Intelligent power grid multi-target operation optimization method and system for defending network attack
By employing real-number coding and iterative optimization methods, the problem of optimizing multiple performance indicators in smart grid systems when defending against network attacks was solved, achieving high security, reliability, and economical operation, improving power supply reliability, and reducing costs.
Patent Information
- Application Number
- CN202411005181.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-25
AI Technical Summary
Existing technologies struggle to achieve high security, reliability, and economical operation across multiple performance indicators when defending smart grid systems against cyberattacks, particularly in addressing the challenges of precisely setting weighting coefficients and optimizing overall performance through compromises.
A multi-objective operation optimization model for smart grids is constructed using a real-number encoding method. A population is generated through iteration and mutation operations, and the number of security defense components and generator parameters are optimized by combining dominance relationships and congestion distance ranking to achieve multi-objective optimization.
It has achieved high security, reliability and economical operation of smart grids in defending against network attacks, improving power supply reliability, security defense and reducing operating costs.
Smart Images

Figure CN119009950A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of operation optimization technology in the field of information security of smart grid systems, and in particular to a multi-objective operation optimization method and system for smart grids that defend against network attacks. Background Art
[0002] As electricity consumption continues to rise, infrastructure is prone to failures and reliability issues, and traditional power systems are being replaced by smart grids. The emergence of smart grids provides energy suppliers with greater transparency and accessibility, enabling them to conduct efficient real-time monitoring and management. As the dependence on communication networks gradually increases, the possibility of smart grids being attacked by cyber attacks is also increasing. Cyber attacks such as denial of service attacks and false data injection attacks will cause smart grid security monitoring technology to face huge challenges in detection accuracy, computational complexity, and credibility. Therefore, how to ensure that smart grids can defend against external cyber attacks while taking into account the economy of system operation is a technical problem that needs to be solved urgently.
[0003] Since the security defense capability of the smart grid system is closely related to the number of attacked buses and the security defense components installed on the bus, the defense status of the security defense components of the smart grid system should be systematically analyzed during the planning, design and operation optimization of the smart grid system, and the following factors should be comprehensively considered: (1) the load demand loss of the smart grid; (2) the procurement and operation cost of the security defense components; (3) the total fuel cost of power generation. However, current technical research is limited to the optimization of a single objective for the above situation. The common method of assigning weight coefficients to scale multiple objectives to form a single objective optimization or direct single objective optimization has the problem that the weight coefficients are difficult to accurately set and the comprehensive performance is difficult to compromise and optimize. In addition, there are a small number of studies that start from the perspective of multi-objective optimization, but they often start from the aspects of economy and system power supply loss probability, and only optimize the number of components in the smart grid system. The multi-objective optimization algorithm used is also relatively complex. Therefore, how to optimize the planning and design of the number of security defense components for defending against cyber attacks in the smart grid system, the generator active output vector, the generator voltage vector, the transformer decomposition ratio vector, the reactive power generation vector, the number of transformer taps and the number of reactive units, so as to ultimately achieve high security, reliability and economic operation of the smart grid with compromise optimization of multiple performance indicators, has become one of the key technical problems that need to be urgently solved in the field of smart grid system defense against cyber attacks. Summary of the invention
[0004] The purpose of the present invention is to provide a multi-objective operation optimization method and system for a smart grid that defends against network attacks in view of the deficiencies in the prior art.
[0005] The objective of the present invention is achieved through the following technical solutions: In a first aspect, an embodiment of the present invention provides a multi-objective operation optimization method for a smart grid for defending against network attacks, comprising the following steps:
[0006] S1: Construct a multi-objective operation optimization model of smart grid for defense against cyber attacks and its constraints;
[0007] S2: setting a plurality of first parameters;
[0008] S3: According to the first parameter, a real number coding method is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks, and after iterating the initial population, a first population P is generated. I ;
[0009] S4: Set the initial value of the number of iterations t to 1, record the number of iterations t, and for the first population P of the tth iteration I Each individual in the mutates one by one to obtain the second population P I+1 ;
[0010] S5: For the second population P I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 Dominance relations between offspring individuals within a group;
[0011] S6: According to the dominance relationship determined in step S5, from the second population P I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals are stored in the external archive A;
[0012] S7: According to the maximum number of archives L of the external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A;
[0013] S8: In the external archive A, sort in ascending order according to the crowding distance sorting results described in step S7, and select the first individuals, replacing the third population P n Randomly selected from individuals, forming the fourth population; where |A| represents the number of individuals stored in the external archive A, Indicates that Round up;
[0014] S9: Determine whether the number of iterations t satisfies t≥G max , if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals, including the procurement and operation costs of security defense components, the total fuel cost of power generation, and the load demand loss of the smart grid; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and repeat steps S4 to S8; where G max Indicates the maximum number of iterations.
[0015] Furthermore, the multi-objective operation model of the smart grid for defending against network attacks includes:
[0016] f1=(K1+b1)×(W1+W2+W3+W4+W5)+(K2+b2)×(W6+W7+W8+W9+W 10 ) (1)
[0017]
[0018]
[0019] Among them, f1, f2, and f3 represent the first, second, and third optimization objective functions respectively, f1 describes the purchase and operation costs of security defense components, f2 describes the total fuel cost of power generation, f3 describes the load demand loss of the smart grid, K1 and K2 represent the purchase costs of the first and second security defense components respectively, b1 and b2 represent the operating costs of the first and second security defense components respectively, e1 and e2 represent the reliability of the first and second security defense components respectively, z1 and z2 represent the costs of attacking the first and second security defense components respectively, W1, W2, W3, W4, and W5 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively, and W6, W7, W8, W9, and W10 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively. 10 are the number of Class 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively. G Indicates the number of generators connected to the grid, P Gv represents the actual output power of the vth generator, r v 、s v ,t v They represent the fuel cost curve coefficients corresponding to the vth generator, H Loss represents the number of buses associated with the load loss, Denotes the load demand of busbar m, D mN represents the probability of a successful network attack on bus m. re Indicates the probability that the re-type defense component successfully defends against network attacks.
[0020] Furthermore, the constraint conditions include instantaneous power balance and cost limit for defense attack, and the specific expressions thereof include:
[0021]
[0022] δ ij =δ i -δ j (8)
[0023] (K1+b1)×W m +(K2+b2)×W m+5 <6947 m=1,2,…,5 (9)
[0024]
[0025] Among them, δ i represents the voltage angle at busbar i, δ j represents the voltage angle at busbar j, δ ij represents the voltage angle difference between busbar i and busbar j, NB represents the number of buses, and They represent the active load demand and reactive load demand at bus i respectively, and They represent the active power generation and reactive power generation of the power source connected to bus i, V i Represents the voltage at bus i, V j represents the voltage at busbar j, G ij represents the transmission conductance between bus i and bus j, B ij represents the susceptance between bus i and bus j, K1 and K2 represent the purchase cost of the first and second category security defense components, b1 and b2 represent the operating cost of the first and second category security defense components, W1, W2, W3, W4, and W5 represent the number of first category security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively, and W6, W7, W8, W9, and W 10 Respectively represent the number of type 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, P Gk represents the active power of the generator of the kth bus, and They represent the lower and upper limits of the active power of the generator of the kth bus, Q Cn represents the parallel compensation of the nth busbar, and They represent the lower and upper limits of the parallel compensation of the nth busbar, T td represents the tap ratio vector of the voltage regulating transformer on the dth branch, and They represent the lower and upper limits of the tap ratio vector of the voltage regulating transformer on the dth branch, V Gf represents the generator voltage amplitude of the fth bus, and They represent the lower and upper limits of the generator voltage amplitude of the fth bus, respectively, W a Indicates the number of installed defense components. and They represent the lower and upper limits of the number of installed defense components, a=1, 2,…, 10.
[0026] Furthermore, the first parameter includes the purchase cost K1 of the first type of security defense component, the operating cost b1 of the first type of security defense component, the reliability e1 of the first type of security defense component, the cost z1 of attacking the first type of security defense component, the purchase cost K2 of the second type of security defense component, the operating cost b2 of the second type of security defense component, the reliability e2 of the second type of security defense component, the cost z2 of attacking the second type of security defense component, the maximum number of iterations G max , population size NP, external archive size L max , multi-point non-uniform variation parameter bc, and polynomial variation parameter q.
[0027] Furthermore, the real number coding method is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks, and after iterating the initial population, the first population P is generated. I , including:
[0028] The first population P with uniform distribution is randomly generated by real number coding method. I = {P s ,s=1,2,…,NP}, where P I represents the initial first population, P s represents the sth individual, NP represents the population size, and the sth individual P s =(P G1s ,P G2s ,P G3s ,P G4s ,P G5s ,V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s ,T 11s ,T 12s,T 15s ,T 36s ,Q C10s ,Q C12s ,Q C15s ,Q C17s ,Q C20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s ,W 1s ,W 2s ,W 3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s ), P G1s ,P G2s ,P G3s ,P G4s ,P G5s Respectively represent the active power of the generator of the 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s Respectively represent the generator voltage amplitudes of the 1st, 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, T 11s ,T 12s ,T 15s ,T 36s The sth individual corresponds to the tap ratio vector of the voltage regulating transformer on the 11th, 12th, 15th, and 36th branches, Q C10s ,Q C12s ,Q C15s ,Q C17s ,Q C20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s Respectively represent the parallel reactive power compensation of the 10th, 12th, 15th, 17th, 20th, 21st, 23rd, 24th, and 29th buses corresponding to the sth individual, W 1s ,W 2s ,W 3s ,W 4s ,W 5s represents the number of first-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual, respectively. 6s ,W 7s ,W 8s ,W9s ,W 10s Respectively represent the number of second-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual. 1s ,W 2s ,W 3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s Round to the nearest integer.
[0029] Furthermore, the mutation operation specifically includes:
[0030] Randomly generate 6 random numbers ranging from 0 to 1, labeled rn1, rn2, rn3, rn4, rn5, rn6;
[0031] If rn1 is greater than rn4, a multi-point non-uniform mutation operation is performed, and the formula of the multi-point non-uniform mutation operation is specifically:
[0032]
[0033] If rn1 is less than or equal to rn4, a polynomial mutation operation is performed, and the formula of the polynomial mutation operation is specifically:
[0034]
[0035] Among them, t represents the number of iterations, G max represents the maximum number of iterations, bc represents the multi-point non-uniform variation parameter, P cd (t) represents the first population P before mutation I The cdth variable of the sth individual in cd (t+1) represents the second population P after mutation I+1 The cdth variable of the sth individual in cd LB Represents the first population P I The lower limit allowed, P cd UB Represents the first population P I The upper limit of the allowable cd LB Depend on Composition, P cd UB Depend on Composition, q represents the polynomial variation parameter, represents the intensity coefficient of random variation, δ maxIt represents the farthest distance of an individual from the boundary, and B(t) represents the mutation intensity under the remaining number of iterations.
[0036] Furthermore, the step S5 includes the following sub-steps:
[0037] S51: The second population P I+1 By {P s,I+1 ,s=1,2,…,NP}, P s,I+1 Represents the population P I+1 The sth individual in the second population P I+1 Whether all individuals in the group violate the constraints described in formula (6) to formula (14), and then calculate the cost performance according to formula (17);
[0038]
[0039] in, represents the price / performance ratio, PF represents the second population P I+1 The number of feasible solutions, NP represents the population size;
[0040] S52: If the second population P I+1 All individuals in the second population P are infeasible solutions. I+1 The constraint violation amounts of all individuals in the group are normalized, and the normalized constraint violation degree is calculated according to formula (18) to formula (25), and the corresponding constraint violation degree index is marked as H(P s,I+1 ); according to the constraint violation degree, the first judgment method corresponding to formula (23) is used to judge the second population P I+1 The dominant relationship between any two individuals in the Labeled as the first non-dominated individual P d,I+1 ;
[0041]
[0042] in, Indicates that in individual P s,I+1 The corresponding o1th constraint in the case is, Indicates that in individual P s,I+1 The corresponding constraint violation degree index in the case of o1, o1 represents the sequence number of the constraint, δ represents the tolerance of constraint violation, s1 and s2 both represent P I+1 The number of individuals in the optimization objective function is d1 = 1, 2, ..., NP, s2 = 1, 2, ..., NP, but s1 ≠ s2, d1 represents the number of the optimization objective function, d1 = 1, 2, 3, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s1th individual in , Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s2th individual in , ε represents the maximum tolerance value of the normalized index of the constraint violation degree, ≤ ε represents the ε constraint dominance, Represents an individual For individuals Satisfy the ε constraint, represents any mathematical symbol, It means that when d1 takes any value among 1, 2, and 3, the corresponding relationship in formula (23) must hold. cp is the feedback coefficient. is the cost performance, t is the current iteration number, G max is the maximum number of iterations, rn7 is a randomly generated number in the range of 0 to 1;
[0043] S53: If the second group P I+1 There are some feasible solutions in the memory. The evaluation method corresponding to formula (26) to formula (28) is used to evaluate all individual P s,I+1 ,s=1,2,…,NP corresponding optimization objective function According to the first evaluation result, the first judgment method corresponding to formula (23) is used to judge the second population P I+1 The dominant relationship between any two individuals in the Labeled as the second non-dominated individual P d,I+1 ;
[0044]
[0045] in, Represents individual P s,I+1 The corresponding d1th new optimization objective function, H(P s,I+1 ) represents individual P s,I+1 The corresponding normalized constraint violation degree, Represents individual P s,I+1 The corresponding d1th normalized optimization objective function, P fea,I+1 represents a feasible solution, Represents the population P I+1 The minimum function value of the d1th optimization objective function corresponding to the feasible solution in mvol,I+1 Represents the population P I+1 The individual corresponding to the maximum value of the normalized index of constraint violation degree, Represents individual P mvol,I+1 The corresponding d1th optimization objective function value, For individual P s,I+1 The corresponding d1th optimization objective function value, ε represents the maximum tolerance value of the normalized index of constraint violation degree, mk is the penalty coefficient, For cost-effectiveness;
[0046] S54: If the second population P I+1 All are feasible solutions. The evaluation method corresponding to formula (29) is used to evaluate the second population P I+1 The dominant relationship of all individuals in the Labeled as the third non-dominated individual P d,I+1 ;
[0047]
[0048] in, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s2th individual, d1 and d2 both represent the serial numbers of the optimization objective functions, d1 = 1, 2, 3, d2 = 1, 2, 3; A mathematical symbol that indicates existence.
[0049] Further, according to the maximum number of archives L of the external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A, including:
[0050] If the maximum number of archives in the external archive A is L max Greater than the second population P I+1 The number of all non-dominated individuals in the second population P I+1 All non-dominated individuals in update the external archive A;
[0051] If the maximum number of archives in the external archive A is L max Less than or equal to the second population P I+1 The number of all non-dominated individuals in the second population P I+1In the above example, all non-dominated individuals are sorted by crowding distance, and the first L individuals in the most crowded position are left. max Individuals update external archive A.
[0052] Furthermore, the process of ranking the congestion distance specifically includes:
[0053]
[0054] The current size of the external archive is marked as |A|, A(s) represents the sth individual in the external archive, and the three new optimization objective functions corresponding to all individuals in the external archive {A(s), s=1,2,…,|A|} are Sort in ascending order so that in, They represent the first, second and |A|th serial numbers of the sorted index of the d1th optimization objective function, respectively. Indicates that the d1th optimization objective function value is sorted as The corresponding external archived individual; and The crowding distances are marked as and For s=2,L,|A|-1, calculate according to formula (30) The crowding distance
[0055] A second aspect of an embodiment of the present invention provides a multi-objective operation optimization system for a smart grid for defending against network attacks, which is used to implement the multi-objective operation optimization method for a smart grid for defending against network attacks. The system includes:
[0056] A building module for constructing a multi-objective operation optimization model of a smart grid for defense against cyber attacks and its constraints;
[0057] A setting module, used for setting a plurality of first parameters;
[0058] The first population initialization module is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks according to the first parameter by using a real number encoding method, and after iterating the initial population, generate a first population P I ;
[0059] The second population generation module is used to record the number of iterations t. For the first population P of the tth iteration I Each individual in the mutates one by one to generate the second population P I+1 ;
[0060] A dominance relationship judgment module is used to determine the dominance relationship of the second population P according to the I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 Dominance relations between offspring individuals within a group;
[0061] The archive module is used to extract the second population P according to the dominance relationship. I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals are stored in the external archive A;
[0062] The archive update module is used to update the second population P I+1 All non-dominated individuals and crowding distances in the external archive A are updated;
[0063] The fourth group generation module is used to sort the external archive A in ascending order according to the crowding distance and select the top individuals, replacing the third population P n Randomly selected from individuals, forming a fourth population; and
[0064] Iteration judgment module, used to judge whether the number of iterations t satisfies t≥G max , if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and return to the second population generation module to re-execute the iterative process.
[0065] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention comprehensively considers three important indicators, namely, the load demand loss of the smart grid, the procurement and operation costs of security defense components, and the total fuel cost of power generation, and designs a multi-objective adaptive operation optimization method for the smart grid that defends against network attacks, which efficiently realizes the high safety, reliability and economic operation of the smart grid with compromise optimization of multiple performance indicators; the present invention can efficiently realize the high safety, reliability and economic operation of the smart grid with compromise optimization of multiple performance indicators, and has higher power supply reliability, higher security defense and lower operating cost. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1It is a schematic diagram of the topological structure of an IEEE 30-node power system in an implementation case of the present invention;
[0067] Figure 2 It is a flow chart of the multi-objective adaptive operation optimization method of a smart grid for defending against network attacks of the present invention;
[0068] Figure 3 is the optimal Pareto frontier graph obtained after iterative optimization of the present invention;
[0069] Figure 4 A structural schematic diagram of the multi-objective adaptive operation optimization system of a smart grid for defending against network attacks of the present invention. DETAILED DESCRIPTION
[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0071] This embodiment is specifically described in conjunction with an IEEE 30-node power system. Figure 1 It is a topological diagram of the IEEE 30-node power system, including 5 thermal generators.
[0072] In this embodiment, the network attack used in the simulation is a false data injection attack. The goal of the network attack is to bypass bad data detection and maximize the load demand loss of the smart grid through erroneous scheduling information, thereby causing safety problems or physical damage. In this embodiment, it is assumed that in the IEEE 30-node power system, the 2nd, 5th, 7th, 8th, and 21st buses are attacked by false data injection.
[0073] See also Figure 2 The multi-objective adaptive operation optimization method of the smart grid for defending against network attacks specifically includes the following steps:
[0074] S1: Construct a multi-objective operation optimization model for smart grids to defend against cyber attacks and its constraints.
[0075] Furthermore, the multi-objective operation model of smart grid for defending against cyber attacks includes:
[0076] f1=(K1+b1)×(W1+W2+W3+W4+W5)+(K2+b2)×(W6+W7+W8+W9+W 10 ) (1)
[0077]
[0078]
[0079] Among them, f1, f2, and f3 represent the first, second, and third optimization objective functions respectively, f1 describes the purchase and operation costs of security defense components, f2 describes the total fuel cost of power generation, f3 describes the load demand loss of the smart grid, K1 and K2 represent the purchase costs of the first and second security defense components respectively, b1 and b2 represent the operating costs of the first and second security defense components respectively, e1 and e2 represent the reliability of the first and second security defense components respectively, z1 and z2 represent the costs of attacking the first and second security defense components respectively, W1, W2, W3, W4, and W5 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively, and W6, W7, W8, W9, and W10 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively. 10 are the number of Class 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively. G Indicates the number of generators connected to the grid, P Gv represents the actual output power of the vth generator, r v 、s v ,t v They represent the fuel cost curve coefficients corresponding to the vth generator, H Loss represents the number of buses associated with the load loss, Denotes the load demand of busbar m, D m N represents the probability of a successful network attack on bus m. re Indicates the probability that the re-type defense component successfully defends against network attacks.
[0080] Furthermore, the constraints include instantaneous power balance and the cost limit of defense attack, and the specific expressions include:
[0081]
[0082] δ ij =δ i -δ j (8)
[0083] (K1+b1)×W m +(K2+b2)×W m+5 <6947 m=1,2,…,5 (9)
[0084]
[0085] Among them, δ i represents the voltage angle at busbar i, δ j represents the voltage angle at busbar j, δ ijrepresents the voltage angle difference between busbar i and busbar j, NB represents the number of buses, and They represent the active load demand and reactive load demand at bus i respectively, and They represent the active power generation and reactive power generation of the power source connected to bus i, V i Represents the voltage at bus i, V j represents the voltage at busbar j, G ij represents the transmission conductance between bus i and bus j, B ij represents the susceptance between bus i and bus j, K1 and K2 represent the purchase cost of the first and second category security defense components, b1 and b2 represent the operating cost of the first and second category security defense components, W1, W2, W3, W4, and W5 represent the number of first category security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively, and W6, W7, W8, W9, and W 10 Respectively represent the number of type 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, P Gk represents the active power of the generator of the kth bus, and They represent the lower and upper limits of the active power of the generator of the kth bus, Q Cn represents the parallel compensation of the nth busbar, and They represent the lower and upper limits of the parallel compensation of the nth busbar, T td represents the tap ratio vector of the voltage regulating transformer on the dth branch, and They represent the lower and upper limits of the tap ratio vector of the voltage regulating transformer on the dth branch, V Gf represents the generator voltage amplitude of the fth bus, and They represent the lower and upper limits of the generator voltage amplitude of the fth bus, respectively, W a Indicates the number of installed defense components. and They represent the lower and upper limits of the number of installed defense components, a=1, 2,…, 10.
[0086] S2: Set a plurality of first parameters.
[0087] In this embodiment, the first parameter includes the purchase cost K1 of the first type of security defense component, the operating cost b1 of the first type of security defense component, the reliability e1 of the first type of security defense component, the cost z1 of attacking the first type of security defense component, the purchase cost K2 of the second type of security defense component, the operating cost b2 of the second type of security defense component, the reliability e2 of the second type of security defense component, the cost z2 of attacking the second type of security defense component, and the maximum number of iterations G max , population size NP, external archive size L max , multi-point non-uniform variation parameter bc, and polynomial variation parameter q.
[0088] It should be noted that the first parameter can be set according to actual conditions. For example, in this embodiment, the first parameter is set to: K1 = 239.52, b1 = 143.71, e1 = 0.75, z1 = 209.58, K2 = 958.08, b2 = 86.23, e2 = 0.90, z2 = 838.32, G max =1000,NP=500,L max =100, bc=3, q=0.1.
[0089] S3: According to the first parameter, a real number coding method is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks, and after iterating the initial population, a first population P is generated. I .
[0090] Specifically, the first population P with uniform distribution is randomly generated using real number coding method. I = {P s ,s=1,2,…,NP}, where P I represents the initial first population, P s represents the sth individual, NP represents the population size, and the sth individual P s =(P G1s ,P G2s ,P G3s ,P G4s ,P G5s ,V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s ,T 11s ,T 12s ,T 15s ,T 36s ,Q C10s ,Q C12s ,Q C15s ,Q C17s ,QC20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s ,W 1s ,W 2s ,W 3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s ), P G1s ,P G2s ,P G3s ,P G4s ,P G5s Respectively represent the active power of the generator of the 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s Respectively represent the generator voltage amplitudes of the 1st, 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, T 11s ,T 12s ,T 15s ,T 36s The sth individual corresponds to the tap ratio vector of the voltage regulating transformer on the 11th, 12th, 15th, and 36th branches, Q C10s ,Q C12s ,Q C15s ,Q C17s ,Q C20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s Respectively represent the parallel reactive power compensation of the 10th, 12th, 15th, 17th, 20th, 21st, 23rd, 24th, and 29th buses corresponding to the sth individual, W 1s ,W 2s ,W 3s ,W 4s ,W 5s represents the number of first-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual, respectively. 6s ,W 7s ,W 8s ,W 9s ,W 10s Respectively represent the number of second-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual. 1s ,W 2s ,W3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s Round to the nearest integer.
[0091] S4: Set the initial value of the number of iterations t to 1, record the number of iterations t, and for the first population P of the tth iteration I Each individual in the mutates one by one to obtain the second population P I+1 .
[0092] Furthermore, the mutation operation specifically includes:
[0093] Randomly generate 6 random numbers ranging from 0 to 1, labeled rn1, rn2, rn3, rn4, rn5, rn6;
[0094] If rn1 is greater than rn4, a multi-point non-uniform mutation operation is performed. The formula for the multi-point non-uniform mutation operation is as follows:
[0095]
[0096] If rn1 is less than or equal to rn4, perform the polynomial mutation operation. The formula for the polynomial mutation operation is:
[0097]
[0098] Among them, t represents the number of iterations, G max represents the maximum number of iterations, bc represents the multi-point non-uniform variation parameter, P cd (t) represents the first population P before mutation I The cdth variable of the sth individual in cd (t+1) represents the second population P after mutation I+1 The cdth variable of the sth individual in cd LB Represents the first population P I The lower limit allowed, P cd UB Represents the first population P I The upper limit of the allowable cd LB Depend on Composition, P cd UB Depend on Composition, q represents the polynomial variation parameter, represents the intensity coefficient of random variation, δ maxIt represents the farthest distance of an individual from the boundary, and B(t) represents the mutation intensity under the remaining number of iterations.
[0099] S5: For the second population P I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 The dominance relationship between offspring individuals within a group.
[0100] Specifically, step S5 includes the following sub-steps:
[0101] S51: The second population P I+1 By {P s,I+1 ,s=1,2,…,NP}, P s,I+1 Represents the population P I+1 The sth individual in the second population P I+1 Whether all individuals in the group violate the constraints described in formula (6) to formula (14); then calculate the cost performance according to formula (17):
[0102]
[0103] in, represents the price / performance ratio, PF represents the second population P I+1 NP is the number of feasible solutions and NP is the population size.
[0104] It should be understood that each individual in the second population is composed of the generator active power of the corresponding bus, the generator voltage amplitude of the bus, the tap ratio vector of the voltage-regulating transformer on the branch, the parallel reactive compensation of the bus, and the number of installed Class 1 and Class 2 safety defense components in the bus. Based on formulas (6) to (14), calculations are performed according to the composition of the individual to determine whether the individual violates the constraints described in formulas (6) to (14).
[0105] It should be noted that the second population P I+1 The individuals that do not violate the constraints described in formulas (6) to (14) are feasible solutions.
[0106] S52: If the second population P I+1 All individuals in the second population P are infeasible solutions. I+1 The constraint violation amounts of all individuals in the group are normalized, and the normalized constraint violation degree is calculated according to formula (18) to formula (25), and the corresponding constraint violation degree index is marked as H(P s,I+1 ); According to the degree of constraint violation, the first judgment method corresponding to formula (23) is used to judge the second population P I+1The dominant relationship between any two individuals in the formula (23) will satisfy the individual of the first judgment method. Labeled as the first non-dominated individual P d,I+1 .
[0107]
[0108]
[0109] in, Indicates that in individual P s,I+1 The corresponding o1th constraint in the case is, Indicates that in individual P s,I+1 The corresponding constraint violation degree index in the case of o1, o1 represents the sequence number of the constraint, δ represents the tolerance of constraint violation, s1 and s2 both represent P I+1 The number of individuals in the optimization objective function is d1 = 1, 2, ..., NP, s2 = 1, 2, ..., NP, but s1 ≠ s2, d1 represents the number of the optimization objective function, d1 = 1, 2, 3, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s1th individual in , Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s2th individual in , ε represents the maximum tolerance value of the normalized index of the constraint violation degree, ≤ ε represents the ε constraint dominance, Represents an individual For individuals Satisfy the ε constraint, represents any mathematical symbol, It means that when d1 takes any value among 1, 2, and 3, the corresponding relationship in formula (23) must hold. cp is the feedback coefficient. is the cost performance, t is the current iteration number, G max is the maximum number of iterations, and rn7 is a randomly generated number in the range of 0 to 1.
[0110] S53: If the second group P I+1 There are some feasible solutions in the memory. The evaluation method corresponding to formula (26) to formula (28) is used to evaluate all individual P s,I+1 ,s=1,2,…,NP corresponding optimization objective function According to the first evaluation result, the first judgment method corresponding to formula (23) is used to judge the second population P I+1 The dominant relationship between any two individuals in the formula (23) will satisfy the individual of the first judgment method. Labeled as the second non-dominated individual P d,I+1 .
[0111]
[0112] in, Represents individual P s,I+1 The corresponding d1th new optimization objective function, H(P s,I+1 ) represents individual P s,I+1 The corresponding normalized constraint violation degree, Represents individual P s,I+1 The corresponding d1th normalized optimization objective function, P fea,I+1 represents a feasible solution, Represents the population P I+1 The minimum function value of the d1th optimization objective function corresponding to the feasible solution in mvol,I+1 Represents the population P I+1 The individual corresponding to the maximum value of the normalized index of constraint violation degree, Represents individual P mvol,I+1 The corresponding d1th optimization objective function value, For individual P s,I+1 The corresponding d1th optimization objective function value, ε represents the maximum tolerance value of the normalized index of constraint violation degree, mk is the penalty coefficient, For cost performance.
[0113] S54: If the second population P I+1 All are feasible solutions. The evaluation method corresponding to formula (29) is used to evaluate the second population P I+1 The dominant relationship of all individuals in the Labeled as the third non-dominated individual P d,I+1 .
[0114]
[0115] in, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s2th individual, d1 and d2 both represent the serial numbers of the optimization objective functions, d1 = 1, 2, 3, d2 = 1, 2, 3; A mathematical symbol that indicates existence.
[0116] S6: According to the dominance relationship determined in step S5, from the second population P I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals in are stored in the external archive A.
[0117] S7: According to the maximum number of archives L of external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A.
[0118] Furthermore, according to the maximum number of archives L of the external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A, including:
[0119] If the maximum number of archives in external archive A is L max Greater than the second population P I+1 The number of all non-dominated individuals in the second population P I+1 All non-dominated individuals in update the external archive A;
[0120] If the maximum number of archives in external archive A is L max Less than or equal to the second population P I+1 The number of all non-dominated individuals in the second population P I+1 In the above example, all non-dominated individuals are sorted by crowding distance, and the first L individuals in the most crowded position are left. max Individuals update external archive A.
[0121] Furthermore, the process of crowding distance sorting specifically includes: marking the current size of the external archive as |A|, A(s) represents the sth individual in the external archive, and three new optimization objective functions corresponding to all individuals in the external archive {A(s), s=1,2,…,|A|} Sort in ascending order so that in, They represent the first, second and |A|th serial numbers of the sorted index of the d1th optimization objective function, respectively. Indicates that the d1th optimization objective function value is sorted as The corresponding external archived individual; and The crowding distances are marked as and For s=2,L,|A|-1, calculate according to formula (30) The crowding distance
[0122]
[0123] S8: In the external archive A, sort in ascending order according to the crowding distance sorting results described in step S7, and select the first individuals, replacing the third population P n Randomly selected from individuals, forming the fourth population; where |A| represents the number of individuals stored in the external archive A, Indicates that Round up.
[0124] S9: Determine whether the number of iterations t satisfies t≥G max , if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals, including the procurement and operation costs of security defense components, the total fuel cost of power generation, and the load demand loss of the smart grid; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and repeat steps S4 to S8; where G max Indicates the maximum number of iterations.
[0125] In this embodiment, after 1000 iterations, the Pareto frontier is obtained as follows: Figure 3 As shown, the real number encoding corresponding to the optimal compromise solution is p best =[58.7383,20.9685,35,18.2862,17.3430,1.0744,1.0599,1.0310,1.0398,1.0644,1.0526,1,1,1,1,4.7048,6,4.0534,6,4.3517,6,2.8310,6,3.8047,5,1,6,1,4,1,5,1,7,1]. The specific values of the three optimization objectives are: [15568.76,972.3013,2.709421].
[0126] In the real number coding corresponding to the optimal compromise individual, 58.7383, 20.9685, 35, 18.2862, and 17.3430 represent the generator active power of the 2nd, 5th, 8th, 11th, and 13th buses respectively, 1.0744, 1.0599, 1.0310, 1.0398, 1.0644, and 1.0526 represent the generator voltage amplitude of the 1st, 2nd, 5th, 8th, 11th, and 13th buses respectively, 1, 1, 1, 1 represent the tap ratio vector of the voltage regulating transformer on the 11th, 12th, 15th, and 36th branches respectively, 4.7048, 6, 4.0534, 6, 4.3517, and 6, 2.8310 , 6, and 3.8047 represent the parallel reactive power compensation of the 10th, 12th, 15th, 17th, 20th, 21st, 23rd, 24th, and 29th buses respectively; 5, 6, 4, 5, and 7 represent the number of type 1 safety defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively; 1, 1, 1, 1, and 1 represent the number of type 2 safety defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively; among the three target values corresponding to the optimal compromise individuals, the load demand loss of the smart grid is 2.709421MW, the procurement and operation cost of the safety defense components is US$15,568.76, and the total fuel cost of power generation is US$972.3013.
[0127] In summary, through the above technical scheme, the present invention comprehensively considers the three important performance indicators of the load demand loss of the smart grid, the procurement and operation cost of the security defense components, and the total fuel cost of power generation, and designs a multi-objective operation optimization adaptive method for the smart grid to defend against network attacks. Through the smart grid system that defends against network attacks, not only the defense time of the smart grid system to respond to network attacks is shortened, but also the defense cost and fuel cost are minimized while ensuring the safe operation of the smart grid system against network attacks, and the balance between the safety reliability and economy of the smart grid system is effectively achieved; through formula (1) to formula (29) combined with steps S1-S9 of the present invention, the high safety, reliability and economic operation of the smart grid with compromise optimization of multiple performance indicators is finally efficiently achieved. Compared with the prior art, the present invention achieves higher power supply reliability, higher security defense and lower operating cost.
[0128] It is worth mentioning that the present invention also provides a smart grid multi-objective operation optimization system for defending against network attacks, which is used to implement a smart grid multi-objective adaptive operation optimization method for defending against network attacks in the above-mentioned embodiment.
[0129] See also Figure 4 The system includes a construction module 10, a setting module 20, a first population initialization module 30, a second population generation module 40, a dominance relationship judgment module 50, an archive module 60, an archive update module 70, a fourth population generation module 80 and an iteration judgment module 90.
[0130] In this embodiment, the construction module 10 is used to construct a smart grid multi-objective operation optimization model and its constraint conditions for defending against network attacks.
[0131] In this embodiment, the setting module 20 is used to set a plurality of first parameters.
[0132] In this embodiment, the first population initialization module 30 is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks according to the first parameter using a real number encoding method, and after iterating the initial population, generate a first population P I .
[0133] In this embodiment, the second population generation module 40 is used to record the number of iterations t. For the first population P of the tth iteration I Each individual in the mutates one by one to generate the second population P I+1 .
[0134] In this embodiment, the dominance relationship judgment module 50 is used to determine the second population P I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 The dominance relationship between offspring individuals within a group.
[0135] In this embodiment, the archive module 60 is used to obtain the second population P according to the dominance relationship. I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals in are stored in the external archive A.
[0136] In this embodiment, the archive update module 70 is used to update the archive update module 70 according to the second population P. I+1 Update the external archive A with all non-dominated individuals and crowding distances.
[0137] In this embodiment, the fourth population generation module 80 is used to sort the external archive A in ascending order according to the crowding distance, and select the top individuals, replacing the third population P n Randomly selected from individuals, forming a fourth population.
[0138] In this embodiment, the iteration judgment module 90 is used to judge whether the number of iterations t satisfies t≥G max, if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and return to the second population generation module to re-execute the iterative process.
[0139] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc. Various media that can store program codes.
[0140] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A multi-objective operation optimization method for a smart grid to defend against network attacks, characterized in that: The following steps are involved: S1: Construct a multi-objective operation optimization model of smart grid for defense against cyber attacks and its constraints; S2: setting a plurality of first parameters; S3: According to the first parameter, a real number coding method is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks, and after iterating the initial population, a first population P is generated. I ; S4: Set the initial value of the iteration number t to 1, record the iteration number t, and for the first population P of the tth iteration I Each individual in the mutates one by one to obtain the second population P I+1 ; S5: For the second population P I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 Dominance relations between offspring individuals within a group; S6: According to the dominance relationship determined in step S5, from the second population P I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals are stored in the external archive A; S7: according to the maximum number of archives L of the external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A; S8: In the external archive A, sort in ascending order according to the crowding distance sorting results described in step S7, and select the first individuals, replacing the third population P n Randomly selected from individuals, forming the fourth population; where |A represents the number of individuals stored in the external archive A, Indicates that Round up; S9: Determine whether the number of iterations t satisfies t≥G max , if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals, including the procurement and operation costs of security defense components, the total fuel cost of power generation, and the load demand loss of the smart grid; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and repeat steps S4 to S8; where G max Indicates the maximum number of iterations.
2. The multi-objective operation optimization method for smart grids for defending against network attacks according to claim 1 is characterized in that: The multi-objective operation model of the smart grid for defending against network attacks includes: <h2 style=";text-align:left;direction:ltr">f1 = (K1+b1)×(W1+W2+W3+W4+W5)+(K2+b2)×(W6+W7+W8+W9+W<h2 style=";text-align:left;direction:ltr"> 10 <h2 style=";text-align:left;direction:ltr"> ) (1) Among them, f1, f2, and f3 represent the first, second, and third optimization objective functions respectively, f1 describes the purchase and operation costs of security defense components, f2 describes the total fuel cost of power generation, f3 describes the load demand loss of the smart grid, K1 and K2 represent the purchase costs of the first and second security defense components respectively, b1 and b2 represent the operating costs of the first and second security defense components respectively, e1 and e2 represent the reliability of the first and second security defense components respectively, z1 and z2 represent the costs of attacking the first and second security defense components respectively, W1, W2, W3, W4, and W5 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively, and W6, W7, W8, W9, and W10 are the numbers of the first security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses respectively. 10 are the number of Class 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively. G Indicates the number of generators connected to the grid, P Gv represents the actual output power of the vth generator, r v 、s v ,t v They represent the fuel cost curve coefficients corresponding to the vth generator, H Loss represents the number of buses associated with the load loss, Denotes the load demand of busbar m, D m N represents the probability of a successful network attack on bus m. re Indicates the probability that the re-type defense component successfully defends against network attacks.
3. The multi-objective operation optimization method for smart grids for defending against network attacks according to claim 1 is characterized in that: The constraints include instantaneous power balance and cost limit for defense attacks, and their specific expressions include: d ij =d i -d j (8) Among them, δ i represents the voltage angle at busbar i, δ j represents the voltage angle at busbar j, δ ij represents the voltage angle difference between busbar i and busbar j, NB represents the number of buses, and They represent the active load demand and reactive load demand at bus i respectively, and They represent the active power generation and reactive power generation of the power source connected to bus i, V i Represents the voltage at bus i, V j represents the voltage at busbar j, G ij represents the transmission conductance between bus i and bus j, B ij represents the susceptance between bus i and bus j, K1 and K2 represent the purchase cost of the first and second category security defense components, b1 and b2 represent the operating cost of the first and second category security defense components, W1, W2, W3, W4, and W5 represent the number of first category security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, respectively, and W6, W7, W8, W9, and W 10 Respectively represent the number of type 2 security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses, P Gk represents the active power of the generator of the kth bus, and They represent the lower and upper limits of the active power of the generator of the kth bus, Q Cn represents the parallel compensation of the nth busbar, and They represent the lower and upper limits of the parallel compensation of the nth busbar, T td represents the tap ratio vector of the voltage regulating transformer on the dth branch, and They represent the lower and upper limits of the tap ratio vector of the voltage regulating transformer on the dth branch, V Gf represents the generator voltage amplitude of the fth bus, and They represent the lower and upper limits of the generator voltage amplitude of the fth bus, respectively, W a Indicates the number of installed defense components. and They represent the lower and upper limits of the number of installed defense components, a=1, 2,…, 10.
4. The multi-objective operation optimization method for a smart grid for defending against network attacks according to claim 1 is characterized in that: The first parameters include the purchase cost K1 of the first type of security defense component, the operating cost b1 of the first type of security defense component, the reliability e1 of the first type of security defense component, the cost z1 of attacking the first type of security defense component, the purchase cost K2 of the second type of security defense component, the operating cost b2 of the second type of security defense component, the reliability e2 of the second type of security defense component, the cost z2 of attacking the second type of security defense component, and the maximum number of iterations G max , population size NP, external archive size L max , multi-point non-uniform variation parameter bc, and polynomial variation parameter q.
5. The multi-objective operation optimization method for smart grids for defending against network attacks according to claim 1 is characterized in that: The real number coding method is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks, and after iterating the initial population, the first population P is generated. I , specifically including: The first population P with uniform distribution is randomly generated by real number coding method. I = {P s ,s=1,2,...,NP}, where P I represents the initial first population, P s represents the sth individual, NP represents the population size, and the sth individual P s =(P G1s ,P G2s ,P G3s ,P G4s ,P G5s ,V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s ,T 11s ,T 12s ,T 15s ,T 36s ,Q C10s ,Q C12s ,Q C15s ,Q C17s ,Q C20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s ,W 1s ,W 2s ,W 3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s ), P G1s ,P G2s ,P G3s ,P G4s ,P G5s Respectively represent the active power of the generator of the 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, V G1s ,V G2s ,V G5s ,V G8s ,V G11s ,V G13s Respectively represent the generator voltage amplitude of the 1st, 2nd, 5th, 8th, 11th, and 13th buses corresponding to the sth individual, T 11s ,T 12s ,T 15s ,T 36s The sth individual corresponds to the tap ratio vector of the voltage regulating transformer on the 11th, 12th, 15th, and 36th branches, Q C10s ,Q C12s ,Q C15s ,Q C17s ,Q C20s ,Q C21s ,Q C23s ,Q C24s ,Q C29s Respectively represent the parallel reactive power compensation of the 10th, 12th, 15th, 17th, 20th, 21st, 23rd, 24th, and 29th buses corresponding to the sth individual, W 1s ,W 2s ,W 3s ,W 4s ,W 5s represents the number of first-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual, respectively. 6s ,W 7s ,W 8s ,W 9s ,W 10s Respectively represent the number of second-class security defense components installed in the 2nd, 5th, 7th, 8th, and 21st buses corresponding to the sth individual. 1s ,W 2s ,W 3s ,W 4s ,W 5s ,W 6s ,W 7s ,W 8s ,W 9s ,W 10s Round to the nearest integer.
6. The multi-objective operation optimization method for a smart grid for defending against network attacks according to claim 1 is characterized in that: The mutation operation specifically includes: Randomly generate 6 random numbers ranging from 0 to 1, labeled rn1, rn2, rn3, rn4, rn5, rn6; If rn1 is greater than rn4, a multi-point non-uniform mutation operation is performed, and the formula of the multi-point non-uniform mutation operation is specifically: If rn1 is less than or equal to rn4, a polynomial mutation operation is performed, and the formula of the polynomial mutation operation is specifically: Among them, t represents the number of iterations, G max represents the maximum number of iterations, bc represents the multi-point non-uniform variation parameter, P cd (t) represents the first population P before mutation I The cdth variable of the sth individual in cd (t+1) represents the second population P after mutation I+1 The cdth variable of the sth individual in cd LB Represents the first population P I The lower limit allowed, P cd UB Represents the first population P I The upper limit of the allowable cd LB Depend on Composition, P cd UB Depend on Composition, q represents the polynomial variation parameter, represents the intensity coefficient of random variation, δ max It represents the farthest distance of an individual from the boundary, and B(t) represents the mutation intensity under the remaining number of iterations.
7. The multi-objective operation optimization method for a smart grid for defending against network attacks according to claim 1 is characterized in that: The step S5 comprises the following sub-steps: S51: The second population P I+1 By {P s,I+1 ,s=1,2,...,NP}, P s,I+1 Represents the population P I+1 The sth individual in the second population P I+1 Whether all individuals in the group violate the constraints described in formula (6) to formula (14), and then calculate the cost performance according to formula (17); in, represents the price / performance ratio, PF represents the second population P I+1 The number of feasible solutions, NP represents the population size; S52: If the second population P I+1 All individuals in the second population P are infeasible solutions. I+1 The constraint violation amounts of all individuals in the group are normalized, and the normalized constraint violation degree is calculated according to formula (18) to formula (25), and the corresponding constraint violation degree index is marked as H(P s,I+1 ); according to the constraint violation degree, the first judgment method corresponding to formula (23) is used to judge the second population P I+1 The dominant relationship between any two individuals in the Labeled as the first non-dominated individual P d,I+1 ; in, Indicates that in individual P s,I+1 The corresponding o1th constraint in the case is, Indicates that in individual P s,I+1 The corresponding constraint violation degree index in the case of o1, o1 represents the sequence number of the constraint, δ represents the tolerance of constraint violation, s1 and s2 both represent P I+1 The number of individuals in the optimization objective function is d1 = 1, 2, ..., NP, s2 = 1, 2, ..., NP, but s1 ≠ s2, d1 represents the number of the optimization objective function, d1 = 1, 2, 3, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s1th individual in , Indicates P I+1 The normalized index of the constraint violation degree corresponding to the s2th individual in , ε represents the maximum tolerance value of the normalized index of the constraint violation degree, ≤ ε represents the ε constraint dominance, Represents an individual For individuals Satisfy the ε constraint, represents any mathematical symbol, It means that when d1 takes any value among 1, 2, and 3, the corresponding relationship in formula (23) must hold. cp is the feedback coefficient. is the cost performance, t is the current iteration number, G max is the maximum number of iterations, rn7 is a randomly generated number in the range of 0 to 1; S53: If the second group P I+1 There are some feasible solutions in the memory. The evaluation method corresponding to formula (26) to formula (28) is used to evaluate all individual P s,I+1 ,s=1,2,...,NP corresponding optimization objective function s=1,2,...,NP to evaluate and obtain a first evaluation result; according to the first evaluation result, the first judgment method corresponding to formula (23) is used to judge the second population P I+1 The dominant relationship between any two individuals in the Labeled as the second non-dominated individual P d,I+1 ; in, Represents individual P s,I+1 The corresponding d1th new optimization objective function, H(P s,I+1 ) represents individual P s,I+1 The corresponding normalized constraint violation degree, Represents individual P s,I+1 The corresponding d1th normalized optimization objective function, P fea,I+1 represents a feasible solution, Represents the population P I+1 The minimum function value of the d1th optimization objective function corresponding to the feasible solution in mvol,I+1 Represents the population P I+1 The individual corresponding to the maximum value of the normalized index of constraint violation degree, Represents individual P mvol,I+1 The corresponding d1th optimization objective function value, For individual P s,I+1 The corresponding d1th optimization objective function value, ε represents the maximum tolerance value of the normalized index of constraint violation degree, mk is the penalty coefficient, For cost-effectiveness; S54: If the second population P I+1 All are feasible solutions. The evaluation method corresponding to formula (29) is used to evaluate the second population P I+1 The dominant relationship of all individuals in the Labeled as the third non-dominated individual P d,I+1 ; in, Indicates P I+1 The d1th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d1th optimization objective function corresponding to the s2th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s1th individual in is, Indicates P I+1 The d2th optimization objective function corresponding to the s2th individual, d1 and d2 both represent the serial numbers of the optimization objective functions, d1 = 1, 2, 3, d2 = 1, 2, 3; A mathematical symbol that indicates existence.
8. The multi-objective operation optimization method for a smart grid for defending against network attacks according to claim 1 is characterized in that: According to the maximum number of archives L of the external archive A max And the crowding distance sorting results, determine whether to use the second population P I+1 All non-dominated individuals in update the external archive A, including: If the maximum number of archives in the external archive A is L max Greater than the second population P I+1 The number of all non-dominated individuals in the second population P I+1 All non-dominated individuals in update the external archive A; If the maximum number of archives in the external archive A is L max Less than or equal to the second population P I+1 The number of all non-dominated individuals in the second population P I+1 In the above example, all non-dominated individuals are sorted by crowding distance, and the first L individuals in the most crowded position are left. max Individuals update external archive A.
9. The multi-objective operation optimization method for a smart grid for defending against network attacks according to claim 8 is characterized in that: The process of the congestion distance sorting specifically includes: The current size of the external archive is marked as |A|, A(s) represents the sth individual in the external archive, and the three new optimization objective functions corresponding to all individuals in the external archive {A(s), s=1,2,…,|A|} are Sort in ascending order so that in, They represent the first, second and |A|th serial numbers of the sorted index of the d1th optimization objective function, respectively. Indicates that the d1th optimization objective function value is sorted as The corresponding external archived individual; and The crowding distances are marked as and For s = 2, L, A | -1, calculate according to formula (30) The crowding distance 10. A multi-objective operation optimization system for a smart grid for defending against network attacks, used to implement the multi-objective operation optimization method for a smart grid for defending against network attacks as claimed in any one of claims 1 to 9, characterized in that: The system comprises: A building module for constructing a multi-objective operation optimization model of a smart grid for defense against cyber attacks and its constraints; A setting module, used for setting a plurality of first parameters; The first population initialization module is used to encode the optimization variables in the multi-objective operation optimization model of the smart grid to be optimized for defense against network attacks according to the first parameter by using a real number encoding method, and after iterating the initial population, generate a first population P I ; The second population generation module is used to record the number of iterations t. For the first population P of the tth iteration I Each individual in the mutates one by one to generate the second population P I+1 ; A dominance relationship judgment module is used to determine the dominance relationship of the second population P according to the I+1 The offspring individuals in the second population P are processed according to the dominant method of the constraint mechanism. I+1 The offspring individual constraints within the group are used to determine the degree of constraint violation and the second group P I+1 Dominance relations between offspring individuals within a group; The archive module is used to extract the second population P according to the dominance relationship. I+1 A non-dominated individual is randomly selected from the individuals, and the non-dominated individual is named P d , stored in the third population P n Used to update individuals in the next iteration, and at the same time, the second population P I+1 All non-dominated individuals are stored in the external archive A; The archive update module is used to update the second population P I+1 All non-dominated individuals and crowding distances in the external archive A are updated; The fourth group generation module is used to sort the external archive A in ascending order according to the crowding distance and select the top individuals, replacing the third population P n Randomly selected from individuals, forming a fourth population; and Iteration judgment module, used to judge whether the number of iterations t satisfies t≥G max , if satisfied, all non-dominated individuals stored in the external archive A are output as the optimal Pareto solution set, and the Pareto optimal individuals are selected according to the actual engineering requirements of the smart grid operation to obtain the optimization objective function value corresponding to the Pareto optimal individuals; otherwise, set t = t + 1, unconditionally replace the next generation population with the fourth population, and return to the second population generation module to re-execute the iterative process.
Citation Information
Patent Citations
Multi-target reactive power optimization method for electric system
CN102611119A
Multi-objective PQ optimization control method for three-phase grid-connected inverter
CN106655256A
Power information physical cooperative attack analysis method considering false injection of load data
CN108234492A
High-dimensional multi-objective constraint optimization method for micro-grid ductile economic dispatching
CN118100312A
