Mitchell K. Colby

Optimizing ballast design of wave energy converters using evolutionary algorithms

Wave energy converters promise to be a viable alternative to current electrical generation methods. However, these generators must become more efficient before wide-scale industrial use can become cost-effective. The efficiency of these devices is primarily dependent upon their geometry and ballast configuration which are both difficult to evaluate, due to slow computation time and high computation cost of current models. In this paper, we use evolutionary algorithms to optimize the ballast geometry of a wave energy generator using a two step process. First, we generate a function approximator (neural network) to predict wave energy converter power output with respect to key geometric design variables. This is a critical step as the computation time of using a full model (e.g., AQWA) to predict energy output prohibits the use of an evolutionary algorithm for design optimization. The function approximator reduced the computation time by over 99% while having an average error of only 1.5%. The evolutionary algorithm then optimized the weight distribution of a wave energy generator, resulting in an 84% improvement in power output over a ballast-free wave energy converter.

Implicit adaptive multi-robot coordination in dynamic environments

Multi-robot teams offer key advantages over single robots in exploration missions by increasing efficiency (explore larger areas), reducing risk (partial mission failure with robot failures), and enabling new data collection modes (multi-modal observations). However, coordinating multiple robots to achieve a system-level task is difficult, particularly if the task may change during the mission. In this work, we demonstrate how multiagent cooperative coevolutionary algorithms can develop successful control policies for dynamic and stochastic multi-robot exploration missions. We find that agents using difference evaluation functions (a technique that quantifies each individual agents contribution to the team) provides superior system performance (up to 15%) compared to global evaluation functions and a hand-coded algorithm.

An Evolutionary Game Theoretic Analysis of Difference Evaluation Functions

One of the key difficulties in cooperative coevolutionary algorithms is solving the credit assignment problem. Given the performance of a team of agents, it is difficult to determine the effectiveness of each agent in the system. One solution to solving the credit assignment problem is the difference evaluation function, which has produced excellent results in many multiagent coordination domains, and exhibits the desirable theoretical properties of alignment and sensitivity. However, to date, there has been no prescriptive theoretical analysis deriving conditions under which difference evaluations improve the probability of selecting optimal actions. In this paper, we derive such conditions. Further, we prove that difference evaluations do not alter the Nash equilibria locations or the relative ordering of fitness values for each action, meaning that difference evaluations do not typically harm converged system performance in cases where the conditions are not met. We then demonstrate the theoretical findings using an empirical basins of attraction analysis.

Design of Complex Engineered Systems Using Multi-Agent Coordination

In complex engineering systems, complexity may arise by design, or as a by-product of the system’s operation. In either case, the cause of complexity is the same: the unpredictable manner in which interactions among components modify system behavior. Traditionally, two different approaches are used to handle such complexity: (i) a centralized design approach where the impacts of all potential system states and behaviors resulting from design decisions must be accurately modeled and (ii) an approach based on externally legislating design decisions, which avoid such difficulties, but at the cost of expensive external mechanisms to determine trade-offs among competing design decisions. Our approach is a hybrid of the two approaches, providing a method in which decisions can be reconciled without the need for either detailed interaction models or external mechanisms. A key insight of this approach is that complex system design, undertaken with respect to a variety of design objectives, is fundamentally similar to the multi-agent coordination problem, where component decisions and their interactions lead to global behavior. The results of this paper demonstrate that a team of autonomous agents using a cooperative coevolutionary algorithm (CCEA) can effectively design a complex engineered system. This paper uses a system model of a Formula SAE racing vehicle to illustrate and simulate the methods and potential results. By designing complex systems with a multi-agent coordination approach, a design methodology can be developed to reduce design uncertainty and provide mechanisms through which the system level impact of decisions can be estimated without explicitly modeling such interactions. [DOI: 10.1115/1.4038158]

Coordination and control for large distributed sensor networks

As the complexity of power plants increase, so does the difficulty in accurately modeling the interactions among the subsystems. Distributed sensing and control offers a possible solution to this problem, but introduces a new one: how to ensure that each subsystem satisfying its control objective leads to the safe and reliable operation of the entire power plant. In this work we present a distributed coordination algorithm that offers safe, reliable, and scalable control of a distributed system. In this approach, each system component uses a reinforcement learning algorithms to achieve its own objectives, but those objectives are derived to coordinate implicitly and achieve the system level objective. We show that in a Time-Extended Defect Combination Problem where the agents need to determine when and whether or not they should be sensing in order to maintain QoS in a system, the proposed method outperforms traditional methods by up to two orders of magnitude.

Multiobjective Neuroevolutionary Control for a Fuel Cell Turbine Hybrid Energy System

Increased energy demands are driving the development of new power generation technologies with high efficient. Direct fired fuel cell turbine hybrid systems are one such development, which have the potential to dramatically increase power generation efficiency, quickly respond to transient loads (and are generally flexible), and offer fast start up times. However, traditional control techniques are often inadequate in these systems because of extremely high nonlinearities and coupling between system parameters. In this work, we develop multi-objective neural network controller via neuroevolution and the Pareto Concavity Elimination Transformation (PaCcET). In order for the training process to be computationally tractable, we develop a computationally efficient plant simulator based on physical plant data, allowing for rapid fitness assignment. Results demonstrate that the multi-objective algorithm is able to develop a Pareto front of control policies which represent tradeoffs between tracking desired turbine speed profiles and minimizing transient operation of the fuel cell.

Neuroevolution of a Hybrid Power Plant Simulator

Ever increasing energy demands are driving the development of high-efficiency power generation technologies such as direct-fired fuel cell turbine hybrid systems. Due to lack of an accurate system model, high nonlinearities and high coupling between system parameters, traditional control strategies are often inadequate. To resolve this problem, learning based controllers trained using neuroevolution are currently being developed. In order for the neuroevolution of these controllers to be computationally tractable, a computationally efficient simulator of the plant is required. Despite the availability of real-time sensor data from a physical plant, supervised learning techniques such as backpropagation are deficient as minute errors at each step tend to propagate over time. In this paper, we implement a neuroevolutionary method in conjunction with backpropagation to ameliorate this problem. Furthermore, a novelty search method is implemented which is shown to diversify our neural network based-simulator, making it more robust to local optima. Results show that our simulator is able to achieve an overall average error of 0.39% and a maximum error of 1.26% for any state variable averaged over the time-domain simulation of the hybrid power plant.

Design of Complex Engineering Systems Using Multiagent Coordination

In complex engineering systems, complexity may arise by design, or as a by-product of the system’s operation. In either case, the root cause of complexity is the same: the unpredictable manner in which interactions among components modify system behavior. Traditionally, two different approaches are used to handle such complexity: (i) a centralized design approach where the impacts of all potential system states and behaviors resulting from design decisions must be accurately modeled; and (ii) an approach based on externally legislating design decisions, which avoid such difficulties, but at the cost of expensive external mechanisms to determine trade-offs among competing design decisions. Our approach is a hybrid of the two approaches, providing a method in which decisions can be reconciled without the need for either detailed interaction models or external mech⇤Address all correspondence to this author. anisms. A key insight of this approach is that complex system design, undertaken with respect to a variety of design objectives, is fundamentally similar to the multiagent coordination problem, where component decisions and their interactions lead to global behavior. The design of a race car is used as the case study. The results of this paper demonstrate that a team of autonomous agents using a cooperative coevolutionary algorithm can effectively design a Formula racing vehicle. INTRODUCTION Complex engineering systems, such as state-of-the-art aircraft, advanced power systems, unmanned aerial vehicles, and autonomous automobiles, are required to operate dependably in an ever widening variety of environmental conditions, over a wide range of missions. Such systems must be cost-effective Proceedings of the ASME 2016 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference IDETC/CIE 2016 August 21-24, 2016, Charlotte, North Carolina

Learning Based Control of a Fuel Cell Turbine Hybrid Power System

Increased demands for energy are driving development of new technologies for power generation with high efficiency. Direct fired fuel cell turbine hybrid systems are one such development, which promise to drastically increase power generation efficiency, respond to transient loads quickly, and offer fast start up times. However, traditional control techniques are inadequate in these hybrid energy systems because of high nonlinearities and coupling between system parameters, as well as the lack of an accurate system model. In this work, we evolve a neural network controller for a hybrid fuel cell turbine system. In order to allow for neuroevolution to be computationally tractable, we develop a computationally efficient simulator based on real plant data. Results demonstrate that the proposed controller can accurately control plant parameters such as the fuel cell inlet flow rate to within 0.04% of a desired setpoint, and is robust to noise in both system sensors and plant actuators.