Frank Ferrese

Online Optimal Generation Control Based on Constrained Distributed Gradient Algorithm

In traditional power system, economic dispatch and generation control are separately applied. Online generation adjustment is necessary to regulate generation reference for real-time control to realize economic operation of power systems. Since most economic dispatch solutions are centralized, they are usually expensive to implement, susceptible to single-point-failures, and inflexible. To address the above-mentioned problems, this paper proposed a multi-agent system based distributed control solution that can realize optimal generation control. The solution is designed based upon an improved distributed gradient algorithm, which can address both equality and inequality constraints. To improve the reliability of multi-agent system, the N-1 rule is introduced to design the communication network topology. Compared with centralized solutions, the distributed control solution not only can achieve comparable solutions but also can respond timely when the system experiences change of operating conditions. MAS based real-time simulation results demonstrate the effectiveness of the proposed solution.

Distributed Subgradient-Based Coordination of Multiple Renewable Generators in a Microgrid

For a microgrid with high renewable energy penetration to work autonomously, it must maintain its own supply-demand balance of active power. Maximum peak power tracking algorithms, which emphasize high renewable energy utilization, may cause a supply-demand imbalance when the available renewable generation is more than demanded, especially for autonomous microgrids. Currently, droop control is one of the most popular decentralized methods for sharing active and reactive loads among the distributed generators. However, conventional droop control methods suffer from slow and oscillating dynamic response and steady state deviations. To overcome these problems, this paper proposes a distributed subgradient-based solution to coordinate the operations of different types of distributed renewable generators in a microgrid. By controlling the utilization levels of renewable generators, the supply-demand balance can be well maintained and the system dynamic performance can be significantly improved. Simulation results demonstrate the effectiveness of the proposed control solution.

Fully Distributed Coordination of Multiple DFIGs in a Microgrid for Load Sharing

When wind power penetration is high, the available generation may be more than needed, especially for wind-powered microgrids working autonomously. Because the maximum peak power tracking algorithm may result in a supply-demand imbalance, an alternative algorithm is needed for load sharing. In this paper, a fully distributed control scheme is presented to coordinate the operations of multiple doubly-fed induction generators (DFIGs) in a microgrid. According to the proposed control strategy, each bus in a microgrid has an associated bus agent that may have two function modules. The global information discovery module discovers the total available wind generation and total demand. The load sharing control module calculates the generation reference of a DFIG. The consensus-based algorithm can guarantee convergence for microgrids of arbitrary topologies under various operating conditions. By controlling the utilization levels of DFIGs to a common value, the supply-demand balance can be maintained. In addition, the detrimental impact of inaccurate and outdated predictions of maximum wind power can be alleviated. The generated control references are tracked by coordinating converter controls and pitch angle control. Simulation results with a 5-DFIG microgrid demonstrate the effectiveness of the proposed control scheme.

Distributed Multiple Agent System Based Online Optimal Reactive Power Control for Smart Grids

Under high penetration of renewable energy resources, more and more reactive power control devices are integrated into power grids. The large-scale deployment of these devices requires upgrading existing reactive power control solutions. To improve energy efficiency and voltage profiles of the power grids under different operating conditions, this paper proposes a fully distributed multiple agent system based optimal reactive control solution. To update its control setting, a reactive controller only needs information measured locally or obtained from its neighboring buses. The updating rules of the subgradient based algorithm are derived under mild assumptions. The solution is able to provide comparable steady state performance as that of centralized optimization solutions. Due to the simplicity of communication topology and the reduced amount of data to process, the solution can provide timely response to changes of operating conditions. Simulation studies with power systems of different sizes demonstrate the effectiveness of proposed control solution.

Performance analysis of mobile agent-based wireless sensor network

In this paper, we describe a reliability model which can be used to analyze the performance and power consumption in resource constrained, data rate scarce, mobile agent-based wireless sensor network (WSN) systems. The primary model is referred to as a generalize access structure congestion (GGC) system [3] which is an extended model from a circular sequential k-out-of-n congestion (CSknC) [2]. There are many other reliability models which can be used to study WSN systems, but they are not suitable to analyze and address mobile agent-based multisensory WSN systems. These systems are not based on a centralized architecture because they use mobile agent technologies to distribute decision tasks at local nodes. By employing mobile agent technologies, the systems can make accurate decisions quickly and reduce data rate and data redundancy problems. An important research problem is to determine how to maintain efficient duty cycle by using multiple types of sensors without centralized architecture and with mobile agent technologies. From the GGC model, we can develop a method to determine an optimal power management scheme by computing an efficient duty cycle in mobile agent-based multisensory WSN systems.

A BDI multi-agent approach for power restoration

The objective of this paper is to design and develop a Belief-Desire-Intention (BDI) agent-based approach for power system restoration. We describe a multiple bus electrical power system (multi-bus power system) as a market environment that consists of BDI bus agents representing two different characters, consumer and producer. These bus agents are able to balance the power system between power generation and load consumption while consumers explore the market and trade the power resource with producers. In addition, the power system will be naturally and efficiently split into power branches by the bus agents. When a fault occurs in the system, the bus agents can maximize the capacity of the served loads or minimize the loss of power loads (when load shedding may be the only option) in a timely manner. The proposed BDI multi-agent approach can be applied to any size or structure of the multi-bus power systems. It is shown from our simulation and results comparison that the proposed approach becomes more effective and efficient when the scale of the multi-bus power system expands.

Market-based resource allocation for distributed data processing in wireless sensor networks

In recent years, improved wireless technologies have enabled the low-cost deployment of large numbers of sensors for a wide range of monitoring applications. Because of the computational resources (processing capability, storage capacity, etc.) collocated with each sensor in a wireless network, it is often possible to perform advanced data analysis tasks autonomously and in-network, eliminating the need for the post-processing of sensor data. With new parallel algorithms being developed for in-network computation, it has become necessary to create a framework in which all of a wireless networks scarce resources (CPU time, wireless bandwidth, storage capacity, battery power, etc.) can be best utilized in the midst of competing computational requirements. In this study, a market-based method is developed to autonomously distribute these scarce network resources across various computational tasks with competing objectives and/or resource demands. This method is experimentally validated on a network of wireless sensing prototypes, where it is shown to be capable of Pareto-optimally allocating scarce network resources. Then, it is applied to the real-world problem of rupture detection in shipboard chilled water systems.

Resilient consensus control for linear systems in a noisy environment

This paper presents multi-agent based control of networked systems in the presence of environmental noise. The subsystems are assumed to be linear time invariant with Gaussian white noise appearing as an exogenous input. The control protocol is based on output information received from other subsystems through the communication channel. We show that the multi-agent system arrives at a collective weak consensus in the sense that the sum of the mean square state errors between various subsystems converges to a small bound. Resilience is demonstrated in that the controller maintains collective stability in the event of communication or subsystem failures, with a degradation of performance. Simulation results are presented to illustrate the methodology.

Market-based computational task assignment within autonomous wireless sensor networks

In recent years, improved wireless technologies have enabled the low-cost deployment of large numbers of sensors for a variety of applications across different engineering disciplines. Because of the computational resources (processing capability, storage capacity, etc.) distributed throughout these sensing networks, it is often possible to perform advanced data analysis tasks autonomously and in-network, eliminating the need for the post-processing of sensor data. With new parallel algorithms being developed for in-network computation, it has become necessary to create a framework in which the computational resources available throughout a wireless sensing network can be best utilized in the midst of competing computational requirements. In this study, a Pareto-optimal market-based method is developed in order to autonomously distribute various computational tasks with competing objectives and/or resource demands across available network resources. This method is experimentally validated on a network of wireless sensing prototypes.

Resiliency of linear system consensus in the presence of channel noise

This paper presents multi-agent based control of networked linear time invariant systems in a noisy environment. The control protocol is based on output information received from other subsystems through the communication channel, which imparts noise to the sensor data. We show that the sum of the mean square state errors between various subsystems converges to a small bound for the multi-agent system. It is apparent that a higher controller gain tends to make the networked system arrive at a consensus faster, while at the same time has the detrimental effect of enlarging the radius of consensus. Resilience of consensus is demonstrated in that the controller maintains collective stability in the event of communication or subsystem failures.