A system and method for predicting the remaining life of a ship power equipment

Abstract

The present invention discloses a method for predicting the remaining life of a ship power equipment, and the steps are as follows: step 1: collect various data of the ship power equipment; step 2: form two remaining life prediction samples; step 3: form two multi-dimensional grayscale remaining life prediction samples; step 4: substitute the training set into the improved TCN model, and convert the multi-dimensional data into a one-dimensional feature vector; step 5: extract the deep-level features of the data, and the multi-layer fusion model outputs the weights of the parameters with great influence; step 6: obtain the remaining life prediction model, input the test set of the remaining life prediction sample into the remaining life prediction model after normalization, and output the health factor; step 7: perform linear regression prediction on the output health factor to obtain the health factor at the current operating time; step 8: make a real-time remaining life prediction for the ship power equipment. The present invention reduces the prediction error because the equipment component corrosion data is introduced on the basis of the traditional life prediction data.

Description

Technical Field

[0001] The present invention relates to the technical field of remaining life prediction of ship equipment, and in particular to a system and method for predicting remaining life of ship power equipment. Background Art

[0002] Marine power equipment is an important supporting equipment for ships. These equipment are distributed in all positions of the ship, with a large number and wide distribution. However, most of the current power equipment is still in simple control or only realizes status monitoring. In order to avoid sudden accidents and economic losses caused by the sudden failure of power equipment, it is necessary to predict the remaining life of the equipment and prepare emergency treatment plans in advance to provide safe and reliable guarantees for its normal operation. Due to the uncertainty of the operating conditions of the equipment and the complexity of the working environment, the collected data must have some abnormal points and noise. These noises cannot reflect the true state of the equipment and will inevitably have an error effect on the final prediction results. Therefore, it is necessary to denoise the original data to eliminate the error effect of noise on the prediction results as much as possible.

[0003] In order to solve these problems and ensure the normal operation of ship power equipment, it is urgent to timely predict its remaining life. In recent years, the technology of equipment remaining life prediction has been gradually studied, including mechanism-based, data-based and hybrid model-based methods, especially deep learning has been gradually introduced into the field of remaining life prediction. For example, Yuan Ye and others from Huazhong University of Science and Technology invented and disclosed a method and system for predicting the remaining life of mechanical equipment, combining convolutional neural network with bidirectional gated recurrent unit to form a hybrid neural network, so as to effectively extract time and space features and improve the remaining life prediction accuracy (Yuan Ye, Huang Hong, Li Jiaqi. A method and system for predicting the remaining life of mechanical equipment [P]. Chinese Patent: CN113094822A: 2021-07-09), Hu Changhua and others from the Rocket Force Engineering University of the Chinese People's Liberation Army invented and disclosed a method and system for predicting the remaining life of equipment, based on massive data analysis, using deep belief network to realize quantitative prediction of remaining life (Hu Changhua, Pei Hong, Si Xiaosheng, etc. A method and system for predicting the remaining life of equipment [P]. Chinese Patent: CN110781592 A).

[0004] However, these methods all have obvious shortcomings. They only predict the remaining life of the equipment from a certain aspect of condition monitoring or fault diagnosis. The prediction data source is single and not universal. There is no distribution calculation of the remaining life of the equipment, and it is impossible to make real-time remaining life prediction of the equipment in a timely and accurate manner. Summary of the invention

[0005] The present invention provides a system and method for predicting the remaining life of a ship power device, so as to solve the problem in the prior art that it is impossible to timely and accurately predict the remaining life of a ship power device in real time.

[0006] The present invention provides a system for predicting the remaining life of a ship power equipment, comprising: a power equipment, a data acquisition module, a data processing module, a model fusion module, and a remaining life prediction module;

[0007] The power equipment is connected to the data acquisition module; the data acquisition module is connected to the data processing module; the data acquisition module is respectively connected to the model fusion module and the remaining life prediction module; the model fusion module is connected to the remaining life prediction module;

[0008] The data acquisition module is used to collect historical operation cycle data and real-time operation data of the ship power equipment;

[0009] The data processing module processes the historical operation cycle data and the 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;

[0010] The model fusion module is used to perform model training on the data processed by the data processing module and to construct equipment degradation trends;

[0011] The remaining life prediction module is used for the fusion model of the model fusion module, imports the test set of the data processing module into the remaining life prediction model, and predicts the remaining life of the ship power equipment.

[0012] The present invention also provides a method for predicting the remaining life of a ship power equipment, which is applicable to the above-mentioned system for predicting the remaining life of a ship power equipment, and comprises the following steps:

[0013] Step 1: Collect historical operation cycle data, real-time operation data, historical cycle component corrosion data, and real-time component corrosion data of ship power equipment;

[0014] Step 2: Preprocess the data collected in step 1, extract the data with small variation, and the remaining data constitute two remaining life prediction samples;

[0015] Step 3: pre-process the remaining life prediction samples into a matrix to form two multi-dimensional grayscale remaining life prediction samples;

[0016] Step 4: Substitute the training set of remaining life prediction samples into the improved TCN model, extract long-term and short-term time series features and degradation information from the feature data, and convert the multidimensional data into a one-dimensional feature vector;

[0017] Step 5: Add the attention module and soft threshold in the residual mode to the DRSN model to form a multi-layer fusion model. Train the multi-layer fusion model based on step 4. Use the multi-layer fusion model to reduce the loss of sample time series features in the multi-layer training process, extract the deep features of the data, and output the weights of the parameters with great influence.

[0018] Step 6: Train and optimize the multi-layer fusion model to obtain the remaining life prediction model, input the test set of the remaining life prediction samples into the remaining life prediction model after normalization, and output the health factor;

[0019] Step 7: Perform linear regression prediction on the output health factor to obtain the health factor at the current running time;

[0020] Step 8: Make a real-time remaining life prediction for the ship power equipment based on the health factor at the current operating moment.

[0021] Furthermore, after step 5 and before step 6, the following steps are further included:

[0022] Optimize the health degradation trend of ship power equipment in a multi-layer fusion model.

[0023] Furthermore, the improved TCN model in step 4 includes an input layer, four residual units, an attention layer, a fully connected layer and an output layer connected in sequence.

[0024] Furthermore, the four residual units are connected in sequence, wherein the first residual unit has the same structure as the third residual unit; and the second residual unit has the same structure as the fourth residual unit.

[0025] Furthermore, the first and third residual units include: a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence;

[0026] Second, the four residual units include: a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence.

[0027] Furthermore, the optimization of the health degradation trend of the ship power equipment in the multi-layer fusion model is specifically: adding the corrosion rate to the health degradation trend.

[0028] Furthermore, the calculation formula of the health factor is as follows:

[0029]

[0030] Among them, HI represents the health factor, t i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j represents the damage weight of different faults of ship power equipment to its health state, T represents the total operation time of ship power equipment, n is the time series, v k Represents the corrosion rate of equipment components, α k It represents the weight of the impact of corrosion of different components on the equipment life. k It represents the design life of equipment components, and N represents the number of equipment components.

[0031] Furthermore, the calculation formula for the remaining service life of ship power equipment is as follows:

[0032]

[0033] Among them, RUL represents the remaining service life of the ship power equipment, HI represents the health factor, and t i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j It represents the damage weight of different faults of ship power equipment to its health state, T represents the total operation time of ship power equipment, and n is the time series.

[0034] Beneficial effects of the present invention:

[0035] The present invention provides a system and method for predicting the remaining life of a ship power equipment. Since the corrosion data of equipment components is introduced on the basis of the traditional life prediction data, the prediction error can be reduced. By introducing the DRSN and improved TCN models in deep learning into the traditional field of remaining life prediction, the redundant information in the data can be effectively removed, the high-level feature learning ability of noisy data can be improved, the historical monitoring data can be effectively used, and the time series characteristics of the data can be fully mined. In addition, the one-dimensional dilated convolution can prevent the leakage of information and expand the receptive field of the observed data, so the remaining life prediction of the ship power equipment can be realized in time. Since the corrosion rate is added to the health degradation trend, the health degradation trend of the ship power equipment established by the fusion model is optimized, so the accuracy of the health factor calculation can be improved, and then the remaining life prediction of the ship power equipment can be realized in time and accurately, a scientific maintenance strategy is given, the working efficiency of the ship power equipment is improved, and the maintenance cost is saved. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the present invention in any way. In the accompanying drawings:

[0037] Figure 1 is a system overall flow chart of a specific embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the structure of an improved TCN model according to a specific embodiment of the present invention;

[0039] Figure 3 are the first and third residual units of the specific embodiment of the present invention;

[0040] Figure 4 are the second and fourth residual units in the specific embodiment of the present invention;

[0041] Figure 5 is a structural diagram of a DRSN model according to a specific embodiment of the present invention;

[0042] Figure 6 It is a diagram of the attention module under the residual mode of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0044] The embodiment of the present invention provides a system for predicting the remaining life of a ship power equipment. Figure 1 As shown, it includes: power equipment, data acquisition module, data processing module, model fusion module and remaining life prediction module.

[0045] The power equipment is connected to the data acquisition module; the data acquisition module is connected to the data processing module; the data acquisition module is respectively connected to the model fusion module and the remaining life prediction module; the model fusion module is connected to the remaining life prediction module.

[0046] The data acquisition module is used to collect historical operation cycle data and real-time operation data of the ship power equipment, wherein the data acquisition module includes a historical operation cycle data unit and a real-time data unit;

[0047] The data processing module processes the historical operation cycle data and real-time operation data from the data acquisition module, and divides the processed historical operation cycle data and real-time operation data into a training set and a test set. The data processing module includes a data preprocessing unit, a grayscale remaining life prediction sample unit, a training set unit, a test set unit, and a normalization unit.

[0048] The model fusion module is used to train the model based on the data processed by the data processing module and construct the equipment degradation trend. The model fusion module includes an improved TCN model unit, a DRSN model unit, an attention unit in residual mode, a soft threshold unit, and an equipment degradation trend unit.

[0049] The remaining life prediction module is used for the fusion model of the model fusion module, imports the test set of the data processing module into the remaining life prediction model, and predicts the remaining life of the ship power equipment. The remaining life prediction module includes a remaining life prediction model unit, a linear regression unit, and a remaining life unit.

[0050] The specific embodiment of the present invention also provides a remaining life prediction method of a remaining life prediction system for a marine power equipment;

[0051] The remaining life prediction method includes the following steps:

[0052] Step 1: The data acquisition module collects historical operation cycle data and real-time operation data of the ship power equipment; uses an industrial endoscope to detect the ship power equipment, collects historical cycle component corrosion data, and uses a corrosion monitor to collect real-time operation cycle component corrosion data;

[0053] In particular, under the current deteriorating world economic environment, shipowners usually reduce engine load and speed to reduce fuel consumption. In addition, the main engine cylinder liner is easily eroded by seawater, and the main engine is running at partial load for a long time, which is bound to cause poor combustion and carbon deposits in the combustion chamber and exhaust system. The reduction of the main engine's daily fuel consumption will aggravate exhaust gas corrosion and low-temperature corrosion, which will cause serious wear of key components such as the engine cylinder and combustion chamber, and cause continuous mechanical knocking in the main engine cylinder. Under low-load conditions, corrosion and wear often cause greater damage to the engine than mechanical wear.

[0054] However, the current remaining life prediction system and method of marine power equipment directly predicts its life or uses historical cycle operation data and failure data, but fails to consider that the equipment components may have been corroded, including the location, degree, and type of corrosion, resulting in a large discrepancy between the life prediction result and the actual life. Therefore, the equipment component corrosion data is introduced on the basis of the traditional life prediction data to reduce the prediction error;

[0055] Step 2: The data processing module pre-processes the collected historical operation cycle data and component corrosion data of the corresponding cycle, real-time operation data and component corrosion data of the corresponding cycle, removes the data with a small degree of change, and the retained data constitute the remaining life prediction samples P(A) and P(B);

[0056] Step 3: Matrix the remaining life prediction samples P(A) and P(B) to form multi-dimensional grayscale remaining life prediction samples X and Y, respectively. That is, a sample is represented as x1 = (x 1i ,x 2i ,…x ni ),y1=(y 1i ,y 2i ,…y ni ), where n is the number of sample features;

[0057] Multidimensional grayscale remaining life prediction sample

[0058] Multidimensional grayscale remaining life prediction sample

[0059] Where m is the number of samples and d is the number of sample features;

[0060] The multi-dimensional grayscale remaining life prediction sample X is used as the training set, and the multi-dimensional grayscale remaining life prediction sample Y is used as the test set;

[0061] Step 4: Substitute the training set into the improved TCN model, extract long-term and short-term time series features and degradation information from the feature data, and convert the multidimensional data into a one-dimensional feature vector;

[0062] In particular, the traditional TCN model includes dilated convolution and causal convolution, which takes more time to process data for large-scale equipment life prediction, and the range of the receptive field is limited, and it cannot receive information in more time periods. Therefore, the TCN model is improved, and the attention layer and asymmetric residual unit are added to increase the range of the receptive field to control the long and short memory sequence, shorten the calculation time and reduce the amount of calculation;

[0063] Combination Figure 2 , the improved temporal convolutional network model sequentially connects the input layer, four residual units, attention layer, fully connected layer and output layer.

[0064] Four residual units, including: a first residual unit, a second residual unit, a third residual unit, and a fourth residual unit connected in sequence, wherein the third residual unit has the same structure as the first residual unit, and the fourth residual unit has the same structure as the second residual unit;

[0065] Combination Figure 3 , the first residual unit, including: a one-dimensional dilated convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional dilated convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence;

[0066] Combination Figure 4 , the second residual unit, comprising: a one-dimensional dilated convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional dilated convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional dilated convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence;

[0067] Step 5: Based on the improved TCN model, the attention module and soft threshold in the residual mode are added to the DRSN model for training. The multi-layer fusion model is used to reduce the loss of sample time series features in the multi-layer training process, extract the deep features of the data, and strengthen the weights of the parameters that have a greater impact on the output of the fusion model;

[0068] Combination Figure 5 , the deep residual contraction network sequentially connects the input layer, convolutional layer, four residual contraction units, pooling layer, and fully connected layer;

[0069] Combination Figure 6 ,The attention module in the residual mode includes the residual shrinkage unit and the attention unit.

[0070] The residual shrinkage unit is a residual shrinkage unit with different thresholds between channels; it includes two batch normalization BNs, two PReLU activation functions, two convolutional layers, a nonlinear shrinkage unit and an identity mapping; the nonlinear shrinkage unit includes a global average pooling GAP and two fully connected layers;

[0071] The attention unit consists of a global mean pooling layer, a fully connected layer, a ReLU function, a fully connected layer, a Softmax function, a Scale function, and a summation unit;

[0072] Optimization of the fusion model to establish the health degradation trend of ship power equipment;

[0073] Step 6: Continuously train and optimize the DRSN model to obtain the remaining life prediction model, insert the test set into the remaining life prediction model after normalization, and output the health factor;

[0074] Introduced component corrosion rate v k Component corrosion is an early sign of equipment abnormality and failure. It may cause corrosion of the engine cylinder liner or even cylinder knocking and deformation of combustion chamber components. Therefore, the corrosion rate is added to the health degradation trend, and the fusion model is optimized to establish the health degradation trend of ship power equipment. Specifically:

[0075] The health indicator HI is expressed as follows:

[0076]

[0077] 1 represents that the ship's power equipment is completely healthy, and 0 represents that the ship's power equipment has been scrapped; i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j represents the damage weight of different faults of ship power equipment to its health state, T represents the total operation time of ship power equipment, n is the time series, v k Represents the corrosion rate of equipment components, α k It represents the weight of the impact of corrosion of different components on the equipment life. k represents the design life of the equipment components, N represents the number of equipment components, and if the abnormal alarm causes damage to the engine health status The damage caused by equipment failure to the engine health status is As the corrosion degree increases, the engine operating condition gradually deteriorates, the abnormal alarm time and the failure rate increase, that is, D1 and D2 increase accordingly, further affecting the health of the engine;

[0078] In particular, the traditional health factor is calculated based on the normal operation time of the equipment, and fails to take into account the fact that the equipment can still operate normally under some abnormal and fault conditions, resulting in deviations in the health factor, which ultimately leads to a large discrepancy between the life prediction result and the actual result. Therefore, the health factor calculation method is optimized to calculate based on the abnormal and fault time of the equipment, excluding the normal operation under abnormal and fault conditions, and at the same time introducing the corrosion rate to improve the accuracy of health factor calculation;

[0079] Step 7: Perform linear regression prediction on the output health factor to obtain the health factor at the current running time;

[0080] The linear regression equation is: y = ax + b,

[0081] Among them, a and b are linear parameters, and x is the normal operation time of the ship power equipment, that is, y is the current health factor, that is

[0082] Furthermore, the remaining useful life (RUL) of the ship power equipment can be derived from the health factor:

[0083]

[0084] HI stands for health factor, t i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j represents the damage weight of different faults of ship power equipment to its health state, T represents the total operating time of ship power equipment, and n is the time series;

[0085] Step 8: Based on the results of linear regression prediction, make a real-time remaining life prediction for the ship power equipment to provide safety protection for its operating status and a reliable basis for daily maintenance.

[0086] The remaining service life of the ship power equipment can be used to obtain its remaining service life probability distribution function F(t|y 1:n ) and the probability density function f(t|y 1:n ):

[0087] F(t|y 1:n)=∫F(t|μ,y 1:n ;η,Λ(t))p(μ|y 1:n ;ω,k)dμ

[0088] f(t|y 1:n )=∫f(t|μ,y 1:n ;η,Λ(t))p(μ|y 1:n ;ω,k)dμ

[0089] Among them, t represents the time series, y 1:n represents the equipment health factor corresponding to the 1:n time series, μ represents the random parameter of the equipment degradation trend, p represents the cumulative distribution function, η, ω, k all represent distribution parameters, and Λ(t) represents the monotonic growth function of the time series t.

[0090] 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 present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A system for predicting the remaining life of a ship power equipment, characterized in that: include: Power equipment, data acquisition module, data processing module, model fusion module, remaining life prediction module; The power equipment is connected to the data acquisition module; the data acquisition module is connected to the data processing module; the data acquisition module is respectively connected to the model fusion module and the remaining life prediction module; the model fusion module is connected to the remaining life prediction module; The data acquisition module is used to collect historical operation cycle data and real-time operation data of the ship power equipment; The data processing module processes the historical operation cycle data and the 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 to perform model training on the data processed by the data processing module and to construct equipment degradation trends; The remaining life prediction module is used for the fusion model of the model fusion module, imports the test set of the data processing module into the remaining life prediction model, outputs the health factor and predicts the remaining life of the ship power equipment, The calculation formula of the health factor is as follows: Among them, HI represents the health factor, t i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j represents the damage weight of different faults of ship power equipment to its health state, T represents the total operation time of ship power equipment, n is the time series, v k Represents the corrosion rate of equipment components, α k It represents the weight of the impact of corrosion of different components on the equipment life. k It represents the design life of equipment components, and N represents the number of equipment components.

2. A method for predicting the remaining life of a ship power equipment, applicable to the system for predicting the remaining life of a ship power equipment as claimed in claim 1, characterized in that: The method for predicting the remaining life of a ship power equipment comprises the following steps: Step 1: Collect 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: Preprocess the data collected in step 1, extract the data with small variation, and the remaining data constitute two remaining life prediction samples; Step 3: Matrix preprocess the remaining life prediction samples to form two multi-dimensional grayscale remaining life prediction samples; Step 4: Substitute the training set of remaining life prediction samples into the improved TCN model, extract long-term and short-term time series features and degradation information from the feature data, and convert the multidimensional data into a one-dimensional feature vector; Step 5: Add the attention module and soft threshold in the residual mode to the DRSN model to form a multi-layer fusion model. Train the multi-layer fusion model based on step 4. Use the multi-layer fusion model to reduce the loss of sample time series features in the multi-layer training process, extract the deep features of the data, and output the weights of the parameters with great influence. Step 6: Train and optimize the multi-layer fusion model to obtain the remaining life prediction model, input the test set of the remaining life prediction samples into the remaining life prediction model after normalization, and output the health factor; Step 7: Perform linear regression prediction on the output health factor to obtain the health factor at the current running time; Step 8: Make a real-time remaining life prediction for the ship power equipment based on the health factor at the current operating moment.

3. The method for predicting the remaining life of a ship power equipment according to claim 2, characterized in that: After step 5 and before step 6, the following steps are also included: Optimize the health degradation trend of ship power equipment in a multi-layer fusion model.

4. The method for predicting the remaining life of a ship power equipment according to claim 2, characterized in that: The improved TCN model in step 4 includes an input layer, four residual units, an attention layer, a fully connected layer and an output layer connected in sequence.

5. The method for predicting the remaining life of a ship power equipment according to claim 4, characterized in that: The four residual units are connected in sequence, wherein the first residual unit has the same structure as the third residual unit; and the second residual unit has the same structure as the fourth residual unit.

6. The method for predicting the remaining life of a ship power equipment according to claim 5, characterized in that: First, the three residual units include: a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence; Second, the four residual units include: a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a one-dimensional expanded convolution layer, a PReLU activation function, a batch normalization, a Dropout layer, a convolution layer, and a summation unit connected in sequence.

7. The method for predicting the remaining life of a ship power equipment according to claim 3, characterized in that: The optimization of the health degradation trend of the ship power equipment in the multi-layer fusion model is specifically: adding the corrosion rate to the health degradation trend.

8. The method for predicting the remaining life of a ship power equipment according to claim 2, characterized in that: The calculation formula for the remaining service life of ship power equipment is as follows: Among them, RUL represents the remaining service life of the ship power equipment, HI represents the health factor, and t i It represents the time when various abnormal alarms occur in the ship power equipment. i It represents the damage weight of different abnormalities of ship power equipment to its health status, t j It represents the time when various failures occur in the ship power equipment. j It represents the damage weight of different faults of ship power equipment to its health state, T represents the total operation time of ship power equipment, and n is the time series.

Images