Shafting-machinery-hull system coupling dynamic load online identification method and forecasting system

Abstract

The invention relates to a shaft system-machine-hull system coupling dynamic load online identification method and a forecasting system. The method comprises the following steps: S1, determining the positions, directions and the number of action points of dynamic loads acted by all force sources in a propulsion shaft system and a mechanical system; s2, determining the positions, the directions and the number of measuring points of dynamic response required by dynamic load identification; s3, constructing a cross-point frequency response function matrix between the dynamic load and the dynamic response, and performing de-noising processing on elements in the frequency response function matrix; s4, determining the position, direction and number of working condition dynamic response phase decomposition reference measuring points; s5, forming an auto-power spectrum and cross-power spectrum combined matrix by reference measurement inching response; and S6, constructing a phase reference matrix of dynamic response and the like. The invention provides a shaft system-machine-ship body system coupling dynamic load online identification method and a forecasting system aiming at the defects in the existing production technology, and through the method, a high-precision identification model and a quasi-real-time forecasting system which can consider multi-source coupling and multi-directional coupling are established.

Description

Technical Field

[0001] The invention relates to the technical field of load identification, in particular to an online identification method and a prediction system for coupled dynamic loads of a shafting-machinery-hull system. Background Art

[0002] When ships, submarines, UUVs and other navigation platforms are sailing in the water, the propulsion shaft system is subject to the propeller hydrodynamic excitation force and the main engine output coupling vibration excitation force, as well as the shaft system self-rotation will form misalignment and unbalanced excitation force, etc., which will be coupled with the dynamic excitation force formed by the main engine, auxiliary engine and other mechanical equipment and pipeline system near the shaft system through the hull structure. The coupled dynamic load will be transmitted to the wet vibration radiation surface of the hull through the support or installation structure channel of the propulsion shaft system, main engine and auxiliary engine, and together form underwater radiation noise. The underwater noise they form is mainly broadband and line spectrum components of medium and low frequencies, which greatly affects the acoustic stealth performance of the ship.

[0003] Among them, the broadband driven loads acting on the hull structure through the propulsion shaft system through the supporting bearings, the mechanical equipment through the mounting base, and the pipeline through the horse foot support are not only the key dynamic parameters for calculating and evaluating the structural sound radiation surface vibration and underwater noise, but also the indispensable calculation parameters for calculating the vibration power flow transmitted to the hull. In addition, they are also important characterization parameters for dynamic monitoring and fault diagnosis. They have the characteristics of multi-source (multiple force source locations, multi-point excitation distribution at a single location), multi-directional (multiple transmission path intersections, multiple direction intersections), and mixed coherence (different types of force source coherence components exist in the dynamic response phase) coupling. Therefore, the online real-time and high-precision acquisition, evaluation and consideration of such dynamic loads are not only of great significance, but also of great difficulty.

[0004] Traditional dynamic load measurement is usually divided into two methods: direct measurement and indirect identification:

[0005] (1) Direct measurement method: In a laboratory environment (such as an onshore bench model), dynamic force sensors can be integrated into the base for direct measurement. Related research has been done. However, the force sensor needs to be inserted in series into the structure, which will change the form and safety of the original structure and is not suitable for complex models and real ship structures.

[0006] (2) Indirect measurement method: Through indirect observations such as strain or acceleration, the dynamic load of the force source is obtained based on the inversion of dynamic principles. There are related studies, but they only focus on simple force source models (such as single direction, single type, etc.) and do not consider the effects of complex coupling.

[0007] At the same time, Chinese patent CN202210625118.8, an indirect measurement method of bearing dynamic load considering foundation excitation, also involves this content, but there are still the following deficiencies in this solution:

[0008] ① The patent simplifies the bearing and bearing seat into a spring-damper unit, ignoring the inertial effect caused by the mass and the standing wave effect caused by the actual size of the bearing structure; ② The patent uses the acceleration frequency response data of the connection between the foundation and the bearing base to replace the influence of the foundation, and only considers the origin equivalence of the bearing position, but fails to consider the cross-point path coupling, direction coupling and phase coherent coupling between the foundations of different bearing positions; ③ The acceleration sensor needs to be arranged at the upper and lower interfaces of each support, where the upper interface is a rotating component, which is difficult to implement in actual operation; in addition, the upper and lower interfaces arranged in the patent should theoretically be the two end positions of the spring-damper unit, but the actual situation corresponds to the contact position between the rotating shaft and the bearing, which is difficult to implement in actual operation; ④ The patent only considers the excitation of the shaft system itself, and does not consider the excitation coupling with other mechanical equipment or mechanical systems.

[0009] Another Chinese patent CN202111390635.3, a method for reducing the pathological nature of the dynamic load identification model of rotating machinery bearings, also involves this content, but there are still the following deficiencies in this solution:

[0010] ① The patent proposes to reduce the pathological nature of the identification model in order to respond to the pathological nature caused by the combination of measurement point positions, without considering the pathological nature caused by the coupling of cross-point paths between different excitation directions, different excitation types, and different bearing positions; ② The patent also only considers the excitation of the shaft system itself, without considering the coupling with the excitation of other mechanical equipment or mechanical systems.

[0011] There is also a paper titled "Inverse Identification Method of Dynamic Loads on Propulsion Shafting Bearings Based on Measured Data" that involves the above method, but the paper also has the following shortcomings:

[0012] ① The paper only considers the case of shaft system excitation (single type of external excitation), and fails to consider the case of coupling with other mechanical equipment or mechanical systems and other types of excitations;

[0013] ② The paper only considered the case of using a single-direction response (vertical in the paper) to invert a single-direction excitation force (vertical in the paper), and failed to consider the case of multi-directional excitation forces coupling with each other;

[0014] ③ In this paper, when performing inversion, Tikhonov-GCV regularization is directly performed to reduce the possibility of pathological conditions, and no judgment is made before regularization.

[0015] To this end, we propose an online identification method and prediction system for the coupled dynamic loads of the shafting-machinery-hull system. Summary of the invention

[0016] In view of the shortcomings of the above-mentioned existing production technology, the applicant provides an online identification method and prediction system for the coupled dynamic loads of the shafting-machinery-hull system. Through this method, a high-precision identification model and a quasi-real-time prediction system that can take into account multi-source coupling and multi-directional coupling are established.

[0017] The technical solution adopted by the present invention is as follows:

[0018] A method for online identification of coupled dynamic loads of a shafting-machinery-hull system comprises the following steps:

[0019] S1. Determine the position, direction and number of points of action of dynamic loads from all sources of force in the propulsion shaft system and mechanical system, where dynamic loads include forces and moments;

[0020] S2. Determine the location, direction and number of measuring points of dynamic response required for dynamic load identification;

[0021] S3, constructing a cross-point frequency response function matrix between dynamic load and dynamic response, and performing denoising on the elements in the frequency response function matrix;

[0022] S4. Determine the position, direction and number of reference measuring points for phase decomposition of dynamic response of the working condition;

[0023] S5, the reference measurement point dynamic response forms a combination matrix of auto-power spectrum and cross-power spectrum;

[0024] S6, constructing a phase reference matrix of the dynamic response, and testing and denoising the elements in the matrix;

[0025] S7, separate the independent irrelevant components and decompose the dynamic response into components;

[0026] S8. For the working condition response generated by each independent component excitation obtained by decomposition, combined with the cross-point frequency response function matrix, the virtual dynamic load is obtained by inversion calculation;

[0027] S9, determine and process the ill-posedness in the inversion calculation process;

[0028] S10, performing energy superposition on the dynamic load of each order of decoupling according to the decoupling incoherent characteristics to obtain the actual dynamic load;

[0029] S11. Perform synthetic calculation on the dynamic response of the additional focus position.

[0030] Furthermore, in S1, for the propulsion shaft system, the position where the rotating part contacts or is connected to the non-rotating part is determined as the position where the force source acts on the dynamic load, and the number of discrete equivalent force source application points at each position is determined based on the characteristic size of the contact or connection; for mechanical equipment and pipelines, the position where the equipment foot and pipeline support are in contact or connected to the hull mounting base is determined as the position where the force source acts on the dynamic load, and the number of discrete equivalent force source application points at each position is determined based on the characteristic size of the contact or connection.

[0031] Furthermore, in S2, dynamic response measurement points are arranged near the position of each force source action point, the number of dynamic response measurement points is not less than the number of equivalently discretized force source action points to be identified, and the dynamic characteristics between them should have a certain degree of distinction.

[0032] Furthermore, in S3, a cross-point frequency response function matrix between dynamic loads and dynamic responses is constructed through testing or simulation, and the elements in the frequency response function matrix are denoised using methods such as optimal estimation.

[0033] Furthermore, in S4, multiple dynamic response phase reference signals that can characterize the characteristics of the respective incoherent components are selected according to the force source characteristics, and the total number of reference signals is not less than the number of incoherent types existing in the dynamic response.

[0034] Furthermore, in S7, singular value decomposition and decoupling methods are used to decompose the dynamic response into components to obtain multiple non-correlated independent component signals.

[0035] Furthermore, in S9, the discrete PICARD method is used to determine the ill-posedness of the inversion process. If the problem is ill-posed, regularization methods such as Tikhonov-GCV are used to improve and suppress the ill-posedness.

[0036] Furthermore, in S10, when there are a large number of incoherent components, the important first few orders are intercepted to achieve problem reduction.

[0037] A shafting-machinery-hull system coupled dynamic load prediction system, comprising:

[0038] Sensors for collecting measured dynamic responses required for identification and phase coherence decomposition under various operating conditions;

[0039] A / D conversion module, used to convert analog signals into digital signals;

[0040] Heterogeneous data fusion and synchronous acquisition module, used to unify the format and clock synchronization of signal data of different physical architectures;

[0041] The digital operation module performs fast operation according to the method of the aforementioned steps S5 to S11 on the sensor signals collected in real time and the cross-point frequency response function matrix data obtained a priori;

[0042] Algorithm calibration module, which quickly compares and verifies the calculation results with the calibration signal;

[0043] The post-processing rendering and display module is used to display the recognition results.

[0044] Furthermore, the system further includes a calibration database for storing calibration signals and historical data so that the algorithm calibration module can perform rapid comparison and verification; the system can handle the situation of mixed coherence of dynamic response phases under multi-source and multi-directional excitation, and can separate and solve different coherent types of excitations caused by shaft systems and equipment.

[0045] The beneficial effects of the present invention are as follows:

[0046] The invention has a compact and reasonable structure and is easy to operate. Through this method, a high-precision identification model and a quasi-real-time prediction system are established that can consider multi-source (multiple force source positions, multi-point excitation distribution at a single position) coupling and multi-directional (multiple transmission path intersections, multiple direction intersections) coupling; at the same time, for the excitations of different coherent types caused by the shaft system, equipment, and piping system, the problem of mixed coherence of the dynamic response phase (that is, there are different types of force source coherence components in the dynamic response phase) caused by them can be separated to improve the accuracy; in addition, when solving the inverse problem at each frequency, a pre-criteria is used to determine whether to perform regularization processing to improve the ill-posedness.

[0047] At the same time, the present invention also has the following advantages:

[0048] 1) All multi-source and multi-directional dynamic loads in the complex coupling system of shafting, equipment and hull can be obtained through one calculation;

[0049] 2) The method is theoretically complete and can obtain an exact solution;

[0050] 3) Considered and solved the practical problems such as pathological and mixed problems in the actual measurement operation process;

[0051] 4) The shaft system and equipment will cause excitations of different coherent types. The phase mixed coherence caused by them in the dynamic response (i.e., there are different types of force source coherent components in the dynamic response phase) can be separated and solved to obtain higher precision results;

[0052] 5) At each frequency, the pre-criteria will be used to make a pathological judgment to determine whether regularization inversion processing is required to avoid over-constraint caused by blind regularization or under-constraint caused by non-regularization;

[0053] 6) An online quasi-real-time identification and prediction system has been formed. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a schematic diagram of the excitation of the hull by the propulsion shaft system, main engine, auxiliary engine and pipelines in the present invention.

[0055] Figure 2 This is a principle block diagram of the online identification and prediction system of the present invention.

[0056] Figure 3 It is an implementation flow chart of the present invention.

[0057] Figure 4 This is a comparison diagram of the amplitude of the same signal when using signals with different phase characteristics as references under multi-source excitation in the present invention.

[0058] Figure 5 This is a force and frequency relationship diagram without regularization processing in the present invention.

[0059] Figure 6 This is a force and frequency relationship diagram after regularization processing of the present invention.

[0060] Figure 7 This is a relationship diagram between the force and frequency without considering the directional coupling factor in the present invention.

[0061] Figure 8 The force and frequency relationship diagram under the directional coupling factor is taken into account for the present invention.

[0062] Fig. 9 This is a force and frequency relationship diagram of the present invention without considering the response phase mixing coherence factor caused by different excitations.

[0063] Fig.10 This is a relationship diagram between the force and frequency of the present invention under phase mixing coherent coupling factors caused by different excitations. DETAILED DESCRIPTION

[0064] The specific implementation of the present invention will be described below in conjunction with the accompanying drawings.

[0065] The present invention discloses a dynamic load identification method for a complex dynamic coupling system of a propulsion shaft system, a mechanical system and a hull structure. Through this method, a high-precision identification model and a quasi-real-time prediction system are established, which can consider multi-source (multiple force source positions, multi-point excitation distribution at a single position) coupling and multi-directional (multiple transmission path intersections, multiple direction intersections) coupling; at the same time, for the excitations of different coherent types caused by the shaft system, equipment, and piping system, the problem of mixed coherence of the dynamic response phase (that is, there are different types of force source coherence components in the dynamic response phase) caused by them can be separated to improve the accuracy; in addition, when solving the inverse problem at each frequency, a pre-criteria is used to determine whether to perform regularization processing to improve the ill-posedness.

[0066] Embodiment 1: A method for online identification of coupled dynamic loads of a shafting-machinery-hull system, comprising the following steps:

[0067] S1. Determine the position, direction and number of points of action of dynamic loads from all sources of force in the propulsion shaft system and mechanical system, where dynamic loads include forces and moments;

[0068] S2. Determine the location, direction and number of measuring points of dynamic response required for dynamic load identification;

[0069] S3, constructing a cross-point frequency response function matrix between dynamic load and dynamic response, and performing denoising on the elements in the frequency response function matrix;

[0070] S4. Determine the position, direction and number of reference measuring points for phase decomposition of dynamic response of the working condition;

[0071] S5, the reference measurement point dynamic response forms a combination matrix of auto-power spectrum and cross-power spectrum;

[0072] S6, constructing a phase reference matrix of the dynamic response, and testing and denoising the elements in the matrix;

[0073] S7, separate the independent irrelevant components and decompose the dynamic response into components;

[0074] S8. For the working condition response generated by each independent component excitation obtained by decomposition, combined with the cross-point frequency response function matrix, the virtual dynamic load is obtained by inversion calculation;

[0075] S9, determine and process the ill-posedness in the inversion calculation process;

[0076] S10, performing energy superposition on the dynamic load of each order of decoupling according to the decoupling incoherent characteristics to obtain the actual dynamic load;

[0077] S11. Perform synthetic calculation on the dynamic response of the additional focus position.

[0078] Specific combination of Figure 1-Figure 3, fully explain the specific steps:

[0079] S1 in this embodiment: determining the propulsion shafting-mechanical system force source identification model, that is, the position, direction and number of action points of the dynamic loads acting on the hull structure by all force sources.

[0080] Dynamic loads are generalized forces, that is, they include forces and moments.

[0081] a) For propulsion shafting:

[0082] The position where the rotating part contacts or connects with the non-rotating part (mainly the bearing position) is determined as the position where the force source acts on the dynamic load; for the radial support in the radial bearing and thrust bearing, the vertical, horizontal and torsional directions are determined as the directions of the dynamic load acted by the force source, and the number of discrete equivalent force source action points in each support should be no less than N1 (assuming that the characteristic dimension of the radial bearing is L1, and the commonly used characteristic dimensions are the maximum values ​​of the bearing length l, the inner diameter d, the aspect ratio l / d, etc.; assuming that the bending wave wavelength corresponding to the upper limit of the frequency of concern for the dynamic load at this location is λ1; then N1 is taken as 6L1 / λ1 and rounded up); for the axial support part in the thrust bearing, the bending direction along the axial direction and around the vertical and horizontal directions is determined as the direction of the dynamic load acted by the force source, and the number of discrete equivalent force source action points in each support should be no less than N2 (assuming that the characteristic dimension of the axial support is L2, and the commonly used characteristic dimensions are the maximum values ​​of the circumference and diameter of the axial support; assuming that the longitudinal wave wavelength corresponding to the upper limit of the frequency of concern for the dynamic load at this location is λ2; then N2 is taken as 6L2 / λ2 and rounded up).

[0083] b) For mechanical equipment and pipelines:

[0084] If rigid installation is adopted, the position where the equipment installation foot or pipeline support foot contacts or is connected to the hull installation base is determined as the position where the force source acts on the dynamic load; if elastic installation is adopted, the position where the elastic element connecting the equipment foot or pipeline support foot contacts or is connected to the hull installation base is determined as the position where the force source acts on the dynamic load; the direction of the dynamic load applied by the force source is determined according to the actual situation of the equipment or pipeline support; the number of discrete equivalent force source application points at each support position should be no less than N3 (assuming that the characteristic dimension of the contact or connection position is L3, and the commonly used characteristic dimension is the maximum value of the three-dimensional dimensions of the contact or connection position; assuming that the bending wave wavelength corresponding to the upper limit of the frequency of concern of the dynamic load at this position is λ3; then N3 is taken as 6L3 / λ3 and rounded up).

[0085] In summary, the frequency domain complex signals of dynamic loads (including forces and moments) of all force sources such as propulsion shafting, mechanical equipment, and pipelines that need to be identified are written in matrix form:

[0086]

[0087] Where F is the dynamic load force in the direction of translational freedom, M is the dynamic load moment in the direction of rotational freedom, m=s+t.

[0088] S2: Determine the location, direction and number of measurement points of dynamic response required for dynamic load identification.

[0089] Dynamic response measurement points (such as acceleration dynamic response, strain dynamic response, etc., and mixed arrangement of different types of dynamic response measurement points is also allowed) are arranged near the location of each force source action point; the number of dynamic response measurement points in each degree of freedom direction of the dynamic load should not be less than the number of force source identification points after equivalent discretization to be identified, and the dynamic characteristics between them should have a certain degree of distinction. The location and number of dynamic response measurement points are determined by selection or optimization methods.

[0090] First, the dynamic response time domain signal obtained from the test for inversion identification is written in column matrix form:

[0091]

[0092] Where a is the time domain dynamic response in the direction of translational freedom, s is the time domain dynamic response in the direction of rotational freedom, and n=u+v.

[0093] Then, the complex frequency domain signal of the dynamic response to be determined for inversion identification is expressed in the form of a column matrix:

[0094]

[0095] Where A is the frequency domain complex dynamic response in the direction of translational freedom, S is the frequency domain complex dynamic response in the direction of rotational freedom, and n = u + v.

[0096] Since the working condition dynamic response needs to be obtained by actual measurement, and when performing dynamic load identification, it is necessary to apply the complex dynamic response, that is, it is necessary to obtain the real part and the imaginary part of the dynamic response. For the measured data of the dynamic response, on the one hand, in order to remove the obvious signal noise, on the other hand, in order to reflect the steady-state characteristics in the sense of statistical average and eliminate the uncertainty of truncated initial phase, while considering the ability to perform batch measurement of mobile sensors when the number of sensors is insufficient, the dynamic response cannot be processed by conventional fast Fourier transform, but needs to be processed by the spectral analysis method that needs to consider the reference phase proposed in the subsequent S4~7 of the present invention.

[0097] S3: Construct the cross-point frequency response function matrix H between dynamic load and dynamic response through testing or simulation n×m (f), and use the optimal estimation method to denoise the elements in the frequency response function matrix:

[0098]

[0099] S4: Determine the position, direction and number of reference measuring points for phase decomposition of the dynamic response of the operating condition.

[0100] According to the characteristics of the force source, select multiple dynamic response phase reference signals that can characterize the characteristics of each incoherent component. The position of the reference signal should be selected as close as possible to each excitation source with independent coherent characteristics and with a better signal-to-noise ratio. The total number of selected reference signals should not be less than the number of incoherent types existing in the dynamic response, and the dynamic characteristics between them should have a certain degree of distinction.

[0101] S5: The dynamic response of the reference measurement point forms a combination matrix of the auto-power spectrum and the cross-power spectrum.

[0102] After selecting and measuring the frequency domain signal S4 used as the phase reference, the auto-power spectrum and cross-power spectrum of each reference signal and each other are calculated and combined to form a complex matrix G RR (f):

[0103]

[0104] Where, R(f)=[R1(f)R2(f)…R p (f)] T is the complex matrix composed of the reference signal spectrum, p is the number of reference signals, R p (f) is the complex spectrum of the p-th reference signal at frequency f, the superscript “H” represents the Hermitian transpose and “*” represents the conjugate.

[0105] S6: For different operating conditions, a phase reference matrix of the dynamic response is constructed, and the elements in the matrix are tested and denoised.

[0106] For different operating conditions, the cross-power spectrum complex matrix G between all dynamic response signals in S2 and each reference signal in S4 is calculated. XR (f):

[0107]

[0108] S7: Separate independent and irrelevant components and decompose the dynamic response

[0109] For S5 G RR (f) Perform singular value decomposition and decoupling, and the original complex signal coupling matrix is ​​decomposed into multiple non-correlated independent component signals:

[0110]

[0111] Where U p×p is the unitary matrix obtained by decomposition, Σ p×p =diag(σ1,σ2,…,σP ) is the singular value matrix.

[0112] First, construct the virtual autospectral matrix of the reference response

[0113]

[0114] Then, construct the virtual cross-spectral matrix of the reference response and the measured response for:

[0115]

[0116] Finally, decoupling is performed according to the singular values ​​of the virtual cross-spectrum matrix, and the operating response generated by the excitation of the i-th independent component is obtained as follows:

[0117]

[0118] In the formula, is the virtual cross-spectrum of the e-th measured working condition response and the i-th reference response, is the virtual autospectrum of the ith reference response.

[0119] Depend on The i-th order virtual working condition response matrix after decoupling is:

[0120]

[0121] S8: The operating response generated by the excitation of the i-th independent component decomposed in S7 Combined with the cross-point frequency response function matrix H obtained in S3 n×m (f) Through inversion calculation, the virtual dynamic load caused by the independent component formed by the i-th order decoupling is obtained as follows:

[0122]

[0123] In the formula,

[0124]

[0125] S9: When S8 is inverted, in order to reduce the ill-posedness, the discrete PICARD method is first used to determine the ill-posedness of the inversion process when calculating at each frequency. If the problem is ill-posed, regularization methods such as Tikhonov-GCV are required to improve and suppress the ill-posedness; if the problem is not ill-posed, the generalized inverse method is used to directly perform the inversion operation.

[0126] S10: Perform energy superposition on each dynamic load of each order decoupling obtained by S8 and S9 according to the decoupling incoherent characteristics, and obtain the actual dynamic load:

[0127]

[0128] When there are a large number of incoherent components, in the process of superimposing the independent components in the above formula, the important first few orders can be truncated (usually the orders are truncated according to the retained energy component of not less than 90%) to achieve problem reduction.

[0129] S11: After obtaining each dynamic load, further synthetic calculation is performed on other dynamic responses of additional focus positions (indicated by the subscript "r"). First, according to each virtual dynamic load of each order And their cross-point frequency response functions H at the dynamic response position r (f) Perform amplitude superposition to calculate the virtual dynamic response of each order, and then perform energy superposition according to the incoherent characteristics formed by decoupling to obtain the actual result AS of the dynamic response of interest r (f):

[0130]

[0131] Similarly, when there are a large number of incoherent components, in the process of superimposing the independent components in the above formula, the important first few orders can be truncated (usually the orders are truncated according to the retained energy component of not less than 90%) to achieve problem reduction.

[0132] S4 to S7 in the scheme of the present invention take into account the most complex situation - the situation of mixed coherence of the dynamic response phase under multi-source and multi-directional excitation. The situations of complete coherence and complete incoherence in multi-source excitation can be regarded as special cases of the mixed coherence situation in the scheme of the present invention, and can be processed and solved by the unified method in the scheme of the present invention, with the same effect.

[0133] Embodiment 2: A shafting-machinery-hull system coupled dynamic load prediction system;

[0134] Based on the first embodiment, this embodiment establishes a prediction system for the coupled dynamic load of the shafting-machinery-hull system, and the prediction system includes:

[0135] Sensors for collecting measured dynamic responses required for identification and phase coherence decomposition under various operating conditions;

[0136] A / D conversion module, used to convert analog signals into digital signals;

[0137] Heterogeneous data fusion and synchronous acquisition module, used to unify the format and clock synchronization of signal data of different physical architectures;

[0138] A digital operation module, for performing fast operation on the sensor signals collected in real time and the cross-point frequency response function matrix data obtained a priori according to the method of steps S5 to S11 in claim 1;

[0139] Algorithm calibration module, which quickly compares and verifies the calculation results with the calibration signal;

[0140] The post-processing rendering and display module is used to display the recognition results.

[0141] At the same time, in this embodiment, the prediction system also includes a calibration database for storing calibration signals and historical data so that the algorithm calibration module can perform rapid comparison and verification; the system can handle the situation of mixed coherence of dynamic response phases under multi-source and multi-directional excitation, and can separate and solve different coherent types of excitations caused by shaft systems and equipment.

[0142] Specifically, the method further includes the following step S12: constructing a quasi-real-time online identification and prediction system;

[0143] S3 is the inherent property of the system to be identified, which can be acquired a priori through testing or simulation; S2, S4~S11 can achieve quasi-real-time online identification and prediction by building hardware and software systems (including sensors, A / D conversion modules, heterogeneous data fusion and synchronous acquisition modules, digital operation modules, algorithm calibration modules, post-processing rendering and display modules, etc.).

[0144] Among them, the sensor is used to collect the measured dynamic response required for identification and phase coherent decomposition under various working conditions; the A / D conversion module is used to convert analog signals into digital signals; the heterogeneous data fusion and synchronous acquisition module is used to unify the format and clock synchronization of signal data of different physical architectures (such as voltage signals, charge signals, optical signals, etc.); the digital operation module performs rapid calculations on the real-time collected sensor signals and the a priori cross-point frequency response function matrix data according to the methods of S5 to S11; the algorithm calibration module quickly compares and verifies the calculation results with the calibration signals; the post-processing rendering and reality module is used to display the identification results.

[0145] In summary, the present invention has the following effects:

[0146] 1) All multi-source and multi-directional dynamic loads in the complex coupling system of shafting, equipment and hull can be obtained through one calculation;

[0147] 2) The method is theoretically complete and can obtain an exact solution;

[0148] 3) Considered and solved the practical problems such as pathological and mixed problems in the actual measurement operation process;

[0149] 4) The shaft system and equipment will cause excitations of different coherent types. The phase mixed coherence caused by them in the dynamic response (i.e., there are different types of force source coherent components in the dynamic response phase) can be separated and solved to obtain higher precision results;

[0150] 5) At each frequency, the pre-criteria will be used to make a pathological judgment to determine whether regularization inversion processing is required to avoid over-constraint caused by blind regularization or under-constraint caused by non-regularization;

[0151] 6) An online quasi-real-time identification and prediction system has been formed.

[0152] Experimental example, by comparing with the solution in the background technology, we can know that:

[0153] When there are incoherent or partially coherent dynamic response components caused by multiple sources of excitation, such as dynamic response components caused by other independent excitation source devices other than shaft system excitation, if the autospectral power spectrum or cross-spectral power spectrum is estimated according to the traditional method, the incoherent components between them will be statistically averaged, resulting in the loss of signal components, such as Figure 4 shown.

[0154] During the dynamic load identification process, Figure 5 This is the comparison between the traditional method (without regularization) and the direct measurement results. Figure 6 This is the result after regularization processing using the method of the present application and comparison with the direct measurement result.

[0155] During the dynamic load identification process, Figure 7 This is the comparison between the traditional method (without considering the directional coupling factor) and the direct measurement results. Figure 8 This is the situation in which the method of the present application takes directional coupling factors into consideration and is compared with the direct measurement results.

[0156] During the dynamic load identification process, Fig. 9 This is the comparison between the identification results of the traditional method (without considering the mixed coherence factors of the response phase caused by different excitations) and the direct measurement results. Fig.10 This is the case where the inversion results are compared with the direct measurement results after adopting the method of the present application, that is, taking into account the phase mixing coherent coupling factor.

[0157] The above description is an explanation of the present invention, not a limitation of the present invention. The scope of the present invention is defined in the claims. Any form of modification may be made within the scope of protection of the present invention.

Claims

1. An online identification method for coupled dynamic loads of a shafting-machinery-hull system, characterized in that: The following steps are involved: S1. Determine the position, direction and number of points of action of dynamic loads from all sources of force in the propulsion shaft system and mechanical system, where dynamic loads include forces and moments; S2. Determine the location, direction and number of measuring points of dynamic response required for dynamic load identification; S3, constructing a cross-point frequency response function matrix between dynamic load and dynamic response, and performing denoising on the elements in the frequency response function matrix; S4. Determine the position, direction and number of reference measuring points for phase decomposition of dynamic response of the working condition; S5, the reference measurement point dynamic response forms a combination matrix of auto-power spectrum and cross-power spectrum; S6, constructing a phase reference matrix of the dynamic response, and testing and denoising the elements in the matrix; S7, separate the independent irrelevant components and decompose the dynamic response into components; S8. For the working condition response generated by each independent component excitation obtained by decomposition, combined with the cross-point frequency response function matrix, the virtual dynamic load component is obtained by inverse calculation; S9, determine and process the ill-posedness in the inversion calculation process; S10, performing energy superposition on the dynamic load of each order of decoupling according to the decoupling incoherent characteristics to obtain the actual dynamic load; S11. Perform synthetic calculation on the dynamic response of the additional focus position.

2. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S1, for the propulsion shaft system, the position where the rotating part contacts or connects with the non-rotating part is determined as the position where the force source acts on the dynamic load, and the number of discrete equivalent force source application points at each position is determined based on the characteristic size of the contact or connection; for mechanical equipment and pipelines, the position where the equipment foot, pipeline support and hull mounting base contact or connect is determined as the position where the force source acts on the dynamic load, and the number of discrete equivalent force source application points at each position is determined based on the characteristic size of the contact or connection.

3. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S2, dynamic response measurement points are arranged near the position of each force source action point. The number of dynamic response measurement points is not less than the number of equivalently discretized force source action points to be identified, and the dynamic characteristics between them should have a certain degree of distinction.

4. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S3, a cross-point frequency response function matrix between dynamic loads and dynamic responses is constructed through testing or simulation, and the elements in the frequency response function matrix are denoised using methods such as optimal estimation.

5. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S4, multiple dynamic response phase reference signals that can characterize the characteristics of their respective incoherent components are selected according to the force source characteristics, and the total number of reference signals is not less than the number of incoherent types existing in the dynamic response.

6. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S7, singular value decomposition and decoupling methods are used to decompose the dynamic response into components to obtain multiple non-correlated independent component signals.

7. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S9, the discrete PICARD method is used to determine the ill-posedness of the inversion process. If the problem is ill-posed, regularization methods such as Tikhonov-GCV are used to improve and suppress the ill-posedness.

8. The online identification method of coupled dynamic load of shafting-machinery-hull system according to claim 1 is characterized in that: In S10, when there are a large number of incoherent components, the important first few orders are intercepted to achieve problem reduction.

9. A shafting-machinery-hull system coupled dynamic load prediction system, characterized in that: include: Sensors for collecting measured dynamic responses required for identification and phase coherence decomposition under various operating conditions; A / D conversion module, used to convert analog signals into digital signals; Heterogeneous data fusion and synchronous acquisition module, used to unify the format and clock synchronization of signal data of different physical architectures; A digital operation module, for performing fast operation on the sensor signals collected in real time and the cross-point frequency response function matrix data obtained a priori according to the method of steps S5 to S11 in claim 1; Algorithm calibration module, which quickly compares and verifies the calculation results with the calibration signal; The post-processing rendering and display module is used to display the recognition results.

10. A shafting-machinery-hull system coupled dynamic load prediction system according to claim 9, characterized in that: The system further includes a calibration database for storing calibration signals and historical data so that the algorithm calibration module can perform rapid comparison and verification; the system can handle the situation of mixed coherence of dynamic response phases under multi-source and multi-directional excitation, and can separate and solve different coherent types of excitations caused by shaft systems and equipment.

Images