Instrument Failure Detection Research Articles

Generating Analytical Redundancy for Fault Detection Systems using Flowgraphs

Abstract The paper concerns the analytical redunduncy methods of diagnosis of monitored physical systems. Redundancy relations, describing how plant variables are nominally related to each other, are used to generate residuals. To address the problem of generating the redundancy relations that are unaffected by unmeasurable disturbances, we propose to apply the flow graph description that can reflect the internal structure of a given physical system, as a basis for the derivation of the primary redundancy relation. By reduction of the system's internal variables, a “concise” flow graph, having only the variables of interest (inputs, outputs, and disturbances), may be obtained. A resulting set of connected subgraphs containing only measurable variables provides the equations of the disturbance-decoupled redundancy relations. All the system variables can be identified as control inputs, measured inputs, sub-key variables (inputs and outputs), and key variables (sole outputs). The proposed general design methodology is suitable for instrument failure detection and isolation systems monitoring linear dynamic systems. It may also be useful for a class of nonlinear systems.

Instrument failure detection and reconstruction methodology in space-time nuclear reactor dynamic systems with fixed in-core detectors

The detection and isolation of instrument failures in nuclear reactors equipped with fixed in-core detectors were studied in order to improve reliability and safety. This was done by representing the reactor as a linear stochastic distributed parameter system. A bank of detection observers based on the Kalman filter concept was constructed in order to isolate component failures via robust observation. Each observer was sensitive to only one specified component failure. This was done to minimize the covariance matrix error. However, because observed deviations may be attributed to changes in system behavior, failure decisions are confirmed by a multiple consecutive miscomparison (MCM) counter. Following failure detection, the failed sensor's output is replaced with an estimate from the failure-free filter. Various simulations were performed to verify this failure detection method as applied to reactor instrument failures. It was demonstrated that the method can detect both single- and common-mode failures. Also, without hardware redundancy, it can describe the system dynamics in the event of failures.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A hierarchical multiple model adaptive control of discrete-time stochastic systems for sensor and actuator uncertainties

A hierarchical multiple model adaptive control (MMAC) is described for discrete-time stochastic systems with unknown sensor and actuator parameters, where the decentralized structure consists of a central processor and of m local processors which do not communicate between each other. A major assumption in this study is that the central and any local stations have different knowledge of the hypotheses on the unknown parameters. This leads to a flexible design algorithm for passively adaptive control strategies. Furthermore, the coordinator algorithm in evaluating the global a posteriori probability is relatively simple to implement. The result is applied to the design problem of an instrument failure detection and identification (FDI) system.

Instrument failure detection and estimation methodology for the nuclear power plant

To detect instrument failures in a nuclear power plant, a failure detection and isolation (FDI) method based on the Kalman filter is developed. Each filter is designed to be insensitive to the failed measurements by decreasing the Kalman gain artificially. Since it is mainly dependent upon the dynamic model and averaged outputs, it can exactly indicate the direction of failures. Even though this concept minimizes the number of filters, it performs the role of analytic redundancy for estimation. As soon as the residual exceeds the predetermined bound, the Kalman filter indicates the possibility of failures. However, since the measurement may show false indication owing to abrupt noises, it must be confirmed several times by the multiple consecutive miscomparison counter, which is strongly dependent on measurement history. Then, if it is not in accordance with other measurements, detailed information on the status of the measurements is provided to help the operator's decision. Various simulations were performed to verify and validate the FDI logic in detecting steam generator and pressurizer instrument failures. It is shown that the FDI technique can detect not only a single failure but also simultaneous common-mode and sequential multiple failures of several direct redundancies. It can correctly estimate the physical states in real time, and the remaining time may be used for control with signal validation.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Distributed Instrument Failure Detection, Estimation and Isolation for Linear Time-Varying Process

In most cases, failure detection methods are based on the use of analytical redundancy. This suppose that a reference model, wich described the normal behaviour of the process, is known. Classical methods are not suitable when the process is a large-scale time-varying system.In this paper we present a distributed failure detection, estimation and isolation scheme wich allows to detect systematic errors (bias, drift) in time-varying process.The proposed method involves two levels :-a local level : each subsystem parameters are determined by use of a bias and drift-free identification method. Multiple hypothesis tests are applied on the prediction error obtained with identified parameters in order to detect failures and estimate their magnitudes.-a global level : each local failure informations are centralized and a structural method is used to isolate instrument failures.An example is indued to illustrate the presented method.

An Application of Nonlinear Filtering to Instrument Failure Detection in a Pressurized Water Reactor

The application of a model-based nonlinear filtering algorithm to the problem of instrument failure detection in a pressurized water reactor is considered. Detailed simulation results are presented to illustrate the performance and robustness properties of the detection algorithms developed.

Development of a Combined Algorithm of On-Line Instrument Failure Detection with an Improved Generalized Likelihood Ratio Method and Suboptimal Control on a PWR Pressurizer

In nuclear power plants, instrument failures must be detected and compensated for before a significant decrease in plant performance occurs. The improved generalized likelihood ratio method to detect and identify a failure and a control algorithm in the event of an instrument failure is developed. The failure detection algorithm and the control algorithm are combined and applied to a pressurized water reactor pressurizer. Computer simulation tests show that on-line prompt failure detection and compensation through control action can be performed very successfully even when there is an instrument failure.

Adaptive Kalman Gain Approach to On-Line Instrument Failure Detection with Improved GLR Method and Suboptimal Control on Loft Pressurizer

A method to obtain Kalman gain independent of process noise and initial state estimate error covariances during system operation is developed through the adaptive control concept in case that output measurement matrix is non-singular square matrix. A failure detection scheme of the conventional GLR method is modified to overcome much time consumption in the failure identification. Therefore it is possible to accomplish the control purpose in the state of transients in real time. The computer simulation tests show that the on-line instrument failure detection and suboptimal control can be performed very successfully.

On-line failure detection in nuclear power plant instrumentation

The functional redundancy approach to detecting instrument failures in nuclear power plant instrumentation is described and evaluated. This real-time method uses a bank of Kalman filters (one for each instrument) to generate optimal estimates of the plant state. By performing consistency checks between the output of each filter, failed instruments can be identified. The technique is used to design an instrument failure detection system for a pressurized water reactor (PWR) pressurizer. Actual pressurizer data are used to demonstrate the capabilities of the functional redundancy methods.

Instrument Failure Detection in Partially Observable Systems

Instrument failure detection using the dedicated observer scheme (DOS) depends on partial state observability through each instrument which is monitored. For instrument fault detection by the DOS technique, a quantitative measure of partial state observability is established for each instrument and used to determine a necessary condition on the output structure of the system. This measure, called the internal redundancy of the instrument, indicates the complexity of the logic required for failure detection, and it also indicates where some hardware redundancy can be introduced into the system to improve the fault detection capability of the DOS. The principles developed are applied to a simulation of the pitch axis autopilot of the A7 jet aircraft.

Application of Observers to Instrument Failure Detection

This paper discusses the application of Luenberger observers to instrument failure detection (IFD). Several methods to generate functional redundancy developed in recent years are reviewed. The effect of process parameter variations is considered. With the aid of sensitivity analysis a motivation for Clark's simplified IFD scheme is given. Clark's dedicated observer scheme (DOS) is modified by application of robust observer design (R0S). The results of a simulation study of instrument failure detection by ROS in a nuclear power plant are presented.

Sensitivity Discriminating Observer Design for Instrument Failure Detection

A procedure is presented for the design of a pair of sensitivity discriminating Luenberger observers for instrument failure detection (IFD). The proposed IFD scheme permits detection of instrument malfunctions with reduced sensitivity to process parameter variations. The efficiency of detecting instrument malfunctions and the influence of parameter variations are illustrated with an example.

Evaluation of Parity Equations for Gyro Failure Detection and Isolation

A set of seven unique parity equations (linear dependence relationships) are derived for an octahedron inertial measurement unit (IMU) having six skew-redundant sensing axes. Single and dual axis gyro failure detection and isolation (FDI) algorithms, utilizing the parity equations, are evaluated for a typical Earth launch to geosynchronous orbit mission. The IMU implemented with single degree-ot-freedoni gyros had better FDI characteristics but was relegated in favor of the more cost effective IMU with two degree-of-freedom gyros. A life cycle cost analysis showed that to be more cost effective the latter IMU was required to yield an FDI probability greater than 0.98 with false alarm rate less than 0.001. This capability was verified by Monte Carlo simulation. HE requirement for high reliability and low life-cycle cost in a production program involving many vehicles makes the selection of a skew-redundant strapdown inertial measurement unit (IMU) an attractive choice. Gimbaled orthogonal platform systems which provide single-point failure protection require triple modular redundancy (TMR) implementation with a significant penalty in size, weight, power, and production cost. The decision to go skewed is complicated somewhat by the need to provide a highly effective instrument failure detection and isolation (FDI) algorithm. A TMR system can simply vote each set of three independent redundant axes and thereby avoid the need for special algorithms for failure detection and isolation. Soft failures are, by definition, undetectable with built-in test equipment (BITE). They gradually degrade IMU performance and, if undetected, may cause end-of-mission (EOM) accuracy specifications to be violated. The effectiveness of an FDI algorithm is measured by how likely it is to detect and correctly isolate (cover) a soft failure before it has a chance to jeopardize mission success. At the same time the algorithm must not have a high probability for false alarm (i.e., inadvertently causing the removal of a good instrument). Several different types of FDI algorithms have been developed and/or evaluated by numerous authors.J'3 One of the most attractive, from the point of view of ease of implementation in an airborne digital computer (ADC), is the parity equation algorithm. This paper presents results of an analysis of parity equation efficacy for two strapped-down skewed IMU's. A. Description of Skewed IMU's Figure 1 illustrates the IMU type discussed in the paper. This is the so-called octahedron configuration. The body frame is shown as XNB, YNB, ZNB with the instrument input axes labeled numerically 1 through 6. Each axis contains a gyro and an accelerometer. Complimentary axes in the octahedron IMU (i.e., 1 and 2, 3 and 4, and 5 and 6) make angles of 90 deg with each other and are symmetrically placed with respect to the body system. That is to say, instruments 1 and 2 both are inclined 45 deg with respect to the ZNB body axis, 3 and 4, 45 deg with respect