مدیریت ریزشبکه با پیکربندی مجدد احتمالاتی و تعهد واحد

ABSTRACT Microgrid operation and management using probabilistic reconfiguration and unit commitment

Distribution networks are reconfigured in order to power loss reduction, load balancing and service restoration in critical operational conditions [1]. Impact of MG on Distribution Network Reconfiguration (DNR) is discussed in [2–4]. To solve the optimal DNR problem for power loss minimization, the PSO algorithm using some scenarios generated by MCS is presented in [2]. Load Economic Dispatch (ED) and DNR, considering costs of generation and storage in MG, utility and network power loss as OF, are studied in [3,4]. The Distributed Generators (DGs) are considered by the stochastic nature according to forecasting weather data. However, load ED and DNR are not considered within the same time intervals.

MG reconfiguration is analyzed in [5–8]. A new algorithm is proposed to solve MG reconfiguration problem based on an ordered binary decision diagram in order to minimize power loss cost in [5]. A hybrid programming technique to solve MG reconfiguration problem to minimize power loss and service restoration is proposed in [6]. Considering the operational requirements, load maximizing and demand supply priority after fault, some methodologies that are based on genetic algorithm (GA) and graph theory, are used to reconfigure MG in [7]. Neglecting network power loss and line capacity, a scheme to recover much more loads with minimum switching operation is presented in [8]. However, stochastic nature of renewable energy resources and load demand in MGs are neglected in these studies.

MG UC and ED are discussed in [9–14]. A stochastic model for considering wind power uncertainty is investigated in [9]. Different scenarios are generated by MCS and are applied to solve UC problem. A probabilistic approach including point estimate method for handling uncertainties and a self-adaptive optimization algorithm for optimal energy management of MGs is proposed in [10]. The offered mutation technique makes the solution able to meet global optimum. A new algorithm based on an adaptive modified PSO algorithm to optimize multi-objective management of MG is presented in [11]. Three optimization algorithms are developed for optimal MG operation in [12]. A multi-objective optimization, using weight coefficient to coordinate the proportion of generation and environmental costs, is applied to the environmental and economic problem of MG in [13]. A new stochastic method using the probability distribution function of variables and the roulette wheel mechanism is proposed in [14]. Some scenarios are generated by using of the proposed probabilistic method. Then similar scenarios are eliminated. A self-adaptive bee swarm optimization algorithm is proposed in [15]. Different uncertainty modeling methods are reviewed and the 2m + 1 point estimate method is used for modeling uncertainties of load demands, market price, WT and photovoltaic systems. UC is investigated on a MG including several grid parallel PEM-fuel cell power plants in [16,17]. The objective is optimal sizing of storage devices and committed units’ output power is scheduled with 15 min step over a day. A two stage algorithm is proposed to solve the complexity of the problem imposed by stochastic nature of electrical/thermal load, photovoltaic and WT output power and market price.

Some novel aspects in DNR studies are represented in [18–24]. Time varying data characteristic is considered in [18,20]. A method to determine annual reconfiguration scheme, considering switching cost and time-dependent variables such as load profiles, is proposed in [18]. The best topology for each hour is determined in [19], aimed at minimizing power loss and switching cost. Considering time-varying loads, a probabilistic approach for optimal DNR to reduce the total cost of operation, including power loss and switching cost, is presented in [20]. The proposed method can obtain an optimal balance between the number of switching and the power loss.

Reconfiguration with different kinds of uncertain data is presented in [21,22]. Different kinds of uncertainty are modeled in [21], in order to assess stochastic distribution feeder reconfiguration in the presence of fuel cell power plants. Interval analysis is used in [22], to deal with imprecision and uncertainties in reliability input, electrical parameters and load data to present a reliability oriented reconfiguration method in order to enhance distribution network performance. A methodology to convert a distribution network to an autonomous MG is presented in [23]. The methodology determines number, site and size of DGs and structural modifications in distribution network. Multi-scenario analysis handled with decision theory concepts is applied in [24], to determine intraday distribution configuration. The determined configurations are then used to formulate a demand response scheme, aimed at demand reduction to further decreasing in distribution network losses.

However, none of the above-mentioned papers has considered the reconfiguration and UC simultaneously. As the review shows, probabilistic reconfiguration and UC for MG optimal management, is a novel operation scheme. Considering wind power and load demand uncertainties, simultaneous reconfiguration and UC for hourly scheduling is used to estimate MG’s benefit in the uncertain environment. The paper includes five sections. In section ‘Problem formulation’, problem formulation is introduced. In section ‘Proposed algorithm’, proposed algorithm is described. Section ‘Simulation results’ indicates simulation results and section ‘Conclusion’ is dedicated to conclusion.