Fuzzy Pattern Recognition Technique Research Articles

A Novel Traffic Analysis Model for Botnet Discovery in Dynamic Network

In this paper, we propose a collaborative pattern-based filtering algorithm which is a behavior-based approach to detect bots in association with case-based reasoning and fuzzy pattern recognition techniques. Network traces are used as a pivotal element to inspect bot-relevant domain names and IP addresses. Particularly, this method extracts the features, and making use of such features along with the IP address, the case-based reasoning is performed. If the address is known, it will be classified as a known bot, whereas if it is unknown, the fuzzy-based mapping is performed to detect botnet. This proposed approach especially reduces the search time and enhances the prediction accuracy up to 96%, and it is also observed that it improves the knowledge repository.

Fuzzy pattern recognition technique for crack propagation on earplate connection of guyed mast under wind load

A 2-stage damage identification technique is presented for identifying crack lengths on cable-tower connections of guyed masts. It is based on fuzzy pattern recognition theory and uses the cable force and the strain at the key locations of the earplate connection as input parameters. The methodology is developed with a numerical model under stochastic wind excitation and verified by numerical simulation and experiment. First, the relationships of cable force, strain on the connection, and crack length are explored. The wind-induced responses of guyed mast are analyzed under different wind speed and directions to form a database of cable forces. The responses of a typical connection with different crack lengths are simulated to investigate the strain sensitivity to crack propagation. Its results are used to form a second-stage strain database. The first stage of the fuzzy identification consists in identifying the interval of cable forces on the basis of cable force measurements. Then, a subjection degree function is formed between the measured strain and the developed strain database to judge the damage state of the connection by identifying the existence of a crack and estimating its length. The results of a numerical case study and further experiment are presented and show the potential and practicability of the technique to estimate the right range of crack length on earplates. Therefore, the proposed technique is practical and feasible to predict the damage state of earplate and can provide valuable prewarning information for structural safety.

A three-step methodology for GI classification, GL estimation of foods by fuzzy c-means classification and fuzzy pattern recognition, and an LP-based diet model for glycaemic control

In this paper, a three-step methodology is proposed for assigning foods with measured glycaemic index (GI) values to GI classes by using the fuzzy c-means classification technique, assigning foods with no measured GI values to GI classes by using the fuzzy pattern recognition technique, and estimating the glycaemic load (GL) values of foods with no measured GI values. In this methodology, the decision rules for menu planning are also defined, and a Linear Programming-based (LP-based) diet model is developed with the objective function of minimizing the total dietary glycaemic load and the constraints of the daily nutritional requirements. An application based on the real data of GI, GL, and nutritional values of the foods is also provided.

A fuzzy pattern-based filtering algorithm for botnet detection

Botnet has become a popular technique for deploying Internet crimes. Although signature-based bot detection techniques are accurate, they could be useless when bot variants are encountered. Therefore, behavior-based detection techniques become attractive due to their ability to detect bot variants and even unknown bots. In this paper, we propose a behavior-based botnet detection system based on fuzzy pattern recognition techniques. We intend to identify bot-relevant domain names and IP addresses by inspecting network traces. If domain names and IP addresses used by botnets can be identified, the information can be further used to prevent protected hosts from becoming one member of a botnet. To work with fuzzy pattern recognition techniques, we design several membership functions based on frequently observed bots’ behavior including: (1) generate failed DNS queries; (2) have similar DNS query intervals; (3) generate failed network connections; and (4) have similar payload sizes for network connections. Membership functions can be easily altered, removed, or added to enhance the capability of the proposed system. In addition, to improve the overall system performance, we develop a traffic reduction algorithm to reduce the amount of network traffic required to be inspected by the proposed system. Performance evaluation results based on real traces show that the proposed system can reduce more than 70% input raw packet traces and achieve a high detection rate (about 95%) and a low false positive rates (0–3.08%). Furthermore, the proposed FPRF algorithm is resource-efficient and can identify inactive botnets to indicate potential vulnerable hosts.

Multi-Axle Moving Train Loads Identification by Using Fuzzy Pattern Recognition Technique

Identification of multi-axle moving loads on bridge is very important for bridge design, construction, and maintenance in engineering field. It is complicated and time consuming to identify the multi-axle moving train loads with general identification methods and far away from practical practice. Based on the theory of fuzzy pattern recognition, the fuzzy pattern recognition method for multi-axle moving train loads identification on bridge is presented in this paper. The multi-axle moving loads pattern library on a simply supported bridge is established with numerical methods. Effect of measurement noise on the proposed method is investigated in three situations. The results show that the proposed identification method has a certain resistance to measurement noise and can realize moving train loads identification with high accuracy.

ECT and LS-SVM Based Void Fraction Measurement of Oil-Gas Two-Phase Flow

A method based on Electrical Capacitance Tomography (ECT) and an improved Least Squares Support Vector Machine (LS-SVM) is proposed for void fraction measurement of oil-gas two-phase flow. In the modeling stage, to solve the two problems in LS-SVM, pruning skills are employed to make LS-SVM sparse and robust; then the Real-Coded Genetic Algorithm is introduced to solve the difficult problem of parameters selection in LS-SVM then. In the measurement process, the flow pattern of oil-gas two-phase flow is identified by using fast back-projection image reconstruction and a fuzzy pattern recognition technique and the void fraction is computed using the void fraction model corresponding to the identified flow pattern. Experimental results demonstrate that both the improvement of LS-SVM and the parameter optimization are effective. The results also show that the real-time performance of the proposed void fraction measurement method is good, and the measurement precision can satisfy the application requirement.

From driving cycle analysis to understanding battery performance in real-life electric hybrid vehicle operation

This paper proposes a methodology and approach to understand battery performance and life through driving cycle and duty cycle analyses from electric and hybrid vehicle (EHV) operation in real-world situations. Conducting driving cycle analysis with trip data collected from EHV operation in real life is very difficult and challenging. In fact, no comprehensive approach has been accepted to date, except those using standard driving cycles on a dynamometer or a track. Similarly, analyzing duty cycle performance of a battery under real-life operation faces the same challenge. A successful driving cycle analysis, however, can significantly enhance our understanding of EHV performance in real-life driving. Likewise, we also expect similar results through duty cycle analysis for batteries. Since 1995, we have been developing tools to analyze EHV and power source performance. In particular, we were able to collect data from a fleet of 15 Hyundai Santa Fe electric sports utility vehicles (e-SUVs) operated on Oahu, Hawaii; from July 2001 to June 2003 to allow driving and duty cycle analyses in order to understand battery pack performance from a variety of EHV operating conditions. We thus developed a comprehensive approach that comprises fuzzy logic pattern recognition (FL-PR) techniques to perform driving and duty cycle analyses. This approach has been successfully applied to EHV performance analysis via the creation of a compositional driving profile called “driving cycle profile” (DrCP) for each trip. The same approach was used to analyze battery performance via the construction of “duty cycle profile” (DuCP) to express battery usage under various operating conditions. The combination of the two analyses enables us to understand both the usage profile of EHV and battery performance in synergetic details and in a systematic manner using a pattern recognition technique.

Voidage Measurement of Gas-Oil Two-phase Flow

A new method for the voidage measurement of gas-oil two-phase flow was proposed. The voidage measurement was implemented by the identification of flow pattern and a flow pattern specific voidage measurement model. The flow pattern identification was achieved by combining the fuzzy pattern recognition technique and the crude cross-sectional image reconstructed by the simple back projection algorithm. The genetic algorithm and the partial least square method were applied to develop the voidage measurement models. Experimental results show that the proposed method is effective. It can overcome the influence of flow pattern on the voidage measurement, and also has the advantages of simplicity and speediness.

An Online Flow Pattern Identification System for Gas–Oil Two-Phase Flow Using Electrical Capacitance Tomography

An online and noninvasive gas-oil two-phase flow pattern identification system is proposed based on the electrical capacitance tomography (ECT) technique. A fuzzy pattern recognition technique is introduced to solve the flow pattern identification problem because the flow pattern and its transition are fuzzy in nature. Images captured by the ECT system are analyzed through image processing and a pattern recognition technique. The flow pattern identification results are achieved by fuzzy sets theory. The online flow pattern identification system is established, and typical flow patterns, such as the homogeneous flow, the stratified flow, the annular flow, and the slug flow, in horizontal pipeline can be identified. The experimental results show that the online flow pattern identification system is effective

A fuzzy ARTMAP fault classifier for impulse testing of power transformers

The work presents an artificial intelligence (AI) based impulse test technique for oil filled power transformers. Determination of exact nature and location of faults, during impulse testing of large power transformer is of practical importance to the transformer manufacturers as well as designers. The presently available impulse test techniques more or less depend on expertise of the test personnel, and in many cases lead to ambiguity and controversy. The new AI approach presented in the paper overcomes the limitations of conventional test methods. This new technique relies on high discrimination power and excellent generalization ability of fuzzy neural networks in complex pattern classification problem. The proposed method employs a fuzzy ARTMAP pattern recognition technique to recognize the frequency responses of the winding admittance of high voltage transformers under healthy and different faulty conditions of winding insulation, and learns to establish the correlations between the nature and physical location of occurrence of an internal insulation fault in a transformer winding and its associated frequency response. The technique was tested on the winding model of typical high voltage transformer and yielded high diagnostic accuracy by successful detection and discrimination of faults of different nature and different site of occurrence in the high voltage winding.

Fuzzy pattern recognition of circadian cycles in ecosystems

Many ecological variables show a wide range of fluctuations, the most important of which is the diurnal variation. This cycling may contain important information regarding the ecosystem’s functioning and, if properly interpreted, can represent a valuable predictive tool in ecosystems management. This paper describes a simple algorithm for extracting meaningful information from daily cycles using fuzzy pattern recognition techniques. The algorithm is organised in three parts: in the first, typical patterns are extracted from experimental data to form the knowledge-base upon which the algorithm operates. The second step is to condense the information contained in the knowledge-base into mathematical objects, referred to in the paper as fuzzy masks. The third step is the set-up of an inferential set of fuzzy rules, using the fuzzy masks as antecedents. Depending on how the inference engine is structured, the algorithm output can be viewed as an assessment of the current daily cycle with respect to a set of “typical” system behaviours and decisions can be made accordingly. Two applications are presented to demonstrate the algorithm. In the first a medium-scale wastewater treatment plant with minimum instrumentation is considered. Based on diurnal flow variations of the last 2 days, the algorithm determines the current treatment level in terms of sludge recycling. The other application is the prediction of the eutrophication level in a coastal lagoon. In this case, the algorithm considers the diurnal fluctuations of basic water quality parameters such as dissolved oxygen, oxidation–reduction potential, pH and temperature to detect the possibility of a macroalgae bloom. In both cases, the algorithm provides information about the shift of the system from one trend to another on the basis of the analysis of diurnal cycles. The algorithm performance is assessed against practical situations and the results discussed.

Fuzzy pattern recognition of tungsten inert gas weld quality

In this paper, a fuzzy pattern recognition technique is applied to classifying aluminium weld quality in tungsten inert gas (TIG) welding. The pattern vector includes three components, that is, the front height, the back height, and the front width of weld. Based on the values of the pattern vector, good, fair, and poor weld qualities can be automatically classified by using the fuzzy pattern recognition technique. Experimental results under different welding parameters are presented to illustrate the proposed method.

Identifying Accident-Prone Locations Using Fuzzy Pattern Recognition

This paper describes a method to identify accident-prone locations (APLs) based on an assessment of factors that contribute to accidents. Current methods to identify APLs make no distinction between accidents that result from road- and nonroad-related factors. Combining accidents that are treatable and nontreatable by road improvements can be misleading and may lead to a misallocation of funds by road authorities. This paper presents a computerized procedure that uses safety experts' knowledge on classifying accidents into a finite set of categories. In practice, the categories can include any one or a combination of the three basic highway system components: the driver, the vehicle, and the road environment. Realizing the complex interaction of these components within the accident environment, the procedure employs fuzzy pattern recognition techniques for the classification process. Accidents that do not belong to the road environment category are excluded from the identification of APLs. The method is tested using data from the accident database of the British Columbia Ministry of Transportation and Highways. The method and results are described.

Tool Wear Monitoring in Diamond Turning by Fuzzy Pattern Recognition

This paper introduces a fuzzy pattern recognition technique for monitoring single crystal diamond tool wear in the ultraprecision machining process. Selected features by which to partition the cluster of patterns were obtained by time series AR modeling of dynamic cutting force signals. The wear on a diamond tool edge appears to be classifiable into two types, micro-chipping and gradual, both very small compared to conventional tool wear. In this regard, we used a fuzzy technique in pattern recognition, which considers the ambiguity in classification as well as the weakness of the cutting force variation, to monitor the diamond tool wear status, with satisfactory results.

Knowledge-Based Expert System for Ballscrew Grinding

This paper discusses the ballscrew grinding process and provides a newly structured knowledge-based expert system (KBES) developed to automate the process based on fuzzy pattern recognition techniques for shape control with recommendations for further optimization considering practical machining constraints. The KBES offers several advantages including the use of semiskilled operators who can monitor several machines simultaneously, the consistent production of high quality parts, and the significant reduction in cycle time due to automatic adjustment of steadies. A new stepper motor steady design and in-process measuring system are also described.

Fuzzy Pattern Recognition for Tool Wear Monitoring in Diamond Turning

Abstract In this paper, to monitor single crystal diamond tool wear in ultraprecision machining process, a fuzzy pattern recognition technique was introduced. Some selected features to partition the cluster of pattern were obtained from time series AR modeling of dynamic cutting force signals. The wear on a diamond tool edge appears to be classifiable into two types, one of which is micro-chipping and the other gradual, both very small quantities to detect compared to conventional tool wear. In this regard, a fuzzy technique in pattern recognition, which considers the ambiguity in classification as well as the weakness of the cutting force variation, was used for monitoring the diamond tool wear status.

False color data imaging: A new pattern recognition technique for analyzing chromatographic profile data

Fuzzy pattern recognition techniques were used to analyze the gas chromatograms of hexane soaks obtained from red carpenter ants. The soaks were found to convey information about the caste and social experience of the ants. In this study a new mapping and display technique, which depends upon high-resolution computer graphics for the presentation of results, was used to develop these relationships.

Monitoring Drilling Wear States by a Fuzzy Pattern Recognition Technique

This paper introduces a new approach for on-line monitoring of drill wear states by using a fuzzy C-means algorithm. Experimental and simulation results have shown that drill wear conditions can be represented by four fuzzy grades. They are: “initial,” “small,” “normal,” and “severe.” The grade “severe” is proposed to be used as the prediction of tool replacement. This fuzzy technique is more adequate than conventional pattern recognition technqiues.