Active distribution network robust recovery control method

1. Problem background: Uncertain factors in the active distribution network bring challenges to recovery control

1.1 Distribution network recovery control principle

The distribution network usually operates in a radial manner. When the grid fails, it will cause all the downstream areas of the line where the fault is located to be powered off. The switches in the distribution network are generally divided into two types: normally closed sectional switches and normally open communication switches. After the faults are located and isolated, the non-faulty power outage areas can be coordinated by coordinating the opening and closing states of the two types of switches. The load is transferred to Other feeders or the live branch of the same feeder to restore power. Therefore, the recovery control is essentially a switch combination optimization problem that satisfies the constraints of the distribution network operation.

1.2 Uncertainty in active distribution network

Since the operation of many devices requires manual execution, the actual power recovery is a time-consuming process. The main uncertainties during the period are:

1) Distributed power output fluctuations: The output of distributed power (DG), represented by photovoltaics and wind power, fluctuates due to weather and environmental factors. As the penetration rate of distributed power sources in the distribution network continues to increase, the impact of output uncertainty on recovery control will become increasingly significant.

2) Node load fluctuation: The node load value depends on the user's power consumption, and is a time variable that cannot be accurately predicted.

3) Load estimation error: Since there are few load measurement devices installed in the distribution network, most non-measurement nodes can only obtain pseudo-measurement data through typical load curve method or short-term load forecasting method, so the estimated value and true There may be a large deviation between the load values.

1.3 Problems with traditional deterministic algorithms

In the traditional recovery control algorithm, the distributed power output and load are represented by single-point prediction values, which are regarded as deterministic parameters. This determination decision model has security risks. For example, if the actual load during recovery is higher than the predicted value, the distributed power output is lower than the predicted value, and the recovery strategy generated by the deterministic algorithm may cause line overload or voltage overrun, which eventually leads to power failure recovery. Therefore, uncertainties in active distribution networks present challenges to traditional deterministic algorithms. This paper proposes a new robust recovery control method, which makes the recovery strategy always guarantee when these uncertain parameters (distributed power output, load fluctuation and estimation error) fluctuate within a given range. The grid is safe and can be used.

2. Algorithm model: Conservatively adjustable robust recovery control optimization model

Common deterministic modeling methods can be divided into stochastic models and robust models, but stochastic models require a given variable probability distribution, and there are computational difficulties. To this end, this paper constructs a two-stage robust optimization method for the distribution network restoration control decision problem, which includes the following steps:

Step 1: Describe the general recovery control decision as a mixed integer linear programming model whose objective function is to maximize the recovery of the de-energized load. The constraints include radial topology constraints, power balance constraints, line capacity constraints, and node voltage security constraints. . The model does not consider the uncertainty factor and is a deterministic recovery control optimization model (DROM).

Step 2: Based on historical data, construct an uncertainty interval for distributed power output and load.

Step 3: Based on the DROM and the uncertainty interval, establish a two-stage robust recovery control optimization model (RROM); the first stage is to generate the optimal power loss recovery strategy, and the second stage is in the uncertainty interval. Search for the worst volatility scenes.

Step 4: Introduce the Budget of Uncertainty technique to adjust the conservativeness of the model, so that the method can be traded off between robustness and conservatism.

Finally, the decision-making of recovery control considering distributed power output uncertainty and load uncertainty is established as a two-stage robust optimization model with conservative conservativeness. For such a two-stage problem, this paper uses the column constraint generation method to decompose it into the main problem and the sub-problem to solve iteratively. The recovery strategy generated by the model can guarantee that the distribution network operation constraints are not destroyed for all the fluctuation scenarios in the uncertainty interval.

3, the case verification

The PG&E 69-node power distribution system is used to verify the results. The worst-case fluctuation scenario test and Monte Carlo simulation are used to compare and analyze the recovery control effects of the deterministic model DROM and the robust model RROM.

1) The worst volatility scene test

An "N-1" fault scan was performed on the entire test system to extract the worst-case fluctuation scenario generated in the sub-problem as the test environment. The fault recovery situation of the deterministic model and the robust model is shown in Figure 1. The deterministic model represented by red has multiple recovery failures in the worst case scenario, and additional partial load is required (the recovery load is negative) The recovery strategy is feasible; and the robust method represented by blue guarantees the feasibility and optimization of the decision in the case of all line failures.

Figure 1 "N-1" scan results of the PG&E 69-node system

2) Monte Carlo simulation test

The actual distributed power output and node load are normal distributions with the predicted value as the center, and 3000 fluctuation scenarios are randomly generated for each fault situation as the test scenario. The load recovery of four typical line fault cases is shown in Table 1. The robust method always maintains a 100% recovery success rate and an optimal recovery load, while the deterministic method will have recovery performance in many scenarios. Poor, even recovering from failure.

4, the conclusion

In this paper, a conservative two-stage robust recovery control optimization model is proposed, which is established as a form of mixed integer linear programming. Based on a given uncertainty interval, the robust model can search for the worst fluctuation scenarios and generate recovery control strategies. The results of the example verify the robustness and optimization of the proposed method. This method can be used to solve the problem that the recovery control strategy is not feasible due to fluctuations in distributed power output, load changes and load measurement errors during the recovery process.