We propose a new method for fuzzy rule extraction from data by a genetic algorithm and a fine tuning of the extracted membership functions by a constrained nonlinear optimization. This approach is able to select the m...
详细信息
We propose a new method for fuzzy rule extraction from data by a genetic algorithm and a fine tuning of the extracted membership functions by a constrained nonlinear optimization. This approach is able to select the most significant rules out of a set of all possible ones, that is it learns the rule structure by itself. The genetic algorithm does not limit the kind of operator and the number and form of the membership functions for the inputs. However, in order to utilize linear optimization techniques, singletons and center of gravity defuzzification are used on the output side. Since each rule premise may include a conjunction of a variable number of inputs (between one and the input dimension), the "curse of dimensionality" can be overcome, that is the number of rules does not increase exponentially with the input dimension. This feature makes the proposed algorithm especially attractive for interpretation of high dimensional nonlinear mappings that are hard to visualize. The strategy followed by the nonlinear optimization of the fuzzy input membership functions focuses on a good interpretability rather than on best approximation performance. This will be demonstrated on a real world data example.
A general procedure for model based fault detection and diagnosis is first described. It comprises the steps of a knowledge based approach by merging analytical and heuristic knowledge. Model based fault detection met...
详细信息
A general procedure for model based fault detection and diagnosis is first described. It comprises the steps of a knowledge based approach by merging analytical and heuristic knowledge. Model based fault detection methods are summarized together with underlying fault models and appropriate combinations are proposed. Then fault diagnosis based on analytical and heuristic symptoms by using methods of approximate reasoning with fuzzy-logic is briefly described. Finally the practical application for an automotive electro-mechanical actuator is shown.
作者：
Rolf IsermannInstitute of Automatic Control
Laboratory of Control Engineering and Process Automation Technical University of Darmstadt Landgraf-Georg-Str 4 D-64283 Darmstadt Germany
The integration of mechanical processes and microelectronics towards mechatronic systems opens new possibilities as well for the design of mechanical components as for automatic functions. The contribution discusses f...
详细信息
The integration of mechanical processes and microelectronics towards mechatronic systems opens new possibilities as well for the design of mechanical components as for automatic functions. The contribution discusses first the involved mechanical components and machines and the ways of integration. Then the different automation functions are described in the frame of intelligent control systems which contain multilevel control functions, a knowledge base, and inference mechanisms. Multilevel feedback control for mechanical systems comprises lower level and higher level control, including e.g. nonlinear adaptive control and fuzzy control. The inclusion of model based supervision and fault diagnosis is a further development step. Two examples of mechatronic systems for cars are shown, like an adaptive suspension system and selftuning damping of drive chain oscillations.
Faults which appear in technical processes can often be described as additive or multiplicative faults with respect to the process model. To perform fast detection of these faults continuous-time parity equations are ...
详细信息
Faults which appear in technical processes can often be described as additive or multiplicative faults with respect to the process model. To perform fast detection of these faults continuous-time parity equations are used. Parameter deviations are estimated directly from residuals providing information about their size. The combined method enables to distinguish between additive and parametric faults. In case of time variant processes with slow parameter changes the coefficients of the parity equations can also be adapted with respect to the tracked parameters. The problem of persistent process excitation for estimation is by-passed. The fault detection scheme is demonstrated at a permanently excited *** on a laboratory rig. Its properties and the experimental results are discussed.
Servo systems play an important role in many automated processes. In order to fulfil the hard demands on reliability and fast and precise operation, intelligent concepts for the control, supervision and (reconfigurati...
详细信息
Servo systems play an important role in many automated processes. In order to fulfil the hard demands on reliability and fast and precise operation, intelligent concepts for the control, supervision and (reconfiguration are necessary. In this paper, an approach is presented which integrates different levels of signal processing in an electromechanical servo system. The digital controller and the model-based fault detection scheme are designed taking into account model-uncertainty and the time variant process behaviour, which is caused by temperature influences. After a brief description of the theoretical basis an experimental application shows results for an automobile servo system which is driven by a d.c. motor.
After a short view on the historical development of model-based fault detection some proposals for the terminology in the field of supervision, fault detection and diagnosis are stated based on the work within the IFA...
详细信息
After a short view on the historical development of model-based fault detection some proposals for the terminology in the field of supervision, fault detection and diagnosis are stated based on the work within the IFAC Technical Committee SAFEprocess. Some basic fault detection and diagnosis methods are briefly considered. Then, an evaluation of publications during the last 5 years shows some trends in the application of model-based fault detection and diagnosis methods.
作者：
H. KonradIsermann R.Technical University of Darmstadt
Institute of Automatic Control Laboratory of Control Engineering and Process Automation Landgraf-Georg-Str. 4 D-64283 Darmstadt Germany Phone: +496151 163927 Fax: +49 6151 293445
A new method of fault detection in milling is described. The method uses exclusively drive signals and is based on models for the feed drive and the milling process. Using parameter estimation features are generated w...
详细信息
A new method of fault detection in milling is described. The method uses exclusively drive signals and is based on models for the feed drive and the milling process. Using parameter estimation features are generated which are independent of cutting conditions. A subsequent classifier evaluates the process state and provides a reliable diagnosis of the milling process.
As individual and commercial traffic flow on roads and highways grows enormously, rhe number of accidents increases as well. Therefore, modern vehicle research is focused on improving driving comfort as well as passen...
详细信息
As individual and commercial traffic flow on roads and highways grows enormously, rhe number of accidents increases as well. Therefore, modern vehicle research is focused on improving driving comfort as well as passengers' safety. Aiming at that, recent advances in controlengineering and modern computer technology enable the engineer to design special control and supervision systems supporting the driver. Exemplary, this contribution presents two possible solutions. On the one hand, an Adaptive Cruise control system which assists the driver during highway traffic, whereas a vehicle supervision method is applied to detect critical driving situations and sensor faults.
作者：
H. KonradTechnical University of Darmstadt
Institute of Automatic Control Laboratory of Control Engineering and Process Automation Landgraf-Georg-Str.4 D-64283 Darmstadt Germany Phone: +49 6151 163927 Fax: +49 6151 293445
In this paper a new method of fault detection in milling is reported. Based on measured cutting forces, model parameters are estimated for each insert of the milling cutter. Using a classifier, the patterns of these e...
详细信息
In this paper a new method of fault detection in milling is reported. Based on measured cutting forces, model parameters are estimated for each insert of the milling cutter. Using a classifier, the patterns of these estimated parameters are processed further and the state of the milling process is determined. The method is first tested with simulated data and then verified with measurements on a machining center.
Due to the rising consciousness of safety aspects the supervision of vehicles' tire pressure is a major effort to improve active car safety. Therefore, in this contribution a method for monitoring the tire pressur...
详细信息
Due to the rising consciousness of safety aspects the supervision of vehicles' tire pressure is a major effort to improve active car safety. Therefore, in this contribution a method for monitoring the tire pressure is presented using body acceleration signals. Analysing the frequency spectrum of the virtual transfer function between the body acceleration at the front and the rear wheel of one side of the vehicle characteristic features are generated. Thereby, external interferences to the spectrum and their influences to the symptoms are discussed. Then, a neuro-fuzzy classification of the characteristics is applied to quantify the tire pressure.
暂无评论