Two-stage probabilistic reserve optimization method considering renewable energy reserve and frequency security

Abstract

The present invention discloses a two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety, specifically: adopting a two-stage optimization process method, optimizing the unit combination considering conventional operation constraints and EENS constraints in the first stage; the second stage optimization is based on the optimization result of the first stage, that is, the start-stop plan of the synchronous generator in each scheduling time period, the synchronous generator and the new energy reserve, considering the nonlinearity of the output power limit of each generator set, optimizing the virtual inertia and droop control coefficient of the new energy to meet the frequency safety constraint; through the iteration of the two-stage optimization process, the start-stop state of the synchronous generator in each scheduling time period, the synchronous generator and the new energy reserve, and the virtual inertia and droop control coefficient of the new energy participating in the frequency regulation are optimized. Compared with the existing frequency safety constraint unit combination method, the present invention has better economy.

Description

Technical Field

[0001] The present invention belongs to the field of power system optimization and dispatching, and in particular relates to a two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety. Background Art

[0002] Considering the risk of load loss in the unit combination optimization model requires calculating EENS, and calculating EENS requires calculating the prediction error of the net load. In current research, the probability distribution of prediction errors of new energy and load are both considered to be normal distributions, so the expected value and standard deviation of the probability distribution of net load prediction error are obtained by adding their respective expected values ​​and standard deviations. However, new energy and load prediction errors such as wind power and photovoltaic power usually obey different probability distributions, and the distribution parameters of the net load prediction error cannot be directly obtained by adding the distribution parameters of each prediction error.

[0003] In low-inertia power systems with high penetration of renewable energy, there are frequency safety risks even if the total reserve is sufficient. Existing studies have considered frequency response models in unit combinations to calculate frequency dynamic values, so that the optimization results meet frequency safety constraints. The maximum power limit of the generator causes output power overshoot, which triggers nonlinearity in the frequency response model. Some studies do not consider the nonlinearity of the maximum power limit of each generator set, while others limit the output power of the generator set so that it does not reach the maximum power limit during the frequency dynamic process, thereby avoiding nonlinearity caused by overshoot. However, in the dynamic process, limiting the output power size to prevent overshoot will narrow the feasible solution space of the optimization problem and usually require more reserves, thereby increasing operating costs.

[0004] In addition, frequency regulation coefficients such as virtual inertia and droop control coefficients of renewable energy are usually fixed. Only a few literatures consider several sets of optional virtual inertia or droop control coefficients in unit commitment to optimize dynamic frequency. In general, existing papers rarely explore the potential of flexible virtual inertia and droop control coefficients of renewable energy in unit commitment optimization considering frequency dynamic safety. Summary of the invention

[0005] In view of the above problems, the present invention provides a two-stage probabilistic reserve optimization method taking into account new energy reserve and frequency safety.

[0006] A two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety of the present invention is characterized by comprising the following steps:

[0007] Step 1: Establish the first-stage unit commitment probability optimization model, including the unit commitment optimization objective function considering new energy backup, basic operation safety constraints and EENS (Expected energy not supplied) constraints.

[0008] Step 2: Establish the second-stage nonlinear optimization model, transform the frequency safety constraint into the optimization objective function through the penalty function, and take into account the nonlinear frequency response model of the output limit.

[0009] Step 3: Initially set or iteratively update the limit values ​​of EENS and new energy reserve ratio, perform the first-stage unit combination optimization, and solve the start-stop status of synchronous generators in each scheduling period, as well as synchronous generators and new energy reserves.

[0010] Step 4: Based on the results of the first-stage unit combination optimization, the second-stage optimization is carried out to solve the virtual inertia and droop control coefficients of the new energy participating in frequency regulation in each scheduling period.

[0011] Step 5: Check whether each scheduling period meets the post-disturbance dynamic frequency safety constraint; if not, return to step 3, update the limit of the EENS constraint or the new energy reserve ratio constraint and iterate; otherwise, end the optimization and obtain the start-stop status of the synchronous generator, the synchronous generator and the new energy reserve, as well as the virtual inertia and droop control coefficient of the new energy participating in the frequency regulation in each scheduling period.

[0012] Furthermore, step 1 is specifically as follows:

[0013] The optimization objective of unit combination considering new energy backup is:

[0014]

[0015] In the formula, is the operating cost of the synchronous machine and new energy to be optimized; C ss , C so , and C wind They are the start-up and shutdown costs of the synchronous machine, the operation and standby costs of the synchronous machine, and the operation and standby costs of the new energy, which are calculated by the following equation:

[0016]

[0017] In the formula, C up and C down is the start-stop cost of the synchronous generator; δ i,t and z i,t are binary variables to be optimized for controlling the start and stop of the synchronous machine; a i and b i is the coal consumption coefficient of the synchronous generator; C r g,i P is the price coefficient of synchronous generator standby; sc_max and P sc_min is the maximum output limit and minimum output of the synchronous generator; u i,tis a binary variable to be optimized representing the state of the synchronous generator; sc,i,t is the synchronous generator dispatch output variable to be optimized; C w_r and C w are the reserve and operation price coefficients of wind farms respectively; p w is the predicted power of the wind farm; p o w,j,t It is the output variable of the new energy dispatching to be optimized.

[0018] The basic operational safety constraints are:

[0019] P sc_min,i u i,t ≤p sc,i,t

[0020] P sc_max,i u i,t ≥p sc,i,t

[0021] δ i,t -z i,t =u i,t+1 -u i,t ,δ i,t +z i,t ≤1

[0022]

[0023] p sc,i,t -p sc,i,t-1 ≥D min,i

[0024] p sc,i,t -p sc,i,t-1 ≤D max,i

[0025]

[0026]

[0027]

[0028]

[0029]

[0030]

[0031]

[0032] In the formula, τ on i and are the minimum start-up and shutdown time of the synchronous generator respectively; Dmin , D max is the minimum and maximum value of the generator climbing speed; P w_min is the minimum output limit of the wind farm; P res,t is the minimum total spare capacity that satisfies the EENS constraint; p s,k,t For the predicted output of photovoltaic, p load is the predicted total load power, the set are the collections of all synchronous generators, wind farms and photovoltaics respectively; the collection They are synchronous machine, wind farm and photovoltaic of bus n respectively; B nm is the susceptance of the transmission line; θ n is the voltage phase angle; F min nm and F max nm are the maximum and minimum capacity of the line respectively; k d_load,t The proportion of new energy reserves.

[0033] The EENS constraints are:

[0034]

[0035] Among them, EENS i,t Calculated by the following steps:

[0036] (1) Obtain the discrete series of forecast error power and probability of wind farms, photovoltaics, and loads;

[0037]

[0038]

[0039]

[0040]

[0041]

[0042] In the formula, is the set of discrete sequence indices of the prediction error power; j Δ , k Δ , l Δ , o are the subscripts of the discrete power series of wind power, photovoltaic power, load, and net load forecast error respectively; N w , N s , N l are the total number of discrete sequence footnotes of wind power, photovoltaic power, and load forecast error power; p erW , p erS , p erLis the forecast error power discrete sequence of wind power, photovoltaic power and load; f pw,t , f ps,t , f pl,t are the prediction error probability density functions of wind power, photovoltaic power, and load, respectively; are the discrete probability sequences of wind power, photovoltaic power and load forecast errors respectively; Δp w , Δp s , Δp load is the step size of the discrete sequence of the forecast error power of wind power, photovoltaic power and load.

[0043] (2) Using sequence operation method to obtain the discrete sequence of net load forecast error probability;

[0044]

[0045] In the formula, is the discrete sequence of the calculated net load forecast error probability.

[0046] (3) Calculate the EENS considering the new energy forecast error, load forecast error, and synchronous machine "N-1" failure risk, and obtain the constraint function of the EENS;

[0047]

[0048] In the formula, f sc_out is the probability of synchronous generator failure; σ o,1 and σ o,2 is a binary variable.

[0049] Furthermore, step 2 is specifically as follows:

[0050] The objective function that transforms the frequency safety constraint into the optimization target through the penalty function is:

[0051] min{|α1+α2+α3|+Φ}

[0052] Where Φ is the penalty function; α1, α2, α3 are the differences between the frequency dynamic characteristic value and the frequency dynamic characteristic safety threshold, which are obtained by the following equation:

[0053]

[0054]

[0055]

[0056]

[0057] Φ=Ф1+Ф2+Ф3

[0058] In the formula, and They are the lowest frequency point, steady-state frequency and the safety threshold of frequency change rate; f nadir , f ss and |df / dt| max are the absolute values ​​of the lowest frequency point, steady-state frequency and frequency change rate, respectively, which are calculated by the nonlinear frequency response model taking into account the output limit; k Ф An extremely large number.

[0059] The nonlinear frequency response model includes the equivalent rotor motion equation and load damping link, the prime mover and speed regulator control link, the converter control link, and the output limiting link.

[0060] The equivalent rotor motion equation and the description equation of the load damping link are:

[0061]

[0062] Where M g is the equivalent moment of inertia constant, D is the load damping, Δf is the frequency change, P M is the sum of the mechanical power change of the synchronous generator and the output power change of the new energy source, ΔP E is the power disturbance.

[0063] The descriptive equations of the prime mover and speed governor control link are:

[0064]

[0065] Where, T Ri 、F Hi , K mi are the reheating time constant, reheating coefficient, and mechanical power gain of the i-th synchronous generator, 1 / R i is the droop control coefficient of the speed regulator of the i-th synchronous generator, p SC_out,i is the mechanical power of the i-th synchronous generator.

[0066] The description equation of the control link of the new energy converter is:

[0067]

[0068] Where M vj and 1 / R drj is the virtual inertia and droop control coefficient of the new energy source participating in frequency modulation of the jth station, T dj and T vj is the time constant of the new energy converter, D vj and K dj is the damping and gain of the new energy converter, p re_out,j is the output power of the jth renewable energy source participating in frequency modulation.

[0069] The description equation of the limiting link is:

[0070]

[0071] Where P SC_max,i is the maximum mechanical power of the ith synchronous machine, p w,j is the predicted output power of the jth renewable energy source.

[0072] Furthermore, the method for iteratively updating the limit values ​​of EENS and new energy reserve ratio in step 3 is as follows: for any scheduling period that does not meet the post-disturbance dynamic frequency safety constraint, first increase the new energy reserve ratio; when the new energy reserve ratio has reached the maximum value, reduce the EENS constraint limit value of the corresponding scheduling period.

[0073] Furthermore, in step 5, the criterion for checking whether each scheduling period satisfies the post-disturbance dynamic frequency safety constraint is:

[0074] min{|α1+α2+α3|+Φ}≤k Ф

[0075] If it holds, it is satisfied; otherwise, it is not satisfied.

[0076] The beneficial technical effects of the present invention compared with the prior art are:

[0077] 1. The present invention takes into account the nonlinearity of the maximum power limit of each generator set in the frequency response model. Compared with the commonly used method of limiting output power to avoid overshoot, it reduces the spare capacity that a single generator must provide and expands the feasible solution space, thereby achieving better economy.

[0078] 2. The present invention achieves better economy while satisfying frequency safety constraints by optimizing the proportion of new energy reserve in each scheduling time period, optimizing the virtual inertia and droop control coefficient of new energy participating in frequency regulation.

[0079] 3. The two-stage optimization method proposed in the present invention combines the above nonlinear frequency response model with the fast and flexible frequency response optimization of new energy. In addition, it is also compatible with the unit combination without considering frequency safety constraints, reducing the cost of considering frequency safety, and retaining the advantages of better economy of the general unit combination method without considering frequency safety constraints as much as possible.

[0080] 4. The present invention establishes an EENS model and introduces a sequence operation method to calculate the probability distribution of the net load forecast error, wherein the forecast error probability distributions of wind power, photovoltaic power and load can be different types of probability distributions or even non-parametric probability distributions. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 It is a flow chart of the two-stage probabilistic standby optimization method considering frequency safety constraints proposed by the present invention.

[0082] Figure 2 is a nonlinear frequency response model used to calculate frequency dynamics.

[0083] Figure 3 is the EENS value obtained by Monte Carlo simulation.

[0084] Figure 4 It is a backup for the three optimized examples.

[0085] Figure 5 The RoCoF, the lowest frequency point and the steady-state frequency of each scheduling period under a given disturbance are obtained based on the optimization results of three examples.

[0086] Figure 6 are the virtual inertia and droop control coefficients of the new energy sources participating in frequency regulation in each scheduling period optimized in Example 1.

[0087] Figure 7 It is the running cost.

[0088] Figure 8 These are the frequency curves and power responses of Example 1 and Example 2 under given power disturbances. DETAILED DESCRIPTION

[0089] The present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0090] The present invention provides a two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety, such as Figure 1 As shown, a two-stage optimization process method is adopted. In the first stage, the optimization is for the unit combination considering the conventional operation constraints and EENS constraints. The second stage optimization is based on the optimization results of the first stage, that is, the start-up and shutdown plans of synchronous generators in each scheduling time period, as well as synchronous generators and new energy reserves, considering the nonlinearity of the output power limit of each generator set, and optimizing the virtual inertia and droop control coefficient of new energy to meet the frequency safety constraints. Through the iteration of the two-stage optimization process, the start-up and shutdown plans of synchronous generators in each scheduling time period, synchronous generators and new energy reserves, as well as the virtual inertia and droop control coefficients of new energy are optimized. The specific steps are as follows:

[0091] Step 1: Establish the first-stage unit commitment probability optimization model, including the unit commitment optimization objective function considering new energy reserve, basic operation safety constraints and EENS constraints.

[0092] The optimization objective of unit combination considering new energy backup is:

[0093]

[0094] In the formula, is the operating cost of the synchronous machine and new energy to be optimized; C ss , C so , and C wind They are the start-up and shutdown costs of the synchronous machine, the operation and standby costs of the synchronous machine, and the operation and standby costs of the new energy, which are calculated by the following equation:

[0095]

[0096] In the formula, C up and C down is the start-stop cost of the synchronous generator; δ i,t and z i,t are binary variables to be optimized for controlling the start and stop of the synchronous machine; a i and b i is the coal consumption coefficient of the synchronous generator; C r g,i P is the price coefficient of synchronous generator standby; sc_max and P sc_min is the maximum output limit and minimum output of the synchronous generator; u i,t is a binary variable to be optimized representing the state of the synchronous generator; sc,i,t is the synchronous generator dispatch output variable to be optimized; C w_r and C w are the reserve and operation price coefficients of wind farms respectively; p w is the predicted power of the wind farm; p o w,j,t It is the output variable of the new energy dispatching to be optimized.

[0097] The basic operational safety constraints are:

[0098] P sc_min,i u i,t ≤p sc,i,t

[0099] P sc_max,i u i,t ≥p sc,i,t

[0100] δ i,t -z i,t =u i,t+1 -u i,t ,δ i,t +z i,t ≤1

[0101]

[0102] p sc,i,t -psc,i,t-1 ≥D min,i

[0103] p sc,i,t -p sc,i,t-1 ≤D max,i

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] In the formula, τ on i and are the minimum start-up and shutdown time of the synchronous generator respectively; D min , D max is the minimum and maximum value of the generator climbing speed; P w_min is the minimum output limit of the wind farm; P res,t is the minimum total spare capacity that satisfies the EENS constraint; p s,k,t For the predicted output of photovoltaic, p load is the predicted total load power, the set are the collections of all synchronous generators, wind farms and photovoltaics respectively; the collection They are synchronous machine, wind farm and photovoltaic of bus n respectively; B nm is the susceptance of the transmission line; θ n is the voltage phase angle; F min nm and F max nm are the maximum and minimum capacity of the line respectively; k d_load,t The proportion of new energy reserves.

[0112] The EENS constraints are:

[0113]

[0114] Among them, EENS i,t Calculated by the following steps:

[0115] (1) Obtain the discrete series of forecast error power and probability of wind farms, photovoltaics, and loads;

[0116]

[0117]

[0118] In the formula, is the set of discrete sequence indices of the prediction error power; j Δ , k Δ , l Δ , o are the subscripts of the discrete power series of wind power, photovoltaic power, load, and net load forecast error respectively; N w , N s , N l are the total number of discrete sequence footnotes of wind power, photovoltaic power, and load forecast error power; p erW , p erS , p erL is the forecast error power discrete sequence of wind power, photovoltaic power and load; f pw,t , f ps,t , f pl,t are the prediction error probability density functions of wind power, photovoltaic power, and load, respectively; are the discrete probability sequences of wind power, photovoltaic power and load forecast errors respectively; Δp w , Δp s , Δp load is the step size of the discrete sequence of the forecast error power of wind power, photovoltaic power and load.

[0119] (2) Using sequence operation method to obtain the discrete sequence of net load forecast error probability;

[0120]

[0121] In the formula, is the discrete sequence of the calculated net load forecast error probability.

[0122] (3) Calculate the EENS considering the new energy forecast error, load forecast error, and synchronous machine "N-1" failure risk, and obtain the constraint function of the EENS;

[0123]

[0124] In the formula, f sc_out is the probability of synchronous generator failure; σ o,1 and σ o,2 is a binary variable.

[0125] Step 2: Establish the second-stage nonlinear optimization model, transform the frequency safety constraint into the optimization objective function through the penalty function, and take into account the nonlinear frequency response model of the output limit.

[0126] The objective function that transforms the frequency safety constraint into the optimization target through the penalty function is:

[0127] min{|α1+α2+α3|+Φ}

[0128] Where Φ is the penalty function; α1, α2, α3 are the differences between the frequency dynamic characteristic value and the frequency dynamic characteristic safety threshold, which are obtained by the following equation:

[0129]

[0130]

[0131] Φ=Ф1+Ф2+Ф3

[0132] In the formula, and They are the lowest frequency point, steady-state frequency and the safety threshold of frequency change rate; f nadir , f ss and |df / dt| max are the absolute values ​​of the lowest frequency point, steady-state frequency and frequency change rate, respectively, which are calculated by the nonlinear frequency response model taking into account the output limit; k Ф An extremely large number.

[0133] The nonlinear frequency response model includes the equivalent rotor motion equation and load damping link, the prime mover and speed regulator control link, the converter control link, and the output limiting link.

[0134] The equivalent rotor motion equation and the description equation of the load damping link are:

[0135]

[0136] Where M g is the equivalent moment of inertia constant, D is the load damping, Δf is the frequency change, P M is the sum of the mechanical power change of the synchronous generator and the output power change of the new energy source, ΔP E is the power disturbance.

[0137] The descriptive equations of the prime mover and speed governor control link are:

[0138]

[0139] Where, T Ri 、F Hi , K mi are the reheating time constant, reheating coefficient, and mechanical power gain of the i-th synchronous generator, 1 / R i is the droop control coefficient of the speed regulator of the i-th synchronous generator, p SC_out,iis the mechanical power of the i-th synchronous generator.

[0140] The description equation of the control link of the new energy converter is:

[0141]

[0142] Where M vj and 1 / R drj is the virtual inertia and droop control coefficient of the new energy source participating in frequency modulation of the jth station, T dj and T vj is the time constant of the new energy converter, D vj and K dj is the damping and gain of the new energy converter, p re_out,j is the output power of the jth renewable energy source participating in frequency modulation.

[0143] The description equation of the limiting link is:

[0144]

[0145] Where P SC _max,i is the maximum mechanical power of the i-th synchronous machine, and pw,j is the predicted output power of the j-th new energy source.

[0146] Step 3: Initially set or iteratively update the limit values ​​of EENS and new energy reserve ratio, perform the first-stage unit combination optimization, and solve the start-stop status of synchronous generators in each scheduling period, as well as synchronous generators and new energy reserves.

[0147] The method for iteratively updating the limits of EENS and new energy reserve ratio is as follows: for any scheduling period that does not meet the post-disturbance dynamic frequency safety constraint, first increase the new energy reserve ratio; when the new energy reserve ratio has reached the maximum value, reduce the EENS constraint limit of the corresponding scheduling period.

[0148] Step 4: Based on the results of the first-stage unit combination optimization, the second-stage optimization is carried out to solve the virtual inertia and droop control coefficients of the new energy participating in frequency regulation in each scheduling period.

[0149] Step 5: Check whether each scheduling period meets the post-disturbance dynamic frequency safety constraint; if not, return to step 3, update the limit of the EENS constraint or the new energy reserve ratio constraint and iterate; otherwise, end the optimization and obtain the start-stop status of the synchronous generator, the synchronous generator and the new energy reserve, as well as the virtual inertia and droop control coefficient of the new energy participating in the frequency regulation in each scheduling period.

[0150] The criterion for checking whether each scheduling period meets the post-disturbance dynamic frequency safety constraint is:

[0151] min{|α1+α2+α3|+Φ}≤k Ф

[0152] If it holds, it is satisfied; otherwise, it is not satisfied.

[0153] Experimental verification:

[0154] The following implementation cases verify the correctness and superiority of the two-stage probabilistic standby optimization method proposed in the present invention. Three examples are provided for comparison.

[0155] Example 1: The two-stage probabilistic reserve optimization method proposed in this invention considering new energy reserve and frequency security constraints.

[0156] Example 2: Existing unit commitment method considering frequency security constraints.

[0157] Example 3: Unit commitment method without considering frequency safety constraints.

[0158] The test is conducted in the modified IEEE 6bus test system. The installed capacity of wind turbines is 190MW, and the installed capacity of photovoltaics is 50MW, connected to bus 3 and bus 5 respectively. The total installed capacity of synchronous generators is 220MW. It is assumed that all photovoltaics are absorbed, and the new energy backup is provided by the wind farm. In Example 2, 1 / R dr and M v The values ​​of are set to 12 and 5s respectively. max The initial value is set to 1MWh, k d_load,t The initial value is set to 5%.

[0159] The first stage unit combination optimization model is solved by CPLEX solver. The second stage nonlinear optimization model is solved by genetic algorithm, and the nonlinear frequency response model of the frequency dynamic eigenvalue is solved as follows: Figure 2 As shown in the figure. The initial population of the genetic algorithm is 20 chromosomes, each chromosome consists of 10×2 genes, which are used to optimize the two variables of virtual inertia of new energy and droop control coefficient involved in frequency modulation. The maximum number of generations of inheritance is 50, and the generation interval is set to 0.9. To simplify the calculation, the virtual inertia of new energy and droop control coefficient to be optimized are both integers.

[0160] First, the Monte Carlo simulation is used to verify the correctness of the proposed EENS calculation considering the net load prediction error and "N-1". Considering the probability distribution of net load error and synchronous machine failure, 10,000 power disturbance samples are generated to calculate the EENS under these 10,000 power disturbances. The EENS calculated by this Monte Carlo sampling is considered to be the true EENS. The results are shown in Figure 3 shown.

[0161] Optimize the backup of synchronous machines and new energy sources under each scheduling time period. Figure 4 A power disturbance ΔP is applied in the optimized operation mode in each scheduling time period of the three examples. E The lowest frequency point, quasi-steady-state frequency and frequency change rate of the three examples are obtained. The results are as follows Figure 5 shown. Figure 6 The frequency modulation parameters of renewable energy in each scheduling period are optimized for Example 1. The optimized operating costs of the three examples are as follows: Figure 7 shown.

[0162] from Figure 7 It can be seen that Example 3 has the lowest operating cost, but Figure 5 As shown in Figure 3, the optimization result of Example 3 cannot meet the frequency security constraint. Both Example 1 and Example 2 consider the frequency security constraint, but the cost of Example 1 is 8.71% less than that of Example 2. Therefore, the proposed two-stage probabilistic reserve optimization method achieves better economic benefits while ensuring frequency security.

[0163] like Figure 4 As shown in Figure 2, in different scheduling periods, the optimized total reserve of Example 1 and Example 3 is the same. Figure 5 It can be seen that the optimization result of Example 1 can meet the dynamic frequency safety, while Example 3 cannot. This means that even if the total reserve of the synchronous machine is sufficient to cope with power disturbances, the lowest frequency point may not be guaranteed. Optimizing the new energy reserve and the new energy virtual inertia and droop control coefficient can ensure frequency safety in a system that originally could not guarantee frequency safety under the same total reserve.

[0164] In addition, during the period of 01:00-14:00 and 19:00-24:00, the total reserve of Example 1 is less than that of Example 2, while during the period of 15:00-18:00, the total reserve of Example 1 is equal to that of Example 2. Taking the result of 16:00, a scheduling period with the same total reserve, as an example, Figure 8 The frequency response and power variation after disturbance at 16:00 for Example 1 ((a)-(e)) and Example 2 ((f)-(j)) are given. A power deficit of 26MW is applied at 2.0s. Figure 8 The dashed lines in (b)-(e) and (g)-(j) represent the maximum power of each generator set. In Example 2, synchronous generator 2 and wind farm participate in the frequency response model. Figure 8 As shown in (e), the optimization result of Example 1 causes the wind farm to overshoot. At this time, the reserve of the wind farm in Example 1 is 13.2MW. In Example 2, Figure 8As shown in (j), the wind farm did not overshoot, and the reserve was 22.8MW. In Example 2, since the output power of the wind turbines and synchronous machines involved in frequency regulation was constrained to be low enough to avoid overshoot in the optimization, more renewable energy reserves were needed. When the total reserve is the same, less renewable energy reserves means better economic performance, because the reserve cost of renewable energy is higher than that of synchronous machines, while the operating cost of renewable energy is relatively low.

[0165] In summary, the two-stage probabilistic reserve optimization method considering new energy reserve and frequency security constraints proposed in the present invention can not only ensure the dynamic frequency security of the power system, but also reduce the additional cost brought to the power system by considering frequency security constraints.

Claims

1. A two-stage probabilistic reserve optimization method considering new energy reserve and frequency security, characterized in that: The following steps are involved: Step 1: Establish the first-stage unit commitment probability optimization model, including the unit commitment optimization objective function considering new energy reserve, basic operation safety constraints and EENS constraints; The optimization objective of unit combination considering new energy backup is: Where c is the operating cost of the synchronous machine and new energy to be optimized; C ss , C so , and C wind They are the start-up and shutdown costs of the synchronous machine, the operation and standby costs of the synchronous machine, and the operation and standby costs of the new energy, which are calculated by the following equation: In the formula, C up and C down is the start-stop cost of the synchronous generator; δ i,t and z i,t are binary variables to be optimized for controlling the start and stop of the synchronous machine; a i and b i is the coal consumption coefficient of the synchronous generator; C r g,i P is the price coefficient of synchronous generator standby; sc_max and P sc_min is the maximum output limit and minimum output of the synchronous generator; u i,t is a binary variable to be optimized representing the state of the synchronous generator; sc,i,t is the synchronous generator dispatch output variable to be optimized; C w_r and C w are the reserve and operation price coefficients of wind farms respectively; p w is the predicted power of the wind farm; p o w,j,t Dispatching output variables for new energy sources to be optimized; Step 2: Establish the second-stage nonlinear optimization model, transform the frequency safety constraint into the optimization objective function through the penalty function, and take into account the nonlinear frequency response model of the output limit; The objective function that transforms the frequency safety constraint into the optimization target through the penalty function is: min{|α1+α2+α3|+Φ} Where Φ is the penalty function; α1, α2, α3 are the differences between the frequency dynamic characteristic value and the frequency dynamic characteristic safety threshold, which are obtained by the following equation: Φ=Ф1+Ф2+Ф3 In the formula, f ref nadir , f ref ss and f ref RoCoF They are the lowest frequency point, steady-state frequency and the safety threshold of frequency change rate; f nadir , f ss and |df / dt| max are the absolute values ​​of the lowest frequency point, steady-state frequency and frequency change rate, respectively, which are calculated by the nonlinear frequency response model taking into account the output limit; k Ф is a very large number; Step 3: Initially set or iteratively update the limit values ​​of EENS and new energy reserve ratio, perform the first-stage unit combination optimization, and solve the start-stop status of synchronous generators in each scheduling period, as well as synchronous generators and new energy reserves; Step 4: Based on the results of the first-stage unit combination optimization, the second-stage optimization is carried out to solve the virtual inertia and droop control coefficients of the new energy sources participating in frequency regulation in each dispatching period; Step 5: Check whether each dispatch period meets the post-disturbance dynamic frequency safety constraint; if not, return to step 3, update the limit of the EENS constraint or the renewable energy reserve ratio constraint, and then iterate; otherwise, end the optimization and obtain the start-stop status of the synchronous generator, the synchronous generator and renewable energy reserve, and the virtual inertia and droop control coefficient of the renewable energy participating in the frequency regulation in each dispatch period; The criterion for checking whether each scheduling period meets the post-disturbance dynamic frequency safety constraint is: min{|α1+α2+α3|+Φ}≤k Ф If it holds, it is satisfied; otherwise, it is not satisfied.

2. A two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety according to claim 1, characterized in that: The basic operational safety constraints are: P sc_min,i in i,t ≤p sc,i,t P sc_max,i in i,t ≥p sc,i,t δ ii,t -z i,t =u i,t+1 -u i,t ,δ i,t +z i,t ≤1 p sc,i,t -p sc,i,t-1 ≥D min,i p sc,i,t -p sc,i,t-1 ≤D max,i In the formula, τ on i and τ off i are the minimum start-up and shutdown time of the synchronous generator respectively; D min , D max is the minimum and maximum value of the generator climbing speed; P w_min is the minimum output limit of the wind farm; P res,t is the minimum total spare capacity that satisfies the EENS constraint; p s,k,t For the predicted output of photovoltaic, p load is the predicted total load power, the set are the collections of all synchronous generators, wind farms and photovoltaics respectively; the collection They are synchronous machine, wind farm and photovoltaic of bus n respectively; B nm is the susceptance of the transmission line; θ n is the voltage phase angle; F min nm and F max nm are the maximum and minimum capacity of the line respectively; k d_load,t The proportion of energy used as reserve for new energy; The EENS constraints are: Among them, EENS i,t Calculated by the following steps: (1) Obtain the discrete series of forecast error power and probability of wind farms, photovoltaics, and loads; In the formula, is the set of discrete sequence indices of the prediction error power; j Δ , k Δ , l Δ , o are the subscripts of the discrete power series of wind power, photovoltaic power, load, and net load forecast error respectively; N w , N s , N l are the total number of discrete sequence footnotes of wind power, photovoltaic power, and load forecast error power; p erW , p erS , p erL is the forecast error power discrete sequence of wind power, photovoltaic power and load; f pw,t , f ps,t , f pl,t are the prediction error probability density functions of wind power, photovoltaic power, and load, respectively; are the probability discrete sequences of wind power, photovoltaic power and load forecast errors respectively; Δp w , Δp s , Δp load is the step size of the discrete sequence of wind power, photovoltaic power and load prediction error power; (2) Using sequence operation method to obtain the discrete sequence of net load forecast error probability; In the formula, is the discrete sequence of the calculated net load forecast error probability; (3) Calculate the EENS considering the new energy forecast error, load forecast error, and synchronous machine "N-1" failure risk, and obtain the constraint function of the EENS; In the formula, f sc_out is the probability of synchronous generator failure; σ o,1 and σ o,2 is a binary variable.

3. A two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety according to claim 2, characterized in that: The nonlinear frequency response model includes an equivalent rotor motion equation and a load damping link, a prime mover and a speed regulator control link, a converter control link, and an output limit link; The equivalent rotor motion equation and the description equation of the load damping link are: Where M g is the equivalent moment of inertia constant, D is the load damping, Δf is the frequency change, P M is the sum of the mechanical power change of the synchronous generator and the output power change of the new energy source, ΔP E is the power disturbance; The descriptive equations of the prime mover and speed governor control link are: Where, T Ri 、F Hi , K mi are the reheating time constant, reheating coefficient, and mechanical power gain of the i-th synchronous generator, 1 / R i is the droop control coefficient of the speed regulator of the i-th synchronous generator, p SC_out,i is the mechanical power of the i-th synchronous generator; The description equation of the control link of the new energy converter is: Where M vj and 1 / R drj is the virtual inertia and droop control coefficient of the new energy source participating in frequency modulation of the jth station, T dj and T vj is the time constant of the new energy converter, D vj and K dj is the damping and gain of the new energy converter, p re_out,j is the output power of the jth renewable energy source participating in frequency modulation; The description equation of the limiting link is: Where P SC_max,i is the maximum mechanical power of the ith synchronous machine, p w,j is the predicted output power of the jth renewable energy source.

4. According to claim 1, a two-stage probabilistic reserve optimization method considering new energy reserve and frequency safety is characterized in that: The method for iteratively updating the limit values ​​of EENS and new energy reserve ratio in step 3 is: for any scheduling period that does not meet the post-disturbance dynamic frequency safety constraint, first increase the new energy reserve ratio; when the new energy reserve ratio has reached the maximum value, reduce the EENS constraint limit value of the corresponding scheduling period.

Images