How to achieve fully distributed optimal scheduling under multiple microgrid clusters?

With the advancement of the electricity market, microgrid clusters are an important application scenario for electricity market transactions. In subsequent research, we will study the optimization and operation of microgrid clusters in the spot market environment.

Main content

1) Microgrid group optimization scheduling model.

With the extensive application of renewable energy distributed generation technology, microgrids have attracted widespread attention as an effective access method for distributed power sources. However, a single microgrid has the disadvantages of limited capacity and weak anti-interference ability. Interconnecting multiple microgrids and operating them in a cluster is conducive to improving power supply reliability. The microgrid group studied in this paper is not connected to the large power grid, and there is power exchange and information interaction between adjacent microgrids. The microgrid contains distributed power sources such as photovoltaics, wind turbines, diesel generators, batteries, and power loads.

The optimization strategy in this paper is mainly to minimize the operating cost of the entire microgrid group by optimizing controllable variables such as diesel generator power and battery charging and discharging power. The optimization goal of the microgrid group is to minimize the total operating cost of all microgrids, including the battery charging and discharging losses of each sub-microgrid, diesel engine fuel consumption, and the cost of exchange power. The constraints of the microgrid group optimization problem can be divided into internal constraints of the sub-microgrid and global constraints of the microgrid group. The internal constraints of the sub-microgrid include internal power balance constraints, upper and lower limits of diesel engine output constraints, upper and lower limits of exchange power constraints, and battery operation constraints; the global constraints of the microgrid group refer to the exchange power balance constraints between microgrids, that is, the sum of the input or output power of all sub-microgrids is zero.

2) Synchronous ADMM algorithm and finite time consistency algorithm.

The iterative process of the standard ADMM method is that the sub-problems are carried out alternately in a pre-arranged order. The solution of the previous sub-problem will be substituted into the next sub-problem for optimization and solution. After all sub-problems have completed one iteration, the Lagrange multiplier will be updated.

The iterative process of two-region distributed optimization is written as follows:

For the optimization problem of microgrid groups, the exchange power balance constraint condition includes the exchange power variables of all sub-microgrids. In the distributed optimization process, the expected exchange power optimization results of other sub-microgrids need to be obtained in each iteration. In order to avoid iterative confusion, the synchronous ADMM algorithm is adopted to convert each sub-microgrid from asynchronous calculation to synchronous calculation during the iteration process. When performing optimization calculation on a sub-microgrid, the correction value of the previous expected exchange power calculation result is selected as the reference value for the next iteration.

The distributed algorithm used in this paper needs to obtain the average value of the expected exchange power of each microgrid during the iteration process. The limited consistency algorithm can transmit necessary information without a control center. In addition, the limited time consistency algorithm can still effectively transmit information when the communication topology changes.

3) Distributed optimization scheduling based on MPC.

Figure 1 MPC rolling optimization process

The rolling optimization process of MPC is shown in Figure 1. If the current system is running at time , the system state at time is collected. Based on the prediction information of the system disturbance variables in the future time domain, the optimal output of the diesel engine and the battery in the prediction time domain is obtained with the goal of minimizing the operating cost, but the optimization result is only used in one control time domain, and the optimization process is repeated at time . Here, only the energy storage SOC value is related to the state value of the previous moment, and the state variable that needs to be collected at each moment is the SOC value of the battery; the disturbance variable is the random wind and solar load power; the output of the diesel engine and the battery is the control variable of the system, and it is also the decision variable in the optimization problem.

MPC is used as the optimization scheduling strategy for the microgrid group. The whole day is divided into 96 moments for rolling optimization. The synchronous ADMM is used to perform distributed solution of the optimization model at each moment. During the distributed solution process of the synchronous ADMM, it is necessary to obtain the consistency information of each microgrid (that is, the average expected exchange power). The consistency algorithm is used to obtain the global consistency information when only communicating with adjacent microgrids.

4) Case analysis.

A microgrid group consisting of 4 MGs is taken as an example for analysis. Figure 2 shows the comparison of the operating costs of the microgrid group at each moment when the method in this paper and the real-time single-step optimization strategy are used. It can be seen that the difference in operating costs between the two methods mainly occurs from the 70th to the 96th moment, during which the load power is at a higher level. Each step of MPC optimization takes into account the state of the system for a period of time in the future. During the peak load period, the diesel engine and the battery are used to supply power at the same time, while in the single-step optimization, the battery is used first and then the diesel engine to supply power, which can effectively avoid the high-load operation of the diesel engine, thereby reducing the operating cost. The total operating cost of the microgrid group within 24 hours is calculated, and the use of MPC reduces the operating cost by 6.59% compared with the use of single-step optimization.

Figure 2 Comparison of operating costs under the two methods

Innovation

1) A synchronous ADMM is used to solve the optimization problem, and a finite-time consistency algorithm is introduced for information interaction to transmit the information required in the iterative process of the distributed algorithm to achieve fully distributed computing.

2) Considering the time coupling and multi-disturbance characteristics of the microgrid system, MPC and distributed algorithms are combined and applied to the optimization scheduling problem of the microgrid group. MPC can consider the system operation status in the future, reduce the impact of uncertain factors such as wind and solar power through rolling optimization, achieve better control effects, and reduce the total operating cost of the microgrid group.

Further research directions

The distributed optimization scheduling strategy of the microgrid group proposed in this paper is mainly aimed at the active power scheduling problem of the microgrid group, and verifies the effectiveness of MPC in the coordination of active power of the microgrid group, but does not involve the content of reactive power optimization. Therefore, MPC will be considered to be applied to the reactive power optimization of the microgrid group in the future, so as to better coordinate the reactive power distribution between each sub-microgrid to maintain voltage stability. In addition, with the advancement of the power market, the microgrid group is an important application scenario for power market transactions. In subsequent research, we will study the optimization and operation of the microgrid group under the spot market environment.