APPARATUS FOR DETERMINING THE ACTUAL STATE AND/OR THE REMAINING SERVICE LIFE OF STRUCTURAL COMPONENTS OF A WORK MACHINE

Abstract

The present invention relates to an apparatus for determining the actual state and/or the remaining service life of structural components, for example large-diameter rolling bearings, of a work machine, in particular a construction machine, a material-handling machine and/or a conveyor machine, comprising a sensor system for acquiring state information relating to the structural component, and an analytical device for analyzing the acquired state information and determining the actual state and/or the remaining service life on the basis of a comparison with predetermined damage characteristics, wherein an active database device is provided for storing the damage characteristics, to which database device a determination device for determining the damage characteristics from design data of the structural component, and an adjustment device for adjusting the predetermined damage characteristics on the basis of the state and/or the remaining service life information determined by the evaluation device are connected.

Claims

1. An apparatus for determining an actual state and/or a remaining service life of structural components comprising large-diameter rolling bearings of a work machine, wherein the work machine comprises a construction machine, a material-handling machine and/or a conveyor machine, comprising: a sensor system for acquiring state information relating to a structural component; an evaluation device for evaluating acquired state information and determining the actual state and/or the remaining service life by a comparison with predetermined damage characteristics; an active database device for storing the damage characteristics; a determination device for determining the damage characteristics from design data of the structural component; and an adjustment device for adjusting the predetermined damage characteristics on the basis of the state and/or the remaining service life information determined by the evaluation device; wherein the adjustment device and the determination device are connected to the active database device.

2. The apparatus according of claim 1, wherein the determination device is configured to determine kinematic frequencies of the structural component from geometry data of the structural component to generate a damage frequency image for the structural component from the kinematic frequencies.

3. The apparatus of claim 2, wherein the determination device comprises an adjustment module configured to transform frequency patterns corresponding to different damage patterns and/or types into adapted frequency patterns corresponding to different damage patterns and/or types of the specific structural component based on the geometry data of the structural component.

4. The apparatus of claim 3, wherein the determination device comprises a structural analysis module for determining fundamental oscillation of the structural component and comprises a superimposition module for superimposing frequency patterns indicative of different damage patterns and/or types on the determined fundamental oscillation in order to generate synthetically generated damage characteristics adapted to the structural component by the superimposition.

5. The apparatus of claim 4, wherein the evaluation device and/or the adjustment device are configured as a self-learning system and/or as part of a self-learning system which feeds back the sensorily detected state information and/or the actual states and/or remaining service lives derived therefrom to the database and/or integrates them into the damage characteristics stored by the database.

6. The apparatus of claim 5, wherein the self-learning system comprises a regression analysis module for determining the influence of determined damage patterns and/or determined actual states on damage characterizing parameters of the structural component such as structure-borne sound signal reference patterns, roll-over frequency patterns or tooth mesh frequency patterns of the structural component by regression analysis.

7. The apparatus of claim 6, wherein the self-learning system comprises a Kl-based estimation module for estimating correlations between acquired actual state information patterns and synthetically generated damage characteristics and/or between acquired actual state information patterns and a damage pattern or a remaining service life of the structural component.

8. The apparatus according to claim 7, wherein a combination module for combining the synthetically generated damage characteristics and combinatorial damage characteristics is associated with the determination device, and wherein the evaluation device is configured to match the state information acquired by the sensor system with the combinatorial damage characteristics.

9. The apparatus according to claim 1, further comprising a weighting module for weighting the damage characteristics on the basis of an occurrence probability of a damage event corresponding to the damage characteristic, and wherein the weighting module is associated with the determination device.

10. The apparatus according to claim 1, wherein the sensor system comprises at least one sensor from the group of sensors consisting of: oscillation sensors, temperature sensors, lubricant sensors, structure-borne sound sensors, acceleration sensors, displacement sensors and speed sensors; and wherein the adjustment device is configured to adjust the damage characteristics stored by the active database device depending on at least one signal from the at least one sensor.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The invention is explained in more detail below on the basis of a preferred exemplary embodiment and the corresponding drawings. The drawings show:

[0034] FIG. 1: shows a representation of an apparatus for determining the actual state and/or the remaining service life of structural components of a work machine according to an advantageous embodiment of the invention;

[0035] FIG. 2: shows a representation of the apparatus of FIG. 1 with supplementary details in the individual components of the apparatus;

[0036] FIG. 3: shows a representation of the frequency image corresponding to a damage pattern and its transformation to a frequency damage signature specifically adapted to the system of interest; and

[0037] FIG. 4: shows a schematic representation of the synthetic generation of a damage characteristic starting from a fundamental oscillation image generated by structural analysis and its transformation to a specific damage characteristic image.

DETAILED DESCRIPTION

[0038] As shown in the figures, the Condition Monitoring System 1 comprises an active, self-implementing database 2 that stores a plurality of synthetically generated damage characteristics as reference examples 3 in the form of data records indicating various damage patterns of a structural component 4 and containing parameters on which the respective damage pattern is based or which are characteristic of the respective damage pattern.

[0039] As shown in FIG. 2, a reference example 3 designated as Sample I may contain the roll-over frequency of a bearing ring, for example of a bearing outer ring of a rolling bearing, wherein said roll-over frequency may contain said frequency spectrum, for example for different degrees of damage and possibly also for the undamaged state for one or more speeds.

[0040] Another reference example 3, referred to as Sample II, may contain the damage characteristics of a drive module, for example in the form of the gear meshing frequency of a drive shaft.

[0041] Various other reference examples can include, for example, a temperature curve of a bearing over the running time and/or over the time after shutdown, or an acoustic emission spectrum of a rolling bearing, or a vibration pattern of a component or other characteristic damage patterns.

[0042] As shown in the figures, a determination device 5 can be connected to the database 2, which automatically generates said synthetic damage characteristics from design data provided on the respective structural component 4, in particular calculated from geometry and/or drawing and/or material and/or material data. Such a determination device can also estimate the damage characteristics for a particular structural component, if necessary, with the aid of stored data sets for similar structural components, and/or estimate the synthetic damage characteristics for the structural component of interest from known, typical damage characteristics for a particular structural component, for example by interpolation and/or extrapolation based on the geometry data.

[0043] For the synthetic generation of damage characteristics, the kinematics of the specific system of interest can first be calculated. In particular, kinematic frequencies can be calculated for any machine components such as rolling bearings, gears or shafts. Kinematics refers to a description of the system by its geometry and time-varying parameters independent of forces and energies.

[0044] If, for example, a rolling bearing is considered, the kinematics can be determined as follows, wherein, for example, the number of rolling elements z, the rolling element diameter D.sub.W, the pitch circle diameter D.sub.PW under contact angle a can be assumed to be known as geometric data or, if necessary, can be obtained from a design database. In this respect, to determine the kinematics of the rolling bearing there can be carried out the following steps: [0045] Calculating the frequency of rotation of the rolling element

[00001]$.fwdarw.n_W=?\frac{n}{2}.Math.\left(\frac{D_{pw}}{D_W}-\frac{D_W.Math.{\cos }^2?}{D_{pw}}\right)$ [0046] Calculating the roll-over frequency on the inner ring

[00002]$.fwdarw.n_j=\frac{z}{2}.Math.\left(1+\frac{D_W.Math.\cos ?}{D_{pw}}\right)$ [0047] Calculating the roll-over frequency on the outer ring

[00003]$.fwdarw.n_A=\frac{z}{2}.Math.\left(1-\frac{D_W.Math.\cos ?}{D_{pw}}\right)$

[0048] In order to generate a damage pattern, for example for damage to the outer ring of the rolling bearing, the following procedure can be followed, cf. in particular also FIG. 3. [0049] Generating spectra with corresponding damage information [0050] From calculation of the roll-over frequency at the outer ring uA, damage signatures can be generated in the frequency range for bearing outer ring damage, for example (damage patterns are known (from literature) or are available from other measurements.

[0051] A normalized data set can be converted to geometric variants by means of kinematics.

[0052] As a result, therethrough there can be obtained the damage pattern according to the upper part of FIG. 3. [0053] Transforming the damage signature in the frequency domain (f) to the time domain (f) X(jw).fwdarw.x(t) to generate the excitation bursts corresponding to the machine damage characteristic f(uA), cf. FIG. 3, lower part.

[0054] 14 Alternatively, or additionally, the following procedure can be used to create a damage pattern reflecting outer ring damage of the rolling bearing: [0055] Generating spectra with corresponding damage information [0056] From the structural analysis (e.g. from FE simulation) there can be determined the natural oscillation mode of the structure. [0057] .fwdarw.Generating fundamental oscillation f(Sch.sub.Ani), cf. FIG. 4 upper part Superimposition of the machine damage characteristics f(u.sub.A) and the fundamental oscillation f(Sch.sub.Ani) to produce synthetically generated damage characteristics/samples, cf. FIG. 4 lower part.

[0058] Advantageously, by combining the synthetic damage characteristics or samples, the structure-borne sound signature of a system or an interested component can be generated already in the design phase, thus avoiding the time-consuming learning necessary in previous monitoring systems by means of real machine failures and possibly executed using artificial intelligence. Without synthetic generation of a structure-borne sound signature or damage characterization risk based on the design data, at least 7 to 8 real systems corresponding to the system of interest would have to be tested in experiments with known failures. In contrast, the synthetically generated damage characteristics already provide the AI module with more than ? of the expected damage patterns, which significantly reduces the learning time and effort required to refine the system. For example, the remaining damage characteristics that are still missing or not synthetically generated can be fed back by a self-learning system, for example, in the order of 20%.

[0059] In order to reflect more complex damage patterns or wear phenomena, a combination module 6 is advantageously associated with the active, self-implementing database 2, which combines the damage characteristics synthetically determined by the determination device 5 from the design data and thereby generates combinatorial damage characteristics.

[0060] Advantageously, the damage characteristics, for example the synthetically generated damage characteristics and/or the combinatorially determined damage characteristics, can be given a weighting which can be generated by a weighting module 7, in particular on the basis of the probability of occurrence of a respective damage event.

[0061] Said database 2 and/or the modules associated therewith for determining the damage characteristics, i.e. in particular the determination device 5 and/or the combination module 6 and/or the weighting module 7 can be part of a self-learning AI system 8 or be formed by such an AI system 8 which is equipped with artificial intelligence and can estimate or determine a relationship between a determined condition parameter or several condition parameters of a structural component and a damage pattern of the structural component and/or its actual state and/or remaining service life, wherein the AI system can comprise, for example, a regression analysis module in order to adjust said relationship between a parameter or a parameter set and the actual state or the remaining service life of the structural component on the basis of changes that occur.

[0062] As the figures further show, the Condition Monitoring System 1 further comprises a sensor system 9 which may comprise various sensors for measuring or acquiring relevant condition variables or parameters of the structural component 4 of interest, wherein said sensors may be of different types depending on the structural component.

[0063] For example, said sensor system 9 may include a structure-borne sound sensor and/or a displacement sensor and/or a velocity sensor and/or an acceleration sensor and/or a temperature sensor for acquiring corresponding state information on the structural component 4 or surrounding components connected thereto, for example, oscillation data, temperature data, lubricant data, noise emission data, or other relevant state information of the structural component 4.

[0064] Said state information acquired and provided by the sensor system 9 can be evaluated by an evaluation device 10 and compared with the damage characteristics provided by the database 2 to determine the actual state and/or the remaining service life of the structural component 4. As shown in FIG. 2, said evaluation device 10 may comprise an evaluation module 11 which compares real measured damage or condition characteristics with the synthetic damage characteristics from the reference examples 3 or the combinatorial damage characteristics formed therefrom. Upstream of such an evaluation module 11, a pre-analysis and/or processing module 12 can be provided, which processes and/or pre-analyzes the sensory acquired state information, for example by means of a filter or other signal processing modules.

[0065] On the basis of the evaluation of the evaluation device 10, a prognosis and/or trend analysis module 13 can provide a prognosis for the actual state and/or a trend for the actual state of the structural component and/or the entire machine, see FIG. 1 and FIG. 2.

[0066] As shown in the figures, the Condition Monitoring System 1 further comprises an adjustment device 14 that adjusts the damage characteristics stored by the database 2 on the basis of the evaluations of the evaluation device 10 and/or the determined state and/or the remaining service life information.

[0067] Said adjustment device 14 is preferably part of a self-learning AI system or is formed by such an AI system 8, which provides for feeding back real machine condition data and/or integration into the existing reference examples 3 by means of artificial intelligence.

[0068] In particular, the AI system 8 can adjust the synthetic damage characteristics and/or the damage characteristics combinatorially formed therefrom depending on relevant machine condition and/or environmental parameters, in particular, for example, adjust them depending on the age of the structural component 4, the machine and/or operating condition, and changes in environmental influences. This allows the forecast to be kept accurate at all times and the error rate to be minimized.

[0069] The Condition Monitoring System 1 thus makes particular use of a data analysis model based on synthetically generated machine operating characteristics.

[0070] For this purpose, starting from the design of a component/product, synthetic operating characteristics are generated artificially in the form of samples. Each sample corresponds to a specific damage characteristic of a determined component (e.g. the roll-over frequency of bearing outer ring, the gear meshing frequency of a drive shaft, etc.) which can be calculated from geometry/drawing data.

[0071] By combining the specific, synthetic samples there is created a complex picture of all possible, measurable forms of damage. Afterwards, this database can be compared with the measured damage characteristics (e.g. due to structure-borne noise, etc.) and the condition of the component can be evaluated. Due to the large number of variations, the comparison takes place by means of artificial intelligence (AI) or comparable Feature Recognition.

[0072] The samples or reference examples of the damage characteristics, as well as the combination thereof, can be pre-stored in a matching database and associated with the respective component.

[0073] In a further configuration level, each sample can be weighted according to the occurrence probability (frequency of the cause of failure of the respective component).

[0074] Depending on the learning ability of the system, the sample database of damage features can be expanded. Additionally, through the secondary damage analysis, the system can be rewarded to be able to improve the detection rate for related samples (e.g., of other, but similar components).