CN114881342A - System and method for predicting residual life of ship power equipment - Google Patents
System and method for predicting residual life of ship power equipment Download PDFInfo
- Publication number
- CN114881342A CN114881342A CN202210550575.5A CN202210550575A CN114881342A CN 114881342 A CN114881342 A CN 114881342A CN 202210550575 A CN202210550575 A CN 202210550575A CN 114881342 A CN114881342 A CN 114881342A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- residual
- residual life
- power equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for predicting the residual life of ship power equipment, which comprises the following steps: step 1: collecting data of ship power equipment; step 2: forming two residual life prediction samples; and step 3: forming two multi-dimensional gray scale residual life prediction samples; and 4, step 4: substituting the training set into an improved TCN model, and converting multidimensional data into one-dimensional characteristic vectors; and 5: extracting deep level features of the data, and outputting the weight of parameters with large influence by the multi-layer fusion model; step 6: obtaining a residual life prediction model, inputting a test set of a residual life prediction sample into the residual life prediction model after normalization processing, and outputting a health factor; and 7: performing linear regression prediction on the output health factor to obtain the health factor at the current operation moment; and 8: and (4) carrying out real-time residual life prediction on the ship power equipment. According to the invention, the corrosion data of the equipment parts is introduced on the basis of the traditional life prediction data, so that the prediction error is reduced.
Description
Technical Field
The invention relates to the technical field of prediction of residual life of ship equipment, in particular to a system and a method for predicting residual life of ship power equipment.
Background
The ship power equipment is important corollary equipment of the ship, and the equipment distributes in all positions of whole ship, and is in large quantity, wide distribution, but most power equipment still remains simple control or has only realized state monitoring at present. In order to avoid sudden accidents and economic losses caused by sudden failure of power equipment, the residual service life of the equipment needs to be predicted, an emergency treatment scheme is made in advance, and safe and reliable guarantee is provided for normal operation of the equipment. Due to uncertainty of operating conditions of the equipment and complexity of working environments, certain abnormal points and noises inevitably exist in acquired data, and the noises cannot reflect the real state of the equipment and inevitably generate error influence on the final prediction result. Therefore, the original data needs to be denoised, and the error influence of noise on the prediction result is eliminated as much as possible.
In order to solve the problems, in order to guarantee the normal operation of the ship power equipment, the residual life prediction of the ship power equipment needs to be realized in time, in recent years, technologies related to the residual life prediction of the equipment are gradually researched, and particularly, deep learning is gradually introduced into the field of residual life prediction, wherein the technologies include methods based on mechanisms, data and mixed models. For example, the invention of China scientific and technological university, YuanYe, and the like, discloses a method and a system for predicting the residual life of mechanical equipment, wherein a convolutional neural network and a bidirectional gated cyclic unit are combined to form a mixed neural network so as to effectively extract time and spatial characteristics and improve the prediction accuracy of the residual life (Yuan, Huanghong, Li Jia Qi, a method and a system for predicting the residual life of mechanical equipment [ P ]. Chinese patent CN113094822A:2021-07-09), and the invention of Huchang Hua, and the like, which are Chinese people's liberation military and fire arrow military engineering university, discloses a method and a system for predicting the residual life of equipment, and the invention of the depth confidence network is used for realizing the quantitative prediction of the residual life (Huchang Hua, Si Xiaosheng, and the like) on the basis of mass data analysis.
However, these methods have obvious disadvantages, and only predict the remaining life of the device from some aspect of state monitoring or fault diagnosis, and the prediction data has a single source and no universality, and the distribution calculation of the remaining life of the device is not performed, so that the real-time remaining life prediction of the device cannot be performed accurately in time.
Disclosure of Invention
The invention provides a system and a method for predicting the residual life of ship power equipment, which aim to solve the problem that the real-time residual life prediction of the ship power equipment cannot be timely and accurately carried out in the prior art.
The invention provides a system for predicting the residual life of ship power equipment, which comprises: the system comprises power equipment, a data acquisition module, a data processing module, a model fusion module and a residual life prediction module;
the power equipment is connected with the data acquisition module; the data acquisition module is connected with the data processing module; the data acquisition module is respectively connected with the model fusion module and the residual life prediction module; the model fusion module is connected with the residual life prediction module;
the data acquisition module is used for acquiring historical operation cycle data and real-time operation data of the ship power equipment;
the data processing module processes historical operation cycle data and real-time operation data from the data acquisition module, and divides the processed historical operation cycle data and the real-time operation data into a training set and a test set;
the model fusion module is used for performing model training on the data processed by the data processing module and constructing an equipment degradation trend;
the residual life prediction module is used for a fusion model of the model fusion module, importing a test set of the data processing module into the residual life prediction model, and predicting the residual life of the ship power equipment.
The invention also provides a method for predicting the residual life of the ship power equipment, which is suitable for the system for predicting the residual life of the ship power equipment and comprises the following steps:
step 1: collecting historical operation cycle data, real-time operation data, historical cycle component corrosion data and real-time component corrosion data of ship power equipment;
step 2: preprocessing the data acquired in the step 1, providing data with small change degree, and forming two residual life prediction samples by the left data;
and step 3: performing matrix preprocessing on the residual life prediction samples to form two multi-dimensional gray residual life prediction samples;
and 4, step 4: substituting the training set of the residual life prediction sample into an improved TCN model, extracting long-term and short-term time sequence characteristics and degradation information of characteristic data, and converting multidimensional data into one-dimensional characteristic vectors;
and 5: adding the attention module and the soft threshold value in the residual error mode into the DRSN model to form a multilayer fusion model, training the multilayer fusion model on the basis of the step 4, reducing sample time sequence characteristic loss in the multilayer training process through the multilayer fusion model, extracting deep level characteristics of data, and outputting the weight of parameters with large influence by the multilayer fusion model;
step 6: training and optimizing the multilayer fusion model to obtain a residual life prediction model, inputting a test set of residual life prediction samples into the residual life prediction model after normalization processing, and outputting health factors;
and 7: performing linear regression prediction on the output health factor to obtain the health factor at the current operation moment;
and 8: and according to the health factor at the current running time, predicting the residual life of the ship power equipment in real time.
Further, after the step 5 and before the step 6, the method further comprises:
and optimizing the health degradation trend of the ship power equipment in the multilayer fusion model.
Further, the improved TCN model in step 4 includes an input layer, four residual error units, an attention layer, a full connection layer, and an output layer, which are connected in sequence.
Furthermore, the four residual error units are connected in sequence, wherein the first residual error unit and the third residual error unit have the same structure; the second residual error unit and the fourth residual error unit have the same structure.
Further, the first and the third residual error units comprise sequentially connected: one-dimensional extended convolutional layer, PReLU activation function, batch normalization, Dropout layer, convolutional layer, and summing unit;
the second residual error unit and the fourth residual error unit are sequentially connected: one-dimensional extended convolutional layer, PReLU Activate function, batch normalization, Dropout layer, convolutional layer, and summing unit.
Further, the optimizing the health degradation trend of the ship power equipment in the multilayer fusion model specifically comprises: the corrosion rate is added to the health degradation trend.
Further, the health factor is calculated as follows:
wherein HI represents a health factor, t i Representing the time, alpha, at which various abnormal alarms occur in the power plant of the vessel i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j Representing the damage weight of different faults of the ship power equipment to the health state of the ship power equipment, T representing the total running time of the ship power equipment, n being a time sequence, v k Typical is the corrosion rate of the plant parts, α k Representing the weight of the effect of different component corrosion conditions on the life of the plant, L k The representative is the design life of the equipment components, and N is the number of the equipment components.
Further, the calculation formula of the residual service life of the ship power equipment is as follows:
wherein RUL represents the residual service life of the ship power equipment, HI represents the health factor, t i Representing the time, alpha, at which various abnormal alarms occur in the power plant of the vessel i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j The damage weight of different faults of the ship power equipment to the health state of the ship power equipment is represented, T represents the total running time of the ship power equipment, and n is a time sequence.
The invention has the beneficial effects that:
according to the system and the method for predicting the residual life of the ship power equipment, the corrosion data of the equipment parts is introduced on the basis of the traditional life prediction data, so that the prediction error can be reduced. The DRSN and the improved TCN model in the deep learning are introduced into the field of traditional residual life prediction, so that redundant information in data can be effectively removed, the advanced feature learning capability of noisy data is improved, historical monitoring data is effectively utilized, time sequence features of the data are fully mined, information leakage can be prevented through one-dimensional expansion volume, the receptive field of observation data is expanded, and residual life prediction of ship power equipment can be realized in time. Because the corrosion rate is added into the health degradation trend, the health degradation trend of the ship power equipment established by the fusion model is optimized, the calculation accuracy of the health factor can be improved, the residual service life of the ship power equipment can be predicted timely and accurately, a scientific maintenance strategy is provided, the working efficiency of the ship power equipment is improved, and the maintenance cost is saved.
Drawings
The features and advantages of the present invention will be more clearly understood by reference to the accompanying drawings, which are illustrative and not to be construed as limiting the invention in any way, and in which:
FIG. 1 is a flowchart of the overall system of an embodiment of the present invention;
FIG. 2 is a schematic diagram of an improved TCN model according to an embodiment of the present invention;
FIG. 3 is a first, three residual units of an embodiment of the present invention;
FIG. 4 is a second, four residual units of an embodiment of the present invention;
FIG. 5 is a diagram of a DRSN model architecture in accordance with an embodiment of the present invention;
FIG. 6 is a diagram of an attention module in residual mode in accordance with an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An embodiment of the present invention provides a system for predicting a remaining life of a ship power plant, as shown in fig. 1, including: the device comprises power equipment, a data acquisition module, a data processing module, a model fusion module and a residual life prediction module.
The power equipment is connected with the data acquisition module; the data acquisition module is connected with the data processing module; the data acquisition module is respectively connected with the model fusion module and the residual life prediction module; and the model fusion module is connected with the residual life prediction module.
The data acquisition module is used for acquiring historical operation period data and real-time operation data of the ship power equipment, and comprises a historical operation period data unit and a real-time data unit;
the data processing module processes historical operation cycle data and real-time operation data from the data acquisition module and divides the processed historical operation cycle data and the real-time operation data into a training set and a test set, wherein the data processing module comprises a data preprocessing unit, a gray scale residual life prediction sample unit, a training set unit, a test set unit and a normalization unit.
The model fusion module is used for performing model training on the data processed by the data processing module and constructing equipment degradation trend, wherein the model fusion module comprises an improved TCN model unit, a DRSN model unit, an attention unit in a residual error mode, a soft threshold unit and an equipment degradation trend unit;
the residual life prediction module is used for a fusion model of the model fusion module, importing a test set of the data processing module into the residual life prediction model, and predicting the residual life of the ship power equipment, wherein the residual life prediction module comprises a residual life prediction model unit, a linear regression unit and a residual life unit.
The specific embodiment of the invention also provides a residual life prediction method of the residual life prediction system of the ship power equipment;
the residual life prediction method comprises the following steps:
step 1: the data acquisition module acquires historical operation cycle data and real-time operation data of the ship power equipment; detecting ship power equipment by using an industrial endoscope, acquiring historical period component corrosion data, and acquiring component corrosion data of an operation period in real time by using a corrosion monitor;
particularly, under the condition of the deterioration of the economic environment in the world, the shipowner generally reduces the load and the speed of the engine to reduce the fuel consumption, and in addition, the cylinder sleeve of the engine host is easy to be eroded by seawater, and the long-term partial load operation of the host inevitably generates poor combustion and carbon deposition in a combustion chamber and an exhaust system. The daily oil consumption of the main engine is reduced, so that the exhaust gas corrosion, the low-temperature corrosion and the like are aggravated, further, key parts such as an engine cylinder, a combustion chamber and the like are seriously abraded, and faults such as continuous mechanical cylinder knocking and the like in the main engine cylinder occur. Under low load conditions, corrosive wear tends to cause greater damage to the engine than mechanical wear.
However, the existing system and method for predicting the residual life of the ship power equipment directly predict the life of the ship power equipment or adopt historical periodic operation data and failure data, and fail to consider that the equipment parts are possibly corroded, including positions, degrees, types and the like of corrosion, so that the difference between the life prediction result and the actual life is large, therefore, the corrosion data of the equipment parts is introduced on the basis of the traditional life prediction data, and the prediction error is reduced;
step 2: the data processing module preprocesses the collected historical operation period data, the part corrosion data of the corresponding period, the real-time operation data and the part corrosion data of the corresponding period, eliminates the data with small change degree, and respectively forms residual life prediction samples P (A) and P (B) by the retained data;
and step 3: performing matrix preprocessing on the residual life prediction samples P (A), P (B) to respectively form multi-dimensional gray scale residual life prediction samples X, Y, i.e. one sample is individually expressed as x 1 =(x 1i ,x 2i ,…x ni ),y 1 =(y 1i ,y 2i ,…y ni ) Wherein n is a sample feature number;
Wherein m is the number of samples, and d is the characteristic number of the samples;
taking a multi-dimensional gray scale residual life prediction sample X as a training set and taking a multi-dimensional gray scale residual life prediction sample Y as a test set;
and 4, step 4: substituting the training set into an improved TCN model, extracting long-term and short-term time sequence characteristics and degradation information from the characteristic data, and converting the multidimensional data into one-dimensional characteristic vectors;
particularly, the traditional TCN model comprises expansion convolution and causal convolution, more time is needed when data are predicted and processed on large-scale equipment, the range of a receptive field is limited, and information in more time periods cannot be received, so that the TCN model is improved, an attention layer and an asymmetric residual error unit are added, the range of the receptive field is increased to control a long and short memory time sequence, the calculation time is shortened, and the calculation amount is reduced;
and with reference to fig. 2, the improved time convolution network model sequentially includes an input layer, four residual error units, an attention layer, a full connection layer and an output layer.
Four residual units comprising: the system comprises a first residual error unit, a second residual error unit, a third residual error unit and a fourth residual error unit which are sequentially connected, wherein the third residual error unit has the same structure as the first residual error unit, and the fourth residual error unit has the same structure as the second residual error unit;
referring to fig. 3, the first residual unit includes: the device comprises a one-dimensional extended convolutional layer, a PReLU activation function, batch standardization, a Dropout layer, a convolutional layer and a summation unit which are connected in sequence;
referring to fig. 4, the second residual unit includes: the device comprises a one-dimensional extended convolutional layer, a PReLU active function, a batch normalization layer, a Dropout layer, a convolutional layer and a summation unit which are connected in sequence;
and 5: on the basis of improving the TCN model, an attention module and a soft threshold under a residual error mode are added into a DRSN model for training, sample time sequence characteristic loss in the multi-layer training process is reduced through a multi-layer fusion model, deep-layer characteristics of data are extracted, and the weight of parameters which have large influence on the output of the fusion model is enhanced;
with reference to fig. 5, the input layer, the convolution layer, the four residual shrinkage units, the pooling layer, and the full-link layer are sequentially connected by the depth residual shrinkage network;
referring to fig. 6, the attention module in the residual mode includes a residual contraction unit and an attention unit.
The residual contraction unit is a residual contraction unit with different threshold values among channels; the method comprises the steps of including two batches of standardized BN, two PReLU activation functions, two convolution layers, a nonlinear contraction unit and identity mapping; wherein the nonlinear contraction unit comprises a global mean pooling GAP and two full link layers;
the attention unit comprises a global mean pooling layer, a full link layer, a ReLU function, a full link layer, a Softmax function, a Scale function and a summation unit;
optimizing the health degradation trend of the ship power equipment established by the fusion model;
step 6: continuously training and optimizing the DRSN model to obtain a residual life prediction model, substituting the test set into the residual life prediction model after normalization processing, and outputting a health factor;
introduction of component corrosion Rate v k The method is characterized in that component corrosion is used as an early characteristic of equipment abnormity and fault, corrosion of an engine cylinder sleeve is caused if the corrosion is slight, cylinder knocking, deformation of combustion chamber components and the like are caused if the corrosion is severe, so that corrosion rate is added into a health degradation trend, and optimization of the health degradation trend of the ship power equipment established by a fusion model is specifically as follows:
health factor hi (health indicator) is represented as follows:
wherein 1 represents that the marine power plant is completely healthy, and 0 represents that the marine power plant has been scrapped; t is t i Representing the time, alpha, at which the ship's power plant has various abnormal alarms i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j Representing the damage weight of different faults of the ship power equipment to the health state of the ship power equipment, T representing the total running time of the ship power equipment, n being a time sequence, v k Typical is the corrosion rate of the plant parts, α k Representing the weight of the effect of different component corrosion conditions on the life of the plant, L k The typical is the design life of the equipment components, N is the number of the equipment components, and if abnormal alarm causes damage to the health state of the engine, the damage isThe damage to the health of the engine caused by equipment failure isWith the increase of the corrosion degree, the working condition of the engine gradually becomes worse, and the abnormal alarming time and the failure rate are increased, namely D 1 、D 2 The health state of the engine is further influenced along with the increase of the engine;
particularly, the traditional health factor is calculated according to the normal operation time of the equipment, the condition that the equipment can still normally operate under some abnormal and fault states cannot be considered, so that the health factor has deviation, and finally the service life prediction result and the actual access are large, so that the health factor calculation method is optimized, the normal operation conditions under the abnormal and fault states are eliminated according to the calculation of the abnormal and fault times of the equipment, the corrosion rate is introduced, and the calculation accuracy of the health factor is improved;
and 7: performing linear regression prediction on the output health factor to obtain the health factor at the running time;
the linear regression equation is: y is equal to ax + b and,
wherein a and b are linear parameters, and x is the normal operation time of the ship power equipment, namelyy is the current health factor, i.e.
Further, the Remaining service Life (RUL) of the ship power plant can be derived from the health factors as:
wherein HI represents a health factor, t i Representing the time, alpha, at which various abnormal alarms occur in the power plant of the vessel i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j Representing damage weight of different faults of the ship power equipment to the health state of the ship power equipment, T representing total running time of the ship power equipment, and n representing a time sequence;
and 8: according to the result of the linear regression prediction, the real-time residual life prediction is carried out on the ship power equipment, so that the safety guarantee is provided for the running state of the ship power equipment, and the reliable basis is provided for the daily maintenance.
The probability distribution function F (t | y) of the residual service life of the ship power equipment can be obtained from the residual service life of the ship power equipment 1:n ) And a probability density function f (t | y) 1:n ):
F(t|y 1:n )=∫F(t|μ,y 1:n ;η,Λ(t))p(μ|y 1:n ;ω,k)dμ
f(t|y 1:n )=∫f(t|μ,y 1:n ;η,Λ(t))p(μ|y 1:n ;ω,k)dμ
Wherein t represents a time series, y 1:n The device health factor corresponding to the 1: n time series is represented, mu represents a random parameter of the degradation trend of the device, p represents a cumulative distribution function, eta, omega and k represent distribution parameters, and Λ (t) represents a monotone increasing function relative to the time series t.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Claims (9)
1. A marine power plant residual life prediction system, comprising: the system comprises power equipment, a data acquisition module, a data processing module, a model fusion module and a residual life prediction module;
the power equipment is connected with the data acquisition module; the data acquisition module is connected with the data processing module; the data acquisition module is respectively connected with the model fusion module and the residual life prediction module; the model fusion module is connected with the residual life prediction module;
the data acquisition module is used for acquiring historical operation cycle data and real-time operation data of the ship power equipment;
the data processing module processes historical operation cycle data and real-time operation data from the data acquisition module, and divides the processed historical operation cycle data and the real-time operation data into a training set and a test set;
the model fusion module is used for performing model training on the data processed by the data processing module and constructing an equipment degradation trend;
the residual life prediction module is used for a fusion model of the model fusion module, importing a test set of the data processing module into the residual life prediction model, and predicting the residual life of the ship power equipment.
2. A method for predicting the remaining life of a marine power plant, which is applied to the system for predicting the remaining life of a marine power plant according to claim 1, wherein the method for predicting the remaining life of a marine power plant comprises the following steps:
step 1: collecting historical operation cycle data, real-time operation data, historical cycle component corrosion data and real-time component corrosion data of ship power equipment;
step 2: preprocessing the data acquired in the step 1, providing data with small change degree, and forming two residual life prediction samples by the left data;
and step 3: performing matrix preprocessing on the residual life prediction samples to form two multi-dimensional gray residual life prediction samples;
and 4, step 4: substituting the training set of the residual life prediction sample into an improved TCN model, extracting long-term and short-term time sequence characteristics and degradation information of characteristic data, and converting multidimensional data into one-dimensional characteristic vectors;
and 5: adding the attention module and the soft threshold value in the residual error mode into the DRSN model to form a multilayer fusion model, training the multilayer fusion model on the basis of the step 4, reducing sample time sequence characteristic loss in the multilayer training process through the multilayer fusion model, extracting deep level characteristics of data, and outputting the weight of parameters with large influence by the multilayer fusion model;
step 6: training and optimizing the multilayer fusion model to obtain a residual life prediction model, inputting a test set of residual life prediction samples into the residual life prediction model after normalization processing, and outputting health factors;
and 7: performing linear regression prediction on the output health factor to obtain the health factor at the current operation time;
and 8: and according to the health factor at the current running time, predicting the residual life of the ship power equipment in real time.
3. The method for predicting the remaining life of a marine power plant according to claim 2, further comprising, after step 5 and before step 6:
and optimizing the health degradation trend of the ship power equipment in the multilayer fusion model.
4. The method for predicting the remaining life of the marine power plant according to claim 2, wherein the modified TCN model in step 4 includes an input layer, four residual error units, an attention layer, a full-link layer, and an output layer, which are connected in sequence.
5. The method for predicting the residual life of the marine power plant according to claim 4, wherein four residual error units are connected in sequence, wherein the first residual error unit and the third residual error unit have the same structure; the second residual error unit and the fourth residual error unit have the same structure.
6. The method for predicting the remaining life of the marine power plant of claim 5, wherein the first and third residual error units comprise, connected in sequence: one-dimensional extended convolutional layer, PReLU activation function, batch normalization, Dropout layer, convolutional layer, and summing unit;
the second and the fourth residual error units comprise the following components which are connected in sequence: one-dimensional extended convolutional layer, PReLU Activate function, batch normalization, Dropout layer, convolutional layer, and summing unit.
7. The method for predicting the remaining life of the marine power plant according to claim 3, wherein the optimizing the health degradation trend of the marine power plant in the multi-layer fusion model specifically comprises: the corrosion rate is added to the health degradation trend.
8. The method of predicting the remaining life of a marine power plant of claim 7, wherein the health factor is calculated as follows:
wherein HI represents a health factor, t i Representing the time, alpha, at which various abnormal alarms occur in the power plant of the vessel i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j Representing the damage weight of different faults of the ship power equipment to the health state of the ship power equipment, T representing the total running time of the ship power equipment, n being a time sequence, v k Typical is the corrosion rate of the plant parts, α k The weight of the effect of different component corrosion conditions on the life of the installation, L k The representative is the design life of the equipment components, and N is the number of the equipment components.
9. The method for predicting the remaining service life of the marine power plant according to claim 8, wherein the calculation formula of the remaining service life of the marine power plant is as follows:
wherein RUL represents the residual service life of the ship power equipment, HI represents the health factor, t i Representing the time, alpha, at which various abnormal alarms occur in the power plant of the vessel i Representing the weight of damage to the health of the vessel power plant due to different anomalies, t j Representing the time, alpha, at which various faults occur in the power plant of the vessel j The damage weight of different faults of the ship power equipment to the health state of the ship power equipment is represented, T represents the total running time of the ship power equipment, and n is a time sequence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210550575.5A CN114881342A (en) | 2022-05-18 | 2022-05-18 | System and method for predicting residual life of ship power equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210550575.5A CN114881342A (en) | 2022-05-18 | 2022-05-18 | System and method for predicting residual life of ship power equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114881342A true CN114881342A (en) | 2022-08-09 |
Family
ID=82676849
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210550575.5A Pending CN114881342A (en) | 2022-05-18 | 2022-05-18 | System and method for predicting residual life of ship power equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114881342A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115796059A (en) * | 2023-02-07 | 2023-03-14 | 中国电建集团山东电力建设第一工程有限公司 | Electrical equipment service life prediction method and system based on deep learning |
CN116011109A (en) * | 2023-01-13 | 2023-04-25 | 北京控制工程研究所 | Spacecraft service life prediction method and device, electronic equipment and storage medium |
CN116300837A (en) * | 2023-05-25 | 2023-06-23 | 山东科技大学 | Fault diagnosis method and system for unmanned surface vehicle actuator |
CN116502544A (en) * | 2023-06-26 | 2023-07-28 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
CN117129332A (en) * | 2023-08-29 | 2023-11-28 | 中国航发湖南动力机械研究所 | Method and system for acquiring strength life input data of engine complete machine corrosion test |
-
2022
- 2022-05-18 CN CN202210550575.5A patent/CN114881342A/en active Pending
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116011109A (en) * | 2023-01-13 | 2023-04-25 | 北京控制工程研究所 | Spacecraft service life prediction method and device, electronic equipment and storage medium |
CN116011109B (en) * | 2023-01-13 | 2023-09-08 | 北京控制工程研究所 | Spacecraft service life prediction method and device, electronic equipment and storage medium |
CN115796059A (en) * | 2023-02-07 | 2023-03-14 | 中国电建集团山东电力建设第一工程有限公司 | Electrical equipment service life prediction method and system based on deep learning |
CN116300837A (en) * | 2023-05-25 | 2023-06-23 | 山东科技大学 | Fault diagnosis method and system for unmanned surface vehicle actuator |
CN116300837B (en) * | 2023-05-25 | 2023-08-18 | 山东科技大学 | Fault diagnosis method and system for unmanned surface vehicle actuator |
CN116502544A (en) * | 2023-06-26 | 2023-07-28 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
CN116502544B (en) * | 2023-06-26 | 2023-09-12 | 武汉新威奇科技有限公司 | Electric screw press life prediction method and system based on data fusion |
CN117129332A (en) * | 2023-08-29 | 2023-11-28 | 中国航发湖南动力机械研究所 | Method and system for acquiring strength life input data of engine complete machine corrosion test |
CN117129332B (en) * | 2023-08-29 | 2024-05-10 | 中国航发湖南动力机械研究所 | Method and system for acquiring strength life input data of engine complete machine corrosion test |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114881342A (en) | System and method for predicting residual life of ship power equipment | |
CN112834211A (en) | Fault early warning method for transmission system of wind turbine generator | |
CN115017826B (en) | Method for predicting residual service life of equipment | |
Boullosa-Falces et al. | Monitoring of fuel oil process of marine diesel engine | |
Liu et al. | Bayesian long short-term memory model for fault early warning of nuclear power turbine | |
CN115717590B (en) | Intelligent abnormality detection method and related device for compressor | |
CN114444582A (en) | Mechanical equipment fault diagnosis method based on convolutional neural network and Bayesian network | |
CN116878902A (en) | Ship diesel engine fault early warning method and system based on deep learning CNN-BiLSTM | |
Tian et al. | Operation status monitoring of reciprocating compressors based on the fusion of spatio-temporal multiple information | |
Tong et al. | Fault prediction of marine diesel engine based on time series and support vector machine | |
Chen et al. | Remaining useful life prediction of bearings with two-stage LSTM | |
CN115774847B (en) | Diesel engine performance evaluation and prediction method and system | |
CN117523793A (en) | Power plant equipment fault early warning method and computer equipment | |
CN116610999A (en) | Gas turbine abnormality detection method and system based on federal life learning | |
CN115962944A (en) | VMD-based real-time diagnosis method for bearing fault of cutting part of intelligent coal mining machine group | |
Zaman et al. | The combination of reliability and predictive tools to determine ship engine performance based on condition monitoring | |
Li et al. | Multiple fault diagnosis of aeroengine control system based on autoassociative neural network | |
Sun et al. | Anomaly detection of marine diesel engine valve system based on Mahalanobis distance | |
Morgan et al. | Predicting future states with $ n $-dimensional Markov chains for fault diagnosis | |
Ayman et al. | Fault Detection in Wind Turbines using Deep Learning | |
Liang et al. | An investigation of unsupervised data-driven models for internal combustion engine condition monitoring | |
Vuong et al. | Anomaly Detection and Breakdown Diagnosis for Condition Monitoring of Marine Engines | |
CN109976316B (en) | Fault-related variable selection method | |
Tan et al. | Health assessment of marine gas turbine propulsion system under cross-working conditions based on transfer learning | |
Wei et al. | Condition monitoring for the marine diesel engine economic performance analysis with degradation contribution |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |