CN112883638A - Online estimation method for temperature distribution of super capacitor module - Google Patents

Online estimation method for temperature distribution of super capacitor module Download PDF

Info

Publication number
CN112883638A
CN112883638A CN202110140366.9A CN202110140366A CN112883638A CN 112883638 A CN112883638 A CN 112883638A CN 202110140366 A CN202110140366 A CN 202110140366A CN 112883638 A CN112883638 A CN 112883638A
Authority
CN
China
Prior art keywords
temperature
module
loss
super capacitor
neural network
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
Application number
CN202110140366.9A
Other languages
Chinese (zh)
Other versions
CN112883638B (en
Inventor
韦莉
张婉婷
吴铭
闫孟迪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN202110140366.9A priority Critical patent/CN112883638B/en
Publication of CN112883638A publication Critical patent/CN112883638A/en
Application granted granted Critical
Publication of CN112883638B publication Critical patent/CN112883638B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/13—Energy storage using capacitors

Abstract

The invention relates to an online estimation method for temperature distribution of a super capacitor module, which can realize online estimation of the temperature of all monomers in the module by measuring the temperature of a few monomers in the module. The method avoids using a large number of temperature sensors, not only greatly reduces the hardware cost, but also solves the problem that the temperature sensors are not easy to install in partial areas inside the module. Meanwhile, the algorithm occupies less computing resources, is suitable for online estimation and can meet higher computing precision. The method has important significance for heat management optimization and safe operation of the super capacitor.

Description

Online estimation method for temperature distribution of super capacitor module
Technical Field
The invention relates to an online estimation method for temperature distribution of a super capacitor module.
Background
The super capacitor is used as a novel energy storage device, and has wide application prospect in the fields of renewable clean energy power generation, energy-saving operation of electric automobiles, urban rails, ships and the like. Along with the work of the super capacitor, the accumulation of the internal temperature of the super capacitor can continuously raise the temperature of the monomer, influence the electrical characteristics of the super capacitor, cause the performance to be reduced, accelerate the aging of the super capacitor and further shorten the service life of the super capacitor. The thermal runaway of the whole system can be caused after the temperature of the super capacitor is too high or is out of control, so that the online estimation of the temperature of the super capacitor is particularly important, the management of the super capacitor is more accurate and efficient, and the safety of an energy storage system is improved.
With the wide application of the super capacitor energy storage system, the temperature of each super capacitor monomer needs to be monitored on line in more and more application occasions. However, in practical applications, the temperature sensor is limited by the problems of cost, installation and the like, and often only detects the temperature of a few monomers in the module, and cannot monitor the temperature of each monomer. In addition, finite element thermal simulation software is generally used for researching the temperature distribution of the module, but the method is complex in model, large in calculation amount and long in calculation time, and is not beneficial to online operation.
Disclosure of Invention
Aiming at the defects of the conventional supercapacitor module temperature estimation method, the invention provides the supercapacitor module temperature online estimation method which can effectively utilize a small amount of monomer temperature to estimate all the monomer temperatures in the module and has a simple calculation process. The method combines the characteristics of monomer arrangement in the module, optimizes and selects the installation position of the temperature sensor, considers the correlation among monomers of the super capacitor module and establishes a neural network model for online temperature estimation. The method is realized by the following steps as shown in figure 1:
the method comprises the following steps: the method comprises the following steps of (1) taking the temperature correlation of monomers in a super capacitor module into consideration, and partitioning the module by combining the arrangement characteristics of the monomers;
step two: selecting a tested monomer according to the subareas for testing to obtain temperature and current data of the selected monomer;
step three: constructing a neural network model, designing a training algorithm, training and optimizing the neural network model to obtain an estimation model, and inputting measured data into the temperature online estimation model for calculation;
step four: and outputting the estimated value of each monomer temperature in the module.
In the third step, the neural network model is trained by adopting an algorithm based on Levenberg-Marquardt, and the training process is as follows:
(1) initializing neural network parameters θ0Updating the step length lambda and the step length reduction coefficient beta according to the maximum iteration number K, wherein the current iteration number K is 0, the training error limit epsilon and the early-stop initial value P are 0, and early-stop tolerance coefficient P is obtained;
(2) selecting Sigmoid function as activation function
Figure BDA0002928538390000021
Computational neural network forward propagation
Figure BDA0002928538390000022
Wherein the content of the first and second substances,
Figure BDA0002928538390000023
value, K, representing the ith neuron of the mth layerm-1Represents the number of neurons in the m-1 layer,
Figure BDA0002928538390000024
represents the weight of the jth input neuron of the (m-1) th layer to the ith neuron of the m layer,
Figure BDA0002928538390000025
is the bias term for the ith neuron in the mth layer. Selecting a loss function of
Figure BDA0002928538390000026
Wherein the content of the first and second substances,
Figure BDA0002928538390000027
is the square of the error of sample i, and l is the number of samples; calculating the losskIf losskIf ∈ or K ═ K or P ═ P, then step (5) is carried out;
(3) according to the formula theta-lambda [ JT(θ)J(θ)+μI]-1JT(theta) e (theta) calculating network parameters, wherein mu is a constant larger than zero, I is an identity matrix, and J is a Jacobian matrix; furthermore, the utility modelObtaining new parameters
Figure BDA0002928538390000031
And calculating the loss value loss of the model after updating the one-step parameterk+1(ii) a If lossk+1＜losskIf yes, the parameters are updated, and
Figure BDA0002928538390000032
and p is 0, k is k +1, and λ is λ/β, and the process goes to step (2). Otherwise, turning to the step (4);
(4) making λ ═ λ β, p ═ p +1, and going to step (2);
(5) and outputting the temperature values estimated by the T1, T2, T4, T5, T7, T8 and T9 monomers.
Compared with the prior art, the invention has the following beneficial effects:
the temperature of all the monomers in the super capacitor module can be effectively estimated through a small number of temperature measurement values, and the number and the installation cost of the temperature sensors are greatly reduced. The method has simple process, is easy to realize on line and has good engineering application prospect.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a schematic layout diagram of a super capacitor module according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a neural network model according to an embodiment of the present invention.
Detailed Description
The method comprises the following specific implementation processes:
the module partitioning method in the first step is described by taking the super capacitor module shown in fig. 2 as an example. Under the condition of no forced heat dissipation, in consideration of the symmetry of the module, 9 sections of monomers of half of the module are selected as research objects and respectively marked as T1-T9. By performing 30A, 50A, 70A and 90A constant current experiments on the super capacitor module, the temperature distribution of the tested monomer can be mainly divided into three areas. The temperature distribution result of each monomer and the arrangement mode of the supercapacitor module can be analyzed, and the monomer temperature rising rate is higher when the monomer is closer to the inner arrangement, and the monomer temperature rising rate is higher when the monomer is more in contact with the rest of the monomers.
In an example, as shown:
the temperatures of T1 and T3 were the lowest, and the number of contact surfaces with the remaining monomers was 2, so the first region was defined.
The temperature of the T2, T6, T7, T9 and T4 monomers is similar, the number of contact surfaces with the rest monomers is 3, and the monomers can be divided into a second area.
The temperatures of T5 and T8 were the highest, and the number of contact surfaces with the remaining monomers was 4, which was defined as the third region.
The selection method of the test monomer in the step two comprises the following steps: and selecting a reasonable number of monomers for testing according to the requirements on algorithm precision and occupied computing resources, and recommending one monomer to be selected in each temperature partition. In case of limited computational resources, at least one monomer may be selected. In this example, T3 and T6 were selected as input monomers.
And the module temperature online estimation model in the third step is realized based on a neural network. And taking the temperature and current I of the input monomers T3 and T6 selected in the step two as the input of the module temperature online estimation model, wherein the output result is the temperature estimation values of the rest monomers T1, T2, T4, T5, T7, T8 and T9 in the module, and the precision of the temperature estimation values is related to the structure of the neural network model and the number of training samples. And determining the structure of the neural network by off-line training of the neural network model to obtain the temperature on-line estimation model. The neural network model built by the example is shown in fig. 3, and comprises 3 inputs, 7 outputs and 2 successive hidden layers in the middle, and the number of neurons in the hidden layers is 28 and 14 in sequence.
The neural network model is trained by adopting an algorithm based on Levenberg-Marquardt, and the Levenberg-Marquardt algorithm utilizes approximate second-order gradient information of errors, so that the method is more accurate and efficient in theory compared with a traditional gradient descent method. The training process is as follows:
(1) initializing neural network parameters θ0The maximum iteration number K, the updating step length lambda, the step length reduction coefficient beta, the current iteration number K is 0, the training error limit epsilon, the early-stop initial value p is 0, and the early-stop endurance coefficientP。
(2) Selecting Sigmoid function as activation function
Figure BDA0002928538390000041
Computational neural network forward propagation
Figure BDA0002928538390000051
Wherein the content of the first and second substances,
Figure BDA0002928538390000052
value, K, representing the ith neuron of the mth layerm-1Represents the number of neurons in the m-1 layer,
Figure BDA0002928538390000053
represents the weight of the jth input neuron of the (m-1) th layer to the ith neuron of the m layer,
Figure BDA0002928538390000054
is the bias term for the ith neuron in the mth layer. Selecting a loss function of
Figure BDA0002928538390000055
Wherein the content of the first and second substances,
Figure BDA0002928538390000056
is the square of the error for sample i and l is the number of samples. Calculating the losskIf losskIf ∈ or K ═ K or P ═ P, then step (5) is executed.
(3) According to the formula theta-lambda [ JT(θ)J(θ)+μI]-1JT(θ) e (θ) calculating network parameters, where μ is a constant greater than zero, I is an identity matrix, and J is a Jacobian matrix. Update parameter derivation
Figure BDA0002928538390000057
And calculating the loss value loss of the model after updating the one-step parameterk+1. If lossk+1＜losskIf yes, the parameters are updated, and
Figure BDA0002928538390000058
and p is 0, k is k +1, and λ is λ/β, and the process goes to step (2). Otherwise, the step (4) is carried out.
(4) And (3) enabling the step (2) to be shifted to the step (2), wherein the step (lambda) is equal to the step (beta) and the step (p) is equal to the step (p + 1).
(5) And outputting the temperature values estimated by the T1, T2, T4, T5, T7, T8 and T9 monomers.
The present invention is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can make various corresponding changes and modifications according to the present invention without departing from the spirit and the essence of the present invention, but those changes and modifications should fall within the protection scope of the appended claims.

Claims (2)

1. An online estimation method for temperature distribution of a super capacitor module is characterized by comprising the following steps:
the method comprises the following steps: the method comprises the following steps of (1) taking the temperature correlation of monomers in a super capacitor module into consideration, and partitioning the module by combining the arrangement characteristics of the monomers;
step two: selecting a tested monomer according to the subareas for testing to obtain temperature and current data of the selected monomer;
step three: constructing a neural network model, designing a training algorithm, training and optimizing the neural network model to obtain an estimation model, and inputting measured data into the temperature online estimation model for calculation;
step four: and outputting the estimated value of each monomer temperature in the module.
2. The online estimation method for the temperature distribution of the supercapacitor module group according to claim 1, characterized in that in the third step, the neural network model is trained by using an algorithm based on Levenberg-Marquardt, and the training process is as follows:
(1) initializing neural network parameters θ0Updating the step length lambda and the step length reduction coefficient beta according to the maximum iteration number K, wherein the current iteration number K is 0, the training error limit epsilon and the early-stop initial value P are 0, and early-stop tolerance coefficient P is obtained;
(2) selecting Sigmoid function as activation function
Figure FDA0002928538380000011
Computational neural network forward propagation
Figure FDA0002928538380000012
Wherein the content of the first and second substances,
Figure FDA0002928538380000013
value, K, representing the ith neuron of the mth layerm-1Represents the number of neurons in the m-1 layer,
Figure FDA0002928538380000014
represents the weight of the jth input neuron of the (m-1) th layer to the ith neuron of the m layer,
Figure FDA0002928538380000015
is the bias term for the ith neuron in the mth layer. Selecting a loss function of
Figure FDA0002928538380000016
Wherein the content of the first and second substances,
Figure FDA0002928538380000017
is the square of the error of sample i, and l is the number of samples; calculating the losskIf losskIf ∈ or K ═ K or P ═ P, then step (5) is carried out;
(3) according to the formula theta-lambda [ JT(θ)J(θ)+μI]-1JT(theta) e (theta) calculating network parameters, wherein mu is a constant larger than zero, I is an identity matrix, and J is a Jacobian matrix; update parameter derivation
Figure FDA0002928538380000018
And calculating the loss value loss of the model after updating the one-step parameterk+1(ii) a If lossk+1＜losskIf yes, the parameters are updated, and
Figure FDA0002928538380000019
and p is 0, k is k +1, and λ is λ/β, and the process goes to step (2). Otherwise, turning to the step (4);
(4) making λ ═ λ β, p ═ p +1, and going to step (2);
(5) and outputting the temperature values estimated by the T1, T2, T4, T5, T7, T8 and T9 monomers.
CN202110140366.9A 2021-02-02 2021-02-02 Online estimation method for temperature distribution of super capacitor module Active CN112883638B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110140366.9A CN112883638B (en) 2021-02-02 2021-02-02 Online estimation method for temperature distribution of super capacitor module

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110140366.9A CN112883638B (en) 2021-02-02 2021-02-02 Online estimation method for temperature distribution of super capacitor module

Publications (2)

Publication Number Publication Date
CN112883638A true CN112883638A (en) 2021-06-01
CN112883638B CN112883638B (en) 2022-08-09

Family

ID=76052500

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110140366.9A Active CN112883638B (en) 2021-02-02 2021-02-02 Online estimation method for temperature distribution of super capacitor module

Country Status (1)

Country Link
CN (1) CN112883638B (en)

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103675694A (en) * 2012-09-26 2014-03-26 同济大学 Fuel cell pile online regional detection system and method
EP3096389A1 (en) * 2015-05-18 2016-11-23 Nokia Technologies Oy An apparatus and associated methods for electrical storage
CN109088129A (en) * 2018-08-28 2018-12-25 华霆（合肥）动力技术有限公司 A kind of radiator and battery modules
CN109239489A (en) * 2018-08-29 2019-01-18 中车株洲电力机车有限公司 A kind of health status on-line analysis, device and medium for energy-storage system
CN110008557A (en) * 2019-03-28 2019-07-12 江苏理工学院 Stable and convergent supercapacitor service life On-line Estimation method
CN110474119A (en) * 2019-08-13 2019-11-19 许昌学院 A kind of Li-ion batteries piles heat management system and its thermal management algorithm
CN110687452A (en) * 2019-09-05 2020-01-14 南京理工大学 Lithium battery capacity online prediction method based on K-means clustering and Elman neural network
CN111474487A (en) * 2020-04-13 2020-07-31 重庆大学 Battery state of charge-internal temperature joint online estimation method
WO2020178616A1 (en) * 2019-03-02 2020-09-10 Qatar University Carbon nanotubes decorated with carbon nanospheres

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103675694A (en) * 2012-09-26 2014-03-26 同济大学 Fuel cell pile online regional detection system and method
EP3096389A1 (en) * 2015-05-18 2016-11-23 Nokia Technologies Oy An apparatus and associated methods for electrical storage
CN109088129A (en) * 2018-08-28 2018-12-25 华霆（合肥）动力技术有限公司 A kind of radiator and battery modules
CN109239489A (en) * 2018-08-29 2019-01-18 中车株洲电力机车有限公司 A kind of health status on-line analysis, device and medium for energy-storage system
WO2020178616A1 (en) * 2019-03-02 2020-09-10 Qatar University Carbon nanotubes decorated with carbon nanospheres
CN110008557A (en) * 2019-03-28 2019-07-12 江苏理工学院 Stable and convergent supercapacitor service life On-line Estimation method
CN110474119A (en) * 2019-08-13 2019-11-19 许昌学院 A kind of Li-ion batteries piles heat management system and its thermal management algorithm
CN110687452A (en) * 2019-09-05 2020-01-14 南京理工大学 Lithium battery capacity online prediction method based on K-means clustering and Elman neural network
CN111474487A (en) * 2020-04-13 2020-07-31 重庆大学 Battery state of charge-internal temperature joint online estimation method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
戴朝华等: "不同空间结构的有轨电车超级电容热行为", 《西南交通大学学报》 *
肖璆: "锂离子超级电容组高可靠性数据采集系统设计与研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 *
黄业伟: "电动汽车锂离子动力电池健康状态估计方法研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 *

Also Published As

Publication number Publication date
CN112883638B (en) 2022-08-09

Similar Documents

Publication Publication Date Title
Ge et al. A review on state of health estimations and remaining useful life prognostics of lithium-ion batteries
CN106295003B (en) Lithium battery life prediction method based on degradation track coordinate reconstruction and multiple linear regression
CN111595485B (en) Lithium ion battery online temperature distribution observer design method based on reduced order model
Tian et al. Feature fusion-based inconsistency evaluation for battery pack: Improved Gaussian mixture model
CN114548368A (en) Modeling method, prediction model and prediction method of lithium battery temperature field prediction model based on multilayer nuclear overrun learning machine
CN111191824B (en) Power battery capacity attenuation prediction method and system
Wang et al. Health diagnosis for lithium-ion battery by combining partial incremental capacity and deep belief network during insufficient discharge profile
CN103413038A (en) Vector quantization based long-term intuitionistic fuzzy time series prediction method
Li et al. Remaining useful life prediction of lithium battery using convolutional neural network with optimized parameters
CN115407207A (en) Online prediction method and system for lithium ion battery charge state
CN111723982A (en) Medium-and-long-term power load combined prediction method based on gray-Markov chain
Huang et al. Correlation analysis and prediction of PEM fuel cell voltage during start-stop operation based on real-world driving data
Tang et al. Novel solution framework for inverse problem considering interval uncertainty
Lin et al. Algorithm of BPNN‐UKF based on a fusion model for SOC estimation in lithium‐ion batteries
Cai et al. Remaining useful life prediction method of EV power battery for DC fast charging condition
Li et al. Degradation prediction of proton exchange membrane fuel cell based on the multi-inputs Bi-directional long short-term memory
Hongwei et al. An interpretable data-driven method for degradation prediction of proton exchange membrane fuel cells based on temporal fusion transformer and covariates
CN112883638B (en) Online estimation method for temperature distribution of super capacitor module
CN112016241B (en) Soft measurement modeling method based on mLASSO-MLP model
CN107301266B (en) LOC estimation method and system for lithium iron phosphate battery
CN112731183A (en) Lithium ion battery life prediction method based on improved ELM
Li et al. Predicting Software Quality by Optimized BP Network Based on PSO.
CN115455842A (en) Method for estimating state of charge of super capacitor based on variable-temperature model weighted fusion
CN113688465B (en) Aircraft structural strength digital twin method based on combination of load and state
Li et al. Machine learning algorithm based battery modeling and management method: A Cyber-Physical System perspective

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