Reliability evaluation method for integrated electricity-hydrogen energy system based on data-driven polynomial chaos expansion

Abstract

The application discloses a reliability evaluation method for an electric-hydrogen integrated energy system based on data-driven polynomial chaos expansion, and comprises the following steps: 1) constructing an orthogonal polynomial base based on multiple moments; 2) constructing a chaos polynomial surrogate model; 3) calculating the optimal load reduction of the electric-hydrogen integrated energy system according to sample values of matching points; 4) solving each coefficient of the chaos polynomial surrogate model by using a least square method; 5) acquiring current random variables and inputting the current random variables into the chaos polynomial surrogate model to calculate the optimal load reduction of the current electric-hydrogen integrated energy system; and 6) calculating a reliability index of the electric-hydrogen integrated energy system. The chaos polynomial surrogate model is constructed through historical data of random variables, a large number of mixed random variables can be efficiently processed, information such as probability characteristics of the random variables is not required, and fast and accurate reliability evaluation can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of reliability evaluation, in particular to an electric hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion. BACKGROUND

[0002] With the continuous maturity and cost reduction of electrolytic hydrogen production technology, as well as the deep coupling of hydrogen energy "production-storage-transportation-use" and electric energy "generation-transmission-distribution-variation" in the whole process and whole chain, the electric hydrogen integrated energy system has become an important support for building a new type of power system. At the same time, the wide application of hydrogen energy in the energy consumption terminal will have a great impact on social productivity. Therefore, by studying the reliability evaluation method of the electric hydrogen integrated energy system, it can provide theoretical support and decision basis for the stable operation and scientific planning of the electric hydrogen integrated energy system, and is the premise of building a safe and reliable electric hydrogen integrated energy system.

[0003] In the aspect of reliability evaluation method of electric hydrogen integrated energy system, the existing research mainly includes analytical method represented by enumeration method and simulation method represented by Monte Carlo simulation method, but the evaluation speed of these two methods will decrease greatly with the increase of the scale of electric hydrogen integrated energy system operation state. In the problem of probability evaluation, chaos polynomial expansion (PCE) as a fast and high-precision global polynomial approximation method represents the input of random variables and the response output of optimization model as the weighted sum of a set of orthogonal polynomials, which avoids repeated calculation in probability analysis problem, thereby greatly improving the solution speed of probability analysis problem and has been widely applied. In addition, the data-driven polynomial chaos expansion (DPCE) method based on the construction of standard orthogonal basis from data statistics can efficiently handle a large number of mixed (continuous and discrete) random inputs (such as wind and light uncertainty and line random interruption), and does not require any probability distribution information. At present, the chaos polynomial expansion method has not been applied to the reliability evaluation problem of electric hydrogen integrated energy system.

[0004] In summary, in the aspect of reliability evaluation method of electric hydrogen integrated energy system, in order to solve the problem of low solution efficiency of reliability evaluation calculation caused by large scale of system operation state, it is urgent to verify the applicability of data-driven polynomial chaos expansion method in reliability evaluation problem, and to explore the efficient application of the method in the reliability evaluation problem of electric hydrogen integrated energy system, to improve the solution speed of reliability evaluation of electric hydrogen integrated energy system, to provide reliability reference index for the operation planning of electric hydrogen integrated energy system, and to help the construction of new type of power system. SUMMARY

[0005] The application aims to provide a data-driven polynomial chaos expansion-based reliability evaluation method for an electric-hydrogen integrated energy system, comprising the following steps:

[0006] 1) reading basic parameters of the electric-hydrogen integrated energy system;

[0007] 2) constructing a multi-order moment-based orthogonal polynomial basis based on historical data of mixed random variables in the basic parameters of the electric-hydrogen integrated energy system;

[0008] 3) constructing a chaotic polynomial surrogate model based on linear combinations of orthogonal polynomials;

[0009] 4) selecting collocation points based on the linear independence principle, and calculating the optimal load reduction of the electric-hydrogen integrated energy system according to sample values of the collocation points;

[0010] 5) taking the historical data of mixed random variables and the optimal load reduction of the electric-hydrogen integrated energy system as inputs and outputs of the chaotic polynomial surrogate model, and solving the coefficients of the chaotic polynomial surrogate model by using the least square method;

[0011] 6) obtaining current random variables and inputting them into the chaotic polynomial surrogate model to calculate the optimal load reduction of the current electric-hydrogen integrated energy system;

[0012] 7) calculating reliability indexes of the electric-hydrogen integrated energy system.

[0013] Further, the basic parameters of the electric-hydrogen integrated energy system include electric-hydrogen integrated energy system parameters, device element reliability parameters, and a mixed random variable historical data set.

[0014] Further, the electric-hydrogen integrated energy system parameters include power line resistance and reactance parameters of a power system, generator output upper and lower limit parameters, wind turbine and photovoltaic unit parameters, line topology connection parameters, electrolyzer, fuel cell, hydrogen storage tank, and hydrogen dispenser parameters of a hydrogen energy system, and electric load and hydrogen load data;

[0015] The device element reliability parameters include failure rates and repair rates of generator units, lines, electrolyzers, fuel cells, hydrogen storage tanks, and hydrogen dispensers;

[0016] The mixed random variable historical data set includes historical data of wind turbine and photovoltaic unit outputs, and device failure-operation historical data of generator units, lines, electrolyzers, and fuel cells.

[0017] Further, in step 2), the step of constructing a multi-order moment-based orthogonal polynomial basis comprises:

[0018] 2.1) constructing a multi-dimensional PCE model, i.e.:

[0019]

[0020] where c κ is the coefficient of the κth expansion term Φ κ ; M is the number of terms of the random response Y; M = (H + N)! / (H!N!); Φ κ is the full tensor product of one-dimensional polynomials; κ = 1, 2, …, M; ξ1, ξ2, …, ξ N are random variables; N is the number of random variables; H is the order.

[0021] 2.2) Construct the orthogonal polynomial basis based on the multi-order moments using the multi-dimensional PCE model, i.e.,

[0022]

[0023] where α κ,i is the order of the ith one-dimensional polynomial.

[0024] Further, in step 3), the step of constructing the chaotic polynomial surrogate model comprises:

[0025] 3.1) Convert the historical data set of the mixed random variables in the basic parameters of the integrated energy system of electricity and hydrogen into the form of multi-order moments, and construct the one-dimensional orthogonal polynomial P i (l) (ξ i ), i.e.,

[0026]

[0027] where ξ i κ is the historical data of the mixed random variables; l = 0, 1, …, H.

[0028] where 0 ~ l order arbitrary polynomial basis is as follows:

[0029]

[0030] where μ k,i is the kth origin moment of the random variable; M p is the sampling number of the random variable.

[0031] 3.2) Normalize the one-dimensional orthogonal polynomial P to construct the chaotic polynomial surrogate model P , i.e.,

[0032]

[0033] where ||·|| represents the two-norm of the polynomial.

[0034] Further, in step 4), the step of calculating the optimal load shedding amount of the electric-hydrogen integrated energy system according to the sample values of the collocation points comprises:

[0035] 4.1) Calculate the third-order root of the one-dimensional orthogonal polynomial ;

[0036] 4.2) Randomly combine "0" and the third-order root of the one-dimensional orthogonal polynomial to obtain an initial collocation point set Ω IC ;

[0037] 4.3) Select the same number of collocation points as the number of undetermined coefficients M from the set Ω IC in turn to form a collocation point combination Ω C ;

[0038] 4.4) Calculate the coefficient matrix Φ of the collocation point combination Ω C , that is:

[0039]

[0040] 4.4) Record the rank of the coefficient matrix Φ as R Φ . If R Φ =M, it is considered that the current collocation point combination Ω C is the optimal collocation point combination, the collocation point selection operation ends, and step 4.5) is entered; otherwise, remove M-R C linearly dependent collocation points in the current collocation point combination Ω Φ , select M-R IC collocation points from the unselected collocation points in the set Ω Φ , write the selected M-R Φ collocation points into the collocation point combination Ω C to form a new collocation point combination Ω C , and return to step 4.4);

[0041] 4.5) Construct an optimal load shedding model; input the sample values of the selected collocation points into the optimal load shedding model in turn to obtain the optimal load shedding amount of the electric-hydrogen integrated energy system corresponding to each collocation point.

[0042] Further, the objective function f of the optimal load shedding model is as follows:

[0043]

[0044] In the formula, c DG is the unit penalty cost of abandoned wind and light; Ω WG and Ω PV are the wind and light unit grid node sets respectively; ΔP WG,it and ΔP PV,it are the abandoned wind and light powers respectively; T is the scheduling period; n is the node number; and cp L p,it L q L q,it L

[0045] Further, the constraint conditions of the optimal load shedding model include an electric power balance constraint, a hydrogen balance constraint, an electric-hydrogen load shedding constraint, an operation constraint of a relationship between hydrogen production and power consumption of an electrolyzer, and an operation constraint of a relationship between hydrogen consumption and power output of a fuel cell;

[0046] The electric power balance constraint is shown as follows:

[0047] P GEN,it +P WG,it +P PV,it +P FC,it =P EL,it +P load,it -L p,it +Bθ i,t (10)

[0048] In the formula, P GEN,it is active power output by a generator set; P WG,it and P PV,it are active power actually output by a wind-solar generator set; P FC,it and P EL,it are active power output by a fuel cell and active power consumed by an electrolyzer, respectively; P load,it and L p,it are active load and load shedding power, respectively; B is a node susceptance matrix for direct current flow calculation, and θ i,t is a node voltage phase angle.

[0049] The hydrogen balance constraint is shown as follows:

[0050] Q DP,in,it =Q EL,it -Q HT,in,it +Q HT,out,it -Q FC,it (11)

[0051] Q DP,out,it =Q load,it -L q,it (12)

[0052] In the formula, Q DP,in,it and Q DP,out,it are hydrogen input and output by a hydrogenation unit, respectively; Q EL,it and Q FC,itrespectively, are hydrogen production and hydrogen consumption of the fuel cell; Q HT,in,it , respectively, are hydrogen charging and hydrogen discharging of the hydrogen storage tank; Q HT,out,it , respectively, are hydrogen production and hydrogen consumption of the fuel cell; Q load,it , respectively, are hydrogen charging and hydrogen discharging of the hydrogen storage tank; Q q,it , respectively, are hydrogen load demand and reduction.

[0053] The hydrogen load reduction constraint of electricity is shown as follows:

[0054] 0≤L p,it ≤P load,it (13)

[0055] 0≤L q,it ≤Q load,it (14) The hydrogen production and power consumption relationship operation constraint of the electrolyzer is shown as follows:

[0056]

[0057] μ EL,it P EL,min ≤P EL,it ≤μ EL,it P EL,max (17)

[0058] In the formula, is the standard density of hydrogen; μ EL,it is the start-stop state variable of the electrolyzer, and if it is 1, it indicates the start state, and if it is 0, it indicates the stop state; P EL,max and P EL,min are the upper and lower limits of the power consumption of the electrolyzer, respectively; is a relationship function; m EL,t is the hydrogen production of the electrolyzer.

[0059] The hydrogen consumption and power output relationship operation constraint of the fuel cell is shown as follows:

[0060] Q FC,it = ρ H2 m FC,it (18)

[0061] P FC,it = g(m FC,it ) (19)

[0062] μ FC,it Q FC,min ≤Q FC,it ≤μ FC,it Q FC,max (20)

[0063] In the formula, μ FC,it is the start-stop state variable of the fuel cell, and if it is 1, it indicates the start state, and if it is 0, it indicates the stop state; Q FC,max, Q FC,min are upper and lower limits of hydrogen consumption of fuel cell, respectively, is the standard density of hydrogen; g(m FC,it ) is a relationship function; m FC,it is the hydrogen consumption of fuel cell.

[0064] Further, in step 5), the step of solving each coefficient of the chaotic polynomial surrogate model by using the least square method comprises:

[0065] 5.1) substituting the sample value of the random variable and the value of the response function into the following formula to solve the chaotic expansion coefficient:

[0066]

[0067] 5.2) calculating the chaotic expansion coefficient by residual sum of squares minimization, as shown in the following formula:

[0068]

[0069] In the formula, J(C) is the residual sum of squares;

[0070] 5.3) deriving the chaotic expansion coefficient to obtain each coefficient of the chaotic expansion, that is:

[0071]

[0072] In the formula, is the solved each coefficient of the chaotic expansion.

[0073] Further, in step 7), the step of calculating the reliability index of the electric-hydrogen integrated energy system comprises:

[0074] 7.1) projecting Y(ξ) onto the orthogonal polynomial Φ1 by using the Galerkin projection method to obtain:

[0075]

[0076] In the formula, E(·) is an expectation operation, Φ1(ξ) is the first expansion term of the polynomial, c κ is the coefficient of the κth expansion term Φ κ ;

[0077] 7.2) calculating the expectation μ of the response function Y(ξ) as the reliability index of the electric-hydrogen integrated energy system;

[0078] The expectation μ is as follows:

[0079]

[0080] The technical effect of the present application is self-evident, and the data-driven polynomial chaos expansion-based reliability evaluation method of an electric-hydrogen integrated energy system helps improve the reliability evaluation efficiency of the electric-hydrogen integrated energy system and realizes fast and accurate reliability evaluation.

[0081] The present application applies the data-driven polynomial chaos expansion method to the reliability evaluation problem of an electric-hydrogen integrated energy system. By constructing an arbitrary polynomial orthogonal basis and orthogonal polynomials through historical data, a chaotic polynomial surrogate model is constructed, which can output statistical moment information of random variables, efficiently process a large number of mixed random variables, and does not require information such as probability characteristics of random variables, thereby improving the solution efficiency of reliability evaluation and efficiently evaluating the reliability level of the electric-hydrogen integrated energy system. BRIEF DESCRIPTION OF DRAWINGS

[0082] Fig. 1 A flowchart of the data-driven polynomial chaos expansion-based reliability evaluation method of an electric-hydrogen integrated energy system is shown.

[0083] Fig. 2 A topology diagram of an electric-hydrogen integrated energy system based on IEEE-RTS79 is shown. DETAILED DESCRIPTION

[0084] The present application will be further described below in conjunction with examples, but should not be understood as limiting the above-mentioned subject matter of the present application to the following examples. Various substitutions and modifications can be made according to ordinary technical knowledge and conventional means in the art without departing from the above-mentioned technical idea of the present application, and all such substitutions and modifications should be included in the protection scope of the present application.

[0085] Example 1

[0086] Referring to Figs. 1-2 The data-driven polynomial chaos expansion-based reliability evaluation method of an electric-hydrogen integrated energy system includes the following steps:

[0087] 1) Read the basic parameters of the electric-hydrogen integrated energy system;

[0088] 2) Based on the historical data of the mixed random variables in the basic parameters of the electric-hydrogen integrated energy system, construct an orthogonal polynomial basis based on multiple moments;

[0089] 3) Construct a chaotic polynomial surrogate model based on the linear combination of orthogonal polynomials;

[0090] 4) Select collocation points based on the linear independence principle, and calculate the optimal load reduction of the electric-hydrogen integrated energy system according to the sample values of the collocation points;

[0091] 5) The mixed random variable historical data and the optimal load reduction amount of the electric-hydrogen integrated energy system are taken as the input and output of the chaotic polynomial surrogate model, and the least square method is used to solve the coefficients of the chaotic polynomial surrogate model;

[0092] 6) The current random variable is obtained and input into the chaotic polynomial surrogate model to calculate the optimal load reduction amount of the current electric-hydrogen integrated energy system;

[0093] 7) The reliability index of the electric-hydrogen integrated energy system is calculated.

[0094] The basic parameters of the electric-hydrogen integrated energy system include electric-hydrogen integrated energy system parameters, device element reliability parameters, and mixed random variable historical data sets.

[0095] The electric-hydrogen integrated energy system parameters include power line resistance and reactance parameters of the power system, generator output upper and lower limit parameters, wind turbine and photovoltaic unit parameters, line topology connection parameters, electrolyzer, fuel cell, hydrogen storage tank, and hydrogen dispenser parameters of the hydrogen energy system, and electric load and hydrogen load data;

[0096] The device element reliability parameters include the failure rate and repair rate of the generator unit, line, electrolyzer, fuel cell, hydrogen storage tank, and hydrogen dispenser;

[0097] The mixed random variable historical data set includes wind turbine and photovoltaic unit output historical data, and generator unit, line, electrolyzer, fuel cell device failure-operation historical data.

[0098] In step 2), the step of constructing the orthogonal polynomial basis based on the multiple moments includes:

[0099] 2.1) Construct a multi-dimensional PCE model, that is:

[0100]

[0101] In the formula, c κ is the coefficient of the κth expansion term Φ κ ; M is the number of terms of the random response Y; M = (H+N)! / (H!N!); Φ κ is the full tensor product of one-dimensional polynomials; κ = 1, 2, …, M; ξ1, ξ2, …, ξ N are random variables; N is the number of random variables; H is the order;

[0102] 2.2) Construct an orthogonal polynomial basis based on multiple moments using the multi-dimensional PCE model, that is:

[0103]

[0104] In the formula, α κ,iDegree of the i-th one-dimensional polynomial.

[0105] In step 3), the steps of constructing the chaotic polynomial surrogate model include:

[0106] 3.1) Convert the historical data set of the mixed random variable in the basic parameter of the electric hydrogen integrated energy system into the form of multiple moments, and construct one-dimensional orthogonal polynomial P i (l) (ξ i ), that is:

[0107]

[0108] In the formula, ξ i κ is the historical data of the mixed random variable; l = 0, 1,..., H;

[0109] Wherein, 0~l order arbitrary polynomial basis As follows:

[0110]

[0111] In the formula, μ k,i is the k-th central moment of the random variable; M p is the sampling number of the random variable;

[0112] 3.2) Normalize the one-dimensional orthogonal polynomial , thereby constructing the chaotic polynomial surrogate model ψ i (l) (ξ i ), that is:

[0113]

[0114] In the formula, ||·|| represents the two-norm of the polynomial.

[0115] In step 4), the steps of calculating the optimal load reduction amount of the electric hydrogen integrated energy system according to the sample value of the collocation point include:

[0116] 4.1) Calculate the three-order root of the one-dimensional orthogonal polynomial

[0117] 4.2) Randomly combine "0" and the three-order root of the one-dimensional orthogonal polynomial , to obtain the initial collocation point set Ω IC ;

[0118] 4.3) From the set Ω IC , select the same number of collocation points as the number of undetermined coefficients M in turn, to form the collocation point combination Ω C ;

[0119] ​4.4) Calculate the coefficient matrix Φ of the current combination of selected points Ω C , i.e.

[0120]

[0121] 4.4) Record the rank R of the coefficient matrix Φ Φ , if R Φ = M, consider the current combination of selected points Ω C as the optimal combination of selected points, and the selected point selection operation ends, and step 4.5) is entered; otherwise, remove M-R C linearly dependent selected points in the current combination of selected points Ω Φ , select M-R IC selected points from the unselected points in the set Ω Φ , write the selected M-R Φ selected points into the combination of selected points Ω C , form a new combination of selected points Ω C , and return to step 4.4);

[0122] 4.5) Construct an optimal load reduction model; input the sample values of the selected points into the optimal load reduction model in turn, and obtain the optimal load reduction amount of the electric-hydrogen integrated energy system corresponding to each selected point.

[0123] The objective function f of the optimal load reduction model is as follows:

[0124]

[0125] In the formula, c DG is the unit penalty cost of abandoned wind and light; Ω WG and Ω PV are the wind and light grid-connected node sets; ΔP WG,it and ΔP PV,it are the abandoned wind and light powers; T is the scheduling period; n is the number of nodes; c p is the unit reduction penalty cost of the electric load, and L p,it is the electric load reduction power; c q is the unit reduction penalty cost of the hydrogen load, and L q,it is the hydrogen load reduction power. Subscripts i and t represent node i and time t, respectively; and Δt is the time step.

[0126] The constraint conditions of the optimal load reduction model include the electric power balance constraint, the hydrogen balance constraint, the electric-hydrogen load reduction constraint, the operation constraint of the relationship between the hydrogen production amount and the consumption power of the electrolyzer, and the operation constraint of the relationship between the hydrogen consumption amount and the output power of the fuel cell.

[0127] The electric power balance constraint is as follows:

[0128] PGEN,it +P WG,it +P PV,it +P FC,it =P EL,it +P load,it -L p,it +Bθ i,t (10)

[0129] where P GEN,it is the active power output by the generator set; P WG,it , P PV,it are the active power actually output by the wind-solar generator set; P FC,it , P EL,it are the active power output by the fuel cell and the active power consumed by the electrolyzer, respectively; P load,it , L p,it are the active load and the load curtailment power, respectively; B is the nodal susceptance matrix for DC power flow calculation, and θ i,t is the nodal voltage phase angle.

[0130] The hydrogen balance constraint is shown as follows:

[0131] Q DP,in,it =Q EL,it -Q HT,in,it +Q HT,out,it -Q FC,it (11)

[0132] Q DP,out,it =Q load,it -L q,it (12)

[0133] where Q DP,in,it , Q DP,out,it are the hydrogen input and output of the hydrogenation unit, respectively; Q EL,it , Q FC,it are the hydrogen production of the electrolyzer and the hydrogen consumption of the fuel cell, respectively; Q HT,in,it , Q HT,out,it are the hydrogen charging and discharging of the hydrogen storage tank, respectively; Q load,it , L q,it are the demand and curtailment of the hydrogen load, respectively.

[0134] The electrical and hydrogen load curtailment constraint is shown as follows:

[0135] 0≤L p,it ≤P load,it (13)

[0136] 0≤L q,it ≤Q load,it (14)

[0137] The operation constraint of the hydrogen production of the electrolytic cell and the consumed power is shown as follows:

[0138]

[0139] μ EL,it P EL,min ≤P EL,it ≤μ EL,it P EL,max (17)

[0140] wherein, ρ is the standard density of hydrogen; μ EL,it is the start-stop state variable of the electrolytic cell, and if 1, it indicates the start state, and if 0, it indicates the stop state; P EL,max and P EL,min are the upper and lower limits of the consumed power of the electrolytic cell, respectively; is a relational function; m EL,t is the hydrogen production of the electrolytic cell.

[0141] The operation constraint of the hydrogen consumption of the fuel cell and the output power is shown as follows:

[0142] Q FC,it = ρ H2 m FC,it (18)

[0143] P FC,it = g(m FC,it ) (19)

[0144] μ FC,it Q FC,min ≤ Q FC,it ≤ μ FC,it Q FC,max (20)

[0145] wherein, μ FC,it is the start-stop state variable of the fuel cell, and if 1, it indicates the start state, and if 0, it indicates the stop state; Q FC,max , Q FC,min are the upper and lower limits of the hydrogen consumption of the fuel cell, respectively, ρ is the standard density of hydrogen; g(m FC,it ) is a relational function; m FC,it is the hydrogen consumption of the fuel cell.

[0146] In step 5), the step of solving the coefficients of the chaotic polynomial proxy model by using the least square method comprises:

[0147] 5.1) substituting the sample values of the random variables and the values of the response function into the following formula to solve the coefficients of the chaotic expansion:

[0148]

[0149] 5.2) Calculate the coefficients of the chaotic expansion by minimizing the residual sum of squares, as shown in the following formula:

[0150]

[0151] In the formula, J(C) is the residual sum of squares;

[0152] 5.3) Derive the coefficients of the chaotic expansion to obtain the coefficients of each term of the chaotic expansion, that is:

[0153]

[0154] In the formula, is the solved coefficient of each term of the chaotic expansion.

[0155] In step 7), the steps for calculating the reliability index of the integrated electric-hydrogen energy system include:

[0156] 7.1) Project Y(ξ) onto the orthogonal polynomial Φ1 using the Galerkin projection method to obtain:

[0157]

[0158] In the formula, E(·) is the expectation operation, Φ1(ξ) is the first expansion term of the polynomial, c κ is the coefficient of the κth expansion term Φ κ ;

[0159] 7.2) Calculate the expectation μ of the response function Y(ξ) as the reliability index of the integrated electric-hydrogen energy system;

[0160] The expectation μ is as follows:

[0161]

[0162] Example 2:

[0163] The method for evaluating the reliability of an integrated electric-hydrogen energy system based on data-driven polynomial chaotic expansion includes the following steps:

[0164] 1) Read the basic parameters of the integrated electric-hydrogen energy system;

[0165] 2) Based on the historical data of the mixed random variables in the basic parameters of the integrated electric-hydrogen energy system, construct an orthogonal polynomial basis based on multiple moments;

[0166] 3) Construct a chaotic polynomial surrogate model based on the linear combination of orthogonal polynomials;

[0167] 4) Select collocation points based on the linear independence principle, and calculate the optimal load reduction of the integrated electric-hydrogen energy system according to the sample values of the collocation points;

[0168] 5) The mixed random variable historical data and the optimal load reduction amount of the electric-hydrogen integrated energy system are taken as the input and output of the chaotic polynomial surrogate model, and the least square method is used to solve the coefficients of the chaotic polynomial surrogate model;

[0169] 6) The current random variable is obtained and input into the chaotic polynomial surrogate model to calculate the optimal load reduction amount of the current electric-hydrogen integrated energy system;

[0170] 7) The reliability index of the electric-hydrogen integrated energy system is calculated.

[0171] Embodiment 3:

[0172] The electric-hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion has the same technical content as embodiment 2, and further, the electric-hydrogen integrated energy system basic parameters include electric-hydrogen integrated energy system parameters, device element reliability parameters, and mixed random variable historical data set.

[0173] Embodiment 4:

[0174] The electric-hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion has the same technical content as any one of embodiments 2-3, and further, the hydrogen integrated energy system parameters include power system power line resistance reactance parameters, generator output upper and lower limit parameters, wind turbine and photovoltaic unit parameters, line topology connection parameters, hydrogen energy system electrolyzer, fuel cell, hydrogen storage tank, hydrogen refueling machine parameters, and electric load, hydrogen load data;

[0175] The device element reliability parameters include the failure rate and repair rate of the generator set, the line, the electrolyzer, the fuel cell, the hydrogen storage tank, and the hydrogen refueling machine;

[0176] The mixed random variable historical data set includes wind turbine and photovoltaic unit output historical data, and generator set, line, electrolyzer, fuel cell device fault-operation historical data.

[0177] Embodiment 5:

[0178] The electric-hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion has the same technical content as any one of embodiments 2-4, and further, in step 2), the step of constructing the orthogonal polynomial basis based on the multiple moments includes:

[0179] 2.1) Construct a multi-dimensional PCE model, that is:

[0180]

[0181] In the formula, c κ is the κth expansion term Φ κcoefficient of the i-th one-dimensional polynomial; M is the number of terms of the random response Y; M=(H+N)! / (H!N!); Φ κ is the full tensor product of one-dimensional polynomials; κ=1,2,…,M;

[0182] 2.2) Construct the orthogonal polynomial basis based on the multi-order moments using the multi-dimensional PCE model, i.e.,

[0183]

[0184] where, α κ,i is the order of the i-th one-dimensional polynomial.

[0185] Embodiment 6:

[0186] The method for reliability evaluation of an electric-hydrogen integrated energy system based on a data-driven polynomial chaos expansion, the technical content of which is the same as that of any one of embodiments 2-5, further, in step 3), the steps for constructing the chaos polynomial surrogate model include:

[0187] 3.1) Convert the historical data set of the mixed random variables in the basic parameters of the electric-hydrogen integrated energy system into a multi-order moment form, and construct a one-dimensional orthogonal polynomial, i.e.,

[0188]

[0189] where, ξ i κ is the historical data of the mixed random variables; l=0,1,…,H;

[0190] where, 0~l order arbitrary polynomial basis is as follows:

[0191]

[0192] where, μ k,i is the k-th origin moment of the random variable; M p is the sampling number of the random variable.

[0193] 3.2) Normalize the one-dimensional orthogonal polynomial to construct a chaos polynomial surrogate model i.e.,

[0194]

[0195] where, ||·|| represents the two-norm of the polynomial.

[0196] Embodiment 7:

[0197] The technical content is the same as any one of embodiments 2-6, further, in step 4), the step of calculating the optimal load shedding amount of the electric-hydrogen integrated energy system according to the sample values of the collocation points comprises:

[0198] 4.1) Calculate the third-order root of the one-dimensional orthogonal polynomial ;

[0199] 4.2) Randomly combine "0" and the third-order root of the one-dimensional orthogonal polynomial to obtain an initial collocation point set Ω IC ;

[0200] 4.3) Select the same number of collocation points as the number M of undetermined coefficients from the set Ω IC in turn to form a collocation point combination Ω C ;

[0201] 4.4) Calculate the coefficient matrix Φ of the collocation point combination Ω C , that is:

[0202]

[0203] 4.4) Record the rank of the coefficient matrix Φ as R Φ , if R Φ =M, it is considered that the current collocation point combination Ω C is the optimal collocation point combination, the collocation point selection operation is ended, and step 4.5) is entered; otherwise, remove M-R Φ linearly dependent collocation points in the current collocation point combination Ω C , select M-R Φ collocation points from the unselected collocation points in the set Ω IC , write the selected M-R Φ collocation points into the collocation point combination Ω C to form a new collocation point combination Ω C , and return to step 4.4);

[0204] 4.5) Construct an optimal load shedding model; input the sample values of the selected collocation points into the optimal load shedding model in turn to obtain the optimal load shedding amount of the electric-hydrogen integrated energy system corresponding to each collocation point.

[0205] Embodiment 8:

[0206] The technical content is the same as any one of embodiments 2-7, further, the objective function f of the optimal load shedding model is as follows:

[0207]

[0208] In the formula, cDG Penalty cost of units for curtailment of wind and solar power; Ω WG , Ω PV are respectively the set of wind-solar unit grid-connected nodes; ΔP WG,it , ΔP PV,it are respectively the curtailment of wind and solar power; T is the scheduling period; n is the number of nodes; c p is the penalty cost of unit reduction of electrical load, L p,it is the reduction of electrical load power; c q is the penalty cost of unit reduction of hydrogen load, L q,it is the reduction of hydrogen load power. Subscripts i and t respectively represent node i and time t; Δt is the time step.

[0209] Example 9:

[0210] The reliability evaluation method of the electric-hydrogen integrated energy system based on data-driven polynomial chaos expansion, the technical content is the same as any one of embodiments 2-8, further, the constraint conditions of the optimal load reduction model include the electrical power balance constraint, the hydrogen balance constraint, the electrical-hydrogen load reduction constraint, the operation constraint of the relationship between the hydrogen production of the electrolyzer and the consumption power, the operation constraint of the relationship between the hydrogen consumption of the fuel cell and the output power;

[0211] The electrical power balance constraint is as follows:

[0212] P GEN,it + P WG,it + P PV,it + P FC,it = P EL,it + P load,it - L p,it + Bθ i,t (10)

[0213] In the formula, P GEN,it is the active power output by the generator set; P WG,it , P PV,it are respectively the active power actually output by the wind-solar unit; P FC,it , P EL,it are respectively the active power output by the fuel cell and the active power consumed by the electrolyzer; P load,it , L p,it are respectively the active load and the load reduction power; B is the node susceptance matrix for direct current flow calculation, and θ is the node voltage phase angle.

[0214] The hydrogen balance constraint is as follows:

[0215] Q DP,in,it = Q EL,it - Q HT,in,it + Q HT,out,it - Q FC,it (11)

[0216] Q DP,out,it = Q load,it -L q,it (12)

[0217] wherein Q DP,in,it , Q DP,out,it are the input and output hydrogen amount of hydrogenation unit; Q EL,it , Q FC,it are the hydrogen production amount of electrolyzer and hydrogen consumption amount of fuel cell; Q HT,in,it , Q HT,out,it are the hydrogen charging amount and hydrogen discharging amount of hydrogen storage tank; Q load,it , L q,it are the demand amount and reduction amount of hydrogen load, respectively.

[0218] The hydrogen load reduction constraint is shown as follows:

[0219] 0 ≤ L p,it ≤ P load,it (13)

[0220] 0 ≤ L q,it ≤ Q load,it (14)

[0221] The operation constraint of hydrogen production amount and power consumption of electrolyzer is shown as follows:

[0222]

[0223] μ EL,it P EL,min ≤ P EL,it ≤ μ EL,it P EL,max (17)

[0224] wherein ρ is the standard density of hydrogen; μ EL,it is the start-stop state variable of electrolyzer, which is 1 if in start state and 0 if in stop state; P EL,max and P EL,min are the upper and lower limits of power consumption of electrolyzer, respectively; is a relational function; m EL,t is the hydrogen production amount of electrolyzer.

[0225] The operation constraint of hydrogen consumption amount and output power of fuel cell is shown as follows:

[0226] Q FC,it = ρ H2 m FC,it (18)

[0227] P FC,it = g(mFC,it (19)

[0228] μ FC,it Q FC,min ≤Q FC,it ≤μ FC,it Q FC,max (20)

[0229] In the formula, μ FC,it This is the start / stop state variable for the fuel cell; a value of 1 indicates the start-up state, and a value of 0 indicates the stop-up state. FC,max Q FC,min These represent the upper and lower limits of hydrogen consumption for fuel cells. The standard density of hydrogen gas is g(m). FC,it ) represents a relational function; m FC,it for.

[0230] Example 10:

[0231] The reliability assessment method for an integrated electric-hydrogen energy system based on data-driven polynomial chaotic expansion is the same as any one of Examples 2-9. Further, in step 5), the step of solving the coefficients of the chaotic polynomial surrogate model using the least squares method includes:

[0232] 5.1) Substitute the sample values ​​of the random variable and the values ​​of the response function into the following formula to solve for the coefficients of the chaotic expansion:

[0233]

[0234] 5.2) The coefficients of the chaotic expansion are calculated by minimizing the sum of squared residuals, as shown in the following equation:

[0235]

[0236] 5.3) Differentiating the coefficients of the chaotic expansion yields the coefficients of each term in the chaotic expansion, i.e.:

[0237]

[0238] In the formula, These are the coefficients of the terms in the chaotic expansion obtained by solving.

[0239] Example 11:

[0240] The reliability assessment method for the integrated electric-hydrogen energy system based on data-driven polynomial chaotic expansion is the same as any one of Examples 2-10. Further, in step 7), the steps for calculating the reliability index of the integrated electric-hydrogen energy system include:

[0241] 7.1) Projecting Y(ξ) onto the orthogonal polynomial Φ1 using the Galerkin projection method, we obtain:

[0242]

[0243] where E(·) is the expectation operation, Φ1(ξ) is the first expansion term of the polynomial, c κ is the coefficient of the κth expansion term Φ κ ;

[0244] 7.2) Calculate the expectation μ of the response function Y(ξ) as the reliability index of the integrated electricity-hydrogen energy system;

[0245] The expectation μ is shown as follows:

[0246]

[0247] Embodiment 12:

[0248] The method for evaluating the reliability of the integrated electricity-hydrogen energy system based on the data-driven polynomial chaos expansion mainly includes the following steps:

[0249] 1) Read the integrated electricity-hydrogen energy system parameters, equipment element reliability parameters, and mixed random variable historical data set information;

[0250] The integrated electricity-hydrogen energy system parameters, equipment element reliability parameters, and mixed random variable historical data set information, wherein the integrated electricity-hydrogen energy system parameters include the power line resistance reactance parameters of the power system, the upper and lower limit parameters of the generator output, the wind turbine and photovoltaic unit parameters, the line topology connection parameters, the electrolytic cell, fuel cell, hydrogen storage tank, and hydrogen dispenser parameters of the hydrogen energy system, and the electric load and hydrogen load data; the equipment element reliability parameters include the failure rate and repair rate parameters of the generator unit, line, electrolytic cell, fuel cell, hydrogen storage tank, and hydrogen dispenser; and the mixed random variable historical data set includes the wind turbine and photovoltaic unit output historical data and the generator unit, line, electrolytic cell, and fuel cell equipment failure-operation historical data.

[0251] 2) Based on the historical measured data of the mixed random variable, establish an orthogonal polynomial basis based on multiple moments, and construct a chaotic polynomial surrogate model based on the linear combination of the orthogonal polynomial;

[0252] The historical measured data of the mixed random variable is used to establish an orthogonal polynomial basis based on multiple moments, and a chaotic polynomial surrogate model is constructed based on the linear combination of the orthogonal polynomial, as shown below:

[0253] The multi-dimensional independent random variable ξ = [ξ1,..., ξ N ] is used as the input of the PCE model, and the random response Y can be represented as a multi-dimensional PCE model with H-order truncation:

[0254]

[0255] wherein c κ is the coefficient of the k-th expansion term Φ κ , M is the number of terms of the expansion Y, M=(H+N)! / (H!N!), and the k-th standard orthogonal polynomial Φ κ (k=1,2,...,M) is a one-dimensional polynomial.

[0256]

[0257] wherein α κ,i is the order of the i-th one-dimensional polynomial.

[0258] The standard orthogonal polynomial basis described in the above formula is constructed by the multiple moments of the original data. The random variable ξ i The one-dimensional orthogonal polynomial is defined as the sum of any polynomial basis p (l)κ,i (l=0,1,...,H) of 0-l order.

[0259]

[0260] Any polynomial basis p (l)κ,i (l=0,1,...,H) is obtained by solving the moment matching equation shown in the following formula:

[0261]

[0262] wherein μ k,i is the k-th central moment of the random variable.

[0263] The one-dimensional orthogonal polynomial is normalized as shown in the following formula:

[0264]

[0265] wherein ||·|| represents the two-norm of the polynomial.

[0266] 3) Select collocation points based on the linear independence principle, and calculate the optimal load reduction amount of the integrated electricity and hydrogen energy system according to the sample values of the collocation points;

[0267] The collocation points are selected based on the linear independence principle, and the optimal load reduction amount of the integrated electricity and hydrogen energy system is calculated according to the sample values of the collocation points, as shown in the following formula:

[0268] Step 1) Solve the three-order roots of the one-dimensional orthogonal polynomial to combine as the to-be-selected collocation points, and randomly combine "0" and the three-order roots of the one-dimensional orthogonal polynomial to obtain an initial collocation point set Ω IC ;

[0269] Step 2) From set Ω IC Select the same number of collocation points as the number M of undetermined coefficients to form the collocation combination Ω. C And calculate the coefficient matrix Φ;

[0270]

[0271] Step 3) Let R be the rank of the coefficient matrix Φ. Φ If R Φ =M, then the current collocation combination Ω is considered to be M. C If the optimal collocation combination is found, the collocation selection operation ends; otherwise, the current collocation combination Ω is discarded. C MR Φ ) linearly dependent collocation points, and then from the set Ω IC Select from the remaining collocation points (MR) Φ ) collocation points form a new collocation combination Ω C Repeat step 3).

[0272] Based on the selected location points, the optimal load reduction model is calculated sequentially, and the model output response Y is recorded. The optimal load reduction model is as follows:

[0273] Objective function:

[0274]

[0275] In the formula, c DG The unit penalty cost for curtailing wind and solar power; Ω WG Ω PV These are the sets of grid-connected nodes for wind and solar turbines; ΔP WG,it ΔP PV,it These represent the power of wind and solar power curtailment; T is the scheduling period; n is the number of nodes; c p To reduce penalty costs per unit of electricity load, L p,it To reduce power consumption for electrical loads; c q To reduce penalty costs per unit of hydrogen load, L q,it Power reduction for hydrogen load. Subscripts i and t represent node i and time t, respectively.

[0276] Constraints:

[0277] Electric power balance constraints:

[0278] P GEN,it +P WG,it +P PV,it +P FC,it =P EL,it +P load,it -L p,it +Bθ i,t (10)

[0279] In the formula, P GEN,it P is the active power output of the generator set. WG,it P PV,it These represent the actual active power output of the wind and solar turbine units; P FC,it P EL,it These represent the active power output of the fuel cell and the active power consumed by the electrolyzer, respectively; P load,it L p,it These represent active load and load reduction power, respectively; B is the nodal susceptance matrix used for DC power flow calculation, and θ is the nodal voltage phase angle.

[0280] Hydrogen balance constraints:

[0281] Q DP,in,it =Q EL,it -Q HT,in,it +Q HT,out,it -Q FC,it (11)

[0282] Q DP,out,it =Q load,it -L q,it (12)

[0283] In the formula, Q DP,in,it Q DP,out,it These represent the input and output hydrogen quantities of the hydrogen refueling unit, respectively; Q EL,it Q FC,it These represent the hydrogen production from the electrolyzer and the hydrogen consumption from the fuel cell, respectively; Q HT,in,it Q HT,out,it These represent the hydrogen filling and discharging volumes of the hydrogen storage tank, respectively; Q load,it L q,it These represent the demand and reduction of hydrogen load, respectively.

[0284] Constraints on reducing hydrogen load:

[0285] 0≤L p,it ≤P load,it (13)

[0286] 0≤L q,it ≤Q load,it (14)

[0287] The operational constraints on the relationship between hydrogen production and power consumption in the electrolyzer are shown below:

[0288]

[0289] μ EL,it P EL,min ≤P EL,it ≤μ EL,it P EL,max (17)

[0290] wherein ρ H2 is the standard density of hydrogen (kg / Nm 3 ) ; μ EL,it is the start-stop state variable of the electrolyzer, 1 if it is in start state, 0 if it is in stop state; P EL,max and P EL,min are the upper and lower limits of the power consumption of the electrolyzer, respectively. m EL,t is the hydrogen production of the electrolyzer.

[0291] The operation constraint of the hydrogen consumption and output power relationship of the fuel cell is shown as follows:

[0292] Q FC,it = ρ H2 m FC,it (18)

[0293] P FC,it = g(m FC,it ) (19)

[0294] μ FC,it Q FC,min ≤ Q FC,it ≤ μ FC,it Q FC,max (20)

[0295] wherein μ FC,it is the start-stop state variable of the fuel cell, 1 if it is in start state, 0 if it is in stop state; Q FC,max , Q FC,min are the upper and lower limits of the hydrogen consumption of the fuel cell, respectively, ρ H2 is the standard density of hydrogen (kg / Nm 3 ). g(m FC,it ) is a relationship function; m FC,it is the hydrogen consumption of the fuel cell.

[0296] 4) solving the chaos polynomial surrogate model coefficients corresponding to the input mixed random variables and the output load reduction response according to the least square method;

[0297] The chaos polynomial surrogate model coefficients corresponding to the input mixed random variables and the output load reduction response are solved according to the least square method as follows:

[0298] The sample values of the random variables and the values of the response function are substituted into the following formula to solve the chaos expansion coefficient:

[0299]

[0300] The chaos expansion coefficient is calculated by minimizing the residual sum of squares as follows:

[0301]

[0302] The least square solution is obtained by derivation of the above formula as follows:

[0303]

[0304] wherein, is the coefficient of each term of the chaotic expansion.

[0305] 5) Based on the constructed chaotic polynomial surrogate model, the statistical moment information of the output random variable is calculated, and the reliability index of the electric hydrogen integrated energy system is calculated.

[0306] Based on the constructed chaotic polynomial surrogate model, the statistical moment information of the output random variable is calculated, and the reliability index of the electric hydrogen integrated energy system is calculated, as shown below:

[0307] The Galerkin projection method is used to project Y(ξ) onto the orthogonal polynomial Φ1, and according to the definition of inner product and the orthogonality of the orthogonal polynomial, the following formula can be obtained:

[0308]

[0309] Then the expectation μ of the response function Y(ξ) can be calculated by the following formula:

[0310]

[0311] The expectation μ of the response function Y(ξ) of the chaotic polynomial surrogate model is the expected electricity not supplied (EENS) or the expected hydrogen not supplied (EHNS) of the electric hydrogen integrated energy system.

[0312] Example 13:

[0313] Referring to Fig. 2 A simulation example of an electric hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion is provided, which is based on the IEEE-RTS79 system, and the electric hydrogen integrated energy system composed of hydrogen energy systems for storage and use at nodes 5, 8, 17, 18, and 20 is taken as an example to verify the feasibility and effectiveness of the proposed electric hydrogen integrated energy system reliability evaluation method based on data-driven polynomial chaos expansion. The following four methods are used for comparative analysis to verify the advantages of the proposed method:

[0314] Method 1: Monte Carlo simulation method;

[0315] Method 2: Monte Carlo simulation method based on Latin hypercube sampling;

[0316] Method 3: Chaos polynomial expansion method (PCE method);

[0317] Method 4: proposed data-driven chaos polynomial expansion method (DPCE method).

[0318] The load curtailment expectation value calculated by Method 1 is taken as a comparison benchmark to verify the accuracy and speed of the method in this paper. The calculation results and solving speed are shown in Tables 1 and 2.

[0319] Table 1 Comparison of system power / hydrogen reliability evaluation results of different methods

[0320]

[0321] Table 2 Comparison of evaluation speed of different methods

[0322]

[0323] In Method 1, the load curtailment energy expectation variance coefficient is used as the convergence criterion, and when 2x105 sampling calculations are performed, Method 1 achieves convergence. Method 2 uses the Latin hypercube sampling method to accelerate the convergence speed of the Monte Carlo simulation method while ensuring the accuracy of the solution. However, the essence of Method 2 is still the same as Method 1, which is to achieve reliability evaluation through a large number of sampling calculations.

[0324] In contrast, the proxy model construction method based on the chaos polynomial expansion method can complete the calculation of the load curtailment expectation value faster, such as Method 3 and Method 4. However, Method 3 cannot directly use historical data for evaluation calculation and cannot handle mixed random variables, so it is not applicable. Method 4 is the data-driven chaos polynomial expansion method proposed in this paper, which handles random variables including the output of photovoltaic and wind power units connected to nodes 5, 8, 17, 18, and 20, as well as the random failure of electrolytic cells, fuel cells, and other equipment, and the random failure of 33 generator units and 38 lines in the IEEE-RTS79 system. This method can handle both continuous and discrete random variables and does not require prior knowledge of the probability distribution information of the random variables. As can be seen from the results in Table 1, the data-driven chaos polynomial expansion method proposed in this paper significantly improves the calculation speed compared to Method 1 and Method 2 within 10% of the solution accuracy, because the data-driven chaos polynomial expansion method only needs to extract part of the distribution point combination (4000 groups in this paper), and the undetermined coefficients of the data-driven chaos polynomial expansion method are calculated, which can calculate the expectation of the electric and hydrogen load curtailment value. This proxy model-based method significantly reduces the number of times the optimal load curtailment model is solved, thereby saving calculation time.

Claims

1. A method for reliability evaluation of an integrated electricity-hydrogen energy system based on data-driven polynomial chaos expansion, characterized in that , comprising the following steps: 1) reading basic parameters of the electric-hydrogen integrated energy system; 2) constructing an orthogonal polynomial basis based on multiple moments based on historical data of mixed random variables in the basic parameters of the electric-hydrogen integrated energy system; 3) constructing a chaotic polynomial surrogate model based on a linear combination of orthogonal polynomials; 4) selecting collocation points based on the linear independence principle, and calculating the optimal load shedding amount of the electric-hydrogen integrated energy system according to the sample values of the collocation points; 5) taking the historical data of the mixed random variables and the optimal load shedding amount of the electric-hydrogen integrated energy system as the input and output of the chaotic polynomial surrogate model, and solving the coefficients of the chaotic polynomial surrogate model by the least square method; 6) obtaining the current random variable and inputting it into the chaotic polynomial surrogate model to calculate the current optimal load shedding amount of the electric-hydrogen integrated energy system; 7) calculating the reliability index of the electric-hydrogen integrated energy system; In step 2), the step of constructing an orthogonal polynomial basis based on multiple moments includes: 2.1) constructing a multi-dimensional PCE model, that is: where c κ is the coefficient of the κth expansion term Φ κ ; M is the number of terms of the random response Y; M = (H + N)! / (H!N!); Φ κ is the full tensor product of one-dimensional polynomials; κ = 1, 2, …, M; ξ1, ξ2, …, ξ N are random variables; N is the number of random variables; and H is the order. 2.2) constructing an orthogonal polynomial basis based on multiple moments using the multi-dimensional PCE model, that is: In the formula, α κ,i is the order of the i-th one-dimensional polynomial; In step 4), the step of calculating the optimal load shedding amount of the electric-hydrogen integrated energy system according to the sample values of the collocation points includes: 4.1) Calculate the cubic roots of the one-dimensional orthogonal polynomials of order 2. One-dimensional orthogonal polynomials P i (l) (ξ i ) are as follows: wherein ξ i κ is the history data of the mixed random variable; l = 0, 1,..., H; is an arbitrary polynomial basis of order 0~l; 4.2) Randomly combine "0" and the roots of the one-dimensional orthogonal polynomial of the third order to obtain the initial collocation point set Ω IC ;​ 4.3) from the set Ω IC Selecting in turn the same number M of collocation points as the number of coefficients to be determined, forming the collocation point combination Ω C ; 4.4) Calculate the combination of points Ω C the coefficient matrix Φ, i.e.: 4.4) Consider the rank of the coefficient matrix Φ as R Φ , if R Φ = M, consider the current combination of collocation points Ω C as the optimal combination of collocation points, the collocation point selection operation ends, and enter step 4.5); otherwise, eliminate M-R Φ linearly dependent collocation points in the current combination of collocation points Ω C , select M-R Φ collocation points from the unselected collocation points in the set Ω IC , write the selected M-R Φ collocation points into the combination of collocation points Ω C , form a new combination of collocation points Ω C , and return to step 4.4); 4.5) constructing an optimal load shedding model; sequentially inputting the sample values of the selected collocation points into the optimal load shedding model to obtain the optimal load shedding amount of the electric-hydrogen integrated energy system corresponding to each collocation point; In step 7), the step of calculating the reliability index of the electric-hydrogen integrated energy system includes: 7.1) projecting Y(ξ) onto the orthogonal polynomial Φ1 using the Galerkin projection method to obtain: where E(·) is the expectation operation, Φ1(ξ) is the first expansion term of the polynomial, c κ is the coefficient of the κth expansion term Φ κ . 7.2) calculating the expectation μ of the response function Y(ξ) as the reliability index of the electric-hydrogen integrated energy system; The expectation μ is as follows:

2. The data-driven polynomial chaos expansion based reliability assessment method for an integrated power-hydrogen energy system according to claim 1, wherein The basic parameters of the electric-hydrogen integrated energy system include electric-hydrogen integrated energy system parameters, equipment element reliability parameters, and mixed random variable historical data sets.

3. The data-driven polynomial chaos expansion based reliability assessment method for an integrated power-hydrogen system according to claim 2, wherein The electric-hydrogen integrated energy system parameters include power line resistance and reactance parameters of the power system, generator output upper and lower limit parameters, wind turbine and photovoltaic unit parameters, line topology connection parameters, electrolyzer, fuel cell, hydrogen storage tank, and hydrogen dispenser parameters of the hydrogen energy system, and electric load and hydrogen load data; The equipment element reliability parameters include failure rates and repair rates of the generator set, line, electrolyzer, fuel cell, hydrogen storage tank, and hydrogen dispenser; The mixed random variable historical data set includes wind turbine and photovoltaic unit output historical data, and generator set, line, electrolyzer, and fuel cell equipment failure-operation historical data.

4. The method of claim 1, wherein, In step 3), the step of constructing a chaotic polynomial surrogate model includes: 3.1) Convert the historical dataset of mixed random variables in the basic parameters of the integrated energy system of electric hydrogen into the form of multiple order moments, and construct one-dimensional orthogonal polynomials P i (l) (ξ i ), that is: where ξ i κ is the mixed random variable history data; l = 0, 1,..., H; where 0th~lth order arbitrary polynomial basis As shown below: wherein μ k,i is the kth order central moment of the random variable; M p is the number of samples of the random variable; 3.2) Normalizing the one-dimensional orthogonal polynomials to construct the chaotic polynomial surrogate model i.e. In the formula, ||·|| represents the two-norm of the polynomial.

5. The data-driven polynomial chaos expansion based reliability assessment method for integrated electrical and hydrogen energy systems according to claim 1, wherein, The objective function f of the optimal load shedding model is as follows: In the formula, c DG is the unit penalty cost of abandoned wind and light; Ω WG , Ω PV are respectively the wind and light unit penalty cost; ΔP WG,it , ΔP PV,it are respectively the abandoned wind and light power; T is the scheduling period; n is the node number; c p is the unit penalty cost of reduced electrical load, L p,it is the reduced electrical load power; c q is the unit penalty cost of reduced hydrogen load, L q,it is the reduced hydrogen load power; the subscripts i and t respectively represent the node i and the time t; and Δt is the time step.

6. The data-driven polynomial chaos expansion based reliability assessment method for an integrated power-hydrogen energy system according to claim 5, wherein, The constraint conditions of the optimal load shedding model include electric power balance constraints, hydrogen balance constraints, electric-hydrogen load shedding constraints, operating constraints of the relationship between hydrogen production and power consumption of the electrolyzer, and operating constraints of the relationship between hydrogen consumption and output power of the fuel cell; The electric power balance constraint is as follows: P GEN,it +P WG,it +P PV,it +P FC,it =P EL,it +P load,it -L p,it +Bθ i,t (10) where P GEN,it is the active power output of the generator set; P WG,it , P PV,it are the active power output of the wind-solar generator set respectively; P FC,it , P EL,it are the active power output of the fuel cell and the active power consumed by the electrolyzer respectively; P load,it , L p,it are the active load and the load curtailment power respectively; B is the nodal susceptance matrix for DC power flow calculation, θ i,t is the nodal voltage phase angle; The hydrogen balance constraint is as follows: Q DP,in,it = Q EL,it - Q HT,in,it + Q HT,out,it - Q FC,it (11) Q DP,out,it = Q load,it - L q,it (12) In the formula, Q DP,in,it , Q DP,out,it are the hydrogen input and output of the hydrogenation unit respectively; Q EL,it , Q FC,it are the hydrogen production of the electrolytic cell and the hydrogen consumption of the fuel cell respectively; Q HT,in,it , Q HT,out,it are the hydrogen charging and discharging of the hydrogen storage tank respectively; Q load,it , L q,it are the demand and reduction of hydrogen load respectively; The electric-hydrogen load shedding constraint is as follows: 0 < L p,it ≤ P load,it (13) 0 < L q,it ≤ Q load,it (14) The operation constraint of the relationship between the hydrogen production of the electrolytic cell and the consumed power is shown as follows: m EL,it = f sEL,it (P EL,it ) (16) μ EL,it P EL,min ≤P EL,it ≤μ EL,it P EL,max (17) wherein is the standard density of hydrogen; μ EL,it is the start-stop state variable of the electrolyzer, which is 1 if it is in the start state and 0 if it is in the stop state; P EL,max and P EL,min are the upper and lower limits of the power consumption of the electrolyzer, respectively; is a relationship function; m EL,t is the hydrogen production of the electrolyzer; The operation constraint of the relationship between the hydrogen consumption of the fuel cell and the output power is shown as follows: Q FC,it = p H2 m FC,it (18) P FC,it = g(m FC,it ) (19) μ FC,it Q FC,min ≤Q FC,it ≤μ FC,it Q FC,max (20) In the formula, μ FC,it is a start-stop state variable of the fuel cell, and if it is 1, it indicates a start state, and if it is 0, it indicates a stop state; Q FC,max , Q FC,min are respectively upper and lower limits of the hydrogen consumption of the fuel cell, is the standard density of hydrogen; g(m FC,it ) is a relationship function; m FC,it is the hydrogen consumption of the fuel cell.

7. The method of claim 1, wherein, In step 5), the step of solving each coefficient of the chaotic polynomial surrogate model by using the least square method comprises: 5.1) substituting the sample values of the random variables and the values of the response function into the following formula to solve the coefficients of the chaotic expansion: 5.2) calculating the coefficients of the chaotic expansion by residual sum of squares minimization, as shown in the following formula: In the formula, J(C) is the residual sum of squares; 5.3) deriving the coefficients of the chaotic expansion to obtain the coefficients of each term of the chaotic expansion, that is: wherein are the coefficients of the sought chaotic expansion.

Images