Method for calibrating a balancing machine

Abstract

A method is described for calibrating a balancing machine, in which a rotor (1) to be corrected is rotatably mounted in bearings (2) and a correction run k is performed, wherein at least one measuring sensor (3) determines an initial vibration of the rotor (1) prior to an imbalance correction and transmits this vibration to an evaluation device (4), which stores the measured value as vibration vector {right arrow over (s)}.sub.0. After an imbalance on the rotor (1) is corrected, a residual vibration of the rotor (1) is measured by measuring sensor (3), transmitted to the evaluation unit (4) and stored as vibration vector {right arrow over (s)}.sub.1. The difference {right arrow over (s)}={right arrow over (s)}.sub.1{right arrow over (s)}.sub.0 formed from the measurement data and the corrected imbalance is stored for compensation run k as {right arrow over (s)}.sub.k and {right arrow over (u)}.sub.k by the evaluation unit (4). To undertake a calibration of the machine, one can either determine a process calibration matrix K by solving the equation system S=UK.sup.T using the collected measurement data or one can select an already available process calibration matrix using the initial vibration {right arrow over (s)}.sub.0 and/or imbalance vector {right arrow over (u)} and store it as a calibration matrix in the evaluation unit (4) and use it to calculate an unknown imbalance vector of a rotor (1).

Claims

1. A method for calibrating a balancing machine in a correction run k for rotors manufactured in a series process, in which a rotor to be corrected is rotatably mounted in bearings and accelerated by means of a drive unit, comprising the following steps: a. in which at least one measuring sensor determines an initial vibration of the rotor prior to the imbalance correction and transmits it to an evaluation device, which stores the measured value as vibration vector {right arrow over (s)}.sub.0, b. in which an imbalance on the rotor is corrected in a correction step, wherein a currently available process calibration matrix K is used to calculate the imbalance, c. in which a residual vibration of the rotor is measured by the measuring sensor, is transmitted to the evaluation unit and is stored as vibration vector {right arrow over (s)}.sub.1, d. in which the difference {right arrow over (s)}={right arrow over (s)}.sub.1{right arrow over (s)}.sub.0 and the corrected imbalance for the correction run k is stored as {right arrow over (s)}.sub.k and {right arrow over (u)}.sub.k by the evaluation unit and are used to improve the process calibration matrix K or in which an already available process calibration matrix is selected using the initial vibration {right arrow over (s)}.sub.0 and/or imbalance vector {right arrow over (u)}, and e. in which the process calibration matrix according to step d. is stored as a calibration matrix in the evaluation unit and is used to calculate an unknown imbalance vector of a subsequent rotor by solving the equation system S=UK.sup.T, wherein for every balancing step, matrix S contains a line having a transposed vibration vector Sk={right arrow over (s)}kT for step k and that matrix U contains in every line the transposed imbalance vector Uk={right arrow over (u)}kT of step k and the process calibration matrix is calculated by transposing K.sup.T.

2. The method according to claim 1, wherein steps a. to d. are repeated with n-rotors for calculating process calibration matrix K.

3. The method according to claim 1, wherein the equation system S=UK.sup.T is solved according to the linear correction calculation procedure.

4. The method according to claim 1, wherein already balanced rotors are categorized into clusters according to their initial vibration {right arrow over (s)}.sub.0 and/or their imbalance vector {right arrow over (u)} and from their initial vibrations and/or imbalance vectors, one calculates a cluster process calibration matrix and one selects a cluster process calibration matrix as the already available process calibration matrix, which comprises rotors whose initial vibrations and/or imbalance vectors are at least similar to the initial vibration and/or imbalance vector of the rotor to be balanced.

5. The method according to claim 4, wherein a cluster process calibration matrix is interpolated from selected cluster process calibration matrices and that matrix is used as the already available process calibration matrix.

6. The method according to claim 1, wherein the calibration is either performed one time for a defined number of rotors or is updated for every run of the balancing machine.

Description

BRIEF DESCRIPTION OF THE DRAWING

(1) An embodiment of the present invention is illustrated in the accompanying drawing and will be described in greater detail in the following. In the drawing,

(2) FIG. 1 shows a rotor mounted in a balancing machine.

DETAILED DESCRIPTION OF THE DRAWINGS

(3) FIG. 1 schematically depicts a balancing machine, in which a driven rotor 1 is mounted in bearings 2. Bearings 2 can be soft or rigid bearings. Vibrations of rotor 1 can be calculated from measurement values of a measuring sensor 3 (a vibration sensor, for example). The recorded measurement values are supplied to an evaluation device 4. Additional sensors, such as a rotational speed and angle sensors for example, may be provided. A measuring sensor 3 designed as a vibration sensor can measure vibrations, on bearing stands for example, directly on the rotor or on a housing. The position of measuring sensor 3 is not critical when executing this procedure. The number of sensors can also be adapted to the application in each case.

(4) A drive of the imbalance-measuring device accelerates rotor 1 to a corresponding rotational speed, wherein the drive can also be controlled in such a manner that fixed rotational speeds for balancing can be specified. The measurement values to be recorded can also be recorded in a normal operating mode, so that for example when starting up rotor 1 given predefinable balancing rotational speeds, the measurement values are recorded and supplied to evaluation device 4. Rotor 1 can also be accelerated to its operating rotational speed.

(5) In the method according to the invention, instead of a calibration run using a special master, a process calibration is performed, in which the calibration occurs during the actual balancing of rotor 1, by using the balancing run, in other words the correction run, as a calibration run. In this sense, the method according to the invention can also be referred to as process calibration. The calibration can be used for subsequent rotors. The rotor used for the calibration can benefit from the process calibration when measuring the imbalance after the correction.

(6) In the method according to the invention, the vibration vector is measured for a correction step prior to the correction using a suitable measuring sensor 3 and stored as {right arrow over (s)}.sub.0 by evaluation device 4. The imbalance is calculated using the currently valid calibration of the balancing machine, i.e., using the calibration, which is already stored in the balancing machine. At the start of production, there are as yet no prior balancing steps that can be used for calculating the calibration. The first calibration and consequently calculating the imbalance can take place using a master calibration. Alternatively, test balancing processes can be undertaken without knowing the actual imbalance. All additional calibrations are performed using the method according to the invention.

(7) Calculated imbalance {right arrow over (u)} is corrected on rotor 1 and the vibration is measured again as a residual vibration and stored as {right arrow over (s)}.sub.1. The difference {right arrow over (s)}={right arrow over (s)}.sub.1{right arrow over (s)}.sub.0 and the corrected imbalance for correction run k is stored as {right arrow over (s)}.sub.k and {right arrow over (u)}.sub.k, wherein a correction run or correction step may contain either the same or a new rotor. To calculate process calibration matrix K, one solves the equation system S=UK.sup.T. For every balancing step, matrix S hereby contains a line with the transposed vibration vector of step k where Sk={right arrow over (s)}.sub.k.sup.T. Correspondingly, matrix U contains in every line the transposed imbalance vector of a step U.sub.k={right arrow over (u)}.sub.k.sup.T.

(8) The equation system contains on the right side a matrix U as well as the matrix of K.sup.T as an unknown and can be solved individually for every column S and K.sup.T. Since the equation system generally has more lines than columns (more balance steps or rotors 1 are used than imbalanced planes), it is over-determined and can be solved using an optimisation procedure, such as the linear correction calculation method. The process calibration matrix is then determined by transposing K.sup.T.

(9) Even though in a normal calibration using a rotor 1, P+1 calibration runs are required for P planes, the method according to the invention also works using one correction step per rotor 1, regardless of the number of planes. This is one advantage of the method because in series processes, a short cycle rate is important and thus there is often the requirement to have only one correction step. It has been shown that using the method according to the invention, a short cycle rate can be achieved.

(10) In the method, the process calibration matrix is used to calculate the imbalance utilising the influence coefficient procedure. In this context, it is pointed out that the influence coefficient procedure is often used with shaft-elastic rotors. However, the influence coefficient procedure can be understood to be a general procedure that is used when balancing rigid as well as shaft-elastic rotors. In regard to rigid rotors, the size of the influence coefficient matrix, which is also referred to as process calibration matrix within the meaning of the invention, is simply smaller and only a fixed rotational speed is used. Alternatively, another optimisation procedure can be used to calculate the imbalance.

(11) Since a large number of rotors 1 are used in the process calibration, one obtains good results for measurement values, despite a high error count. Scattering errors are decreased when solving the over-determined equation system and only the systematic transmission behaviour of rotor 1 is determined. This is a major advantage compared to using a single master rotor, which due to its individual bearing properties, does not need to be representative for production diversity, so that an exact and error-free calibration cannot be ensured.

(12) Process calibration can either be performed one time for a certain selection of rotors 1 or correction steps, or be updated continuously. Steps a. to d. of the method for determining the measurement data and calculating process calibration matrix K can be repeated using n-rotors. In this type of an updated calibration, a certain number of previous measurement runs are always used (for example, the last 200 rotors). One can hereby ensure that the calibration takes into account slow changes in the process (for example, the wearing of a tool or modifications in the upstream production step). In regard to selecting the steps, additional rules can be defined, for example that only parts are to be used whose input parameters lie within certain limits. One can hereby exclude particularly small correction imbalances for example, whose uncertainty range is larger than their benefit. If the update occurs automatically without any intervention by the operator, one can refer to the process as self-learning.

(13) In the series-related balancing of turbocharger body groups, it has been shown that the vibration behaviour depends in a non-linear manner on the input imbalances, which results in that systematic balancing errors are generated as a function of the magnitude of the input vibrations. Various calibration matrices result when input data of various amplitudes is used. Systematic correction errors increase, particularly for large initial imbalances. In regard to this, it has proven to be advantageous that rotors are organised into clusters according to their initial vibration {right arrow over (s)}.sub.0 and/or their imbalance vector {right arrow over (u)}, and their initial vibrations and/or imbalance vectors form a cluster process calibration matrix, which is used as an already available process calibration matrix. This means that one can use already existing process calibration matrices. By means of the initial vibration in the initial state for example, one decides for which calibration a respective run can be used, i.e., which cluster process calibration matrix is selected or whether the current process calibration matrix would be advantageous. In calculating the imbalance, the stored process calibration matrices are selected using the magnitude of the initial vibrations.

(14) To select the input value, one must calculate a scalar (or multidimensional) selection value from the vibration level. This can be the imbalance quantity for example using an averaged process calibration matrix or the maximum amounts of vibration. To select the process calibration matrix for calculating the imbalance, one can either select the cluster process calibration matrix of the respective cluster or one can interpolate an individual process calibration matrix from the available cluster process calibration matrix.