Optimized Configuration of Distributed Wind-Solar-Storage System Based on Wind-Solar Scene Generation

To achieve large-scale, high-proportion, high-quality sustainable development of new energy such as wind and solar, the integration of wind, solar, and storage is imperative. In this paper, the Latin hypercube sampling and rapid backward reduction methods are used to generate and reduce the wind-solar output data. Then, an optimization model based on distributed wind-solar-storage system is established, in which the storage capacity is the configurable variable and the system benefit is the optimization objective. Subsequently, the wind-solar output data is substituted into the model for solving, and the Cplex solver is used to solve the model. The results show that the optimal storage capacity of the model built in this paper is 5 MWh.


Introduction
The annual installed capacity of wind power and PV is more than 100 million kilowatts, and the annual power generation is more than 1 trillion kilowatt hours, which has become the new normal.Entering 2023, China's wind-solar-storage industry integration has accelerated significantly, a number of more than 10 billion yuan wind-solar storage base projects have landed one after another [1].The wind-solarstorage industry integration development is imperative [2].
In recent years, the new energy industry represented by wind-solar-storage has developed rapidly, and the installed capacity and power generation have broken through a record high.However, wind and solar power generation have randomness and volatility [3,4], which cannot independently support the demand for stable operation of power load.The overall and coordinated development of wind, solar, and storage integration requires a deep understanding of the reform logic of China's power industry and solving the high-cost problem faced by the integrated development of wind, solar, and storage.
At present, certain research has been carried out on the allocation of wind-solar-storage capacity.Li et al. proposed a joint planning method for wind, solar, and storage based on the analysis of unit output characteristics [5].Based on the working characteristics of the pumping unit in the oilfield scenario, Ding et al. established the calculation model of the output power of wind turbine, solar photovoltaic power generation, and energy storage device in the wind-solar-storage integration mode [6].Yang et al. proposed an optimal configuration scheme for a wind-solar-storage system considering the operating status and demand response of the distribution network under the integrated park electric energy substitution [7].Yang et al. constructed a microgrid power optimization configuration model for the independent operation of wind power, photovoltaic power generation, and battery energy storage devices [8].Wu et al. analyzed the capacity allocation and battery capacity of wind solar power generation under different operation strategies through the reliability test system [9].All the above studies have carried out some research on the allocation of wind-solar-storage capacity, but the research on distributed wind-solar-storage capacity allocation aiming at wind-solar uncertainty is still insufficient.
To alleviate the uncertainty of wind-solar-storage in the research of wind-solar-storage configuration, this paper adopts the Latin hypercube sampling method to simulate the generation of wind-solar scenes, substitutes the generated wind-solar scene data into the distributed wind-solar-storage model constructed in this paper with system revenue as the optimization objective to solve it, and uses Cplex to solve the model.

Distributed wind-solar-storage system structure
The distributed wind-solar-storage system is jointly operated by distributed wind power, photovoltaic, and energy storage.The distributed photovoltaic system mainly includes photovoltaic modules and photovoltaic inverters.Distributed wind power system mainly includes fans and inverters.Distributed energy storage system mainly includes energy storage units and converter equipment.The structure diagram is shown in Figure 1.

Wind-solar output scene generation method
Latin hypercube sampling is used to randomly generate corresponding samples [10].It is supposed that the sample number is n and the random number is m, and it can be expressed as The steps for sampling wind power and photovoltaic power are as follows [11]: 1) It is supposed that the probability distribution function of the values j x in the interval [ , ] F x ] of the probability density function is divided into n parts of equal probability intervals.
3) For any probability interval [ ], we randomly select any interval , where r is a uniformly distributed random variable in the interval [0,1].4) The corresponding sample value is obtained by inverting the normal distribution, and its value is 3. Optimal operation model for distributed wind-solar-storage system

Objective function
The system revenue of distributed wind solar storage is taken as the optimization objective of the system, whose objective function is as shown in Formula (1).
where S C is the system revenue, sell C is the revenue from wind and solar power sales, ess C is the cost of storage losses, and pun C is the cost of system output penalty deviation.The system revenue of distributed wind-solar-storage system is mainly from the revenue of the sale of electricity from the system, which can be expressed as Formula (2).The loss cost of energy storage is closely related to the battery cycle life.This paper considers that the main influence of battery cycle life is the depth of discharge.The loss cost is shown in Formula (3).
where N is the number of cycles of discharge of storage, % i O is the rate of life loss of the single discharge, ess D is the cost per unit of storage, and bat E is the rated capacity of storage.When the joint output of the distributed photovoltaics and distributed energy storage deviates from the planned output, the distributed wind-solar-storage system requires a shortfall penalty.The cost of output deviation penalty for the distributed optical storage system is shown in Formula (4).
where pun E is the penalty cost per unit of the output deviation and ( ) grid P t is the grid-connected power of the distributed solar-wind-storage system at time t.

Distributed photovoltaic operation mode
This paper analyses the contact line adjustment mode of distributed photovoltaics.The contact line adjustment mode is based on the predicted output curve reported by the power station end, and the dispatching center makes the photovoltaic generation output curve according to the power flow constraints sent from the power station.Its dispatching curve is shown in Figure 2.

Constraint conditions 1)
System balance constraint The distributed solar-wind-storage system needs to satisfy the power balance constraints of the system, which can be expressed as Formula ( 5).where max char P is the maximum value of the charge of storage and max dis P is the maximum value of discharge of storage.
The storage cannot be charged and discharged simultaneously, and its constraint conditions can be expressed as Formula (9).( ) ( ) 0 The SOC of storage and its capacity also need to satisfy the upper and lower limit constraints, which can be expressed as Formula (10) P t is the planned wind-solar-storage system output.

Scene generation and reduction
The paper uses the Latin hypercube to generate wind-solar scenes and sets the number of generated scenes to be 1000, totally generating 1000 sets of the date of wind-solar output, as shown in Figure 3.

Analysis of the optimal configuration of the system
Based on the above generation and reduction of the wind-solar scenes, the analysis of the optimal configuration of the distributed wind-solar-storage system constructed in this paper is carried out.The simulation parameters of the storage are shown in Table 1.The generation values of wind-solar scenes refer to the forecasted values of wind-solar output, and the photovoltaic forecasted values, wind power forecasted values, and the curve of grid-connected power values are shown in Figure 5.As is shown in Table 2, the optimal capacity value for distributed energy storage power plant is 5 MWh.The charging and discharging states of storage and the charge capacity in each time period are shown in Figure 6.As is shown in Figure 5 and Figure 6, when the predicted output of distributed photovoltaics is greater than the planned output of photovoltaics, the energy storage stores excess energy.When the sum of predicted outputs of distributed photovoltaics and distributed wind power is greater than the planned output of the system, the storage is charged.When the predicted output of the distributed optical storage system is lower, the storage is discharged to maintain the planned output of the system.
The power deviation of the distributed wind-soler-storage system is shown in Figure 7.

Conclusion
The Latin hypercube sampling method is used to simulate the uncertainty of wind and solar, and an optimization model based on distributed wind-solar-storage system is established.This model takes energy storage capacity as a configurable variable and system benefit as the optimization goal, which solves the problem that the independent operation of wind, solar, and storage is not high.The joint operation mode of wind, solar, and storage is adopted, which effectively improves the system's benefit and rationally allocates energy storage capacity.

Figure 1 .
Figure 1.Structure diagram of distributed wind-solar-storage system

Figure 2 .
Figure 2. Diagram of the scheduling curve of the liaison line adjustment mode

Figure 3 .
Figure 3. Diagram of the generation of wind-solar scenes

Figure 4 .
Figure 4. Diagram of reduced wind-solar scenes

Figure 5 .
Figure 5. Diagram of power values

Figure 6 .
Figure 6.Diagram of the state of storage

Table 1 .
Parameters of Storage.