Search

The estimation of the height of the centre of gravity of vehicle is significantly important for various applications in modern car management. A theoretical model of a full - scale half car including suspension and a crank axle at the front and the rear axle is developed. A longitudinal model of the vehicle is described to compute the forces at the wheels curing braking and accelerating manoeuvres. Based on nonlinear spring curves and measured deflection at the front and the rear axle the centroidal height of the vehicle is estimated. Some experimental results with a space wagon are shown.

An approach for supervision of vehicle dynamics is presented which may be used for intelligent vehicle control and state monitoring. In particular, an on-board process parameter estimation was implemented which allows one to compute the physical coefficients of lateral vehicle models and their changes during operation. In addition, tire- and velocity-dependent look-up tables in presently used vehicle models were replaced by feedforward neural networks. In the phase of driving state monitoring, a set of these hybrid models-each of them trained for a special driving situation-predict the vehicle motion as a result of the actual steering angle and velocity. In a further step, suitable classification algorithms were used to detect the actual driving state by processing the residual output.

For the fault detection of technical processes different methods can be applied based on the information extracted from direct measured signals, from signal models and process models. Examples for signal model based fault detection methods are spectral analysis or parameter estimation of ARMA models, examples for process model based methods are parameter estimation, state estimation or parity equation approaches. A comparison of these methods shows that they have different properties with regard to the detection of faults in the process, the actuators and the sensors. By a proper integration of different fault detection methods mainly their advantages can be used to generate a number of different analytical symptoms. For fault diagnosis a knowledge based procedure is required, because also qualitative information in form of heuristic symptoms have to be taken into account. Based on heuristic process knowledge as fault-symptom causalities and a unified representation of all symptoms an integrated fault diagnosis can be performed. This comprises the treatment of the symptoms as uncertain facts and approximate diagnostic reasoning via if-then rules either in a probabilistic or a fuzzy-logic (possibilistic) frame. The described methodology was verified by experiments with several technical processes like electric motors, actuators, pumps, machine tools, robots, heat exchangers, combustion engines and vehicles.

The unknown parameters of physical laws which govern the dynamics of processes can be gained from measured input and output signals in a two step procedure. First the parameters of differential equations are estimated and then the physical parameters are calculated based on algebraic relations. After considering the model structures a suitable parameter estimation method for continuous time signals is described and the influence of a priori known parameters on the convergence is shown. Then the determination of physical parameters is considered based on nonlinear algebraic relations known from theoretically modeling. Applying the implicit function theorem identifiability conditions for the unique determination of physical parameters are given. Then applications are shown for the estimation of the ratios of inertia, d.c. motor damping coefficients and Coulomb friction coefficients of an 6-axis-industrial robot and the use of parameter changes for the detection of faults.

On presente une enquete d'introduction sur l'estimation des parametres: on considere d'abord les tâches d'identification des systemes dynamiques et les caracteristiques principales des differentes methodes. Puis, on s'interesse a l'estimation des parametres des systemes discrets ou continus, statiques, multivariables et non lineaires. On insiste sur les algorithmes d'estimation recursive compatibles avec des applications temps reel, en ligne

An identification method is described which first identifies a linear nonparametric model (crosscorrelation function, impulse response) by correlation analysis and then estimates the parameters of a parametric model (discrete transfer function) and also includes a method for the detection of the model order and the time delay. The performance, the computational expense and the overall reliability of this method is compared with five other identification methods. This two-step identification method, which can be applied off-line or on-line, is especially suited to identification by process computers, since it has the properties: Little a priori knowledge about the structure of the process model; very short computation time; small computer storage; no initial values of matrices and parameters are necessary and no divergence is possible for the on-line version. Results of an on-line identification of an industrial process with a process computer are shown.

Methods for on-line process identification in noisy closed loops using correlation and least squares parameter estimation are developed. It is first shown how the problem of correlation between noise and process input caused by the feedback can be solved. Then two modified methods are shown which are applicable also for additional perturbations. Using perturbations, the convergence of the parameter estimates can be improved, so that almost the same accuracy results as for open loop identification. Further modifications are given for changing reference values and low frequent disturbances. The methods are especially suited for process computers because of recursive data processing.

In order to compare and evaluate identification and parameter estimation methods, three simulated linear discrete processes-a second-order oscillatory process, a second-order nonminimum phase process, and a third-order, low pass, process with delay-were identified with the following on-line methods. 1.(1) Least squares 2.(2) Generalized least squares 3.(3) Instrumental variables 4.(4) Stochastic approximation 5.(5) Correlation analysis with least squares parameter estimation 6.(6) Fourier analysis using a model with three unknown parameters. The test processes correspond to the ''multi solution test cases'' we have proposed for the 3rd IFAC-Symposium on Identification and System Parameter Estimation. The variances of parameter estimates and impulse responses are given for simulation runs using all six parameter estimation methods for two noise-to-signal ratios and three measurement time periods. The system order is assumed to be known exactly. Since the ultimate application of the parameter estimates is important in choosing an identification procedure, on example of a final goal, the design of a digital control algorthm, was chosen for these evaluations. Hence the identified models are used to optimize the parameters of a three-mode controller. Both the controller parameters and the r.m.s. error of the closed loop controlled variable are compared for the identified and the exact model.

For on-line identification and parameter estimation of industrial processes with process computers an identification program package was developed. Three appropriate identification methods can be selected: recursive least squares, recursive instrumental variables and recursive correlation analysis with least squares. The program package also includes: signal generation, determination of model order and time delay, data filtering for the elimination of low frequent disturbances, model verification and plotting of intermediate and final results. Practical results and comparisons with the identification package are shown for an industrial size steam-heated heat exchanger.