CN115718880A - Prediction method for degradation stage of complex equipment - Google Patents
Prediction method for degradation stage of complex equipment Download PDFInfo
- Publication number
- CN115718880A CN115718880A CN202211399371.2A CN202211399371A CN115718880A CN 115718880 A CN115718880 A CN 115718880A CN 202211399371 A CN202211399371 A CN 202211399371A CN 115718880 A CN115718880 A CN 115718880A
- Authority
- CN
- China
- Prior art keywords
- model
- prediction
- complex equipment
- degradation
- data
- 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
Abstract
The invention relates to the technical field of prediction of a degradation stage of complex equipment, and discloses a prediction method of the degradation stage of the complex equipment, which comprises the steps of collecting signals of a multi-source sensor of the complex equipment; multi-source sensor parameter relevance analysis of complex equipment; performing k-means clustering on the multi-source sensor data in the whole life cycle, establishing a clustering model, and determining a degradation stage center; establishing a linear autoregressive model and a nonlinear neural network mixed multi-dimensional parameter prediction model; training the constructed multi-dimensional parameter prediction model until the prediction error meets the prediction precision requirement; inputting the latest monitoring data of the complex equipment into a multi-dimensional parameter prediction model to obtain the predicted value of each key parameter; and calculating a prediction result, inputting the prediction result into a clustering model, and judging a degradation stage to be entered by the complex equipment. The method and the device predict the degradation stage of the complex equipment to be entered, improve the accuracy of the prediction of the degradation stage of the complex equipment and improve the comprehensive guarantee capability of the complex equipment.
Description
Technical Field
The invention relates to the technical field of prediction of a degradation stage of complex equipment, in particular to a prediction method of the degradation stage of the complex equipment.
Background
The complex equipment has a plurality of parts, the sensor monitoring parameter relevance coupling is complex, and various mechanical, hydraulic, electrical and electronic devices are integrated. Under the influence of workload and working environment, the complex equipment inevitably degrades, and the degradation stage is generally divided into a healthy stage (slow degradation), a degradation stage (fast degradation) and an impending failure stage. Therefore, the accurate and timely prediction of the degradation stage can predict the state of the complex equipment, the preventive maintenance of the equipment is carried out in advance, and the operation of the complex equipment is guaranteed.
The maintenance of the complex equipment depends on regular maintenance for a long time, the regular maintenance has the characteristics of multiple times and high frequency, and the equipment guarantee expenditure is increased. With the development of fault prediction and health management technologies, preventive maintenance and guarantee become the mainstream of equipment maintenance.
Therefore, how to provide a method for predicting the degradation stage of the complex equipment, which can improve the accuracy and the rapidity of the prediction of the degradation stage of the complex equipment, becomes a technical problem which needs to be solved urgently by a person skilled in the art.
Disclosure of Invention
The invention aims to provide a method for predicting the degradation stage of complex equipment so as to solve the problem.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a method of predicting a degradation phase of complex equipment, the method comprising the steps of:
step S1: establishing a linear autoregressive model and a nonlinear neural network mixed multi-dimensional parameter prediction model;
step S2: establishing a k-means clustering model for outputting a prediction result of a degradation stage of the complex equipment;
and step S3: calculating the predicted value of each parameter according to the input online monitoring data of the complex equipment by the multi-dimensional parameter prediction model in the step S1;
and step S4: and (4) inputting the predicted values of the parameters obtained in the step (S3) into the k-means clustering model in the step (S2) to measure the distances between the prediction state and the degradation centers, performing degradation stage matching according to the distances between the prediction state and the degradation centers, and outputting the prediction result of the degradation stage of the complex equipment.
Further, the nonlinear neural network comprises a hidden layer and an output layer, wherein the hidden layer comprises a linear autoregressive model, and the linear autoregressive model is parallel to the nonlinear neural network model.
Further, the specific steps of establishing the multidimensional parameter prediction model are as follows:
step S1.1: forming a historical database by signal data acquired by a multi-source sensor in the whole life cycle process of the complex equipment;
step S1.2: carrying out data preprocessing on the data of the historical database;
step S1.3: performing correlation analysis on the multidimensional parameters of the complex equipment through a Pearson coefficient, and performing feature selection;
step S1.4: and (4) combining the input data processed by the S1.2 and the S1.3 to construct a multi-dimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network.
Furthermore, the online monitoring data is subjected to data preprocessing before being input into the multi-dimensional parameter prediction model, and relevant features are selected as input.
Further, the nonlinear neural network model is trained through historical data, and the model structure and the model parameters are adjusted.
Further, the establishment of the k-means clustering model comprises the following steps,
s2.1, selecting K objects in a monitoring data space as initial centers, namely centers of all degradation stages;
s2.2, dividing the data objects into degradation stages corresponding to the nearest clustering centers according to the Euclidean distance between the data objects and the centers of the degradation stages in S2.1 and the nearest criterion;
s2.3, updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category, and calculating the value of a target function;
s2.4, judging whether the values of the clustering center and the objective function are changed or not, if not, outputting the result, and if so, re-classifying the categories.
Further, the data preprocessing method comprises missing value filling, outlier rejection and data normalization and noise removal.
Further, the monitoring data is multidimensional time series data.
Further, the historical data training comprises the steps of,
step one, constructing a supervised sample based on time sequence monitoring data according to an input/output structure of a constructed prediction model, and dividing a training set and a test set;
setting a model hyper-parameter, initializing a neuron parameter in a network, and designing a loss function, a parameter optimizer and a training convergence finishing condition;
inputting a training sample, and performing forward propagation calculation;
step four, calculating a model loss function and judging whether the loss function is converged;
step five, if the loss function is converged, training is finished; otherwise, the error is reversely propagated, the model parameters are optimized, the third step and the fourth step are iterated until the loss function meets the requirement or meets the training end condition, and whether the prediction precision meets the requirement or not is judged.
A system for predicting a degradation phase of a complex equipment, the system comprising:
a first calculation module: the multi-dimensional parameter prediction model is used for establishing a linear autoregressive model and a nonlinear neural network mixture;
a second calculation module: establishing a k-means clustering model for outputting a prediction result of a degradation stage of the complex equipment;
a first execution module: the multi-dimensional parameter prediction model is used for calculating the predicted value of each parameter according to the input online monitoring data of the complex equipment through the first calculation module;
a second execution module: and the prediction value of each parameter obtained by the first execution module is input into the k-means clustering model of the second calculation module to measure the distance between the prediction state and each degradation center, the degradation stage matching is carried out according to the distance between the prediction state and each degradation center, and the prediction result of the degradation stage of the complex equipment is output.
Has the advantages that:
the invention discloses a prediction method of a degradation stage of complex equipment, which clusters parameters of a multi-dimensional sensor based on a k-means clustering method to obtain a degradation center of each degradation stage, predicts key parameters based on a multi-dimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network, and inputs a prediction result into the k-means clustering model to predict the degradation stage.
Drawings
Fig. 1 is a flowchart of a method for predicting a degradation stage of complex equipment according to the present invention.
FIG. 2 is a flow chart of the k-means degradation staging division of the present invention.
FIG. 3 is a diagram of a multi-dimensional parameter prediction model based on a linear autoregressive model and a nonlinear neural network.
FIG. 4 is a flow chart of the multi-dimensional parametric prediction model training 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.
Example 1
The invention discloses a method for predicting a degradation stage of complex equipment, which comprises the following steps:
s1, accumulating signal data acquired by a multi-source sensor of the complex equipment in the whole life cycle process to form a historical database;
s2, performing data preprocessing on the original data, wherein the data preprocessing comprises noise removal and normalization operations;
s3, performing correlation analysis on the multi-dimensional parameters of the complex equipment through a Pearson coefficient, and performing feature selection;
s4, establishing a k-means clustering model, taking the data processed in the S2 and the S3 as input, and adjusting model parameters to obtain clustering centers of all degradation stages;
s5, combining the input data processed in S2 and S3 to construct a multi-dimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network;
s6, training the constructed neural network model by using historical data, and adjusting the model structure and model parameters until the prediction precision meets the requirement;
s7, inputting online monitoring data of the complex equipment into the hybrid network prediction model obtained in the S6 to obtain a prediction value of each parameter;
s8, inputting the prediction result in the S7 into the k-means clustering model in the S4 to measure the distance between the prediction state and each degradation center;
and S9, performing degradation stage matching according to the calculation result of the S8, and outputting a prediction result of the degradation stage of the complex equipment.
In this embodiment, the step S2 specifically includes:
preprocessing the data, including missing value filling and normalization processing;
preprocessing data, including missing value filling, abnormal value removing, data normalization and noise removing;
filling the missing values by adopting a moving average method, namely calculating an average value of a section of data before the missing values as the missing values; carrying out abnormity judgment on the abnormal value by adopting a 3 sigma rule, and filling by adopting a moving average method; all the variables are normalized by the formula
Noise removal filters high frequency noise using a butterworth filter.
In this embodiment, the step S3 specifically includes:
for each monitoring variable, sequentially calculating a Pearson coefficient between the monitoring variable and other monitoring variables; for time series X and Y of two monitoring variables, pearsonThe coefficient is calculated by the formula:
where ρ is X,Y To calculate the Pearson coefficient, cov (X, Y) is the covariance of the two, μ X And mu Y Is the mean of the two, σ X And σ Y Is the variance of the two;
and calculating a Pearson coefficient matrix of each monitoring variable, wherein only one variable with higher correlation is reserved.
In this embodiment, the step S4 specifically includes:
selecting K objects in a monitoring data space as initial centers, namely the centers of all degradation stages according to experience; for the data objects in the sample, according to Euclidean distances between the data objects and the clustering centers, classifying the data objects into degradation stages corresponding to the clustering centers (most similar) closest to the data objects according to the closest criterion; updating a clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category, and calculating the value of a target function; and judging whether the values of the clustering center and the objective function are changed or not, if not, outputting the result, and if so, re-classifying the categories.
In this embodiment, the step S5 specifically includes:
the constructed multi-dimensional parameter prediction model is a deep learning model designed for multi-variable time series prediction, the multi-dimensional time series is firstly input into a convolutional layer, a local dependency mode among multi-dimensional input variables is discovered by utilizing the advantages of the convolutional layer, and then the multi-dimensional parameter prediction model is input into a recursive layer to capture a complex long-term dependency relationship. Meanwhile, the traditional linear autoregressive model and a nonlinear neural network are partially paralleled by the model, and an input time sequence is automatically decomposed into a linear low frequency block and a nonlinear high frequency block for prediction, so that the nonlinear deep learning model has higher robustness for the time sequence with scale change.
In this embodiment, the step S6 specifically includes:
and (3) performing training set and test set division on the historical data, if the historical data volume is not large enough, training the model by adopting a five-fold cross validation method, and completing model training when the model prediction is converged and the prediction performance meets the requirements.
In this embodiment, the step S7 specifically includes:
and (3) acquiring online data of the complex equipment, executing step (S2) to perform data preprocessing, selecting corresponding parameters according to the conclusion of step (S3), and inputting the parameters into the mixed model obtained by training in step (S6) to obtain the prediction result of each parameter.
In this embodiment, the step S8 specifically includes:
and (4) inputting the prediction result obtained in the step (S7) into the k-means clustering model obtained by learning in the step (S4) to obtain the Euclidean distance between the prediction result and each clustering center.
In this embodiment, the step S9 specifically includes:
and sequencing the Euclidean distance calculation results in the step S8 and matching with the degradation stage.
Example 2
As shown in fig. 1, the present invention provides a method for predicting the degradation stage of complex equipment, in case 2, an aircraft engine is used as an application case, and the method comprises the following steps:
step 1, accumulating running state data collected by a multi-source sensor in the historical working process of complex equipment. The monitoring data of the complex equipment mainly comprises the temperature, pressure, rotating speed and other information of the system for characterizing the running state of the equipment.
And 2, performing data cleaning on the acquired various data, specifically comprising missing value filling, abnormal value removing, noise removing and normalization processing.
Specifically, the steps of data preprocessing are described in detail as follows:
step 2.1, filling missing data values: for the position with missing value, using the average value of the first five values of the position as the value of the missing value;
step 2.2, removing abnormal data values: calculating the mean value and the variance of the data, regarding the numerical value with the difference value of the mean value larger than 3 sigma as an abnormal value, removing the numerical value, and filling missing values;
step 2.3, noise removal: for data recorded by a sensor and having high-frequency noise, a low-pass filter is designed by considering factors such as acquisition frequency and the like to filter the high-frequency noise;
step 2.4, normalization: all the variables are normalized by the formula of
And 3, performing correlation analysis on the multi-dimensional parameters of the complex equipment through the Pearson coefficient, performing feature selection, and screening out parameters with different change rules.
Step 3.1, sequentially calculating the Pearson coefficient between each monitoring variable and other monitoring variables; for example, for a time series X and Y of two monitored variables, the pearson coefficient is calculated as:
where ρ is X,Y To calculate the Pearson coefficient, cov (X, Y) is the covariance of the two, μ X And mu Y Is the mean of the two, σ X And σ Y Is the variance of the two; the coefficient characterizes the direct correlation of the two variables;
and 3.2, calculating the Pearson coefficient matrix of each monitoring variable, and only one variable with higher correlation is reserved, so that the aim of data reduction is fulfilled.
Step 4, establishing a k-means clustering model, taking the data processed in the steps S2 and S3 as input, and adjusting model parameters to obtain clustering centers of all degradation stages;
specifically, as shown in fig. 2, the detailed flow of the k-means model construction is as follows:
step 4.1, selecting K objects in the monitoring data space as initial centers according to experience, namely the centers of all degradation stages;
step 4.2, for the data objects in the sample, according to Euclidean distances between the data objects and the clustering centers, dividing the data objects into degradation stages corresponding to the clustering centers (most similar) closest to the data objects according to the closest criterion;
step 4.3, updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category, and calculating the value of a target function;
and 4.4, judging whether the values of the clustering center and the objective function are changed or not, if not, outputting a result, and if so, re-classifying the categories.
Step 5, combining the processed historical input data to construct a multi-dimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network;
specifically, as shown in fig. 3, the detailed structure of the model construction is as follows:
and 5.1, inputting and outputting the network to comprise N-dimensional parameters, wherein the input is data subjected to data preprocessing and relevance analysis, and the input data is multidimensional time sequence data.
And 5.2, the time step of inputting the multidimensional data is m, the time step of outputting the multidimensional data is 1, and the sensor data of the next step 1 is predicted by using the monitored sensor data of the previous step m.
And 5.3, alternately connecting hidden layers of the network by convolution layers and circulation layers, wherein the hidden layers comprise linear autoregressive models, inputting the convolution layers in a multi-dimensional time sequence, discovering a local dependency mode among multi-dimensional input variables by using the advantages of the convolution layers, and then inputting the recursive layers to capture complex long-term dependency relationships. Meanwhile, the model parallels the traditional linear autoregressive model and a nonlinear neural network part, so that the nonlinear deep learning model is more robust to time sequences with scale changes.
Step 6, training the constructed neural network model by using historical data, and adjusting the model structure and the model parameters until the prediction precision meets the requirement;
specifically, as shown in fig. 4, the detailed structure of the model construction is as follows:
6.1, constructing a supervised sample based on time sequence monitoring data according to an input/output structure of the constructed prediction model, and dividing a training set and a test set;
step 6.2, setting model hyper-parameters, initializing neuron parameters in the network, and designing a loss function, a parameter optimizer and a training convergence finishing condition;
step 6.3, inputting a training sample and carrying out forward propagation calculation;
6.4, calculating a model loss function and judging whether the loss function is converged;
step 6.5, if the loss function is converged, finishing the training; otherwise, the error is reversely propagated, the model parameters are optimized, the steps 6.3 and 6.4 are iterated until the loss function meets the requirements or meets the training end conditions, and the predicted value is calculated
Root mean square error of sum true value y
And accuracy of alignment
And calculating the prediction error and accuracy, and judging whether the prediction accuracy meets the requirement.
Step 7, inputting online monitoring data of the complex equipment into the prediction model of the hybrid network obtained in the step 6 to obtain the predicted value of each parameter;
specifically, the detailed flow of multi-dimensional parameter prediction is as follows:
step 7.1, data preprocessing is carried out on the online data, and relevant characteristics are selected as input;
and 7.2, predicting by using the trained prediction model.
And 8, inputting the prediction result obtained in the step 7 into the k-means clustering model obtained in the step 4 to obtain Euclidean distances between the prediction result and each clustering center.
And 9, performing degradation stage matching according to the calculation result of the S8, and outputting a prediction result of the degradation stage of the complex equipment.
According to the method, the advantages of the linear model and the nonlinear model are fused by establishing the hybrid prediction model of the linear model and the nonlinear model, and the problem that the multi-dimensional data multi-input multi-output prediction effect is poor in the traditional method is solved; the method of k-means clustering overcomes the defect that only fixed threshold value and artificial experience degradation stage division are relied on.
Example three:
a system for predicting a degradation phase of a complex equipment, the system comprising:
a first calculation module: the multi-dimensional parameter prediction model is used for establishing a linear autoregressive model and a nonlinear neural network mixture;
a second calculation module: establishing a k-means clustering model for outputting a prediction result of a degradation stage of the complex equipment;
a first execution module: the multi-dimensional parameter prediction model is used for calculating predicted values of all parameters according to input online monitoring data of the complex equipment through the first calculation module;
a second execution module: and the prediction value of each parameter obtained by the first execution module is input into the k-means clustering model of the second calculation module to measure the distance between the prediction state and each degradation center, the degradation stage matching is carried out according to the distance between the prediction state and each degradation center, and the prediction result of the degradation stage of the complex equipment is output.
The working principle is as follows:
clustering degradation stages based on k-means clustering, and predicting multidimensional parameters of a sensor based on a multidimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network; and predicting the degradation stage of the complex equipment to be entered based on the prediction result of the mixed model and the mean value clustering model. Based on the deep learning, statistical model prediction and unsupervised learning theory, the degradation state of the complex equipment to be entered is predicted, and powerful guarantee is provided for preventive maintenance and stable and efficient operation of the complex equipment.
Aiming at the defects of the existing method, the invention provides a degradation stage prediction method based on deep learning, statistical model prediction and unsupervised learning theory. On the basis of carrying out data preprocessing and relevance analysis on the complex equipment multi-dimensional monitoring data, predicting the multi-dimensional monitoring data generated by the complex equipment based on a multi-dimensional parameter prediction model, and then carrying out degradation stage division on the predicted data according to the established k-means clustering model to realize degradation stage prediction on the complex equipment.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for predicting a degradation phase of complex equipment, the method comprising the steps of:
step S1: establishing a linear autoregressive model and a nonlinear neural network mixed multidimensional parameter prediction model;
step S2: establishing a k-means clustering model for outputting a prediction result of a degradation stage of the complex equipment;
and step S3: calculating the predicted value of each parameter according to the input online monitoring data of the complex equipment by the multi-dimensional parameter prediction model in the step S1;
and step S4: and (4) inputting the predicted values of the parameters obtained in the step (S3) into the k-means clustering model in the step (S2) to measure the distances between the prediction state and the degradation centers, performing degradation stage matching according to the distances between the prediction state and the degradation centers, and outputting the prediction result of the degradation stage of the complex equipment.
2. The method of claim 1, wherein the nonlinear neural network comprises an implicit layer and an output layer, the implicit layer comprises a linear autoregressive model and a cyclic layer, the linear autoregressive model is parallel to the nonlinear neural network model, the linear autoregressive model models the autoregressive predictive linear component with a fully connected layer, the nonlinear network predicts high-frequency nonlinear changes, and the sum of the two results is used as a prediction vector.
3. The method for predicting the degradation stage of complex equipment according to claim 1, wherein the multidimensional parameter prediction model is established by the following specific steps:
step S1.1: forming a historical database by using signal data acquired by a multi-source sensor in the whole life cycle process of the complex equipment;
step S1.2: carrying out data preprocessing on the data of the historical database;
step S1.3: performing correlation analysis on multi-dimensional parameters of the complex equipment through a Pearson coefficient, and performing feature selection;
step S1.4: and (4) combining the input data processed by the S1.2 and the S1.3 to construct a multi-dimensional parameter prediction model mixed by a linear autoregressive model and a nonlinear neural network.
4. The method as claimed in claim 1, wherein the online monitored data is pre-processed before being input into the multidimensional parameter prediction model, and the relevant features are selected as input.
5. The method as claimed in claim 1, wherein the nonlinear neural network model is trained from historical data to adjust model structure and model parameters.
6. The method for predicting the degradation stage of complex equipment according to claim 1, wherein the k-means clustering model is established by the following steps,
s2.1, selecting K objects in a monitoring data space as initial centers, namely centers of all degradation stages;
s2.2, dividing the data objects into degradation stages corresponding to the nearest clustering centers according to the Euclidean distance between the data objects and the centers of the degradation stages in S2.1 and the nearest criterion;
s2.3, updating the clustering center: taking the mean values corresponding to all the objects in each category as the clustering center of the category, and calculating the value of a target function;
s2.4, judging whether the values of the clustering center and the objective function are changed or not, if not, outputting the result, and if so, re-classifying the categories.
7. The method for predicting the degradation stage of complex equipment as claimed in claim 3 or 4, wherein the method for preprocessing the data comprises missing value filling, outlier rejection and data normalization and noise removal.
8. The method of claim 1, wherein the monitoring data is multidimensional time series data.
9. The method of claim 5, wherein the historical data training comprises the steps of,
step one, constructing a supervised sample based on time sequence monitoring data according to an input/output structure of a constructed prediction model, and dividing a training set and a test set;
setting a model hyper-parameter, initializing a neuron parameter in a network, and designing a loss function, a parameter optimizer and a training convergence finishing condition;
inputting a training sample, and performing forward propagation calculation;
step four, calculating a model loss function and judging whether the loss function is converged;
step five, if the loss function is converged, training is finished; otherwise, the error is propagated reversely, the model parameters are optimized, the third step and the fourth step are repeated iteratively until the loss function meets the requirement or meets the training end condition, and whether the prediction precision meets the requirement or not is judged.
10. A system for predicting a degradation phase of complex equipment, the system comprising:
a first calculation module: the multi-dimensional parameter prediction model is used for establishing a linear autoregressive model and a nonlinear neural network mixture;
a second calculation module: establishing a k-means clustering model for outputting a prediction result of a degradation stage of the complex equipment;
a first execution module: the multi-dimensional parameter prediction model is used for calculating predicted values of all parameters according to input online monitoring data of the complex equipment through the first calculation module;
a second execution module: and the prediction value of each parameter obtained by the first execution module is input into the k-means clustering model of the second calculation module to measure the distance between the prediction state and each degradation center, the degradation stage matching is carried out according to the distance between the prediction state and each degradation center, and the prediction result of the degradation stage of the complex equipment is output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211399371.2A CN115718880A (en) | 2022-11-09 | 2022-11-09 | Prediction method for degradation stage of complex equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211399371.2A CN115718880A (en) | 2022-11-09 | 2022-11-09 | Prediction method for degradation stage of complex equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115718880A true CN115718880A (en) | 2023-02-28 |
Family
ID=85255438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211399371.2A Pending CN115718880A (en) | 2022-11-09 | 2022-11-09 | Prediction method for degradation stage of complex equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115718880A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117370870A (en) * | 2023-12-05 | 2024-01-09 | 浙江大学 | Knowledge and data compound driven equipment multi-working condition identification and performance prediction method |
-
2022
- 2022-11-09 CN CN202211399371.2A patent/CN115718880A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117370870A (en) * | 2023-12-05 | 2024-01-09 | 浙江大学 | Knowledge and data compound driven equipment multi-working condition identification and performance prediction method |
| CN117370870B (en) * | 2023-12-05 | 2024-02-20 | 浙江大学 | Knowledge and data compound driven equipment multi-working condition identification and performance prediction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111222290B (en) | Multi-parameter feature fusion-based method for predicting residual service life of large-scale equipment | |
| CN113779496B (en) | Power equipment state evaluation method and system based on equipment panoramic data | |
| CN114048600A (en) | Digital twin-driven multi-model fusion industrial system anomaly detection method | |
| CN107273924B (en) | Multi-data fusion power plant fault diagnosis method based on fuzzy clustering analysis | |
| CN110571792A (en) | Analysis and evaluation method and system for operation state of power grid regulation and control system | |
| CN110757510B (en) | Method and system for predicting remaining life of robot | |
| CN112987666B (en) | Power plant unit operation optimization regulation and control method and system | |
| CN112116198A (en) | Data-driven process industrial state perception network key node screening method | |
| CN114218872B (en) | DBN-LSTM semi-supervised joint model-based residual service life prediction method | |
| CN114282443B (en) | Residual service life prediction method based on MLP-LSTM supervised joint model | |
| CN109240276B (en) | Multi-block PCA fault monitoring method based on fault sensitive principal component selection | |
| Silva et al. | Damage detection in a benchmark structure using AR-ARX models and statistical pattern recognition | |
| Feng et al. | A new time–space attention mechanism driven multi-feature fusion method for tool wear monitoring | |
| CN115587543A (en) | Federal learning and LSTM-based tool residual life prediction method and system | |
| CN113325721A (en) | Model-free adaptive control method and system for industrial system | |
| CN115718880A (en) | Prediction method for degradation stage of complex equipment | |
| Oliveira-Santos et al. | Combining classifiers with decision templates for automatic fault diagnosis of electrical submersible pumps | |
| CN117639602A (en) | Self-adaptive motor running state adjusting method and system | |
| CN113726558A (en) | Network equipment flow prediction system based on random forest algorithm | |
| CN117076869A (en) | Time-frequency domain fusion fault diagnosis method and system for rotary machine | |
| Sallehuddin et al. | Forecasting small data set using hybrid cooperative feature selection | |
| Alimohammadi et al. | Predict the remaining useful life in HVAC filters using a hybrid strategy | |
| CN116595889B (en) | Processing method and system for thin rib uniform distribution structure based on PEEK material | |
| Xu et al. | Multiple-model-based Convolutional Neural Network for Machine Remaining Useful Life Prediction under Multiple Operation Modes | |
| CN117669373A (en) | Energy consumption prediction method and system for hydraulic system of forging forming equipment |
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 |