ^P2G 、c ^HS 、And c ^FCS HR respectivelyS, P2G, HS, annual investment costs for hydrogen pipelines and FCS,and->The investment capacities of HRS, P2G, HS and FCS at node i respectively,is a node set;

the operation cost comprises light rejection penalty, electric energy trade with the main network, unsatisfied charging load penalty, traffic jam and unsatisfied charging load penalty.

Further, the step 1.2 further includes: it is assumed that a traffic network with a limited range comprises a plurality of starting points o and a plurality of ending points d, and a plurality of paths p are arranged to enable the electric automobile to pass from the starting points o to the ending points d, wherein each path p consists of a plurality of links l; establishing link traffic at each moment in a traffic networkAnd (4) path traffic->Is a model of the relationship:

to ensure that each electric vehicle performs a charging operation through the traffic network, the following constraints are added, representing the flow on the link at the fast charging station to which the grid node i is connectedAnd (4) path traffic->Is a model of the relationship:

according to the united states road bureau function, the relationship between the travel time of each vehicle on link l and the link traffic is:

the total congestion time on link l is:

Referring to the traffic flow optimization management problem, according to the Wardrop user balancing principle, when the optimization problem obtains an optimal decision, no vehicle in the traffic network can change the self-running decision to lower the total cost; this optimization problem is equivalent to the following Karush-Kuhn-turner (KKT) condition:

linearizing equations (12) and (11) using the large M method and the piecewise linearization method, respectively, and replacing equations (13) and (14), respectively:

further, the step 1.3 further includes: for HS constraint, equation (15) is an HS energy balance equation, equation (16) limits the maximum hydrogen injection and release of HS, and the state of charge and recirculation conditions of HS are constrained by equations (17) and (18):

H _i,0 ＝H _i,T (18)

regarding the planning decisions for HRS, P2G and hydrogen pipelines, there are the following limitations:

in order to make the final optimization problem easy to handle, a hydrogen pipeline modeling method divided by region concepts is applied; assuming that each candidate hydrogen node is analogically to a region and the net flow per region is balanced, a balance constraint (21) and a hydrogen flow rate limit (22) are derived:

meanwhile, the hydrogen pipelines have the capability of wire-package storage, and each hydrogen pipeline storage dynamic equation gives pipeline capacity constraint and circulation condition for (23), (24) and (25) respectively:

The equilibrium constraints of the hydrogen network are represented by formula (26), which characterizes the conservation relationship of hydrogen production, hydrogen demand, piping, storage, and purchase:

equation (27) constrains the upper and lower limits of hydrogen purchase per node:

in the constraint (28) of the first and second members,indicating whether there is a hydrogen pipe connecting the hydrogen source and node i is a binary variable requiring additional constraints to be added in a particular simulation case depending on the particular network configuration:

constraint (29) ensures non-negativity of the hydrogen rejection requirement:

constraint (30) specifies upper and lower limits for the total hydrogen demand of the HFCV:

in formula (30)Is an uncertain variable.

Further, the step 1.4 further includes: equations (31) and (32) are active power balance and line capacity constraints:

constraint (33) limits the upper and lower limits of power each node can purchase from the main network:

formulas (34) and (35) represent upper and lower limits of P2G capacity and PV waste amount, respectively:

constraint (36) ensures non-negativity of the abandoned EV charging demand:

constraint (37) limits the overall charge demand of the EV:

constraints (38) - (42) represent relevant constraints for ES:

E _e,0 ＝E _e,T (41)

setting the actual output of the PV as an exogenous uncertainty variable, described by an uncertainty set V; wherein the predicted value of the photovoltaic output isThe fluctuation interval is +.> Threshold for uncertainty variable, robustness for adjusting photovoltaic output:

The step 1.5 further comprises: coupling constraints (44) and (45) represent the relationship between vehicle flow and charge demand and hydrogen demand, respectively:

the hydrogen production of P2G is expressed as a steady-state linear function of input power as shown in equation (46):

further, the step 2.1 further includes: setting an uncertainty variable hydrogen demand toN _b N is the possible value of the uncertainty quantity at each node for the total number of nodes _s There is a scene number->The index is n; in DRO-DDUs it is often assumed that the DDU set has a limited decision independent support, since assigning a 0 probability to a specific element amounts to excluding it from the support, a mapping of points to sets is achieved, where the decision dependency is reflected by a varying probability quality function; for limited support The candidate probability distribution in (a) is p _n (w ^hy ) Corresponding scene->And p _n (w ^hy )|| ₁ ＝1；

Definition of the neisseria metricThe following are provided:

wherein, the liquid crystal display device comprises a liquid crystal display device,for supporting the probability distribution set of upper xi, pi is u ₁ And u ₂ Is the joint probability distribution of (a), the edge distribution is P respectively ₁ And P ₂ The method comprises the steps of carrying out a first treatment on the surface of the The terms are arbitrary norms, u ₁ -u ₂ The term "is used to refer to the unit mass from the distribution P ₁ Move to P ₂ Cost of (2);

the fuzzy set is defined as follows:

the fuzzy set can be regarded as empirically distributedAs a center, a sphere with r as a radius, wherein the radius r can clearly control the conservation of a decision result;

Consider u _t Empirical distribution and decision w of (2) ^hy Related, i.eWherein control u _t Probability measure of ε -XI ∈>Is decision w ^hy And sample space xi and w ^hy Is irrelevant; DDU ambiguity set is written in the form:

since the hydrogen demand is a discrete support, the equivalent form of the fuzzy set described above is:

further, the step 2.2 further includes: the investment optimization problem constructed in the first step is equivalent to the following distributed robust optimization

Wherein f (·) and h (·) are investment and operating costs of the HESI, respectively; deriving a reconstructed form of the DDU-DRO model in combination with the ambiguity set (50) set forth in said step 2.1:

ε≥0 (52c)

further, the step 3.1 further includes: said step 2.1 already mentions u as having dimension N _b The possible implementation of u is N _s The number of discrete values is chosen to be the number of discrete values, i.e., {0,2,.. _s -1, index k, increasing from low to high; the number of scenes isThe index is n; />Decision making for investment at node i; given parameter vector +.>Andrespectively->The probability of the possible value of the hydrogen charging demand at the node i when taking 0 and 1 is obtained by historical data, and the relationship exists inside +.>And->The probability of scene n at node i is:

wherein, the liquid crystal display device comprises a liquid crystal display device,decision-dependent probability distribution +.>The probability of the scene n is:

the probability expression (54) is a high-dimensional nonlinear function of the decision variable, and the number of scenes is exponentially related to the number of nodes; using an improved distribution shaping theory, a set of linear constraints are used to characterize the scene probabilities associated with the decisions, and by introducing a truncated vector, a linear form with polyhedral features is derived:

Wherein, the liquid crystal display device comprises a liquid crystal display device,a reference probability representing scene n; then in the question (52)Items are equivalent to

s.t. (55)

Further, the step 3.2 further includes: assuming that the norm term is a 1-norm, the constraint (52 b) is translated into:

wherein T is [ T ]],i∈[N _b ],j,n∈[N]The method comprises the steps of carrying out a first treatment on the surface of the Order theThen

Introducing auxiliary variables willThe linear approximation is:

the nonlinear term is included in the method, variable substitution is carried out by using a Mickey envelope linearization method, and the non-convex problem is relaxed into a convex problem:

the formula (60) is simplified to obtain:

further, the step 3.3 further includes: constraint (30) is a soft constraint in the coupled network, introducing the following opportunistic constraints to reduce conservation of the final decision:

it is characterized by the meaning of in fuzzy setThe probability that the hydrogen charging demand actually satisfied at the lower node i satisfies the constraint (14) should be not less than 1-sigma _i The method comprises the steps of carrying out a first treatment on the surface of the In fuzzy set +.>In this regard, the derived opportunity constraint is equivalent to z=z ₁ ∪Z ₂ Wherein

Wherein, xi, eta, z _n Is an auxiliary variable which is used to control the operation of the motor, I.I _* Is the dual norm; the opportunity constraint consists of two inequalities, equation (62) is changed to vector form:

thus, in the opportunistic constraint equivalence transformation process, the vector function in (63) is replaced with the following:

z of formula (64) ₂ Constraints are not considered in this scenario, since when a (w ^hy ) =0, the opportunity constraint of the problem will be automatically satisfied; z is Z ₁ The constraint is obviously not convex, and the sigma is small enough and sigma epsilon (0, 1/N)]In the time-course of which the first and second contact surfaces,is true, replace->Z is then ₁ Is approximated in the form:

to reduce the number of auxiliary variables, the formula (67) is relaxed to the formula (68) to significantly reduce the number of problem variables:

compared with the prior art, the invention has at least the following beneficial technical effects:

1. the invention aims to solve the problem of 'prior chicken or prior egg' of vehicles and stations in the early development stage of hydrogen fuel vehicles; experiments show that the method provided by the invention can effectively improve the energy utilization rate of the infrastructure, reduce the investment and the operation cost, effectively avoid the energy waste in the initial development stage of the hydrogen car, better adapt to the increase of the hydrogen demand of the hydrogen fuel car in the future, and provide effective reference for the development of the hydrogen car and the hydrogen infrastructure in the future;

2. according to the invention, through comparison of four simulation examples, the method provided by the invention is verified to be capable of reducing the photovoltaic reduction of the power grid, improving the energy utilization rate and effectively relieving the traffic jam while obtaining more economic and reasonable infrastructure site selection and scale planning decisions;

3. and using an improved distribution shaping technology to linearize the high-dimensional nonlinear decision-related scene probability introduced by the decision-dependent uncertainty so that the final optimization problem can be solved.

The conception, specific structure, and technical effects of the present invention will be further described with reference to the accompanying drawings to fully understand the objects, features, and effects of the present invention.

Drawings

FIG. 1 is a flow chart of a method of a preferred embodiment of the present invention;

FIG. 2 is a diagram of a coupled network framework in accordance with a preferred embodiment of the present invention;

FIG. 3 is a traffic network topology of a preferred embodiment of the present invention;

FIG. 4 is a grid topology of a preferred embodiment of the present invention;

FIG. 5 is a hydrogen pipeline topology of a preferred embodiment of the present invention;

FIG. 6 is a schematic diagram of the price of an electro-hydrogen transaction according to a preferred embodiment of the present invention;

FIG. 7 is photovoltaic power generation forecast data for a preferred embodiment of the present invention;

FIG. 8 is a decision diagram of HESI device location planning for examples 1-4 in accordance with a preferred embodiment of the present invention;

FIG. 9 is a schematic diagram of the impact of the number of HESI planning nodes selected on investment costs in accordance with a preferred embodiment of the present invention;

FIG. 10 is a schematic diagram of the effect of the number of alternative HESI programming nodes on the undershot hydrogen load in accordance with a preferred embodiment of the present invention.

Detailed Description

The following description of the preferred embodiments of the present invention refers to the accompanying drawings, which make the technical contents thereof more clear and easy to understand. The present invention may be embodied in many different forms of embodiments and the scope of the present invention is not limited to only the embodiments described herein.

In the drawings, like structural elements are referred to by like reference numerals and components having similar structure or function are referred to by like reference numerals. The dimensions and thickness of each component shown in the drawings are arbitrarily shown, and the present invention is not limited to the dimensions and thickness of each component. The thickness of the components is exaggerated in some places in the drawings for clarity of illustration.

As shown in fig. 1, a method flow chart of a preferred embodiment of the present invention comprises the steps of:

step one, modeling three networks, namely a power grid, a hydrogen network and a traffic network. The embodiment fully utilizes the synergistic effect of the three-network coupling network of the power grid, the hydrogen network and the traffic network, fully consumes redundant renewable energy sources of the power grid through site selection and scale planning of the infrastructure, and redistributes traffic flow of the traffic network to realize joint optimization of the three networks. The specific operation is as follows:

s1, constructing a coupling network physical scene and establishing an optimization target. In the hydrogen-containing hybrid energy supply infrastructure planning, the hydrogen adding station, the charging station, the electric conversion device and the hydrogen pipeline are planned according to the charging and hydrogen adding requirements and the economic benefits of the node, and the investment decisions are respectivelyAnd w _l Are binary decision variables. Assuming that the hydrogen charge of each hydrogen fuelled automobile is fixed, the hydrogen load is a discrete variable. The specific architecture of the three-network coupling network of the power grid, the hydrogen network and the traffic network is shown in fig. 2. The hydrogen net is coupled with the power grid through P2G, renewable energy sources of the power grid are fully utilized, the hydrogen produced by the water electrolysis device is compressed through the compressor and stored in the hydrogen storage tank (HS), and the hydrogen is supplied to the HFCV through HRS after energy conversion and storage. When the PV output is surplus, the part which cannot be stored by the energy storage device (ES) reduces the light rejection rate through P2G hydrogen production. When the PV output is insufficient, the hydrogen supply system and the hydrogen supply method realize hydrogen purchasing to a hydrogen source and flexible hydrogen transmission among hydrogen nodes by planning a hydrogen pipeline for the hydrogen load which cannot be met by P2G. In addition, in this embodiment, the hydrogen output from the hydrogen storage tank is not considered to be discharged through the fuel cell and returned to the power grid, because the efficiency of performing the electric hydrogen conversion multiple times is extremely low, unnecessary resource waste is caused, and the fuel cell has high cost and can not be adopted in the initial infrastructure planning.

The optimization objective is to minimize the sum of the HESI annual investment cost and the annual average operating cost.

Wherein c ^hy 、c ^P2G 、c ^HS 、And c ^FCS Annual investment costs for HRS, P2G, HS, hydrogen pipeline and FCS respectively, And->The investment capacities of HRS, P2G, HS and FCS at node i respectively,is a set of nodes. The operation cost comprises light rejection penalty, electric energy trade with the main network, unsatisfied charging load penalty, traffic jam and unsatisfied charging load penalty.

S2, constructing traffic network constraint. It is assumed that a limited range of traffic network includes a plurality of start points o and a plurality of end points d, and that there are a plurality of paths p, each of which is composed of a plurality of links l, for the electric vehicle from the start point o to the end point d. Establishing link traffic at each moment in a traffic networkAnd (4) path traffic->Is a model of the relationship:

/>

further, in order to ensure that each electric automobile performs one-time charging operation through the traffic network, the following constraint is addedRepresenting the flow on a link at a quick charging station to which a grid node i is connectedAnd (4) path traffic->Is a model of the relationship:

according to the united states road bureau function, the relationship between the travel time of each vehicle on link l and the link traffic is:

further, the total congestion time on link l is:

referring to the traffic flow optimization management problem, according to the Wardrop user balancing principle, when the optimization problem obtains an optimal decision, no vehicle in the traffic network can change the self-running decision so as to lower the total cost. This optimization problem may be equivalent to the following Karush-Kuhn-turner (KKT) condition:

The large M method and the piecewise linearization method are applied to linearize equations (12) and (11), respectively, and replaced with equations (13) and (14), respectively.

S3, constructing hydrogen net constraint. First, for HS constraints, the modeling form is similar to ES, equation (15) is an HS energy balance equation, equation (16) limits the maximum hydrogen injection and release of HS, and the state of charge and recirculation conditions of HS are constrained by equations (17) and (18).

/>

H _i,0 ＝H _i,T (18)

Assume that the precondition for planning a hydrogen pipeline is that HRS is planned at both ends of the candidate pipeline. Further, the present embodiment sets that if P2G is planned at a node, HRS must be planned, but even if HRS is planned, P2G may not be planned, which node may purchase hydrogen from a hydrogen source. Thus, with regard to HRS, P2G and hydrogen pipeline planning decisions, there are the following limitations:

to make the final optimization problem easy to handle, a hydrogen pipeline modeling approach that is divided by regional concepts is applied. Assume that each candidate hydrogen node is analogically to a region and that the net flow per region is balanced, resulting in a balance constraint (21) and hydrogen flow rate limit (22).

Meanwhile, the hydrogen pipelines have the capability of wire-package storage, and each hydrogen pipeline storage dynamic equation gives pipeline capacity constraint and circulation condition for (23), (24) and (25) respectively.

The equilibrium constraints of the hydrogen net are represented by formula (26), which characterizes the conservation relationship of hydrogen production, hydrogen demand, piping, storage, and purchase. Equation (27) constrains the upper and lower limits of hydrogen purchase per node. In the constraint (28) of the first and second members,indicating whether there is a hydrogen conduit connecting the hydrogen source and node i is a binary variable requiring additional constraints to be added in a particular simulation case depending on the particular network configuration. Constraint (29) ensures non-negativity of the hydrogen rejection requirement. Constraint (30) defines the upper and lower limits of the HFCV total hydrogen demand.

In formula (30)To further reduce the decision conservation of the optimization problem, for the uncertainty variables considered in this embodiment, allow +.>There is a probability that the constraint (30) is not satisfied, i.e., a small amount of hydrogen demand may exceed the upper planning limit for the node. Thus, the soft constraint (30) may be configured in the form of an opportunistic constraint, the specific form and processing method being described in step three. />

S4, constructing power grid constraint. In the embodiment, the running characteristic of the power grid is simulated by using the direct current power flow, although the alternating current power flow is more accurate, a large number of nonlinear items are introduced to the optimization problem, and although the direct current power flow has a certain error, the current scene is a facility early planning problem, and power flow parameters such as voltage and the like have no obvious influence on the model and output. Therefore, it is reasonable to apply a direct current power flow here. Equations (31) and (32) are active power balance and line capacity constraints. Constraints (33) limit the upper and lower limits of power that each node can purchase from the main network. Equations (34) and (35) represent the upper and lower limits of the P2G capacity and the PV waste, respectively. Constraint (36) ensures non-negativity of the abandoned EV charging demand. Constraints (37) limit the overall charge demand of the EV. About (38) - (42) represent relevant constraints of ES.

E _e,0 ＝E _e,T (41)

The actual output of PV is set to an exogenous uncertainty variable, described by uncertainty set V. Which is a kind ofWherein the predicted value of the photovoltaic output isThe fluctuation interval is +.> Is a threshold value of an uncertainty variable and is used for adjusting the robustness of the photovoltaic output.

S5, constructing coupling constraint. Coupling constraints (44) and (45) represent the relationship between vehicle flow and charge demand and hydrogen demand, respectively. Additionally, the hydrogen production of P2G may be expressed as a steady state linear function of input power, as shown in equation (46).

/>

And step two, constructing a decision-dependent distribution robust optimization model. To more effectively and reasonably plan the HESI before the full popularity of the HFCV, this embodiment builds a hydrogen demand DDU fuzzy set. On the basis, a DDU-DRO planning model is established. The uncertainty of the hydrogen charging demand of the hydrogen vehicle is represented by skillfully utilizing a decision-dependent mechanism, and the problems of the hydrogen vehicle in the early development stage, the mutual restriction of the hydrogen vehicle development and the hydrogen infrastructure planning, low utilization rate of the hydrogen infrastructure and difficult planning are effectively solved. The specific operation is as follows:

s1, constructing a hydrogen demand decision-dependent fuzzy set. For convenience of description, the uncertainty variable hydrogen requirement is set toN _b N is the possible value of the uncertainty quantity at each node for the total number of nodes _s The number of scenes isThe index is n. In DRO-DDUs it is often assumed that the DDU set has a limited decision independent support, since assigning a 0 probability to a specific element amounts to excluding it from the support, a point-to-set mapping is achieved, where the decision dependency is reflected by a varying probability quality function. For limited support-> The candidate probability distribution in (a) is p _n (w ^hy ) Corresponding to the sceneAnd p _n (w ^hy )|| ₁ ＝1。

Further, a gas metric is definedThe following are provided:

wherein, the liquid crystal display device comprises a liquid crystal display device,to support the probability distribution set of the upper xi, the pi is u ₁ And u ₂ Is the joint probability distribution of (a), the edge distribution is P respectively ₁ And P ₂ . The terms are arbitrary norms, u ₁ -u ₂ The term "is used to refer to the unit mass from the distribution P ₁ Move to P ₂ In this embodiment, 1 norm is adopted, which has better numerical disposability. Further, the fuzzy set is defined as follows:

the fuzzy set can be regarded as empirically distributedThe sphere with r as radius is used as the center, and the radius r can clearly control the conservation of the decision result.

Consider u _t Empirical distribution and decision w of (2) ^hy Related, i.eWherein control u _t Probability measure of ε -XI ∈>Is decision w ^hy And sample space xi and w ^hy Irrespective of the fact that the first and second parts are. Further, the DDU ambiguity set can be written as follows:

since this embodiment considers the hydrogen requirement as a discrete support, the equivalent form of the fuzzy set described above is:

S2, distributing robust optimization framework construction. The investment optimization problem constructed in step one can be equivalently the following distributed robust optimization

Wherein f (·) and h (·) are investment and operating costs of the HESI, respectively. In order to convert the worst case expectations of (51) into a processable form, in combination with the ambiguity set (50) proposed by S1, a reconstructed form of the DDU-DRO model is derived:

and thirdly, reconstructing and solving a distributed robust optimization problem. To solve the DDU-DRO problem (52), it is also subjected to the following conversions, including decision-dependent scene probabilitiesLinearization of the norm term, and equivalent transformation of the opportunity constraint implicit in the constraint (30).

S1, constructing a decision-dependent scene probability. Improved distribution shaping techniques are proposed, extending conventional distribution shaping linearization techniques to derive scenarios where the uncertainty in the present invention is a non-binary variable. Unlike the prior art, the planning decision variables of this embodiment are binary, and the uncertainty amount is a non-binary discrete variable. S1 in step two has mentioned u as N in dimension _b The possible implementation of u is N _s The number of discrete values is chosen to be the number of discrete values, i.e., {0,2,.. _s -1, index k, increasing from low to high. The number of scenes isThe index is n. / >Is the investment decision at node i. Given parameter vector +.>Andrespectively->Get the hydrogen demand at node i when 0 and 1Probability of possible value is obtained from history data and there is a relation +.>And->The probability of scene n at node i is

Wherein, the liquid crystal display device comprises a liquid crystal display device,decision-dependent probability distribution +.>The probability of the scene n is

The probability expression (54) is a high-dimensional nonlinear function of the decision variables, and the number of scenarios is exponentially related to the number of nodes, resulting in an excessively large formula. Using an improved distribution shaping theory, a set of linear constraints are used to characterize the scene probabilities associated with the decisions, and by introducing a truncated vector, a linear form with polyhedral features is derived:

wherein, the liquid crystal display device comprises a liquid crystal display device,representing the reference probability of scene n. Then in the question (52)An item may be equivalent to->

s.t. (55)

S2, linearizing the norm term. Assuming that the norm term is a 1-norm, the constraint (52 b) is translated into

Wherein T is [ T ]],i∈[N _b ],j,n∈[N]. Order theThen

Introducing auxiliary variables willLinear approximation of

The nonlinear term is included in the method, variable substitution is carried out by using a Mickey envelope linearization method, and the non-convex problem is relaxed into a convex problem:

simplifying the formula (60)

S3, opportunistic constraint reconstruction. Constraint (30) is a soft constraint in the coupled network, which may introduce the following opportunistic constraints to reduce conservation of the final decision:

It is characterized by the meaning of in fuzzy setThe probability that the hydrogen charging demand actually satisfied at the lower node i satisfies the constraint (14) should be not less than 1-sigma _i . In fuzzy set +.>In this regard, the derived opportunity constraint is equivalent to z=z ₁ ∪Z ₂ Wherein->

Wherein, xi, eta, z _n Is an auxiliary variable which is used to control the operation of the motor, I.I _* Is the dual norm. The opportunity constraint consists of two inequalities, equation (62) may be changed to vector form

Therefore, in the opportunistic constraint equivalence transformation process, the vector function in (63) is replaced with the following formula

Z of formula (64) ₂ Constraints may not be considered in this scenario, since when a (w ^hy ) =0, the opportunity constraint of the problem will be automatically satisfied. Z is Z ₁ The constraint is obviously not convex, and the sigma is small enough and sigma epsilon (0, 1/N)]In the time-course of which the first and second contact surfaces,is true, can be replaced by->Z is then ₁ Is of the approximate form of

Because the solution process has introduced some auxiliary variables, when the amount of historical sample data is large, the excessive computational burden may prevent us from fully utilizing the data. To reduce the number of auxiliary variables, equation (67) may be relaxed to equation (68) to significantly reduce the number of problem variables.

And step four, performing example simulation analysis on the three-network coupling system. In this embodiment, to verify the effectiveness of the proposed planning method, we apply it to a traffic network coupling system with 33 bus power networks and 12 nodes. The system comprises 6 HESI candidate nodes, the numbers of the HESI candidate nodes are from H1 to H6, the nodes are connected by 8 candidate hydrogen pipelines, and the topology of the three networks is shown in fig. 3, 4 and 5. Table 1 provides the main investment parameters of the HESI plant, candidate nodes H2 and H4 apply type 2 devices, while other nodes apply type 1 devices. Fig. 6 and 7 show predicted data of the unit purchase price of electricity and hydrogen and photovoltaic power generation.

TABLE 1 major investment parameters for HESI plant

Apparatus and method for controlling the operation of a device	Type 1	Type 2
			HRS capacity (kg)	5000	4600
P2G Capacity (MW)	50	40
			HS Capacity Limit (kg) (maximum/minimum)	6000/200	5500/200
ES Capacity limitation (MW) (maximum/minimum)	40/7	30/5
			FCS capacity (MW)	50	40
P2G conversion efficiency	0.79	0.79
			P2G electro-hydrogen transfer factor (kg/MW)	28.7	28.7

The present embodiment provides four optimization examples to verify the superiority of the proposed method. Calculation example 1: the hydrogen demand of each node is set to a known parameter, which is an empirical value of historical data; calculation example 2: the hydrogen demand is set to a robust optimization uncertainty set, and the uncertainty threshold is set to 6; calculation example 3: constructing a hydrogen demand fuzzy set by using a DRO model considering decision independent uncertainty, wherein the empirical probability distribution of the fuzzy set of the gas metric is assumed to be a Dirac metric; calculation example 4: the hydrogen demand DRO model with decision-dependent fuzzy set is applied. Next, the planning decisions and investment running costs of the HESI are analyzed for how they are affected by the different modeling methods and the number of candidate nodes in the coupled network.

Fig. 8 and table 2 show the equipment planning site selection, investment scale and total investment costs for the four examples. Table 3 shows four example operating costs and total planning costs, which are the sum of the investment costs and the average operating costs of the HESI. As can be seen from fig. 8 and table 2, the facility scale and investment cost of example 1 and example 2 are the same, being the highest of the four examples. In table 3, since the operation cost of example 2 is higher, the total planning cost thereof is higher than that of example 1 because it takes into account the worst case of photovoltaic power generation, resulting in a corresponding increase in the operation cost. In table 2, the DRO model is used in the calculation example 3 and the calculation example 4, and the partial probability information of the uncertain parameters is considered, so that the conservation of the final optimization decision can be effectively reduced. Whereas in table 2, the operating cost of example 3 is highest among the four examples, because it has fewer P2G plans, purchases less hydrogen, and gives up a large hydrogen demand to reduce investment costs, resulting in higher operating costs. The method of example 3 can still effectively reduce the conservation of decisions and reduce investment costs in early infrastructure planning. In example 4, the facility planning scale, investment cost, and operating cost are all the lowest of the four examples. Meanwhile, each of the operation costs in example 4, including the traffic congestion cost, the hydrogen purchase cost, and the unsatisfied hydrogen demand cost, is at a low level. The number of HRS, P2G, FCS and hydrogen pipes planned is smaller compared to other examples and reduces the operating costs while ensuring that there is no excessive load rejection. Furthermore, only example 4 reduced the investment amount of hydrogen piping, indicating that the local P2G hydrogen production at certain nodes was able to meet the hydrogen load. The DDU-DRO modeling method is applied in the calculation example 4, and the influence of planning decisions on uncertain parameter probability distribution is considered, so that the investment scale is properly reduced while the future requirements are met, and the facility utilization rate is improved.

TABLE 2 four example investment Scale and Total investment costs

Investment scale and cost	Example 1	EXAMPLE 2	EXAMPLE 3	EXAMPLE 4
					HRS	6	6	6	5
P2G	6	6	3	3
					FCS	4	4	2	2
Hydrogen pipeline	5	5	5	4
					HESI investment cost (M$)	13.29	13.29	8.72	8.13

Table 3 four example operating costs and total planning costs

Cost of operation	Example 1	EXAMPLE 2	EXAMPLE 3	EXAMPLE 4
					Traffic congestion cost (M$)	0.27	0.88	0.64	0.40
Abandon light cost (M$)	0	0	0	0
					Purchasing hydrogen into productBook (M$)	6.93	7.54	5.08	3.70
Hydrogen rejection cost (M$)	1.73	3.81	11.24	2.52
					Total cost of operation (M$)	8.93	12.23	16.96	6.62
Total planning cost (M$)	22.22	25.52	25.68	14.75

In the coupled network, different planning decisions for HRS and FCS may redistribute traffic flows. Of the four above examples, example 4 has the lowest traffic congestion cost. This shows that the planning decision obtained by the DDU-DRO modeling method provided by this embodiment can further optimize traffic network flow distribution, and effectively alleviate traffic congestion. In addition, the photovoltaic cut-down costs for the four examples are all zero, which indicates that the four examples have preferentially planned enough P2G to utilize zero-running cost photovoltaic power generation, however, only example 4 achieves the same renewable energy consumption as the other examples with the lowest investment and running costs.

In addition, for the four above examples, the present example compares the HESI investment cost with the hydrogen supply to the hydrogen vehicle under the different total number of candidate nodes. Simulation results fig. 9 and 10 show that the investment cost and the hydrogen rejection demand cost of example 1 and example 2 are greatly affected by the total number of candidate nodes, and the trend of change is almost linear. The fluctuations of example 3 and example 4 are relatively small and the investment costs are almost the same under the restriction of the number of different candidate nodes. However, the cost of hydrogen rejection requirements for example 4 decreases significantly with increasing number of candidate nodes, while the cost of hydrogen rejection requirements for example 3 does not decrease significantly. This is because example 4 considers the increased hydrogen demand that depends on the decision to open more candidate nodes, reducing the number of candidate nodes would impair the performance of the algorithm to accommodate the increased demand. While for other examples, the cost of hydrogen rejection requirements decreases as the number of candidate nodes increases, they incur significant capital equipment investment costs. This shows that the DDU-DRO modeling approach proposed by this embodiment has significant advantages in accommodating future increases in hydrogen demand.

Aiming at the problems of the initial development stage of the hydrogen vehicle, the mutual restriction of the development of the hydrogen vehicle and the planning of the hydrogen infrastructure, low utilization rate of the hydrogen infrastructure and difficult planning, the invention develops a method for planning the hydrogen-containing hybrid energy supply infrastructure by considering the uncertainty of the decision-making of the hydrogen demand. When the hydrogen vehicle is not popular, the method can effectively reduce conservation of planning decisions, reduce investment and operation cost to the maximum extent on the premise of meeting energy requirements, and can better adapt to the increase of the future hydrogen fuel vehicle requirements. In addition, the method can effectively reduce the photovoltaic light rejection rate of the power grid, reduce the traffic jam cost and has important significance for promoting the popularization of hydrogen fuel automobiles. The invention mainly contributes to the following:

in order to improve the operation flexibility of the energy system, a three-network coupling network of a hydrogen network, a traffic network and a power grid is established. The hydrogen network is coupled with the power grid through P2G, and hydrogen is supplied to the HFCV through HRS to coordinate with the traffic network. The proposed HESI planning model utilizes P2G technology to promote renewable energy consumption, and redistributes traffic network traffic through infrastructure site selection planning to alleviate traffic congestion.

The hydrogen demand decision-making dependent uncertainty fuzzy set is constructed, and the actual probability distribution of hydrogen demand is set as a gas sphere centered on an empirical distribution that depends on the HRS's site selection and scale planning decisions. Furthermore, a two-stage decision-dependent distribution robust planning model is constructed, an improved distribution shaping method is introduced to process a high-dimensional nonlinear function caused by a decision-dependent mechanism, a robust peer-to-peer conversion method and an opportunity constraint processing method are applied, and a resolvable form of the DRO problem is deduced.

The example simulation result shows that the method provided by the invention can obtain more economic and reasonable HESI site selection and scale planning decisions in the initial stage of hydrogen vehicle development, reduce photovoltaic reduction, improve the energy utilization rate and effectively relieve traffic jams. And the simulation result can infer that the DDU-DRO method can better adapt to the increase of the hydrogen filling requirement of the future hydrogen vehicle compared with other traditional planning methods.

The foregoing describes in detail preferred embodiments of the present invention. It should be understood that numerous modifications and variations can be made in accordance with the concepts of the invention without requiring creative effort by one of ordinary skill in the art. Therefore, all technical solutions which can be obtained by logic analysis, reasoning or limited experiments based on the prior art by the person skilled in the art according to the inventive concept shall be within the scope of protection defined by the claims.

Claims (10)

1. A method of planning a hydrogen-containing hybrid energy supply infrastructure, the method comprising the steps of:

step 1, modeling three networks, namely a power grid, a hydrogen network and a traffic network;

step 2, constructing a decision-dependent distribution robust optimization model;

step 3, reconstructing and solving a distributed robust optimization problem;

Step 4, performing example simulation analysis on the three-network coupling system;

the step 1 further includes:

step 1.1, constructing a physical scene of a coupling network and establishing an optimization target;

step 1.2, constructing a traffic network constraint;

step 1.3, constructing hydrogen net constraint;

step 1.4, constructing power grid constraint;

step 1.5, constructing coupling constraint;

the step 2 further includes:

step 2.1, constructing a hydrogen demand decision-dependent fuzzy set;

step 2.2, distributing robust optimization framework construction;

the step 3 further includes:

step 3.1, constructing decision-dependent scene probability;

step 3.2, linearizing a norm term;

and 3.3, opportunistic constraint reconstruction.

2. The hydrogen-containing hybrid energy supply infrastructure planning method of claim 1, wherein step 1.1 further comprises: in the hydrogen-containing hybrid energy supply infrastructure planning, the hydrogen adding station, the charging station, the electric conversion device and the hydrogen pipeline are planned according to the charging and hydrogen adding requirements and the economic benefits of the node, and the investment decisions are respectivelyAnd w _l Are binary decision variables; assuming that the hydrogen charge of each hydrogen fuelled automobile is fixed, then the hydrogen load is a discrete variable; the hydrogen net is coupled with the power grid through an electric gas conversion device P2G, renewable energy sources of the power grid are fully utilized, the hydrogen produced by the water electrolysis device is compressed through a compressor and stored in a hydrogen storage tank (HS), and the hydrogen is supplied to a hydrogen fuel automobile (HFCV) through a hydrogenation station (HRS) after energy conversion and storage; when the PV output is surplus, the part which cannot be stored by the energy storage device (ES) reduces the light rejection rate through P2G hydrogen production; hydrogen loading that is not met for P2G when PV output is insufficient;

The optimization objective is to minimize the sum of the HESI annual investment cost and the annual average operating cost as follows:

wherein c ^hy 、c ^P2G 、c ^HS 、And c ^FCS Annual investment costs for HRS, P2G, HS, hydrogen pipeline and FCS respectively,and->The investment capacities of HRS, P2G, HS and FCS at node i respectively,is a node set;

the operation cost comprises light rejection penalty, electric energy trade with the main network, unsatisfied charging load penalty, traffic jam and unsatisfied charging load penalty.

3. The hydrogen-containing hybrid energy supply infrastructure planning method of claim 2, wherein step 1.2 further comprises: it is assumed that a traffic network with a limited range comprises a plurality of starting points o and a plurality of ending points d, and a plurality of paths p are arranged to enable the electric automobile to pass from the starting points o to the ending points d, wherein each path p consists of a plurality of links l; establishing link traffic at each moment in a traffic networkAnd (4) path traffic->Is a model of the relationship:

to ensure that each electric vehicle performs a charging operation through the traffic network, the following constraints are added, representing the flow on the link at the fast charging station to which the grid node i is connectedAnd (4) path traffic->Is a model of the relationship:

according to the united states road bureau function, the relationship between the travel time of each vehicle on link l and the link traffic is:

The total congestion time on link l is:

referring to the traffic flow optimization management problem, according to the Wardrop user balancing principle, when the optimization problem obtains an optimal decision, no vehicle in the traffic network can change the self-running decision to lower the total cost; this optimization problem is equivalent to the following Karush-Kuhn-turner (KKT) condition:

linearizing equations (12) and (11) using the large M method and the piecewise linearization method, respectively, and replacing equations (13) and (14), respectively:

4. the hydrogen-containing hybrid energy supply infrastructure planning method of claim 3, wherein step 1.3 further comprises: for HS constraint, equation (15) is an HS energy balance equation, equation (16) limits the maximum hydrogen injection and release of HS, and the state of charge and recirculation conditions of HS are constrained by equations (17) and (18):

H _i,0 ＝H _i,T (18)

regarding the planning decisions for HRS, P2G and hydrogen pipelines, there are the following limitations:

in order to make the final optimization problem easy to handle, a hydrogen pipeline modeling method divided by region concepts is applied; assuming that each candidate hydrogen node is analogically to a region and the net flow per region is balanced, a balance constraint (21) and a hydrogen flow rate limit (22) are derived:

meanwhile, the hydrogen pipelines have the capability of wire-package storage, and each hydrogen pipeline storage dynamic equation gives pipeline capacity constraint and circulation condition for (23), (24) and (25) respectively:

The equilibrium constraints of the hydrogen network are represented by formula (26), which characterizes the conservation relationship of hydrogen production, hydrogen demand, piping, storage, and purchase:

equation (27) constrains the upper and lower limits of hydrogen purchase per node:

in the constraint (28) of the first and second members,indicating whether there is a hydrogen pipe connecting the hydrogen source and node i is a binary variable requiring additional constraints to be added in a particular simulation case depending on the particular network configuration:

constraint (29) ensures non-negativity of the hydrogen rejection requirement:

constraint (30) specifies upper and lower limits for the total hydrogen demand of the HFCV:

in formula (30)Is an uncertain variable.

5. The hydrogen-containing hybrid energy supply infrastructure planning method of claim 4, wherein step 1.4 further comprises: equations (31) and (32) are active power balance and line capacity constraints:

constraint (33) limits the upper and lower limits of power each node can purchase from the main network:

formulas (34) and (35) represent upper and lower limits of P2G capacity and PV waste amount, respectively:

constraint (36) ensures non-negativity of the abandoned EV charging demand:

constraint (37) limits the overall charge demand of the EV:

constraints (38) - (42) represent relevant constraints for ES:

E _e,0 ＝E _e,T (41)

setting the actual output of the PV as an exogenous uncertainty variable, described by an uncertainty set V; wherein the predicted value of the photovoltaic output is The fluctuation interval is +.> Threshold for uncertainty variable, robustness for adjusting photovoltaic output:

the step 1.5 further comprises: coupling constraints (44) and (45) represent the relationship between vehicle flow and charge demand and hydrogen demand, respectively:

the hydrogen production of P2G is expressed as a steady-state linear function of input power as shown in equation (46):

6. the hydrogen-containing hybrid energy supply infrastructure planning method of claim 5, wherein step 2.1 further comprises: setting an uncertainty variable hydrogen demand toN _b N is the possible value of the uncertainty quantity at each node for the total number of nodes _s There is a scene number->The index is n; in DRO-DDUs it is often assumed that the DDU set has a limited decision independent support, since assigning a 0 probability to a specific element amounts to excluding it from the support, a mapping of points to sets is achieved, where the decision dependency is reflected by a varying probability quality function; for limited support->The candidate probability distribution in (a) is p _n (w ^hy ) Corresponding scene->And p _n (w ^hy )|| ₁ ＝1；

Definition of the neisseria metricThe following are provided:

wherein, the liquid crystal display device comprises a liquid crystal display device,for supporting the probability distribution set of upper xi, pi is u ₁ And u ₂ Is the joint probability distribution of (a), the edge distribution is P respectively ₁ And P ₂ The method comprises the steps of carrying out a first treatment on the surface of the The terms are arbitrary norms, u ₁ -u ₂ The term "is used to refer to the unit mass from the distribution P ₁ Move to P ₂ Cost of (2);

the fuzzy set is defined as follows:

the fuzzy set can be regarded as empirically distributedAs a center, a sphere with r as a radius, wherein the radius r can clearly control the conservation of a decision result;

consider u _t Empirical distribution and decision w of (2) ^hy Related, i.eWherein control u _t Probability measure of EIs decision w ^hy And sample space xi and w ^hy Is irrelevant; DDU ambiguity set is written in the form:

since the hydrogen demand is a discrete support, the equivalent form of the fuzzy set described above is:

7. the hydrogen-containing hybrid energy supply infrastructure planning method of claim 6, wherein step 2.2 further comprises: the investment optimization problem constructed in the first step is equivalent to the following distributed robust optimization

Wherein f (·) and h (·) are investment and operating costs of the HESI, respectively; deriving a reconstructed form of the DDU-DRO model in combination with the ambiguity set (50) set forth in said step 2.1:

ε≥0(52c)

8. as claimed inThe hydrogen-containing hybrid energy supply infrastructure planning method of claim 7, wherein the step 3.1 further comprises: said step 2.1 already mentions u as having dimension N _b The possible implementation of u is N _s The number of discrete values is chosen to be the number of discrete values, i.e., {0,2,.. _s -1, index k, increasing from low to high; the number of scenes is The index is n; />Decision making for investment at node i; given parameter vector +.>And->Respectively->The probability of the possible value of the hydrogen charging demand at the node i when taking 0 and 1 is obtained by historical data, and the relationship exists inside +.>Andthe probability of scene n at node i is:

wherein, the liquid crystal display device comprises a liquid crystal display device,decision-dependent probability distribution +.>The probability of the scene n is:

the probability expression (54) is a high-dimensional nonlinear function of the decision variable, and the number of scenes is exponentially related to the number of nodes; using an improved distribution shaping theory, a set of linear constraints are used to characterize the scene probabilities associated with the decisions, and by introducing a truncated vector, a linear form with polyhedral features is derived:

wherein, the liquid crystal display device comprises a liquid crystal display device,a reference probability representing scene n; then in the question (52)Items are equivalent to

s.t.(55)

9. The hydrogen-containing hybrid energy supply infrastructure planning method of claim 8, wherein step 3.2 further comprises: assuming that the norm term is a 1-norm, the constraint (52 b) is translated into:

wherein T is [ T ]],i∈[N _b ],j,n∈[N]The method comprises the steps of carrying out a first treatment on the surface of the Order theThen

Introducing auxiliary variables willThe linear approximation is:

the nonlinear term is included in the method, variable substitution is carried out by using a Mickey envelope linearization method, and the non-convex problem is relaxed into a convex problem:

The formula (60) is simplified to obtain:

10. the hydrogen-containing hybrid energy supply infrastructure planning method of claim 9, wherein step 3.3 further comprises: constraint (30) is a soft constraint in the coupled network, introducing the following opportunistic constraints to reduce conservation of the final decision:

it is characterized by the meaning of in fuzzy setThe hydrogen charging requirement actually met at the lower node iThe probability of satisfying constraint (14) should be not less than 1-sigma _i The method comprises the steps of carrying out a first treatment on the surface of the In fuzzy set +.>In this regard, the derived opportunity constraint is equivalent to z=z ₁ ∪Z ₂ Wherein

Wherein, xi, eta, z _n Is an auxiliary variable which is used to control the operation of the motor, I.I _* Is the dual norm; the opportunity constraint consists of two inequalities, equation (62) is changed to vector form:

thus, in the opportunistic constraint equivalence transformation process, the vector function in (63) is replaced with the following:

z of formula (64) ₂ Constraints are not considered in this scenario, since when a (w ^hy ) =0, the opportunity constraint of the problem will be automatically satisfied; z is Z ₁ The constraint is obviously not convex, and the sigma is small enough and sigma epsilon (0, 1/N)]In the time-course of which the first and second contact surfaces,is true, replace->Z is then ₁ Is approximated in the form:

to reduce the number of auxiliary variables, the formula (67) is relaxed to the formula (68) to significantly reduce the number of problem variables: