Feedforward Procedure Research Articles

An Artificial Neural Network-Based Model for Effective Software Development Effort Estimation

In project management, effective cost estimation is one of the most crucial activities to efficiently manage resources by predicting the required cost to fulfill a given task. However, finding the best estimation results in software development is challenging. Thus, accurate estimation of software development efforts is always a concern for many companies. In this paper, we proposed a novel software development effort estimation model based both on constructive cost model II (COCOMO II) and the artificial neural network (ANN). An artificial neural network enhances the COCOMO model, and the value of the baseline effort constant A is calibrated to use it in the proposed model equation. Three state-of-the-art publicly available datasets are used for experiments. The backpropagation feedforward procedure used a training set by iteratively processing and training a neural network. The proposed model is tested on the test set. The estimated effort is compared with the actual effort value. Experimental results show that the effort estimated by the proposed model is very close to the real effort, thus enhanced the reliability and improving the software effort estimation accuracy.

Common vs. independent limb control in sequential vertical aiming: The cost of potential errors during extensions and reversals

The following study explored movement kinematics in two-component aiming contexts that were intended to modulate the potential cost of overshoot or undershoot errors in up and down directions by having participants perform a second extension movement (Experiment 1) or a reversal movement (Experiment 2). For both experiments, the initial movement toward a downward target took longer, and had lower peak acceleration and peak velocity than upward movements. These movement characteristics may reflect a feedback-based control strategy designed to prevent energy-consuming limb modifications against gravitational forces. The between-component correlations of displacement at kinematic landmarks (i.e., trial-by-trial correlation between the first and second components) increased as both components unfolded. However, the between-component correlations of extensions were primarily negative, while reversals were positive. Thus, movement extensions appear to be influenced by the use of continuous on-line sensory feedback to update limb position at the second component based on the position attained in the first component. In contrast, reversals seem to be driven by pre-planned feedforward procedures where the position of the first component is directly replicated in the second component. Finally, the between-component correlations for the magnitude of kinematic landmarks showed that aiming up generated stronger positive correlations during extensions, and weaker positive correlations toward the end of the first component during reversals. These latter results suggest the cost of potential errors associated with the upcoming second component directly influence the inter-dependence between components. Therefore, the cost of potential errors is not only pertinent to one-component discrete contexts, but also two-component sequence aims. Together, these findings point to an optimized movement strategy designed to minimize the cost of errors, which is specific to the two-component context.

Surprise, Attraction, and Propagation

The management of off-normal operational events, such as an aircraft in-flight loss of control, present a formidable challenge to the stakeholders of commercial aviation. This paper posits that present solutions to off-normal events are grounded in linear models of accident causation that are insufficient in today’s complex socio-technical domain—where change propagates through systems to produce instability and system state shifts. These prior methods of system safety are predominately predicated upon the notion that system failure can be mitigated to acceptable levels of risk by using defenses, constraints, and controls in design and practice; such perspectives also influence post-accident investigation and remediation. This paper provides a new theory of aircraft accident causation, Catastrophic Information Entropy Theory (CIET), which offers that an increase in information entropy during an off-normal event, produces emergent changes in system state that prevent a return to normal flight. Therefore, in the face of run-time accident factor propagation, surprise, and attractors, the management of off-normal events, utilizing rigid feed forward procedures may make recovery inaccessible. CIET suggest that existing safety strategies need to be expanded to include the recognition of the time-critical role of the pilot, who must attempt to understand and control a complex system which is behaving in a novel and emergent way. CIET offers a theoretical perspective from which aviation stakeholders can design tools for actionable solutions for pilots during real time emergent events and tools for post-event analytic approaches.

A feedback-based [formula omitted]-norm limiter for periodic signals

Limiting the rms value ( L 2 norm) of periodic signals can be a necessity in some controllers for power electronic devices. The prevailing existing solution is simply to compute, off-line, the L 2 norm and then to multiply the signal by the factor that ensures the desired bound is achieved. This feedforward procedure could yield unsatisfactory results during the transients of the system. In this paper, a feedback-based algorithm that carries out the computations on-line is proposed. Besides the on-line computation of the L 2 norm, the second component of the algorithm is a classical proportional–integral loop with an anti-windup mechanism. Although the overall algorithm is described by complicated nonlinear, non-autonomous, delay-differential equations, a detailed stability analysis is carried out, proving its robust performance. Simulation results, that confirm the theoretical results, are presented.

Simulation of floor response spectra in shake table experiments

AbstractAmong several different experimental techniques, used to test the response of structures and to verify their seismic performance, the shake table testing allows to reproduce the conditions of true effects of earthquake ground motions in order to challenge complex model structures and systems. However, the reproduction of dynamic signals, due to the dynamics of the shake table and of the specimen, is usually imperfect even though closed‐loop control in a shake table system is used to reduce these errors and obtain the best fidelity reproduction. Furthermore, because of the dynamic amplifications in the specimen, the signal recorded at desired locations could be completely different from the expected effect of shake table motion. This paper focuses on the development of practical shake table simulations using additional ‘open loop’ feedforward compensation in form of inverse transfer functions (i.e. the ratio of the output structural response to an input base motion in the frequency domain) in order to obtain an acceptable reproduction of desired acceleration histories at specific locations in the specimen. As the first step, a well‐known global feedforward procedure is reformulated for the compensation of the table motion distortions due to the servo‐hydraulic system. Subsequently, the same concept is extended to the table‐structure system to adjust the shake table input in order to achieve a desired response spectrum at any floor of the specimen. Implementations show how such a method can be used in any experimental facility. Copyright © 2010 John Wiley & Sons, Ltd.

Concatenated beam splitters, optical feed-forward, and the nonlinear sign gate

We consider a nonlinear sign gate implemented using a sequence of two beam splitters, and consider the use of further sequences of beam splitters to implement feed-forward so as to correct an error resulting from the first beam splitter. We obtain similar results to Scheel et al. [Phys. Rev. A 73, 034301 (2006)], in that we also find that our feed-forward procedure is only able to produce a very minor improvement in the success probability of the original gate.

Comparison of Different Multivariable Control Design Methods Applied on Half Car Test Setup

The multivariable tracking accuracy on an automotive vibration test rig can be improved by extending the current industrial off-line iterative feedforward procedure with a real time feedback controller. This article compares three MIMO feedback control design procedures with respect to design complexity and obtained performance: (1) DK-iteration, the most common μ-synthesis method to design robust MIMO-controllers, (2) a control design based on static decoupling, which is a combination of a static input – output transformation to decouple the system, and independent SISO-controllers designed for the diagonal elements of the transformed system and (3) an inverse-based control design, which is also based on a decoupling transformation, but the decoupling is performed by a dynamic precompensator. The performance of these controllers is compared experimentally on a half car test setup. The tracking performance of the controller designs based on static or dynamic decoupling is comparable with the performance of the controller based on μ-synthesis. The decoupling-based controller designs, using standard SISO H∞ techniques, are however much simpler and more straightforward.

Multivariable feedback design to improve tracking performance on tractor vibration test rig

The multivariable tracking accuracy on an automotive vibration test rig can be improved by extending the current industrial off-line iterative feedforward procedure with a real-time multiple-input-multiple-output (MIMO) feedback controller. This paper shows that for the considered 4 × 4 application, the two most common approaches, namely full MIMO identification and μ -synthesis on one hand and decentralised control on the other hand fail to design a robust MIMO controller with acceptable performance. Therefore, a control design based on static decoupling is applied. This is a combination of (1) a transformation of the inputs and the outputs to decouple the system, and (2) independent single-input-single-output (SISO) controllers designed for the diagonal elements of the transformed system, using standard SISO identification and H ∞ -techniques. To optimise the closed-loop performance, a new decoupling algorithm is developed. The new presented algorithm outperforms the existing decoupling algorithms, yielding satisfactory multivariable closed-loop performance.

Experimental validation of different MIMO-feedback controller design methods

The multivariable tracking performance of an automotive vibration test rig can be improved by extending the current industrial off-line iterative feedforward procedure with a real-time feedback controller. This paper tests and validates three different MIMO-control schemes on an industrial test rig. Firstly decentralized (or diagonal) control is used, which is a combination of independently designed SISO-controllers based on the diagonal elements of the MIMO-system model. Secondly full MIMO H ∞ -control, based on a MIMO-model is tested. Finally inverse-based control is applied to the test rig, this is a combination of a pre-compensator dealing with the interaction in the system and a decentralized controller, designed for the compensated system. The performance of these controllers is compared experimentally, and the best controller is combined with the off-line feedforward control, showing a significant reduction of the tracking error.

1108 フィードフォワード/フィードバック機能を併合した行動探索による学習(GS-10.16 制御手法)

Making the use of the feedforward and feedback controls, an action-based learning algorithm is proposed. The modern control theory comes up with a new interpretation of an artificial intelligence. Our learning algorithm contains the feedforward procedure where we can iteratively improve our behavior better than before, and feedback procedure where we can reject any kind of disturbances. Based on Genetic Programming some simulation results are reported in order to illustrate the usefulness of our learning algorithm.

1212 フィードフォワード/フィードバック機能を併合した行動探索型学習アルゴリズム

In this paper, we present an Action-based Learning algorithm making the use of the feedforward and feedback controls. The modern control theory gives us a new interpretation of an artificial intelligence. Our learning algorithm contains the feedforward procedure where we can iteratively improve our behavior better than before, and feedback procedure where we can reject any kind of disturbances that might be met. Some experiment results on an inverted pendulum are reported in order to illustrate the usefulness of our learning algorithm.

Building thermal analyses: What the industry needs: The Architect's perspective

Drawing on experience as a practitioner, the author argues that a static need definition is necessarily inferior to a dynamic need definition attained by a feed-back and feed-forward procedure built on general systems theory. The designer's approach, working from the whole towards the detail, is contrasted with the way analytical models are typically structured. Recent advances, problems and future work are identified.

Industrial adaptive controllers based on frequency response techniques

This paper describes several adaptive techniques based on frequency response analysis, which are used in industrial controllers. Automatic tuning is used to obtain PID controller parameters, to build up gain schedules, and to initialize adaptive algorithms. A simple adaptive feedback algorithm which is based on the idea of tracking a point on the Nyquist curve is presented as well as an adaptive feedforward procedure. The adaptive techniques have been implemented in various industrial products. Industrial experiences from using the controllers are summarized in the paper.

A Frequency Domain Approach to Adaptive Control

This paper compares adaptive controllers based on parametric models and adaptive controllers based on a frequency response models. Several uses of the adaptive technique based on frequency response models are presented. Autotuning is used to obtain PID controller parameters, to build up gain schedules, and to initialize adaptive algorithms. Simple adaptive feedback and feedforward procedures are presented. Experiences from the use of these algorithms in industrial products are also given in the paper.