CN110059377A - A kind of fuel battery service life prediction technique based on depth convolutional neural networks - Google Patents
A kind of fuel battery service life prediction technique based on depth convolutional neural networks Download PDFInfo
- Publication number
- CN110059377A CN110059377A CN201910260003.1A CN201910260003A CN110059377A CN 110059377 A CN110059377 A CN 110059377A CN 201910260003 A CN201910260003 A CN 201910260003A CN 110059377 A CN110059377 A CN 110059377A
- Authority
- CN
- China
- Prior art keywords
- data
- service life
- battery service
- fuel battery
- life prediction
- 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.)
- Granted
Links
Classifications
-
- 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
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Abstract
The present invention discloses a kind of fuel battery service life prediction technique based on depth convolutional neural networks, including concentrates initial data selected characteristic variable as mode input data test data of experiment;Data normalization is carried out to initial data;Sample data is extracted, multiple batches of division is carried out;The parameter of the fuel battery service life prediction model is set, enters convolutional layer later and calculates;After completing the calculating of a convolutional layer, before entering next convolutional layer, the processing of maximum value pondization is made to obtained Feature Mapping matrix;After the calculating of multilayer convolutional layer, global average pond is carried out, and the multilayer feature mapping matrix of single feature is made into full connection operation;It is divided according to batch and is practiced through excessive training in rotation, export prediction result.The present invention can satisfy the demand of fuel battery service life prediction, and predicted time is fast and precision of prediction is high;The fuel cell remaining life Prediction Parameters of high efficient and reliable can be provided for fuel cell management, improve the efficiency of management.
Description
Technical field
The invention belongs to field of fuel cell technology, more particularly to a kind of fuel electricity based on depth convolutional neural networks
Pond life-span prediction method.
Background technique
Although fuel cell has the advantages that cleaning, efficiently etc. multiple, its commercialization process is still in the starting stage, and it
Service life be then an important factor for limiting its large-scale application.For this bad defect of fuel cell durability, prediction
Become the popular method that current fuel cell health state evaluation and concrete surplus working life are predicted with health control technology (PHM).
Health control technology is to pass through the information of prediction or diagnosis, goes out appropriate determine using maintenance action of the Maintenance Resource to next step
Plan, remaining life predict that (RUL) is an indispensable ring in PHM, are formulated according to prediction result system equipment and managed
Strategy can effectively improve the reliability of power supply system.
Method currently used for prediction mainly has data-driven, model-driven and blending algorithm three classes, is based on data-driven
Prediction technique using mass data complete nonlinear fitting, do not need the fuel cell module degradation model of priori, predicting
Aspect has greater advantage.But the prediction algorithm based on data generally uses statistical technique, by algorithm such as particle filter,
The fitting prediction techniques such as local weighted projection recurrence, the precision of prediction is lower, and time cost is high;Carrying out, fuel battery service life is pre-
When survey, due to itself speciality of fuel electronics, the prediction technique of mass data is needed to establish in real time and is capable of handling, and it is current pre-
Survey method is not able to satisfy the demand of fuel battery service life prediction, and predicted time is long and precision of prediction is low, can not be fuel cell
Management provides the fuel cell remaining life Prediction Parameters of high efficient and reliable.
Summary of the invention
To solve the above-mentioned problems, the invention proposes a kind of fuel battery service life based on depth convolutional neural networks is pre-
Survey method can satisfy the demand of fuel battery service life prediction, and predicted time is fast and precision of prediction is high;It can be fuel cell tubes
Reason provides the fuel cell remaining life Prediction Parameters of high efficient and reliable, improves the efficiency of management.
In order to achieve the above objectives, the technical solution adopted by the present invention is that: a kind of fuel based on depth convolutional neural networks
Battery life predicting method, comprising steps of
S100 obtains test data of experiment collection and is pre-processed;
S200, constructs according to test data of experiment collection and fuel battery service life of the training based on depth convolutional neural networks is pre-
Survey model;By the way that test data is inputted the fuel battery service life prediction model by training, prediction result is exported;
The training and prediction process of the fuel battery service life prediction model, comprising steps of
S201 concentrates initial data selected characteristic variable as mode input data test data of experiment;
S202 pre-processes initial data, carries out data normalization;
S203, according to treated, initial data extracts sample data, carries out multiple batches of division;
S204, the parameter that the fuel battery service life prediction model is arranged include convolution nuclear volume, convolution kernel length and swash
Function living enters convolutional layer later and calculates;
S205, after completing the calculating of a convolutional layer, before entering next convolutional layer, to obtained Feature Mapping
Matrix makees the processing of maximum value pondization;
S206 carries out global average pond, and the multilayer feature of single feature is mapped after the calculating of multilayer convolutional layer
Matrix makees full connection operation;
S207 is divided according to batch and is practiced through excessive training in rotation, exports prediction result.
Further, access time variable parameter include fuel cell pile air inlet/outlet Hydrogen Vapor Pressure and temperature,
Fuel cell pile air inlet/outlet air themperature, fuel cell pile inlet and outlet cooling water temperature and fuel cell pile electric current are close
Degree is used as input data;Setting exports electricity using fuel cell pile voltage and time as output data with fuel cell pile
The performance degradation index as fuel cell pile is pressed, to predict the timing node of beginning to failure threshold timing node as in fact
The remaining life on border, to predict fuel cell remaining life.
Further, dimension lesser variable value in part is larger since the dimension between variable is different, it will weaken
The effect of the biggish fractional value variable of dimension, in preprocessing process, at the normalization of test data of experiment collection initial data
Reason can also facilitate the comparison and weighting between different variables using progress nondimensionalization processing;
The nondimensionalization processing uses Min-Max standardized calculation method, conversion formula are as follows:
Wherein, x is initial data, xstdFor the data after standardization, xmaxFor the maximum value in initial data, xminFor original
Minimum value in beginning data.Both the Weighted problem for having avoided different dimension variables, also can be improved the training speed of model.
Further, by treated, initial data is divided into several segments, often during the extraction sample data
One piece of data is a batch data;
A batch process once is trained for a batch data completion, primary training is completed to all batch datas and is known as
One wheel is conducive to improve training speed and computational accuracy.
Further, the fuel battery service life prediction model includes 4 convolutional layers;Wherein, preceding 3 convolutional layers are used for
Feature extraction, after obtained Feature Mapping matrix is extracted two dimensional character mapping matrix with single convolution kernel by the last one convolutional layer
The multilayer feature mapping matrix of single feature is obtained, the multilayer feature mapping matrix of single feature is made into full connection operation.
Further, input data passes through convolution, nonlinear activation letter in the fuel battery service life prediction model
Number mapping and Chi Huaquan attended operation, it is successively abstract;By input data by feed forward operation, by advanced features successively from original
It is stripped out in data;After full articulamentum, according to the difference of goal task, loss function is chosen, calculates output data
Error between sample data;And the error is carried out to layer-by-layer backpropagation by back-propagation algorithm, constantly more
New each layer of parameter;A wheel parameter and then secondary carry out feed forward operation are being updated, and so on, until network model is received
It holds back, model training finishes at this time.
Further, in the feed forward operation, by convolution kernel in convolutional layer from left to right, mobile time from top to bottom
Go through input;The equal dimensional matrix of each of convolution kernel and input data carry out discrete convolution, and the numerical value that each convolution obtains is led to
It crosses activation primitive and carries out Nonlinear Mapping, and be stored in Feature Mapping matrix, the size of Feature Mapping matrix depends on traversal
Mobile number needed for entire input matrix;By multiple convolution kernels, multiple Feature Mapping matrixes are superimposed, multilayer feature is obtained
Mapping matrix, the input as next convolutional layer.
Further, the fuel battery service life prediction model by the error being calculated by layer-by-layer backpropagation,
By error averaging to each convolutional layer, the weight matrix parameter and offset parameter of more new model make minimization of loss;It completes
After the backpropagation of error, the feed forward operation is repeated, until network convergence, network losses in model is made to be reduced to permission
Within accuracy rating.
It include band in each convolution kernel further, being stacked in each described convolutional layer by multiple convolution kernels
There is the two-dimensional matrix of fixed weight neuron, data processing is in the data stored in the two-dimensional matrix i.e. receptive field
Number;Convolution kernel is enabled to carry out feature extraction to all data in the peripheral zero padding of input data matrix.Utilize convolution kernel
Local receptor field and weight are shared, and local receptor field takes the entire input data set of traversal using the lesser convolution kernel of dimension,
The object of input data convolution is all the same convolution kernel in ergodic process, i.e. the weight matrix of participation convolution algorithm is all
It is consistent, it can be effectively reduced in deep neural network need trained number of parameters in this way, to more huge data
Collection is when being trained, and can effectively shorten the training time, reduces hsrdware requirements.
Further, the quantity of the convolution kernel is set as 16, since fuel cell operation data are Multivariate Time
Sequence, therefore be essentially the splicing of multiple One-dimension Time Series, the size of each convolution kernel is 5 × 1 here, by convolution
The step-length of core sliding is set as 1;Activation primitive uses Relu function.
Using the technical program the utility model has the advantages that
The present invention is by fuel battery service life prediction model of the building based on depth convolutional neural networks, to collected combustion
Material battery operation data carry out layer-by-layer feature extraction, by the backpropagation of error, constantly update network parameter;It can obtain
High-precision forecast fuel cell output voltage and its time for reaching failure threshold, using fuel cell pile output voltage as property
Can degenerate index, using the timing node of prediction beginning to failure threshold timing node as actual remaining life, thus
Predict fuel cell remaining life;So that the prediction result of fuel cell remaining life is more accurate, calculated
Cheng Gengjia is efficient, can the remaining life to fuel cell predicted in real time.It provides for fuel cell management and efficiently may be used
The fuel cell remaining life Prediction Parameters leaned on, improve the efficiency of management, provide for the reliable and stable operation of fuel cell
Powerful support.
Detailed description of the invention
Fig. 1 is that a kind of process of fuel battery service life prediction technique based on depth convolutional neural networks of the invention is illustrated
Figure;
Fig. 2 is the training of fuel battery service life prediction model in the embodiment of the present invention and the flow diagram of prediction process
Fig. 3 is the structural schematic diagram of fuel battery service life prediction model in the embodiment of the present invention;
Fig. 4 is predicted voltage and virtual voltage comparison of wave shape figure in the prediction result of the embodiment of the present invention;
Fig. 5 is predicted voltage error map in the prediction result of the embodiment of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention is made into one with reference to the accompanying drawing
Step illustrates.
In the present embodiment, referring to shown in Fig. 1 and 2, the invention proposes a kind of combustions based on depth convolutional neural networks
Expect battery life predicting method, comprising steps of
S100 obtains test data of experiment collection and is pre-processed;
S200, constructs according to test data of experiment collection and fuel battery service life of the training based on depth convolutional neural networks is pre-
Survey model;By the way that test data is inputted the fuel battery service life prediction model by training, prediction result is exported;
The training and prediction process of the fuel battery service life prediction model, comprising steps of
S201 concentrates initial data selected characteristic variable as mode input data test data of experiment;
S202 pre-processes initial data, carries out data normalization;
S203, according to treated, initial data extracts sample data, carries out multiple batches of division;
S204, the parameter that the fuel battery service life prediction model is arranged include convolution nuclear volume, convolution kernel length and swash
Function living enters convolutional layer later and calculates;
S205, after completing the calculating of a convolutional layer, before entering next convolutional layer, to obtained Feature Mapping
Matrix makees the processing of maximum value pondization;
S206 carries out global average pond, and the multilayer feature of single feature is mapped after the calculating of multilayer convolutional layer
Matrix makees full connection operation;
S207 is divided according to batch and is practiced through excessive training in rotation, exports prediction result.
As the prioritization scheme of above-described embodiment, access time variable parameter includes the hydrogen of fuel cell pile air inlet/outlet
Atmospheric pressure and temperature, fuel cell pile air inlet/outlet air themperature, fuel cell pile inlet and outlet cooling water temperature and fuel
Battery stack current density is as input data;Setting is using fuel cell pile voltage and time as output data, with fuel
Performance degradation index of the battery stack output voltage as fuel cell pile, to predict the timing node started to failure threshold
Timing node is as actual remaining life, to predict fuel cell remaining life.
Since the dimension between variable is different, dimension lesser variable value in part is larger, it will it is biggish to weaken dimension
The effect of fractional value variable uses the normalized of test data of experiment collection initial data and carries out in preprocessing process
Nondimensionalization processing, while the comparison and weighting between different variables can also be facilitated;
The nondimensionalization processing uses Min-Max standardized calculation method, conversion formula are as follows:
Wherein, x is initial data, xstdFor the data after standardization, xmaxFor the maximum value in initial data, xminFor original
Minimum value in beginning data.Both the Weighted problem for having avoided different dimension variables, also can be improved the training speed of model.
It, will treated initial data during the extraction sample data as the prioritization scheme of above-described embodiment
Several segments are divided into, every one piece of data is a batch data;
A batch process once is trained for a batch data completion, primary training is completed to all batch datas and is known as
One wheel is conducive to improve training speed and computational accuracy.
As the prioritization scheme of above-described embodiment, as shown in figure 3, the fuel battery service life prediction model includes 4 volumes
Lamination;Wherein, preceding 3 convolutional layers are used for feature extraction, the Feature Mapping that the last one convolutional layer will be obtained with single convolution kernel
Matrix obtains the multilayer feature mapping matrix of single feature after extracting two dimensional character mapping matrix, by the multilayer feature of single feature
Mapping matrix makees full connection operation.
In the fuel battery service life prediction model, input data by convolution, nonlinear activation function mapping and
Chi Huaquan attended operation, it is successively abstract;By input data by feed forward operation, advanced features are successively removed from initial data
Out;After full articulamentum, according to the difference of goal task, loss function is chosen, calculates output data and sample data
Between error;And the error is carried out to layer-by-layer backpropagation by back-propagation algorithm, constantly update each layer
Parameter;A wheel parameter and then secondary carry out feed forward operation is being updated, and so on, until network model is restrained, model at this time
Training finishes.
In the feed forward operation, by convolution kernel in convolutional layer, mobile traversal is inputted from left to right, from top to bottom;Convolution
The equal dimensional matrix of each of core and input data carry out discrete convolution, the numerical value that each convolution is obtained by activation primitive into
Row Nonlinear Mapping, and be stored in Feature Mapping matrix, the size of Feature Mapping matrix depends on traversing entire input matrix
Required mobile number;By multiple convolution kernels, multiple Feature Mapping matrixes are superimposed, multilayer feature mapping matrix is obtained, made
For the input of next convolutional layer.
The fuel battery service life prediction model by the error being calculated by layer-by-layer backpropagation, extremely by error averaging
Each convolutional layer, the weight matrix parameter and offset parameter of more new model make minimization of loss;Complete the reversed biography of error
After broadcasting, repeat the feed forward operation, until network convergence, make network losses in model be reduced to permission accuracy rating it
It is interior.
As the prioritization scheme of above-described embodiment, stacked in each described convolutional layer by multiple convolution kernels, it is each described
It include the two-dimensional matrix with fixed weight neuron in convolution kernel, in the data stored in the two-dimensional matrix i.e. receptive field
The coefficient of data processing;Convolution kernel to carry out feature to all data in the peripheral zero padding of input data matrix to mention
It takes.Shared using convolution kernel local receptor field and weight, local receptor field takes traversal entire using the lesser convolution kernel of dimension
Input data set, the object of input data convolution is all the same convolution kernel in an ergodic process, i.e. participation convolution algorithm
Weight matrix be all consistent, can be effectively reduced in deep neural network need trained number of parameters in this way, right
When more huge data set is trained, the training time can be effectively shortened, reduce hsrdware requirements.
The quantity of the convolution kernel is set as 16, since fuel cell operation data are Multivariate Time Series, this
It is the splicing of multiple One-dimension Time Series in matter, the size of each convolution kernel is 5 × 1 here, the step that convolution kernel is slided
Length is set as 1;Activation primitive uses Relu function.
The fuel battery service life prediction model described in Python platform construction is simultaneously trained it, carries out in Python
The curve comparison of the training and prediction of model, obtained predicted voltage and practical pile output voltage is as shown in figure 4, from Fig. 4
As can be seen that predicted voltage and virtual voltage coincide substantially, losing is 0.003, is maintained in lesser error range.
Average value processing is made to the voltage prediction deviation of every 300 data sampled points, is represented in the sampling area with the mean value
Voltage prediction deviation.From figure 5 it can be seen that the value that predicted voltage deviates virtual voltage maintains a lower number always
Value is horizontal, and maximum relative error is 5.41%, and the relative error of entire test data set is in precision allowed band.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.The claimed scope of the invention by appended claims and its
Equivalent thereof.
Claims (10)
1. a kind of fuel battery service life prediction technique based on depth convolutional neural networks, which is characterized in that comprising steps of
S100 obtains test data of experiment collection and is pre-processed;
S200, constructs according to test data of experiment collection and fuel battery service life of the training based on depth convolutional neural networks predicts mould
Type;By the way that test data is inputted the fuel battery service life prediction model by training, prediction result is exported;
The training and prediction process of the fuel battery service life prediction model, comprising steps of
S201 concentrates initial data selected characteristic variable as mode input data test data of experiment;
S202 pre-processes initial data, carries out data normalization;
S203, according to treated, initial data extracts sample data, carries out multiple batches of division;
S204, the parameter that the fuel battery service life prediction model is arranged includes convolution nuclear volume, convolution kernel length and activation letter
Number enters convolutional layer later and calculates;
S205, after completing the calculating of a convolutional layer, before entering next convolutional layer, to obtained Feature Mapping matrix
Make the processing of maximum value pondization;
S206 carries out global average pond after the calculating of multilayer convolutional layer, and by the multilayer feature mapping matrix of single feature
Make full connection operation;
S207 is divided according to batch and is practiced through excessive training in rotation, exports prediction result.
2. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 1, special
Sign is that access time variable parameter includes the Hydrogen Vapor Pressure and temperature, fuel cell pile of fuel cell pile air inlet/outlet
Air inlet/outlet air themperature, fuel cell pile inlet and outlet cooling water temperature and fuel cell pile current density are as input number
According to;Setting is electric using fuel cell pile output voltage as fuel using fuel cell pile voltage and time as output data
The performance degradation index of pond pile is used using the timing node for predicting to start to failure threshold timing node as actual residue
Service life, to predict fuel cell remaining life.
3. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 2, special
Sign is, in preprocessing process, uses and is carried out at nondimensionalization to the normalized of test data of experiment collection initial data
Reason;
The nondimensionalization processing uses Min-Max standardized calculation method, conversion formula are as follows:
Wherein, x is initial data, xstdFor the data after standardization, xmaxFor the maximum value in initial data, xminFor original number
Minimum value in.
4. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 3, special
Sign is, during the extraction sample data, by treated, initial data is divided into several segments, and every one piece of data is one
A batch data;
A batch process once is trained for a batch data completion, primary training is completed to all batch datas and is known as one
Wheel.
5. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 4, special
Sign is that the fuel battery service life prediction model includes 4 convolutional layers;Wherein, preceding 3 convolutional layers are used for feature extraction, most
The latter convolutional layer obtains single spy after obtained Feature Mapping matrix is extracted two dimensional character mapping matrix with single convolution kernel
The multilayer feature mapping matrix of single feature is made full connection operation by the multilayer feature mapping matrix of sign.
6. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 1, special
Sign is, in the fuel battery service life prediction model, input data passes through convolution, nonlinear activation function mapping and pond
Change full attended operation, it is successively abstract;By input data by feed forward operation, advanced features are successively separated from initial data
Come;After full articulamentum, according to the difference of goal task, choose loss function, calculate output data and sample data it
Between error;And the error is carried out to layer-by-layer backpropagation by back-propagation algorithm, constantly update each layer of ginseng
Number;A wheel parameter and then secondary carry out feed forward operation are being updated, and so on, until network model is restrained, model is instructed at this time
White silk finishes.
7. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 6, special
Sign is, in the feed forward operation, by convolution kernel in convolutional layer, mobile traversal is inputted from left to right, from top to bottom;Convolution
The equal dimensional matrix of each of core and input data carry out discrete convolution, the numerical value that each convolution is obtained by activation primitive into
Row Nonlinear Mapping, and be stored in Feature Mapping matrix, the size of Feature Mapping matrix depends on traversing entire input matrix
Required mobile number;By multiple convolution kernels, multiple Feature Mapping matrixes are superimposed, multilayer feature mapping matrix is obtained, made
For the input of next convolutional layer.
8. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 7, special
Sign is, the fuel battery service life prediction model by the error being calculated by layer-by-layer backpropagation, extremely by error averaging
Each convolutional layer, the weight matrix parameter and offset parameter of more new model make minimization of loss;Complete the reversed biography of error
After broadcasting, repeat the feed forward operation, until network convergence, make network losses in model be reduced to permission accuracy rating it
It is interior.
9. any a kind of fuel battery service life prediction side based on depth convolutional neural networks in -8 according to claim 1
Method, which is characterized in that stacked in each described convolutional layer by multiple convolution kernels, include with fixation in each convolution kernel
The two-dimensional matrix of weight neuron, the coefficient of data processing in the data stored in the two-dimensional matrix i.e. receptive field;Defeated
The peripheral zero padding for entering data matrix enables convolution kernel to carry out feature extraction to all data.
10. a kind of fuel battery service life prediction technique based on depth convolutional neural networks according to claim 9, special
Sign is that the quantity of the convolution kernel is set as 16, and the size of each convolution kernel is 5 × 1, the step-length that convolution kernel is slided
It is set as 1;Activation primitive uses Relu function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910260003.1A CN110059377B (en) | 2019-04-02 | 2019-04-02 | Fuel cell life prediction method based on deep convolutional neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910260003.1A CN110059377B (en) | 2019-04-02 | 2019-04-02 | Fuel cell life prediction method based on deep convolutional neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110059377A true CN110059377A (en) | 2019-07-26 |
CN110059377B CN110059377B (en) | 2022-07-05 |
Family
ID=67318162
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910260003.1A Active CN110059377B (en) | 2019-04-02 | 2019-04-02 | Fuel cell life prediction method based on deep convolutional neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110059377B (en) |
Cited By (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110927584A (en) * | 2019-12-09 | 2020-03-27 | 天津市捷威动力工业有限公司 | Neural network-based battery life extension prediction method |
CN111144021A (en) * | 2019-12-30 | 2020-05-12 | 新源动力股份有限公司 | Fuel cell service life prediction method and system |
CN111310387A (en) * | 2020-02-19 | 2020-06-19 | 电子科技大学 | Fuel cell life prediction method |
CN111598218A (en) * | 2020-05-08 | 2020-08-28 | Oppo广东移动通信有限公司 | Data processing method and device based on convolutional neural network and storage medium |
CN111639749A (en) * | 2020-05-25 | 2020-09-08 | 上海智殷自动化科技有限公司 | Industrial robot friction force identification method based on deep learning |
CN112033463A (en) * | 2020-09-02 | 2020-12-04 | 哈尔滨工程大学 | Nuclear power equipment state evaluation and prediction integrated method and system |
CN112069741A (en) * | 2020-09-25 | 2020-12-11 | 国网四川省电力公司经济技术研究院 | Fuel cell stack transient aging prediction method, device, equipment and storage medium |
CN112133941A (en) * | 2020-10-10 | 2020-12-25 | 西南交通大学 | Rapid fault diagnosis method for locomotive proton exchange membrane fuel cell system |
CN112580263A (en) * | 2020-12-24 | 2021-03-30 | 湖南工业大学 | Turbofan engine residual service life prediction method based on space-time feature fusion |
CN112666479A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery service life prediction method based on charging cycle fusion |
CN112666480A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery life prediction method based on charging process characteristic attention |
CN112731183A (en) * | 2020-12-21 | 2021-04-30 | 首都师范大学 | Lithium ion battery life prediction method based on improved ELM |
CN112782584A (en) * | 2019-11-05 | 2021-05-11 | 奥动新能源汽车科技有限公司 | Method, system, medium, and device for predicting remaining usage limit of battery power |
DE102020202881A1 (en) | 2020-03-06 | 2021-09-09 | Robert Bosch Gesellschaft mit beschränkter Haftung | Method for monitoring the operation of at least one fuel cell device and fuel cell device |
DE102020202878A1 (en) | 2020-03-06 | 2021-09-09 | Robert Bosch Gesellschaft mit beschränkter Haftung | Method for monitoring the operation of at least one fuel cell device and fuel cell device |
CN113406439A (en) * | 2021-06-23 | 2021-09-17 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Power distribution network fault positioning method |
CN113406505A (en) * | 2021-07-22 | 2021-09-17 | 中国第一汽车股份有限公司 | Method and device for predicting residual life of fuel cell |
CN113657024A (en) * | 2021-07-22 | 2021-11-16 | 武汉理工大学 | Fuel cell life prediction method |
CN113657012A (en) * | 2021-07-21 | 2021-11-16 | 西安理工大学 | TCN and particle filter-based method for predicting residual life of key equipment |
CN113777496A (en) * | 2021-09-06 | 2021-12-10 | 北京化工大学 | Lithium ion battery residual life prediction method based on time convolution neural network |
CN113779750A (en) * | 2021-07-22 | 2021-12-10 | 广东劲天科技有限公司 | Battery life prediction method and system based on charging state and charging pile |
CN114683964A (en) * | 2022-03-29 | 2022-07-01 | 北京芯虹科技有限责任公司 | Battery state information determining method and charging equipment |
CN116432543A (en) * | 2023-06-13 | 2023-07-14 | 湖南大学 | Method for predicting remaining life of power semiconductor module, terminal device and storage medium |
CN116544458A (en) * | 2023-07-04 | 2023-08-04 | 国家电投集团氢能科技发展有限公司 | Fault warning method and device for fuel cell system |
CN117031307A (en) * | 2023-08-17 | 2023-11-10 | 广州汽车集团股份有限公司 | Storage battery life prediction method and device and storage medium thereof |
CN117374469A (en) * | 2023-12-07 | 2024-01-09 | 深圳市普裕时代新能源科技有限公司 | Control method based on immersed liquid cooling energy storage system |
CN114683964B (en) * | 2022-03-29 | 2024-04-26 | 北京芯虹科技有限责任公司 | Battery state information determining method and charging equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106786606A (en) * | 2017-03-17 | 2017-05-31 | 西南交通大学 | A kind of computational methods of the Probabilistic Load based on various stochastic variables |
CN108287317A (en) * | 2018-01-24 | 2018-07-17 | 上海电气分布式能源科技有限公司 | Power of battery prediction model generation method and system, power forecasting method and system |
CN108647716A (en) * | 2018-05-09 | 2018-10-12 | 北京理工大学 | A kind of diagnosing failure of photovoltaic array method based on composite information |
CN108830188A (en) * | 2018-05-30 | 2018-11-16 | 西安理工大学 | Vehicle checking method based on deep learning |
CN108932567A (en) * | 2018-08-10 | 2018-12-04 | 燕山大学 | A kind of more energy consumption index prediction techniques of cement burning assembly procedure based on convolutional neural networks |
WO2019021095A1 (en) * | 2017-07-26 | 2019-01-31 | 株式会社半導体エネルギー研究所 | System for controlling charging of secondary cell and method for detecting abnormality in secondary cell |
-
2019
- 2019-04-02 CN CN201910260003.1A patent/CN110059377B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106786606A (en) * | 2017-03-17 | 2017-05-31 | 西南交通大学 | A kind of computational methods of the Probabilistic Load based on various stochastic variables |
WO2019021095A1 (en) * | 2017-07-26 | 2019-01-31 | 株式会社半導体エネルギー研究所 | System for controlling charging of secondary cell and method for detecting abnormality in secondary cell |
CN108287317A (en) * | 2018-01-24 | 2018-07-17 | 上海电气分布式能源科技有限公司 | Power of battery prediction model generation method and system, power forecasting method and system |
CN108647716A (en) * | 2018-05-09 | 2018-10-12 | 北京理工大学 | A kind of diagnosing failure of photovoltaic array method based on composite information |
CN108830188A (en) * | 2018-05-30 | 2018-11-16 | 西安理工大学 | Vehicle checking method based on deep learning |
CN108932567A (en) * | 2018-08-10 | 2018-12-04 | 燕山大学 | A kind of more energy consumption index prediction techniques of cement burning assembly procedure based on convolutional neural networks |
Non-Patent Citations (3)
Title |
---|
JIAWEI LIU等: "Remaining useful life prediction of PEMFC based on long short-term memory recurrent neural networks", 《INTERNATIONAL JOURNAL OF HYDROGEN ENERGY》 * |
XIANG LI 等: "Remaining useful life estimation in prognostics using deep convolution neural networks", 《RELIABILITY ENGINEERING AND SYSTEM SAFETY》 * |
吴涛 等: "基于LeNet-5 模型的太阳能电池板缺陷识别分类", 《热力发电》 * |
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112782584A (en) * | 2019-11-05 | 2021-05-11 | 奥动新能源汽车科技有限公司 | Method, system, medium, and device for predicting remaining usage limit of battery power |
CN110927584A (en) * | 2019-12-09 | 2020-03-27 | 天津市捷威动力工业有限公司 | Neural network-based battery life extension prediction method |
CN110927584B (en) * | 2019-12-09 | 2022-05-10 | 天津市捷威动力工业有限公司 | Neural network-based battery life extension prediction method |
CN111144021A (en) * | 2019-12-30 | 2020-05-12 | 新源动力股份有限公司 | Fuel cell service life prediction method and system |
CN111144021B (en) * | 2019-12-30 | 2023-08-01 | 新源动力股份有限公司 | Fuel cell life prediction method and system |
CN111310387A (en) * | 2020-02-19 | 2020-06-19 | 电子科技大学 | Fuel cell life prediction method |
DE102020202878A1 (en) | 2020-03-06 | 2021-09-09 | Robert Bosch Gesellschaft mit beschränkter Haftung | Method for monitoring the operation of at least one fuel cell device and fuel cell device |
DE102020202881A1 (en) | 2020-03-06 | 2021-09-09 | Robert Bosch Gesellschaft mit beschränkter Haftung | Method for monitoring the operation of at least one fuel cell device and fuel cell device |
CN111598218A (en) * | 2020-05-08 | 2020-08-28 | Oppo广东移动通信有限公司 | Data processing method and device based on convolutional neural network and storage medium |
CN111598218B (en) * | 2020-05-08 | 2024-04-16 | Oppo广东移动通信有限公司 | Image data processing method and device based on convolutional neural network and storage medium |
CN111639749A (en) * | 2020-05-25 | 2020-09-08 | 上海智殷自动化科技有限公司 | Industrial robot friction force identification method based on deep learning |
CN112033463B (en) * | 2020-09-02 | 2022-09-06 | 哈尔滨工程大学 | Nuclear power equipment state evaluation and prediction integrated method and system |
CN112033463A (en) * | 2020-09-02 | 2020-12-04 | 哈尔滨工程大学 | Nuclear power equipment state evaluation and prediction integrated method and system |
CN112069741A (en) * | 2020-09-25 | 2020-12-11 | 国网四川省电力公司经济技术研究院 | Fuel cell stack transient aging prediction method, device, equipment and storage medium |
CN112133941A (en) * | 2020-10-10 | 2020-12-25 | 西南交通大学 | Rapid fault diagnosis method for locomotive proton exchange membrane fuel cell system |
CN112666479A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery service life prediction method based on charging cycle fusion |
CN112666480A (en) * | 2020-12-02 | 2021-04-16 | 西安交通大学 | Battery life prediction method based on charging process characteristic attention |
CN112666480B (en) * | 2020-12-02 | 2023-04-28 | 西安交通大学 | Battery life prediction method based on characteristic attention of charging process |
CN112731183A (en) * | 2020-12-21 | 2021-04-30 | 首都师范大学 | Lithium ion battery life prediction method based on improved ELM |
CN112731183B (en) * | 2020-12-21 | 2023-04-21 | 首都师范大学 | Improved ELM-based lithium ion battery life prediction method |
CN112580263A (en) * | 2020-12-24 | 2021-03-30 | 湖南工业大学 | Turbofan engine residual service life prediction method based on space-time feature fusion |
CN113406439A (en) * | 2021-06-23 | 2021-09-17 | 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 | Power distribution network fault positioning method |
CN113657012A (en) * | 2021-07-21 | 2021-11-16 | 西安理工大学 | TCN and particle filter-based method for predicting residual life of key equipment |
CN113406505A (en) * | 2021-07-22 | 2021-09-17 | 中国第一汽车股份有限公司 | Method and device for predicting residual life of fuel cell |
CN113657024A (en) * | 2021-07-22 | 2021-11-16 | 武汉理工大学 | Fuel cell life prediction method |
CN113779750A (en) * | 2021-07-22 | 2021-12-10 | 广东劲天科技有限公司 | Battery life prediction method and system based on charging state and charging pile |
CN113777496A (en) * | 2021-09-06 | 2021-12-10 | 北京化工大学 | Lithium ion battery residual life prediction method based on time convolution neural network |
CN113777496B (en) * | 2021-09-06 | 2023-10-24 | 北京化工大学 | Lithium ion battery residual life prediction method based on time convolution neural network |
CN114683964A (en) * | 2022-03-29 | 2022-07-01 | 北京芯虹科技有限责任公司 | Battery state information determining method and charging equipment |
CN114683964B (en) * | 2022-03-29 | 2024-04-26 | 北京芯虹科技有限责任公司 | Battery state information determining method and charging equipment |
CN116432543A (en) * | 2023-06-13 | 2023-07-14 | 湖南大学 | Method for predicting remaining life of power semiconductor module, terminal device and storage medium |
CN116432543B (en) * | 2023-06-13 | 2023-09-05 | 湖南大学 | Method for predicting remaining life of power semiconductor module, terminal device and storage medium |
CN116544458A (en) * | 2023-07-04 | 2023-08-04 | 国家电投集团氢能科技发展有限公司 | Fault warning method and device for fuel cell system |
CN117031307A (en) * | 2023-08-17 | 2023-11-10 | 广州汽车集团股份有限公司 | Storage battery life prediction method and device and storage medium thereof |
CN117374469A (en) * | 2023-12-07 | 2024-01-09 | 深圳市普裕时代新能源科技有限公司 | Control method based on immersed liquid cooling energy storage system |
CN117374469B (en) * | 2023-12-07 | 2024-02-09 | 深圳市普裕时代新能源科技有限公司 | Control method based on immersed liquid cooling energy storage system |
Also Published As
Publication number | Publication date |
---|---|
CN110059377B (en) | 2022-07-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110059377A (en) | A kind of fuel battery service life prediction technique based on depth convolutional neural networks | |
CN104573879A (en) | Photovoltaic power station output predicting method based on optimal similar day set | |
Semeraro et al. | Digital twin application in energy storage: Trends and challenges | |
CN108287317A (en) | Power of battery prediction model generation method and system, power forecasting method and system | |
CN107423839A (en) | A kind of method of the intelligent building microgrid load prediction based on deep learning | |
CN109446028B (en) | Method for monitoring state of refrigeration dehumidifier based on genetic fuzzy C-mean clustering | |
CN113837464A (en) | Load prediction method of cogeneration boiler based on CNN-LSTM-Attention | |
CN114035098A (en) | Lithium battery health state prediction method integrating future working condition information and historical state information | |
CN115656857A (en) | Multi-scale fusion prediction method for remaining service life of lithium ion battery | |
CN110782071A (en) | Method for predicting wind power by convolutional neural network based on time-space characteristic fusion | |
CN110738429A (en) | electric energy meter state evaluation method and device | |
Liu et al. | Research on the strategy of locating abnormal data in IOT management platform based on improved modified particle swarm optimization convolutional neural network algorithm | |
CN116611702A (en) | Integrated learning photovoltaic power generation prediction method for building integrated energy management | |
CN114636212B (en) | GRNN-based running control method for multiple water chilling unit systems | |
CN116819372A (en) | Method for calculating health state of energy storage battery by neural network | |
CN116842337A (en) | Transformer fault diagnosis method based on LightGBM (gallium nitride based) optimal characteristics and COA-CNN (chip on board) model | |
CN116449218A (en) | Lithium battery health state estimation method | |
CN113959071B (en) | Centralized water chilling unit air conditioning system operation control optimization method based on machine learning assistance | |
Li et al. | Optimizing combination of aircraft maintenance tasks by adaptive genetic algorithm based on cluster search | |
CN112821456B (en) | Distributed source-storage-load matching method and device based on transfer learning | |
Udeogu et al. | Remaining useful life prediction for supercapacitors using an optimized end-to-end deep learning approach | |
CN114254828A (en) | Power load prediction method based on hybrid convolution feature extractor and GRU | |
CN111815085B (en) | Method for realizing comprehensive energy efficiency evaluation of rail transit energy management system | |
CN113780686A (en) | Distributed power supply-oriented virtual power plant operation scheme optimization method | |
Wei et al. | Predicting Electricity Usage Based on Deep Neural Network |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |