François Delmotte

Target identification based on the transferable belief model interpretation of dempster-shafer model

This paper explains how multisensor data fusion and target identification can be performed within the transferable belief model (TBM), a model for the representation of quantified uncertainty based on belief functions. We present the underlying theory, in particular the general Bayesian theorem needed to transform likelihoods into beliefs and the pignistic transformation needed to build the probability measure required for decision making. We present how this method applies in practice. We compare its solution with the classical one, illustrating it with an embarrassing example, where the TBM and the probability solutions completely disagree. Computational efficiency of the belief-function solution was supposedly proved in a study that we reproduce and we show that in fact the opposite conclusions hold. The results presented here can be extended directly to many problems of data fusion and diagnosis.

Continuous Takagi–Sugeno's Models: Reduction of the Number of LMI Conditions in Various Fuzzy Control Design Technics

Fuzzy models of the Takagi-Sugeno type enable representation of a wide class of nonlinear models. Conditions about their stabilization can be derived systematically. Usually these conditions are written as linear matrix inequality (LMI) problems. In this paper, we use a collection of properties concerning matrices to extend the area of solutions involving the feedback control laws. We will work out with continuous fuzzy models, and supply results about the stabilization, the regulator problem, and also Hinfin control. Solutions based on a reduced number of decision variables, implying a reduced number of LMI conditions, are also proposed. We show that a good tradeoff can been obtained.

Belief functions contextual discounting and canonical decompositions

In this article, the contextual discounting of a belief function, a classical discounting generalization, is extended and its particular link with the canonical disjunctive decomposition is highlighted. A general family of correction mechanisms allowing one to weaken the information provided by a source is then introduced, as well as the dual of this family allowing one to strengthen a belief function.

Fouling detection in a heat exchanger by observer of Takagi-Sugeno type for systems with unknown polynomial inputs

This paper proposes a new method for fouling detection in a heat exchanger. It is based on the modeling of the system in a fuzzy Takagi-Sugeno representation derived from a physical model. With this representation, the design of a fuzzy observer with unknown inputs of polynomial types is obtained via a LMI formulation. Main advantages of the proposed method are that neither specific sensor, excepted standard ones, nor special operating conditions such as steady state regime are required. Some realistic simulations show the efficiency of the proposed technique.

Detection of defective sources in the setting of possibility theory

Possibility theory offers appealing tools to manage uncertain and imprecise data. This paper studies the problem of fusing information stemming from several sources. Different operators already exist but they have problems with conflicting data. The discounting approach weights the respective impacts of sources and solves most of these problems. But we need to assess the discounting factors correctly. A solution is proposed with the assumption that conflicts come from defective sources. In this paper defective means that we trust a source, and we give it a high reliability, but suddenly it supplies wrong reports that conflict with the reports from other sources. Our algorithm detects such a failure and improves the fusion step. Meanwhile a new fusion rule is introduced. Indeed, the discounting approach extends the support of the resulting distribution to the reference set, which is debatable. A few comparisons are provided.

Discrete Takagi–Sugeno's Fuzzy Models: Reduction of the Number of LMI in Fuzzy Control Techniques

Conditions about the stabilization of Takagi-Sugeno fuzzy models can systematically be derived and solved usually using linear matrix inequality problems. Current research tries to lower any pessimism. However, often it leads to an important increase in the number of decision variables, and problems become unsolvable. In this correspondence, we choose to reduce the number of decision variables while not raising the conservatism in comparison with previous results. This correspondence deals with the discrete case, which is harder to solve. We supply results about the stabilization, the regulator problem, and also Hinfin control.

Object tracking and credal classification with kinematic data in a multi-target context

Abstract This article proposes a method to classify multiple maneuvering targets at the same time. This task is a much harder problem than classifying a single target, as sensors do not know how to assign captured observations to known targets. This article extends previous results scattered in the literature and unifies them in a single global framework with belief functions. Through two examples, it is shown that the full algorithm using belief functions improves results obtained with standard Bayesian classifiers and that it can be applied to a large variety of applications.

A fuzzy multi-criteria evaluation method for designing fashion oriented industrial products

In this paper, we present a fuzzy multi-criteria decision making method for evaluating a set of fashion oriented industrial products in order to design new products meeting specific market requirements. Human perceptions at two levels (basic product perception and complex fashion perception), evaluated by a group of evaluators, have been integrated into the related evaluation procedure. For a specific product, the three first fuzzy evaluation criteria are its conformity degrees related to the specific consumer’s requirement in fashion themes, basic product perception and functional properties. The degree of conformity between the basic product perception and the complex fashion perception, and the price of the product constitute the two remaining evaluation criteria. The previous conformity degrees are formalized according to the measures of dissimilarity between products as well as dissimilarity and inclusion between different fashion themes. The weights of the evaluation criteria are linguistic variables generated from the results of market classification obtained by a parametric identification method. These weights can effectively characterize the relationship between sales volumes of products and their components (price, fashion style, physical features, and basic perception). Finally, the set of all existing products can be evaluated and ranked by aggregating the previous fuzzy evaluation criteria with linguistic weights. The proposed fuzzy multi-criteria evaluation method has been applied to select the most relevant industrial products for different markets. Moreover, as the general aggregated evaluation criterion can be considered as a quality function of design parameters (functional properties, basic and fashion complex perceptions) for a specific market, we can estimate this function by evaluating all existing products in order to design new consumer oriented products.

Comparative study of supervised classification algorithms for the detection of atmospheric pollution

Abstract The management of atmospheric pollution using video is not yet widespread. However it is an efficient way to evaluate the polluting rejects coming from large industrial facilities when traditional captors are not usable. This paper presents a comparison of different classifiers for a monitoring system of polluting smokes. The data used in this work are stemming from a system of video analysis and signal processing. The database includes the pollution level of puffs of smoke defined by an expert. Six machine learning techniques are tested and compared to classify the puffs of smoke: k-nearest neighbour, naive Bayes classifier, artificial neural network, decision tree, support vector machine and a fuzzy model. The parameters of each type of classifier are split into three categories: learned parameters, parameters determined by a first step of the experimentation, and parameters set by the programmer. We compare the results of the best classifier of each type depending on the size of the learning set. A part of the discussion concerns the robustness of the classifier facing the case where classes of interest are under-represented, as the high level of pollution in our data.

A high-level application using belief functions for exchanging and managing uncertain events on the road in vehicular ad hoc networks

This article introduces a high-level system using belief functions for exchanging and managing imperfect information about events on the road in vehicular ad hoc networks. The main purpose of this application is to provide the most reliable information for the driver from multiple messages received informing the driver about events on the roads. This system and some variants are tested using a MATLAB™ simulator. An implementation with Android smartphones using a Bluetooth technology to exchange the messages is also introduced.