The invention relates to a method for identifying an acoustic source in a component, in particular for identifying an acoustic source in a component of a power plant.

Such acoustic sources have their origin in or are caused for example by defects of the component that increase its noise level or by an impact of loose parts of the component and/or upon the component. Such a component is for example a component of a nuclear power plant, exemplarily a machine housing or a pressure vessel of a primary circuit of the nuclear power plant, that is difficult to access. Therefore methods that allow identifying a position and cause of an acoustic source from outside or without disassembling parts of the component or power plant are desired .

It is known from the prior art to use several sensors in order to detect an acoustic source respectively its structure-borne noise signal occurring within and propagating through the component at several measuring points. In order to determine the position of the acoustic source and to correlate the acoustic source to a specific event or defect within the component, at best three sensors are necessary. Therefore known methods have the disadvantage that several sensors are required which increases costs and effort of such measurement and surveilling methods. EP 0 279 184 Al for example discloses a method for detecting the position of acoustic sources respectively the position of a so-called "burst" with a structure-borne noise signal monitoring system, wherein the frequency range of a measured signal is further analyzed by mathematical methods in order to determine the position of the acoustic source with a reduced amount of struc- ture-borne noise sensors. However, said method can only be applied with sufficient reliability for plat-shaped components and several structure-borne noise sensors are required further on. On this basis, it is an object of this invention to provide an improved method for identifying acoustic sources in components, wherein technical costs and time requirement can be reduced. The object of the present invention is achieved by a method for identifying an acoustic source in a component, in particular a component of a power plant according to claim 1.

The method comprises several steps, wherein in a first step (step a)) at least one simulated structure-borne noise signature of a simulated acoustic source is determined for at least one observation point by use of a simulation model of the component. Said determined simulated structure-borne noise signature is at least related to a position of the simulated acoustic source, which is known from the simulation model as the acoustic source is simulated for said position.

In other words : For determining a simulated respectively theoretical structure- borne noise signature, with applying an acoustic source having an excitation point at any position within the component, more precise within the model of component, a simulation by use of a mathematical model is performed. A structure-borne noise signal that is caused by the simulated acoustic source is calculated for an observation point, being also positioned at any position within the component respectively its model. Therefore, the simulated structure- borne noise signature has to be understood as a structure-borne noise signal that is related to and therefore is characteristic for at least a position within the component at which the structure-borne noise signal had its origin. At each position of the component, a temporal course and form of the structure- borne noise can be determined or calculated with the aid of the simulation model. By use of a simulation model of the component, acoustic sources at different positions and the propagation of their structure-borne noise through the component, thus structure-borne noises, can be simulated for components with arbitrary structure, thus independent from a geometrical structure of the component. The propagation of the structure-borne noises and its characteris- tic depends on different physical correlations for example reflection or absorption that can be simulated and modeled by the simulation model.

In a second step (step b)) at least one real structure-borne noise signature of a real acoustic source is detected, in particular with at least one structure- borne noise sensor, at at least one measuring point. The at least one measuring point corresponds to the at least one observation point, for which at least one simulated structure-borne noise signature has been determined before. Such a real acoustic source might for example be caused by a loose part of the component during the operation of the power plant. Therefore, the real structure-borne noise signature has to be understood as a structure-borne noise signal that is measured at a defined measuring point, wherein said measuring point matches with an observation point that has been used in step a) for determining a simulated structure-borne noise signature.

In a third step (step c)) the real structure-borne noise signature detected at said measuring point is compared to or correlated with the at least one simulated structure-borne noise signature simulated for the corresponding observation point.

Identification of at least a position respectively point of origin of the real acoustic source in the component is afterwards carried out based on the result of the comparison between the real structure-borne noise signature measured at measuring point and the simulated structure-borne noise signature simulat- ed for the corresponding observation point, thus the comparison carried out in step c). Comparing the real structure-borne noise signature detected at a measuring point with the simulated structure-borne noise signature simulated for the corresponding observation point is for example carried out by correlation of the signatures or by a self-learning system.

In other words : A simulation model of the component is used for calculating at least one or several simulated structure-borne noise signatures, that means the form and temporal course of structure-borne noise signals that are re- ceived at at least one observation point by simulating acoustic sources at at least one position within the component. Subsequently, said simulated structure-borne noise signature or simulated structure-borne noise signatures is or are used as a reference signature or reference signatures in order to identify acoustic sources by comparison of a real and measured structure-borne noise signature with the simulated respectively calculated structure-borne noise signatures. Due to this approach, the invention has the advantage that the amount of sensors that are required in order to identify an acoustic source, in particular to identify the position and the kind of an acoustic source, for ex- ample caused by loose components or loose parts of components, can be - depending on the form and geometrical structure of the component - significantly reduced without deteriorating the quality or preciseness of identification of the acoustic source. On the other hand, once the amount of sensors remains the same, the quality or preciseness of identification can be improved.

In summary, the idea of the invention is therefore to use calculated and/or theoretical values for different acoustic sources, namely corresponding simulated structure-borne noise signatures calculated for at least one observation point and to compare real detected structure-borne sound signals to said simulated structure-borne noise signatures that have been derived from the simulation model in order to identify acoustic sources, thus to determine at least the position of a real acoustic source within the component. Thereby, the method has the advantage that the amount of sensors that are required in order to detect or measure structure-borne noise signals in a component in order to identify an acoustic source can be significantly reduced. Therefore said method can also be carried out for components having so far only too few sensors for previously known methods.

The expression "identifying" has first of all to be understood in such a way that at least a position of a real acoustic source is determined . Additionally a classification of the real acoustic source is carried out according to a preferred embodiment of the invention. Therefore, in step a), different excitation functions are applied within the simulation model, respectively different excitation functions are simulated, to determine at least one simulated structure-borne noise signature of the simulated acoustic source by use of the simulation model that is further related to a specific kind of acoustic source. Using different excitation functions allows a more specific identification of real acoustic sources having a same point of origin, as the simulated structure-borne noise signatures also considers further characteristics of a structure-borne noise signal, for example different masses or different materials having for example different elasticities. In other words : Different excitation functions are used to simulated different kinds of acoustic sources that have the same point of origin but another cause so that they differ further in their form and/or temporal course. According to this embodiment, not only the point of origin of the acoustic source but also the cause, for example which loosen part has hit the surface or is at least partially damaged, can be identified by comparison of real and simulated structure-borne noise signals.

The method can be used for identifying acoustic sources in any components, in particular in component of a power plant, particularly preferred a component of nuclear power plants, conventional power plants or wind power plants. In case of performing the method for identifying acoustic source in a component of a nuclear power plant, the component is in particular a component of the primary circuit of the nuclear power plant.

In a further preferred embodiment, several simulated structure-borne noise signatures each corresponding to a simulated acoustic source, in particular to a position and/or a specific kind of an acoustic source, are determined at the at least one observation point. This allows a more precise identification of a real detected acoustic source, as the real structure-borne noise signature is compared to each of said several simulated structure-borne noise signatures and the closest matching can be determined in order to identify the acoustic source.

According to a further preferred embodiment at least one, in particular several simulated structure-borne noise signatures are determined for several obser- vation points. This might be useful, if several measuring points are present within the component, so that the simulation model comprises simulated structure-borne noise signatures for several observation points, each corresponding to one of said measuring points. This even allows a more precise identification of an acoustic source.

In order to cover the component nearly completely for being able to identify acoustic sources at any position, it is advantageous if the positions of several simulated acoustic sources are distributed over the entire component, in par- ticular a respective distance between the positions of the several simulated acoustic sources is chosen depending on a curvature of the component in order to ensure a best possible covering of the entire component. For example the respective distance between two positions of said several simulated acoustic sources is increased with increasing curvature of the component and/or said several simulated acoustic sources are positioned with equidistant distances from each other within an area having a constant curvature.

The simulated structure-borne noise signature of a simulated acoustic source and corresponding position and/or kind of the simulated acoustic sources and/or corresponding observation point are preferably stored as reference signatures, for example within a database or an artificial neural network being linked with the simulation model and an evaluation unit.

A further advantageous embodiment includes that real structure-borne noise signatures of a real acoustic source are used for adapting or extending the simulation model, so that identification is improved from time to time.

The invention will be described in further detail hereinafter with reference to the drawings being illustrated in a schematic manner, and in which

FIG 1 shows a flow-chart comprising the steps of a method for identifying an acoustic source in a component, FIG 2 shows a component respectively a simulation model of the component,

FIG 3 shows several simulated structure-borne noise signals KSi,j and a measured structure-borne noise signal KR at an observation point BSi,

Fig . 1 shows a flow-chart comprising the steps of a method according to the invention for identifying an acoustic source in a component 2. Said component 2 is for example a component of a nuclear power plant, for example a component of a primary circuit like a pressure vessel or any tube. For convenience only, the component 2 respectively a model of said component 2 is illustrated in Fig . 2 as a two-dimensional arbitrary form.

In a first step (step a)), a simulation model S of the component 2 that simulates the structure and form of said component 2 is provided and several simu- lated structure-borne noise signatures KS of simulated acoustic sources are determined for several observation points, exemplary a first observation point BSi and a second observation point BS ₂ , by use of said simulation model S. According to Fig . 2, five acoustic sources having different excitations points respectively positions ASi, AS ₂ , AS ₃ , AS ₄ , AS ₅ are simulated within the simula- tion model. The positions ASi, AS ₂ , AS ₃ , AS ₄ , AS ₅ of said simulated acoustic sources are distributed over the component 2, whereby the positions ASi, AS ₂ , AS ₃ of corresponding acoustic sources are exemplary positioned with distances d to each other in an area of the component 2 having an increased curvature. In this example, five simulated structure-borne noise signature KSi , KS _2/J , KS ₃ ,j, KS ₄ ,j, KS ₅ ,j are calculated for each observation point BSi, BS ₂ , each of said simulated structure-borne noise signature KSij, KS _2/J , KS _3/J , KS _4/J , KS _5/J corresponding to a position ASi, AS ₂ , AS ₃ , AS ₄ , AS ₅ of a simulated acoustic source.

In order to further identify, thus to classificate a specific kind Fj of acoustic source, e.g. its cause, each of the simulated acoustic sources is caused by different excitation function F _j that might consider different masses or material characteristics of the acoustic source. Therefore the simulated structure-borne noise signature KS of the simulated acoustic sources are additionally related to a specific kind Fj of acoustic source. Two different excitation functions F _J; in the following Fi and F ₂ , would exemplary lead to ten simulated structure borne noise signatures KSi,i, KS ₂ ,i, KS ₃ ,i, KS ₄ ,i, KS ₅ ,i, KSi, ₂ , KS _2/2 , KS _3/2 , KS _4/2 , KS _5/2 , for each observation point BSi, BS ₂ .

The simulated structure-borne noise signatures KSi,i, KS ₂ ,i, KS ₃ ,i, KS ₄ ,i, KS ₅ ,i, KSi, ₂ , KS _2/2 , KS _3/2 , KS _4/2 , KS _5/2 , for each of both observation points BSi, BS ₂ and corresponding position ASi, AS ₂ , AS ₃ , AS ₄ , AS ₅ and kind Fj respectively Fi and F ₂ of the simulated acoustic sources are stored exemplarily in a database being linked to the simulation model S and an evaluation unit as reference signatures. For convenience only, different positions AS,, but only one kind of acoustic source, thus characterized by excitation functions Fj will be considered for further explanation in the following. During the lifetime or during operation of a nuclear power plant, a loose part of the component 2 or the component 2 itself might act as an acoustic source and therefore cause structure-borne noise waves that propagate through the component 2. In the example shown in Fig. 2, sensors are positioned at two measuring points, a first measuring point BRi and a second measuring point BR ₂ , wherein the measuring points BRi, BR ₂ correspond to the observation points BSi, BS ₂ for which simulated structure-borne noise signatures KS have been determined in the previous step by use of the simulation model S. Due to a real acoustic source caused for example by a hit of the surface of a loose part, a first real structure-borne noise signature KRi is detected at the first measuring point BRi and a second real structure-borne noise signature KR ₂ is detected at the second measuring point BR ₂ (step b)).

In a third step (step c)) the real detected structure-borne noise signatures KRi detected at measuring point BRi is compared with the simulated structure- borne noise signatures KSij, KS _2/J , KS _3/J , KS _4/J , KS ₅ ,j simulated for the corresponding observation point BSi. The real detected structure-borne noise signatures KR ₂ detected at measuring point BR ₂ is compared with the simulated structure-borne noise signatures KSij, KS ₂ ,j, KS ₃ ,j, KS ₄ ,j, KS _5/J simulated for the corresponding observation point BS ₂ .

Fig . 3 shows exemplary five structure-borne noise signatures KSi , KS _2/J , KS _3/J , KS ₄ ,j, KS ₅ ,j having a certain temporal course and form simulated for observation point BSi (dashed lines) for the component 2 and the real structure-borne noise signature KRi having a certain temporal course and form that has actually been detected at measuring point BRi (solid line). As can be seen in Fig. 3, a comparison of the simulated structure-borne noise signatures KSi , KS _2/J , KS ₃ ,j, KS ₄ ,j, KS ₅ ,j and the real structure-borne noise signature KRi shows that the temporal course and the form of the real structure-borne noise signature KRi matches essentially with that of simulated and calculated structure-borne noise signature KSij. This allows an identification of the point of origin, thus the position of the real acoustic source in the component 2 and its cause, for example which part of the component 2 has caused the acoustic source. As the simulated structure- borne noise signature KSi,j is associated with an acoustic source at position ARi and a specific kind F _J; one can determine that the real source that has caused the real structure-borne noise signature KRi corresponds to an acoustic source being near to position ARi and having characteristics of specific kind Fj.

In order to improve the simulation model S and therefore an identification of acoustic sources, the real structure-borne noise signatures KR, of a real acous- tic source are preferably used for adapting or extending the simulation model S. References

2 component

KSi simulated structure-borne noise signature

KRi measured structure-borne noise signature

AS, simulated acoustic source

AR| real acoustic source

BS _k observation point in the simulation model

BR _k real observation point

F _j excitation function

S simulation model