CN-121981553-A - Port equipment multidimensional intelligent health assessment method based on damage mechanism and machine learning

CN121981553ACN 121981553 ACN121981553 ACN 121981553ACN-121981553-A

Abstract

The invention discloses a multi-dimensional intelligent health assessment method for port equipment based on a damage mechanism and machine learning, and relates to the technical field of port equipment health assessment. The method comprises the steps of constructing a harbour equipment salt spray-temperature change-dynamic load coupling multi-physical field simulation model through finite elements, quantifying stress distribution, fatigue hot spots and damage evolution paths, generating a dynamic interactive intelligent damage map, extracting performance degradation characteristics based on the intelligent damage map, dividing equipment health states through a spectral clustering algorithm, identifying key risk indexes through a Transformer time sequence network, outputting safety, reliability and performance attenuation degree multidimensional scores based on a Bayesian network, and dynamically updating an evaluation threshold according to harbour throughput and environmental change. The invention greatly improves the accuracy of the health assessment of the equipment, improves the fault early warning time, reduces the maintenance cost, supports the whole life cycle health management of the port equipment, and provides a core decision basis for predictive maintenance.

Inventors

LI XIANRUI
LIU JIAN
LIU LEILEI
XU BIN
LI BINGSHUAI
ZHAO HAOXU
Zhou Sanbo
WANG ZHE
ZHANG LEI
LI ZHANGYUN

Assignees

交通运输部天津水运工程科学研究所

Dates

Publication Date: 20260505
Application Date: 20260203

Claims (7)

1. A multi-dimensional intelligent health assessment method for port equipment based on damage mechanism and machine learning is characterized by comprising the following steps: S1, constructing a harbour equipment salt spray-temperature change-dynamic load coupling multi-physical-field simulation model through finite element analysis, quantifying stress distribution, fatigue hot spots and damage evolution paths, and generating a dynamic interactable intelligent damage map; s2, extracting performance degradation characteristics based on an intelligent damage map, dividing the health state of the equipment into four levels of normal, prompt, important and urgent by using a spectral clustering algorithm, and identifying key risk indexes through a transducer timing network; And S3, inputting real-time sensor data, health grade, key risk indexes and damage mechanism priori knowledge based on a Bayesian network, outputting safety, reliability and performance attenuation degree multidimensional scores, and dynamically updating an evaluation threshold according to port throughput and environmental change.
2. The harbour site multi-dimensional intelligent health assessment method based on damage mechanism and machine learning according to claim 1, wherein, The specific contents of the salt spray-temperature change-dynamic load coupling multi-physical field simulation model constructed in the S1 are as follows: establishing a three-dimensional geometric model of port equipment, and setting material properties, boundary conditions and load working conditions; The stress field, the temperature field and the corrosion damage field are calculated through finite element analysis by considering the coupling effect of salt spray corrosion environment, temperature cyclic variation and dynamic load, and the method specifically comprises the following steps: stress field balance equation: Wherein, the As a function of the stress tensor, As a vector of the volume force, In order to achieve a material density of the material, As the acceleration vector, the acceleration vector is calculated, Is a divergence operator; Temperature field heat conduction equation: Wherein, the Is the specific heat capacity of the material, In order to be able to determine the temperature, In order to be of a thermal conductivity coefficient, For the intensity of the internal heat source, Is the rate of change of temperature; Salt spray corrosion damage evolution equation: wherein D (t) is the corrosion damage degree at the time t, In order to saturate the degree of damage, In order for the corrosion rate constant to be high, The salt fog concentration is given, and n is the corrosion reaction stage number; and identifying a high stress area, a fatigue hot spot position and a crack initiation and propagation path based on the multi-physical field coupling calculation result.
3. The harbour site multi-dimensional intelligent health assessment method based on damage mechanism and machine learning according to claim 1, wherein, The specific content of the dynamic interactive intelligent damage map generated in the S1 is as follows: presenting a stress cloud picture, fatigue damage distribution and crack evolution process in a three-dimensional visual form; inquiring the stress value, damage degree and residual life prediction of any position by adopting a user interaction mode, wherein the residual life is predicted by the following formula: Wherein, the In order for the remaining life to be sufficient, For the number of fatigue limit cycles, In order to have been subjected to the number of cycles, The service life is designed; And dynamically updating the map state according to the real-time monitoring data, and realizing the visualization of the damage evolution process.
4. The harbour site multi-dimensional intelligent health assessment method based on damage mechanism and machine learning according to claim 1, wherein, The specific content of extracting performance degradation characteristics based on the intelligent damage map in the S2 is as follows: extracting static characteristics, dynamic characteristics and statistical characteristics from the intelligent damage map; carrying out standardization processing on the extracted features to construct a multidimensional feature vector; the normalized formula is: Wherein, the For the normalized characteristic value of the sample, As the original characteristic value of the object is obtained, Is the characteristic average value of the characteristic, Is the standard deviation of the features.
5. The method for intelligent health assessment of harbour site in multiple dimensions based on damage mechanism and machine learning according to claim 1 or 4, wherein, In S2, the specific contents of the four grades of normal, prompt, important and urgent equipment health states are divided by using a spectral clustering algorithm: the similarity matrix W is constructed based on the performance degradation characteristics and is defined as: Wherein, the 、, Feature vectors of the ith sample and the jth sample respectively, wherein sigma is a Gaussian kernel bandwidth parameter; By Laplacian matrix , Performing feature decomposition on L for the degree matrix, and taking feature vectors corresponding to the first k minimum non-zero feature values to form a low-dimensional matrix Y; clustering in a low-dimensional space by using a K-means algorithm, and dividing the health state of the equipment into four grades of normal, prompt, important and urgent; Setting a characteristic threshold range corresponding to each level As a state criterion, the following are satisfied: Normal rating: < Prompting the grade: ≤ < Importance level: ≤ < Emergency grade: ≤T4。
6. The harbour site multi-dimensional intelligent health assessment method based on damage mechanism and machine learning according to claim 1, wherein, The specific content of the key risk index identified through the transducer timing network in the S2 is as follows: Constructing a transducer network comprising an encoder-decoder structure, inputting as a time series of historical monitoring data ; Capturing a long time sequence dependency relationship through a self-attention mechanism, wherein a self-attention calculation formula is as follows: wherein Q, K, V are query, key, value matrix respectively, Is a key vector dimension; outputting predicted values and abnormal fluctuation characteristics of key risk indexes, wherein the key risk indexes comprise stress mutation frequencies Injury acceleration factor And rate of performance degradation The key risk index calculation formula is: Wherein, the As the threshold value for the abrupt stress change, Taking 1 for indicating function and meeting condition, otherwise taking 0; Wherein, D (t) is the accumulated damage degree at the time t, D (t+Deltat) is the accumulated damage degree at the time t+Deltat, deltat is the time interval; Wherein P (t) is the performance parameter value at time t, For the moment of time Is used for the performance parameter values of (a), Is the initial performance parameter value.
7. The harbour site multi-dimensional intelligent health assessment method based on damage mechanism and machine learning according to claim 1, wherein, In the S3, based on the Bayesian network, real-time sensor data, health grade, key risk indexes and damage mechanism priori knowledge are input, and the specific contents of the multi-dimensional scores of safety, reliability and performance attenuation degree are output as follows: constructing a directed acyclic graph comprising sensor nodes, health state nodes, risk index nodes and multidimensional scoring nodes; node conditional probability distribution setting based on damage mechanism priori knowledge , wherein, Is a node Is a parent node set of (a); The posterior probability is updated through real-time data, and a Bayesian formula is followed: outputting probability distribution of safety, reliability and performance attenuation degree, and grading and quantifying formula: Security scoring: Reliability scoring: performance decay degree score: Wherein, the 、 Respectively a safety and reliability threshold value, Is the maximum allowable performance degradation.

Description

Port equipment multidimensional intelligent health assessment method based on damage mechanism and machine learning Technical Field The invention relates to the technical field of port equipment health assessment, in particular to a multi-dimensional intelligent port equipment health assessment method based on a damage mechanism and machine learning. Background The harbour equipment (such as a shore bridge and a field bridge) is in complex working conditions of high salt fog, large temperature difference, dynamic load and the like for a long time, and the damage evolution has the characteristics of multi-factor coupling and nonlinear accumulation. Traditional health assessment methods rely on static thresholds or single sensor data, have difficulty accurately capturing trends in device performance degradation, and lack collaborative assessment of multi-dimensional health status (e.g., safety, reliability, performance decay). In the prior art, although the evaluation model based on data driving can process a large amount of monitoring data, the constraint of a physical mechanism is often ignored, so that the generalization capability is insufficient under extreme working conditions or small sample scenes. Therefore, a multi-dimensional intelligent health assessment method for port equipment based on damage mechanism and machine learning is provided to solve the difficulty existing in the prior art, which is a problem to be solved by those skilled in the art. Disclosure of Invention In view of the above, the invention provides a multi-dimensional intelligent health assessment method for port equipment based on a damage mechanism and machine learning, which greatly improves the accuracy of equipment health assessment, improves the fault early warning time, reduces the maintenance cost, supports the whole life cycle health management of port equipment, and provides a core decision basis for predictive maintenance. In order to achieve the above object, the present invention provides the following technical solutions: A harbour site multidimensional intelligent health assessment method based on damage mechanism and machine learning comprises the following steps: S1, constructing a harbour equipment salt spray-temperature change-dynamic load coupling multi-physical-field simulation model through finite element analysis, quantifying stress distribution, fatigue hot spots and damage evolution paths, and generating a dynamic interactable intelligent damage map; s2, extracting performance degradation characteristics based on an intelligent damage map, dividing the health state of the equipment into four levels of normal, prompt, important and urgent by using a spectral clustering algorithm, and identifying key risk indexes through a transducer timing network; And S3, inputting real-time sensor data, health grade, key risk indexes and damage mechanism priori knowledge based on a Bayesian network, outputting safety, reliability and performance attenuation degree multidimensional scores, and dynamically updating an evaluation threshold according to port throughput and environmental change. Optionally, the specific contents of the salt spray-temperature change-dynamic load coupling multi-physical field simulation model constructed in the S1 are as follows: establishing a three-dimensional geometric model of port equipment, and setting material properties, boundary conditions and load working conditions; The stress field, the temperature field and the corrosion damage field are calculated through finite element analysis by considering the coupling effect of salt spray corrosion environment, temperature cyclic variation and dynamic load, and the method specifically comprises the following steps: stress field balance equation: Wherein, the As a function of the stress tensor,As a vector of the volume force,In order to achieve a material density of the material,As the acceleration vector, the acceleration vector is calculated,Is a divergence operator; Temperature field heat conduction equation: Wherein, the Is the specific heat capacity of the material,In order to be able to determine the temperature,In order to be of a thermal conductivity coefficient,For the intensity of the internal heat source,Is the rate of change of temperature; Salt spray corrosion damage evolution equation: wherein D (t) is the corrosion damage degree at the time t, In order to saturate the degree of damage,In order for the corrosion rate constant to be high,The salt fog concentration is given, and n is the corrosion reaction stage number; and identifying a high stress area, a fatigue hot spot position and a crack initiation and propagation path based on the multi-physical field coupling calculation result. Optionally, the specific content of generating the dynamic interactive intelligent damage map in S1 is: presenting a stress cloud picture, fatigue damage distribution and crack evolution process in a three-dimensional visual form; inquiring the st