Inverse-time overcurrent protection optimization method for inverter-type distributed power distribution network

Abstract

This invention discloses an optimization method for inverse-time overcurrent protection in inverter-type distributed energy distribution networks. Considering the switching strategy of energy storage systems and the disconnection process of inverter power sources, and using protection selectivity requirements as constraints, a multi-objective optimization model for inverse-time overcurrent protection is established with speed and sensitivity as optimization objectives. Based on this multi-objective optimization model, an adaptive inertia weight, learning factor, grid density, and Cauchy variation factor are introduced using the standard multi-objective particle swarm optimization algorithm. An optimization scheme for the inverse-time overcurrent protection settings of distribution networks containing inverter power sources is proposed. An ideal approximation method based on the analytic hierarchy process (AHP) and objective weighting method to obtain comprehensive weights is then used to solve the multi-objective optimization model for inverse-time overcurrent protection. This invention, by optimizing the inverse-time overcurrent protection settings, can effectively reduce the impact of phase-to-phase faults and inverter-type distributed power sources, better meeting the backup protection requirements of low-resistance grounded distribution networks, and has a wide range of applications.

Description

Technical Field

[0001] This invention relates to the field of power system relay protection and fault analysis technology, and more specifically, to a method for optimizing the inverse-time overcurrent protection settings of a distribution network with inverter-type distributed energy resources based on an improved multi-objective particle swarm optimization algorithm. Background Technology

[0002] Distributed power sources, such as wind power and photovoltaics, and energy storage systems can form inverter-type power sources. Due to the uncertainty of inverter-type power sources, the source-load duality, and the fact that their output characteristics are closely related to the converter control strategy, and the low-voltage ride-through capability requirement that inverter-type power sources disconnect from the grid after a certain period of time, the power flow direction and short-circuit current level will change fundamentally after a distribution network fault. This may lead to maloperation or failure to operate of conventional inverse-time overcurrent protection in distribution networks, posing challenges to the setting and coordination of protection in new distribution networks.

[0003] However, current optimization models do not address the source-load duality of inverter-based distributed energy resources (IDERs) and inverter control strategies, and protection schemes remain at the single-objective optimization stage, exhibiting weakened speed and sensitivity, as well as selectivity mismatch issues. Therefore, further research is needed on inverse-time overcurrent protection optimization models for distribution networks that consider the impact of inverter-based power sources and satisfy multiple objectives to address the weakened speed and sensitivity, and selectivity mismatch problems present in existing optimization models. Summary of the Invention

[0004] The purpose of this invention is to overcome the problems of weakened speed and sensitivity and selective mismatch in optimization models with action time as the single objective, and to provide an inverse time-limited overcurrent protection setpoint optimization method based on an improved multi-objective particle swarm optimization algorithm with adaptive inertia weight, learning factor, adaptive grid density and Cauchy mutation operator.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] An optimization method for inverse-time overcurrent protection in inverter-type distributed energy distribution networks includes the following steps:

[0007] S1. Establish an optimization model for inverse time overcurrent protection, determine the objective function and constraints for optimization of inverse time overcurrent protection, and use the multi-objective particle swarm optimization algorithm to solve the objective parameters of inverse time overcurrent protection;

[0008] S2. Based on the constraints, construct the initial population for the multi-objective particle swarm algorithm, calculate the fitness of the population particles, update the inertia weight and learning factor, and calculate the fitness value of the objective parameter.

[0009] S3. Update the population, calculate the weighted fitness of the population particles, calculate the adaptive grid density and selection probability, and perform mutation operations with Cauchy operators on the population particles.

[0010] S4. Compare whether the number of particles in the external archive exceeds the maximum allowed value. If it does, update the external archive and the global optimal solution. If it does not exceed the maximum allowed value, update the global optimal solution directly.

[0011] S5. Compare whether the current iteration count exceeds the maximum iteration count. If it does not exceed the maximum iteration count, continue to update the iteration. If it does exceed the maximum iteration count, end the iteration and obtain the Pareto optimal front set.

[0012] S6. Calculate the comprehensive weight using the analytic hierarchy process and the objective weighting method, and calculate the ideal solution using the approximation ideal solution sorting method to obtain the required inverse time protection setting.

[0013] Furthermore, in step S1, the iteration parameters of the multi-objective particle swarm optimization algorithm include the number of external archive particles, the number of Pareto solution sets, and the maximum number of iterations.

[0014] Furthermore, in step S1, the objective function of the inverse-time overcurrent protection optimization model is:

[0015]

[0016]

[0017] In the formula, N p Number of primary protections; N b Number of backup protections; The action time of the i-th main protection when the i-th protected segment fails; Let N be the action time of the k-th backup protection when the i-th protected section fails; obj1 is the first optimization objective; N bc K represents the total number of distribution network protection devices BC; senj is the sensitivity coefficient of the j-th main protection; obj2 is the second optimization objective.

[0018] Furthermore, the constraints of the inverse time overcurrent protection optimization model are:

[0019]

[0020] In the formula, ΔT is the time difference of the action; T p The time coefficient for inverse time overcurrent protection; I op The starting current for inverse-time overcurrent protection; t CTImax Let ΔT be the upper limit, and t be the lower limit. CTImin This is the lower limit of ΔT; T pmax For T p The upper limit of T pminFor T p The lower limit; I Lmax For I op The upper limit of I Lmin For I op The lower limit.

[0021] Furthermore, in step S2, if the inverter-type distributed energy source experiences a grid disconnection process, the action time of the inverse-time overcurrent protection is updated through iterative calculation as follows:

[0022]

[0023] In the formula, The time of the i-th protection action when the i-th protected segment fails before disconnection from the network; N represents the time of the i-th protection action when the i-th protected segment fails before all inverter-type distributed energy sources are disconnected from the grid; f N represents the number of inverter-type distributed energy sources that experience grid disconnection processes; IB The number of inverter-type distributed energy sources; The time of the i-th protection action when the i-th protected section fails before the j-th inverter-type distributed energy source is disconnected from the grid;

[0024] The running time of the h-th inverter-type distributed power source after the j-th inverter-type distributed power source is disconnected from the grid; The running time of the h-th inverter-type distributed power source before the j-th inverter-type distributed power source is disconnected from the grid; The time of the i-th protection action when the i-th protected section fails after the j-th inverter-type distributed energy source is disconnected from the grid.

[0025] Furthermore, in step S2, the inertia weights are updated using the following formula:

[0026]

[0027] In the formula, w min The minimum value of the initial inertia weight; w max Δf is the maximum value of the initial inertia weight; Δf is the average weight variable. Let be the fitness value of the p-th particle in the g-th iteration; is the average fitness of the population at the g-th iteration; Let be the inertial weight of the p-th particle in the g-th iteration;

[0028] The learning factor is updated using the following formula:

[0029]

[0030] In the formula, This represents the minimum value of the individual learning factor; G represents the maximum value of the individual learning factor; G is the maximum number of iterations. This represents the individual learning factor at the g-th iteration. This represents the minimum value of the social learning factor. This represents the maximum value of the social learning factor. Let g be the social learning factor at the g-th iteration;

[0031] This represents the minimum fitness of the population at the g-th iteration. Let p be the position variable of the p-th particle in the g-th iteration; Let p be the position variable of the p-th particle in the (g+1)-th iteration; Let be the velocity variable of the p-th particle in the (g+1)-th iteration; This is the local optimal solution at the g-th iteration; This is the globally optimal solution at the g-th iteration.

[0032] Further, in step S3, the population is updated, and the weighted fitness of the population particles is calculated using the following formula:

[0033]

[0034] In the formula, ζ λ Let λ be the fitness coefficient of the λ-th objective. Let λ be the fitness of the p-th particle and the λ-th target at the g-th iteration. Let be the weighted fitness of the p-th particle in the g-th iteration.

[0035] Further, in step S3, the adaptive mesh density and selection probability are calculated using the following formula:

[0036]

[0037] In the formula, N q N represents the number of particles in the q-th grid. g N represents the number of grids in the external archive. R The number of particles in the external archive; [] is the round-up function; Let g be the adaptive grid density of the q-th grid in the g-th iteration; Let q be the selection probability of the q-th grid in the g-th iteration; This represents the minimum fitness value of the λ-th objective at the g-th iteration. This represents the maximum fitness value of the λ-th objective during the g-th iteration.

[0038] Furthermore, in step S4, a mutation operation with a Cauchy operator is performed on the population particles, as shown in the formula:

[0039]

[0040]

[0041] In the formula, C e For the standard Cauchy mutation operator; C r P is a random number within the interval [0,1]. S η is the mutation probability; η is the adjustment parameter; Let be the position variable of the p-th particle after the mutation strategy in the g-th iteration.

[0042] Further, in step S6, the comprehensive weights are calculated using the analytic hierarchy process (AHP) and the objective weighting method, and the ideal solution is calculated using the approximation of the ideal solution ranking method, with the following formula:

[0043]

[0044] In the formula, N1 is the dimension of the subjective weight matrix Z1; ρ iλ ρ is the element in the i-th row of the λ-th objective in the subjective weight matrix Z1; ki ω is the element in the k-th row and i-th column of the subjective weight matrix Z1; AHP,λ Let λ be the weight of the analytic hierarchy process for the λ-th objective.

[0045]

[0046] In the formula, N p The subjective weight matrix Z p Dimension; S λ Let r be the Pareto standard deviation of the λth objective; iλ ω is the correlation coefficient of the i-th Pareto solution set for the λ-th objective; CRI,λ The objective weighting method weights for the λ-th objective;

[0047]

[0048] In the formula, D Tp ω is the approximate ideal solution evaluation value of the Pareto solution set; λ The weight of the λth objective; z p x is the average solution of the Pareto solution set; p N represents the position variable of the p-th particle in the Pareto solution set. z z is the number of Pareto solution sets; p,min For z p The minimum value of z; p,max For z p The maximum value of α; A For ω AHP,λ Correlation coefficient; α C For ω CRI,λ The correlation coefficient.

[0049] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0050] 1. By optimizing the inverse time overcurrent protection settings, this invention can effectively reduce the impact of phase-to-phase faults and IIDG (Inverter Interfaced Distributed Generator), and can better meet the backup protection requirements of low-resistance grounded distribution networks.

[0051] 2. This invention has great versatility and can reliably and efficiently take protective actions for different fault types and the connection of inverter power supplies.

[0052] 3. This invention has the advantages of low implementation cost, wide applicability and strong practicality. This method can be applied to power distribution lines with communication automation functions. Attached Figure Description

[0053] Figure 1 This is a flowchart of the optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to the present invention.

[0054] Figure 2 The flowchart illustrates the optimization method for inverse-time overcurrent protection in an inverter-type distributed energy distribution network, as shown in the example.

[0055] Figure 3 A simplified model diagram of a low-resistance grounded distribution network.

[0056] Figure 4 This is a model diagram of a low-resistance grounded distribution network.

[0057] Figure 5 The curves showing the variation of ΔT under different fault locations and optimized models.

[0058] Figure 6 The target value is the optimization value for different inverter-type distributed energy access locations and capacities. Detailed Implementation

[0059] The following description, in conjunction with the accompanying drawings and specific embodiments, further illustrates the optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to the present invention.

[0060] Please see Figure 1 and Figure 2 This invention discloses an optimization method for inverse-time overcurrent protection in inverter-type distributed energy distribution networks, comprising the following steps:

[0061] S1. Establish an optimization model for inverse time overcurrent protection, determine the objective function and constraints for optimization of inverse time overcurrent protection, and use the multi-objective particle swarm optimization algorithm to solve the objective parameters of inverse time overcurrent protection;

[0062] S2. Based on the constraints, construct the initial population for the multi-objective particle swarm algorithm, calculate the fitness of the population particles, update the inertia weight and learning factor, and calculate the fitness value of the objective parameter.

[0063] S3. Update the population, calculate the weighted fitness of the population particles, calculate the adaptive grid density and selection probability, and perform mutation operations with Cauchy operators on the population particles.

[0064] S4. Compare whether the number of particles in the external archive exceeds the maximum allowed value. If it does, update the external archive and the global optimal solution. If it does not exceed the maximum allowed value, update the global optimal solution directly.

[0065] S5. Compare whether the current iteration count exceeds the maximum iteration count. If it does not exceed the maximum iteration count, continue to update the iteration. If it does exceed the maximum iteration count, end the iteration and obtain the Pareto optimal front set.

[0066] S6. Calculate the comprehensive weight using the analytic hierarchy process and the objective weighting method, and calculate the ideal solution using the approximation ideal solution sorting method to obtain the required inverse time protection setting.

[0067] Specifically, in step S1, the iteration parameters of the multi-objective particle swarm algorithm include the number of external archive particles, the number of Pareto solution sets, and the maximum number of iterations.

[0068] The objective function of the inverse time overcurrent protection optimization model is:

[0069]

[0070]

[0071] In the formula, N p Number of primary protections; N b Number of backup protections; The action time of the i-th main protection when the i-th protected segment fails; Let N be the action time of the k-th backup protection when the i-th protected section fails; obj1 is the first optimization objective; N bc K represents the total number of distribution network protection devices BC; senj is the sensitivity coefficient of the j-th main protection; obj2 is the second optimization objective.

[0072] The constraints of the inverse time overcurrent protection optimization model are:

[0073]

[0074] In the formula, ΔT is the time difference of the action; T p The time coefficient for inverse time overcurrent protection; I op The starting current for inverse-time overcurrent protection; tCTImax Let ΔT be the upper limit, and t be the lower limit. CTImin This is the lower limit of ΔT; T pmax For T p The upper limit of T pmin For T p The lower limit; I Lmax For I op The upper limit of I Lmin For I op The lower limit.

[0075] Specifically, in step S2, if the inverter-type distributed energy source experiences a grid disconnection process, the action time of the inverse-time overcurrent protection is updated through iterative calculation as follows:

[0076]

[0077] In the formula, The time of the i-th protection action when the i-th protected segment fails before disconnection from the network; N represents the time of the i-th protection action when the i-th protected segment fails before all inverter-type distributed energy sources are disconnected from the grid; f N represents the number of inverter-type distributed energy sources that experience grid disconnection processes; IB The number of inverter-type distributed energy sources; The time of the i-th protection action when the i-th protected section fails before the j-th inverter-type distributed energy source is disconnected from the grid;

[0078] The running time of the h-th inverter-type distributed power source after the j-th inverter-type distributed power source is disconnected from the grid; The running time of the h-th inverter-type distributed power source before the j-th inverter-type distributed power source is disconnected from the grid; The time of the i-th protection action when the i-th protected section fails after the j-th inverter-type distributed energy source is disconnected from the grid.

[0079] In step S2, the inertia weights are updated using the following formula:

[0080]

[0081] In the formula, w min The minimum value of the initial inertia weight; w max Δf is the maximum value of the initial inertia weight; Δf is the average weight variable. Let be the fitness value of the p-th particle in the g-th iteration; is the average fitness of the population at the g-th iteration; Let be the inertial weight of the p-th particle in the g-th iteration;

[0082] The learning factor is updated using the following formula:

[0083]

[0084] In the formula, This represents the minimum value of the individual learning factor; G represents the maximum value of the individual learning factor; G is the maximum number of iterations. This represents the individual learning factor at the g-th iteration. This represents the minimum value of the social learning factor. This represents the maximum value of the social learning factor. Let g be the social learning factor at the g-th iteration;

[0085] This represents the minimum fitness of the population at the g-th iteration. Let p be the position variable of the p-th particle in the g-th iteration; Let p be the position variable of the p-th particle in the (g+1)-th iteration; Let be the velocity variable of the p-th particle in the (g+1)-th iteration; This is the local optimal solution at the g-th iteration; This is the globally optimal solution at the g-th iteration.

[0086] Specifically, in step S3, the population is updated, and the weighted fitness of the population particles is calculated using the following formula:

[0087]

[0088] In the formula, ζ λ Let λ be the fitness coefficient of the λ-th objective. Let λ be the fitness of the p-th particle and the λ-th target at the g-th iteration. Let be the weighted fitness of the p-th particle in the g-th iteration.

[0089] The adaptive mesh density and selection probability are calculated using the following formula:

[0090]

[0091] In the formula, N q N represents the number of particles in the q-th grid. g N represents the number of grids in the external archive. R The number of particles in the external archive; [] is the round-up function; Let g be the adaptive grid density of the q-th grid in the g-th iteration; Let q be the selection probability of the q-th grid in the g-th iteration; This represents the minimum fitness value of the λ-th objective at the g-th iteration. This represents the maximum fitness value of the λ-th objective during the g-th iteration.

[0092] Specifically, in step S4, a mutation operation with a Cauchy operator is performed on the population particles, as shown in the formula:

[0093]

[0094]

[0095] In the formula, C e For the standard Cauchy mutation operator; C r P is a random number within the interval [0,1]. S η is the mutation probability; η is the adjustment parameter; Let be the position variable of the p-th particle after the mutation strategy in the g-th iteration.

[0096] Specifically, in step S6, the comprehensive weight is calculated using the Analytic Hierarchy Process (AHP) and the Criteria Importance Through Intercrieria Correlation (CRITIC) method, and the ideal solution is calculated using the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS). The formula is as follows:

[0097]

[0098] In the formula, N1 is the dimension of the subjective weight matrix Z1; ρ iλ ρ is the element in the i-th row of the λ-th objective in the subjective weight matrix Z1; ki ω is the element in the k-th row and i-th column of the subjective weight matrix Z1; AHP,λ Let λ be the weight of the analytic hierarchy process for the λ-th objective.

[0099]

[0100] In the formula, N p The subjective weight matrix Z p Dimension; S λ Let r be the Pareto standard deviation of the λth objective; iλ ω is the correlation coefficient of the i-th Pareto solution set for the λ-th objective; CRI,λ The objective weighting method weights for the λ-th objective;

[0101]

[0102] In the formula, D Tp ω is the approximate ideal solution evaluation value of the Pareto solution set; λ The weight of the λth objective; zp x is the average solution of the Pareto solution set; p N represents the position variable of the p-th particle in the Pareto solution set. z z is the number of Pareto solution sets; p,min For z p The minimum value of z; p,max For z p The maximum value of α; A For ω AHP,λ Correlation coefficient; α C For ω CRI,λ The correlation coefficient.

[0103] In this embodiment, the protection device settings of the 6-node inverter-type distributed energy distribution network system are optimized and adjusted, as shown in the schematic diagram below. Figure 4 As shown in the figure, the line length between each node is 2km, and the unit impedance is 0.06 + j0.089Ω / km; the power factor of each load is 0.92. The inverter-type distributed power sources IIDG1, IIDG2, and IIDG3 each have a capacity of 4.5MW, with active power of 1.5MW, 1.5MW, and 0.5MW respectively, and reactive power of 0MVAR. The energy storage systems ESS1 and ESS2 each have a capacity of 3.53MW, with active power of 1MW and reactive power of 1MVAR. The switching strategies α1 and α2 are set to 1 by default.

[0104] T in the constraint conditions pmin For 0.1s, T pmax It is 0.5s; t CTImin For 0.2s, t CTImax It is 0.4s; I Lmax I is the minimum short-circuit current flowing through BC; Lmin This represents the maximum load current flowing through BC. min It is 1.5; w max It is 2.5; =1; It is 2; =1; 2; N R =50; G = 200; η = 0.05; Z1 = [1, 1 / 3; 3, 1]; N p It is 50; α A and α C All are 0.5; the particle population size is 100.

[0105] To highlight the superiority of the improved algorithm in this invention, NSGA-II, MOGADE, standard MOPSO, and improved MOPSO are used to solve multi-objective optimization models. A Spacing metric is introduced to measure Z. p The uniformity of S is shown in Table 1, and the results of different algorithms are also presented.min S is the minimum value of Spacing. std Spacing standard deviation; population size, maximum number of iterations G, and number of particles N in the external archive REP. R Maintain consistency.

[0106] Table 1 Comparison of the solution performance of different algorithms

[0107]

[0108] Comparing the objective values ​​obj1 and obj2 of each algorithm in Table 1, it can be seen that the improved MOPSO (Multi-Objective Particle Swarm Optimization) can escape the search range and find a solution with a smaller objective value, namely [3.487, 2.083]. Compared with the standard MOPSO, the proposed scheme shortens the protection action time by 40ms and reduces the sensitivity by 1.501, effectively solving the problem that the standard MOPSO is prone to getting trapped in local optima. Compared with the solution performance of NSGA-II (Multi-Objective Genetic Algorithm) and MOGADE, the optimization scheme of the improved MOPSO has significant improvements in both speed and sensitivity.

[0109] The index S of the proposed solution in this invention min and S std Smaller than standard MOPSO and MOGADE metrics, ensuring Z p The uniformity is good; although the index is larger than that of NSGA-II, NSGA-II suffers from local optima and cannot guarantee uniformity for Z. p The improved MOPSO proposed in this invention balances the convergence accuracy and search capability in solving the Pareto front in terms of evolutionary mechanism, REP maintenance strategy, and mutation operation strategy, demonstrating the advantages of the proposed inverse time-limit protection scheme in terms of protection speed and sensitivity.

[0110] To verify the good protective performance of the multi-objective optimization model proposed in this invention, the speed, sensitivity, and selectivity of the following three optimization models are compared based on the improved MOPSO: ① Optimization model, i.e. Figure 3 China BC i No motion curve and and respectively by and Decision. ② Considering the single optimization objective of sensitivity and speed, Obj = obj1 + obj2. ③ The multi-objective optimization model of this scheme.

[0111] Comparison of protection speed and sensitivity

[0112] When a two-phase short-circuit fault occurs at the protection outlet of the distribution network, the minimum operating time difference, maximum operating time difference, total operating time of the main protection, total operating time of the backup protection, and the sum of sensitivity based on the above three optimization models are shown in Table 2. The fault transition resistance is 1Ω.

[0113] Table 2 Comparison of protection speed and sensitivity under different optimization models

[0114]

[0115] As shown in Table 2, compared with model ③ proposed in this invention, the minimum action time difference of model ① is reduced to 39ms, the sum of sensitivity is increased by 0.11, and the action time difference is less than t. CTImin (200ms) or greater than t CTImax In the case of (400ms), the mismatch between upper and lower level protection is caused; in Model ②, the total action time of the main protection and the total action time of the backup protection are extended by 65.9ms and 73.1ms respectively, the maximum action time difference is shortened by 17.7ms, the sum of sensitivity increases by 2.79, which cannot guarantee the speed of protection and the sensitivity is severely restricted.

[0116] Comparison of protective selectivity

[0117] Under different fault locations (point F moves from 0.4km to 2km in the interval), the change curve of the action time difference ΔT of the protection coordination relationship CR1 in models ①-③ is as follows: Figure 5 As shown. The fault type is a two-phase short-circuit fault, and the transition resistance is 1Ω. (From...) Figure 5 It can be seen that when point F moves from the beginning to the end of the interval, the ΔT of models ①-③ shows a monotonically decreasing trend. Furthermore, the ΔT of models ② and ③ is greater than t. CTImax (400ms) its motion curve and The two models cannot effectively coordinate and cooperate; however, Model ① can meet the protection coordination requirements across the entire line. Therefore, the proposed optimization model can take into account the selectivity, speed, and sensitivity of the protection, and also reflects the rationality and correctness of considering a multi-objective model.

[0118] As shown in Table D3, inverter-type distributed energy scenarios with different access locations and capacities are set up. The sum of the optimized total operating time of the main protection, the total operating time of the backup protection, and the sensitivity for each scenario is as follows: Figure 6 As shown. By Figure 6 It can be seen that the maximum sum of the total operating time of the main protection, the total operating time of the backup protection, and the sensitivity does not exceed 1.8s, 2.2s, and 2.5s, respectively, which can maintain good speed and sensitivity. Moreover, the effectiveness of the proposed protection scheme is not affected by the type of inverter-type distributed energy, the access location, and the capacity, and it has good adaptability.

[0119] Figure 6 The inverter-type distributed energy source connected to node 3 is experiencing grid disconnection. Figure 6 Table 3 shows the changes in the operating time of the main protection (BC3) and backup protection (BC2) before and after grid disconnection in the five scenarios illustrated. As can be seen from Table 3, the operating times of both the main protection (BC3) and backup protection (BC2) change after the inverter-type distributed energy source is disconnected from the grid. The change in the operating time of BC2 is particularly significant, with a time difference of up to 0.1175 seconds before and after grid disconnection. This demonstrates the effectiveness of the multi-objective optimization model considering the grid disconnection process of the inverter-type distributed energy source, and that the proposed protection scheme still guarantees selectivity and speed.

[0120] Table 3 shows the changes in protection action time during the grid disconnection process of inverter-type distributed energy sources.

[0121]

[0122] The calculation results above show that the proposed method for optimizing the inverse-time overcurrent protection settings of distribution networks with inverter-type distributed energy sources, based on an improved multi-objective particle swarm optimization algorithm, significantly improves the protection's effectiveness. The optimized settings accurately reflect the coordination relationship of the action curves, greatly enhancing protection sensitivity and speed performance. Furthermore, it exhibits strong adaptability to different inverter-type distributed energy source types, access locations, and capacities, thereby improving the absorption capacity of new energy sources. This invention has advantages such as low implementation cost, wide applicability, and strong practicality; it can be applied to distribution lines with communication automation functions.

[0123] The above description is a detailed description of the preferred embodiments of the present invention. However, the embodiments are not intended to limit the scope of the patent application of the present invention. All equivalent changes or modifications made under the technical spirit disclosed in the present invention should fall within the patent scope covered by the present invention.

Claims

1. An optimization method for inverse-time overcurrent protection in inverter-type distributed energy distribution networks, characterized in that, Includes the following steps: S1. Establish an optimization model for inverse time overcurrent protection, determine the objective function and constraints for optimization of inverse time overcurrent protection, and use the multi-objective particle swarm optimization algorithm to solve the objective parameters of inverse time overcurrent protection; S2. Based on the constraints, construct the initial population for the multi-objective particle swarm algorithm, calculate the fitness of the population particles, update the inertia weight and learning factor, and calculate the fitness value of the objective parameter. S3. Update the population, calculate the weighted fitness of the population particles, calculate the adaptive grid density and selection probability, and perform mutation operations with Cauchy operators on the population particles. S4. Compare whether the number of particles in the external archive exceeds the maximum allowed value. If it does, update the external archive and the global optimal solution. If it does not exceed the maximum allowed value, update the global optimal solution directly. S5. Compare whether the current iteration count exceeds the maximum iteration count. If it does not exceed the maximum iteration count, continue to update the iteration. If it does exceed the maximum iteration count, end the iteration and obtain the Pareto optimal front set. S6. Calculate the comprehensive weight using the analytic hierarchy process and the objective weighting method, and calculate the ideal solution using the approximation ideal solution sorting method to obtain the required inverse time protection setting. In step S1, the objective function of the inverse time overcurrent protection optimization model is: ； ； In the formula, N p Number of primary protections; N b Number of backup protections; The action time of the i-th main protection when the i-th protected segment fails; The action time of the k-th backup protection when the i-th protected segment fails; For the first optimization objective; N bc K represents the total number of distribution network protection devices BC; senj Let be the sensitivity coefficient of the j-th main protection. This is the second optimization objective; In step S2, if the inverter-type distributed energy source experiences a grid disconnection process, the action time of the inverse-time overcurrent protection is updated through iterative calculation as follows: ； In the formula, The time of the i-th protection action when the i-th protected segment fails before disconnection from the network; N represents the time of the i-th protection action when the i-th protected segment fails before all inverter-type distributed energy sources are disconnected from the grid; f N represents the number of inverter-type distributed energy sources that experience grid disconnection processes; IB The number of inverter-type distributed energy sources; The time of the i-th protection action when the i-th protected section fails before the j-th inverter-type distributed energy source is disconnected from the grid; The running time of the h-th inverter-type distributed power source after the j-th inverter-type distributed power source is disconnected from the grid; The running time of the h-th inverter-type distributed power source before the j-th inverter-type distributed power source is disconnected from the grid; The time of the i-th protection action when the i-th protected section fails after the j-th inverter-type distributed energy source is disconnected from the grid; In step S2, the inertia weights are updated using the following formula: ； In the formula, w min The minimum value of the initial inertia weight; w max This represents the maximum value of the initial inertia weight; For weighted average variables; Let be the fitness value of the p-th particle in the g-th iteration; is the average fitness of the population at the g-th iteration; Let be the inertial weight of the p-th particle in the g-th iteration; The learning factor is updated using the following formula: ； In the formula, This represents the minimum value of the individual learning factor; G represents the maximum value of the individual learning factor; G is the maximum number of iterations. This represents the individual learning factor at the g-th iteration. This represents the minimum value of the social learning factor. This represents the maximum value of the social learning factor. Let g be the social learning factor at the g-th iteration; This represents the minimum fitness of the population at the g-th iteration. Let p be the position variable of the p-th particle in the g-th iteration; Let p be the position variable of the p-th particle in the (g+1)-th iteration; Let be the velocity variable of the p-th particle in the (g+1)-th iteration; This is the local optimal solution at the g-th iteration; This is the globally optimal solution at the g-th iteration.

2. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 1, characterized in that, In step S1, the iteration parameters of the multi-objective particle swarm optimization algorithm include the number of external archive particles, the number of Pareto solution sets, and the maximum number of iterations.

3. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 1, characterized in that, The constraints of the inverse time overcurrent protection optimization model are: ； In the formula, T is the time difference between actions; p The time coefficient for inverse time overcurrent protection; I op The starting current for inverse-time overcurrent protection; t CTImax for The upper limit, t CTImin for The lower limit of T; pmax For T p The upper limit of T pmin For T p The lower limit; I Lmax For I op The upper limit of I Lmin For I op The lower limit.

4. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 1, characterized in that, In step S3, the population is updated, and the weighted fitness of the population particles is calculated using the following formula: ； In the formula, Let λ be the fitness coefficient of the λ-th objective. Let λ be the fitness of the p-th particle and the λ-th target at the g-th iteration. Let be the weighted fitness of the p-th particle in the g-th iteration.

5. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 4, characterized in that, In step S3, the adaptive mesh density and selection probability are calculated using the following formula: ； In the formula, N q N represents the number of particles in the q-th grid. g N represents the number of grids in the external archive. R The number of particles in the external archive; [ ] represents the floor function; Let g be the adaptive grid density of the q-th grid in the g-th iteration; Let q be the selection probability of the q-th grid in the g-th iteration; This represents the minimum fitness value of the λ-th objective at the g-th iteration. This represents the maximum fitness value of the λ-th objective during the g-th iteration.

6. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 5, characterized in that, In step S3, a mutation operation with the Cauchy operator is performed on the population particles, as shown in the formula: ； ； In the formula, C e For the standard Cauchy mutation operator; C r P is a random number within the interval [0,1]. S η is the mutation probability; η is the adjustment parameter; Let be the position variable of the p-th particle after the mutation strategy in the g-th iteration.

7. The optimization method for inverse-time overcurrent protection of inverter-type distributed energy distribution networks according to claim 6, characterized in that, In step S6, the comprehensive weights are calculated using the analytic hierarchy process (AHP) and the objective weighting method. The ideal solution is then calculated using the approximation of the ideal solution ranking method, with the following formula: ； In the formula, N1 is the dimension of the subjective weight matrix Z1; This refers to the element in the i-th row of the λ-th objective in the subjective weight matrix Z1; This refers to the element in the k-th row and i-th column of the subjective weight matrix Z1; Let λ be the weight of the analytic hierarchy process for the λ-th objective. ； In the formula, N p The subjective weight matrix Z p Dimension; Let λ be the Pareto standard deviation of the λ-th objective; The correlation coefficient of the i-th Pareto solution set for the λ-th objective; The objective weighting method weights for the λ-th objective; ； In the formula, The approximate ideal solution evaluation value for the Pareto solution set; Let λ be the weight of the λ-th objective. The average solution of the Pareto solution set; Let p be the position variable of the p-th particle in the Pareto solution set; The number of Pareto solution sets; for The minimum value; for The maximum value; for The correlation coefficient; for The correlation coefficient.

Images