ABSTRACT The mill fans (MF) are centrifugal fans of the simplest type with flat radial blades ada... more ABSTRACT The mill fans (MF) are centrifugal fans of the simplest type with flat radial blades adapted for simultaneous operation both like fans and also like mills. The key variable that could be used for diagnostic purposes is vibration amplitude of MF corpse. However its mode values include a great deal of randomness. Therefore the application of deterministic dependencies with correcting coefficients is non-effective for MF predictive modeling. Standard statistical and probabilistic (Bayesian) approaches are also inapplicable to estimate MF vibration state due to non-stationarity, non-ergodicity and the significant noise level of the monitored vibrations. Adequate for the case methods of computational intelligence (fuzzy logic, neural networks and more general AI techniques - the precedents' method or machine learning (ML)) must be used. The present paper describes promising initial results on applying the Case Based Reasoning (CBR) approach for intelligent diagnostic of the mill fan working capacity using its vibration state.
A hybrid control system with Case Based Reasoning (CBR) procedure is considered. It is intended f... more A hybrid control system with Case Based Reasoning (CBR) procedure is considered. It is intended for complex industrial plants in abnormal and fault situations where the conventional control systems cannot work autonomously. CBR systems allow capture and use a specific experience by Case Base (CB) creation, similarity assessment, case adaptation and learning new cases. Some new results are incorporated into the traditional CBR cycle, relevant to control tasks - uncertainty, presence of multivariant solutions from different operators, rational distribution of knowledge among autonomous agents, ontologies and CBR-system. Simulation results are represented for an industrial plant with a non-square matrix.
In the present paper we focus on online monitoring system for predictive maintenance based on sen... more In the present paper we focus on online monitoring system for predictive maintenance based on sensor automated inputs. Our subject was a device from Maritsa East 2 power plant-a mill fan. The main sensor information we have access to is based on the vibration of the ...
ABSTRACT In the present paper we focus on online monitoring system for predictive maintenance bas... more ABSTRACT In the present paper we focus on online monitoring system for predictive maintenance based on sensor automated inputs. Our subject was a device from Maritsa East 2 power plant - a mill fan. The main sensor information we have access to is based on the vibration of the nearest to the mill rotor bearing block. Our aim was to create a (nonlinear) model able to predict on time possible changes in vibrations tendencies that can be early signal for system work deterioration. For that purpose recently developed kind of Recurrent Neural Networks named Echo state networks were applied. The preliminary investigations showed their good approximation ability for our purpose. Direction of future work will be increasing of predications time horizon.
The paper focuses on identification issues of the advanced controller ASPECT* that is implemented... more The paper focuses on identification issues of the advanced controller ASPECT* that is implemented on a simple PLC platform with an extra mathematical coprocessor and is intended for the advanced control of complex plants. The model of the controlled plant is obtained by means of experimental modeling using an online learning procedure that combines model identification with pre- and post-identification steps that provide reliable operation. It is shown that acceptable performance of the system is obtained despite difficult conditions that may arise during operation.
The presented self-tuning nonlinear controller ASPECT (advanced control algorithms for programmab... more The presented self-tuning nonlinear controller ASPECT (advanced control algorithms for programmable logic controllers) is intended for control of a practically important class of nonlinear processes that may be presented by a set of low-order local linear models. The model is obtained by means of experimental modelling using an online learning procedure that combines model identification with pre- and post-identification steps that provide reliable operation. There is a choice of several control algorithms suitable for different processes that may be used for control and whose parameters are automatically tuned from the model. The controller monitors the resulting control performance and may react to detected irregularities. The choice of the controller structure and algorithms allows implementation on available industrial hardware, such as programmable logic controllers (PLCs) or open controllers, and simplifies configuration of parameters.
... algorithm. The actual model (the one that resides in MIA) is favoured if the confidence index... more ... algorithm. The actual model (the one that resides in MIA) is favoured if the confidence indexes of the old and the new model are similar. ... 1993. L. Ljung, Sysfem Idenfflcarion, Englcwoad Cliffs NI: Prentice Hall, 1987. M. Sugeno, M. Nishida. ...
An intelligent multicascade control system is proposed using hybrid multi-agent and ontology fram... more An intelligent multicascade control system is proposed using hybrid multi-agent and ontology framework. Fuzzy logic based extension of conventional ontology is implemented to meet uncertainties in the plant, its environment and sensor information. Ant colony optimization is applied to realize trade-off between requirements and control resources as well as for significant reduction of the communication rate among the intelligente agents. To react on unexpected changes in operational conditions certain adaptation functionality of the fuzzy ontology is foreseen. Some simulation results are presented.
This paper presents an advanced control system named ASPECT for control of nonlinear or slowly ti... more This paper presents an advanced control system named ASPECT for control of nonlinear or slowly time varying systems. The described control system was developed and implemented in programmable logic controller to meet industrial demands. The case study, which is control of gas-liquid separation unit, is intended to illustrate the operation and benefits of the controller.
The paper contains a short discussion and example on the possibility for using the apparatus of g... more The paper contains a short discussion and example on the possibility for using the apparatus of generalized nets for modelling of intelligent systems.
The investigation deals with the employment of a Smith predictor as an improved working mechanism... more The investigation deals with the employment of a Smith predictor as an improved working mechanism for overcoming the negative effects arising in control systems with a long time delay. As a model-based algorithm it imposes a stringent restriction over systems behavior by parametric uncertainty. The adaptation algorithm proposed in this paper uses data, derived by analysis of control system performance. It determines a quantitative correction for uncertainty compensation of model parameters and the allowable rate of increasing the gain coefficient of the conventional PI-controller. Special attention is paid and a significant experimental volume of the investigation is carried out for estimation and excluding the effects of the available noise in the system with a specific intensity and spectral density on method accuracy.
A new approach for intelligent control is proposed for complex uncertain plants using synergism b... more A new approach for intelligent control is proposed for complex uncertain plants using synergism between multi-agent and ontology based frameworks. A multi stage procedure is developed for situation recognition, strategy selection and control algorithm parameterization following coordinated objective function. Fuzzy logic based extension of conventional ontology is implemented to meet uncertainties in the plant, its environment and sensor information. Ant colony optimization is applied to realize trade-off between requirements and control resources as well as for significant reduction of the communication rate among the intelligente agents. To react on unexpected changes in operational conditions certain adaptation functionality of the fuzzy ontology is foreseen. A multi-dimensional cascade system is considered and some simulation results are presented for variety of strategies implemented.
An adaptive control system for multivariable time varying plants directly influenced by a nonstat... more An adaptive control system for multivariable time varying plants directly influenced by a nonstationary measurable (estimated) scheduling variable is considered. The plant under control has different behavior depending on the speed of change of the scheduling variable-quasistationary, nonstationary, or stochastic. In order to cover all possible technological modes a parallel structure is proposed. It consists of concurrent feedback and feedforward supervisors, an operating condition identifier, and a scheduling variable estimator. The combined action of these parallel tasks increases the functional robustness of the controlled system for every identified operating range due to the bumpless transfer between the different controllers. A procedure for control law calculation in boundary conditions of adjacent operating modes is given
ABSTRACT An adaptive strategy for knowledge-based control of sophisticated plants is considered. ... more ABSTRACT An adaptive strategy for knowledge-based control of sophisticated plants is considered. The domain knowledge is structured using a set of ontologies – factual ontology and functional ontologies (for tasks, problem solving methods, way to realize the methods, activities – adaptive control, monitoring). The active part of the control system is accomplished using a multiagent system. An Ant Colony Optimization is proposed for fulfilling different optimization procedures. A brief description is given for the structure, the functions and the processes of the developed multiface control system. The software realization is OWL-based for the ontologies and JADE-based for the intelligent agents. An example for adaptive control of a non-square (SITO) plant is considered.
This paper studies the impact of neural network accuracy on the feedback/feed-forward (FB/FF) pH ... more This paper studies the impact of neural network accuracy on the feedback/feed-forward (FB/FF) pH neutralization control system performance. Five neural networks (NNs) are incorporated in a FB/FF control structure in order to perform nonlinear compensation of the non-measurable input disturbances. Two significant problems connected with the application of the FF scheme are discussed - the appearance of oscillations and the offset of the system output from its desired value. A procedure for adding of auxiliary points in the training data set when the NNs accuracy is not sufficient is derived. Simulation results are presented.
ABSTRACT The mill fans (MF) are centrifugal fans of the simplest type with flat radial blades ada... more ABSTRACT The mill fans (MF) are centrifugal fans of the simplest type with flat radial blades adapted for simultaneous operation both like fans and also like mills. The key variable that could be used for diagnostic purposes is vibration amplitude of MF corpse. However its mode values include a great deal of randomness. Therefore the application of deterministic dependencies with correcting coefficients is non-effective for MF predictive modeling. Standard statistical and probabilistic (Bayesian) approaches are also inapplicable to estimate MF vibration state due to non-stationarity, non-ergodicity and the significant noise level of the monitored vibrations. Adequate for the case methods of computational intelligence (fuzzy logic, neural networks and more general AI techniques - the precedents' method or machine learning (ML)) must be used. The present paper describes promising initial results on applying the Case Based Reasoning (CBR) approach for intelligent diagnostic of the mill fan working capacity using its vibration state.
A hybrid control system with Case Based Reasoning (CBR) procedure is considered. It is intended f... more A hybrid control system with Case Based Reasoning (CBR) procedure is considered. It is intended for complex industrial plants in abnormal and fault situations where the conventional control systems cannot work autonomously. CBR systems allow capture and use a specific experience by Case Base (CB) creation, similarity assessment, case adaptation and learning new cases. Some new results are incorporated into the traditional CBR cycle, relevant to control tasks - uncertainty, presence of multivariant solutions from different operators, rational distribution of knowledge among autonomous agents, ontologies and CBR-system. Simulation results are represented for an industrial plant with a non-square matrix.
In the present paper we focus on online monitoring system for predictive maintenance based on sen... more In the present paper we focus on online monitoring system for predictive maintenance based on sensor automated inputs. Our subject was a device from Maritsa East 2 power plant-a mill fan. The main sensor information we have access to is based on the vibration of the ...
ABSTRACT In the present paper we focus on online monitoring system for predictive maintenance bas... more ABSTRACT In the present paper we focus on online monitoring system for predictive maintenance based on sensor automated inputs. Our subject was a device from Maritsa East 2 power plant - a mill fan. The main sensor information we have access to is based on the vibration of the nearest to the mill rotor bearing block. Our aim was to create a (nonlinear) model able to predict on time possible changes in vibrations tendencies that can be early signal for system work deterioration. For that purpose recently developed kind of Recurrent Neural Networks named Echo state networks were applied. The preliminary investigations showed their good approximation ability for our purpose. Direction of future work will be increasing of predications time horizon.
The paper focuses on identification issues of the advanced controller ASPECT* that is implemented... more The paper focuses on identification issues of the advanced controller ASPECT* that is implemented on a simple PLC platform with an extra mathematical coprocessor and is intended for the advanced control of complex plants. The model of the controlled plant is obtained by means of experimental modeling using an online learning procedure that combines model identification with pre- and post-identification steps that provide reliable operation. It is shown that acceptable performance of the system is obtained despite difficult conditions that may arise during operation.
The presented self-tuning nonlinear controller ASPECT (advanced control algorithms for programmab... more The presented self-tuning nonlinear controller ASPECT (advanced control algorithms for programmable logic controllers) is intended for control of a practically important class of nonlinear processes that may be presented by a set of low-order local linear models. The model is obtained by means of experimental modelling using an online learning procedure that combines model identification with pre- and post-identification steps that provide reliable operation. There is a choice of several control algorithms suitable for different processes that may be used for control and whose parameters are automatically tuned from the model. The controller monitors the resulting control performance and may react to detected irregularities. The choice of the controller structure and algorithms allows implementation on available industrial hardware, such as programmable logic controllers (PLCs) or open controllers, and simplifies configuration of parameters.
... algorithm. The actual model (the one that resides in MIA) is favoured if the confidence index... more ... algorithm. The actual model (the one that resides in MIA) is favoured if the confidence indexes of the old and the new model are similar. ... 1993. L. Ljung, Sysfem Idenfflcarion, Englcwoad Cliffs NI: Prentice Hall, 1987. M. Sugeno, M. Nishida. ...
An intelligent multicascade control system is proposed using hybrid multi-agent and ontology fram... more An intelligent multicascade control system is proposed using hybrid multi-agent and ontology framework. Fuzzy logic based extension of conventional ontology is implemented to meet uncertainties in the plant, its environment and sensor information. Ant colony optimization is applied to realize trade-off between requirements and control resources as well as for significant reduction of the communication rate among the intelligente agents. To react on unexpected changes in operational conditions certain adaptation functionality of the fuzzy ontology is foreseen. Some simulation results are presented.
This paper presents an advanced control system named ASPECT for control of nonlinear or slowly ti... more This paper presents an advanced control system named ASPECT for control of nonlinear or slowly time varying systems. The described control system was developed and implemented in programmable logic controller to meet industrial demands. The case study, which is control of gas-liquid separation unit, is intended to illustrate the operation and benefits of the controller.
The paper contains a short discussion and example on the possibility for using the apparatus of g... more The paper contains a short discussion and example on the possibility for using the apparatus of generalized nets for modelling of intelligent systems.
The investigation deals with the employment of a Smith predictor as an improved working mechanism... more The investigation deals with the employment of a Smith predictor as an improved working mechanism for overcoming the negative effects arising in control systems with a long time delay. As a model-based algorithm it imposes a stringent restriction over systems behavior by parametric uncertainty. The adaptation algorithm proposed in this paper uses data, derived by analysis of control system performance. It determines a quantitative correction for uncertainty compensation of model parameters and the allowable rate of increasing the gain coefficient of the conventional PI-controller. Special attention is paid and a significant experimental volume of the investigation is carried out for estimation and excluding the effects of the available noise in the system with a specific intensity and spectral density on method accuracy.
A new approach for intelligent control is proposed for complex uncertain plants using synergism b... more A new approach for intelligent control is proposed for complex uncertain plants using synergism between multi-agent and ontology based frameworks. A multi stage procedure is developed for situation recognition, strategy selection and control algorithm parameterization following coordinated objective function. Fuzzy logic based extension of conventional ontology is implemented to meet uncertainties in the plant, its environment and sensor information. Ant colony optimization is applied to realize trade-off between requirements and control resources as well as for significant reduction of the communication rate among the intelligente agents. To react on unexpected changes in operational conditions certain adaptation functionality of the fuzzy ontology is foreseen. A multi-dimensional cascade system is considered and some simulation results are presented for variety of strategies implemented.
An adaptive control system for multivariable time varying plants directly influenced by a nonstat... more An adaptive control system for multivariable time varying plants directly influenced by a nonstationary measurable (estimated) scheduling variable is considered. The plant under control has different behavior depending on the speed of change of the scheduling variable-quasistationary, nonstationary, or stochastic. In order to cover all possible technological modes a parallel structure is proposed. It consists of concurrent feedback and feedforward supervisors, an operating condition identifier, and a scheduling variable estimator. The combined action of these parallel tasks increases the functional robustness of the controlled system for every identified operating range due to the bumpless transfer between the different controllers. A procedure for control law calculation in boundary conditions of adjacent operating modes is given
ABSTRACT An adaptive strategy for knowledge-based control of sophisticated plants is considered. ... more ABSTRACT An adaptive strategy for knowledge-based control of sophisticated plants is considered. The domain knowledge is structured using a set of ontologies – factual ontology and functional ontologies (for tasks, problem solving methods, way to realize the methods, activities – adaptive control, monitoring). The active part of the control system is accomplished using a multiagent system. An Ant Colony Optimization is proposed for fulfilling different optimization procedures. A brief description is given for the structure, the functions and the processes of the developed multiface control system. The software realization is OWL-based for the ontologies and JADE-based for the intelligent agents. An example for adaptive control of a non-square (SITO) plant is considered.
This paper studies the impact of neural network accuracy on the feedback/feed-forward (FB/FF) pH ... more This paper studies the impact of neural network accuracy on the feedback/feed-forward (FB/FF) pH neutralization control system performance. Five neural networks (NNs) are incorporated in a FB/FF control structure in order to perform nonlinear compensation of the non-measurable input disturbances. Two significant problems connected with the application of the FF scheme are discussed - the appearance of oscillations and the offset of the system output from its desired value. A procedure for adding of auxiliary points in the training data set when the NNs accuracy is not sufficient is derived. Simulation results are presented.
Uploads