Multi-objective voltage optimization method for high penetration photovoltaic distribution network based on genetic algorithm

Abstract

This invention discloses a multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms, belonging to the field of novel power systems. The invention includes: constructing a heterogeneous regulation equipment model comprising on-load tap-changing transformers and photovoltaic inverters to achieve discrete and continuous regulation; constructing a multi-objective optimization function with voltage deviation, network loss, and equipment operating costs as objectives, and taking distribution network power flow balance and node voltage security as constraints; introducing a penalty term to construct a fitness function, balancing global search and local exploitation capabilities through an adaptive crossover probability mechanism; generating N all-day voltage control schemes based on the heterogeneous regulation equipment model, evaluating different all-day voltage control schemes through the optimized multi-objective optimization model, and obtaining the optimal all-day voltage control scheme through iterative optimization. This invention significantly reduces system voltage deviation, network loss, and equipment operating costs while ensuring the distribution network meets node voltage security constraints.

Description

Technical Field

[0001] This invention relates to the field of novel power systems, specifically to a multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms. Background Technology

[0002] In recent years, the installed capacity of photovoltaic (PV) power generation has continued to grow. This trend has presented unprecedented dual challenges to the voltage control and operational reliability of distribution networks: on the one hand, distributed PV output exhibits significant randomness, volatility, and intermittency, which, after integration, can easily lead to frequent voltage exceedances and increased fluctuations in the distribution network, seriously threatening the power supply quality; on the other hand, the distribution network must maintain operational stability and economy under complex dynamic environments, coordinating various discrete and continuous voltage regulation devices to avoid mechanical fatigue and lifespan reduction caused by excessive regulation. Therefore, research on distribution network voltage regulation under high-proportion DPV integration is of great significance.

[0003] However, traditional distribution network control methods often face a dual bottleneck: at the equipment coordination level, discrete equipment such as on-load tap changers (OLTCs) and continuous equipment such as photovoltaic inverters have vastly different response characteristics, making it difficult for traditional methods to handle the time-dependent coupling and nonlinear constraints between the two, resulting in insufficient control accuracy or excessively high equipment operating costs; at the computational efficiency level, traditional numerical optimization is time-consuming and prone to getting trapped in local optima when dealing with nonlinear optimization problems containing multiple mixed integer variables, making it difficult to output a globally optimal strategy that balances voltage compliance rate and operational economy in real time under complex operating scenarios.

[0004] Gao Fengjun and Guan Xiaotian published "An Optimization Method for Distribution Network Loss Reduction Based on Hybrid Encoding Genetic Algorithm" in *Electric Power and Energy*. This scheme is based on a theoretical model of distribution network line losses. Its physical structure mainly includes the distribution network structure to be optimized and reactive power compensation devices deployed within the network. Distributed power sources (such as photovoltaics) mainly serve as power support units in this model, and a power flow model is constructed using the principle of stochastic fuzzy logic to cope with their output fluctuations. The core contribution of this closest similar scheme is that it proposes a hybrid encoding genetic algorithm based on a hierarchical mechanism and an adaptive single-point crossover strategy, solving the problem of simultaneously and efficiently optimizing discrete variables (such as capacitor switching) and continuous variables (such as power source output) in reactive power compensation of distribution networks.

[0005] However, it still has limitations when dealing with the scenario of "high-penetration photovoltaic voltage quality control". First, the coordination between the regulation dimension and the equipment response characteristics is insufficient. This scheme focuses on reducing losses through reactive power compensation, without fully considering the deep coupling relationship between the on-load tap changer (OLTC) tap adjustment and the surplus capacity of the photovoltaic inverter. When facing instantaneous voltage exceedances caused by high-penetration photovoltaics, it lacks an integrated collaborative framework of "discrete coarse adjustment + continuous fine adjustment", making it difficult to achieve globally optimal support for the voltage reference. Second, the scheme's objectives are mainly focused on economic indicators, lacking quantitative characterization of "voltage fluctuation rate" and "OLTC mechanical action frequency", which are extremely critical in high-penetration scenarios. This may lead to frequent transformer voltage regulation, shortening the operating life of expensive equipment, even if grid losses are reduced. In addition, the scheme mostly uses general mechanisms to handle constraints, lacking nonlinear reinforcement penalty design for severe voltage exceedances. In an environment with highly random photovoltaic output, there may be instantaneous voltage exceedance risks during certain prediction periods, and robustness needs to be improved.

[0006] Existing technologies have the following problems when dealing with voltage optimization control in distribution networks with high penetration rates of distributed photovoltaic (PV) integration:

[0007] (1) There is a problem that it is difficult to balance "model dependence and computational timeliness".

[0008] Traditional research on voltage control in distribution networks has proposed methods such as local control, centralized control, and distributed control. However, these methods typically rely heavily on the accuracy of physical model parameters. With the exponential growth in the scale of distribution network topologies and the number of control devices, traditional numerical optimization algorithms are computationally time-consuming when dealing with the dynamic changes in the complex environment of distributed photovoltaic (DPV), making it difficult to guarantee global optimality in real-time control. Although artificial intelligence methods such as deep reinforcement learning attempt to address the model dependency problem, they still face bottlenecks in practical engineering, including complex training processes, insufficient generalization ability, and low transparency of decision-making logic, making it difficult to meet the requirements for safe operation of power systems.

[0009] (2) There is a problem of "lack of response time coupling" and mismatch of action in the voltage regulating equipment.

[0010] While existing genetic algorithm (GA) research possesses strong parallel search capabilities, it often overlooks the deep coupling between discrete and continuous devices in terms of response characteristics. On-load tap changers (OLTCs), as discrete regulating devices, suffer from significantly increased mechanical fatigue due to frequent switching, making them ill-equipped to handle rapid fluctuations in photovoltaic power. Conversely, while photovoltaic inverters offer continuous and rapid regulation capabilities, they cannot independently cope with large-scale voltage deviations without organic coordination with discrete devices. Existing research, although considering multi-device coordination, still focuses on the economic scheduling of discrete regulation, failing to fully utilize the advantages of continuous reactive power support. This leads to disordered operation of different voltage regulating devices, increasing system losses and operating costs.

[0011] (3) There is insufficient consideration of the dynamic voltage risk caused by high-penetration photovoltaic fluctuations.

[0012] With the increasing penetration of distributed photovoltaic (PV) systems, problems such as reverse power flow and voltage exceeding limits are becoming more frequent. Existing control schemes fail to fully consider the combined effects of various uncertainties, including the significant randomness and volatility of PV output and load forecasting deviations, during faults or extreme operating conditions. This results in insufficient robustness of planning and control schemes in real-world operating scenarios, potentially leading to issues such as the failure of preset voltage regulation strategies, insignificant voltage quality improvement, or low equipment regulation efficiency, thus failing to guarantee the power supply reliability of the system under high-proportion renewable energy integration. Summary of the Invention

[0013] To address the shortcomings of existing technologies, the present invention aims to provide a multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms.

[0014] The objective of this invention is achieved through the following technical solution: a multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms, comprising:

[0015] A heterogeneous regulation equipment model is constructed, which includes an on-load tap-changing transformer and a photovoltaic inverter. The on-load tap-changing transformer is used to achieve discrete regulation, and the photovoltaic inverter is used to achieve continuous regulation.

[0016] A multi-objective optimization function is constructed with voltage deviation, network loss and equipment operating cost as objectives, and the power flow balance of the distribution network and node voltage security are used as constraints to construct a multi-objective optimization model.

[0017] A fitness function is constructed, which is defined as the sum of the weighted sum of the multi-objective optimization function and the constraint violation penalty term; an adaptive crossover probability mechanism based on evolutionary algebra is introduced to balance global search and local exploitation capabilities.

[0018] Based on the heterogeneous regulation device model, N all-day voltage control schemes containing discrete and continuous sequences are generated; different all-day voltage control schemes are evaluated and scored using the multi-objective optimization model and fitness function; through iterative optimization and genetic operations, the all-day voltage control scheme with the highest score is obtained, which includes the on-load tap changer adjustment plan and the reactive power output compensation curve of the photovoltaic inverter for each time period of the day.

[0019] Furthermore, the on-load tap-changing transformer is used to achieve discrete regulation by: the on-load tap-changing transformer changing the transformer ratio by adjusting the tap position to achieve regional voltage regulation, and the tap position of the on-load tap-changing transformer is defined as a discrete integer variable that changes with time.

[0020] Furthermore, the photovoltaic inverter is used to achieve continuous regulation by: the reactive power output range of the photovoltaic inverter being jointly determined by the rated apparent power capacity and the active power output.

[0021] Furthermore, the fitness function includes: assigning different weighting coefficients to the voltage deviation, network loss, and equipment operation cost included in the multi-objective optimization function and summing them; and squaring individuals that violate voltage safety constraints to amplify the penalty.

[0022] Furthermore, the adaptive crossover probability mechanism is specifically as follows:

[0023] Preset the maximum and minimum crossover probabilities, and set the maximum number of iterations;

[0024] For the current nth iteration, determine whether the evolution condition is met. If it is met, calculate the current crossover probability. If it is not met, set the crossover probability to the preset minimum crossover probability.

[0025] Furthermore, the specific steps of generating N all-day voltage control schemes containing discrete and continuous sequences based on the heterogeneous control device model are as follows: After inputting grid parameters, photovoltaic load forecasts, and algorithm parameters, based on the heterogeneous control device model, a hybrid coding mechanism is used to initialize the population, and N chromosomes are randomly generated. Each chromosome is composed of a discrete on-load tap changer tap sequence and a continuous photovoltaic inverter reactive power output sequence. Each chromosome represents a voltage control scheme with a 15-minute interval and 96 time periods throughout the day.

[0026] Furthermore, the iterative optimization and genetic operations specifically include:

[0027] The original indicators of a certain scheme are calculated by a multi-objective optimization model, and the original indicators are quantified and scored by a fitness function. The process is then entered into a loop iteration process to determine whether the current iteration number has reached the preset maximum iteration number.

[0028] If the maximum number of iterations is reached, the search stops and the global optimal solution is output as the optimal all-day voltage control scheme. The optimal all-day voltage control scheme includes the on-load tap-changing transformer tap adjustment plan for 96 time periods throughout the day and the reactive power output compensation curve of the photovoltaic inverter at each access point.

[0029] If the maximum number of iterations is not reached, the genetic operation phase is entered, in which tournament selection, adaptive crossover based on the generation number, and mutation operation are executed in sequence. During the operation, the tap position part of the on-load tap changer maintains discrete value attributes, while the reactive power output part of the photovoltaic inverter maintains continuous value attributes. The elite retention strategy ensures that the excellent control genes are passed on to the next generation until the optimal all-day voltage control scheme is found.

[0030] The present invention also provides a multi-objective voltage optimization device for high-penetration photovoltaic distribution networks based on genetic algorithms, comprising one or more processors for implementing the method described.

[0031] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method.

[0032] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0033] The beneficial effects of this invention are as follows:

[0034] 1. This invention constructs a "discrete-continuous" equipment collaborative optimization framework, which unifies the integer range of OLTC and the real reactive power output of photovoltaic inverters under the same evolutionary framework to achieve comprehensive minimization of system voltage deviation, network loss and equipment operating costs while meeting voltage safety constraints.

[0035] 2. This invention improves the encoding mechanism and evolutionary operators of the genetic algorithm by introducing adaptive crossover and mutation probabilities and improved constraint violation penalties. It aims to enhance the robustness of the planning scheme in dealing with the randomness and volatility of photovoltaic power output, and ensure that the globally optimal strategy can be output quickly under various operating conditions.

[0036] 3. By leveraging the advantages of continuous reactive power support and coordinating discrete voltage regulation resources, this invention aims to overcome the computational bottleneck of mixed integer nonlinear optimization problems in large-scale power distribution networks, significantly shorten computation time, and improve the transparency and engineering practicality of decision-making logic. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is a flowchart of the present invention;

[0039] Figure 2 This is a flowchart illustrating the implementation of the improved genetic algorithm voltage optimization strategy proposed in this invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the technical solutions in the embodiments of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art should fall within the protection scope of the embodiments of this application.

[0041] This invention proposes a multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms. The overall goal of this method is to construct an integrated collaborative control framework that deeply integrates "discrete voltage regulation equipment" and "continuous regulation resources," thereby significantly reducing system voltage deviation, network losses, and equipment operating costs, while ensuring that the distribution network meets node voltage safety constraints, and improving its robustness to random fluctuations in photovoltaic output.

[0042] like Figure 1 As shown, this method specifically includes the following four stages:

[0043] Phase 1: Constructing a heterogeneous control device model considering discrete-continuous regulation characteristics: Constructing a heterogeneous control device model, which includes an on-load tap-changing transformer and a photovoltaic inverter, wherein the on-load tap-changing transformer is used to achieve discrete regulation, and the photovoltaic inverter is used to achieve continuous regulation.

[0044] Phase 2: Establishing a multi-objective optimization model integrating voltage quality and economic cost: Constructing a multi-objective optimization function with voltage deviation, network loss and equipment operation cost as objectives, and taking distribution network power flow balance and node voltage security as constraints, to obtain a multi-objective optimization model.

[0045] Phase 3: Designing an improved genetic algorithm based on adaptive evolutionary operators: Constructing a fitness function, defined as the sum of the weighted sum of multi-objective optimization functions and the constraint violation penalty term; wherein, the penalty factor of the dynamic constraint violation penalty term... With the number of iterations It increases non-linearly, and the formula is: This is used to expand the search space in the early stages of iteration and to force the node voltage safety constraints to be met in the later stages of iteration, in order to balance global search and local development capabilities.

[0046] Phase 4: Execution of the collaborative control process of discrete-continuous hybrid coding: Based on the heterogeneous regulation device model, N all-day voltage control schemes containing discrete and continuous sequences are generated; different all-day voltage control schemes are evaluated and scored through the multi-objective optimization model and fitness function; through iterative optimization and genetic operations, the all-day voltage control scheme with the highest score is obtained, which includes the on-load tap changer tap adjustment plan and the reactive power output compensation curve of the photovoltaic inverter for each time period of the day.

[0047] As a preferred embodiment, the first stage aims to establish an accurate and physically supported mathematical model for subsequent voltage optimization control. By finely characterizing the models of key control equipment, it provides the underlying logic for the coordinated optimization of discrete-continuous voltage regulation resources. This invention takes a typical radial distribution network as the research object, with the system containing 33 nodes and L branches. In this system, high-penetration distributed photovoltaic (PV) resources are integrated, and the core control methods consist of discretely regulated on-load tap-changing transformers (OLTCs) and continuously regulated PV inverters. The first stage specifically includes:

[0048] 1.1. Modeling of On-Load Tap Changing Transformer (OLTC)

[0049] OLTC (On-Line Transformer Control) achieves coarse voltage adjustment in a region by changing the transformer's turns ratio through tap positions. In this model, the tap position is defined as a discrete integer variable that varies with time and must satisfy the following constraints:

[0050] ;

[0051] In the formula: Indicates the first Discrete tap positions of on-load tap-changing transformers (OLTCs) during time periods; and These represent the minimum and maximum adjustable settings allowed by the OLTC device, respectively. Represents a set of integers used to constrain the discrete characteristics of gear adjustment.

[0052] Assuming the adjustment step size for each gear is (In this embodiment, it is typically taken as 0.0125 pu), then the voltage of node i is affected by OLTC as follows:

[0053] ;

[0054] In the formula: Indicates the node after adjustment by an on-load tap-changing transformer (OLTC). Voltage; This represents the reference voltage of the node; Indicates OLTC in Discrete tap positions for different time periods; This indicates the step size of the voltage change caused by each adjustment of the OLTC tap.

[0055] 1.2. Photovoltaic (PV) Inverter Modeling

[0056] As a flexible and continuously adjustable resource, photovoltaic inverters can provide continuous reactive power support by utilizing their capacity redundancy, while ensuring priority for active power output. Their reactive power adjustment range is determined by the apparent power capacity, and its expression is:

[0057] ;

[0058] ;

[0059] In the formula: Indicates that the photovoltaic inverter is in Reactive power output during a given time period; and These represent the lower and upper limits of the reactive power regulation capability of a photovoltaic inverter, respectively. This indicates the rated apparent power capacity of the photovoltaic inverter; This indicates the active power output of the photovoltaic inverter at that moment.

[0060] In a preferred embodiment, the second stage integrates multiple operational dimensions to construct quantitative criteria for evaluating the merits of control strategies.

[0061] 2.1. Design a multi-objective optimization function

[0062] The overall optimization objective function of the system includes voltage deviation, network loss, and equipment operating costs, and its expression is as follows:

[0063] ;

[0064] In the formula: For voltage deviation target, For network loss target, This represents the equipment operating cost target. Among the symbols mentioned above, Indicates the optimization time period sequence number. The total number of time periods throughout the day (taken as 96); Indicates the system node number. This represents the total number of nodes in the system. Number the branch roads This represents the total number of branches in the system. For nodes exist Per-unit value of voltage amplitude during a given time period; Reference voltage (taken as 1.0 pu); and Branch roads resistance and Current amplitude over a given period; For discrete taps of on-load tap-changing transformers (OLTC); for The reactive power output of the photovoltaic inverter during a given period; and These are the unit cost coefficients for gear shifting and reactive power adjustment, respectively.

[0065] 2.2. Determine the operational constraint boundaries

[0066] To ensure the physical operational safety of the distribution network, the multi-objective optimization model must satisfy power flow balance constraints and node voltage safety constraints, as expressed below:

[0067] ;

[0068] In the formula: the first three equations represent the power flow constraints of the distribution network, describing the physical balance between branch power and node voltage; the last inequality represents the voltage safety constraint, used to ensure that the voltage of all nodes is within the legally permissible safety range at all times. Among the symbols above: and Branch roads Flow to Node The active and reactive power; For nodes The set of downstream branches of the first node; and They are nodes Net active and reactive power (including distributed photovoltaic output). and Branch roads Resistance and inductance; branch road The current amplitude; and They are nodes and nodes The voltage amplitude; and These are the lower and upper limits of the system's allowed operating voltage, respectively.

[0069] In a preferred embodiment, the third stage improves the algorithm's performance under complex constraints by refining its core logic. Specifically, the constraints are modified to address the characteristics of voltage control problems, and the penalty for constraint violations is improved.

[0070] 3.1. Improve the design of constraint violation penalties

[0071] The fitness function of the system Defined as the sum of the weighted sum of the multi-objective functions and the dynamic constraint violation penalty term, its expression is as follows:

[0072] ;

[0073] In the formula: , , These are the targets for voltage deviation, network loss, and equipment operating cost, respectively. , , These are the corresponding weighting coefficients; This represents the current iteration number; This is the node voltage over-limit function; It is a dynamic nonlinear penalty factor.

[0074] To penalize individuals who violate voltage safety constraints, the improved constraint violation penalty term is constructed as follows:

[0075] ;

[0076] In the formula: This is the initial base penalty coefficient. This is the dynamic amplification factor; this formula ensures that the algorithm has a large search range in the solution space in the early stage of iteration, and forces the population to converge to the feasible solution region with fully qualified voltage by the quadratic increase of the penalty term in the later stage of iteration.

[0077] 3.2. Introducing an adaptive crossover mutation mechanism

[0078] To address the different evolutionary needs of discrete OLTC sequences and continuous PV sequences in chromosomes, a heterogeneous crossover probability calculation mechanism is designed: for the current... In the next iteration, determine whether the evolutionary conditions are met; if so, calculate the two sets of crossover probabilities according to the following formulas:

[0079] (1) Exponential fast decay operator for discrete variables (OLTC level):

[0080] ;

[0081] In the formula: The probability of performing a crossover operation on the discrete gene segment of an on-load tap-changing transformer (OLTC) during the nth iteration; The maximum preset crossover probability is denoted as ; n is the current iteration number of the algorithm. The maximum number of iterations is preset. It is the exponential decay control factor.

[0082] (2) Cosine-type smoothing attenuation operator for continuous variables (PV reactive power output):

[0083] ;

[0084] In the formula: Indicates the first The probability of performing a crossover operation on the reactive continuous gene segment of the photovoltaic inverter in the next iteration; This represents the preset minimum crossover probability.

[0085] (3) Exception boundary handling logic:

[0086] If the current iteration number If the evolutionary conditions are not met, then let and All are taken as the preset minimum value .

[0087] As a preferred embodiment, the fourth stage details the specific execution path of the voltage optimization strategy based on the improved adaptive genetic algorithm (IAGA). This process aims to achieve the dual goals of grid voltage stability and operational economy by coordinating the reactive power output of the on-load tap changer (OLTC) and the photovoltaic (PV) inverter. The implementation process of the proposed improved genetic algorithm voltage optimization strategy is as follows: Figure 2 As shown, the specific steps include the following:

[0088] 4.1. Initialization and Mixed Encoding

[0089] First, the grid parameters, photovoltaic load forecast, and algorithm parameters are input. Then, the population is initialized based on a heterogeneous control device model, randomly generating N chromosomes. To handle both discrete and continuous variables simultaneously, this method employs a hybrid coding mechanism to initialize the population. Each chromosome consists of a discrete OLTC range sequence and a continuous PV inverter reactive power output sequence. This coding method fully represents the comprehensive control strategy across 96 time periods (intervals of 15 minutes) throughout the day.

[0090] 4.2. Fitness Evaluation and Multi-Objective Equilibrium

[0091] For the initially generated population, a multi-objective optimization model is used to calculate various primitive indicators for each chromosome. These primitive indicators are then quantified and scored using a fitness function, and the fitness of each chromosome is calculated individually. The evaluation function comprehensively considers the voltage deviation, overall network loss, and voltage regulation equipment operating costs defined in the first and second stages. A weighting system is used to balance the multiple objectives, and an improved constraint violation penalty term described in the third stage is introduced to suppress the performance of individuals exceeding the safe voltage threshold.

[0092] 4.3. Iterative Optimization and Genetic Operations

[0093] After calculating the fitness, the algorithm enters an iterative process, determining whether the termination condition (the preset number of iterations) is met. If the termination condition is met, the algorithm stops searching and outputs the globally optimal solution. The optimal result includes the OLTC speed adjustment plan for all 96 time periods of the day and the reactive power output compensation curves of the photovoltaic inverters at each access point. If the termination condition is not met, the algorithm enters the genetic operation phase. This phase sequentially executes tournament selection, adaptive crossover based on evolutionary generations, and mutation operations. During the operation, the OLTC speed adjustment part retains discrete value attributes, while the PV reactive power output part retains continuous value attributes. An elite retention strategy ensures that excellent control genes are passed on to the next generation until the globally optimal control scheme is found.

[0094] This invention also provides a multi-objective voltage optimization device for high-penetration photovoltaic distribution networks based on genetic algorithms, including one or more processors for implementing the method described in any embodiment.

[0095] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0096] This invention also provides an electronic device, including a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method described in any embodiment.

[0097] This invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the embodiments.

[0098] The computer-readable storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be any data processing device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0099] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

Claims

1. A multi-objective voltage optimization method for high-penetration photovoltaic distribution networks based on genetic algorithms, characterized in that, Includes the following steps: A heterogeneous regulation equipment model is constructed, which includes an on-load tap-changing transformer and a photovoltaic inverter. The on-load tap-changing transformer is used to achieve discrete regulation, and the photovoltaic inverter is used to achieve continuous regulation. A multi-objective optimization function is constructed with voltage deviation, network loss and equipment operating cost as objectives, and the power flow balance of the distribution network and node voltage security are used as constraints to construct a multi-objective optimization model. A fitness function is constructed, which is defined as the sum of the weighted sum of the multi-objective optimization function and the constraint violation penalty term; an adaptive crossover probability mechanism based on evolutionary algebra is introduced to balance global search and local exploitation capabilities. Based on the heterogeneous control device model, N all-day voltage control schemes containing discrete and continuous sequences are generated. Different all-day voltage control schemes are evaluated and scored using the multi-objective optimization model and fitness function. Through iterative optimization and genetic operations, the all-day voltage control scheme with the highest score is obtained, which includes the on-load tap changer adjustment plan and the reactive power output compensation curve of the photovoltaic inverter for each time period of the day.

2. The method according to claim 1, characterized in that, The on-load tap-changing transformer is used to achieve discrete regulation, including: the on-load tap-changing transformer changes the transformer ratio by adjusting the tap position to achieve regional voltage regulation, and the tap position of the on-load tap-changing transformer is defined as a discrete integer variable that changes with time.

3. The method according to claim 1, characterized in that, The photovoltaic inverter is used to achieve continuous regulation, including: the reactive power output range of the photovoltaic inverter is determined by the rated apparent power capacity and active power output.

4. The method according to claim 1, characterized in that, The fitness function includes: assigning different weighting coefficients to the voltage deviation, network loss, and equipment operation cost included in the multi-objective optimization function and summing them; and squaring individuals that violate voltage safety constraints to amplify the penalty.

5. The method according to claim 1, characterized in that, The adaptive crossover probability mechanism is specifically as follows: Preset the maximum and minimum crossover probabilities, and set the maximum number of iterations; For the current nth iteration, for the discrete OLTC sequence in the chromosome, a genetic operation based on exponentially decaying crossover probability is used to quickly lock the device action plan. For the continuous PV sequence in the chromosome, a genetic operation based on cosine-smoothly decaying crossover probability is used to maintain the continuous fine-tuning capability of reactive voltage. If the maximum number of iterations is reached, the crossover probability is set to the preset minimum crossover probability.

6. The method according to claim 1, characterized in that, The specific steps for generating N all-day voltage control schemes containing discrete and continuous sequences based on the heterogeneous control device model are as follows: After inputting grid parameters, photovoltaic load forecasts, and algorithm parameters, based on the heterogeneous control device model, a hybrid coding mechanism is used to initialize the population, and N chromosomes are randomly generated. Each chromosome is composed of a discrete on-load tap changer tap sequence and a continuous photovoltaic inverter reactive power output sequence. Each chromosome represents a voltage control scheme with a 15-minute interval and 96 time periods throughout the day.

7. The method according to claim 1, characterized in that, The iterative optimization and genetic operations are specifically as follows: The original indicators of a certain scheme are calculated by a multi-objective optimization model, and the original indicators are quantified and scored by a fitness function. The process is then entered into a loop iteration process to determine whether the current iteration number has reached the preset maximum iteration number. If the maximum number of iterations is reached, the search stops and the global optimal solution is output as the optimal all-day voltage control scheme. The optimal all-day voltage control scheme includes the on-load tap-changing transformer tap adjustment plan for 96 time periods throughout the day and the reactive power output compensation curve of the photovoltaic inverter at each access point. If the maximum number of iterations is not reached, the genetic operation phase is entered, in which tournament selection, adaptive crossover based on the generation number, and mutation operation are executed in sequence. During the operation, the tap position part of the on-load tap changer maintains discrete value attributes, while the reactive power output part of the photovoltaic inverter maintains continuous value attributes. The elite retention strategy ensures that the excellent control genes are passed on to the next generation until the optimal all-day voltage control scheme is found.

8. A multi-objective voltage optimization device for high-penetration photovoltaic distribution networks based on genetic algorithms, characterized in that, It includes one or more processors for implementing the method as described in any one of claims 1-7.

9. An electronic device comprising a memory and a processor, characterized in that, The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images