A comprehensive optimization method, system and device for voltage / reactive power control of a photovoltaic inverter distribution network based on multi-parameter discretization

Abstract

The present invention discloses a comprehensive optimization method, system and device for voltage / reactive power control of a photovoltaic inverter distribution network based on multi-parameter discretization. The method minimizes the network loss cost, inverter degradation cost and average bus voltage deviation by centrally optimizing the reactive power set value and Q-V control function of the photovoltaic inverter. The general Q-V control model of the photovoltaic inverter is accurately described by using a piecewise linearization method. The life degradation caused by the inverter supporting voltage / reactive power is analyzed and quantified as a voltage regulation cost, which is incorporated into the optimization objective. The continuous variable discretization method is used to convert the non-convex non-linear constraints in the piecewise linear Q-V control model into linearly constrained ones that can be effectively solved. The voltage control method in the present invention can reduce the operation cost of the distribution network and improve the economy. It can make full use of the remaining capacity of the photovoltaic inverter to participate in voltage control, ensuring the rapidity and flexibility of voltage regulation when the photovoltaic and load outputs are unstable.

Description

Technical Field

[0001] The present invention relates to the field of distribution network voltage control, and in particular to a comprehensive optimization method, system and device for photovoltaic inverter distribution network voltage / reactive power control based on multi-parameter discretization. Background Art

[0002] In recent years, as a clean energy source, photovoltaic has developed rapidly. By the end of 2020, its installed capacity reached 253 million kilowatts. The high proportion of photovoltaic access in the distribution network has led to large fluctuations and even over-limit of the distribution network voltage, posing a huge challenge to the voltage control of the distribution network. Photovoltaic inverters have fast and flexible reactive power output functions and can provide reactive power compensation for the power grid. With the gradual increase in the proportion, the voltage control of photovoltaic inverters has become an adjustment means that cannot be ignored. Due to the low adjustment cost, no additional adjustment device is required, and the response to random voltage fluctuations is faster, which is of great significance for the safe and stable operation of the distribution network.

[0003] However, most of the existing photovoltaic inverter reactive power-voltage (Q-V) control models only optimize some parameters and do not comprehensively consider the model parameters (such as voltage dead zone position, dead zone range, control gain, etc.) that affect the voltage regulation performance of the inverter. In addition, with the increase in the number of parameters to be optimized in the model, the non-linear terms in the model become more complex, exacerbating the computational burden. Summary of the Invention

[0004] Object of the Invention: The present invention aims to provide a comprehensive optimization method for photovoltaic inverter distribution network voltage / reactive power control based on multi-parameter discretization that minimizes the network loss cost, inverter degradation cost and average bus voltage deviation; another object of the present invention is to provide a comprehensive optimization system and device for photovoltaic inverter distribution network voltage / reactive power control based on multi-parameter discretization.

[0005] Technical Solution: The comprehensive optimization method for photovoltaic inverter distribution network voltage / reactive power control based on multi-parameter discretization described in the present invention includes the following steps:

[0006] S1. Based on the reactive power control characteristics of the photovoltaic inverter, establish a general Q-V control model for the photovoltaic inverter;

[0007] S2. Based on the voltage regulation characteristics of the inverter, establish the relationship between the voltage regulation cost and life reduction of the inverter;

[0008] S3. According to the general Q-V control model of the photovoltaic inverter in step S1 and the relationship between the voltage regulation cost and life reduction of the inverter in step S2, considering the distribution network power flow constraint, establish a comprehensive optimization model for distribution network voltage control with the minimum distribution network network loss cost, minimum inverter loss cost and minimum average bus voltage deviation as the objectives;

[0009] S4. According to the comprehensive optimization model of distribution network voltage control in step S3, using the multi-parameter discretization method, convert the non-convex and non-linear constraints in the piecewise linear Q-V control model into linear constraints that can be effectively solved, obtain the comprehensive optimization linear model of distribution network voltage control, and by adjusting the accuracy level, make the output result of the comprehensive optimization linear model of distribution network voltage control gradually approach the optimal solution to obtain the optimal control strategy.

[0010] Further, the general model of Q-V control of the photovoltaic inverter is

[0011]

[0012] w i,t,1,r ≤y i,t,1,r ,w i,t,6,r ≤y i,t,5,r (3)

[0013] w i,t,k,r ≤y i,t,k,r +y i,t,k-1,r ,k = 2, 3, 4, 5 (4)

[0014] w i,t,k,r ≥0,y i,t,k,r ∈{0, 1} (5)

[0015]

[0016] Among them, V i,t,r is the voltage amplitude of the i-th node in the r-th scenario within the t-th scheduling interval; w i,t,k,r is a continuous variable; is the voltage amplitude corresponding to the k-th point on the Q-V control curve; is the reactive power output by the inverter, Q i,t,k is the reactive power corresponding to the k-th point on the Q-V control curve, y i,t,k,r is a binary variable.

[0017] Further, the relationship between the inverter voltage regulation cost and the life reduction is represented by quantifying the photovoltaic inverter voltage regulation cost through the relationship between the reactive power output by the inverter and the unit voltage regulation cost of the inverter, as follows:

[0018]

[0019] Among them, η LR,0 、η LR,1 and η LR,2 are constants.

[0020] Furthermore, considering the power flow constraints of the distribution network, with the minimum network loss cost, inverter loss cost, and average bus voltage deviation of the distribution network as the objectives, the comprehensive optimization model for distribution network voltage control is

[0021]

[0022] s.t. (1)-(6)(9)

[0023]

[0024]

[0025] where obj is the network loss cost; w is the weight coefficient, w ∈ [0, 1]; C loss is the network loss cost, C LR is the reactive power cost caused by the photovoltaic inverter participating in voltage regulation, is the average node voltage offset of the r-th scenario in the t-th scheduling interval, and R is the number of scenarios per scheduling interval; is the time-of-use electricity price; P ij,t,r and Q ij,t,r respectively represent the active power and reactive power on the transmission line; V0 represents the rated voltage of the line; is the rated capacity of the photovoltaic inverter; is the active power output of the photovoltaic; P hi,t,r is the active power of all lines flowing into node i; Q hi,t,r is the reactive power of all lines flowing into node i; V j,t,r is the voltage amplitude of the j-th node in the r-th scenario of the t-th scheduling interval; r ij and x ij respectively represent the resistance and reactance of the line; V i and respectively represent the lower and upper limits of the node voltage; represents the power capacity of the transmission line.

[0026] Furthermore, in step S4, the bilinear term in equation (1) is linearized as follows:

[0027] Introduce the auxiliary variable Redescribe

[0028]

[0029] where, is a continuous variable; Substitute equation (20) into and introduce the auxiliary variable to obtain:

[0030]

[0031] After discretization, it is represented as follows:

[0032]

[0033] where p is a negative integer used to represent the precision level; g is the value at different quantiles; is a binary variable; introducing to ensure it can take any value within the domain;

[0034] Introduce an auxiliary variable and substitute it into formula (21) to get:

[0035]

[0036] Introduce a continuous variable to get:

[0037]

[0038] where the non - linear term is relaxed using the McCormick envelope, w i,t,k,r ∈[0,1], Let the variable The relaxed model is as follows:

[0039]

[0040] Multiply both sides of formula (24) by w i,t,k,r to obtain the following formula:

[0041]

[0042] The bilinear term is expressed as a mixed - integer linear constraint through formulas (20), (22)-(25) and (27)-(33).

[0043] Furthermore, in step S4, the w in formula (2) i,t,k,r Q i,t,k is linearized, and its linear model is as follows:

[0044]

[0045] Furthermore, the integrated optimal linear model for distribution network voltage control obtained in step S4 is

[0046] ​

[0047] such that (9)-(20), (22)-(25), (27)-(45)(47).

[0048] Further, in step S4, the adjustment of the precision level is specifically as follows: First, take p = -1, solve the integrated optimization linear model of the distribution network voltage control to obtain the optimal values of each decision variable, which are used as the initial points of the integrated optimization model of the distribution network voltage control. Then, solve the objective optimal value obj of the integrated optimization model of the distribution network voltage control and the objective optimal value obj of the integrated optimization linear model of the grid voltage control respectively * Compare. When the constraint C is satisfied, the decision is optimal; otherwise, set p = p - 1, repeat the above process until the constraint C is satisfied to end the iteration, and finally obtain the optimal control strategy of the distribution network;

[0049] where the constraint C is

[0050]

[0051] In the formula, ε is a preset threshold, and this threshold is a small positive number.

[0052] The integrated optimization system for the voltage and reactive power control of a photovoltaic inverter distribution network based on multi-parameter discretization according to the present invention includes:

[0053] A model construction module, which is used to construct a general Q-V control model of the photovoltaic inverter based on the reactive power control characteristics of the photovoltaic inverter; is used to establish an integrated optimization model of the distribution network voltage control with the minimum network loss cost, the minimum inverter loss cost, and the minimum average bus voltage deviation of the distribution network as the objectives according to the general Q-V control model of the photovoltaic inverter and the relationship between the inverter voltage regulation cost and the life reduction, considering the distribution network power flow constraint; is used to construct an integrated optimization linear model of the distribution network voltage control by using the multi-parameter discretization method to transform the non-convex and non-linear constraints in the piecewise linear Q-V control model into effectively solvable linear constraints;

[0054] A decision-making and solving module, which is used to gradually approximate the optimal solution according to the output result of the integrated optimization linear model of the distribution network voltage control by adjusting the precision level to obtain the optimal control strategy.

[0055] The computer device according to the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0056] Beneficial effects: Compared with the prior art, the present invention has the following remarkable advantages: 1. By centrally optimizing the reactive power set value and Q-V control function of the photovoltaic inverter, the present invention minimizes the network loss cost, inverter degradation cost, and average bus voltage deviation; 2. Considering the impact of the additional reactive power support provided by the inverter when participating in voltage / reactive power control on its lifespan, the present invention analyzes the relationship between the reduction in inverter lifespan and the voltage regulation cost, analyzes and quantifies the lifespan degradation caused by the inverter's support for voltage / reactive power control as the voltage regulation cost, and incorporates it into the optimization objective; 3. The present invention uses the continuous variable discretization method to transform the non-convex non-linear constraints in the piecewise linear Q-V control model into linearly constrained conditions that can be effectively solved, and tests the proposed method on a 33-bus distribution system. The simulation results verify its effectiveness; 4. In order to balance the solution speed and model accuracy, the present invention adopts an adaptive model accuracy solution method. By gradually adjusting the accuracy, the result of the linearized model gradually approaches the optimal solution, achieving fast regulation and precise response of the distribution network voltage. Description of the Drawings

[0057] Figure 1 is the flowchart of the present invention;

[0058] Figure 2 is the relationship curve between the unit reactive power cost of the inverter and the output reactive power;

[0059] Figure 3 is the topology diagram of the 33-node distribution system;

[0060] Figure 4 is the voltage comparison diagram between the present invention and the method without control;

[0061] Figure 5 is the comparison diagram of the solution results of the comprehensive optimization linear model for distribution network voltage control. Detailed Embodiment

[0062] The present invention will be further described below with reference to the drawings.

[0063] As Figure 1 shown, the method for comprehensive optimization of voltage / reactive power of a photovoltaic inverter-based distribution network based on multi-parameter discretization according to the present invention includes the following steps:

[0064] S1. Reactive power constraint of the photovoltaic inverter

[0065]

[0066] Among them, is the maximum reactive power output / absorption of the photovoltaic inverter, is the inverter capacity, is the photovoltaic active power output with uncertainty;

[0067] General Model of Q-V Control for Photovoltaic Inverters:

[0068]

[0069] Among them, the index t represents a certain scheduling interval in the entire optimization period, r represents a preset scenario within the scheduling interval, the i-index represents the node number, and V i,t,r represents the magnitude of the node voltage. is the upper limit of the reactive power output of the inverter. represents the reactive power value output by the inverter when the node voltage is within the dead zone, that is, the reactive power setting value at the central layer, where represents the abscissa of the 6 points constituting the Q-V control curve.

[0070] It can be seen from Equation (2) that the expressions of the second and fourth segments of the Q-V control curve have relatively complex non-linear terms, making it difficult to linearize them. To facilitate the relaxation of the non-linear terms, the Q-V control curve is accurately modeled by the method of piecewise linearization. By introducing continuous variables w i,t,k,r (k = 1, 2,..., 6) and binary variables y i,t,k,r (k = 1, 2,..., 5), the Q-V control curve can be transformed into the following model:

[0071]

[0072] w i,t,1,r ≤y i,t,1,r , w i,t,6,r ≤y i,t,5,r (6)

[0073] w i,t,k,r ≤y i,t,k,r +y i,t,k-1,r , k = 2, 3, 4, 5 (7)

[0074] w i,t,k,r ≥0, y i,t,k,r ∈{0, 1} (8)

[0075]

[0076] In Equation (5), Q i,t,k (k = 1, 2,..., 6) are the ordinates of the 6 points constituting the Q-V control curve. The auxiliary variable y i,t,k,r limits the value of w i,t,k,r to determine the interval where V i,t,r is located; w i,t,k,r is used to accurately describe the original function of the interval part. S2. The reduction in the inverter's lifespan caused by participating in Q-V control can be calculated through the following formula:

[0077]

[0078] Where OL P is the operating life when the inverter only outputs active power, and OL Q is the operating life when the inverter outputs both active and reactive power. Q represents the reactive power output by the inverter, and v, θ c , μ, λ c are the relevant operating parameters of the inverter at time period c.

[0079] To quantify the inverter life loss, it is necessary to establish the relationship between the inverter life loss LR and the increment LI of the levelized cost of electricity of the photovoltaic system. According to the formula for calculating the levelized cost of electricity of the photovoltaic system, the following relational expression is fitted:

[0080]

[0081] Where η LI,A0 , η LI,A1 , η LI,B0 , η LI,B1 are constants.

[0082] Multiply both sides of formula (11) by which is the average active power of the photovoltaic system during time period c in its life cycle. Therefore, the reactive power cost generated by the inverter life loss is as follows:

[0083]

[0084] Where c c Q,LR is the reactive power cost generated by the inverter life loss, is the reactive power output by the inverter during time period c. Due to the correlation between the inverter reactive power output and the inverter life, therefore, through the fitting method, formulas (11) and (12) can be simplified into the following single-variable polynomial equation:

[0085]

[0086] Where η LR,0 , η LR,1 and η LR,2 are constants. Finally, the relationship between the reactive power output of the photovoltaic inverter and the regulation cost is as Figure 2 shown.

[0087] S3. The integrated optimization problem of the distribution network voltage control is modeled as follows:

[0088]

[0089] s.t.(4)-(9)(15)

[0090]

[0091] The objective function (14) aims to minimize the network loss cost, the inverter voltage regulation cost, and the average node voltage deviation, where C loss represents the network loss cost, and C LR represents the reactive power cost caused by the participation of PV inverters in voltage regulation. represents the average node voltage offset of all preset scenarios, and R represents the number of scenarios for each scheduling interval. w is the weight coefficient, w ∈ [0, 1]; Equations (16)-(18) respectively give the calculation methods of the network loss cost, the inverter voltage regulation cost, and the average node voltage deviation in the objective function.

[0092] Equations (19) and (20) are the reactive power capacity constraints of the inverter. Equations (21) and (22) are the node power balance equations. The voltage relationship between nodes can be expressed as the linear model shown in Equation (23). Equation (24) is the node voltage constraint. Equation (25) is the active power transmission capacity constraint of the line.

[0093] S4. According to the integrated optimization model of distribution network voltage control in step S3, using the multi-parameter discretization method, the non-convex and non-linear constraints in the piecewise linear Q-V control model are transformed into linear constraints that can be effectively solved, obtaining the integrated optimization linear model of distribution network voltage control. By adjusting the accuracy level, the output result of the integrated optimization linear model of distribution network voltage control gradually approaches the optimal solution, and the optimal control strategy is obtained.

[0094] For the bilinear terms and w i,t,k,r Q i,t,k (k = 2, 3, 4, 5), linearize them respectively:

[0095] First, introduce the auxiliary variable Redescribe

[0096]

[0097] Substitute Equation (26) into the original non-linear term, and introduce the auxiliary variable to obtain the following equation:

[0098]

[0099] Substitute and discretize it as follows,

[0100]

[0101] Introduce the auxiliary variable Substitute it into Equation (27) to obtain:

[0102]

[0103] Introduce a new continuous variable Therefore

[0104]

[0105] The newly generated non - linear term in Equation (34) Use the McCormick envelope for relaxation, where w i,t,k,r ∈[0, 1], Then let the variable The relaxed model is shown as follows.

[0106]

[0107] Multiply both sides of Equation (30) by w i,t,k,r The following equation can be obtained:

[0108]

[0109] Therefore, the bilinear term Is represented as mixed - integer linear constraints through Equations (26), (28) - (31) and (33) - (39).

[0110] Similarly, w i,t,k,r Q i,t,k (k = 2, 3, 4, 5) can also be linearized by the above method, and its linear model is shown as follows:

[0111]

[0112]

[0113] Finally, the re - description of the original optimization model is shown as follows

[0114]

[0115] s.t. (14) - (26), (28) - (31), (33) - (51)(53)

[0116] However, according to the above derivation process, it can be found that the optimal solution of the reconstructed model is not necessarily the optimal solution of the original problem, but only provides a lower bound for the optimal solution of the original problem. This is mainly because and This is due to insufficient precision, that is, p cannot take an infinitesimal negative integer. To solve this problem, this method first takes the lowest precision p = -1, solves the reconstructed optimization model to obtain the optimal values of each decision variable as the initial point of the original model, and then solves the original model to obtain its objective optimal value obj and the objective optimal value obj of the reconstructed model * for comparison. When the constraint of Equation (54) is satisfied, the decision is optimal; otherwise, set p = p - 1 and continue the above process until Equation (54) is satisfied and the iteration ends.

[0117]

[0118] To verify the effect of the present invention, the following experiments are given in the embodiments of the present invention:

[0119] As Figure 3 shown is the network topology and photovoltaic position of the test system, and Table 1 shows the rated active power of the photovoltaic and the inverter capacity.

[0120] Table 1

[0121]

[0122] Using the method of the present invention to control the voltage of the test system, as Figure 4 shown is the comparison of voltage deviation with and without the control method.

[0123] The present invention also proposes to relax the bilinear terms using a multi-parameter decomposition method and proposes an adaptive precision solution algorithm Figure 5 for the solution process of the optimization problem.

[0124] It can be seen that using the method of the present invention to control the voltage of the distribution network can effectively reduce the voltage deviation and improve the economy. When the reactive power output of the inverter is too high, it will seriously affect the life of the inverter. However, when the output power of the inverter is small, the life of the inverter is little affected. Therefore, it is very necessary to quantify the life loss caused by the inverter participating in voltage regulation as a cost. In addition, for the non-convex non-linear terms in the Q-V control model, the proposed multi-parameter solution method can continuously iterate and approximate the original optimal solution according to different precision requirements, which also ensures the precision of model solution while improving the solution speed.

[0125] The integrated optimization system for photovoltaic inverter distribution network voltage and reactive power control based on multi-parameter discretization described in the present invention includes:

[0126] A model construction module, which is used to construct a general Q-V control model of a photovoltaic inverter based on the reactive power control characteristics of the photovoltaic inverter; to establish a comprehensive optimization model for distribution network voltage control with the minimum network loss cost, minimum inverter loss cost, and minimum average bus voltage deviation of the distribution network as the objectives by considering the distribution network power flow constraint according to the general Q-V control model of the photovoltaic inverter and the relationship between inverter voltage regulation cost and life reduction; to construct a comprehensive optimization linear model for distribution network voltage control by using a multi-parameter discretization method to convert the non-convex and non-linear constraints in the piecewise linear Q-V control model into effectively solvable linear constraints.

[0127] A decision-making and solving module, which is used to obtain an optimal control strategy by adjusting the accuracy level to make the output result of the comprehensive optimization linear model of the distribution network voltage control gradually approach the optimal solution.

[0128] The computer device of the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

[0129] By centrally optimizing the reactive power set value and Q-V control function of the photovoltaic inverter, the present invention minimizes the network loss cost, inverter degradation cost, and average bus voltage deviation. A general Q-V control model of the photovoltaic inverter is established, and the model is accurately described by using a piecewise linearization method. The life degradation caused by the inverter supporting voltage / reactive power is analyzed and quantified as the voltage regulation cost, which is then incorporated into one of the optimization objectives. On this basis, a continuous variable discretization method is used to convert the non-convex and non-linear constraints in the piecewise linear Q-V control model into effectively solvable linear constraints. The voltage control method in the present invention can reduce the operation cost of the distribution network and improve the economy. It can make full use of the remaining capacity of the photovoltaic inverter to participate in voltage control, ensuring the rapidity and flexibility of voltage regulation when the photovoltaic and load outputs are unstable.

Claims

1. A photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization, characterized in that: The following steps are involved: S1. Based on the reactive power control characteristics of photovoltaic inverters, a general model of photovoltaic inverter QV control is established; S2. Based on the voltage regulation characteristics of the inverter, establish the relationship between the voltage regulation cost and the life reduction of the inverter; S3, according to the photovoltaic inverter QV control general model in step S1 and the relationship between the inverter voltage regulation cost and life reduction in step S2, considering the power flow constraints of the distribution network, with the minimum network loss cost, minimum inverter loss cost and minimum average bus voltage deviation of the distribution network as the goal, establish a comprehensive optimization model for distribution network voltage control; S4. According to the comprehensive optimization model of distribution network voltage control in step S3, a multi-parameter discretization method is used to convert the non-convex nonlinear constraints in the piecewise linear QV control model into effectively solvable linear constraints, and a comprehensive optimization linear model of distribution network voltage control is obtained. By adjusting the accuracy level, the output result of the comprehensive optimization linear model of distribution network voltage control gradually approaches the optimal solution, and the optimal control strategy is obtained; The general model of photovoltaic inverter QV control is: In i,t,1,r ≤y i,t,1,r ,In i,t,6,r ≤y i,t,5,r (3) w i,t,k,r ≤y i,t,k,r +y i,t,k-1,r ,k=2,3,4,5 (4) w i,t,k,r ≥0,y i,t,k,r ∈{0,1} (5) Among them, V i,t,r is the voltage amplitude of the ith node in the rth scenario within the tth scheduling interval; w i,t,k,r is a continuous variable; is the voltage amplitude corresponding to the kth point on the QV control curve; is the reactive power output by the inverter, Q i,t,k is the reactive power corresponding to the kth point on the QV control curve, y i,t,k,r is a binary variable; The relationship between the voltage regulation cost of the inverter and the reduction of its life span is quantified by quantifying the voltage regulation cost of the photovoltaic inverter and the reactive power output by the inverter. And inverter unit voltage regulation cost The relationship is expressed as follows: Among them, η LR,0 , η LR,1 and η LR,2 is a constant; Considering the power flow constraints of the distribution network, with the minimum network loss cost, inverter loss cost and average bus voltage deviation as the goal, the comprehensive optimization model of distribution network voltage control is: st(1)—(6) (9) Where obj is the optimization target; w is the weight coefficient, w∈[0,1]; C loss is the network loss cost, C LR is the reactive power cost caused by the PV inverter participating in voltage regulation, is the average node voltage offset of the rth scenario in the tth scheduling interval, and R is the number of scenarios for each scheduling interval; is the time-of-use electricity price; P ij,t,r and Q ij,t,r They represent the active power and reactive power on the transmission line respectively; V0 represents the rated voltage of the line; is the rated capacity of the PV inverter; The active power output of photovoltaic and Respectively represent the load active power and reactive power; P hi,t,r is the active power flowing into all lines of node i; Q hi,t,r is the reactive power of all lines flowing into node i; V j,t,r is the voltage amplitude of the jth node in the rth scenario within the tth scheduling interval; r ij and x ij Respectively represent the resistance and reactance of the line; V i and V i Respectively represent the lower and upper limits of node voltage; Indicates the power capacity of the transmission line; It is the maximum reactive power output\absorption of the PV inverter.

2. The photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization according to claim 1 is characterized in that: In step S4, the bilinear term in equation (1) is Linearization is performed as follows: Introducing auxiliary variables Redescribe in, is a continuous variable; Substituting formula (20) into And introduce auxiliary variables get: Will After discretization, it is expressed as follows: Among them, p is a negative integer, which is used to represent Accuracy level; g is Values ​​at different quantiles; is a binary variable; ensure Get any value in the domain; Introducing auxiliary variables Substituting it into formula (21), we get: Introducing continuous variables have to: Among them, the nonlinear term Relaxation is performed using the McCormick envelope, w i,t,k,r ∈[0,1], Let the variable The relaxation model is as follows: Multiply both sides of formula (24) by w i,t,k,r The following formula can be obtained: Bilinear term It is expressed as a mixed integer linear constraint by equations (20), (22)-(25) and (27)-(33).

3. The photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization according to claim 2 is characterized in that: In step S4, w in equation (2) is i,t,k,r Q i,t,k After linearization, the linear model is as follows:

4. The photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization according to claim 3 is characterized in that: The comprehensive optimization linear model of distribution network voltage control obtained in step S4 is: st(9)-(20),(22)-(25),(27)-(45) (47).

5. The photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization according to claim 4 is characterized in that: In step S4, the accuracy level is adjusted as follows: the accuracy level is first set to p = -1, and the distribution network voltage control comprehensive optimization linear model is solved to obtain the optimal value of each decision variable, which is used as the initial point of the distribution network voltage control comprehensive optimization model, and the target optimal value obj of the distribution network voltage control comprehensive optimization model and the target optimal value obj of the power grid voltage control comprehensive optimization linear model are solved respectively. * Compare, when constraint C is satisfied, the decision is optimal, otherwise set p = p-1, repeat the above process until constraint C is satisfied and the iteration ends, and finally the optimal control strategy of the distribution network is obtained; The constraint C is Where ε is a preset threshold, which is a small positive number.

6. A system for the photovoltaic inverter distribution network voltage / reactive power control comprehensive optimization method based on multi-parameter discretization according to any one of claims 1 to 5, characterized in that: include: The model building module is used to build a general model of photovoltaic inverter QV control based on the reactive power control characteristics of photovoltaic inverters; based on the general model of photovoltaic inverter QV control and the relationship between the inverter voltage regulation cost and life reduction, considering the power flow constraints of the distribution network, with the minimum network loss cost, the minimum inverter loss cost and the minimum average bus voltage deviation of the distribution network as the goals, to establish a comprehensive optimization model of distribution network voltage control; based on the comprehensive optimization model of distribution network voltage control, using a multi-parameter discretization method, the non-convex nonlinear constraints in the piecewise linear QV control model are converted into effectively solvable linear constraints, and a comprehensive optimization linear model of distribution network voltage control is constructed; The decision-making solution module is used to obtain the optimal control strategy by adjusting the accuracy level so that the output result of the distribution network voltage control comprehensive optimization linear model gradually approaches the optimal solution.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images