Data processing: artificial intelligence – Fuzzy logic hardware – Analog fuzzy computer
Patent
1996-09-23
1998-12-01
Davis, George B.
Data processing: artificial intelligence
Fuzzy logic hardware
Analog fuzzy computer
706903, 706924, G06F 944
Patent
active
058452699
DESCRIPTION:
BRIEF SUMMARY
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to a fuzzy logic control system which uses multiple input modules to reduce the rule base.
2. Description of the Related Art
Fuzzy control systems can be divided into a fuzzification domain, a domain of a rule base and inference formation and into a defuzzification domain.
In a fuzzy control system, the input values are referred to as sharp input values. In the fuzzification, the sharp input values are converted into a fuzzy value in order to be further-processed in the fuzzy control system. Each input of the fuzzy control system is referenced with a linguistic variable, for example body size or weight of a patient. An input range in which the sharp input value can vary is defined for this variable. The input range is subdivided into sub-ranges, what are referred to as the fuzzy sets. Each of these fuzzy sets contains a characteristic designation. A linguistic variable, for example "weight", can be divided into the ranges small, average, big. Each fuzzy set is presented by a function. The membership degree (.mu.) with which a sharp input value is located in the range of a fuzzy set is determined with this function. The result of the determination of membership degrees is not a boolean value but ranges between 0 and 1.
The system behavior is determined in the domain of the rule base and inference formation. The rule base represents a collection of verbally formulated rules that link the input and output fuzzy sets. Given two inputs and one output, the rules have the following form: a.sub.im or, respectively, a.sub.jm represent a fuzzy set of the linguistic input variables A.sub.i or, respectively, A.sub.j. B.sub.k stands for a linguistic output variable with the fuzzy set b.sub.kn.
A rule determines what output fuzzy sets should be weighted by what input fuzzy sets. A membership value is assigned to an output fuzzy set in the weighting. An output set can only be weighted with one value. Given more than one input fuzzy set, the resultant weighting factor must be determined first. To that end, the inputs can be operated with a number of operators.
The inference formation implements the weighting of the output set. Two methods are thereby possible. The first is to cut the output fuzzy set off at the height of the resultant membership degree. This is thereby refer red to as the "min-max method". The second method would be to proportionally reduce the output fuzzy set from its normal size to the height that corresponds to the identified membership degree. This is the "max-prod method".
In the defuzzification domain, the weighted output fuzzy sets obtained in the inference formation are brought into a form as a system result that can be understood by the "non-fuzzy world". Fuzzification methods are the maximum method, the left-(right-)max method and the center of gravity method. Which method is selected is entirely dependent on how the fuzzy result is to be interpreted.
When the combination in which the input values are adjacent is not taken into account in a fuzzy control system, then each combination must be considered to be possible. Given, for example, six different inputs with a total of 37 fuzzy sets, 47,250 rules must be erected. Due to the multitude of rules to be erected, the risk is high that some rules will not have been formulated. A check must be carried out to see whether all rules were in fact formulated, this resulting in a high time expenditure.
European published application EP-A1-0 063 644 discloses to set the exposure parameters (voltage/current, Mas product, the blackening, the measuring field) allocated to organ or group of organs. The ultimate setting of these parameters ensues under the influence of radiation.
French published application FR-A1-2 608 416 discloses an exposure control of an X-ray diagnostics apparatus wherein logarithmic characteristics are employed for the linearization of the control circuit. The data acquisition for the linearization of the control circuit ensues under the infl
REFERENCES:
patent: 5148032 (1992-09-01), Hernandez
Proceedings of the IEEE 1990 Custom Integrated Circuits Conference, Dettloff et al, "A VLSI Fuzzy Logic Interference Engine for Real-Time Processng Control", May 1989.
Xiang et al., "The Processing and Recognition of X-Rays of Spine Tumor", 9th International Conference on Pattern Recognition, Nov. 1988, Italy, 14-17.
Franz Michael
Kortge Detlef
Davis George B.
Siemens Aktiengesellschaft
LandOfFree
Fuzzy control system, particularly for dose rate control in an X does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Fuzzy control system, particularly for dose rate control in an X, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Fuzzy control system, particularly for dose rate control in an X will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2403416