Browsing by Author "Daniel, H. A."
Now showing 1 - 10 of 10
Results Per Page
Sort Options
- Adaptive generalized predictive control algorithm implemented over a DSP networkPublication . Daniel, H. A.; Ruano, AntonioIn this paper a parallel implementation of an Adaprtive Generalized Predictive Control (AGPC) algorithm is presented. Since the AGPC algorithm needs to be fed with knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and a GPC predictor is discussed here.
- Adaptive generalized predictive control algorithm implemented over an heterogeneous parallel architecturePublication . Daniel, H. A.; Ruano, AntonioIn this paper a parallel implementation of an Adaprtive Generalized Predictive Control (AGPC) algorithm is presented. Since the AGPC algorithm needs to be fed with knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and a GPC predictor is discussed here.
- Automatic parallelization of an adaptive generalized predictive control algorithm using MAPS 1.0 environmentPublication . Daniel, H. A.; Ruano, AntonioParallelization of real time control algorithms is a problem that the control engineer must consider to meet tighter specifications in terms of plant sampling time. However, due to the lack of appropriate tools, the development time and prototyping of an efficient parallel algorithm is much higher than its sequential equivalent. It is also common that such implementation requires knowledge of system programming. Assuming that a control engineer must concentrate in implementing an efficient control strategy, rather than work around system dependent issues, the Matrix Algorithms Automatic Parallelization System –MAPS- programming environment was developed. If the control algorithm can be represented in a matricial form, as is the case of many, this programming environment puts in the hands of the control engineer the power of parallel processing at the cost of the sequential programming model. In this paper the automatic parallelization of an Adaptive Generalized Predictive Control - AGPC – algorithm is employed as an example, using the MAPS environment. It will be shown that, as long as MAPS supports the target hardware, porting an algorithm is just a matter of describing the parallel network topology in a simple block diagram. Finally, the performance of this parallel AGPC algorithm, mapped over some network topologies will be presented and discussed.
- Automatic parallelization of matricial algorithmsPublication . Daniel, H. A.; Ruano, AntonioThe introduction of parallel processing architectures allowed the real time impelemtation of more sophisticated control algorithms with tighter specifications in terms of sampling time. However, to take advantage of the processing power of these architectures the control engeneer, due to the lack of appropriate tools, must spend a considerable amount of time in the parallelizaton of the control algorithm.
- Implementation of an daptive generalized predictive control algorithm over an heterogeneous parallel architecturePublication . Daniel, H. A.; Ruano, Antonio; Fleming, P. J.In this paper a parallel implementation of an Adaprtive Generalized Predictive Control (AGPC) algorithm is presented. Since the AGPC algorithm needs to be fed with knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and a GPC predictor is discussed here.
- Paralelização automática de algoritmos matriciaisPublication . Daniel, H. A.; Ruano, A. E.A introdução de arquitecturas de processamento paralelo permitiu que o tempo de processamento de um algoritmo possa ser reduzido dividindo o esforço computacional por mais do que um processador. Todavia para se tirar partido destas arquitecturas, devido à falta de ferramentas apropriadas, o projectista despende uma considerável quantidade de tempo na paralelização do algoritmo sequencial. Outro problema normalmente encontrado, no modelo de programação paralelo, relaciona-se com o facto de a paralelização destes algoritmos ser altamente dependente da arquitectura objecto. Assim, a portabilidade e adaptabilidade destas aplicações são tarefas consumidoras de tempo de desenvolvimento. Pelas razões apontadas, o tempo de implementação de um algoritmo paralelo é muito superior ao tempo de implementação sequencial do mesmo algoritmo. Tais condições constituíram a motivação para o trabalho desenvolvido nesta tese, o qual consiste num sistema de paralelização automático de algoritmos matriciais. Este sistema é visto como um conjunto de níveis de abstracção que gradualmente se afastam do modelo de processamento paralelo e se aproximam do modelo sequencial. No nível mais elevado basta uma descrição do algoritmo, numa linguagem sequencial, e um diagrama de blocos da rede de processadores, para que o sistema, automaticamente, gere o código paralelo para a rede objecto. Esta implementação, baseada em sucessivos níveis de abstracção, permite um elevado grau de portabilidade e flexibilidade do sistema, de modo que a introdução de novos processadores, com diferentes especificações de computação e comunicação, ou de operações matriciais não incluídas na biblioteca matricial que acompanha o sistema, seja uma tarefa facilitada. Finalmente é estudada a paralelização automática de dois algoritmos, de modo a demonstrar o modelo de programação proposto bem como o desempenho dos algoritmos paralelos automaticamente gerados.
- Parallel implementation of an adaptive generalized predictive control algorithmPublication . Daniel, H. A.; Ruano, AntonioThe Adaptive Generalized Predictive Control (GPC) algorithm can be speeded up using parallel processing. Since the GPC algorithm needs to be fed with knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and a GPC predictor is discussed here.
- Parallel implementation of an adaptive generalized predictive control algorithmPublication . Ruano, Antonio; Daniel, H. A.The Adaptive Generalized Predictive Control (AGPC) algorithm can be speeded up using parallel processing. Since the AGPC algorithm needs to be fed with the knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and a GPC predictor is discussed here.
- Performance comparison of parallel architectures for real-time controlPublication . Daniel, H. A.; Ruano, AntonioIn this paper the performance of several parallel architectures for the implementation of matrix-intensive control algorithms is compared. To investigate their performance, a parallel version of an Adaptive Generalized Predictive Control algorithm (AGPC) is mapped over these architectures. Since this algorithm needs to be fed with the knowledge of the plant transfer function, the parallelization of a standard Recursive Least Squares (RLS) estimator and of a GPC predictor is discussed here. The former step operates over a small set of data, while the latter uses a larger set of data, therefore making this algorithm a valid benchmark to measure the performance of such architectures for these two different situations. Two homogeneous architectures built of T805 transputers and TMS320C40 DSPs are investigated. Also two different networks built of T805 and TMS320C40 are used to measure the performance of heterogeneous architectures. Execution times and efficiency results of the RLS and GPC steps presented illustrate that the TMS320C40 network is the fastest network tested. However, for low complexity algorithms, depending on the sampling time, even the slow transputer homogeneous architecture may still have the required performance. In the general case, there is no benefit in using both heterogeneous architectures tested, since they cannot outperform the TMS320C40 homogeneous architectures. (C) 1999 Elsevier Science B.V. All rights reserved.
- Speeding up a learning algorithm for multilayer perceptrons using the MAPS EnvironmentPublication . Daniel, H. A.; Ruano, AntonioArtificial neural networks, as non-linear adaptive elements, have been proposed for applications in adaptive control. Their ability to accurately approximate large classes of non-linear functions made them also a valuable tool for non-linear systems identification. However, in some cases, the parameter estimation phase may take considerable amount of time, and this is crucial in real-time applications. One way of speeding up these learning algorithms consists in executing them over a multiprocessor system. In this paper an implementation over MAPS integrated development environment, which automatically generates a parallel application from a sequential description of a learning algorithm for multilayer perceptrons is presented.