Neural Speed Controller Trained Online by Means of Modified RPROP Algorithm

Pajchrowski, Tomasz; Zawirski, Krzysztof; Nowopolski, Krzysztof

doi:10.1109/tii.2014.2359620

Cited by 68 publications

(23 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The training stops when it reaches the maximum number of epochs or time, or when the best performance has been reached [28,29]. Some of the advantages of RPROP are: (i) the performance is better than other techniques used for adaptation [30] and (ii) it has fast convergence and low memory usage [31].…”

Section: Discussionmentioning

confidence: 99%

Development of Artificial Neural Network Models to Assess Beer Acceptability Based on Sensory Properties Using a Robotic Pourer: A Comparative Model Approach to Achieve an Artificial Intelligence System

Viejo¹,

Torrico²,

Dunshea³

et al. 2019

Beverages

View full text Add to dashboard Cite

Artificial neural networks (ANN) have become popular for optimization and prediction of parameters in foods, beverages, agriculture and medicine. For brewing, they have been explored to develop rapid methods to assess product quality and acceptability. Different beers (N = 17) were analyzed in triplicates using a robotic pourer, RoboBEER (University of Melbourne, Melbourne, Australia), to assess 15 color and foam-related parameters using computer-vision. Those samples were tested using sensory analysis for acceptability of carbonation mouthfeel, bitterness, flavor and overall liking with 30 consumers using a 9-point hedonic scale. ANN models were developed using 17 different training algorithms with 15 color and foam-related parameters as inputs and liking of four descriptors obtained from consumers as targets. Each algorithm was tested using five, seven and ten neurons and compared to select the best model based on correlation coefficients, slope and performance (mean squared error (MSE). Bayesian Regularization algorithm with seven neurons presented the best correlation (R = 0.98) and highest performance (MSE = 0.03) with no overfitting. These models may be used as a cost-effective method for fast-screening of beers during processing to assess acceptability more efficiently. The use of RoboBEER, computer-vision algorithms and ANN will allow the implementation of an artificial intelligence system for the brewing industry to assess its effectiveness.

show abstract

Section: Discussionmentioning

confidence: 99%

Development of Artificial Neural Network Models to Assess Beer Acceptability Based on Sensory Properties Using a Robotic Pourer: A Comparative Model Approach to Achieve an Artificial Intelligence System

Viejo¹,

Torrico²,

Dunshea³

et al. 2019

Beverages

View full text Add to dashboard Cite

show abstract

“…This algorithm is selected due to its rapidity, better convergence and high performance in pattern recognition problems. [24][25][26][27] In order to determine the corresponding coupling class for each two pixels, we must first of all initialize the artificial neural network structure used to know the number of hidden layers, the number of nodes per hidden layer and the activation function for each node. The artificial neural network adopted in our approach includes two hidden layers (excluding the input layer and the output layer).…”

Section: Step 3: the Artificial Neural Network Classifiermentioning

confidence: 99%

“…[19][20][21][22][23] The supervised learning with the multilayer feed-forward network architecture is chosen together with The resilient backpropagation known as RPROP algorithm. [24][25][26][27] The main aim of this algorithm is to identify the appropriate electromagnetic coupling operator between each two pixels among all pixels of the discretized surface and to optimize the calculation time for a large huge mesh surface. The formulation of the iterative method for nondescript geometric form structure is highlighted.…”

Section: Introductionmentioning

confidence: 99%

An efficient algorithm for electromagnetic scattering by a set of perfect conducting cylindrical objects using the artificial neural network

Ines¹,

Ammar²,

Aguili³

et al. 2018

Int J RF Microw Comput Aided Eng

View full text Add to dashboard Cite

In this article an efficient algorithm based on the wave concept iterative process is formulated and applied in order to investigate the electromagnetic scattering by a set of conducting arbitrarily shaped objects placed in the free space. In this approach, the cylindrical coordinates system is used as a modal base in order to develop the modal coefficients of the diffraction operator. This operator models the reaction of the environment expressing the electromagnetic coupling between each two pixels of the discretized surface. A study of electromagnetic coupling between two pixels positioned and oriented somehow in the free space is highlighted so as to determine the coupling operator for different positions and orientations of the emitted and received waves. Twelve coupling operators are developed. For a complex geometric shape structure, the determination of these characteristics involved the determination of interactions between each pixel and the others that constitute the discretized surface. This interaction involves the wave formulation already highlighted for each couple of pixels which implies a higher calculation time. In order to optimize the calculation time, the artificial neural networks are adopted with the feed‐forward architecture. The supervised learning has been chosen with the use of the resilient backpropagation algorithm.

show abstract

“…A good proof of concept is given by [11], in which a two-degree of freedom PID for the speed control of a PM synchronous motor has been emulated by a neural speed controller, trained online on a high-end ac drive. In that case, the implementation on the laboratory prototype was eased by the adoption of a resilient back-propagation training algorithm, well suited for the application to a linear system and anyway modified on purpose to achieve the astaticism of the controller.…”

Section: Introductionmentioning

confidence: 99%

Adaptive Maximum Torque per Ampere Control of Synchronous Reluctance Motors by Radial Basis Function Networks

Ortombina

Tinazzi

Zigliotto

2019

IEEE J. Emerg. Sel. Topics Power Electron.

View full text Add to dashboard Cite

As neodymium and other rare earth materials become a critical commodity, the exploitation of reluctance torque in synchronous motors is certainly an interesting option. Alternative motor topologies span from internal permanent magnet motors to pure reluctance machines. Anyway, any anisotropic structure suffers of some magnetic nonlinearity that call for more sophisticated models to get an efficient torque control. Neural network based algorithms are good candidates for modelling the current-to-flux linkages curves of synchronous reluctance motors, but so far their use was limited by the inherent complexity and the computational burden. This paper proposes the use of a special kind of neural networks, namely, the radial basis function networks, to get the magnetic model of any synchronous motor, including saturation and cross-coupling effects. Through experimental evidence, it will be shown that the structure is light enough to be implemented, trained and self-updated online on standard high-end ac drives. The model is used to track online the maximum torque-per-ampere working point of a synchronous reluctance motor drive.

show abstract

Neural Speed Controller Trained Online by Means of Modified RPROP Algorithm

Cited by 68 publications

References 23 publications

Development of Artificial Neural Network Models to Assess Beer Acceptability Based on Sensory Properties Using a Robotic Pourer: A Comparative Model Approach to Achieve an Artificial Intelligence System

Development of Artificial Neural Network Models to Assess Beer Acceptability Based on Sensory Properties Using a Robotic Pourer: A Comparative Model Approach to Achieve an Artificial Intelligence System

An efficient algorithm for electromagnetic scattering by a set of perfect conducting cylindrical objects using the artificial neural network

Adaptive Maximum Torque per Ampere Control of Synchronous Reluctance Motors by Radial Basis Function Networks

Contact Info

Product

Resources

About