Multi-Time Scale Energy Management and Optimal Dispatch Scheme for Islanded DC Microgrid Operation

DC microgrid has outstanding advantages in accepting new energy generation system and DC load, etc. This paper takes theislandedDC microgrid as the research object, and proposes a multi-time scale optimization scheduling scheme for theislandedDC microgrid. On the basis of satisfying the power balance of the DC bus and considering the factors of economy and environmental protection, this scheme realizes the optimized operation of the DC micro-grid under the isolated island mode. The optimization scheme consists of two stages: day-ahead scheduling and intraday scheduling. According to the short-term forecast data in advance, the day-ahead dispatching takes economic and environmental protection as the target todispatch controllable unit. In this paper, intraday scheduling maintains the power of the battery energy storage and interruptible loads determined by day-ahead scheduling. Taking ultra-short-term prediction data as the premise, intraday scheduling takes the minimum deviation between the output power of controllable power supply and the corresponding value of day-ahead scheduling as the target to formulate scheduling scheme. Finally, an example is given to verify the feasibility of the optimized scheduling scheme by simulating the proposed program on MATLAB.


Introduction
In recent years, the permeability of photovoltaic (PV) and wind turbine (WT) and other new energy sources in the micro-grid has gradually increased, and the power load has diversified, so the DC micro-grid has significant advantages [1].In many cases, the microgrid is isolated island mode, for example, the power supply in remote mountainous areas is inconvenient, the main network is disconnected due to climate reasons, and the main network is scheduled for power outage, etc. Therefore, it is very necessary to optimize the energy scheduling under isolated island operation.
At present, some references have studied the energy management of isolated island operation of micro-grid. Reference [2][3][4] aimed at the micro-grid system under isolated island operation mode, established the operation optimization model of isolated island micro-grid with the lowest total operation cost that can be achieved within the system dispatching cycle as the optimization goal.
Reference [5] took the variation of energy storage as a reference and cut the load according to the importance degree to ensure the reliable operation of the system. Reference [6] combined the GOSSIP distributed algorithm with the cost function of the system, and proposed the optimal scheduling method of distributed economy. References [7][8] studied the uncertainty of renewable energy forecast data. Reference [7] adopted random variables to simulate the uncertainty of forecast data. Reference [8] proposed a two-step optimal scheduling method based on power supply classification, which is based on dividing the power supply into two types: base-load power supply and frequency-modulated power supply. The above references are all about the system operation cost as the target, and more influencing factors are not considered when the system is running.
There are also some references that takes environmental factors into account. References [9][10][11] studied the economic cost and environmental cost of micro-grid. Reference [9] used fitting functions for wind and light energy output and load power to obtain data. Reference [10] adopted the interactive multi-objective optimization method of evaluation function. Reference [11] optimized load distribution. Reference [12] studied the energy efficiency of the cold, heat and power connection system in the isolated island operation. References [13][14][15] comprehensively considered the output power cost, demand side management cost and environmental governance cost of the power supply. Reference [14] studied the planned island operation mode and took the charging and discharging power of the battery as an optimization variable. Reference [15] took the minimum of flexibility insufficiency rate and operating cost as the target. Although the above references have considered the economy and environmental protection of isolated island operation of micro-grid, they have not studied the uncertainty of forecast data.
To this end, this paper proposes a multi-time scale optimal dispatch scheme for the isolated island mode of DC microgrid. Firstly, the structure of DC microgrid and the modeling of each unit are introduced. Secondly, the energy management method of system operation is introduced. It includes day-ahead scheduling and intraday scheduling. Day-ahead scheduling aims at system economy and environmental protection to schedule the operation of controllable units. Based on the day-ahead scheduling to determine the interruptible load and energy storage state, intraday scheduling aims to minimize the deviation between the output power of each controllable power source and the corresponding value of day-ahead scheduling. Finally, using the yalmip toolbox of MATLAB and calling cplex for simulation, the rationality and effectiveness of the proposed scheduling method in this paper are verified.

Unit Modeling
2.2.1. PV unit model. In this paper, the PHOTOVOLTAIC power generation system works in the maximum power point tracking mode. According to the PV maximum output power under standard test conditions provided by the manufacturer, taking solar radiation intensity S and environmental temperature T as input variables, the output power of the photovoltaic power generation system is calculated as follows: PPV is the actual output power of photovoltaic cells, Pm is the maximum output power of photovoltaic cells under standard test conditions, S is the actual solar radiation intensity, Sref is the solar radiation intensity under standard test conditions, T is the actual battery temperature, Tref is the battery temperature under standard test conditions, and kPV is the power temperature coefficient.

WT unit model.
The output power of wind power system is mainly determined by wind speed and wind energy utilization coefficient. In this paper, the generator set adopts the variable pitch control mode, and the relationship between the power and wind speed of the wind turbine is as follows: is the actual output power of wind turbine, v is the actual wind speed, vci is the input wind speed, vr is the rated wind speed, vco is the cut out wind speed, Pr is the rated output power of wind turbine.

2.2.
3. FC unit model. FC adopts natural gas fuel cell. Its active power output is proportional to fuel input. The relationship between fuel cost and output power is as follows: Cfc is the cost of fuel cell, Cfuel is the price of fuel, L fuel is the low calorific value, Pfc is the output power of fuel cell, and ηfc is the power generation efficiency of the FC.

MT unit model.
MT uses natural gas fuel for power generation, and determines the power generated by the micro-turbine according to the fuel volume. The relationship between fuel cost and output power of micro gas turbine is as follows: Cmt is the fuel cost of the micro-turbine, Cfuel is the fuel price, Pmt is the power produced by the micro-turbine, and ηmt is the power efficiency of the MT.

Energy storage unit model.
Battery is selected as the energy storage equipment in this paper. The state of charge (SOC) of battery is as follows: Cr is the remaining capacity in the battery at time t, and CN is the capacity when the battery is fully charged.
is the charging and discharging power of the battery at time t. Pdis(t) is the discharge power. Idis(t) is an integer variable of 0-1, indicating the discharge state, 0 means no discharge, 1 means discharging. Pch(t) is the discharge power. Ich(t) is an integer variable of 0-1, indicating the charging state, 0 means no charging, and 1 means charging.
Δt is the unit of time.

Energy Management Scheme of DC Microgrid under Isolated Island
The scheme includes two stages: day-ahead scheduling and intraday scheduling. According to the short-term forecast data, the operation and maintenance costs and environmental protection costs are taken as objective functions to schedule controllable units. During the intraday phase, the interruptible load and energy storage state determined the previous day remain unchanged. In combination with the ultra-short-term prediction data, the daily planned power deviation of each controllable unit is taken as the target dispatching controllable unit.

Day-ahead scheduling
3.1.1. The objective function. Taking system economy (including interruption compensation cost of interruptible load, system operation and maintenance cost) and environmental protection as the objective function, the formula is as follows: T is the scheduling cycle of the day-ahead scheduling plan, taking 1 hour as the unit time period, where T=24. N is the type of distributed power supply. Pi is the power generated by a distributed power supply. Ci(Pi) is the fuel cost of distributed energy. ki is the operation and maintenance cost coefficient of distributed power supply. M is the number of interruptible loads. Ij is the running state of interruptible load, with 1 for run and 0 for interrupt. CILj is the compensation cost of interruptible load. PILj is the predicted consumption power of interruptible load. L is the type of pollutant. αj is the conversion coefficient of the jth pollutant. Aij is the unit emission of the ith pollutant from the jth distributed power supply.

The constraint.
The stable operation of the system needs to meet some constraints.
Power balance of the system is as follows: PFL is the power consumed by fixed load. Output constraints of the controllable unit are as follows: ,max load max 1 r Pimin is the minimum output of a distributed power supply. Pimax is the maximum output of a distributed power supply. Er is the error coefficient, which is determined according to short-term and ultra-short-term prediction errors of load and renewable energy. ΔPi,max is the maximum difference between short-term and ultra-short-term forecast data of renewable energy output at the same time. Pi,max is the maximum short-term forecast of renewable energy output. ΔPload,max is the maximum difference between short-term and super-short-term forecast data of important loads at the same time. Pload,max is the maximum value of short-term forecast of important loads.
The climbing constraints of controllable unit output are as follows: is controlled unit climbing power limitation of the maximum value.
Reliability constraints of interruptible loads are as follows: ,min 1 ( ) ( ) P

The constraint.
The stable operation of the system needs to meet some constraints.
Ppv(t), Pwt(t) and PFL(t) are ultra-short-term data of PV, WT and FL at time t respectively. Pbat(h) is the charging and discharging power of the battery in the day-ahead scheduling plan. Ij(h) and PIL(h) are respectively the operation state and power of the interruptible load in the day-ahead scheduling plan at h hours.

The Example Analysis
The structure shown in Figure1 is used in this example. The installed photovoltaic capacity is 100kW.The installed capacity of the fan is 180kW.The charging and discharging efficiency of the battery is 90%. Relevant configuration parameters of DC micro grid are shown in Table 1-3. The compensation factor of IL1 is 0.55/kWh. The compensation factor of IL2 is 0.45/kWh. The short-term forecast data are shown in Figure 2   The total cost of the system is 2316.22 yuan/day, of which the economic cost is 1948.31 yuan/day and the environmental cost is 367.91 yuan/day. It can be seen from Figure 3 that the power generation of FC accounts for a large proportion, which is related to the power generation efficiency and pollution control costs of micro-sources. When the load is light, the accumulator, FC and MT jointly maintain the load demand. During the peak load period, the FC generates electricity with full capacity, and the battery neither charges nor discharges, so the MT recharges the remaining power required by the load.
After adopting the method described in the paper, the scheduling results are as follows: . The state of charge of the battery of day-ahead scheduling The day-ahead dispatching cost is 2115.22 yuan/day, among which the operation and maintenance cost is 1833.18 yuan/day and the environmental protection cost is 282.03 yuan/day. When interruptible load participates in scheduling, the operation and maintenance cost and environmental protection cost of the system are significantly reduced. The output of micro sources such as FC and MT is significantly reduced, and the battery plays a greater role in the power balance of the regulation system.
The ultra-short-term prediction data are shown in Figure 9.  Figure 9. Ultra-short-term forecast data of microgrid Figure 10. Comparison of the power of FC and MT during intraday scheduling and day-ahead scheduling It can be seen from Figure 10 that the scheduling results of FC and MT of day-ahead scheduling and intraday scheduling are compared. It can be seen from Figure 10 that the fuel cell is mainly responsible for the intraday power adjustment.Due to the impact of the efficiency and pollution control cost of the micro-turbine, it assumes auxiliary functions for power fluctuation.The sample difference of power fluctuation of intraday dispatching controllable unit is 1.37.The total operating cost of the system is 2,128.99 Yuan/ day. The result of system operation shows that the scheme proposed in this paper not only improves the economy of system operation, but also increases the environmental protection.

Conclusion
This paper first introduces the optimal dispatch of dc microgrid under isolated island, and proposes a multi-time scale optimal dispatch scheme of DC microgrid under isolated island mode. The scheme is studied from two stages: day-ahead scheduling and intraday scheduling. The purpose of day-ahead scheduling is to optimize the operation of the system economically and environmentally. The goal of intraday scheduling is to minimize the deviation from the day-ahead scheduling plan. These two stages cooperate to achieve the goal of environmental protection and economic optimization. A numerical example verifies the feasibility and effectiveness of the proposed method.