Model-based Diagnostics Research Articles

Special Issue on Model-Based Diagnostics

The six papers in this special issue cover different approaches and different applications to model-based diagnostics.

Exhaust System Performance Diagnostic

The paper describes the results of the GM Powertrain study of the Exhaust System Performance Diagnostic (ESPD) that detects the malfunctions of primary and secondary oxygen sensors and catalytic converter. It was achieved by creating diagnostic intrusive actions (DIA), also known as an active diagnosis, which separate component detection from normal coordinated torque control (CTC) action, which otherwise would influence the component outputs. The ESPD algorithm contains three consecutive stages when different fuel to air equivalence ratio (EQR) are commanded, and it is plausibly integrated with CTC architecture to minimize negative driveability and emission impacts during diagnostic intrusive actions. The proposed ESPD test sequence, i.e. deceleration fuel cut-off (DFCO) immediately followed by fuel enrichment ensures the lowest NOx emission impact. In addition, combining three separate intrusive component diagnostics in one test sequence minimizes the overall system intrusiveness, as well. 3.4 model-based diagnostics

Developing computational model-based diagnostics to analyse clinical chemistry data

This article provides methodological and technical considerations to researchers starting to develop computational model-based diagnostics using clinical chemistry data. These models are of increasing importance, since novel metabolomics and proteomics measuring technologies are able to produce large amounts of data that are difficult to interpret at first sight, but have high diagnostic potential. Computational models aid interpretation and make the data accessible for clinical diagnosis. We discuss the issues that a modeller has to take into account during the design, construction and evaluation phases of model development. We use the example of Particle Profiler development, a model-based diagnostic tool for lipoprotein disorders, as a case study, to illustrate our considerations. The case study also offers techniques for efficient model formulation, model calculation, workflow structuring and quality control.

New Methods for Adaptation to Degeneration in Process Models for Process Industries

The implementation of model-based control and diagnostics suffer strongly from the fact that models deteriorate as a function of process and sensor deterioration. Also, changes in the raw material (i.e. wood) may occur and often the process control is not addressing these variations in reality. It is thus vital for the model system to be robust in the sense that it is transparent and easy for the operator to maintain. Robustness is essential in many parts of the system, including measurement, process model validation, the ability of the model to adapt to changes in the process, optimization algorithms, and of course the model itself. In this paper, we first show three real-life applications of the utilization of models for diagnostics and control. Thereafter conditions for on-line adaptation of the models are discussed. The challenges when designing such a system are in achieving operator confidence, filtering of misleading measured data, adaptation of process parameters when the process parameters change, and combining validation of measurements and process models. These challenges are met by using a combination of physical and statistical models and methods based on them such as model predictive control (MPC) and parameter estimation. The model should be maintained by a qualified engineer who should be able to explain the system to the operator so that it is understood and confidence can be maintained.

On-line diagnostics in the Mobile Spatial coordinate Measuring System (MScMS)

Abstract Mobile Spatial coordinate Measuring System (MScMS) is a wireless-sensor-network based system developed at the industrial metrology and quality engineering laboratory of DISPEA – Politecnico di Torino. It has been designed to perform simple and rapid indoor dimensional measurements of large-size volumes (large-scale metrology). It is made up of three basic parts: a “constellation” of wireless devices (Crickets), a mobile probe, and a PC to store and elaborate data. Crickets and mobile probe utilize ultrasound (US) transceivers in order to evaluate mutual distances. The system makes it possible to calculate the position – in terms of spatial coordinates – of the object points “touched” by the probe. Acquired data are then available for different types of elaboration (determination of distances, curves or surfaces of measured objects). In order to protect the system against causes of error such as, for example, US signal diffraction and reflection, external uncontrolled US sources (key jingling, neon blinking, etc.), or software non-acceptable solutions, MScMS implements some statistical tests for on-line diagnostics. Three of them are deeply analyzed in this paper: “energy model-based diagnostics”, “distance model-based diagnostics”, and “sensor physical diagnostics”. For each measurement, if all these tests are satisfied at once, the measured result may be considered acceptable with a specific confidence coefficient. Otherwise, the measurement is rejected. After a general description of the MScMS, the paper focuses on the description of these three on-line diagnostic tools. Some preliminary results achieved by the system prototype are also presented and discussed.

Autoregressive model-based diagnostics for gears and bearings

The autoregressive (AR) model-based fault diagnosis approach employs an AR model of the gear signal as a linear prediction filter. For gear diagnosis, the synchronous signal average is processed by the AR filter with the localised gear fault information being contained in the prediction error signal (residual signal). For bearing diagnosis, the synchronously resampled (not averaged) signal in the angle domain is processed by the AR filter. The non-synchronous bearing fault signal should remain in the residual signal. This paper presents the theoretical background of the AR modelling method and the approaches to gear and bearing diagnosis using the AR modelling method. Application examples of detecting localised gear and bearing faults, such as a gear tooth crack and a bearing raceway spall, are also presented. This approach can be readily applied to fault diagnosis for complex mechanical systems such as helicopter main transmission gearboxes and turbine engines.

Internet Traffic Measurement: A Critical Study of Wavelet Analysis

In communication networks, traffic measurements serve two main purposes: the characterization of traffic-load patterns and the monitoring of performances. This paper is a study of the applicability of traffic-analysis methods as a means of detecting malfunctions and performance changes in packet data networks through measurements of selected parameters. The main contribution is of a methodological nature and is motivated by the fact that wavelet analysis, which has come to be regarded as a standard approach, is by no means a straightforward method. Controversial results can be found that are not always attributable to the complex nature of the measured object. Careful study of the uncertainty of traffic-parameter estimates is also an important issue, which turns out to be somewhat neglected in the literature. The aim of this paper is twofold: on the one hand, to enhance the model validation process and, on the other hand, to provide for objective assessment of the feasibility of network-monitoring procedures that rely on measurement and model-based diagnostics

On case-based control of dynamic industrial processes with the use of fuzzy representation

Model-based diagnostics and control of industrial processes has become even more popular nowadays. To this end, a model of the given (dynamic) process is needed. However, there are several objects involved in processes, which may both cause severe problems when they are subject to a model, thus making a model-based approach inapplicable. An example of such an object is an electric furnace for reduction of copper from slag. Research focused on creating an analytical model of this object was unsuccessful. Therefore, the authors proposed a new approach based on case-based reasoning (CBR). Knowledge among the system is carried by selected realizations of the dynamic non-stationary process. Signals and attributes measured in initial state of the process, during its course and at some control points, are represented using a special process-oriented method that allows reducing unnecessary information. A special fuzzy similarity measure was introduced for efficient search of resembling examples. Properties of both the system of data representation and similarity measure are controlled by means of multiple parameters. To assure sufficient performance of the CBR system, an evolutionary algorithm was efficiently used. The paper ends with some exemplary results obtained by means of software, which implements the developed methodology.

Probabilistic Argumentation Systems and Abduction

Probabilistic argumentation systems are based on assumption-based reasoning for obtaining arguments supporting hypotheses and on probability theory to compute probabilities of supports. Assumption-based reasoning is closely related to hypothetical reasoning or inference through theory formation. The latter approach has well known relations to abduction and default reasoning. In this paper assumption-based reasoning, as an alternative to theory formation aiming at a different goal, will be presented and its use for abduction and model-based diagnostics will be explained. Assumption-based reasoning is well suited for defining a probability structure on top of it. On the base of the relationships between assumption-based reasoning on the one hand and abduction on the other hand, the added value introduced by probability into model based diagnostics will be discussed. Furthermore, the concepts of complete and partial models are introduced with the goal to study the quality of inference procedures. In particular this will be used to compare abductive to possible explanations.

Diagnostics and a qualitative model

First generation expert systems were using shallow knowledge based on heuristic information to solve a diagnostic problem. This approach has many disadvantages, which can be avoided by using deep knowledge. Diagnostic reasoning based on deep knowledge is called model-based diagnostics. Recently, the use of qualitative modeling in relation to deep knowledge in expert systems has become increasingly important. The main purpose of our contribution is to present the model-based diagnostic approach at a formal level. The originality of the presented formalization is the concept of the diagnostic space, the characterization of the minimal diagnoses, and the measurement. The formalization serves as the theoretical background to prove our view to the design of qualitative system models and to establish the diagnostic architecture called DISY. The qualitative system model in our diagnostic approach needs not to be specially adopted for use in the diagnostic domain. The only requirement is that it must simulate the system behavior expressed by normal or abnormal functioning of its components. Proposed DISY architecture is not complex and simply takes into an account the previous diagnostic result to obtain a new one from the additional observation-measurement (medical tests or examinations) of the system.

Using propositional logic to compute probabilities in multistate systems

In this paper we start with presentation of a general language for representing subsets of a cartesian product of finite sets. This language is used to represent the set of diagnoses in the general theory of model-based diagnosis presented by Reiter (R. Reiter, Artif. Intell. 32 (1987) 57–95) when the components have more than two possible operating modes. After having established some general results about Boolean algebras, which turn out to be the appropriate mathematical structure to define the language precisely, they are applied in the special case of propositional logic and product spaces, thereby defining a language for the description of events in product spaces. Then we present three different symbloic methods for computing the probability of a formula in the language without explicitly constructing the corresponding system states. The first two methods are based on the algorithm of Abraham (J.A. Abraham, IEEE Transactions on Reliablity 28 (1979) 58–61) whereas the last method is based on the algorithm of Bertschy-Monney (R. Bertschy, P.A. Monney, J. Comput. Appl. Math. 76 (1996) 55–76). All these methods transform the original formula into an equivalent formula for which it is very simple to compute the probability. The problem of computing the probability of a logical formula appears for example in model-based diagnostics when we need to compute the conditional probability of a diagnosis given the observations made on the system.

Model-based diagnostics and probabilistic assumption-based reasoning

The mathematical foundations of model-based diagnostics or diagnosis from first principles have been laid by Reiter (1987). In this paper we extend Reiter's ideas of model-based diagnostics by introducing probabilities into Reiter's framework. This is done in a mathematically sound and precise way which allows one to compute the posterior probability that a certain component is not working correctly given some observations of the system. A straightforward computation of these probabilities is not efficient and in this paper we propose a new method to solve this problem. Our method is logic-based and borrows ideas from assumption-based reasoning and ATMS. We show how it is possible to determine arguments in favor of the hypothesis that a certain group of components is not working correctly. These arguments represent the symbolic or qualitative aspect of the diagnosis process. Then they are used to derive a quantitative or numerical aspect represented by the posterior probabilities. Using two new theorems about the relation between Reiter's notion of conflict and our notion of argument, we prove that our so-called degree of support is nothing but the posterior probability that we are looking for. Furthermore, a model where each component may have more than two different operating modes is discussed and a new algorithm to compute posterior probabilities in this case is presented.

Turbocharged Diesel Engine Modeling for Nonlinear Engine Control and State Estimation

Engine models that are used for nonlinear diesel engine control, state estimation, and model-based diagnostics are presented in this paper. By collecting, modifying, and adding to current available engine modeling techniques, two diesel engine models, a mean torque production model and a cylinder-by-cylinder model, are summarized for use in the formulation of control and state observation algorithms. In the cylinder-by-cylinder model, a time-varying crankshaft inertia model is added to a cylinder pressure generator to simulate engine speed variations due to discrete combustion events. Fuel injection timing and duration are control inputs while varying engine speed, cylinder pressure, and indicated torque are outputs from simulation. These diesel engine models can be used as engine simulators and to design diesel engine controllers and observers.

Steady-State Optimal State and Input Observer for Discrete Stochastic Systems

Model-based machine diagnostics techniques require the modeled states and machine inputs to be measured. Because measurement of all the states and inputs is not always possible or practical, a simultaneous state and input observer is required. Previous work has developed this type of acausal observer and shown it is susceptible to noise. This paper develops a steady-state optimal observer that minimizes the trace of the steady-state error covariance of the state and input estimates for discrete, linear, time-invariant, stochastic systems with unknown inputs. In addition, a method to distinguish the best measurement set among the available measurement sets is developed. Results from numerical simulations show that the optimal observer can greatly improve estimation results in some cases.

Influential observations in survey sampling

If surveys are analysed using the randomisation distribution then model-based influence diagnostics have no role. Individual values can be influential in randomisation inference either when they are included in the sample or when they are not in the sample. In the former case diagnostics are useful, in the latter only prior knowledge can save the inference. Post-stratification unifies a variety of ad hoc procedures. Some ideas deducible from the influence function provide a justification for the Hajek ratio estimator which is an alternative to the Horvitz-Thompson estimator.