Fig 1 - uploaded by Susan A. Frost
Content may be subject to copyright.
Solar array component of a Solar Powered Satellite, image credit NASA. space systems. The subsystem components of an Evolving System are designed to be stable as free-fliers, or unconnected components, but the Evolving System might fail to inherit stability at any step of the assembly, resulting in an unstable Evolved System. The fundamental topic of stability in Evolving Systems has been a primary focus of our Evolving Systems research (Balas & Frost, 2007; Frost & Balas, 2007a;b; 2008b;a; Balas & Frost, 2008; Frost, 2008). In this chapter, we develop an adaptive key component control method to ensure that stability is inherited in flexible structure Evolving Systems.
Source publication
In this chapter, we presented the motivation and the framework for Evolving Systems, a new area of aerospace research. We developed the adaptive key component controller approach to maintain stability in Evolving Systems that would otherwise fail to inherit the stability traits of their components. We showed that strict passivity, almost strict pas...
Context in source publication
Context 1
... Autonomous assembly of large, complex structures in space, or on-orbit assembly, is an excellent application area for Evolving Systems. For example, the Solar Power Satellite (SPS) is a conceptual space structure that collects solar energy, which is then transmitted to Earth as microwaves (NASA, 1995). The solar array of the SPS, as envisioned in fig. 1, is a complex structure that could be assembled from many actively controlled components to form a new system with a higher purpose. System stability is a trait that could be exhibited by an Evolving System or their components. We say that a subsystem trait is inherited by an Evolving System when the system retains the properties of ...
Citations
... List of multi-sport events 3 (1900s, 1910s, 1920s, 1930s, 1940s, 1950s, and 1960s). The similarity of the reconstructed matrix is progressive with respect to the states (time points). ...
We introduce a
System Evolution Recommender
(SysEvoRecomd) algorithm that uses a novel algorithm
Graph Evolution and Change Learning
(GECL) to do
system network reconstruction
. Internally, GECL uses
Deep Evolution Learner
(DEL) to learn about evolution and changes happened over a system state series. The DEL is an extension of the deep learning algorithm, which uses an
Evolving Connection Matrix
(ECM) representing temporal patterns of the
evolving entity-connections
for training incremental states. The DEL generates a
Deep System Neural Network
(Deep SysNN) to do network (graph) reconstruction. The SysEvoRecomd extracts the evolving characteristic of graph with deep neural network techniques. It aims to learn the evolution and changes of the system state series to reconstruct the system network. Our key idea is to design three variants of GECL based on three remodeled deep learning techniques: Restricted Boltzmann Machine (RBM), Deep Belief Network (DBN), and denoising Autoencoder (dA). Based on proposed SysEvoRecomd algorithm, we developed a SysEvoRecomd-Tool, which is applied on different evolving systems: software, natural language, multisport event, retail market, and IMDb movie genre. We demonstrated the usefulness of intelligent recommendations using three variants of GECL based on
RBM
,
DBN
, and
dA.</italic
... Service Oriented Computing supports loosely coupled distributed systems through the sharing of remote Web Service. An evolving distributed system is a kind of evolving system [1] [2] that often stored in a software repository to support its continuous evolution. Similar to systems in other domains [3], continuous changes to the requirements of distributed systems mandate their ongoing maintenance and evolution. ...
Changeability and evolvability analysis can aid an engineer tasked with a maintenance or an evolution task. This article applies change mining and evolution mining to evolving distributed systems. First, we propose a Service Change Classifier based Interface Slicing algorithm that mines change information from two versions of an evolving distributed system. To compare old and new versions, the following change classification labels are used: inserted, deleted, and modified. These labels are then used to identify subsets of the operations in our newly proposed Interface (WSDL) Slicing algorithm. Second, we proposed four Service Evolution Metrics that capture the evolution of a system's Version Series VS = {V
$_{1}$
, V
$_{2}$
,…,V
$_{N}$
}. Combined the two form the basis of our proposed Service Evolution Analytics model, which includes learning during its development phase. We prototyped the model in an intelligent tool named AWSCM (Automatic Web Service Change Management). Finally, we present results from experiments with two well-known cloud services: Elastic Compute Cloud (EC2) from the Amazon Web Service (AWS), and Cluster Controller (CC) from Eucalyptus. These experiments demonstrate AWSCM's ability to exploit change and evolution mining.
... In this approach, each component or subsystem is equipped with its own local controller which might be designed without any global knowledge about the physical interconnections (vs. decentralized control), and an additional low-dimension controller is designed and augmented to a carefully selected key component of the large-scale system in order to ensure the stability of interconnected system (see Frost and Balas 2,3 for the theoretical developments, and Habets et. al. 4 for an application). ...
Centralized and decentralized strategies have been widely used to control large-scale systems in which, by definition, subsystems are physically coupled to each others. In the former case, a powerful central processor is required to collect and process all subsystem's information, and generate the control commands. In the latter approach, the knowledge about physical interconnection is used to design a set of decentralized low-dimension controllers to be implemented locally around each subsystem. Motivated by the technological improvements in the embedded systems and theoretical advances in the distributed consensus of multiagent systems, we investigate a reference tracking problem for large-scale systems with partially known dynamics in which each subsystem has appropriate sensing, communication, and computation capabilities. We consider three modeling scenarios where the multiagent system is built by first-, second-, or mixed first- and second-order interconnected agents. By approximating the interconnected system dynamics, we develop a discrete-time domain framework for designing fixed and static communication topologies for the considered interconnected systems without local control implementations. For each modeling scenario, we characterize a sufficient condition on the sampling time interval in order to ensure the robust exponential tracking and obtain the guaranteed bounds on the convergence rate and quadratic cost function. We also comment on the discrete-time scenario compared to the equivalent continuous-time results.
... Evolving systems [1] [2][3] [4] [5] are the systems which keep on evolving in a changing, dynamic, and evolving environment. Suppose a state series S = < S1, S2… SN > of an evolving system at different instances of time < t1, t2… tN >. ...
... There are many systems that evolves over time due to changing, dynamic, and evolving environment. Such system are referred as evolving systems [1][2] [3] For example, the evolving systems that we encounter in our daily life are Wikipedia pages, software systems like Microsoft Windows, Facebook, etc. These systems evolve with time, and creates ample amount of information about their evolution. ...
... Evolving Systems are dynamical systems that are self-assembled from actively controlled subsystems. 3 The docking of the chaser with Envisat can be seen as such a system, except that the debris is passive. As mentioned before, docking is a complex procedure, consisting of using a robot arm, tentacles, and pushing rods. ...
... The reasoning behind this is that the rotational dynamics of the chaser have very little impact on the target debris because of the large (two orders of magnitude) inertia difference between the chaser and target. Following the method in Ref. 3, the stability of the linear system is assessed by looking at the location of the poles during the evolution from unconnected to connected. Figure 3 shows the location of the poles for an uncontrolled system. ...
... The adaptive key component controller uses a direct adaptation control law to restore stability to the Evolving System through a subset of the input and output ports on one key component of the Evolving System. Much of the detail of Evolving Systems appears in the chapter [8]. In this paper, we will deal with the situation where persistent disturbances can appear in some components and must be mitigated by the adaptive key component controller. ...
... We say a subsystem trait, such as stability, is inherited when the Evolved System retains the characteristic of the trait from the subsystem. Previous papers have examined the inheritance of stability and shown that stability is not a generally inherited trait [3]- [5] and [8]. Inheritance of almost strict passivity of subsystems is desirable in Evolving Systems that use an adaptive key component controller to restore stability. ...
This paper presents an introduction to Evolving Systems, which are autonomously controlled subsystems which self-assemble into a new Evolved System with a higher purpose. Evolving Systems of aerospace structures often require additional control when assembling to maintain stability during the entire evolution process. This is the concept of Adaptive Key Component Control which operates through one specific component to maintain stability during the evolution. In addition this control must overcome persistent disturbances that occur while the evolution is in progress. We present theoretical results for the successful operation of Adaptive Key Component control in the presence of such disturbances and an illustrative example.
Growing number of evolving systems creates demand for system evolution analytics with modern computational intelligence algorithms and tools. In this paper, we introduce new measures of stability and changeability for system evolution analysis over time. We proposed a Stable Network Evolution Rule Mining and a Changeability Metric for an evolving system. For this, we use two different characteristics of Network Evolution Rules (NERs). First, given a network of a system state S
$_i$
, we call an NER interesting in S
$_i$
if its support and confidence exceed given thresholds (minimum support and minimum confidence). Second, given a set of networks for a set of states (SS), we define the stability of an NER to be the percentage of states in SS in which the rule is interesting. We call an NER stable in SS if its stability exceeds a given threshold named as minimum stability (minStab). Based on this, we developed an intelligent tool, which is used for experiments on evolving systems. We applied our approach to a number of real-world systems including: software system, natural language system, retail market system, and IMDb system. It results Stable NERs and Changeability Metric value for each evolving system.
Era of computation intelligence leads to various kinds of systems that evolve. Usually, an evolving system contains evolving interconnected entities (or components) that make evolving networks for the system State Series
${\text{SS}}\,= \,\{S_{1},\,S_{2}\ldots \,S_{N}\}$
created over time, where S
$_i$
represents the ith system state. In this paper, we introduce an approach for mining Network Evolution Subgraphs such as Network Evolution Graphlets (NEGs) and Network Evolution Motifs (NEMs) from a set of evolving networks. We used graphlets information of a state to calculate System State Complexity (SSC). The System State Complexities (SSCs) represent time-varying complexities of multiple states. Additionally, we also used the NEGs information to calculate Evolving System Complexity (ESC) for a state series over time. We proposed an algorithm named System Network Complexity (SNC) for mining NEGs, SSCs, and ESC, which analyzes a pre-evolved state series of an evolving system. We prototyped the technique as a tool named SNC-Tool, which is applied to six real-world evolving systems collected from open-internet repositories of four different domains: software system, natural language system, retail market basket system, and IMDb movie genres system. This is demonstrated as experimentation reports containing retrieved—NEGs, NEMs, SSCs, and ESC—for each evolving system.
