DC power grid optimal power flow acquisition method, system and device, and storage medium

Abstract

The invention provides a DC power grid optimal power flow acquisition method, system and device, and a storage medium, belongs to the technical field of power system DC power transmission and distribution, and aims to solve the problems of source-load double-side uncertainty and non-convex feasible region fracture caused by high-proportion renewable energy access. According to the method, a main-side coevolution mechanism of a VOA (Vine Optimization Algorithm) and a demand side response (DR) regulation variable are subjected to unified modeling, a penalty function framework with the minimum network loss as the target is constructed, and dynamic balance of global exploration and local development is achieved. The method does not need gradient information, has strong robustness and rapid convergence, can obtain a high-quality solution meeting voltage, power flow and DR constraints in polynomial time, effectively improves the new energy consumption rate and the system operation economy, and provides reliable technical support for real-time scheduling of the DC side of a novel power system.

Description

Technical Field

[0001] This invention belongs to the field of DC power transmission and distribution technology, specifically relating to a method, system, equipment and storage medium for obtaining optimal power flow in a DC power grid. Background Technology

[0002] Driven by the deepening of green and low-carbon development and the accelerated construction of new power systems, high-penetration renewable energy is achieving large-scale grid connection and long-distance transmission through the efficient carrier of DC power grids. With its characteristics of rapid and controllable converters, low line losses, high stability, and natural compatibility with wind and solar DC output, the DC power grid has become a key infrastructure supporting the transformation of the energy structure. Power flow refers to the phenomenon where current or power flows from the power source through various system components to the load and is distributed throughout the power grid during operation. By analyzing and optimizing power flow, we can understand the actual flow of electricity in the grid, thereby optimizing power transmission paths, reducing energy losses, and improving power transmission efficiency.

[0003] However, the strong randomness, intermittency, and anti-peak-shaving characteristics of wind and solar power pose severe challenges to the traditional optimal power flow (OPF) framework based on deterministic models, including dramatic shifts in operating points and non-convex breaks in the feasible region. Sudden drops or surges in source-side power not only easily induce bus voltage overruns, branch overloads, and converter saturation, but also significantly increase system reserve requirements and operating costs, thereby weakening the economic advantages and low-carbon benefits of DC transmission. To mitigate source-side fluctuations, demand-side response is considered the core mechanism for activating load-side elastic resources. It guides users to dynamically adjust their electricity consumption through price or incentive signals, reshaping the net load curve in both time and space, improving renewable energy absorption rates, and delaying infrastructure investment. However, the introduction of demand-side response has transformed the DC power grid OPF problem from a continuous non-convex optimization to a hybrid "continuous-discrete" high-dimensional non-convex optimization: user dispatchable capacity is not only nonlinearly coupled with real-time electricity prices and incentive intensity, but also affected by behavioral habits, equipment physical constraints, and psychological expectations, exhibiting significant time-varying characteristics and behavioral uncertainties; the aggregation effect of flexible loads further leads to a sharp expansion of the power flow equation dimension, and traditional gradient-based algorithms have strict requirements for the continuous differentiability of the objective function and constraints, resulting in slow convergence or even complete failure in feasible region fragmentation scenarios, making it difficult to meet the real-time scheduling requirements under high-proportion uncertain scenarios.

[0004] Swarm intelligence algorithms, due to their lack of gradient dependence and strong global search capabilities, have become a crucial breakthrough in solving the non-convex OPF problem. Particle swarm optimization, differential evolution, and the Grey Wolf algorithm have been successively introduced into the field of DC grid dispatching, alleviating the problem of local extrema to some extent. However, they still reveal inherent defects in high-dimensional uncertain spaces, such as parameter sensitivity, premature convergence, and excessive computational burden. More importantly, existing research generally simplifies wind and solar power output to deterministic curves or uses finite scenario methods for approximation, failing to fully characterize its spatiotemporal uncertainties. Demand-side response modeling is also mostly limited to fixed reduction rates, static electricity price elasticity matrices, or single-period interruptible loads, neglecting multi-stage dynamic game dynamics of users, rebound loads, and the rebound effect of psychological satisfaction, making it difficult to implement optimization results in engineering practice. Furthermore, the DC power grid OPF model itself includes nonlinear power flow equations, converter control mode switching, branch transmission limits, and voltage safety constraints. Its feasible region is non-convex and non-connected. After the introduction of demand-side response, the scale of decision variables increases exponentially, and the coupling of constraints intensifies. Traditional algorithms lack a systematic processing mechanism for complex equality and inequality constraints, making it difficult to obtain high-quality solutions that meet engineering accuracy and are interpretable within a limited time. This significantly reduces the credibility and engineering applicability of the power flow OPF solution, severely restricting the safe, economical, and low-carbon operation of high-penetration renewable energy DC power grids.

[0005] Therefore, under the high penetration rate of wind and solar new energy access, the optimal power flow (OPF) of DC grid with demand-side response faces the continuous-discrete hybrid high-dimensional non-convex optimization problem. The spatiotemporal uncertainty coupling of wind and solar power output, load and demand-side response leads to slow convergence and low computational efficiency of traditional algorithms, ultimately resulting in poor engineering applicability of the optimal power flow solution. Summary of the Invention

[0006] To address the challenge of obtaining high-quality power flow solutions for DC grids that meet engineering accuracy and provide interpretability within a limited timeframe using existing technologies, this invention provides a method for acquiring optimal power flow in DC grids: the Vine Optimization Algorithm (VOA). This method aims to minimize system network losses, using voltage, power, and dispatchable load as unified decision variables. It constructs a high-dimensional non-convex optimization model with constraints on uncertain wind and solar power output, branch power flow, converter capacity, and voltage safety. Through a master-side vine co-evolution mechanism, constrained optimization is transformed into an unconstrained penalty objective. Utilizing a survival-of-the-fittest rule, it dynamically balances global exploration and local development, converging to a high-quality solution in polynomial time. This achieves a synergistic improvement in renewable energy absorption rate and operational economy under source-load interaction, as well as multi-objective collaborative optimization under source-load interaction, making it suitable for real-time scheduling and energy management of complex DC grids.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A method for obtaining optimal power flow in a DC power grid includes the following steps: The wind and solar power output reference values, base load rated power, and DC grid parameters of the target power grid system are obtained, and a joint uncertainty set of wind-solar-load-demand-side response is constructed. The normal distribution error is added to the wind and solar power output reference values ​​to obtain simulated power output data, and the fluctuation is added to the base load rated power to obtain actual load data. At the same time, the power boundary and electricity price elasticity of the demand-side response are transformed into adjustable variable constraints. With the goal of minimizing the total network loss, an initial network loss objective function is constructed based on simulated power output data and actual load data. The maximum / minimum voltage amplitude of each bus in the power grid system, the power conservation of each bus, and the maximum / minimum power output of each external power source are used as constraints. By embedding a fixed penalty factor into the initial network loss objective function, a differentiable penalty objective function is formed. With adjustable variable constraints as boundaries, continuous voltage and power data and discrete demand-side response adjustment data are uniformly encoded to obtain the dimensionality-reduced decision variables. Based on the reduced range of decision variables, the population data for the VOA algorithm is initialized, and the penalty objective function values ​​corresponding to all candidate solutions for the main vine and side vine are calculated. The candidate solution data with the best fitness is selected by pairwise comparison. If the current best candidate solution data meets any threshold, the optimal voltage, power and demand-side response adjustment data are output. If not, the initialization is returned and the iteration continues. At the same time, the input data is updated with real-time measured source load data, and the VOA optimization is repeatedly executed to obtain the VOA algorithm for optimal power flow acquisition. Obtain the maximum input power P of the power grid system in real time. imax Total power generation from new energy sources (P) W +P pv With total load demand P L and P imax and P W +P pv With total load demand P L Comparison is made; based on the comparison results, the optimal power flow (OPF) solution of the power grid is obtained by dynamically optimizing the power grid transmission power setpoint, the output scheme of new energy power plants and the load adjustment of each node through the VOA algorithm used for optimal power flow acquisition.

[0008] Preferably, the construction of the VOA algorithm for obtaining the optimal power flow specifically includes the following steps: Based on the range of decision variables after dimensionality reduction, the population data of the VOA algorithm is initialized; the main vine individual sequentially uses three biomimetic strategies—sine, cosine, and tangent—to perform a large-span scan of the solution space data, generating multiple sets of global candidate solution data; a reverse growth mechanism is introduced to perturb the aggregated candidate solution data; For each set of main vine candidate solution data, 6 side vine data are generated in its neighborhood; each vine represents a set of power grid operation schemes. The side vines combine the top three candidate solution data, population mean data and spiral random data to generate local candidate solutions by performing small step perturbation; for local candidate solution data involving demand-side response adjustment, boundary verification and correction are performed by elastic correction operator. Calculate the penalty objective function values ​​corresponding to all candidate solution data of the main vine and side vine, and select the candidate solution data with the best fitness by comparing them pairwise; if the current best candidate solution data meets any threshold, then output the optimal voltage, power and demand-side response adjustment data.

[0009] Preferably, the initial network loss objective function is: ; in, It is the sum of network losses; It is the modified nodal admittance matrix; It is the voltage of the i-th bus. It is the voltage of the j-th bus. It is the node voltage matrix.

[0010] 10. Preferably, the maximum / minimum voltage amplitude constraints for each bus are as follows: ; in, It is the voltage of the i-th bus; It is the rated voltage; The power conservation constraints for each bus are as follows: ; in, It is the algebraic sum of the input power and load power of the line; It is the modified power node admittance matrix; It is a node voltage matrix; The maximum / minimum output limits for each external power source are as follows: ; ; in, and It is the power injected by the external power source; and These are the upper and lower limits of the line's transmission power.

[0011] Preferably, the penalty objective function is: in, F Indicates fitness. It is a punishment factor. Y 1 is the modified nodal admittance matrix. Y 2 is the modified power node admittance matrix.

[0012] Preferably, obtaining the solution for the optimal power flow (OPF) of the power grid specifically includes the following steps: The system monitors the maximum input power P in real time. imax The sum of new energy power generation P W +P pv and P imax and P W +P pv With total load demand P L Compare; When P imax <P L If the total power is detected to be insufficient to meet the load demand, the demand-side response mechanism will be activated. When P imax >P L , P W +P pv >P L At times, excess electrical energy is permitted to be transmitted back to the power grid for consumption; If P imax >P L , P W +P pv <P L If there is no surplus in new energy power generation but overall power supply is still in surplus, then the power grid and new energy power plants should be coordinated to supply power on demand. The adjustment process takes minimizing network losses as its core objective. It uses the VOA algorithm for optimal power flow acquisition to dynamically optimize the power grid transmission power setpoint, the output scheme of new energy power plants, and the load adjustment of each node to obtain the solution of the optimal power flow (OPF) of the power grid.

[0013] This invention also proposes an optimal power flow acquisition system for a DC power grid, comprising: The dataset construction module is used to obtain the wind and solar power output reference values, base load rated power, and DC grid parameters of the target power grid system, construct a joint uncertainty set of wind-solar-load-demand-side response, and reduce the dimensionality of the data in the joint uncertainty set to obtain the decision variables after dimensionality reduction. The normal distribution error is added to the wind and solar power output reference values ​​to obtain simulated power output data, and the fluctuation amount is added to the base load rated power to obtain actual load data. At the same time, the power boundary and electricity price elasticity of the demand-side response are transformed into adjustable variable constraints as the population data for initializing the Vine Optimization Algorithm (VOA). The model building module is used to construct an initial network loss objective function with the goal of minimizing the total network loss of the system. The maximum / minimum voltage amplitude of each bus in the power grid system, the power conservation of each bus, and the maximum / minimum output of each external power source are used as constraints. By embedding the initial network loss objective function with a fixed penalty factor, a penalty objective function is obtained. The VOA algorithm optimization module initializes the population data of the VOA algorithm based on the range of decision variables after dimensionality reduction, calculates the penalty objective function values ​​corresponding to all candidate solutions of the main vine and side vines, and selects the candidate solution data with the best fitness through pairwise comparison. If the current best candidate solution data meets any threshold, it outputs the optimal voltage, power, and demand-side response adjustment data; if not, it returns to initialization and continues iterating. At the same time, it updates the input data with real-time measured source-load data, repeatedly executes VOA optimization, and obtains the VOA algorithm used for optimal power flow acquisition. The parameter calculation module is used to obtain the maximum input power P of the power grid system in real time. imax Total power generation of new energy sources P W +P pv and P imax and P W +P pv With total load demand P L Comparison; Based on the comparison results, the optimal power flow (OPF) solution of the power grid is obtained by dynamically optimizing the power grid transmission power setpoint, the output scheme of new energy power plants and the load adjustment of each node through the VOA algorithm used for optimal power flow acquisition.

[0014] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement any of the steps in the DC power grid optimal power flow acquisition method.

[0015] The present invention also provides a computer-readable storage medium storing a computer program that, when loaded by a processor, can execute any of the steps in the DC power grid optimal power flow acquisition method.

[0016] The optimal power flow acquisition method for DC power grids provided by this invention has the following beneficial effects: This invention constructs a joint uncertainty set of wind-solar-load-demand-response, comprehensively characterizing the multi-dimensional uncertainties of source load and demand, and simplifies the scale of decision variables by reducing the dimensionality of the data, thus solving the problem of dimensional expansion in continuous-discrete mixed high-dimensional optimization. It uses normal distribution error to simulate wind and solar power output fluctuations and transforms the power boundary and electricity price elasticity of demand-side response into adjustable variable constraints, providing accurate initialization population data for the Vine Optimization Algorithm (VOA). Simultaneously, it employs a fixed penalty factor to embed complex constraints such as voltage and power conservation into the objective function of minimizing network loss, establishing an optimization model adapted to the non-convex feasible region, thus addressing the problem of insufficient constraint handling in traditional algorithms. Through the population iterative screening and rolling optimization mechanism of the VOA algorithm, combined with real-time measurement data, it dynamically updates the input and adjusts the power setpoint, output scheme, and load adjustment, avoiding the slow convergence and easy failure of traditional gradient-based algorithms and the premature convergence of swarm intelligence algorithms, achieving efficient solution of OPF in complex scenarios. Ultimately, this effectively improves the efficiency and accuracy of solving the OPF solution for DC power grids in scenarios with high penetration of renewable energy, ensures the engineering applicability and reliability of the solution, helps the power grid operate safely and economically under source-load fluctuations, increases the renewable energy absorption rate, and reduces operating costs. Attached Figure Description

[0017] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The 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.

[0018] Figure 1 This is a flowchart of the DC power flow acquisition method according to Embodiment 1 of the present invention; Figure 2 This is a topology diagram of a CIGRE B4 DC power grid; Figure 3 A flowchart for power flow calculation after introducing demand-side response is provided. Figure 4 This includes the system's daily line loss and algorithm loss. Figure 5 Calculate the box configuration diagram independently for CS1; Figure 6 Calculate the box configuration independently for CS2; Figure 7 The convergence curve of VOA under CS1; Figure 8 The convergence curve of VOA under CS2; Figure 9 The convergence curve of PSO under CS2; Figure 10 This is the convergence curve of GWO under CS2. Detailed Implementation

[0019] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0020] Example 1 The optimal power flow problem in power grids is non-convex and nonlinear. Traditional optimization algorithms are prone to getting trapped in local optima and exhibiting poor convergence when dealing with uncertainties on both the source and load sides and high-dimensional non-convex constraints. Therefore, it is necessary to construct a novel optimization framework that can simultaneously address high-dimensional uncertainties on the source side, elastic behavior on the load side, and non-convex security constraints of the system, while possessing provable convergence and parameter robustness, to achieve real-time, economical, and safe dispatching of DC power grids under complex operating scenarios. The Vine Optimization Algorithm (VOA), a recently proposed swarm intelligence method, simulates the collaborative growth of main and side branches of a vine, resource competition, and survival of the fittest mechanism. It maintains population diversity while also considering local refined search, providing a novel approach to optimization with high-dimensional non-convex constraints. This invention deeply couples it with the demand-side response mechanism, breaking through the theoretical bottleneck of traditional algorithms in dealing with uncertainties on both the source and load sides and complex equation constraints, providing solid theoretical and technical support for the efficient absorption and low-carbon operation of new power systems.

[0021] This invention integrates the master-side co-evolutionary mechanism of the vine optimization algorithm with demand-side response resources into the optimal power flow model of a DC power grid. It constructs a penalty function with voltage, power flow, and adjustable load constraints with the goal of minimizing network loss, achieving efficient global optimization in a high-dimensional non-convex space. The master vine explores uncertain scenarios through multi-strategy hopping, while the side vine refines the feasible region of demand-side response locally through six modes. The survival-of-the-fittest mechanism dynamically balances source-load interaction, ensuring convergence to an engineering-feasible and economically optimal scheduling scheme within polynomial time.

[0022] Based on this, the present invention provides a method for obtaining the optimal power flow of a DC power grid, specifically involving a method for obtaining the optimal power flow of a DC power grid that integrates the Vine Optimization Algorithm (VOA) and the Demand Response Algorithm (DR).

[0023] First, the mathematical model of OPF will be explained.

[0024] Power flow distribution (OPF) calculation, given the power system network topology, electrical component parameters, and system load conditions, aims to find the optimal control variables that satisfy all constraints, resulting in the optimal power flow distribution for a specific performance index (objective function) of the power system. It is a set of high-order nonlinear algebraic equations. Its main components include a set of variables, constraints, and an objective function. Control variables are controllable independent variables, which in this invention include bus voltages under constant voltage control, active power output from generator units under constant power control, and controllable load power under conditions of insufficient generator power. State variables are the dependent variables of the control variables and must be obtained through power flow calculations; they include the voltages of other buses and the power of each branch, as shown in the following equation.

[0025] ; ; ; ; In the formula: It is the objective function to be optimized in this invention. F The value is fitness; It is an inequality constraint; It is an equality constraint; and These are control variables and state variables, respectively. It is a set of state variables.

[0026] Secondly, the DC power system model and component parameters are explained.

[0027] The DC power system applied in this invention is a novel power system that includes wind turbine generators and photovoltaic power plants. The structural topology of this system is as follows: Figure 2 As shown in the diagram, the system consists of 11 DC buses (#1-#11), two grid power sources (S1 and S2), three photovoltaic power stations (PV1-PV3), three wind turbine generators (W1-W3), and 14 DC lines (1-14). Since wind and solar power generation typically incorporates energy storage to ensure power stability, but current energy storage configurations are insufficient, this invention temporarily includes energy storage in the error term when considering its role. The converter connected to bus #1 uses constant voltage control, while the remaining converters use constant power control. The system's rated voltage is 400kV, and the line resistance is 0.016 ohms.

[0028] Specifically, such as Figure 1 As shown, the implementation of the DC power flow acquisition method proposed in this invention includes the following steps: Step 1: Data Input and Uncertainty Modeling: Obtain the reference values ​​of wind and solar power output, rated power of base load, and DC grid parameters (such as the number of buses, line resistance, and converter control mode) of the target power grid system, and construct a joint uncertainty set of wind-solar-load-demand-side response; add normal distribution error to the reference values ​​of wind and solar power output to obtain simulated power output data, and add fluctuation to the rated power of base load to obtain actual load data. At the same time, transform the power boundary and electricity price elasticity of demand-side response into adjustable variable constraints to achieve unified data dimensions of source-load-side.

[0029] Step 2: Objective Function and Constraint Transformation and Dimensionality Reduction: With the goal of minimizing the total system network loss, an initial network loss objective function is constructed based on simulated power output data, actual load data, and the upgraded node admittance matrix. Constraints such as the maximum and minimum voltage amplitude limits of each bus in the power grid system, bus power conservation, and the upper and lower limits of external power output are embedded into the initial network loss objective function using a fixed penalty factor (value 10000), forming a differentiable penalty objective function. Using adjustable variable constraints as boundaries, continuous voltage and power data and discrete demand-side response adjustment data are uniformly encoded to obtain dimensionality-reduced decision variables, providing standardized input for the VOA algorithm population initialization.

[0030] Step 3: VOA Algorithm Optimization Calculation: Based on the range of decision variables after dimensionality reduction, initialize the population data for the VOA algorithm, calculate the penalty objective function values ​​corresponding to all main and side vine candidate solution data, and select the candidate solution data with optimal fitness through pairwise comparison; if the current optimal candidate solution data meets any threshold, output the optimal voltage, power, and demand-side response adjustment data; if not, return to initialization and continue iteration; simultaneously, update the input data with real-time measured source-load data, repeatedly execute VOA optimization, and obtain the VOA algorithm used for optimal power flow acquisition. Specifically, this includes the following steps:

[0031] Population initialization: Initialize the population data for the VOA algorithm based on the range of decision variables after dimensionality reduction; Global exploration: The main vine individual uses three biomimetic strategies—sine, cosine, and tangent—to scan the solution space and generate global candidate solutions, maintaining population diversity through a reverse growth mechanism.

[0032] Local mining: Each main vine generates 6 side vines. Combining the top three candidate solutions, the population mean, and spiral random data, small-step perturbation is performed. The elastic correction operator ensures that the demand-side response adjustment is within the acceptable range for users.

[0033] Competitive screening: Calculate the fitness (penalty objective function value) of all main-side vine candidate solutions, screen the optimal candidate solution, and eliminate redundant data.

[0034] Convergence determination and rolling correction: Set two thresholds for the maximum number of iterations (e.g., 5000 generations) and convergence accuracy (e.g., algorithm loss less than 10²). If the optimal candidate solution meets either threshold, output the optimal voltage, power and demand-side response adjustment data; otherwise, return to the global exploration step to continue iterating. At the same time, adopt the model predictive control framework to update the input data with real-time measured source load data and perform VOA optimization in a rolling manner.

[0035] Step 4, Dynamic Power Balance Adjustment: Obtain the maximum input power P of the power grid system in real time. imax Total power generation from new energy sources (P) W +P pv With total load demand P L and P imax and P W +P pv With total load demand P L Make a comparison.

[0036] If P imax <P L (Power supply gap), initiate demand-side response to adjust controllable load; if P imax >P L And P W +P pv >P L (Excess renewable energy) is fed back to the grid in reverse; if P imax >P L But P W +P pv <P L (Overall power supply is excessive but renewable energy is insufficient), coordinate the power supply of the power grid and renewable energy power plants; dynamically optimize the power grid transmission power setpoint, renewable energy output scheme and node load adjustment through the VOA algorithm, and finally obtain the optimal power flow OPF solution.

[0037] Specifically, the power balance dynamic adjustment phase uses real-time measured source-load data (including maximum input power P) from the power grid system. imax Total power generation from new energy sources (P) W +P pv Total load demand P LThe algorithm takes the modified node admittance matrix, fixed penalty factor (value 10000), algorithm termination threshold (maximum number of iterations 5000, convergence accuracy 10²), and adjustable variable constraints transformed in step 1 (demand-side response power boundary, range of values ​​corresponding to electricity price elasticity) as inputs. Through the main-side co-evolution mechanism of the VOA algorithm (main vine multi-strategy global exploration + side vine six-mode local refinement + demand-side response elasticity correction operator + survival of the fittest ternary competition), it dynamically optimizes the grid transmission power setpoint, the output scheme of new energy power plants, and the load adjustment of each node, while satisfying the constraints such as upper and lower limits of bus voltage amplitude, bus power conservation, and upper and lower limits of external power output. Finally, it outputs the optimal node voltage stability value, branch power flow distribution, demand-side response adjustment amount, and system network loss value, which is the complete solution of the optimal power flow OPF of the grid.

[0038] During the dynamic adjustment of power balance, the input of the VOA algorithm is real-time and scenario-based source-load data and constraint boundaries, and the output is the optimal scheduling command adapted to the current operating conditions. Both are closely aligned with the real-time operating status of the DC power grid.

[0039] The input data focuses on real-time measurement and scenario adaptation, covering three key types of information: source, load, and network, to ensure that the algorithm optimization closely matches actual working conditions. Real-time source-side data: Total renewable energy power generation P monitored by the system in real time W +P pv (Including real-time output of wind and solar power), replacing the simulated output data in the initial modeling stage, accurately reflecting the real-time fluctuation status of renewable energy.

[0040] Real-time load-side data: Real-time value of total load demand P L It replaces the initial actual load data and reflects the dynamic changes in users' electricity consumption behavior; at the same time, it includes the real-time constraint boundary of demand-side response adjustable load (derived from the power boundary and electricity price elasticity transformed in step 1, and dynamically adjusted according to real-time electricity price or user feedback).

[0041] Real-time grid-side data: Real-time value of maximum grid input power P imax Real-time monitoring values ​​of voltage at each busbar, real-time power flow data of branches, and real-time upper / lower limits of output from external power sources (grid power and new energy power plants) ensure that constraints are synchronized with the grid operation status.

[0042] Fixed foundational data: modified node admittance matrix (constructed based on power grid topology and component parameters, remaining unchanged throughout), fixed penalty factor (value 10000), and algorithm termination threshold (maximum number of iterations 5000, convergence accuracy 10²), providing a stable mathematical foundation and judgment criteria for optimization calculations.

[0043] The output data focuses on power balance, minimum network loss, and constraint satisfaction, and is directly used for real-time grid dispatching. Specifically, it includes three types of core regulation parameters: Grid transmission power setpoint: Optimized grid power injection (as mentioned later) and It can provide positive power supply or reverse absorption according to the operating conditions (such as outputting negative power when there is a surplus of new energy, which represents reverse power supply), to ensure power matching between the grid and the source load.

[0044] New energy power plant output scheme: Optimized real-time output allocation values ​​for wind power and photovoltaic power plants to avoid wind and solar curtailment or insufficient output, and maximize the new energy consumption rate.

[0045] Load adjustment at each node: Demand-side response adjustment (such as P mentioned later) adjust1 - P adjust11 The system dynamically adjusts according to operating conditions, reducing unnecessary loads when there is a power shortage and increasing adjustable loads when there is a surplus of new energy sources, and the adjustment amount is always within the acceptable range for users (verified by the elastic correction operator).

[0046] Auxiliary output data: Optimized stable values ​​of each bus voltage (such as U2-U mentioned later). 11 Ensure that the voltage is within the rated range (around 400kV); the real-time network loss value (Ploss) of the system serves as a quantitative indicator of the optimization effect, ensuring that the core objective of minimizing network loss is met.

[0047] When P imax <P L (Power supply gap): Input data triggers demand-side response start constraints, and the VOA algorithm outputs a solution to reduce the load adjustment of some nodes and optimize the output of external power supply to maintain power balance.

[0048] When P imax >P L And P W +P pv >P L (Excess renewable energy): Input data triggers reverse power transmission constraints, and the VOA algorithm outputs a scheme that combines the grid reverse transmission power setpoint with the increase of adjustable load regulation to absorb excess renewable energy.

[0049] When P imax >P L But P W +P pv <P L (Overall power supply surplus, insufficient new energy): Input data triggers collaborative power supply constraints, and the VOA algorithm outputs a power distribution scheme between the power grid and new energy power plants to minimize network losses while meeting load demand.

[0050] This invention effectively overcomes the bottleneck of finding optimal power flow in high-dimensional non-convex feasible regions of DC grids by deeply coupling the vine optimization algorithm with the demand-side response mechanism. The master-side vine co-evolution strategy achieves a dynamic balance between global exploration and local fine-grained search while maintaining population diversity, significantly improving the convergence speed and optimization accuracy of the algorithm under complex constraints. The introduction of demand-side response incorporates load-side flexible resources into a unified scheduling framework, reshaping the spatiotemporal distribution of net load, effectively mitigating renewable energy fluctuations, reducing system operating losses and reserve requirements, and enhancing the grid's ability to absorb new energy. The combination of adaptive penalty mechanism and rolling correction strategy ensures that key operational constraints of voltage and power flow remain within a strictly feasible range under continuous uncertainties on both the source and load sides, significantly improving the robustness and engineering applicability of the scheduling scheme, and providing reliable technical support for the safe, economical, and low-carbon operation of DC grids under the new power system context.

[0051] The optimal power flow acquisition method for DC power grids proposed in this invention will be further illustrated below through examples. The implementation of this embodiment specifically includes the following steps:

[0052] Step 1: Unified Modeling of Uncertain Sources, Loads, and Demand-Side Responses A joint uncertain set of wind and solar power output, base load, and adjustable capacity of demand-side response is constructed, and the spatiotemporal correlation is characterized by normal perturbation. The power boundary of demand-side response, electricity price elasticity, and user behavior characteristics are transformed into adjustable variable constraints to achieve a three-dimensional expression of source-load-demand-side response.

[0053] Step 2: Reconstruction of Non-convex OPF Penalty Target With the goal of minimizing system network losses, constraints such as voltage safety, converter capacity, branch power flow, and demand-side response adjustment are embedded into the objective function through fixed penalty factors to form a differentiable penalty objective. The decision variables are uniformly encoded as continuous voltage, power, and discrete demand-side response adjustment, thereby achieving dimensionality reduction of the high-dimensional solution space.

[0054] Step 3: Global Exploration of the VOA Main Vine The population is initialized, and the main vine sequentially executes three biomimetic jumping strategies—sine, cosine, and tangent—to perform a large-scale scan of the solution space. A reverse growth mechanism is introduced to prevent population aggregation, maintain diversity, and effectively cope with the search space mutations caused by uncertainty.

[0055] Step 4: Local development of VOA side vines Each main vine generates 6 side vines in its neighborhood. Small-step perturbations are performed using multiple guidance modes, including the top three individuals, population mean, and spiral random. An elastic correction operator is introduced for the demand-side response variables to ensure that the adjustment amount is always within the user's acceptable range, thereby achieving fine mining of the local feasible region.

[0056] Step 5: Main-Side Competition and Survival of the Fittest By comparing the fitness of the main vine with all the side vines, only the best one is retained for the next generation, realizing a three-element competition mechanism of "main-side-demand-side response", reducing redundant calculations and improving convergence efficiency.

[0057] Step 6: Termination Judgment and Rolling Correction The system sets a dual-threshold termination criterion of maximum iteration count and convergence accuracy; it adopts a model predictive control framework to perform VOA optimization on a rolling basis with the latest measurement data, and corrects the demand-side response command and converter setpoint in real time to resist real-time source-load fluctuations.

[0058] Because the power provided by wind and solar power is uncertain, an uncertainty model is constructed. To account for model errors, the impact of energy storage is included in the error term, and a normal perturbation is added to the sampled power points, as shown in the following equation.

[0059] ; ; In the formula: and This is the reference output power for wind and light; and This is a simulated wind and solar power output. and This is the additional wind and solar power that takes into account the error of the normal distribution.

[0060] Furthermore, the loads connected to each busbar also have uncertainties, which are mainly determined by the users' electricity consumption habits and patterns. Therefore, we consider constructing an uncertainty model for the load, as shown in the following formula.

[0061] ; in, It is the actual load power of the i-th bus; It is the rated load power of the i-th busbar; It is the additional load power of the i-th bus considering fluctuations.

[0062] To adapt to the uncertainties inherent in new power systems, this invention considers load participation in power system dispatching. Therefore, load is both a resource for power system dispatching and a decision variable. For each line, to maintain power conservation while ensuring base power, different demand-side response power values ​​are provided. Thus, the load power is modified as described by the following equation.

[0063] ; in, It is the demand-side response power of the i-th bus participating in system scheduling; It takes into account all load power in the demand-side response.

[0064] Description of the objective function and constraints of OPF: This invention primarily addresses the network loss situation in novel DC power systems containing distributed generation sources. Therefore, an initial network loss objective function is constructed to minimize the total network loss. The expression of the objective function is as follows: ; in, It is the sum of network losses; It is the modified nodal admittance matrix; It is the voltage of the i-th bus. It is the voltage of the j-th bus. It is the square of the node voltage matrix.

[0065] The system constraints considered by OPF are divided into two categories: inequality constraints and equality constraints, as detailed below.

[0066] Maximum / minimum voltage amplitude constraints for each bus: ; in, It is the voltage of the i-th bus; That is the rated voltage.

[0067] Power conservation constraints for each bus: ; in, It is the algebraic sum of the input power and load power of the line; It is the modified power node admittance matrix; It is the node voltage matrix.

[0068] Maximum / minimum output constraints for each external power source: ; ; in, and It is the power injected by the external power source; and These are the upper and lower limits of the line's transmission power.

[0069] Power flow calculation formula adapted to VOA algorithm: Like other intelligent optimization algorithms, the VOA algorithm cannot be solved directly when dealing with optimization problems with constraints. It needs to use a penalty term to transform the constraints into part of the objective function. The transformed objective function is shown below, which is the penalty objective function.

[0070] ; in, F For fitness, It is the penalty factor. Since all constraints are treated equally, the same fixed penalty factor with a value of 10000 is used.

[0071] Due to the uncertainty of wind and solar power output, the sum of their output power and thermal power power, as well as the load power, fluctuates significantly within 24 hours. Instantaneous power balance requires adjustments through control strategies; otherwise, wind and solar curtailment or insufficient load capacity may occur. Specific control strategies include: when the sum of grid transmission power and generation power is less than the load power, a demand-side response mechanism is needed to dynamically adjust the controllable load power of each node to meet constraints such as power conservation, while ensuring base power. When the sum of grid transmission power and generation power exceeds the load power, and wind and solar power generation also exceeds the load power, reverse power transmission to the grid is required. Conversely, when wind and solar power generation also exceeds the load power, power is jointly transmitted from the grid and wind / solar power plants to achieve power balance. This mechanism ensures the maximum renewable energy absorption rate. A detailed flowchart is shown below. Figure 3 The diagram shown is the overall flowchart of the DC power flow acquisition method based on the VOA algorithm. The VOA algorithm is the most critical step in finding the minimum network loss. The specific process is as follows.

[0072] First, the system monitors the maximum input power P in real time. imax The sum of new energy power generation P W +P pv and compare it with the total load demand P L Compare them.

[0073] When the total power is detected to be insufficient to meet the load demand (i.e., there is a power supply gap), P imax <P L If the total power supply exceeds the load demand (i.e., DSR=1), then the demand-side response mechanism is activated: While ensuring basic electricity consumption for users, the controllable load power of each node is adjusted (e.g., temporarily reducing the power of non-essential electrical equipment) to rematch the power supply capacity with the load demand. When it is detected that the total power exceeds the load demand and there is a surplus of renewable energy generation (i.e., P...), the demand-side response mechanism is activated. imax >P L , P W +P pv >PL If the amount of new energy generation is not excessive but the overall power supply is still excessive (i.e., P), then it is permissible to reverse the flow of excess electricity back to the grid for consumption; imax >P L , P W +P pv <P L This coordinates the power grid and renewable energy power plants to supply power on demand. The entire adjustment process prioritizes minimizing grid losses, dynamically optimizing three key variables using the Vine Optimization Algorithm (VOA): the setpoint for grid transmission power, the output scheme of renewable energy power plants, and the load adjustment at each node. The VOA algorithm employs a strategy combining global exploration and local search to efficiently solve for the optimal adjustment scheme (such as the voltage U2-U at unbalanced nodes) while satisfying voltage safety constraints, line transmission limits, and power balance. 11 Demand-side response power P adjust1 - P adjust11 Network loss P loss This ensures the stable operation of the system while maximizing the absorption rate and economic efficiency of new energy sources.

[0074] The following specific examples illustrate the proposed method for obtaining the optimal power flow of a DC power grid.

[0075] The OPF problem within a day is solved using the VOA algorithm framework proposed in this invention. To avoid data redundancy and confusing graphical display, this invention selects four time points for display and presents the results of obtaining the optimal target value, decision variables, and control variables in Tables 1 to 3.

[0076] Table 1 shows the initial values ​​of wind power and photovoltaic power generation after power conversion, which is also a predicted value. After considering the prediction error and model error, the data results can basically meet the expected results of a typical day.

[0077] Table 1 Initial wind and solar input power Table 2 Comparative Analysis of Initial and Optimized Values ​​for Different Variables Table 2 shows the initial values ​​corresponding to the values ​​set by the algorithm, representing the upper limit of grid power transmission and the voltages of all nodes except the constant voltage control node. The optimized values ​​represent the optimized values ​​after VOA calculation. First, comparing the changes in grid transmission power pairwise reveals variations in transmission power, including reductions and reverse transmission. Second, comparing the voltage distribution of each node shows that, because the VOA algorithm does not rely on the initial values ​​and randomly allocates node voltages within the voltage limit, the initial voltage distribution is noticeably dispersed, even showing extreme cases where voltages approach the lower limit—unacceptable in a normally operating grid. However, after VOA calculation, the voltages of each node stabilize near their rated voltages and are more evenly distributed. It is also evident that at the 10-hour mark, the grid receives fed-in power because the photovoltaic power gradually increases, exceeding load requirements and transferring excess energy to the grid. Furthermore, the voltages of each node are generally higher than at other times, especially at nodes with distributed generation, consistent with the relationship between power and voltage. At other times, when the power grid is outputting electricity to compensate for wind and solar power, power flows from the grid to the load, and the voltage at each node is relatively low and more evenly distributed.

[0078] Cases that introduce a demand-side response mechanism are referred to as CS1, while cases that do not introduce a demand-side response mechanism are referred to as CS2.

[0079] Table 3 shows the power loss and algorithm loss of the intelligent optimization algorithm in handling the two types of problems, as well as the adjustment amount of the demand-side response. For CS1, it can be seen that at 0, due to the large power of wind power 1, the algorithm adjusts the positive adjustable load to absorb clean energy; at 6, due to the gradual appearance of the sun, the power of photovoltaic power generation is large, and the power of wind power 3 is also large, so the algorithm adjusts the positive adjustable load to absorb clean energy; at 10, since the power of both wind power and photovoltaic power generation is large, it actually exceeds the load, and the maximum demand power is easily obtained at this time. After uncertainty adjustment, the remaining power is fed back to the grid. At 22, wind power and photovoltaic power generation are not significant enough. At this time, the external grid needs to support all the loads in the area. Therefore, the grid is at the transmission limit and still cannot meet the power conservation requirements. So, by adjusting the load, some power is cut off to maintain the stability of the grid. loss The line shows the system's network loss. At 0:00 and 6:00, the grid load is reduced due to the support of distributed power sources, so the power loss flowing through the lines is also reduced. At 10:00 and 22:00, because the distributed power sources are large or small, they need to feed power back to the grid or rely on grid support, resulting in a surge in line losses, and the losses are even greater when feeding back power. C errorThe rows show the degree of violation by the algorithm when dealing with constrained problems, from Figure 4 As can be seen from the results in Table 2, the algorithm loss is less than 10. -2 Therefore, from a computational perspective, the calculation results are accurate. The trend is similar to that of the adjustable load case, with the non-adjustable load case showing a consistent pattern. It can be observed that the overall line loss is lower and the algorithm loss is smaller, both less than 10. -4 .

[0080] Table 3 Comparative Analysis of Optimized Values ​​for Load Side and Line Losses Algorithm stability analysis: To further evaluate the stability of the VOA algorithm in solving the OPF problem, the algorithm was set to search for 1000 individuals and iterate 5000 times, with each iteration performed independently 50 times. Successful computation depends on not violating constraints or having constraints violated to an acceptable degree; therefore, this invention primarily examines the degree of constraint violation when handling constraints to evaluate the algorithm's stability. The computation results are plotted on... Figure 5 and Figure 6 In a box plot, the upper and lower edges represent the maximum and minimum values ​​of the sample, excluding outliers. The rectangular portion of the box plot contains 50% of the sample. The upper and lower boundaries of the matrix represent the upper and lower quartiles of the sample, respectively, and the median represents the median of the sample.

[0081] Through observation Figure 5 It can be seen that the degree of constraint violation in the VOA algorithm's solution of CS1 is basically around 10. -4 The accuracy is sufficient, being close to the order of magnitude. Specifically, the statistical calculation algorithm loss is less than 10. -1 The total number of calculations was 1126, with an accuracy of 93.83%; the algorithm loss was less than 10. -2 The total number of calculations was 1121, with an accuracy of 93.42%; the algorithm loss was less than 10. -3 The total number of calculations was 844, with an accuracy of 70.33%; the algorithm loss was less than 10. -4 The total number of calculations was 571, with an accuracy of 47.58%. This was achieved through observation. Figure 6 It can be seen that the degree of constraint violation in the VOA algorithm's solution of CS2 is basically around 10. -5 Near the order of magnitude. Specifically, the statistical calculation algorithm loss is less than 10. -1 The total number of calculations was 1088, with an accuracy of 90.67%; the algorithm loss was less than 10. -2 The total number of calculations was 1088, with an accuracy of 90.67%; the algorithm loss was less than 10. -3The total number of calculations was 1073, with an accuracy of 89.42%; the algorithm loss was less than 10. -4 The total number of calculations was 1011, and the accuracy was 84.25%. This demonstrates that the algorithm has sufficient precision, high consistency, and good robustness.

[0082] Figure 7 and Figure 8 The convergence curve of the VOA algorithm is shown. It can be seen that although the initial value of the VOA algorithm is large, it has the fastest descent performance, reaching the optimal value around the 500th generation. Then it begins to search finely around the optimal value. It can be seen that the VOA algorithm has a strong optimization ability and convergence.

[0083] Performance comparison of three algorithms: This invention performs runtime convergence tests on PSO, GWO, and VOA respectively.

[0084] Because the optimization search results of the optimization algorithms are random, the outer loop iterations are set to 10, and convergence curves for the three algorithms at four different time points are plotted. To avoid confusion, only the convergence curve for the simpler CS2 case is plotted here, as shown below. Figure 8 and Figure 9 As shown. Observe the algorithm test results graph, from... Figure 8 It can be observed that VOA successfully converges at all four time points and stabilizes after approximately 500 iterations; from Figure 9 It can be observed that PSO converges successfully at only one time point; from Figure 10 It can be observed that GWO fails to converge at any time point. This is partly due to two reasons: firstly, with the continuous updates to intelligent optimization algorithms, traditional intelligent optimization algorithms perform poorly when handling more complex optimization problems; secondly, DC systems contain a large number of equality constraints, and even slight violations of these constraints can cause a sharp increase in the objective function, leading to the optimization problem failing to converge.

[0085] To more thoroughly evaluate the performance of the three algorithms, the program runtime and convergence rate are shown in Table 4, based on the descriptions above. The algorithms are analyzed on a daily basis, with 24 time points per day. The calculations at each time point are independent, resulting in a total of 240 runs for each algorithm. The data recorded in Table 4 represents the total program runtime across the 24 time points within a day. The convergence rate is the ratio of the number of convergent calculations out of the 240 independent calculations to the total number of runs. Observing the data in Table 4, we can see that in CS1 and CS2 cases, PSO has the highest convergence rate of only 29.17% and the shortest runtime of 17.05 seconds; GWO failed to converge. However, VOA has the lowest convergence rate of 79.17% and the longest runtime of 14.84 seconds. Comparing the three algorithms, VOA performs better.

[0086] Table 4 Performance Comparison Analysis of Three Algorithms Based on the same inventive concept, this invention also proposes a DC power flow acquisition system, including a dataset construction module, a model construction module, a VOA algorithm optimization module, and a parameter calculation module.

[0087] Specifically, the dataset construction module is used to obtain the wind and solar power output reference values, base load rated power, and DC grid parameters of the target power grid system, construct a joint uncertainty set of wind-solar-load-demand-side response, reduce the dimensionality of the data in the joint uncertainty set to obtain the decision variables after dimensionality reduction, add normal distribution error to the wind and solar power output reference values ​​to obtain simulated power output data, and add fluctuation amount to the base load rated power to obtain actual load data. At the same time, the power boundary and electricity price elasticity of demand-side response are transformed into adjustable variable constraints as population data for initializing the Vine Optimization Algorithm (VOA).

[0088] The model building module is used to construct an initial network loss objective function with the goal of minimizing the total network loss of the system. The maximum / minimum voltage amplitude of each bus in the power grid system, the power conservation of each bus, and the maximum / minimum output of each external power source are used as constraints. By embedding the initial network loss objective function with a fixed penalty factor, a penalty objective function is obtained.

[0089] The VOA algorithm optimization module initializes the population data for the VOA algorithm based on the range of decision variables after dimensionality reduction, calculates the penalty objective function values ​​corresponding to all candidate solutions for the main and lateral vines, and selects the candidate solution data with the best fitness through pairwise comparison. If the current best candidate solution data meets any threshold, it outputs the optimal voltage, power, and demand-side response adjustment data; if not, it returns to initialization and continues iterating. At the same time, it updates the input data with real-time measured source-load data, repeatedly executes VOA optimization, and obtains the VOA algorithm used for optimal power flow acquisition.

[0090] The parameter calculation module is used to obtain the maximum input power P of the power grid system in real time. imax Total power generation of new energy sources P W +P pv and P imax and P W +P pv With total load demand P L Comparison; Based on the comparison results, the optimal power flow (OPF) solution of the power grid is obtained by dynamically optimizing the power grid transmission power setpoint, the output scheme of new energy power plants and the load adjustment of each node through the VOA algorithm used for optimal power flow acquisition.

[0091] Each module in the aforementioned DC power flow acquisition system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0092] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps in the embodiment of the DC power flow acquisition method. Specific implementation methods can be found in the method embodiments, and will not be repeated here.

[0093] Furthermore, the present invention also provides a non-transitory computer-readable storage medium containing instructions, on which a computer program is stored. For example, a memory containing instructions, which can be executed by a processor of a computer device to complete the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc. When the computer program is executed by the processor, it can implement the steps in the embodiment of the DC power flow acquisition method. Specific implementation methods can be found in the method embodiments, which will not be repeated here.

[0094] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0096] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0098] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the present invention patent. No reference numerals in the claims should be construed as limiting the scope of the claims. Any simple variations or equivalent substitutions of technical solutions that can be readily obtained by those skilled in the art within the scope of the technology disclosed in the present invention are within the protection scope of the present invention.

Claims

1. A method for obtaining optimal power flow of a direct current power grid, characterized in that, The method comprises the following steps: obtaining wind and light output reference values of a target power grid system, basic load rated power and DC power grid parameters, constructing a wind and light-load-demand side response combined uncertainty set; adding normal distribution errors to the wind and light output reference values to obtain simulated output data, superimposing fluctuation on the basic load rated power to obtain actual load data, and converting power boundaries and price elasticity of the demand side response into adjustable variable constraints; taking the minimum of the total system loss as an objective, constructing an initial loss target function based on the simulated output data and the actual load data; taking the maximum / minimum of the voltage amplitude of each bus, the power conservation of each bus, and the maximum / minimum of the output of each external power source as constraint conditions, embedding the initial loss target function through a fixed penalty factor to form a differentiable penalty target function; taking the adjustable variable constraints as boundaries, uniformly encoding continuous voltage and power data and discrete demand side response adjustment data to obtain reduced decision variables; initializing population data of the VOA algorithm according to the range of the reduced decision variables, calculating the penalty target function values corresponding to all main and side vine candidate solution data, and screening the candidate solution data with the optimal fitness through pairwise comparison; if the current optimal candidate solution data meet any threshold, outputting the optimal voltage, power and demand side response adjustment data; if not, returning to initialization for iteration; meanwhile, updating the input data with real-time measured source and load data, repeatedly executing the VOA optimization to obtain the VOA algorithm for optimal power flow acquisition; Obtain the maximum input power P of the power grid system in real time. imax Total power generation from new energy sources P W +P pv With total load demand P L and P imax and P W +P pv With total load demand P L Comparison is made; based on the comparison results, the optimal power flow (OPF) solution of the power grid is obtained by dynamically optimizing the power grid transmission power setpoint, the output scheme of new energy power plants and the load adjustment of each node through the VOA algorithm used for optimal power flow acquisition.

2. The DC grid optimal power flow acquisition method according to claim 1, characterized in that, The construction of the VOA algorithm for optimal power flow acquisition specifically comprises the following steps: initializing population data of the VOA algorithm according to the range of the reduced decision variables; main vine individuals successively adopt three kinds of bionic strategies of sine, cosine and tangent to perform large-span scanning on solution space data, generating multiple sets of global candidate solution data; introducing a reverse growth mechanism to disturb the gathered candidate solution data; for each set of main vine candidate solution data, generating 6 side vine data in its neighborhood; each vine represents a set of power grid operation scheme, and the side vine combines the top three candidate solution data, population mean data and spiral random data to perform small-step disturbance to generate local candidate solutions; the local candidate solution data involving demand side response adjustment are checked and corrected through an elastic correction operator; calculating the penalty target function values corresponding to all main and side vine candidate solution data, and screening the candidate solution data with the optimal fitness through pairwise comparison; if the current optimal candidate solution data meet any threshold, outputting the optimal voltage, power and demand side response adjustment data.

3. The DC grid optimal power flow acquisition method according to claim 1, characterized in that, The initial loss target function is: ； wherein, is the sum of the network losses; is the modified node admittance matrix; is the voltage of the i-th bus, is the voltage of the j-th bus, is the node voltage matrix.

4. The DC grid optimal power flow acquisition method according to claim 1, characterized in that, The maximum / minimum of the voltage amplitude of each bus is: ； wherein, is the voltage of the i-th bus; is the rated voltage; The power conservation of each bus is: ； wherein, is the algebraic sum of the input power and the load power of the line section; is the modified power node admittance matrix; is the node voltage matrix; The maximum / minimum of the output of each external power source is: ； ； wherein, and is the power injected by the external power supply; and are the upper and lower limits of the line transmitted power.

5. The DC grid optimal power flow acquisition method according to claim 3, characterized in that, The penalty target function is: wherein, F represents fitness, is a penalty factor, Y 1 is the modified nodal admittance matrix, Y 2 is the modified power nodal admittance matrix.

6. The DC grid optimal power flow acquisition method of claim 1, wherein, obtaining a solution of the optimal power flow OPF of the power grid, specifically comprising the following steps: The system monitors the maximum input power P in real time imax The sum of the new energy power P W +P pv The system compares P imax and P W +P pv with the total load demand P L ; When P imax <P L is detected, the power sum is insufficient to meet the load demand, and the demand side response mechanism is started; When P imax > P L , P W + P pv > P L , the excess power is allowed to be fed back to the grid for consumption. If P imax > P L , P W +P pv <P L , the new energy power generation is not surplus but the overall power supply is still surplus, then the power grid and the new energy power station are coordinated to supply power on demand; The adjustment process takes the minimization of network loss as the core target, dynamically optimizes the power transmission setting value of the power grid, the output scheme of the new energy power station and the load adjustment amount of each node through the VOA algorithm for optimal power flow acquisition, and obtains the solution of the optimal power flow OPF of the power grid.

7. A direct current grid optimal power flow acquisition system characterized by, The method comprises the following steps: A dataset construction module is configured to obtain wind and light output reference values, basic load rated power and DC power grid parameters of a target power grid system, construct a wind and light-load-demand side response joint uncertainty set, reduce the dimension of data in the joint uncertainty set, and obtain decision variables after dimension reduction; A normal distribution error is added to the wind and light output reference values to obtain simulated output data, a fluctuation is superimposed on the basic load rated power to obtain actual load data, and power boundaries and price elasticity of demand side response are converted into adjustable variable constraints as population data for initialization of a VOA algorithm; A model construction module is configured to take the minimization of total system network loss as a target, construct an initial network loss target function, take the maximum / minimum of bus voltage amplitude, power conservation of each bus, and maximum / minimum of each external power output in the power grid system as constraint conditions, embed a fixed penalty factor into the initial network loss target function, and obtain a penalty target function; A VOA algorithm optimization module is configured to initialize population data of the VOA algorithm according to the range of decision variables after dimension reduction, calculate penalty target function values corresponding to candidate solution data of all main vines and side vines, select the candidate solution data with the optimal fitness through pairwise comparison, output optimal voltage, power and demand side response adjustment amount data if the current optimal candidate solution data meets any threshold, return to initialization for continuous iteration if the current optimal candidate solution data does not meet any threshold, update input data with real-time measured source and load data, repeatedly execute VOA optimization, and obtain the VOA algorithm for optimal power flow acquisition. A parameter calculation module is configured to acquire a maximum input power P imax The sum of the new energy power P W +P pv The sum of the new energy power P imax and P W +P pv The total load demand P L is compared; according to the comparison result, the VOA algorithm for optimal power flow acquisition is used to dynamically optimize the grid transmission power setting value, the new energy power station output scheme and the load regulation amount of each node, so as to obtain the solution of the optimal power flow OPF.

8. A computer device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program comprises instructions that, when executed by the processor, cause the processor to perform the method of any one of claims 1-7. The processor executes the computer program to implement the steps of the method of any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program can execute the steps of the method of any one of claims 1 to 6 when loaded by the processor.

Images