Abstract

International audience; This paper deals with sensors faults detection and isolation using method by observers. The principle is to reconstruct the state vector and measurements of the system by sliding mode observers and compare the estimated outputs with those measured. In this work, a multi observer's technique is used. It consists on construction of many observers, at least one observer for each output (sensor measurement). Each observer must be robust to noises and to other uncertainties but sensitive to sensor faults. The residual is the estimation error witch is the difference between sensor measurement and its estimated. Without failure in the sensor, this residual remains around zero and if a fault is occurred, it deviates significantly from zero. This method is applied to an autonomous electric vehicle called RobuCar. Simulation results are given at the end to show the effectiveness of the approach.

Abstract

International audience; The autonomous navigation of an electric vehicle requires the implementation of a number of sensors and actuators intended to inform it about his environment or his position and velocity and deliver necessary inputs. That's why it is important to detect and locate sensor and actuator faults as soon as possible to enable the operator to run the vehicle in degraded mode or use the fault tolerant control system if it exists. The main purpose of this paper deals with sensors or actuators faults diagnosis of autonomous vehicle. A diagnosis method using a nonlinear model of the vehicle is developed. Nonlinear state space model of the autonomous electric vehicle is used with the method of nonlinear analytical redundancy to detect and to isolate faults occurred on sensors or actuators. Computer simulations are carried out to verify the effectiveness of the method.

Abstract

International audience; Increasing vulnerability of transistors and interconnects due to CMOS technology scaling is continuously challenging the reliability of future electronic circuits and systems. Lifetime reliability is gaining attention over performance as a design factor even for lower-end commodity applications. In this paper we propose an effective hybrid fault-tolerant architecture able to deal with permanent and transient faults in combinational parts of pipelined cores. The principle consists in triplicating the combinational logic parts but, unlike TMR, only two copies run in parallel while the third one remains in standby until an error is detected. We have implemented this approach on a MIPS microprocessor as case study. Experiments show that our approach is comparable to TMR in terms of area with a notable power saving and offers a full protection against transient and permanent faults.

Abstract

This paper presents a digital nonlinearity calibration technique for ADCs with strong input-output discontinuities between adjacent codes, such as pipeline, algorithmic, and SAR ADCs with redundancy. In this kind of converter, the ADC transfer function often involves multivalued regions, where conventional integral-nonlinearity (INL)-based calibration methods tend to miscalibrate, negatively affecting the ADC's performance. As a solution to this problem, this paper proposes a novel INL-based calibration which incorporates information from the ADC's internal signals to provide a robust estimation of static nonlinear errors for multivalued ADCs. The method is fully generalizable and can be applied to any existing design as long as there is access to internal digital signals. In pipeline or subranging ADCs, this implies access to partial subcodes before digital correction; for algorithmic or SAR ADCs, conversion bit/bits per cycle are used. As a proof-of-concept demonstrator, the experimental results for a 1.2 V 23 mW 130 nm-CMOS pipeline ADC with a SINAD of 58.4 dBc (in nominal conditions without calibration) is considered. In a stressed situation with 0.95 V of supply, the ADC has SINAD values of 47.8 dBc and 56.1 dBc, respectively, before and after calibration (total power consumption, including the calibration logic, being 15.4 mW). Peer reviewed