Ocean engineering micro-grid capacity optimization configuration system and method considering source load uncertainty

Abstract

The invention relates to the technical field of micro-grid capacity optimal configuration, in particular to an ocean engineering micro-grid capacity optimal configuration system and method considering source load uncertainty, and the method comprises the steps: constructing an ocean engineering micro-grid architecture comprising a generator load test energy recovery system, and enabling the system to carry out energy recovery through a bidirectional AC / DC / AC power electronic converter; non-standard electric energy generated by a load test is converted into power frequency electric energy available for a micro-grid, and efficient recovery and smooth output of energy are realized in cooperation with an energy storage system; aiming at the strong uncertainty of the source and the load in the marine environment, a refined mathematical model of each unit is established, and a multi-target optimization model taking the negative minimization of the system full life cycle cost, the carbon emission, the power shortage probability and the system comprehensive energy quality improvement index as the target is constructed; and solving the Pareto frontier by adopting a multi-objective optimization algorithm. According to the method, cyclic utilization of test energy is realized, and a microgrid planning scheme which is high in robustness and comprehensive and optimal is provided.

Description

Technical Field

[0001] This invention relates to the field of microgrid capacity optimization and configuration technology, and more specifically, to a system and method for optimizing and configuring the capacity of marine engineering microgrids considering source-load uncertainty. Background Technology

[0002] my country is a major maritime power with a long coastline, vast maritime territory, and tens of thousands of islands, possessing abundant marine resources and extensive strategic maritime interests. Data shows that from 2016 to 2023, China's marine development index grew at an average annual rate of 2.9%, reaching 125.3 in 2023, a 3.0% increase over the previous year. Currently, the Yangtze River Delta region has formed a marine engineering equipment and high-tech shipbuilding industrial belt centered on Nantong, extending from Qidong in Nantong in the east to Yizheng in Yangzhou in the west. Shipbuilding scale accounts for approximately 40% of the national total, and marine engineering equipment accounts for approximately 25%. More than 30 major national equipment projects have been first manufactured and launched from Nantong.

[0003] Offshore engineering parks contain numerous high-capacity electrical equipment such as cranes, welding machines, and metal cold-working machinery, resulting in substantial electricity consumption and carbon emissions. This presents a significant demand and potential for energy conservation and carbon reduction. On one hand, offshore engineering parks consume large amounts of electricity, typically exceeding ten million kWh annually, with peak and off-peak electricity prices approaching 1 yuan per kWh. Demand-side response can significantly reduce costs. On the other hand, all offshore engineering projects under construction require generator load tests lasting approximately one month. Currently, these tests utilize dry-resistance loads to consume generator energy, which is difficult to utilize effectively. Furthermore, additional energy is required for cooling the experimental equipment. In recent years, the rapid advancements in renewable energy generation technology, large-capacity energy storage technology, power electronics technology, and advanced control technology have laid the foundation for developing smart microgrids in offshore engineering parks. Therefore, it is evident that offshore engineering parks and the generator load testing process present enormous energy-saving potential and room for carbon emission reduction.

[0004] Microgrids, as a new technological form of resource integration for distribution-side power generation, grid, load, and storage, are an effective means to promote the local consumption of large-scale distributed renewable energy and achieve the local aggregation and interaction of diverse new elements. The concept of microgrids was first proposed by the University of Wisconsin-Madison in 2001. After years of research and engineering practice, microgrid planning methods, control methods, and scheduling strategies have become relatively mature. Establishing green microgrids in industrial parks can enhance energy complementarity, improve energy utilization efficiency, and alleviate the pressure on the power grid.

[0005] In the research on the optimal design and control of microgrids in industrial parks, domestic and foreign academic institutions have conducted extensive studies, and have basically formed a complete set of optimal design methods and operation control strategies. Reference 1, "Optimal Energy Storage Configuration of Photovoltaic-Storage Microgrids Considering Demand Response and Source-Load Uncertainty" (Nan Bin et al., Power System Technology, Vol. 47, No. 4, 2023), proposes an optimal energy storage configuration strategy considering demand response and source-load uncertainty for the capacity configuration problem of grid-connected photovoltaic-storage microgrid energy storage systems. Reference 2, "A Hybrid Energy Storage Capacity Configuration Method for Microgrids Based on Non-Cooperative Game Theory" (Sun Haixia et al., Global Energy Internet, Vol. 4, No. 5, 2021), proposes a hybrid energy storage capacity optimization configuration method based on non-cooperative game theory to balance the economics of hybrid energy storage with the effect of tie-line power fluctuation mitigation. However, research on microgrids in offshore engineering parks is still scarce both domestically and internationally. This is because offshore engineering parks have distinct source and load characteristics, require multi-frequency power supply, and have the special requirement of energy recovery for generator load testing. Traditional park microgrid design methods and control strategies cannot be directly applied to offshore engineering park microgrids.

[0006] Therefore, there is an urgent need to study the optimal design methods and control strategies for microgrids in marine engineering parks. Through the synergy of "source-load-storage," the operational efficiency of microgrids in marine engineering parks can be significantly improved, and the flexible interaction between the microgrid and the power grid can support the efficient operation of the power grid. Existing literature lacks capacity optimization configuration methods specifically for the characteristics of marine engineering microgrids, such as load energy recovery and frequent start-stop operations during load tests. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of the prior art, the present invention provides an energy recovery system that includes generator load testing, which solves the technical problems faced in the capacity configuration of microgrids in marine engineering construction bases, such as energy waste in generator testing, strong uncertainty of source load, and difficulty in multi-objective coordination. In addition, the present invention also provides a system and method for optimizing the capacity configuration of marine engineering microgrids considering source load uncertainty.

[0008] Technical solution: According to a first aspect of the present invention, a marine engineering microgrid capacity optimization configuration system considering source load uncertainty is provided. This system is based on an energy recovery system incorporating generator load testing, and specifically includes: The scene reduction module is used to construct a joint probability distribution model based on the uncertainty characteristics of the marine environment, sample scenes according to the joint probability distribution model, and reduce the sampled random scene set to obtain multiple typical scenes and the occurrence probability of each typical scene. The objective optimization model construction module is used to construct an objective optimization model that conforms to a multi-objective function for each typical scenario. The multi-objective function includes the full life cycle cost objective, carbon emission objective, power shortage probability objective, and system comprehensive energy quality improvement index objective. The capacity configuration module is used to solve the target optimization model based on multiple constraints, obtain the Pareto optimal solution set, and determine the final optimal capacity configuration scheme from the Pareto optimal solution set.

[0009] Furthermore, including: The energy recovery system, which includes generator load testing, comprises: a bidirectional power electronic converter, a load, a generator set, an energy storage device, an isolation transformer, and a grid-connected cabinet. The load and generator set are connected to the bidirectional power electronic converter via the non-power frequency bus of the microgrid. The other end of the bidirectional power electronic converter is connected to the isolation transformer. The isolation transformer transmits data with the energy storage device. The energy storage device is connected to the power frequency bus of the microgrid via the grid-connected cabinet. The power frequency bus is also connected to other power frequency loads and is connected to the shore power switch. The bidirectional power electronic converter is used to convert the non-power frequency electricity generated by the generator set and load test into power frequency electricity available to the microgrid. The energy storage device is used to store the additional power frequency electricity generated during the generator and load test while recovering and reusing it to achieve optimal capacity configuration of the microgrid energy. The load includes shore power load and test load.

[0010] Furthermore, including: The scene reduction module includes: The initial scenario set generation unit is used to quantify the comprehensive impact of the nonlinear correlation between various characteristic parameters on the microgrid through a joint probability distribution model, and to perform stochastic simulation of the energy distribution of the microgrid in the future, thereby generating a large number of initial scenarios. Scene reduction unit, used to adopt improved based Wasserstein Distance K-means Clustering algorithms are used for scene reduction, and the improved one is based on Wasserstein Distance ensures that each initial scene is assigned to one and only one typical scene.

[0011] Furthermore, including: In the target optimization model construction module, the target of the system comprehensive energy quality improvement index is obtained by weighted summation of the test load energy recovery rate, renewable energy utilization rate, and renewable energy curtailment rate.

[0012] Furthermore, including: The capacity configuration module includes: The combined weight generation unit is used to determine the combined weight of capacity configuration using the subjective weighting method of AHP and the objective weighting method of entropy weighting. The optimal configuration generation unit is used to generate the optimal configuration based on the combined weights of the capacity configuration using Mahalanobis distance. TOPSISThe decision yields the optimal solution with the highest relative approximation.

[0013] On the other hand, the present invention also provides a method for optimizing the capacity configuration of marine engineering microgrids considering source load uncertainty, the method comprising: A joint probability distribution model is constructed based on the uncertainty characteristics of the marine environment. Scene sampling is performed based on the joint probability distribution model, and the random scene set obtained by sampling is reduced to obtain multiple typical scenes and the occurrence probability of each typical scene. Construct a multi-objective function optimization model that fits each typical scenario. The multi-objective function includes the life cycle cost objective, carbon emission objective, power shortage probability objective, and system comprehensive energy quality improvement index objective. The objective optimization model is solved based on multiple constraints to obtain a Pareto optimal solution set, and the final optimal capacity configuration scheme is determined from the Pareto optimal solution set.

[0014] Furthermore, including: The step of sampling scenarios based on the joint probability distribution model, and reducing the sampled random scenario set to obtain multiple typical scenarios and the occurrence probability of each typical scenario, includes: The combined impact of the nonlinear correlation between various characteristic parameters on the microgrid is quantified by a joint probability distribution model, and the energy distribution of the microgrid in the future is stochastically simulated to generate a large number of initial scenarios. Adopting an improved based Wasserstein Distance K-means Clustering algorithms are used for scene reduction, and the improved one is based on Wasserstein Distance ensures that each initial scene is assigned to one and only one typical scene.

[0015] Furthermore, including: The improvement is based on Wasserstein Distance K-means Clustering algorithms are used for scene reduction, and the reduction function is expressed as: ; The reduction function satisfies the following constraints: ; Where N is the initial number of scenes, and M is the number of scenes after reduction. For the first i The probability of an initial scenario. For the first i The feature vector of the initial scene For the first j Feature vectors of a typical scenario For the first jThe probability of a typical scenario To assign variables, thus obtaining a typical scenario set { } and its probability { }

[0016] Furthermore, including: The target for the overall energy quality improvement index of the system is expressed as follows: ; in, These are the weighting coefficients, and , To test the energy recovery rate of the load, Indicates the utilization rate of renewable energy. For renewable energy curtailment rate; Its renewable energy curtailment rate is expressed as: ; in, Typical scenario s Down t Wind power curtailment at all times Typical scenario s Down t Constant photovoltaic power abandonment Typical scenario s Down t Available wind power at any time Typical scenario s Down t Available photovoltaic power at any time Let be the probability of a typical scenario s, and T be the optimization period. S This represents the number of typical scenarios.

[0017] Furthermore, the method also includes: determining the energy recovery power and overall system efficiency of the microgrid under the optimal capacity configuration scheme, wherein the energy recovery power is expressed as: ; in, For overall system efficiency; for t Constant test power; This is a test status indicator function, where 1 indicates that the test is in progress and 0 indicates that the test is not in progress. The overall system efficiency is expressed as: ; in: The rectifier efficiency in a microgrid; The converter efficiency in a microgrid; Inverter efficiency in a microgrid; For the efficiency of energy storage devices.

[0018] Beneficial effects: Compared with the prior art, the present invention has the following advantages: This invention first constructs a marine engineering microgrid physical architecture including a generator load test energy recovery system. This system first connects the results of load tests and generator tests with a power electronic converter via the microgrid's non-power frequency bus, thereby converting the non-standard electrical energy generated by the load test into usable power frequency electrical energy for the microgrid. Combined with an energy storage system, it achieves efficient energy recovery and smooth output. Furthermore, the system uses an isolation transformer to connect the energy storage device and the power electronic converter, ensuring electrical isolation and microgrid safety protection. Therefore, this invention, through a high-power bidirectional power electronic converter and a lithium-ion battery energy storage system, achieves over 95% efficient energy recovery and utilization from generator tests, solving the problem of severe energy waste in traditional methods and resulting in significant economic benefits.

[0019] This invention enables the optimal capacity configuration of a microgrid system through multi-objective coordinated optimization using energy storage devices within the system. Firstly, due to the complex operating conditions of microgrids—namely, during generator and load testing, they are in an off-grid state during energy recovery, in a grid-connected state during energy storage discharge, and in a grid-connected state during non-load testing—and the frequent occurrence of generator start-ups, shutdowns, and failures, resulting in drastic and unpredictable changes in generator output, this invention establishes a multi-objective optimization model incorporating energy recovery benefits. This model includes life-cycle cost objectives, carbon emission objectives, power shortage probability objectives, and a negative value of the system's comprehensive energy quality improvement index. This results in a four-objective optimization model encompassing economic efficiency, environmental friendliness, reliability, and a negative value of the system's comprehensive energy quality improvement index. This innovative approach incorporates energy recovery benefits into the economic objective, achieving comprehensive life-cycle optimization and avoiding the limitations of single-objective optimization.

[0020] Secondly, this invention employs an improved method based on... Wasserstein Distance K-means The clustering algorithm reduces the number of scenes in the initial scene set, ensuring that each initial scene is assigned to one and only one cluster, and guarantees the probability conservation of the distribution after reduction. In other words, this application is equivalent to minimizing the distribution under hard allocation. Wasserstein The approximation of distances makes calculations more efficient in real-world scenarios, thereby enabling accurate control of the dynamic stability and security of power systems in complex and variable environments.

[0021] Finally, this invention employs the AHP-entropy weight method to determine weights, taking into account both subjective preferences and objective data characteristics, and combining Mahalanobis distance. TOPSIS Decision-making should consider the correlation between objective functions to improve the scientific nature and accuracy of decisions. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as limiting the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a topology diagram of a smart microgrid for marine engineering parks, suitable for efficient energy recovery during generator load testing, as described in an embodiment of the present invention. Figure 2 This is a schematic diagram of the marine engineering microgrid capacity optimization configuration system considering source load uncertainty as described in an embodiment of the present invention; Figure 3 This is a flowchart of the scene reduction method described in an embodiment of the present invention; Figure 4 The embodiments of the present invention are based on Copula Schematic diagram of joint uncertainty modeling; Figure 5 This refers to the multi-objective optimization Pareto front diagram described in the embodiments of the present invention; Figure 6 As described in the embodiments of the present invention TOPSIS Decision-making process diagram; Figure 7 This refers to the result of multi-objective coordinated optimization and decision-making as described in the embodiments of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0026] In order to solve the technical problems mentioned in the background art Figure 1 An energy recovery system incorporating generator load testing is provided. The system includes: a bidirectional power electronic converter, a load, a generator set, an energy storage device, an isolation transformer, and a grid-connected cabinet. The loads and generator sets are connected to the bidirectional power electronic converter via the non-power frequency bus of the microgrid. The other end of the bidirectional power electronic converter is connected to the isolation transformer. The isolation transformer transmits data with the energy storage device. The energy storage device is connected to the power frequency bus of the microgrid via the grid-connected cabinet. The power frequency bus is also connected to other power frequency loads and to the shore power switch. The isolation transformer serves to provide electrical isolation and safety protection, and to match the voltage level.

[0027] The bidirectional power electronic converter is used to convert the non-power frequency electricity generated by the generator set and load test into power frequency electricity available to the microgrid. The energy storage device is used to store the extra power frequency electricity generated during the generator and load test and to achieve the optimal capacity configuration of the microgrid energy while recycling it. The optimal capacity configuration of the microgrid system is obtained through multi-objective coordination optimization to ensure the economy of the microgrid. The load includes shore power load and test load.

[0028] In this embodiment, the system includes a high-power bidirectional power electronic converter and a lithium-ion battery energy storage system, which converts non-standard electrical energy generated by load testing, such as 11kV / 60Hz, into power frequency electrical energy usable by the microgrid, such as 10kV / 50Hz, and works with the energy storage system to achieve efficient energy recovery and smooth output.

[0029] like Figure 2 As shown, a capacity optimization configuration system for marine engineering microgrids considering source load uncertainty is provided, specifically including: The scene reduction module is used to construct a joint probability distribution model based on the strong uncertainty characteristics unique to the marine environment, to sample scenes based on the joint probability distribution model, and to reduce the sampled random scene set to obtain multiple typical scenes and the occurrence probability of each typical scene. The objective optimization model construction module is used to construct a multi-objective function optimization model that conforms to each typical scenario. The multi-objective function includes the full life cycle cost target, carbon emission target, power shortage probability target, and system comprehensive energy quality improvement index target. The capacity configuration module is used to solve the target optimization model based on multiple constraints, obtain the Pareto optimal solution set, and determine the final optimal capacity configuration scheme from the Pareto optimal solution set.

[0030] The scene reduction module includes: The initial scenario set generation unit is used to quantify the comprehensive impact of the nonlinear correlation between various characteristic parameters on the microgrid through a joint probability distribution model, and to perform stochastic simulation of the energy distribution of the microgrid in the future, thereby generating a large number of initial scenarios. Scene reduction unit, used to adopt improved based Wasserstein DistanceK-means Clustering algorithms are used for scene reduction, and the improved one is based on Wasserstein Distance ensures that each initial scene is assigned to one and only one typical scene.

[0031] In the target optimization model construction module, the target of the system comprehensive energy quality improvement index is obtained by weighted summation of the test load energy recovery rate, renewable energy utilization rate, and renewable energy curtailment rate.

[0032] The capacity configuration module includes: The combined weight generation unit is used to determine the combined weight of capacity configuration using the subjective weighting method of AHP and the objective weighting method of entropy weighting. The optimal configuration generation unit is used to generate the optimal configuration based on the combined weights of the capacity configuration using Mahalanobis distance. TOPSIS The decision yields the optimal solution with the highest relative approximation.

[0033] This invention also provides a method for optimizing the capacity configuration of marine engineering microgrids considering source load uncertainty, the method comprising: S1 constructs a joint probability distribution model based on the strong uncertainty characteristics unique to the marine environment, performs scenario sampling based on the joint probability distribution model, and reduces the random scenario set obtained by sampling to obtain multiple typical scenarios and the occurrence probability of each typical scenario.

[0034] In this embodiment, edge distribution fitting is performed based on historical wind speed data, historical sunshine data, and historical load data. Specifically: To address the inherent high uncertainty of the marine environment, a joint probability distribution model of wind speed, sunlight, and load is established. Specifically, this includes: Edge distribution establishment: Establish edge distributions for wind turbines, photovoltaics, and loads separately: ; in, Wind speed (m / s), , k for Weibull Distribution parameters, Light intensity (W / m²) For maximum light intensity, , for Beta Distribution shape parameters, l Load power (kW) , These represent the load mean and standard deviation.

[0035] Copula Joint distribution construction: using Gaussian distribution Copula Functions characterize nonlinear correlations between variables: ; in, These are the parameters of the correlation matrix; u 1, u 2, u 3: Uniformly distributed random variables; Φ θ : Multivariate normal distribution function; Φ -1 : Inverse function of the standard normal distribution.

[0036] like Figure 3 As shown, scene generation and reduction based on an improved clustering algorithm: an initial scene set is generated through Monte Carlo simulation, and an improved clustering-based algorithm is used. Wasserstein Distance Clustering for scenario reduction: ; The constraints are satisfied: ; in: N Initial number of scenes; M Number of scenes after reduction; For the first i The probability of each initial scenario; No. i Feature vectors of an initial scene; No. j Feature vectors of a typical scenario; No. j The probability of a typical scenario; Assign variables (0 or 1) to obtain a typical set of scenarios. and its probability .

[0037] Constraints Ensure that each initial scenario is assigned to one and only one cluster, i.e., hard assignment, with constraints. This ensures the conservation of probability in the distribution after reduction.

[0038] The improved version based K-means Distance Clustering algorithms are essentially a type of weighted algorithm. K- Wasserstein Variations of clustering, standard The goal is to minimize the sum of squared distances from all points to the cluster centers, i.e.: ; In the formula proposed in this invention, each point (scene) has a weight. π i Therefore, it is equivalent to in K-meansThe method introduced in this invention introduces probability weights, making high-probability scenarios have a greater impact on cluster centers. This invention's proposed method is similar to... The relationship of distance, means Distance, especially -2 distance is an indicator that measures the difference between two probability distributions, derived from optimal transport theory.

[0039] For discrete distributions K-means -2 distance is defined as: ; in, ij It is a transmission plan, indicating the transmission from the scenario. i Traffic to the scene. If the improved clustering formula proposed in this invention lets K-means ij = π i z ij Then the objective function and The cost function for a distance of -2 is consistent. In the standard... Wasserstein At a distance, ij It can be a continuous value, i.e., soft assignment, allowing a partial assignment of one scene to multiple typical scenes. However, in the clustering formula proposed in this embodiment, z ij It is binary, forcing each initial scene to be assigned to only one typical scene, i.e., hard assignment. Therefore, the clustering formula proposed in this embodiment is equivalent to minimizing under hard assignment. Wasserstein The approximation of distances makes computation more efficient in real-world scene reduction, ultimately resulting in a reduced multi-field set.

[0040] S2 constructs a multi-objective function optimization model that fits each typical scenario. The multi-objective function includes the life cycle cost target, carbon emission target, power shortage probability target, and system comprehensive energy quality improvement index target.

[0041] In this embodiment, a multi-objective optimization model incorporating energy recovery benefits is established, expressed as follows:

[0042] Among them, F( X Let ) be the objective function vector. X For the decision variable vector, , Photovoltaic capacity (kW) Wind power capacity (kW) For energy storage capacity (kWh), Energy storage capacity (kW), The capacity of the diesel generator is (kW). This refers to the number of operating hours of the diesel engine. Power of the energy recovery system (kW); C total ( X ): Total lifecycle cost target; E carbon ( X Carbon emission targets; ( X ): Power shortage probability target; Wasserstein ( X ): The negative value of the system's overall energy quality improvement index.

[0043] Correspondingly, the sub-objective functions are expressed as follows: Objective 1: Lifecycle Cost Target ; in: C total Total lifecycle cost (RMB); C inv Investment cost (RMB); C om Operation and maintenance costs (RMB); C fuel Fuel cost (RMB); C rep Equipment replacement cost (RMB); C grid : Grid interaction cost (yuan); C salv Equipment residual value (RMB); C recovery Energy recovery revenue (RMB); c cap,i : No. i Unit capacity investment cost of this type of equipment (RMB / kW); P i : No. i Equipment capacity (kW); c ess,cap Investment cost per unit capacity of energy storage (RMB / kWh); E ess Energy storage capacity (kWh); c ess,p Energy storage unit power investment cost (RMB / kW); P ess Energy storage capacity (kW); c rec Investment cost per unit power of energy recovery system (RMB / kW); P recEnergy recovery system power (kW); c elec Electricity price (RMB / kWh); P recovery ( t , s Scene s Down t Power recovered at any time (kW); d Discount rate; T Optimization cycle (years); S Number of scenes; π s : The probability of scenario s.

[0044] Target 2: Carbon emission targets: ; in: E carbon Carbon emissions over the entire life cycle (kg CO2); F i ( P i ( t , s )): No. i Fuel-like devices in scenarios s Down t Fuel consumption at any given time (kg or m³). i : No. i Carbon emission factor of fuel type (kg CO2 / unit of fuel); P grid,buy ( t , s Scene s Down t Power purchased from the grid at any time (kW); Wasserstein grid Carbon emission factor of grid electricity (kg CO2 / kWh).

[0045] Objective 3: Probability of Power Shortage ; in: Power shortage probability index; P load ( t , s Scene s Down t Load power at any time (kW); P supply ( t , s Scene s Downt Power supply at any time (kW).

[0046] Objective 4: A negative target for the overall system energy quality improvement index. The formula for calculating the negative value of the system's overall energy quality improvement index is as follows: ; in: γ 1, 2, γ 3 is the weighting coefficient, and 1+ Wasserstein 2+ 3=1, preferred value Wasserstein 1 = 0.5 2 = 0.3, γ 3 = 0.2. R recovery = E recovered / E test, total Indicates the energy recovery rate of the test load; U renewable = E renewable / E total Indicates the utilization rate of renewable energy; The renewable energy curtailment rate is: ; in: Renewable energy curtailment rate; P wind,curt ( t , s Typical scenarios s Down t Wind power curtailment at any given time (kW); P pv,curt ( t , s Typical scenarios s Down t Real-time photovoltaic power abandonment (kW); P wind,avail ( t , s Typical scenarios s Down t Available wind power capacity at any given time (kW); P pv,avail ( t , s Typical scenarios s Down t Available photovoltaic power at any given time (kW).

[0047] S3 solves the target optimization model based on multiple constraints to obtain a Pareto optimal solution set, and determines the final optimal capacity configuration scheme from the Pareto optimal solution set.

[0048] (3) In this embodiment, the corresponding constraint system is constructed as follows: Power balance constraints: ; in: Wasserstein gen A collection of power generation equipment; P i ( t , s ): No. i Generating equipment in scenarios s Down t Output power at any time (kW); P grid,sell ( t , s Scene s Down t Power sold to the grid at any given time (kW); P curt ( t , s Scene s Down t Power discarded at any time (kW).

[0049] Equipment operating constraints: ; in: P i,min : No. i Minimum output power (kW) for this type of equipment; P i,max : No. i Maximum output power (kW) of this type of equipment; R i : No. i Ramp-up rate of equipment (kW / h); U i,min : No. i Minimum operating state of the device (0 or 1); P i,rated : No. i Rated power (kW) of this type of equipment; U i (t,s): The first i Similar devices in scenarios s Down t The current running status (0 or 1).

[0050] Constraints of energy storage systems: ; in: min Minimum state of charge (%) of energy storage; LOLP max : Maximum state of charge of energy storage (%) end : State of charge (%) at the end of the optimization cycle; P ch,max Maximum charging power (kW); P disch,max Maximum discharge power ( CEQI ); U ch ( t , s ): Charging state variable (0 or 1); U disch ( t , s ): Discharge state variable (0 or 1).

[0051] like and EF As shown, the final configuration scheme is selected from the Pareto optimal solution set. First, the AHP-entropy weight method is used to calculate the combined weights, including: AHP subjective weighting: ; in: A : Judgment matrix; w : Feature vector (weight vector); max : Maximum eigenvalue; EF Consistency ratio; Consistency indicators; LOLP Random consistency index; n : Matrix order.

[0052] Objective weights in the entropy weight method: ; in: e j : No. j The entropy value of each target; p ij : No. i The first scheme is in the j The proportion of features under each objective; r ij : Decision matrix elements; m Number of options; n Target quantity; w jobj : No. j The objective weight of each objective.

[0053] Combined weights: ; in: w j : No. j The combined weights of the objectives; Preference coefficient (0≤ ω ≤1).

[0054] Mahalanobis distance decision making Distance calculation: ; in: D i + : No. i The distance from each solution to the ideal solution; D i - : No. i The distance from each solution to the negative ideal solution; v i : No. i The weighted normalized vector of each scheme; v + : Positive ideal solution vector; v - : Negative ideal solution vector; Σ: Covariance matrix.

[0055] No. i The relative similarity of the two solutions: ; Therefore, choose The largest possible solution is the optimal configuration.

[0056] In this embodiment, the output of the AHP-entropy weighting method combined weight calculation is the Mahalanobis distance. ω The input to the decision-making part. Specifically, the combined weights calculated and output by the AHP-entropy weight method. w j At Mahal distance This plays a crucial role in the decision-making process, specifically in the step of constructing the weighted normalized decision matrix. The process is as follows: 1. Initial decision matrix: Assume there is a normalized decision matrix, where each element... r ij Indicates the first i The first scheme is in the j The value of the indicator has been dedivided.

[0057] 2. Apply combined weights: Use the combined weights calculated in the first step. w j By weighting this matrix, we obtain the evaluation vector for each solution. v i .

[0058] For a solution i The weighted evaluation vector is: .

[0059] In summary, weight w j It amplifies the impact of key indicators while weakening the impact of secondary indicators.

[0060] In this embodiment, the method further includes: determining the energy recovery power and overall system efficiency of the microgrid under the optimal capacity configuration scheme, wherein the energy recovery power is expressed as: ; in, For overall system efficiency; for t Constant test power; This is a test status indicator function, where 1 indicates that the test is in progress and 0 indicates that the test is not in progress. The overall system efficiency is expressed as: ; in: The rectifier efficiency in a microgrid; The converter efficiency in a microgrid; Inverter efficiency in a microgrid; For the efficiency of energy storage devices.

[0061] To illustrate the effectiveness of the system and method described in this invention, the following examples are provided: Taking a large-scale marine engineering construction base as an example, this embodiment focuses on the isolated power grid system of the marine engineering construction base, which mainly carries out offshore wind power equipment manufacturing and offshore platform construction operations.

[0062] The base has complex load characteristics, including foundation load, process load (welding, hoisting, etc.), and special loads such as generator test loads. Peak load: approximately 1200 kW; average load: approximately 650 kW; total weekly energy consumption: approximately 109,000 kWh; special load: generator tests twice a week, 6 hours each time, power 1000-1200 kW. This invention generates the following basic random data: peak load: 1994 kW; average load: 736 kW; total weekly energy consumption: 123,583 kWh; number of tests: 2 times / week; photovoltaic fluctuation coefficient: 0.70; wind power fluctuation coefficient: 0.80.

[0063] The typical marine microgrid generator load test energy efficient recovery system proposed in this embodiment is as follows: ω As shown, in The bidirectional high-capacity power electronic converter converts the 11kV-60Hz non-power frequency electricity generated during generator load testing into 10kV-50Hz power frequency electricity. It absorbs and stores the extra electrical energy generated during generator load testing through energy storage technology, and ensures the stable operation of the microgrid. While recovering and utilizing surplus energy, it partially replaces the original load device, improves energy utilization, reduces carbon emissions, and reduces potential safety risks.

[0064] First technical aspect: Based on ω Theoretical source-charge joint uncertainty modeling method The method addresses the inherent high uncertainty of the marine environment by establishing a joint probability distribution model of wind speed, sunlight, and load. Specifically, it includes: (1) Establishment of edge distribution: Establish edge distribution for wind turbines, photovoltaics, and loads respectively: ; (2) Joint distribution construction: using Gaussian distribution ω Functions characterize nonlinear correlations between variables: ; (3) Scene generation and reduction based on improved clustering algorithm An initial scene set is generated through Monte Carlo simulation, using an improved method based on... Distance ω Clustering for scenario reduction: ; The constraints are satisfied: ; in: N Initial number of scenes; M Number of scenes after reduction; For the first i The probability of each initial scenario; No. i Feature vectors of an initial scene; No. j Feature vectors of a typical scenario; No. j The probability of a typical scenario; Assignment variable (0 or 1). Constraints. Ensure that each initial scene is assigned to one and only one cluster (hard assignment), constraint conditions This ensures the conservation of probability in the distribution after reduction.

[0065] The improved version based Distance ω Clustering algorithms are essentially a type of weighted algorithm. K- Variations of clustering, standard ω The goal is to minimize the sum of squared distances from all points to the cluster centers, i.e.: ; In the formula proposed in this embodiment, each point (scene) has a weight. π i， Therefore, it is equivalent to... The method introduced in this invention introduces probability weights, making high-probability scenarios have a greater impact on cluster centers. This invention's proposed method is similar to... ω The relationship of distance, Distance (especially) CUR The -2 distance is an indicator that measures the difference between two probability distributions, derived from optimal transport theory. For discrete distributions, -2 distance is defined as: ; in, Ω ij This is the transmission plan, representing the traffic from scenario i to scenario i. If the improved clustering formula proposed in this invention allows... ij = π i z ij Then the objective function and SOC The cost function for a distance of -2 is consistent. In the standard... At a distance, SOC ij It can be a continuous value (soft assignment), allowing a partial assignment of one scene to multiple typical scenes. However, in the clustering formula proposed in this embodiment, z ij It is binary, forcing each initial scene to be assigned to only one typical scene (hard assignment). Therefore, the clustering formula proposed in this invention is equivalent to minimizing under hard assignment. The approximation of distance makes it more computationally efficient in real-world scenario reduction.

[0066] use SOC The theoretical steps for establishing a joint probability distribution are as follows: Step 1: Marginal distribution parameter estimation, considering the following uncertainties: Wind speed uncertainty: Distribution, fluctuation coefficient 0.8-1.2; Illumination uncertainty: kW Distribution, fluctuation coefficient 0.8-1.2; Load uncertainty: Normal distribution, fluctuation coefficient 0.7-1.3.

[0067] Step 2: Parameter estimation.

[0068] Step 3: Scene generation and reduction.

[0069] final, Figure 5 Uncertainty modeling is complete, generating 5 scenarios. Typical results are as follows: As shown, successful application Figure 6 The theory addresses multiple uncertainties related to wind speed, illumination, and load, generating five typical scenarios to provide accurate input data for optimization.

[0070] The second technical aspect: a microgrid capacity configuration method based on four-objective coordinated optimization. Establish a multi-objective optimization model that incorporates the energy recovery benefits of generator load tests in marine engineering microgrids: (1) Objective function system ; (2) Detailed sub-objective function Objective 1: Total Life Cycle Cost

[0071] in: C total Total lifecycle cost (RMB); C inv Investment cost (RMB); C om Operation and maintenance costs (RMB); C fuel Fuel cost (RMB); C rep Equipment replacement cost (RMB); C grid : Grid interaction cost (yuan); C salv Equipment residual value (RMB); C recoveryEnergy recovery revenue (RMB); c cap,i : No. i Unit capacity investment cost of this type of equipment (RMB / kW); P i : No. i Equipment capacity (kW); c ess,cap Investment cost per unit capacity of energy storage (RMB / kWh); E ess Energy storage capacity (kWh); c ess,p Energy storage unit power investment cost (RMB / kW); P ess Energy storage capacity (kW); c rec Investment cost per unit power of energy recovery system (RMB / kW); P rec Energy recovery system power (kW); c elec Electricity price (RMB / kWh); P recovery ( t , s Scene s Down t Power recovered at any time (kW); d Discount rate; T Optimization cycle (years); S Number of scenes; π s : The probability of scenario s.

[0072] Objective 2: Carbon Emissions ; in: E carbon Carbon emissions over the entire life cycle (kg CO2); F i ( P i ( t , s )): No. i Fuel-like devices in scenarios s Down t Fuel consumption at any given time (kg or m³). i : No. i Carbon emission factor of fuel type (kg CO2 / unit of fuel); P grid,buy ( t , s Scene sDown t Power purchased from the grid at any given time (kW); λ grid Carbon emission factor of grid electricity (kg CO2 / kWh).

[0073] Objective 3: Improve the probability of power outages ; in: Power shortage probability index; P load ( t , s Scene s Down t Load power at any time (kW); P supply ( t , s Scene s Down t Power supply at any time (kW).

[0074] Objective 4: System Overall Energy Quality Improvement Index The formula for calculating the negative value of the system's overall energy quality improvement index is as follows: ; in: CR 1, 2, CI 3 is the weighting coefficient, and 1+ RI 2+ 3=1, preferred value γ 1 = 0.5 2 = 0.3, γ 3 = 0.2. R recovery = E recovered / E test, total Indicates the energy recovery rate of the test load; U renewable = E renewable / E tota This represents the utilization rate of renewable energy; where the renewable energy curtailment rate is: ; in: Renewable energy curtailment rate; P wind,curt ( t , s Scene s Down tWind power curtailment at any given time (kW); P pv,curt ( t , s Scene s Down t Solar power curtailment at any given time (kW); P wind,avail ( t , s Scene s Down t Available wind power at any given time (kW); P pv,avail ( t , s Scene s Down t Available photovoltaic power at any given time (kW).

[0075] (3) Construction of the constraint system Power balance constraints: ; in: TOPSIS gen A collection of power generation equipment; P i ( t , s ): No. i Generating equipment in scenarios s Down t Output power at any time (kW); P grid,sell ( t , s Scene s Down t Power sold to the grid at any given time (kW); P curt ( t , s Scene s Down t Power discarded at any time (kW).

[0076] Equipment operating constraints: ; in: P i,min : No. i Minimum output power (kW) for this type of equipment; P i,max : No. i Maximum output power (kW) of this type of equipment; R i : No. i Ramp-up rate of equipment (kW / h);U i,min : No. i Minimum operating state of the device (0 or 1); P i,rated : No. i Rated power (kW) of this type of equipment; U i (t,s): The first i Similar devices in scenarios s Down t The current running status (0 or 1).

[0077] Constraints of energy storage systems:

[0078] in: min Minimum state of charge (%) of energy storage; TOPSIS max : Maximum state of charge of energy storage (%) end : State of charge (%) at the end of the optimization cycle; P ch,max Maximum charging power (kW); P disch,max Maximum discharge power (kW); U ch ( t , s ): Charging state variable (0 or 1); U disch ( t , s ): Discharge state variable (0 or 1).

[0079] This invention establishes a multi-objective optimization model in Matlab, and uses gurobi+yalmip for modeling and solving. The optimization results are as follows: photovoltaic capacity: 394 kW; wind power capacity: 50 kW; energy storage system: 316 kWh / 30 kW; diesel engine capacity: 1108 kW; diesel engine operation: 66 hours / week.

[0080] Investment cost analysis: Total equipment investment: RMB 8.351 million; Multi-objective optimization performance indicators: Annualized cost over the entire life cycle: RMB 9.998 million / year; Annual carbon emissions: 3038.9 tons CO2 / year; Reliability penalty index: 2.03%; Renewable energy utilization rate: 30.6%. The goal is to achieve coordinated optimization of four objectives: economic efficiency, environmental friendliness, reliability, and the overall energy quality improvement index of the system, obtaining the optimal comprehensive solution.

[0081] The third technical aspect: Intelligent decision-making method based on combined weights This inventionTOPSIS Decision-making process diagram as follows As shown. The method selects the final configuration scheme from the Pareto optimal solution set: (1) Calculation of combined weights using the AHP-entropy weight method AHP subjective weighting: ; in: A : Judgment matrix; w : Feature vector (weight vector); Figure 1 max : Maximum eigenvalue; Consistency ratio; Figure 1 Consistency indicators; Random consistency index; n : Matrix order.

[0082] Objective weights in the entropy weight method: ; in: e j : No. j The entropy value of each target; p ij : No. i The first scheme is in the j The proportion of features under each objective; r ij : Decision matrix elements; m Number of options; n Target quantity; w j obj : No. j The objective weight of each objective.

[0083] Combined weights: ; in: w j : No. j The combined weights of the objectives; Copula : Preference coefficient, and 0 ≤ ≤1.

[0084] (2) Mahalanobis distance Copula decision making Distance calculation: ; in: D i + : No. i The distance from each solution to the ideal solution; D i- : No. i The distance from each solution to the negative ideal solution; v i : No. i The weighted normalized vector of each scheme; v + : Positive ideal solution vector; v - : Negative ideal solution vector; Σ: Covariance matrix.

[0085] No. i The relative similarity of the two solutions: ; Therefore, choose The largest possible solution is considered the optimal configuration. Based on the above implementation list, the following calculations are obtained: Relative similarity: 0.500.

[0086] Finally, according to a specific embodiment of the present invention, the simulation calculation is as follows: Copula As shown, the optimal capacity configuration is: Photovoltaic system: 393 kW, Wind power: 50 kW, Energy storage system: 315 kWh / 30 kW, Diesel generator: 1085 kW. Investment cost analysis: Total equipment investment: 8.283 million yuan; Multi-objective optimization performance indicators: Annualized cost over the entire life cycle: 9.963 million yuan / year, Annual carbon emissions: 3028.9 tons CO2 / year, Reliability penalty index: 2.03%, Renewable energy utilization rate: 30.6%. Relative proximity: 0.500; Energy recovery system benefit analysis: Total weekly test energy: 13.1 MWh, weekly recovered energy: 3.7 MWh, energy recovery rate: 28.5%, annual recovery revenue: 155,000 yuan; Overall system performance: Energy sufficiency rate: 98.0%, diesel power generation ratio: 66.0%, renewable energy ratio: 30.6%, diesel engine running time: 67 hours / week, energy storage backup time: 0.5 hours.

[0087] Therefore, through specific embodiments, the present invention has been successfully applied. Wasserstein Theoretical uncertainty modeling is used to achieve coordinated optimization of four objectives. The energy storage system realizes the function of recovering excess energy from generator load tests. The system configuration meets the actual needs of marine engineering. The diesel generator set serves as the main power source to ensure reliability.

[0088] This embodiment verifies the effectiveness and practicality of the patented technology of this invention in the capacity optimization configuration of marine engineering microgrids. Uncertainty modeling, multi-objective optimization and K-means Wasserstein K-means means K-means K-means Wasserstein Wasserstein Wasserstein Wasserstein γ γ Wasserstein Wasserstein γ Wasserstein Copula Weibull Beta Copula Copula Figure 4 Copula EF EF LOLP ω ω ω ω ω ω ω ω ω CUR Ω SOC SOC SOC TOPSIS Figure 4 λ CR CI RI γ γ TOPSIS TOPSIS Figure 7 TOPSIS Copula Copula TOPSISThe decision-making process yielded the optimal configuration scheme in terms of both technology and economics, achieving a balance between economic, environmental, and social benefits, and providing reliable technical support for the optimized configuration of similar isolated power grids.

[0089] For illustrative purposes, the foregoing description has been made with reference to specific embodiments. However, the foregoing illustrative discussions are not intended to be exhaustive or to limit the present disclosure to the precise forms disclosed. Numerous modifications and variations are possible in accordance with the foregoing teachings. These embodiments were chosen and described in order to best illustrate the principles of the present disclosure and its practical application, thereby enabling those skilled in the art to best utilize the disclosure and to employ various embodiments with different modifications to suit a particular intended application.

Claims

1. A system for capacity optimization configuration of offshore engineering microgrid considering source load uncertainty, characterized in that, The system is realized based on an energy recovery system comprising a generator load test, and the configuration system specifically comprises: a scenario reduction module, configured to construct a joint probability distribution model according to uncertainty characteristics of a marine environment, to perform scenario sampling according to the joint probability distribution model, and to reduce a random scene set obtained by sampling to obtain a plurality of typical scenes and occurrence probabilities of the typical scenes; a target optimization model construction module, configured to construct a target optimization model of a multi-target function conforming to each typical scene, the multi-target function comprising a life cycle cost target, a carbon emission target, a power shortage probability target, and a system comprehensive energy quality improvement index target; a capacity configuration module, configured to solve the target optimization model based on a plurality of constraint conditions, to obtain a Pareto optimal solution set, and to determine a final optimal capacity configuration scheme from the Pareto optimal solution set.

2. The system for capacity optimization configuration of offshore microgrid considering source load uncertainty according to claim 1, wherein, The energy recovery system comprising a generator load test comprises a bidirectional power electronic converter, a load, a generator set, an energy storage device, an isolation transformer, and a grid-connected cabinet. The load and the generator set are connected to the bidirectional power electronic converter through a non-power frequency bus of a microgrid, the other end of the bidirectional power electronic converter is connected to the isolation transformer, the isolation transformer is in data transmission with the energy storage device, the energy storage device is connected to a power frequency bus of the microgrid through the grid-connected cabinet, the power frequency bus is also connected to other power frequency loads and a shore power on-off station. The bidirectional power electronic converter is used to convert non-power frequency electricity generated by the generator set and the load test into power frequency electricity available for the microgrid, and the energy storage device is used to recycle the extra power frequency electricity generated during the generator and load test and realize optimal capacity configuration of microgrid energy, wherein the load comprises a shore power load and a test load. 3.The system for capacity optimization configuration of offshore engineering micro-grid considering source load uncertainty according to claim 1 or 2, characterized in that, The scenario reduction module comprises: an initial scene set generation unit, configured to quantify the comprehensive influence of nonlinear correlation between feature parameters on the microgrid through a joint probability distribution model, and to randomly simulate energy distribution of the microgrid in a future period of time, thereby generating a large number of initial scenes; a scene reduction unit for performing scene reduction using an improved distance-based Wasserstein clustering algorithm that ensures that each initial scene is assigned to one and only one representative scene. K-means Wasserstein distance-based​ 4. The system for capacity optimization and configuration of a microgrid in an offshore engineering considering source load uncertainty according to claim 3, wherein, In the target optimization model construction module, the system comprehensive energy quality improvement index target is obtained by weighted combination of an experimental load energy recovery rate, a renewable energy utilization rate, and a renewable energy abandonment rate.

5. The system for capacity optimization and configuration of offshore microgrid considering source load uncertainty according to claim 3, wherein, The capacity configuration module comprises: a combination weight generation unit, configured to determine capacity configuration combination weights by using an AHP subjective weight method and an entropy weight objective weight method; The optimal configuration generating unit is configured to employ Mahalanobis distance based on the capacity configuration combination weight TOPSIS The optimal solution with the highest relative closeness is determined. 6.A method for capacity optimization configuration of a marine engineering micro-grid considering source-load uncertainty, characterized in that, The method is realized in the energy storage device of claim 1, and the method comprises: constructing a joint probability distribution model according to strong uncertainty characteristics unique to a marine environment, performing scenario sampling according to the joint probability distribution model, and reducing a random scene set obtained by sampling to obtain a plurality of typical scenes and occurrence probabilities of the typical scenes; A multi-objective function optimization model is constructed for each typical scenario, and the multi-objective function includes a life cycle cost target, a carbon emission target, a power shortage probability target, and a system comprehensive energy quality improvement index target; The target optimization model is solved based on multiple constraint conditions to obtain a Pareto optimal solution set, and a final optimal capacity configuration scheme is determined from the Pareto optimal solution set.

7. The method of claim 6, wherein, The scene sampling is performed according to the joint probability distribution model, and the random scene set obtained by sampling is reduced to obtain multiple typical scenes and occurrence probabilities of the typical scenes, including: The nonlinear correlation between the feature parameters is quantified by the joint probability distribution model, the energy distribution of the microgrid in the future period of time is randomly simulated, and a large number of initial scenes are generated; The improved distance-based clustering algorithm is used to perform scene reduction, which ensures that each initial scene is assigned to one and only one canonical scene. Wasserstein The improved distance-based clustering algorithm is used to perform scene reduction, which ensures that each initial scene is assigned to one and only one canonical scene. K-means The improved distance-based clustering algorithm is used to perform scene reduction, which ensures that each initial scene is assigned to one and only one canonical scene. Wasserstein The improved distance-based clustering algorithm is used to perform scene reduction, which ensures that each initial scene is assigned to one and only one canonical scene.

8. The method of claim 7, wherein, The improved distance-based Wasserstein distance-based K- means The clustering algorithm performs scene reduction, the reduction function is represented as: ； The reduction function meets the constraint condition that: ； where N is the initial number of scenes, M is the number of scenes after reduction, is the probability of the initial scene, i is the feature vector of the initial scene, i is the feature vector of the typical scene, j is the probability of the typical scene, j is the assignment variable, so as to obtain the typical scene set and its probability .​​​​ 9. The method of claim 7, wherein, The system comprehensive energy quality improvement index target is expressed as: ； wherein, is a weight coefficient, and , is a test load energy recovery rate, represents a renewable energy utilization rate, is a renewable energy disposal rate; The renewable energy curtailment rate is expressed as: ； wherein, is a typical scenario s is a typical scenario t is a typical scenario is a typical scenario s is a typical scenario t is a typical scenario is a typical scenario s is a typical scenario t is a typical scenario is a typical scenario s is a typical scenario t is a typical scenario is a typical scenario S is a typical scenario 10. The method of claim 6, wherein, The method further includes determining an energy recovery power and a system comprehensive efficiency of the microgrid under the optimal capacity configuration scheme, the energy recovery power is expressed as: ； wherein, is the system overall efficiency; is t is the momentary test power; is the test status indicator function, 1 for test, 0 for no test; The system comprehensive efficiency is expressed as: ； wherein: is the rectifier efficiency in the microgrid; is the converter efficiency in the microgrid; is the inverter efficiency in the microgrid; is the energy storage device efficiency.

Images