High-proportion photovoltaic power distribution network reactive power optimization method considering day-ahead random-intra-day rolling distribution

Abstract

The invention relates to a high-proportion photovoltaic power distribution network reactive power optimization method considering day-ahead random-intra-day rolling distribution. Monte Carlo simulation is combined with an M ILP model to establish a day-ahead optimization model, output strategies of a photovoltaic inverter and a micro gas turbine are predicted and optimized, and network loss and operation cost are minimized; on the basis, DPMU real-time data is used for carrying out intra-day rolling optimization, and a reactive power output strategy is dynamically adjusted; a model is constructed by taking reactive power loss minimization as a target, and reactive power resources are coordinated by using an ADMM algorithm; and finally, realizing continuous optimization through closed-loop control and effect evaluation. The system stability can be improved, the operation cost can be reduced, the electric energy quality can be improved, the adaptability and the optimization precision can be enhanced, and the reactive power optimization problem of the high-proportion photovoltaic power distribution network can be effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization control, and in particular to a reactive power optimization method for a high-proportion photovoltaic distribution network taking into account random day-ahead and intra-day rolling distribution. Background Art

[0002] With the increasing awareness of environmental protection and the transformation of energy structure, solar energy is increasingly used as a clean energy in power systems, and a high proportion of photovoltaic access to distribution networks has become a trend. However, photovoltaic power generation is intermittent and volatile, which brings many challenges to the operation of distribution networks.

[0003] On the one hand, the output power of photovoltaic power generation varies with factors such as light intensity and temperature, making it difficult to accurately match it with load demand, which can easily cause grid power fluctuations and affect the stability of the system. For example, when the light intensity suddenly changes (such as cloud cover), the photovoltaic power generation may drop sharply in an instant. If the system reactive power is not adjusted in time at this time, it may cause voltage fluctuations or even voltage collapse, threatening the safe and stable operation of the power grid.

[0004] On the other hand, the large number of photovoltaic accesses has changed the power distribution of the distribution network, making it difficult for traditional reactive power compensation and voltage regulation methods to meet the needs. At the same time, existing reactive power optimization methods often lack sufficient consideration of the uncertainty of photovoltaic power generation and load, and the optimization model is not accurate enough to effectively cope with the dynamic changes in actual operation. Summary of the invention

[0005] In view of the above deficiencies in the prior art, the purpose of the present invention is to overcome the deficiencies of the prior art in dealing with photovoltaic uncertainty, reactive power optimization accuracy and timeliness, and to effectively improve the operating stability, economy and power quality of the distribution network through precise modeling, real-time rolling optimization and distributed algorithms, so as to achieve efficient and reliable operation of the system.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solutions:

[0007] The reactive power optimization method of a high-proportion photovoltaic distribution network taking into account the random day-ahead and intraday rolling distribution includes the following steps:

[0008] S1. Establishment of the day-ahead optimization model. The Monte Carlo simulation method is combined with the mixed integer linear programming (MILP) model to simulate the uncertainty of photovoltaic power generation and load to achieve the establishment of the day-ahead optimization model. The day-ahead optimization model is used to predict and optimize the output strategy of photovoltaic inverters and micro gas turbines (MT) to minimize the distribution system network loss and photovoltaic operation cost.

[0009] S2. Intraday rolling optimization: Based on the day-ahead optimization results, the real-time data collected by the DPMU device is used to make dynamic adjustments through the distributed computing method of the distribution network partitions. With the help of the rolling optimization mechanism, the photovoltaic power generation and load demand information are updated in real time, and the current and future system operation conditions are comprehensively considered to optimize the reactive power output strategy to cope with power fluctuations and ensure stable and economical operation of the system.

[0010] S3. Reactive power optimization: Based on the intraday rolling optimization, the reactive power optimization model is constructed with the goal of minimizing reactive power network loss, taking full account of the reactive output of photovoltaic inverters and micro gas turbines, and using the alternating direction multiplier method (ADMM) as a distributed optimization algorithm to coordinate reactive resources in the distribution network, optimize reactive power flow distribution, reduce active power loss and maintain a good voltage level.

[0011] S4. Implementation and feedback of results: Through closed-loop control strategy, the reactive power optimization results are sent to each control unit in the distribution network to perform reactive power optimization control; the optimization effect is evaluated by real-time monitoring system, and the optimization model parameters are adjusted according to the evaluation results to improve the optimization accuracy and adaptability, and achieve continuous optimization and improvement.

[0012] Furthermore, the step of establishing the day-ahead optimization model includes:

[0013] S11. Establish a probability distribution model of photovoltaic power generation and load. For photovoltaic power generation, its power generation model is expressed as:

[0014]

[0015] P pv is the photovoltaic power generation; P stc It is the photovoltaic power generation under standard test conditions. It is the rated power generation of photovoltaic panels under specific standard light intensity and temperature, and is a known fixed parameter. stc is the light intensity under standard test conditions, which is a constant; k is the power temperature coefficient, which is a negative value and indicates the rate at which the power generated by the photovoltaic panel decreases when the temperature rises; T stc is the panel temperature under standard test conditions; I is the actual light intensity; T is the actual temperature;

[0016] For load, the probability distribution function is analyzed and fitted based on the historical load data accumulated over a long period of time. The load shows different rules in different time periods, and it is necessary to use segmented fitting or mixed distribution model to accurately describe it. The load is expressed as:

[0017]

[0018] where u Lis the mean value of the load, representing the average load demand in the region during a specific period of time. It is the variance of the load, reflecting the range of variation of load demand.

[0019] S12. Construct a mixed integer linear programming (MILP) model, and the objective function is expressed as:

[0020] min C=C l +C pv ;

[0021]

[0022] min C represents the objective function; G is the optimization time period; N is the number of nodes in the distribution network; C l is the network loss cost of the distribution system; C pv is the photovoltaic operation cost; P l,t,i is the network loss power of node i during period t; c l is the unit network loss cost, that is, the economic cost generated by each unit of power loss; M is the number of photovoltaic power sources; c PV is the unit photovoltaic power operating cost, including the depreciation cost of photovoltaic panels, maintenance cost, inverter loss cost, etc.; P PV,t,j is the output power of the jth photovoltaic power source in period t;

[0023] The constraints include: power balance constraint, photovoltaic power output constraint, micro gas turbine output constraint, voltage constraint, and line flow constraint;

[0024] Power balance constraints:

[0025]

[0026] K is the number of micro turbines (MT); P MT,t,k is the output power of the kth micro gas turbine in period t; this constraint ensures that at any time, the sum of the output power of all power sources in the system is equal to the sum of the load demand power and the network loss power, reflecting the principle of energy conservation; P L,t At time t, the load power of the distribution network;

[0027] Photovoltaic power output constraint: P PV,min,j ≤P PV,t,j ≤P PV,max,j ;

[0028] P PV,min,j and P PV,max,jare the minimum and maximum output limits of the jth photovoltaic power source respectively; the minimum output limit may be affected by factors such as the physical characteristics of the photovoltaic panel and the minimum working power of the inverter; the maximum output limit is mainly determined by the rated capacity of the photovoltaic panel and the maximum conversion capacity of the inverter;

[0029] Micro gas turbine output constraints:

[0030] P MT,min,k ≤P MT,t,k ≤P MT,max,k ;

[0031] P MT,min,k and P MT,max,k are the minimum and maximum output limits of the kth micro gas turbine respectively; the minimum output limit of a micro gas turbine is usually related to the minimum requirements for its startup and stable operation, while the maximum output limit depends on its design capacity, fuel supply capacity, bearing capacity of mechanical components, etc.;

[0032] Voltage constraints, line flow constraints:

[0033] V min,i ≤V t,i ≤V max,i ;I t,ij ≤I max,ij ;

[0034] V min,i and V max,i are the upper and lower limits of the voltage at node i;

[0035] V t,i is the voltage of node i at time period t; I t,ij is the current of line ij during period t; I max,ij is the maximum allowable current of line ij.

[0036] Furthermore, in S2, the step of dynamically adjusting based on the optimization result of the day-ahead optimization model and the real-time data collected by the DPMU device includes:

[0037] S21. Real-time data collection: The DPMU device collects data such as photovoltaic power generation, load demand, voltage, etc. of each node in the distribution network in real time. The DPMU device uses high-precision sensors and fast data processing algorithms to accurately measure the amplitude, phase and other information of electrical quantities;

[0038] S22. Rolling optimization calculation. During intraday rolling optimization, the objective function is:

[0039] min C r ≤C l,r +C PV,r +C adj ;

[0040]

[0041] min C r During intraday rolling optimization, the objective function value; C l,r is the real-time network loss cost; C PV,r is the real-time photovoltaic operation cost; C adj is the adjustment cost, C adj According to the reactive power adjustment and related cost coefficient, the reactive power adjustment is the change in reactive power of the photovoltaic inverter or other reactive power regulating equipment compared with the previous optimization result. PV,r,t,j is the actual output power of the jth node at the current time t; P l,r,t,i is the real-time network loss power of node i at the current time t;

[0042] S23. Constraint update:

[0043] Among them, the power balance constraint is updated as:

[0044]

[0045] P MT,r,t,k is the real-time output power of the kth micro gas turbine at the current moment, obtained by the real-time monitoring device of the micro gas turbine; P L,r,t At time t, the actual load power of the distribution network;

[0046] S24. Rolling optimization mechanism, which performs rolling optimization calculations every same time period. In each calculation, it first calculates based on the real-time data at the current moment and the forecast of the operating conditions in the future. Then, based on the forecast results and real-time data, it re-optimizes the reactive power output strategy of the photovoltaic inverter and micro gas turbine, and adjusts the switching of reactive compensation equipment.

[0047] Furthermore, in S3, the model constructed by reactive power optimization with the goal of minimizing reactive power network loss is expressed as:

[0048]

[0049] Among them, X i is the reactance of node i, is the reactive current of node i during period t; I t,i Current measured in real time; is the power factor angle measured in real time;

[0050] The constraints include: reactive power balance constraint and 3 equipment reactive output constraint;

[0051] Reactive power balance constraints:

[0052]

[0053] Q PV,t,j At time t, the reactive output of the jth PV inverter is Q MT,t,k At time t, the reactive output of the jth micro gas turbine is Q C,t,l At time t, the reactive output of the lth reactive compensation device; Q l,t,i At time t, the reactive network loss of node i; L is the number of reactive compensation devices; The power factor angle of reactive load demand; Q L,t is the reactive load demand during period t; P L,t is the active power of the load;

[0054] Equipment reactive power output constraints:

[0055] Q Q PV,min,j ≤Q PV,t,j ≤Q Q PV,max,j ;

[0056] Q Q MT,min,k ≤Q MT,t,k ≤Q Q MT,max,k ;

[0057] Q Q C,min,l ≤Q C,t,l ≤Q Q C,max,l ;

[0058] Q Q PV,min,j , Q Q PV,max,j They represent the lower and upper limits of the reactive power output of the j-th photovoltaic inverter respectively; Q Q MT,min,k , Q Q MT,max,k are the lower and upper limits of the reactive power output of the kth micro gas turbine respectively; Q Q C,min,l , Q Q C,max,l are respectively the lower limit and upper limit of the reactive output of the lth reactive compensation device.

[0059] Furthermore, in S3, the step of using the alternating direction method of multipliers (ADMM) for distributed optimization includes:

[0060] S31. Decompose the optimization problem, decompose the reactive power optimization problem of the entire distribution network into multiple sub-problems, each sub-problem corresponds to a distributed power source or control node; for a certain photovoltaic power source, its local sub-problem is to minimize its contribution to reactive network loss under the conditions of satisfying its own reactive output constraints and exchanging information with other nodes;

[0061] S32. Introduce dual variables and augmented Lagrangian function to construct augmented Lagrangian function:

[0062]

[0063] L ρ,j (Q,λ) represents the total Lagrangian function; L ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k ) represents the sub-Lagrangian function; Q j represents the reactive power of the jth device (or node); j represents the dual variable of the jth device or subproblem; Q k The reactive power of the kth device or node; λ k represents the dual variable of the kth device or subproblem; ρ is the penalty parameter; A j and A k is the coefficient matrix, which is related to the node admittance matrix and the connection relationship; Q l,j is the reactive network loss of the jth device; b j 、b k Contains information such as load reactive power demand and reactive power injection of other nodes;

[0064] S33. Iterative solution, each sub-problem is solved locally to obtain the local optimal solution, the method is: ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k )About Q j , Q k Taking the derivative, we get:

[0065]

[0066] Solving the above equations yields the local optimal solution Q * j , Q * k ,During the solution process, parameters such as step size are determined according to the ,operating characteristics and constraints of the equipment;

[0067] Update the dual variables through information exchange: Information interaction is achieved through communication networks (such as high-speed Ethernet, wireless communication, etc.), and each node sends the local calculation results (Q * j , Q * k ) is sent to other nodes, and the information received from other nodes is used to update the dual variables; the communication protocol adopts a reliable communication protocol to ensure the accuracy and integrity of data transmission;

[0068] Repeat the above steps until the convergence condition is met. The convergence condition is set as the difference between two adjacent iterative solutions is less than the set threshold, or the change in the value of the augmented Lagrangian function is less than the set threshold, etc.

[0069] Furthermore, the closed-loop control strategy and the steps of evaluating the optimization effect include:

[0070] S41. Results are sent and control is executed. The optimal reactive power output setting value of each device at each moment obtained by reactive power optimization is sent to the corresponding control unit through the high-speed communication network. After receiving the instruction, the control unit controls the device to output the corresponding reactive power;

[0071] The photovoltaic inverter changes the phase or amplitude of the inverter output voltage according to the received reactive power setting value by adjusting the trigger pulse width or frequency of the internal power switch device, thereby outputting the appropriate reactive current;

[0072] Micro gas turbines adjust reactive power output by adjusting the excitation current or changing the output power of the prime mover.

[0073] The reactive power compensation device controls the switching of switches according to the instructions, and switches the corresponding capacitor units to adjust the reactive power output. During the control execution process, the control unit monitors the actual output of the device in real time to ensure that the device outputs reactive power according to the set value.

[0074] S42. Optimization effect evaluation, through the real-time monitoring system of various sensors to collect the distribution network loss, node voltage, power factor and other operating parameters;

[0075] Among them, the network loss is measured by the power sensors installed at the key nodes of the power grid to measure the total input power and the total output power, and the difference between the two is the network loss;

[0076] The node voltage is measured by a voltage transformer;

[0077] The power factor is based on the measured active and reactive power through: Calculated, where PF mon,t is the power factor, P mon,tis the measured active power; Q mon,t is the measured reactive power;

[0078] S43. Calculation of effect evaluation indicators, including network loss reduction rate, voltage qualification rate, and power factor improvement degree, etc. These indicators are used to comprehensively evaluate the optimization effect and determine whether the expected goal has been achieved;

[0079] Among them, the network loss reduction rate Pl,b is the network loss before optimization; Pl,m,t represents the network loss power of the power system obtained through real-time monitoring after the optimization measures are implemented;

[0080] Voltage pass rate N vp Indicates the number of nodes whose voltage is within the qualified range; Nt indicates the total number of nodes;

[0081] Indicators such as power factor improvement degree ΔPF, ΔPF=PF mon,t -PF b , PF b Indicates the power factor before optimization;

[0082] S44. Adjust model parameters. According to the evaluation results, if the optimization effect does not meet expectations, analyze the reasons.

[0083] If it is found that the prediction of PV power generation or load demand in the model is inaccurate, the probability distribution parameters of PV power generation and load are refitted according to real-time data;

[0084] If the constraint conditions are not set properly, such as voltage constraints are often exceeded, adjust the upper and lower limits of the voltage constraints appropriately;

[0085] If certain cost factors have a greater impact in reality, adjust the cost coefficient, etc.;

[0086] Then the optimization calculation is performed again and the next optimization cycle is entered to achieve continuous optimization and improvement.

[0087] In summary, due to the adoption of the above technical solution, the beneficial technical effects of the invention are:

[0088] Improve system stability. By using the Monte Carlo simulation method combined with the mixed integer linear programming (MILP) model to establish a day-ahead optimization model, the uncertainty of photovoltaic power generation and load is fully considered, and the system operation status can be predicted more accurately. In dealing with the problem of photovoltaic power generation being affected by light intensity and temperature, a large number of samples are generated by simulating different light intensity and temperature probability distributions, providing a more reliable data basis for system optimization, so as to formulate reasonable photovoltaic inverter and micro gas turbine (MT) output strategies in advance, effectively respond to power fluctuations, maintain system voltage stability, avoid voltage oscillations and other unstable phenomena caused by intermittent and volatile photovoltaic power generation, enhance the system's ability to resist various interferences, and ensure the stable operation of the system under different working conditions.

[0089] The intraday rolling optimization mechanism uses the real-time data collected by the DPMU device to dynamically adjust the system at fixed time intervals. Combined with the prediction of the system operation in the next 1-2 hours, the photovoltaic power generation and load demand information are updated in real time, and the reactive power output strategy is optimized in time, so that the system can quickly adapt to the dynamic changes in actual operation, such as sudden increase in load or sudden decrease in light intensity, to ensure that the system always maintains a stable operating state, and improve the reliability and robustness of the system operation.

[0090] To reduce operating costs, the optimization model recently adopted the objective function of minimizing the distribution system network loss and photovoltaic operation cost, and optimized the output strategy of photovoltaic inverters and micro gas turbines under constraints. Through precise calculation and optimization, unnecessary power loss and network loss costs were reduced. At the same time, the photovoltaic operation was reasonably arranged, and the unit photovoltaic power operation cost was taken into account, so that the photovoltaic could operate in the most economical state, which improved the utilization efficiency of photovoltaic power generation and reduced the overall operation cost.

[0091] Intraday rolling optimization avoids the irrational flow of reactive power and further reduces network losses by adjusting the reactive output strategy in real time. The reactive output of photovoltaic inverters and micro gas turbines is adjusted in time according to real-time data, so that while meeting the reactive power demand of the system, it reduces the energy loss in the reactive transmission process, thereby reducing operating costs. In addition, by optimizing the switching of reactive compensation equipment, excessive switching or irrational operation of equipment is avoided, the service life of the equipment is extended, and the cost of equipment maintenance and replacement is indirectly reduced.

[0092] Improve the quality of power. In the reactive power optimization stage, the model is constructed with the goal of minimizing reactive network loss, fully considering the reactive output of photovoltaic inverters and micro gas turbines, and using the alternating direction multiplier method (ADMM) to coordinate reactive resources in the distribution network. By optimizing the reactive power flow distribution, the flow of reactive current in the power grid is reduced, thereby reducing the impact of reactive network loss on the system, effectively improving the voltage quality, keeping the node voltage within a reasonable range, reducing voltage fluctuations and flicker, improving the power supply quality, and meeting the user's requirements for power quality.

[0093] In the implementation and feedback phase of the results, the reactive power is monitored and adjusted in real time through a closed-loop control strategy to ensure the effective implementation of the reactive power optimization results. The real-time monitoring system is used to collect operating parameters such as network loss, voltage, and power factor, calculate evaluation indicators, and adjust the optimization model parameters based on the evaluation results. This continuous optimization and improvement mechanism can timely discover and solve power quality problems, continuously improve power quality, and provide users with a more stable and high-quality power supply.

[0094] Enhance system adaptability and optimization accuracy. The rolling optimization mechanism of this technology can continuously adjust the optimization strategy according to the actual operation of the system, and adapt to the changing laws of load and photovoltaic power generation in different time periods. Whether it is long-term seasonal changes or short-term weather changes, the system can dynamically adjust the reactive power output strategy and optimize the model parameters to maintain good operating performance, enhancing the system's adaptability to complex and changing environments.

[0095] The optimization effect is evaluated through a real-time monitoring system, and the optimization model parameters are adjusted according to the evaluation results, thus achieving self-correction and improvement of the optimization model. This feedback adjustment mechanism makes the optimization process more accurate, can continuously improve the optimization accuracy, and make the system always run in the optimal state, further improving the overall performance and economic benefits of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 A flowchart of the reactive power optimization method for high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution;

[0097] Figure 2 A flow chart of the methodology established for the day-ahead optimization model;

[0098] Figure 3 A flow chart of a method for dynamic adjustment based on the optimization results of the day-ahead optimization model and the real-time data collected by the DPMU device;

[0099] Figure 4 A flow chart of the method for distributed optimization using the alternating direction method of multipliers (ADMM);

[0100] Figure 5Flowchart of closed-loop control strategy and method for evaluating optimization effect. DETAILED DESCRIPTION

[0101] In order to make the purpose, technical solution and advantages of the invention more clear, the invention is further described in detail below in combination with the embodiments. It should be understood that the specific embodiments described here are only used to explain the invention and are not used to limit the invention.

[0102] like Figure 1 The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution is shown, and includes the following steps:

[0103] S1. Establishment of the day-ahead optimization model. The Monte Carlo simulation method is combined with the mixed integer linear programming (MILP) model to simulate the uncertainty of photovoltaic power generation and load to achieve the establishment of the day-ahead optimization model. The day-ahead optimization model is used to predict and optimize the output strategy of photovoltaic inverters and micro gas turbines (MT) to minimize the distribution system network loss and photovoltaic operation cost.

[0104] S2. Intraday rolling optimization: Based on the day-ahead optimization results, the real-time data collected by the DPMU device is used to make dynamic adjustments through the distributed computing method of the distribution network partitions. With the help of the rolling optimization mechanism, the photovoltaic power generation and load demand information are updated in real time, and the current and future system operation conditions are comprehensively considered to optimize the reactive power output strategy to cope with power fluctuations and ensure stable and economical operation of the system.

[0105] S3. Reactive power optimization: Based on the intraday rolling optimization, the reactive power optimization model is constructed with the goal of minimizing reactive power network loss, taking full account of the reactive output of photovoltaic inverters and micro gas turbines, and using the alternating direction multiplier method (ADMM) as a distributed optimization algorithm to coordinate reactive resources in the distribution network, optimize reactive power flow distribution, reduce active power loss and maintain a good voltage level.

[0106] S4. Implementation and feedback of results: Through closed-loop control strategy, the reactive power optimization results are sent to each control unit in the distribution network to perform reactive power optimization control; the optimization effect is evaluated by real-time monitoring system, and the optimization model parameters are adjusted according to the evaluation results to improve the optimization accuracy and adaptability, and achieve continuous optimization and improvement.

[0107] In S1, the Monte Carlo simulation method is combined with the principle of mixed integer linear programming model:

[0108] The Monte Carlo simulation method is used to deal with the uncertainty of photovoltaic power generation and load. Its basic idea is to generate a large number of photovoltaic power generation and load value samples according to a known probability distribution model through random sampling. For example, photovoltaic power generation may be affected by factors such as light intensity and temperature. Assuming that light intensity and temperature have a certain probability distribution (such as normal distribution, etc.), different light intensity and temperature values ​​are randomly generated according to these distributions, and then the corresponding photovoltaic power generation samples are calculated.

[0109] The mixed integer linear programming (MILP) model is used to optimize the system under various constraints; its goal is to minimize the distribution system losses and photovoltaic operation costs.

[0110] like Figure 2 As shown, the steps of establishing the day-ahead optimization model include:

[0111] S11. Establish a probability distribution model of photovoltaic power generation and load. For photovoltaic power generation, its power generation model is expressed as:

[0112]

[0113] P pv is the photovoltaic power generation; P stc It is the photovoltaic power generation under standard test conditions. It is the rated power generation of photovoltaic panels under specific standard light intensity and temperature, and is a known fixed parameter. stc is the light intensity under standard test conditions, which is a constant; k is the power temperature coefficient, which is a negative value and indicates the rate at which the power generated by the photovoltaic panel decreases when the temperature rises; T stc is the panel temperature under standard test conditions; I is the actual light intensity; T is the actual temperature;

[0114] For load, the probability distribution function is analyzed and fitted based on the historical load data accumulated over a long period of time. The load presents different rules in different time periods (such as weekdays and holidays, daytime and nighttime, etc.), and needs to be accurately described by piecewise fitting or mixed distribution model. The load is expressed as:

[0115]

[0116] where u L is the mean value of the load, representing the average load demand in the region during a specific period of time. It is the variance of the load, reflecting the range of variation of load demand.

[0117] S12. Construct a mixed integer linear programming (MILP) model, and the objective function is expressed as:

[0118] min C=Cl +C pv ;

[0119]

[0120] min C represents the objective function; G is the optimization time period; N is the number of nodes in the distribution network; C l is the network loss cost of the distribution system; C pv is the photovoltaic operation cost; P l,t,i is the network loss power of node i during period t; c l is the unit network loss cost, that is, the economic cost generated by each unit of power loss; M is the number of photovoltaic power sources; c PV is the unit photovoltaic power operating cost, including the depreciation cost of photovoltaic panels, maintenance cost, inverter loss cost, etc.; P PV,t,j is the output power of the jth photovoltaic power source in period t;

[0121] The constraints include: power balance constraint, photovoltaic power output constraint, micro gas turbine output constraint, voltage constraint, and line flow constraint;

[0122] Power balance constraints:

[0123]

[0124] K is the number of micro turbines (MT); P MT,t,k is the output power of the kth micro gas turbine in period t; this constraint ensures that at any time, the sum of the output power of all power sources in the system (photovoltaic power sources and micro gas turbines) is equal to the sum of the load demand power and the network loss power, reflecting the principle of energy conservation; P L,t At time t, the load power of the distribution network;

[0125] Photovoltaic power output constraint: P PV,min,j ≤P PV,t,j ≤P PV,max,j ;

[0126] P PV,min,j and P PV,max,j are the minimum and maximum output limits of the jth photovoltaic power source respectively; the minimum output limit may be affected by factors such as the physical characteristics of the photovoltaic panel (such as the output capacity under adverse conditions such as low light or high temperature) and the minimum working power of the inverter; the maximum output limit is mainly determined by the rated capacity of the photovoltaic panel and the maximum conversion capacity of the inverter;

[0127] Micro gas turbine output constraints:

[0128] P MT,min,k ≤P MT,t,k ≤P MT,max,k ;

[0129] P MT,min,k and P MT,max,k are the minimum and maximum output limits of the kth micro gas turbine respectively; the minimum output limit of a micro gas turbine is usually related to the minimum requirements for its startup and stable operation, while the maximum output limit depends on its design capacity, fuel supply capacity, bearing capacity of mechanical components, etc.;

[0130] Voltage constraints, line flow constraints:

[0131] V min,i ≤V t,i ≤V max,i ;I t,ij ≤I max,ij ;

[0132] V min,i and V max,i are the upper and lower limits of the voltage at node i;

[0133] V t,i is the voltage of node i at time period t; I t,ij is the current of line ij during period t; I max,ij is the maximum allowable current of line ij.

[0134] In S2, the principle of dynamic adjustment based on the optimization result of the day-ahead optimization model and the real-time data collected by the DPMU device is:

[0135] The DPMU device can accurately collect electrical quantity information of each node in the distribution network at a high sampling frequency (such as dozens of times per second or higher), including photovoltaic power generation, load demand, voltage, etc. These real-time data accurately reflect the actual operating status of the system at the current moment, including the impact of real-time changes in light intensity on photovoltaic power generation, real-time fluctuations in user loads, and real-time dynamics of grid voltage. Taking the day-ahead optimization results as the initial reference, due to the influence of various factors in actual operation (such as sudden weather changes, uncertainty in user behavior, etc.), there must be differences between the system operation and the day-ahead forecast. Through the rolling optimization mechanism, the latest real-time data is used to dynamically adjust the day-ahead optimization results at regular intervals (such as 15 minutes or 30 minutes). In the adjustment process, not only the actual situation at the current moment is considered, but also the prediction of the system operation in the future (such as 1-2 hours) is combined to comprehensively decide the reactive power output strategy of the photovoltaic inverter and micro gas turbine and the switching status of the reactive compensation equipment, so that the system can quickly adapt to real-time changes and ensure the stability of the system operation in a dynamic environment (such as maintaining voltage stability, avoiding power oscillation, etc.) and economy (such as reducing operating costs and reducing network losses, etc.).

[0136] like Figure 3As shown, in S2, the step of dynamically adjusting based on the optimization result of the day-ahead optimization model and the real-time data collected by the DPMU device includes:

[0137] S21. Real-time data acquisition: The DPMU device collects data such as photovoltaic power generation, load demand, and voltage of each node in the distribution network in real time. The DPMU device uses high-precision sensors and fast data processing algorithms to accurately measure the amplitude, phase, and other information of electrical quantities. For example, for the measurement of photovoltaic power generation, it detects the power at the output end of the photovoltaic inverter, and obtains the accurate real-time power value through conversion and calculation; for the measurement of load demand, it is obtained through the power sensor installed on the load side; for voltage measurement, the high voltage is directly converted into a low voltage signal suitable for measurement and processing through the voltage transformer, and then the analog-to-digital converter (ADC) inside the DPMU converts the analog signal into a digital signal for processing; the real-time collected data is transmitted to the control center in a timely manner, providing an accurate data basis for subsequent rolling optimization calculations;

[0138] S22. Rolling optimization calculation. During intraday rolling optimization, the objective function is:

[0139] min C r ≤C l,r +C PV,r +C adj ;

[0140]

[0141] min C r During intraday rolling optimization, the objective function value; C l,r is the real-time network loss cost; C PV,r is the real-time photovoltaic operation cost; C adj is the adjustment cost, C adj According to the reactive power adjustment and related cost coefficient, the reactive power adjustment is the change in reactive power of the photovoltaic inverter or other reactive power regulating equipment compared with the previous optimization result. PV,r,t,j is the actual output power of the jth node at the current time t; P l,r,t,i is the real-time network loss power of node i at the current time t;

[0142] S23. Constraint update:

[0143] Among them, the power balance constraint is updated as:

[0144]

[0145] P MT,r,t,kis the real-time output power of the kth micro gas turbine at the current moment, obtained by the real-time monitoring device of the micro gas turbine; P L,r,t At time t, the actual load power of the distribution network;

[0146] S24. A rolling optimization mechanism performs a rolling optimization calculation every the same time period (such as 15 minutes or 30 minutes, etc.). In each calculation, firstly, the calculation is performed based on the real-time data at the current moment and the prediction of the operation status in the future (such as the next 1-2 hours). Then, based on the prediction results and real-time data, the reactive output strategy of the photovoltaic inverter and the micro gas turbine is re-optimized, and the switching of the reactive compensation equipment is adjusted. For example, if it is predicted that the photovoltaic power generation will increase and the load demand will decrease in the future, according to the reactive power balance principle, the reactive output of the photovoltaic inverter is appropriately adjusted to generate more reactive power, while reducing the reactive output of the micro gas turbine or adjusting the switching state of the reactive compensation equipment to maintain the reactive balance of the system and ensure that the voltage is stable within a reasonable range.

[0147] In S3, an optimization model with the goal of minimizing reactive power losses is established and the principle of the alternating direction multiplier method (ADMM) is adopted:

[0148] Reactive network loss causes additional energy loss in the distribution network, affecting the power quality and system operation efficiency. Unreasonable reactive power distribution will cause the current to produce reactive components. This reactive current will produce active loss (i.e. reactive network loss) on the line resistance during transmission, and will also cause problems such as voltage fluctuation and flicker. By optimizing the reactive output strategy and reasonably allocating the reactive power of equipment such as photovoltaic inverters and micro gas turbines, the flow of reactive current can be reduced and reactive network loss can be reduced. The alternating direction multiplier method (ADMM) is based on the decomposition-coordination idea, which decomposes the reactive optimization problem of complex distribution networks into multiple relatively simple sub-problems, each of which corresponds to a distributed power source or control node. In each sub-problem, the device performs local optimization calculations based on its own operating characteristics and the interactive information with adjacent nodes, and then coordinates the sub-problem solutions through information interaction (such as communication networks) to gradually approach the global optimal solution. The dual variables and augmented Lagrangian functions are introduced to transform the original optimization problem into a decomposable form, so that the sub-problems can be solved independently. The dual variable update mechanism is used to ensure the coordination and consistency between the sub-problems, and finally achieve the global reactive power optimization goal.

[0149] In S3, the model constructed by reactive power optimization with the goal of minimizing reactive power network loss is expressed as:

[0150]

[0151]

[0152] Among them, X i is the reactance of node i, is the reactive current of node i during period t; I t,i Current measured in real time; is the power factor angle measured in real time;

[0153] The constraints include: reactive power balance constraint and 3 equipment reactive output constraint;

[0154] Reactive power balance constraints:

[0155]

[0156] Q PV,t,j At time t, the reactive output of the jth PV inverter is Q MT,t,k At time t, the reactive output of the jth micro gas turbine is Q C,t,l At time t, the reactive output of the lth reactive compensation device; Q l,t,i At time t, the reactive network loss of node i; L is the number of reactive compensation devices; The power factor angle of reactive load demand; Q L,t is the reactive load demand during period t; P L,t is the active power of the load;

[0157] Equipment reactive power output constraints:

[0158] Q Q PV,min,j ≤Q PV,t,j ≤Q Q PV,max,j ;

[0159] Q Q MT,min,k ≤Q MT,t,k ≤Q Q MT,max,k ;

[0160] Q Q C,min,l ≤Q C,t,l ≤Q Q C,max,l ;

[0161] Q Q PV,min,j , Q Q PV,max,j They represent the lower and upper limits of the reactive power output of the j-th photovoltaic inverter respectively; Q Q MT,min,k , Q Q MT,max,k are the lower and upper limits of the reactive power output of the kth micro gas turbine respectively; Q Q C,min,l , Q Q C,max,lare respectively the lower limit and upper limit of the reactive output of the lth reactive compensation device.

[0162] like Figure 4 As shown in Figure 1, the steps of using the alternating direction method of multipliers (ADMM) for distributed optimization include:

[0163] S31. Decompose the optimization problem. Decompose the reactive power optimization problem of the entire distribution network into multiple sub-problems. Each sub-problem corresponds to a distributed power source or control node. For a photovoltaic power source, its local sub-problem is to meet its own reactive power output constraint (Q Q PV,min,j ≤Q PV,t,j ≤Q Q PV,max,j ) and interacting with other nodes (reactualizing the reactive power interaction relationship with other nodes through reactive power balance constraints), minimize its own contribution to reactive network loss;

[0164] S32. Introduce dual variables and augmented Lagrangian function to construct augmented Lagrangian function:

[0165]

[0166] L ρ,j (Q,λ) represents the total Lagrangian function; L ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k ) represents the sub-Lagrangian function; Q j represents the reactive power of the jth device (or node); j represents the dual variable of the jth device or subproblem; Q k The reactive power of the kth device (or node); λ k represents the dual variable of the kth device or subproblem; ρ is the penalty parameter; A j and A k is the coefficient matrix, which is related to the node admittance matrix and the connection relationship; Q l,j is the reactive network loss of the jth device; b j 、b k Contains information such as load reactive power demand and reactive power injection of other nodes;

[0167] S33. Iterative solution, each sub-problem is solved locally to obtain the local optimal solution, the method is: ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k )About Qj , Q k Taking the derivative, we get:

[0168]

[0169] Solving the above equations yields the local optimal solution Q * j , Q * k ,During the solution process, the step size and other parameters are determined according to the ,operating characteristics and constraints of the equipment. For example, for the photovoltaic ,inverter, the power conversion efficiency and response speed and other characteristics need to be considered to ,determine the appropriate step size to ensure the effectiveness and stability of the ,solution;

[0170] Update the dual variables through information exchange: Information interaction is achieved through communication networks (such as high-speed Ethernet, wireless communication, etc.). Each node sends local calculation results (Q*j, Q*k) to other nodes and receives information from other nodes for dual variable update. The communication protocol uses a reliable communication protocol to ensure the accuracy and integrity of data transmission.

[0171] Repeat the above steps until the convergence condition is met. The convergence condition is set as the difference between two adjacent iterative solutions is less than the set threshold, or the change in the value of the augmented Lagrangian function is less than the set threshold, etc.

[0172] In S4, the closed-loop control strategy and the principle of evaluating the optimization effect are as follows: the closed-loop control strategy transmits the reactive power optimization results to each control unit in the distribution network in real time through the communication network, so that the control unit controls the equipment according to the optimized reactive power output strategy, and realizes real-time and accurate regulation of the reactive power of the distribution network. The real-time monitoring system is used to continuously monitor the operating parameters of the distribution network (such as network loss, voltage quality, power factor, etc.), and these actual operating parameters are compared with the optimization target to judge the effectiveness of the optimization strategy. The optimization model parameters are adjusted according to the evaluation results to form a continuous cycle optimization process, thereby continuously improving the optimization accuracy and adaptability, ensuring that the reactive power optimization effect of the system continues to improve, and effectively adapting to various changes in actual operation.

[0173] like Figure 5 As shown, the closed-loop control strategy and the steps of evaluating the optimization effect include:

[0174] S41. Results are sent and controlled. The optimal reactive output setting value of each device (photovoltaic inverter, micro gas turbine, reactive compensation device, etc.) obtained by reactive power optimization at each moment is sent to the corresponding control unit through the high-speed communication network. After receiving the instruction, the control unit controls the device to output the corresponding reactive power;

[0175] The photovoltaic inverter changes the phase or amplitude of the inverter output voltage according to the received reactive output setting value by adjusting the trigger pulse width or frequency of the internal power switch device (such as IGBT, etc.), thereby outputting the appropriate reactive current;

[0176] Micro gas turbines adjust reactive power output by adjusting the excitation current or changing the output power of the prime mover.

[0177] Reactive power compensation equipment (such as capacitor banks) controls the switching of switches according to instructions, switching on and off the corresponding capacitor units to adjust the reactive power output; during the control execution process, the control unit monitors the actual output of the equipment in real time to ensure that the equipment outputs reactive power according to the set value, for example, by measuring the actual output reactive current and voltage through sensors such as current transformers and voltage transformers, calculating the actual reactive power and comparing it with the set value, and adjusting the control signal in time if there is a deviation;

[0178] S42. Evaluation of optimization effect: various sensors (such as power sensors, voltage transformers, current transformers, etc.) in the real-time monitoring system are used to collect the network loss, node voltage, power factor and other operating parameters of the distribution network in real time;

[0179] Among them, the network loss is measured by the power sensors installed at the key nodes of the power grid to measure the total input power and the total output power, and the difference between the two is the network loss;

[0180] The node voltage is measured by a voltage transformer;

[0181] The power factor is based on the measured active power and reactive power (measured by power sensors) through: Calculated, where PF mon,t is the power factor, P mon,t is the measured active power; Q mon,t is the measured reactive power;

[0182] S43. Calculation of effect evaluation indicators, including network loss reduction rate, voltage qualification rate, and power factor improvement degree, etc. These indicators are used to comprehensively evaluate the optimization effect and determine whether the expected goals have been achieved, such as whether the network loss reduction rate has reached the set percentage, whether the voltage qualification rate meets the power supply quality requirements, etc.;

[0183] Among them, the network loss reduction rate P l,b is the network loss before optimization; P l,m,t It indicates the power loss of the power system obtained through real-time monitoring after the optimization measures are implemented;

[0184] Voltage pass rate N vp Indicates the number of nodes whose voltage is within the qualified range; Nt Indicates the total number of nodes;

[0185] Indicators such as power factor improvement degree ΔPF, ΔPF=PF mon,t -PF b , PF b Indicates the power factor before optimization;

[0186] S44. Adjust model parameters. According to the evaluation results, if the optimization effect is found to be not as expected (such as the network loss reduction rate is not obvious, the voltage qualification rate is low, etc.), analyze the reasons;

[0187] If it is found that the prediction of PV power generation or load demand in the model is inaccurate, the probability distribution parameters of PV power generation and load can be refitted based on real-time data;

[0188] If the constraint conditions are not set properly, such as voltage constraints are often exceeded, the upper and lower limits of the voltage constraints can be adjusted appropriately;

[0189] If certain cost factors have a greater impact in practice, the cost coefficient can be adjusted;

[0190] Then the optimization calculation is performed again and the next optimization cycle is entered to achieve continuous optimization and improvement. For example, if the network loss reduction rate is lower than the expected target, and analysis finds that this is due to a large deviation between the actual photovoltaic power generation and the predicted value, then the probability distribution parameters of the photovoltaic power generation are re-estimated using the latest historical data, and the day-ahead optimization, intraday rolling optimization, and reactive power optimization steps are performed again, and the optimization strategy is continuously adjusted until a satisfactory optimization effect is achieved. At the same time, in the process of adjusting parameters, it is necessary to record the content of each adjustment and the corresponding optimization results in order to summarize experience and further improve the accuracy and adaptability of the optimization model.

[0191] The above description is a preferred embodiment of the invention and is not intended to limit the invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the invention should be included in the protection scope of the invention.

Claims

1. A reactive power optimization method for a high-proportion photovoltaic distribution network taking into account random day-ahead and intra-day rolling distribution, characterized in that: The following steps are involved: S1. Establishment of the day-ahead optimization model. The Monte Carlo simulation method is combined with the mixed integer linear programming (MILP) model to simulate the uncertainty of photovoltaic power generation and load to achieve the establishment of the day-ahead optimization model. The day-ahead optimization model is used to predict and optimize the output strategy of photovoltaic inverters and micro gas turbines (MT) to minimize the distribution system network loss and photovoltaic operation cost. S2. Intraday rolling optimization, based on the day-ahead optimization results, using the real-time data collected by the DPMU device, and through the distribution network partition distributed computing method, dynamic adjustment is performed; With the help of rolling optimization mechanism, the photovoltaic power generation and load demand information are updated in real time, the current and future system operation conditions are comprehensively considered, and the reactive power output strategy is optimized; S3. Reactive power optimization: Based on the intraday rolling optimization, the reactive power optimization model is constructed with the goal of minimizing reactive power network loss, fully considering the reactive power output of photovoltaic inverters and micro gas turbines, using the alternating direction multiplier method (ADMM) as a distributed optimization algorithm, coordinating reactive resources in the distribution network, and optimizing reactive power flow distribution; S4. Result implementation and feedback: Through the closed-loop control strategy, the reactive power optimization results are sent to each control unit in the distribution network to perform reactive power optimization control; the optimization effect is evaluated using the real-time monitoring system, and the optimization model parameters are adjusted according to the evaluation results.

2. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution according to claim 1 is characterized in that: The steps of establishing the day-ahead optimization model include: S11. Establish a probability distribution model of photovoltaic power generation and load. For photovoltaic power generation, its power generation model is expressed as: P pv is the photovoltaic power generation; P stc It is the photovoltaic power generation under standard test conditions. It is the rated power generation of photovoltaic panels under specific standard light intensity and temperature, and is a known fixed parameter. stc is the light intensity under standard test conditions, I stc is a constant; k is the power temperature coefficient, which is a negative value, indicating the rate at which the power generated by the photovoltaic panel decreases when the temperature rises; T stc is the panel temperature under standard test conditions; I is the actual light intensity; T is the actual temperature; For load, the probability distribution function is analyzed and fitted based on the historical load data accumulated over a long period of time. The load shows different rules in different time periods, and it is necessary to use segmented fitting or mixed distribution model to accurately describe it. The load is expressed as: where u L is the mean value of the load, representing the average load demand in the region during a specific period of time. is the variance of the load, reflecting the range of variation in load demand; S12. Construct a mixed integer linear programming (MILP) model, and the objective function is expressed as: my C=C l +C pv ; min C represents the objective function; G is the optimization time period; N is the number of nodes in the distribution network; C l is the network loss cost of the distribution system; C pv is the photovoltaic operation cost; P l,t,i is the network loss power of node i during period t; c l is the unit network loss cost, that is, the economic cost generated by each unit of power loss; M is the number of photovoltaic power sources; c PV is the unit photovoltaic power operating cost, including the depreciation cost of photovoltaic panels, maintenance cost, and inverter loss cost; P PV,t,j is the output power of the jth photovoltaic power source in period t; The constraints include: power balance constraint, photovoltaic power output constraint, micro gas turbine output constraint, voltage constraint, and line flow constraint; Power balance constraints: K is the number of micro turbines (MT); P MT,t,k is the output power of the kth micro gas turbine in period t; this constraint ensures that at any time, the sum of the output power of all power sources in the system is equal to the sum of the load demand power and the network loss power, reflecting the principle of energy conservation; P L,t At time t, the load power of the distribution network; Photovoltaic power output constraint: P PV,min,j ≤P PV,t,j ≤P PV,max,j ; P PV,min,j and P PV,max,j are the minimum and maximum output limits of the jth photovoltaic power source respectively; the minimum output limit may be affected by the physical characteristics of the photovoltaic panel and the minimum operating power factor of the inverter; the maximum output limit is mainly determined by the rated capacity of the photovoltaic panel and the maximum conversion capacity of the inverter; Micro gas turbine output constraints: P MT,min,k ≤P MT,t,k ≤P MT,max,k ; P MT,min,k and P MT,max,k are the minimum and maximum output limits of the kth micro gas turbine respectively; the minimum output limit of a micro gas turbine is usually related to the minimum requirements for its startup and stable operation, while the maximum output limit depends on its design capacity, fuel supply capacity, and bearing capacity of mechanical components; Voltage constraints, line flow constraints: V min,i ≤V t,i ≤V max,i ;I t,ij ≤I max,ij ; V min,i and V max,i are the upper and lower limits of the voltage at node i; V t,i is the voltage of node i at time period t; I t,ij is the current of line ij during period t; I max,ij is the maximum allowable current of line ij.

3. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution according to claim 1 is characterized in that: In S2, the step of dynamically adjusting based on the optimization result of the day-ahead optimization model and the real-time data collected by the DPMU device includes: S21. Real-time data collection: The DPMU device collects the photovoltaic power generation, load demand, and voltage data of each node in the distribution network in real time. The DPMU device uses high-precision sensors and fast data processing algorithms to accurately measure the amplitude and phase information of electrical quantities; S22. Rolling optimization calculation. During intraday rolling optimization, the objective function is: my C r ≤C l,r +C PV,r +C adj ; min C r During intraday rolling optimization, the objective function value; C l,r is the real-time network loss cost; C PV,r is the real-time photovoltaic operation cost; C adj is the adjustment cost, C adj According to the reactive power adjustment and related cost coefficient, the reactive power adjustment is the change in reactive power of the photovoltaic inverter or other reactive power regulating equipment compared with the previous optimization result. PV,r,t,j is the actual output power of the jth node at the current time t; P l,r,t,i is the real-time network loss power of node i at the current time t; S23. Constraint update: Among them, the power balance constraint is updated as: P MT,r,t,k is the real-time output power of the kth micro gas turbine at the current moment, obtained by the real-time monitoring device of the micro gas turbine; P L,r,t At time t, the actual load power of the distribution network; S24. Rolling optimization mechanism, which performs rolling optimization calculations every same time period. In each calculation, it first calculates based on the real-time data at the current moment and the forecast of the operating conditions in the future. Then, based on the forecast results and real-time data, it re-optimizes the reactive power output strategy of the photovoltaic inverter and micro gas turbine, and adjusts the switching of reactive compensation equipment.

4. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and rolling distribution within the day according to claim 1 is characterized in that: In S3, the model constructed by reactive power optimization with the goal of minimizing reactive power network loss is expressed as: Among them, X i is the reactance of node i, is the reactive current of node i during period t; I t,i Current measured in real time; is the power factor angle measured in real time; Constraints include: reactive power balance constraints and equipment reactive output constraints; Reactive power balance constraints: Q PV,t,j At time t, the reactive output of the jth PV inverter is Q MT,t,k At time t, the reactive output of the jth micro gas turbine is Q C,t,l At time t, the reactive output of the lth reactive compensation device; Q l,t,i At time t, the reactive network loss of node i; L is the number of reactive compensation devices; The power factor angle of reactive load demand; Q L,t is the reactive load demand during period t; P L,t is the active power of the load; Equipment reactive power output constraints: Q Q PV,min,j ≤Q PV,t,j ≤Q Q PV,max,j ; Q Q MT,min,k ≤Q MT,t,k ≤Q Q MT,max,k ; Q Q C,min,l ≤Q C,t,l ≤Q Q C,max,l ; Q Q PV,min,j , Q Q PV,max,j They represent the lower and upper limits of the reactive power output of the j-th photovoltaic inverter respectively; Q Q MT,min,k , Q Q MT,max,k are the lower and upper limits of the reactive power output of the kth micro gas turbine respectively; Q Q C,min,l , Q Q C,max,l are respectively the lower limit and upper limit of the reactive output of the lth reactive compensation device.

5. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution according to claim 1 is characterized in that: In S3, the steps of using the alternating direction method of multipliers (ADMM) for distributed optimization include: S31. Decompose the optimization problem, decompose the reactive power optimization problem of the entire distribution network into multiple sub-problems, each sub-problem corresponds to a distributed power source or control node; for a certain photovoltaic power source, its local sub-problem is to minimize its contribution to reactive network loss under the conditions of satisfying its own reactive output constraints and exchanging information with other nodes; S32. Introduce dual variables and augmented Lagrangian functions to construct augmented Lagrangian functions: L ρ,j (Q,λ) represents the total Lagrangian function; L ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k ) represents the sub-Lagrangian function; Q j represents the reactive power of the jth device (or node); j represents the dual variable of the jth device or subproblem; Q k The reactive power of the kth device or node; λ k represents the dual variable of the kth device or subproblem; ρ is the penalty parameter; A j and A k is the coefficient matrix, which is related to the node admittance matrix and the connection relationship; Q l,j is the reactive network loss of the jth device; b j , b k Contains reactive power demand of loads and reactive power injection information of other nodes; S33. Iterative solution, each sub-problem is solved locally to obtain the local optimal solution, the method is: ρ,j (Q j ,λ j )、L ρ,k (Q k ,λ k )About Q j , Q k Taking the derivative, we get: Solve the above equations to obtain the local optimal solutions Q*j and Q*k. During the solution process, the step size parameters are determined according to the operating characteristics and constraints of the equipment; Update the dual variables through information exchange: Information interaction is achieved through the communication network. Each node sends the local calculation results (Q*j, Q*k) to other nodes and receives information from other nodes for dual variable update. The communication protocol uses a reliable communication protocol to ensure the accuracy and integrity of data transmission. Repeat the above steps until the convergence condition is met. The convergence condition is set as the difference between two adjacent iterative solutions is less than the set threshold, or the change in the value of the augmented Lagrangian function is less than the set threshold.

6. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and rolling distribution within the day according to claim 1 is characterized in that: The closed-loop control strategy and the steps of evaluating the optimization effect include: S41. Results are sent and controlled, and the optimal reactive output setting value of each device at each moment obtained by reactive optimization is sent to the corresponding control unit through a high-speed communication network; after receiving the command, the control unit controls the device to output the corresponding reactive power; S42. Optimization effect evaluation, through the real-time monitoring system of various sensors to collect the distribution network loss, node voltage, power factor operating parameters; Among them, the network loss is measured by the power sensors installed at the key nodes of the power grid to measure the total input power and the total output power, and the difference between the two is the network loss; The node voltage is measured by a voltage transformer; The power factor is calculated based on the measured active power and reactive power. Calculated, where PF mon,t is the power factor, P mon,t is the measured active power; Q mon,t is the measured reactive power; S43. Calculation of effect evaluation indicators, including network loss reduction rate, voltage qualification rate and power factor improvement index, through which the optimization effect is comprehensively evaluated to determine whether the expected goal is achieved; S44. Adjust model parameters. According to the evaluation results, if the optimization effect does not meet expectations, analyze the reasons. If it is found that the prediction of PV power generation or load demand in the model is inaccurate, the probability distribution parameters of PV power generation and load are refitted according to real-time data; If the constraint conditions are not set properly, such as voltage constraints are often exceeded, adjust the upper and lower limits of the voltage constraints appropriately; If some cost factors have a greater impact in practice, adjust the cost coefficient; Then the optimization calculation is performed again and the next optimization cycle is entered to achieve continuous optimization and improvement.

7. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and rolling distribution within the day according to claim 6 is characterized in that: In S41, the photovoltaic inverter changes the phase or amplitude of the inverter output voltage by adjusting the trigger pulse width or frequency of the internal power switch device according to the received reactive output setting value, thereby outputting a suitable reactive current; The micro gas turbine adjusts the reactive power output by adjusting the excitation current or changing the output power mode of the prime mover; The reactive power compensation device controls the switching of switches according to the instructions, and switches the corresponding capacitor units to adjust the reactive power output; during the control execution process, the control unit monitors the actual output of the equipment in real time to ensure that the equipment outputs reactive power according to the set value.

8. The reactive power optimization method for a high-proportion photovoltaic distribution network taking into account the random day-ahead and intra-day rolling distribution according to claim 6 is characterized in that: In S43, the network loss reduction rate P l,b is the network loss before optimization; P l,m,t It indicates the power loss of the power system obtained through real-time monitoring after the optimization measures are implemented; Voltage pass rate N vp Indicates the number of nodes whose voltage is within the qualified range; N t Indicates the total number of nodes; Power factor improvement index ΔPF, ΔPF = PF mon,t -PF b , PF b Indicates the power factor before optimization.

Images