CN-121997735-A - Marine moon pool resonance inhibition method based on multi-objective collaborative optimization

CN121997735ACN 121997735 ACN121997735 ACN 121997735ACN-121997735-A

Abstract

The invention discloses a multi-target collaborative optimization-based marine moon pool resonance suppression method which comprises the steps of obtaining a moon pool initial flow field model, determining a propagation delay distribution matrix, extracting local vortex intensity and direction characteristics caused by contour interference from the propagation delay distribution matrix, classifying contributions of different mechanism contours to flow field deformation through an improved random forest algorithm to obtain an interference characteristic vector, constructing a time delay interference combined tensor aiming at the interference characteristic vector and the propagation delay distribution matrix, separating the time delay interference combined tensor to determine a dynamic mapping relation, and compensating the time delay mismatch through a counter-propagation adjustment control instruction sequence to obtain a correction instruction set if time sequence deviation in the dynamic mapping relation exceeds a preset threshold value, so as to obtain an optimized execution scheme, and complete marine moon pool resonance suppression. The invention improves the active identification of the moon pool resonance risk and the inhibition capability of key disturbance.

Inventors

ZHANG SHUNING
ZHANG AI
TENG YAO
HE SHILONG
GONG QINGTAO
HU XIN
LI KANGQIANG
SHEN KECHANG
GUO YANLI
HAN YANQING

Assignees

鲁东大学

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (8)

1. The marine moon pool resonance inhibition method based on multi-objective collaborative optimization is characterized by comprising the following steps of: acquiring flow field disturbance data generated by an executing mechanism and water body motion parameters of a moon pool key area through a sensor array, and filtering noise by adopting real-time signal processing to obtain a moon pool initial flow field model; calculating the water density, temperature and salinity distribution on the path from each executing mechanism to the center of the moon pool according to the moon pool initial flow field model, and determining a propagation delay distribution matrix by adopting Kalman filtering to fuse multi-source data; Extracting local vortex intensity and direction characteristics caused by contour interference from a propagation delay distribution matrix, classifying contributions of different mechanism contours to flow field deformation by improving a random forest algorithm, and obtaining interference characteristic vectors; constructing a delay interference combined tensor aiming at the interference characteristic vector and the propagation delay distribution matrix, and separating the delay interference combined tensor to determine a dynamic mapping relation; If the time sequence deviation in the dynamic mapping relation exceeds a preset threshold value, compensating the time delay mismatch through a counter-propagation adjustment control instruction sequence to obtain a correction instruction set; And after the correction instruction set is acquired, merging the cooperative action logic, and injecting profile compensation factors in multi-execution mechanism parallel scheduling to judge the flow field effect convergence time, so as to obtain an optimized execution scheme and complete the suppression of the marine moon pool resonance.
2. The method for suppressing the resonance of the moon pool at sea based on the multi-objective collaborative optimization according to claim 1, wherein the step of filtering noise by adopting real-time signal processing to obtain an initial flow field model of the moon pool comprises the following steps: synchronously acquiring instantaneous speed and pressure signals of an actuating mechanism during action through a multichannel sensor array to obtain original multidimensional time sequence data; performing real-time band-pass filtering and notch processing on the original multidimensional time sequence data by adopting a finite impulse response filter to obtain denoised multidimensional time sequence data; Calculating the instantaneous speed vector and the pressure gradient at each sensor position according to the denoised multidimensional time sequence data to obtain a flow field vector field at the current moment; the flow field vector field at the current moment and the flow field state predicted at the previous moment are fused through a Kalman filter, so that corrected real-time flow field distribution is obtained; extracting average vorticity and speed pulsation intensity in a moon pool key area according to the corrected real-time flow field distribution to obtain hydrodynamic parameters of the moon pool area; if the hydrodynamic force parameter of the moon pool area exceeds the preset threshold range, triggering an adjusting instruction of the control signal of the executing mechanism to obtain a control quantity for suppressing disturbance; And directly updating driving parameters of the executing mechanism according to the control quantity for suppressing disturbance, and simultaneously taking the corrected real-time flow field distribution as the prior state of the Kalman filter of the next period to obtain a continuously tracked moon pool initial flow field model.
3. The method for suppressing resonance in an offshore moon pool based on multi-objective collaborative optimization according to claim 1, wherein the step of determining a propagation delay distribution matrix by fusing multi-source data using kalman filtering comprises: acquiring original distribution data of water density, temperature and salinity on a propagation path from each executing mechanism to the center of the moon pool through a moon pool initial flow field model, and obtaining a preliminary environmental parameter distribution result; Carrying out consistency calibration on the preliminary environmental parameter distribution result, and determining a calibrated multi-source data set; Extracting characteristic values related to the density, the temperature and the salinity of the water body from the calibrated multi-source data set, and constructing an environment variable matrix on a propagation path through characteristic value analysis to obtain the comprehensive characterization of environment variables; carrying out fusion processing on the comprehensive characterization of the environment variable by using a Kalman filtering method, and calculating a time delay estimated value on a propagation path to obtain preliminary distribution of propagation time delay; According to the preliminary distribution of propagation delay, combining the position relation of the moon pool center and the executing mechanism, carrying out spatial correction on the delay estimation value, judging the deviation of the delay distribution, and if the deviation exceeds a preset threshold value, carrying out iterative adjustment on the delay value to obtain corrected delay distribution; and constructing a propagation delay distribution matrix through the corrected delay distribution, and performing smoothing treatment on abnormal points in the matrix to obtain a final propagation delay distribution matrix.
4. The multi-objective co-optimization based marine moon pool resonance suppression method according to claim 1, wherein the classifying contributions of different mechanism contours to the deformation of the flow field by improving the random forest algorithm comprises: Each element in the propagation delay distribution matrix is expressed as disturbance propagation delay of a certain discrete path unit of a certain executing mechanism on a path from the executing mechanism to the center of the moon pool; calculating a local propagation velocity increment for each discrete path element of each actuator; defining a local vortex strength index according to the local propagation speed increment and the path unit length; Extracting three-dimensional fluid velocity vectors corresponding to each propagation path unit from the moon pool initial flow field model, and calculating the local vortex main direction angle of the discrete path unit by using the ratio of two orthogonal components of the three-dimensional fluid velocity vectors in a horizontal plane; combining the local vortex intensity index, the local vortex main direction angle, the disturbance propagation delay and the radial normalization distance into a four-dimensional column vector to obtain a local interference feature vector; The moon pool resonance characteristic period set is obtained through experiments and numerical simulation, and resonance sensitive weight of a sample is calculated based on disturbance propagation delay; Introducing an eddy current intensity reference value and an eddy current amplification coefficient according to the resonance sensitive weight to construct a local comprehensive sample weight; multiplying the local comprehensive sample weight by the ratio of the local eddy current intensity reference value to the eddy current intensity reference value to obtain a local resonance risk index; Constructing an improved random forest model, taking a local interference feature vector, a local resonance risk index and a local comprehensive sample weight as training input of each sample, weighting and combining weighted base-ni unrepeace and weighted resonance risk of the samples into a dividing objective function, and optimizing the dividing objective function to give consideration to classification flow field deformation modes and preferential separation of high resonance risk samples so as to realize node division and tree structure generation based on multi-objective collaborative optimization; Outputting a weighted probability contribution vector of each flow field deformation mode to which each sample belongs for each decision tree; and integrating the output results of all the decision trees according to the weighted probability contribution vector to obtain an interference characteristic vector.
5. The multi-objective collaborative optimization-based marine moon pool resonance suppression method according to claim 1, wherein the constructing a time-lapse interferometry joint tensor and determining a dynamic mapping relationship comprises: corresponding all the outline numbers of the actuating mechanisms, the disturbance propagation path numbers and the interference feature dimensions one by one to obtain a three-dimensional joint array; performing low-rank decomposition on the three-dimensional joint array, wherein the decomposition result comprises three factor arrays; Defining a two-dimensional array of dynamic space-time mapping relation based on the three factor arrays; Performing normalization processing on all elements in the two-dimensional array of the dynamic space-time mapping relation, and calculating average dynamic mapping differences between each pair of actuator contours under all disturbance propagation paths on the basis of normalization results; If the average dynamic mapping difference between any pair of actuator contours exceeds a preset contour coupling deviation threshold, marking the current dynamic mapping relation as a high deviation state, triggering a time delay compensation flow, and recording contour-path pairs exceeding the threshold in the dynamic mapping relation as mapping indexes to be corrected.
6. The multi-objective co-optimization based offshore moon pool resonance suppression method according to claim 1, wherein the compensating for the delay mismatch by the counter-propagation adjustment control command sequence results in a correction command set, comprising: Analyzing the concrete expression form of the time sequence deviation through the dynamic mapping relation to obtain a deviation distribution condition; determining abnormal time sequence points according to the deviation distribution condition aiming at the part exceeding the preset threshold value; if an abnormal time sequence point position is detected, judging a specific interval of time delay mismatch by comparing a preset threshold value with an actual deviation value; aiming at a specific interval of time delay mismatch, performing parameter adjustment on a control instruction sequence by adopting a counter propagation method to generate a preliminary correction instruction; extracting key control points from the preliminary correction instruction, and analyzing the matching degree of the key control points and the instruction sequence to obtain an adjusted instruction set; If the adjusted instruction set still has local delay mismatch, the final correction instruction set is determined by fine tuning the local point positions again.
7. The method for suppressing the resonance of the marine moon pool based on the multi-objective collaborative optimization according to claim 1, wherein the step of determining the convergence time of the flow field effect by injecting the profile compensation factor in the multi-actuator parallel scheduling comprises the following steps: According to the correction instruction set, combining the structural characteristics and the control time sequence of the multiple execution mechanisms to construct an initial action coordination matrix; According to the initial action coordination matrix and the dynamic space-time mapping relation two-dimensional array, calculating the total disturbance superposition intensity at each moment, and judging the disturbance effect convergence moment; Constructing a target offset evaluation function based on the disturbance effect convergence moment candidates and the target resonance avoidance window; introducing a contour compensation factor in the construction of an optimal execution scheme, taking a target offset evaluation function as an optimal target, and performing fine adjustment on the control starting moment of each execution mechanism to minimize the target offset; And outputting a final optimized execution scheme, wherein the requirement that the target offset evaluation function is not higher than a preset upper threshold value is met, and the number of the execution mechanism, the control starting moment and the contour compensation factor are in one-to-one correspondence.
8. The multi-objective collaborative optimization-based marine moon pool resonance suppression method according to claim 1, wherein the final optimization execution scheme comprises: The starting moment of control is adjusted to be T 1 , and the contour compensation factor is set to be C 1 ; The actuating mechanism B controls the starting moment to be adjusted to be T 2 , and the contour compensation factor is set to be C 2 ; the actuating mechanism C controls the starting moment to be adjusted to be T 3 , and the contour compensation factor is set to be C 3 ; Wherein, T 1 、T 2 、T 3 is the control time which finally satisfies the convergence condition of the target offset evaluation function, and C 1 、C 2 、C 3 is the corresponding contour compensation factor.

Description

Marine moon pool resonance inhibition method based on multi-objective collaborative optimization Technical Field The invention relates to the technical field of offshore moon pool, in particular to a multi-objective collaborative optimization-based offshore moon pool resonance inhibition method. Background With the continuous development of deep sea operation and ocean engineering technology, the moon pool structure is increasingly widely applied to various offshore platforms and operation equipment, and a moon pool area provides an access channel and an operation window for an underwater mechanical arm, an operation propeller and measurement and control equipment, so that the moon pool structure becomes an important support for high-precision operation in a deep sea environment. The existing moon pool flow field disturbance control method is mainly focused on simplified prediction and local inhibition of single-mechanism flow field disturbance by means of hydrodynamic modeling and passive regulation of a single actuator, and is difficult to solve the problems of complex disturbance and resonance caused by flow field coupling effect when the multiple actuators are operated cooperatively. In a moon pool limited space, the traditional disturbance modeling mode is difficult to accurately represent space-time interference and flow field response caused by simultaneous operation of multiple execution mechanisms, and the conventional flow field measurement and signal processing technology is limited by the data fusion capability in a high-noise environment, so that the flow field disturbance characteristic extraction is insufficient, and a large error exists in the prediction of a flow field evolution path. In addition, in the prior art, a plurality of execution mechanism scheduling strategies mostly adopt fixed or empirical parameters, the collaborative optimization based on real-time feedback is lacked, the time sequence mismatch caused by the fluid time delay and the structural difference between the mechanisms cannot be dynamically compensated, and the active avoidance and the accurate control of the flow field effect in a high-risk operation window are difficult to realize. Aiming at the identification and suppression of the moon pool regional resonance risk, the existing algorithm model ignores a dynamic coupling mechanism of disturbance propagation and local vortex interference, and the traditional machine learning method has limitations in aspects of feature weighting, risk sensitivity and decision interpretability and cannot realize active sample weighting and multi-target node optimization aiming at the resonance amplification risk. Disclosure of Invention The invention aims to provide a multi-objective collaborative optimization-based marine moon pool resonance suppression method, which improves the active recognition of the moon pool resonance risk and the suppression capability of key disturbance. According to the embodiment of the invention, the marine moon pool resonance suppression method based on multi-objective collaborative optimization comprises the following steps: acquiring flow field disturbance data generated by an executing mechanism and water body motion parameters of a moon pool key area through a sensor array, and filtering noise by adopting real-time signal processing to obtain a moon pool initial flow field model; calculating the water density, temperature and salinity distribution on the path from each executing mechanism to the center of the moon pool according to the moon pool initial flow field model, and determining a propagation delay distribution matrix by adopting Kalman filtering to fuse multi-source data; Extracting local vortex intensity and direction characteristics caused by contour interference from a propagation delay distribution matrix, classifying contributions of different mechanism contours to flow field deformation by improving a random forest algorithm, and obtaining interference characteristic vectors; constructing a delay interference combined tensor aiming at the interference characteristic vector and the propagation delay distribution matrix, and separating the delay interference combined tensor to determine a dynamic mapping relation; If the time sequence deviation in the dynamic mapping relation exceeds a preset threshold value, compensating the time delay mismatch through a counter-propagation adjustment control instruction sequence to obtain a correction instruction set; And after the correction instruction set is acquired, merging the cooperative action logic, and injecting profile compensation factors in multi-execution mechanism parallel scheduling to judge the flow field effect convergence time, so as to obtain an optimized execution scheme and complete the suppression of the marine moon pool resonance. Optionally, the filtering noise by adopting real-time signal processing to obtain the moon pool initial flow field model includes: synchronously acqu