A library of adaptive neural networks for control purposes

Campa, Giampiero; Fravolini, Mario Luca; Napolitano, M.

doi:10.1109/cacsd.2002.1036939

Cited by 36 publications

(11 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We can also take advantage of this scaling by setting the minimum and maximum to reduce the impact of potentially noisy input elements, for example, scaling K dp 2 [20.5, 0.5]. A unique formula (Campa et al 2002) is used in this implementation to determine the distance threshold for the novelty criterion, and is expressed as…”

Section: ) Step 5: Pruning Strategymentioning

confidence: 99%

“…Therefore, it becomes extremely computationally inefficient for high dimensional problems with a large training set. A number of algorithms have been developed to make the RBFN more efficient, and the implementation chosen for this research is adapted from (Campa et al 2002). Only neurons that are close to an input vector will have a noticeable effect on the output, so only those neurons within a given distance from an input training vector are updated in the learning process.…”

Section: B Rbf Neural Networkmentioning

confidence: 99%

“…The implementation used in this work was originally developed in Campa et al (2002). In addition to the core EMRAN-RBF functionality described above, the input and output vectors are scaled so that they fall within [21,1].…”

Section: ) Step 5: Pruning Strategymentioning

confidence: 99%

See 2 more Smart Citations

Simulating Radar Observations of Precipitation at Higher Frequencies from Lower-Frequency Polarimetric Measurements

Fritz

Chandrasekar

2012

Journal of Atmospheric and Oceanic Technology

View full text Add to dashboard Cite

The translation of radar data from one frequency to another based on electromagnetic absorption and scattering models has been used to convert S-band measurements to higher frequencies, such as those within the X or Ku bands, in order to create realistic simulations. As the target frequency of simulations for weather radar increase to X band and above, the size of large raindrops and hail either approach or exceed the radar wavelength, resulting in a radar cross section that is no longer in the linear Rayleigh region. The Mie solution to scattering models is then required. With the advent of dual-polarization systems, a more complete characterization of the effect of precipitation on radar is possible in order to improve model effectiveness. However, typical curve-or surface-fitting methods still have limitations as particle size increases. A more robust solution is presented here in the form of a neural network that incorporates the nonlinear relationship among various polarimetric observation variables and the radar wavelength and look angle. Thus, high-frequency observations of a convective storm containing hail can be simulated using polarimetric ground radar measurements. Adequate polarimetric data from hail storms at high frequencies do not exist, however, so network outputs for this case can only be compared to theoretical observations. An application of the simulation procedure to characterize the effect of precipitation on spaceborne synthetic aperture radar (SAR) using ground observations is presented.

show abstract

Section: ) Step 5: Pruning Strategymentioning

confidence: 99%

Section: B Rbf Neural Networkmentioning

confidence: 99%

See 1 more Smart Citation

Simulating Radar Observations of Precipitation at Higher Frequencies from Lower-Frequency Polarimetric Measurements

Fritz

Chandrasekar

2012

Journal of Atmospheric and Oceanic Technology

View full text Add to dashboard Cite

show abstract

“…In order to reduce the size of the network, a sequential learning algorithm for the RBF neural network called a minimal resource allocation network (M-RAN) is used (Campa et al 2002;Lu et al 1998). M-RAN fulfills the growth criterion and executes the pruning strategy as well as the adjustment of the network parameters with the network input data.…”

Section: Regulatormentioning

confidence: 99%

Modeling biological motor control for human locomotion with functional electrical stimulation

Zhang

Zhu

2006

Biol Cybern

View full text Add to dashboard Cite

This paper develops a novel control system for functional electrical stimulation (FES) locomotion, which aims to generate normal locomotion for paraplegics via FES. It explores the possibility of applying ideas from biology to engineering. The neural control mechanism of the biological motor system, the central pattern generator, has been adopted in the control system design. Some artificial control techniques such as neural network control, fuzzy logic, control and impedance control are incorporated to refine the control performance. Several types of sensory feedback are integrated to endow this control system with an adaptive ability. A musculoskeletal model with 7 segments and 18 muscles is constructed for the simulation study. Satisfactory simulation results are achieved under this FES control system, which indicates a promising technique for the potential application of FES locomotion in future.

show abstract

“…The original MBPM was adapted to real time control applications by Campa et al [10]. He provided a Simulink® block in MATLAB® which models the MPBM of Lewis.…”

Section: Adaptive Neural Network Compensatormentioning

confidence: 99%

A Novel Adaptive Neural Network Compensator as Applied to Position Control of a Pneumatic System

Dehghan

Taghizadeh

Surgenor

et al. 2011

ICA

View full text Add to dashboard Cite

Considerable research has been conducted on the control of pneumatic systems. However, nonlinearities continue to limit their performance. To compensate, advanced nonlinear and adaptive control strategies can be used. But the more successful advanced strategies typically need a mathematical model of the system to be controlled. The advantage of neural networks is that they do not require a model. This paper reports on a study whose objective is to explore the potential of a novel adaptive on-line neural network compensator (ANNC) for the position control of a pneumatic gantry robot. It was found that by combining ANNC with a traditional PID controller, tracking performance could be improved on the order of 45% to 70%. This level of performance was achieved after careful tuning of both the ANNC and PID components. The paper sets out to document the ANNC algorithm, the adopted tuning procedure, and presents experimental results that illustrate the adaptive nature of NN and confirms the performance achievable with ANNC. A major contribution is demonstration that tuning of ANNC requires no more effort than the tuning of PID.

show abstract

A library of adaptive neural networks for control purposes

Cited by 36 publications

References 6 publications

Simulating Radar Observations of Precipitation at Higher Frequencies from Lower-Frequency Polarimetric Measurements

Simulating Radar Observations of Precipitation at Higher Frequencies from Lower-Frequency Polarimetric Measurements

Modeling biological motor control for human locomotion with functional electrical stimulation

A Novel Adaptive Neural Network Compensator as Applied to Position Control of a Pneumatic System

Contact Info

Product

Resources

About