Mark Howell - Academia.edu (original) (raw)
Papers by Mark Howell
IEEE Access, 2014
Regenerative braking is one of the most promising and environmentally friendly technologies used ... more Regenerative braking is one of the most promising and environmentally friendly technologies used in electric and hybrid electric vehicles to improve energy efficiency and vehicle stability. This paper presents a systematic data-driven process for detecting and diagnosing faults in the regenerative braking system of hybrid electric vehicles. The diagnostic process involves signal processing and statistical techniques for feature extraction, data reduction for implementation in memory-constrained electronic control units, and variety of fault classification methodologies to isolate faults in the regenerative braking system. The results demonstrate that highly accurate fault diagnosis is possible with the classification methodologies. The process can be employed for fault analysis in a wide variety of systems, ranging from automobiles to buildings to aerospace systems.
La presente invention concerne un systeme de commande destine a avertir un vehicule suiveur d'... more La presente invention concerne un systeme de commande destine a avertir un vehicule suiveur d'un risque de collision avec un vehicle qui le precede. Le vehicule de tete comprend un systeme de detection qui detecte la presence et la vitesse du vehicule suiveur. Le systeme de commande determine une distance souhaitee entre le vehicule de tete et le vehicule suiveur selon la vitesse du premier. Le systeme de commande soustrait la distance souhaitee de celle reelle pour generer un signal d'erreur de distance, et compare ce signal a une valeur limite. Si le signal d'erreur est superieur a la limite, le systeme de commande peut effectuer une ou plusieurs operations, telles que le declenchement de feux de detresse pour avertir le vehicule suiveur, ou executer d'autres actions en cas de collision imminente, telles que la pretension de ceintures de securite et la fermeture de vitres du vehicule suiveur.
The implementation of the Hough transform using neural networks is described. This highly paralle... more The implementation of the Hough transform using neural networks is described. This highly parallel implementation is fast and suitable for real-time applications. A 'homing-in' version of the algorithm is also described. The susceptibility of these algorithms to noise is examined together with speed and memory considerations.
A method for monitoring the starter engine includes a starter motor resistance value is determine... more A method for monitoring the starter engine includes a starter motor resistance value is determined, which is connected to an engine start event, and that a back-EMF is determined for the starter motor based on the starter motor resistance value. A functional state of the starter motor, which corresponds to the back electromotive force of the starter motor and a Kraftmaschinenankurbelzeitspanne is determined. A start / stop functionality of the combustion engine is controlled based on the functional state of the starter motor.
An application in the automotive filed for the Genetic Learning Automata with Fuzzy Classifier Sy... more An application in the automotive filed for the Genetic Learning Automata with Fuzzy Classifier System is presented in this work. As a non-linear model free based strategy, the major advantages of this approach are its modularity and its extensib ility. A controller designed using this method for a simple longitudinal vehicle dynamic model is applied to a non-linear model with 107 degrees of freedom giving a reasonable performance. Comparisons with a conventional controller are also carried out.
System and related method for monitoring the health status of sensors in an integrated vehicle st... more System and related method for monitoring the health status of sensors in an integrated vehicle stability control system. In one embodiment, the system determines whether a yaw rate sensor, a lateral acceleration sensor or a steering angle sensor has failed. The system uses several models to generate based on the actual sensor readings estimates of the outputs of the sensors. Are generated deviations than the difference between the measured value and each of the estimated values for the particular sensor. The deviations are compared with a threshold to determine for each deviation, if an error flag is set. The threshold for the steering wheel angle sensor is an adaptive threshold, because it has no physical redundancy. If the error flags for the deviations for each sensor have a certain pattern, an error is issued for that sensor.
A system and method for improving vehicle and vehicle guidance algorithms and for improving vehic... more A system and method for improving vehicle and vehicle guidance algorithms and for improving vehicle maintenance practices. The method includes collecting data of vehicle components, subsystems shall and systems and storing the collected data in a database. The charges stored data can be from multiple sources for similar cars or similar components, and various types of trouble codes and labor codes and other information such as operating data and data on the physics of disorders that are fused, included. The process produces for various vehicle components, subsystems and systems classes and builds using data mining techniques from the data stored in the database feature extractors for each class. Furthermore, the method classifiers that classify the features for each class generated. The feature extractors and Merkmalsklassierer be used to determine a vehicle component for a vehicle subsystem or a vehicle system, warm a fault condition has occurred.
A method includes collection of vehicle operating state data from multiple vehicles. Among the se... more A method includes collection of vehicle operating state data from multiple vehicles. Among the several vehicles, a reference group is identified. The vehicle operating state data from the reference group charges go in a cooperative model a for the vehicle operating state, wherein the cooperative model for the vehicle operating state is applicable to a current vehicle to predict a state of at least one component of the present vehicle.
A disc brake system of a vehicle determines the actual temperature of a brake rotor. The disc bra... more A disc brake system of a vehicle determines the actual temperature of a brake rotor. The disc brake system compares the actual temperature of the brake rotor with a critical temperature of the brake rotor. The critical temperature of the brake rotor is a temperature above which damage to the and / or distortion of the brake rotor / s can occur / can. The disc brake system applies a corrective action to prevent damage to the brake rotor when the actual temperature of the brake rotor is higher than the critical temperature of the brake rotor. The corrective action may include that: a warning, a traction control system of the vehicle is adapted and maintenance of the vehicle is planned, but is not limited to this.
Vehicle System Dynamics
The design and analysis process of mechatronic systems requires a wide range of different enginee... more The design and analysis process of mechatronic systems requires a wide range of different engineering methods and tools. The demands on simulation model descriptions and solver functionalities are rising due to the increase of system complexity. A development process can only be carried out in an effective way if all needed models, methods and codes are linked together to an integrated CAE-environment. In the future, the variety of interfaces and types, from a simple model import/export up to a complex co-simulation, will be further extended by new concepts to combine models and codes. A classification of coupling strategies can help to estimate the effects on simulation performance, functionality and accuracy which is important for the decision of an optimal interface type in case of a specific task. For the covering abstract see ITRD E117109.
SAE Technical Paper Series, 2003
An application in the automotive filed for the Genetic Learning Automata Fuzzy Classifier System ... more An application in the automotive filed for the Genetic Learning Automata Fuzzy Classifier System is presented in this work. As a non-linear model free-based strategy, the major advantages of this approach are its modularity and its extensibility. A controller designed using this method for a quarter-car model is applied to a 6-DOF model giving a reasonable performance. Comparisons with the LQR controller are also carried out.
The design and analysis process of mechatronic systems requires a wide range of different enginee... more The design and analysis process of mechatronic systems requires a wide range of different engineering methods and tools. The demands on simulation model descriptions and solver functionalities are rising due to the increase of system complexity. A development process can only be carried out in an effective way if all needed models, methods and codes are linked together to an integrated CAE-environment. In the future, the variety of interfaces and types, from a simple model import/export up to a complex co-simulation, will be further extended by new concepts to combine models and codes. A classification of coupling strategies can help to estimate the effects on simulation performance, functionality and accuracy which is important for the decision of an optimal interface type in case of a specific task. For the covering abstract see ITRD E117109.
In the design of adaptive IIR filters, the multi-modal nature of the error surfaces can limit the... more In the design of adaptive IIR filters, the multi-modal nature of the error surfaces can limit the use of gradient-based and other iterative search methods. Stochastic learning automata have previously been shown to have global optimisation properties making them suitable ...
A new reinforcement learning algorithm is introduced which can be applied over a continuous range... more A new reinforcement learning algorithm is introduced which can be applied over a continuous range of actions. The learning algorithm is reward-inaction based, with a set of probability density functions being used to determine the action set. An experimental study is presented, based on the control of a semi-active suspension system on a road going, four wheeled, passenger vehicle. The control objective is to minimise the mean square acceleration of the vehicle body, thus improving the ride isolation qualities of the vehicle. This represents a difficult class of learning problem, owing to the stochastic nature of the road input disturbance together with unknown high order dynamics, sensor noise and the non-linear (semi-active) control actuators. The learning algorithm described here operates over a bounded continuous action set, is robust to high levels of noise and is ideally suited to operating in a parallel computing environment.
IEE Seminar Learning Systems for Control
IEEE Access, 2014
Regenerative braking is one of the most promising and environmentally friendly technologies used ... more Regenerative braking is one of the most promising and environmentally friendly technologies used in electric and hybrid electric vehicles to improve energy efficiency and vehicle stability. This paper presents a systematic data-driven process for detecting and diagnosing faults in the regenerative braking system of hybrid electric vehicles. The diagnostic process involves signal processing and statistical techniques for feature extraction, data reduction for implementation in memory-constrained electronic control units, and variety of fault classification methodologies to isolate faults in the regenerative braking system. The results demonstrate that highly accurate fault diagnosis is possible with the classification methodologies. The process can be employed for fault analysis in a wide variety of systems, ranging from automobiles to buildings to aerospace systems.
La presente invention concerne un systeme de commande destine a avertir un vehicule suiveur d'... more La presente invention concerne un systeme de commande destine a avertir un vehicule suiveur d'un risque de collision avec un vehicle qui le precede. Le vehicule de tete comprend un systeme de detection qui detecte la presence et la vitesse du vehicule suiveur. Le systeme de commande determine une distance souhaitee entre le vehicule de tete et le vehicule suiveur selon la vitesse du premier. Le systeme de commande soustrait la distance souhaitee de celle reelle pour generer un signal d'erreur de distance, et compare ce signal a une valeur limite. Si le signal d'erreur est superieur a la limite, le systeme de commande peut effectuer une ou plusieurs operations, telles que le declenchement de feux de detresse pour avertir le vehicule suiveur, ou executer d'autres actions en cas de collision imminente, telles que la pretension de ceintures de securite et la fermeture de vitres du vehicule suiveur.
The implementation of the Hough transform using neural networks is described. This highly paralle... more The implementation of the Hough transform using neural networks is described. This highly parallel implementation is fast and suitable for real-time applications. A 'homing-in' version of the algorithm is also described. The susceptibility of these algorithms to noise is examined together with speed and memory considerations.
A method for monitoring the starter engine includes a starter motor resistance value is determine... more A method for monitoring the starter engine includes a starter motor resistance value is determined, which is connected to an engine start event, and that a back-EMF is determined for the starter motor based on the starter motor resistance value. A functional state of the starter motor, which corresponds to the back electromotive force of the starter motor and a Kraftmaschinenankurbelzeitspanne is determined. A start / stop functionality of the combustion engine is controlled based on the functional state of the starter motor.
An application in the automotive filed for the Genetic Learning Automata with Fuzzy Classifier Sy... more An application in the automotive filed for the Genetic Learning Automata with Fuzzy Classifier System is presented in this work. As a non-linear model free based strategy, the major advantages of this approach are its modularity and its extensib ility. A controller designed using this method for a simple longitudinal vehicle dynamic model is applied to a non-linear model with 107 degrees of freedom giving a reasonable performance. Comparisons with a conventional controller are also carried out.
System and related method for monitoring the health status of sensors in an integrated vehicle st... more System and related method for monitoring the health status of sensors in an integrated vehicle stability control system. In one embodiment, the system determines whether a yaw rate sensor, a lateral acceleration sensor or a steering angle sensor has failed. The system uses several models to generate based on the actual sensor readings estimates of the outputs of the sensors. Are generated deviations than the difference between the measured value and each of the estimated values for the particular sensor. The deviations are compared with a threshold to determine for each deviation, if an error flag is set. The threshold for the steering wheel angle sensor is an adaptive threshold, because it has no physical redundancy. If the error flags for the deviations for each sensor have a certain pattern, an error is issued for that sensor.
A system and method for improving vehicle and vehicle guidance algorithms and for improving vehic... more A system and method for improving vehicle and vehicle guidance algorithms and for improving vehicle maintenance practices. The method includes collecting data of vehicle components, subsystems shall and systems and storing the collected data in a database. The charges stored data can be from multiple sources for similar cars or similar components, and various types of trouble codes and labor codes and other information such as operating data and data on the physics of disorders that are fused, included. The process produces for various vehicle components, subsystems and systems classes and builds using data mining techniques from the data stored in the database feature extractors for each class. Furthermore, the method classifiers that classify the features for each class generated. The feature extractors and Merkmalsklassierer be used to determine a vehicle component for a vehicle subsystem or a vehicle system, warm a fault condition has occurred.
A method includes collection of vehicle operating state data from multiple vehicles. Among the se... more A method includes collection of vehicle operating state data from multiple vehicles. Among the several vehicles, a reference group is identified. The vehicle operating state data from the reference group charges go in a cooperative model a for the vehicle operating state, wherein the cooperative model for the vehicle operating state is applicable to a current vehicle to predict a state of at least one component of the present vehicle.
A disc brake system of a vehicle determines the actual temperature of a brake rotor. The disc bra... more A disc brake system of a vehicle determines the actual temperature of a brake rotor. The disc brake system compares the actual temperature of the brake rotor with a critical temperature of the brake rotor. The critical temperature of the brake rotor is a temperature above which damage to the and / or distortion of the brake rotor / s can occur / can. The disc brake system applies a corrective action to prevent damage to the brake rotor when the actual temperature of the brake rotor is higher than the critical temperature of the brake rotor. The corrective action may include that: a warning, a traction control system of the vehicle is adapted and maintenance of the vehicle is planned, but is not limited to this.
Vehicle System Dynamics
The design and analysis process of mechatronic systems requires a wide range of different enginee... more The design and analysis process of mechatronic systems requires a wide range of different engineering methods and tools. The demands on simulation model descriptions and solver functionalities are rising due to the increase of system complexity. A development process can only be carried out in an effective way if all needed models, methods and codes are linked together to an integrated CAE-environment. In the future, the variety of interfaces and types, from a simple model import/export up to a complex co-simulation, will be further extended by new concepts to combine models and codes. A classification of coupling strategies can help to estimate the effects on simulation performance, functionality and accuracy which is important for the decision of an optimal interface type in case of a specific task. For the covering abstract see ITRD E117109.
SAE Technical Paper Series, 2003
An application in the automotive filed for the Genetic Learning Automata Fuzzy Classifier System ... more An application in the automotive filed for the Genetic Learning Automata Fuzzy Classifier System is presented in this work. As a non-linear model free-based strategy, the major advantages of this approach are its modularity and its extensibility. A controller designed using this method for a quarter-car model is applied to a 6-DOF model giving a reasonable performance. Comparisons with the LQR controller are also carried out.
The design and analysis process of mechatronic systems requires a wide range of different enginee... more The design and analysis process of mechatronic systems requires a wide range of different engineering methods and tools. The demands on simulation model descriptions and solver functionalities are rising due to the increase of system complexity. A development process can only be carried out in an effective way if all needed models, methods and codes are linked together to an integrated CAE-environment. In the future, the variety of interfaces and types, from a simple model import/export up to a complex co-simulation, will be further extended by new concepts to combine models and codes. A classification of coupling strategies can help to estimate the effects on simulation performance, functionality and accuracy which is important for the decision of an optimal interface type in case of a specific task. For the covering abstract see ITRD E117109.
In the design of adaptive IIR filters, the multi-modal nature of the error surfaces can limit the... more In the design of adaptive IIR filters, the multi-modal nature of the error surfaces can limit the use of gradient-based and other iterative search methods. Stochastic learning automata have previously been shown to have global optimisation properties making them suitable ...
A new reinforcement learning algorithm is introduced which can be applied over a continuous range... more A new reinforcement learning algorithm is introduced which can be applied over a continuous range of actions. The learning algorithm is reward-inaction based, with a set of probability density functions being used to determine the action set. An experimental study is presented, based on the control of a semi-active suspension system on a road going, four wheeled, passenger vehicle. The control objective is to minimise the mean square acceleration of the vehicle body, thus improving the ride isolation qualities of the vehicle. This represents a difficult class of learning problem, owing to the stochastic nature of the road input disturbance together with unknown high order dynamics, sensor noise and the non-linear (semi-active) control actuators. The learning algorithm described here operates over a bounded continuous action set, is robust to high levels of noise and is ideally suited to operating in a parallel computing environment.
IEE Seminar Learning Systems for Control