CN109870654A - The online method for dynamic estimation of accumulator capacity based on impact load response characteristic - Google Patents

The online method for dynamic estimation of accumulator capacity based on impact load response characteristic Download PDF

Info

Publication number
CN109870654A
CN109870654A CN201910106341.XA CN201910106341A CN109870654A CN 109870654 A CN109870654 A CN 109870654A CN 201910106341 A CN201910106341 A CN 201910106341A CN 109870654 A CN109870654 A CN 109870654A
Authority
CN
China
Prior art keywords
impact load
capacity
energy
battery
accumulator capacity
Prior art date
Application number
CN201910106341.XA
Other languages
Chinese (zh)
Inventor
林琼斌
竺学涛
王武
蔡逢煌
黄捷
Original Assignee
福州大学
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 福州大学 filed Critical 福州大学
Priority to CN201910106341.XA priority Critical patent/CN109870654A/en
Publication of CN109870654A publication Critical patent/CN109870654A/en

Links

Abstract

The present invention discloses a kind of online method for dynamic estimation of the accumulator capacity based on impact load response characteristic, comprising: keeps carrying out on-line real time monitoring to cell voltage, electric current and temperature;When there is impact load, i.e. when battery current moment, which increases amplitude, is greater than preset value, the accumulating voltage acquired during load sudden change, electric current and temperature data are delivered to On-line Estimation device, by merging the accumulator capacity On-line Estimation algorithm of wavelet analysis and Cerebellar Model Articulation Controller, real-time estimation remaining battery capacity.The present invention is not necessarily to put battery progress long-time Man Chongman, without actively injecting other harmonic signals to battery, directly using under impact load effect, the cell voltage of real-time monitoring, electric current and temperature data, the capacity of battery can be carried out estimating quickly, accurately, in real time, it is suitble to uninterruptible power system, battery maintenance cost is reduced, accumulator capacity reduction or Problem of Failure is found in time, improves the reliability of system.

Description

The online method for dynamic estimation of accumulator capacity based on impact load response characteristic

Technical field

The present invention relates to battery detecting fields, and in particular to a kind of accumulator capacity based on impact load response characteristic exists Line method for dynamic estimation.

Background technique

Energy energy-storage units of the battery as uninterruptible power system become electric energy during electric network power-fail and uniquely provide Person, reliability are most important to system.However, battery active volume is gradually with cycle-index increase and long-time aging It reduces or even entirely ineffective.Therefore, the necessary ring that the battery virtual condition comprising active volume is battery service is inspected periodically Section.There are mainly two types of existing battery condition detection technological means: first is that appraising and deciding the surplus of battery using regular charge and discharge system Covolume amount, this method is reliably effective, but that there are labor intensives is more, the time is long, the big disadvantage of energy loss, is additionally present of when electric discharge When low to electricity, there is grid power blackout, then without the risk of enough electric energy support load;Second is that being judged using internal resistance detection method The state of battery, wherein off-line type, it is still desirable to a large amount of artificial routine tests, time-consuming work consuming, during existing simultaneously detection twice Battery failure fails the risk found in time;Online inner walkway method, although it is a large amount of artificial to overcome off-line type consuming Defect, but active harmonic, exist and cause dysgenic hidden danger to system load and other equipment.

Summary of the invention

In view of this, the purpose of the present invention is to provide a kind of accumulator capacity based on impact load response characteristic is online Method for dynamic estimation increases sensor and certain loads without additional, directly negative using impact also not to battery harmonic Under load effect, battery current, the end data such as voltage and temperature are online real-time by trained machine learning intelligent algorithm Estimate remaining battery capacity.

To achieve the above object, the present invention adopts the following technical scheme:

A kind of online method for dynamic estimation of accumulator capacity based on impact load response characteristic, providing a system includes electric power storage Pond, sampling processing circuit, accumulator capacity On-line Estimation device, communication module, remote monitoring system, comprising the following steps:

Step S1: battery voltage, electric current and temperature are acquired in real time by sampling processing circuit;

Step S2: when there is impact load, i.e., electric current is more than preset threshold, by the battery acquired in real time during impact electricity Pressure, electric current and temperature data are sent to remaining battery capacity On-line Estimation module;

Step S3: accumulator capacity On-line Estimation module is according to obtained data, by being built in accumulator capacity On-line Estimation mould It is transferred to machine learning algorithm after wavelet analysis feature extraction in block, real-time remaining battery capacity estimation is carried out, obtains electric power storage Pond residual capacity estimated value;

Step S4: remaining battery capacity estimated value is sent to remote monitoring system by communication module and is monitored in real time.

Further, the wavelet analysis feature extraction specifically:

Step S21: using wavelet transformation extracted valid data feature, selects three rank Daubechies small echos, Decomposition order setting It is 11 layers;

Step S22: being decomposed into approximation coefficient and detail coefficients for voltage signal, current signal and temperature signal, weight after decomposition Then structure calculates the energy of every layer of detail coefficients;

Step S23: scaling function Ф j,k (t) and wavelet function ψ j,k (t) such as formula (1), shown in (2):

(1)

(2)

E dj Indicate thejLayer detail coefficients energy:

i =1, 2,…, N (3)

E a Indicate approximation coefficient portion of energy:

j =1, 2,…, n (4)

E total Indicate gross energy, i.e. detail coefficients energy part and approximation coefficient energy part summation:

(5)

P j Indicate thejLayer detail coefficients energy accounting:

(6)

N is data length in formula, and n is Decomposition order, PjFor jth layer detail coefficients energy accounting;d k,j It isjLayer detail coefficients ThekNumber;a j,n It isnLayer approximation coefficient thejNumber.

Further, the machine learning algorithm uses Cerebellar Model Articulation Controller algorithm, wherein 8 layers of interlayer selecting, By 1,3,9 input as neural network in 11 layers of energy of wavelet decomposition, neural network output is accumulator capacity.

Further, the accumulator capacity is divided into, and 100% ~ 90%, 90% ~ 80%, 80% ~ 70%, 70% ~ 60%, 60% ~ 50%, 50% ~ 0% 6 class;Accumulator electric-quantity SOC is 20% ~ 100%.

Further, the machine learning algorithm building specifically:

Step S31: the storage battery power supply system experimental platform containing impact load is built;

Step S32: under the effect of the same impact load, the response spy in the case of different batteries capacity and different SOC is acquired Linearity curve;

Step S33: adjusting action time and the amplitude of impact load, and return step two obtains the response data under different loads;

Step S34: using collected battery voltage, electric current and temperature data, data sample library is established;

Step S35: using wavelet analysis, carries out data characteristics extraction;

Step S36: the data completed are extracted using data characteristics, add corresponding dash current amplitude and time width, electric power storage Pond temperature carries out off-line training to Cerebellar Model Articulation Controller algorithm, until its estimated accuracy meets preset requirement.

Further, the Cerebellar Model Articulation Controller algorithm interbed selects 8 layers, and choose in voltage decomposition energy 1, 3,9 layers, electric current, temperature decompose the third layer energy in energy, input of totally five characteristic signals as neural network, nerve net Network output is accumulator capacity.

Compared with the prior art, the invention has the following beneficial effects:

The present invention is not necessarily to carry out long-time Man Chongman to battery to put, without actively injecting other harmonic signals to battery, Directly using under impact load effect, the cell voltage of real-time monitoring, electric current and temperature data, can be to the capacity of battery Quickly, accurately, in real time estimate, be suitble to uninterruptible power system, reduce battery maintenance cost, it is timely to find electric power storage pool capacity Amount reduces or Problem of Failure, improves the reliability of system.

Detailed description of the invention

Fig. 1 is that the present invention is based on the online dynamic estimation intelligent apparatus of the accumulator capacity of impact load response characteristic;

Fig. 2 is the method for the present invention flow chart;

Fig. 3 is battery impact load response characteristic in one embodiment of the invention.

Specific embodiment

The present invention will be further described with reference to the accompanying drawings and embodiments.

Fig. 1 is please referred to, the present invention provides a kind of online dynamic estimation of the accumulator capacity based on impact load response characteristic Intelligent method, the online dynamic estimation circuit of accumulator capacity as shown in Figure 1, the system include power grid, rectification circuit, battery, Conventional load, impact load, sampling processing circuit, battery On-line Estimation module, communication network, remote monitoring system.

For the present embodiment impact load selection high-voltage circuitbreaker.When power grid works normally, by power grid to conventional load It powers with high-voltage circuitbreaker;When electric network fault disconnects, powered by battery to constant load and high-voltage circuitbreaker.

Online dynamic estimation intelligent method specific implementation method the following steps are included:

Step S1: it as shown in figure, keeps to cell voltage, electric current and temperature online real-time monitoring;

Step S2: when the high-voltage circuitbreaker movement powered by direct current operative power source, the large impact electricity of a short time can be generated Stream belongs to impact load.

Step S3:, will be during load sudden change when the amplitude for detecting that battery current moment increases is greater than preset value Cell voltage, electric current and the temperature of acquisition are delivered to accumulator capacity On-line Estimation device, include impact load in On-line Estimation device Response characteristic feature extraction and remaining battery capacity On-line Estimation algorithm, the data of acquisition are by being built in small wavelength-division in estimator Machine learning algorithm, which is transferred to, after analysis feature extraction carries out accumulator capacity real-time estimation;

Step S4: by communication network, battery real time capacity is transmitted to remote monitoring end, carries out the processing of next step;It is no Then continue to detect battery voltage, electric current and temperature;

In the present embodiment, impact load is acted on down collected voltage response curves data and electric current and temperature, use are small Wave conversion extracted valid data feature.

Voltage responsive characteristic is as shown in figure 3, three rank Daubechies small echo of case column selection, Decomposition order are set as 11 layers. Wherein scaling function Ф j,k (t) and wavelet function ψ j,k (t) such as formula (1), shown in (2), respectively as low pass and high-pass filter. Voltage signal is broken down into approximation coefficienta j And detail coefficientsd j ,It is reconstructed after decomposition, then calculates the energy of every layer of detail coefficients. Electric current is consistent with voltage decomposition process with temperature signal decomposable process.

(1)

(2)

E dj Indicate thejLayer detail coefficients energy:

i =1, 2,…, N (3)

E a Indicate approximation coefficient portion of energy:

j =1, 2,…, n (4)

E total Indicate gross energy, i.e. detail coefficients energy part and approximation coefficient energy part summation:

(5)

P j Indicate thejLayer detail coefficients energy accounting:

(6)

In the present embodiment, the remaining battery capacity On-line Estimation algorithm, the feature extracted using wavelet analysis, in conjunction with punching Size of current, duration and impact moment battery temp are hit, trained machine learning intelligent algorithm is inputted, completes electric power storage The estimation of pond residual capacity, and the estimated capacity value result of output is transferred to remote monitoring system by communication network.

In the present embodiment, the trained machine learning intelligent algorithm, the trained machine learning intelligent algorithm, It is to be completed in advance by off-line training, as shown in Fig. 2, steps are as follows for machine learning intelligent algorithm off-line training:

Step 1: the storage battery power supply system experimental platform containing impact load is built;

Step 2: under the effect of the same impact load, the response characteristic in the case of different batteries capacity and different SOC is acquired Curve;

Step 3: adjusting action time and the amplitude of impact load, and return step two obtains the response data under different loads;

Step 4: using step 2 and the collected battery voltage of step 3, electric current and temperature data, data sample is established Library;

Step 5: using wavelet analysis, carries out data characteristics extraction;

Step 6: the data completed are extracted using data characteristics, add corresponding dash current amplitude and time width, battery Temperature carries out off-line training to machine learning algorithm, until its estimated accuracy is met the requirements.

In the present embodiment, the online dynamic estimation intelligence side of the accumulator capacity based on impact load response characteristic Method, which is characterized in that step 2 accumulator capacity and electrical parameter, wherein accumulator capacity is divided into, 100% ~ 90%, 90% ~ 80%, 80% ~ 70%, 70% ~ 60%, 60% ~ 50%, 50% ~ 0% 6 class;Accumulator electric-quantity SOC is 20% ~ 100%.

In the present embodiment, the online dynamic estimation intelligence side of the accumulator capacity based on impact load response characteristic Method, which is characterized in that machine learning algorithm described in step 6 selects Cerebellar Model Articulation Controller.Wherein 8 layers of interlayer selecting, and 1,3,9 layers in voltage decomposition energy are chosen, electric current, temperature decompose the third layer energy in energy, and totally five characteristic signals are made For the input of neural network, neural network output is accumulator capacity.

The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with Modification, is all covered by the present invention.

Claims (6)

1. a kind of online method for dynamic estimation of accumulator capacity based on impact load response characteristic, providing a system includes electric power storage Pond, sampling processing circuit, accumulator capacity On-line Estimation device, communication module, remote monitoring system, which is characterized in that including with Lower step:
Step S1: battery voltage, electric current and temperature are acquired in real time by sampling processing circuit;
Step S2: when there is impact load, i.e., electric current is more than preset threshold, by the battery acquired in real time during impact electricity Pressure, electric current and temperature data are sent to remaining battery capacity On-line Estimation module;
Step S3: accumulator capacity On-line Estimation module is according to obtained data, by being built in accumulator capacity On-line Estimation mould It is transferred to machine learning algorithm after wavelet analysis feature extraction in block, real-time remaining battery capacity estimation is carried out, obtains electric power storage Pond residual capacity estimated value;
Step S4: remaining battery capacity estimated value is sent to remote monitoring system by communication module and is monitored in real time.
2. the online method for dynamic estimation of the accumulator capacity according to claim 1 based on impact load response characteristic, It is characterized in that: the wavelet analysis feature extraction specifically:
Step S21: using wavelet transformation extracted valid data feature, selects three rank Daubechies small echos, Decomposition order setting It is 11 layers;
Step S22: being decomposed into approximation coefficient and detail coefficients for voltage signal, current signal and temperature signal, weight after decomposition Then structure calculates the energy of every layer of detail coefficients;
Step S23: scaling function Ф j,k (t) and wavelet function ψ j,k (t) such as formula (1), shown in (2):
(1)
(2)
E dj Indicate thejLayer detail coefficients energy:
i =1, 2,…, N (3)
E a Indicate approximation coefficient portion of energy:
j =1, 2,…, n (4)
E total Indicate gross energy, i.e. detail coefficients energy part and approximation coefficient energy part summation:
(5)
P j Indicate thejLayer detail coefficients energy accounting:
(6)
N is data length in formula, and n is Decomposition order, PjFor jth layer detail coefficients energy accounting;d k,j It isjLayer detail coefficients thekNumber;a j,n It isnLayer approximation coefficient thejNumber.
3. the online method for dynamic estimation of the accumulator capacity according to claim 2 based on impact load response characteristic, Be characterized in that: the machine learning algorithm uses Cerebellar Model Articulation Controller algorithm, wherein 8 layers of interlayer selecting, by small wavelength-division 1,3,9 input as neural network in 11 layers of energy of solution, neural network output are accumulator capacity.
4. the online method for dynamic estimation of the accumulator capacity according to claim 1 based on impact load response characteristic, Be characterized in that: the accumulator capacity is divided into, and 100% ~ 90%, 90% ~ 80%, 80% ~ 70%, 70% ~ 60%, 60% ~ 50%, 50% ~ 0% six class;Accumulator electric-quantity SOC is 20% ~ 100%.
5. the online method for dynamic estimation of the accumulator capacity according to claim 3 based on impact load response characteristic, It is characterized in that: the machine learning algorithm building specifically:
Step S31: the storage battery power supply system experimental platform containing impact load is built;
Step S32: under the effect of the same impact load, the response spy in the case of different batteries capacity and different SOC is acquired Linearity curve;
Step S33: adjusting action time and the amplitude of impact load, and return step two obtains the response data under different loads;
Step S34: using collected battery voltage, electric current and temperature data, data sample library is established;
Step S35: using wavelet analysis, carries out data characteristics extraction;
Step S36: the data completed are extracted using data characteristics, add corresponding dash current amplitude and time width, electric power storage Pond temperature carries out off-line training to Cerebellar Model Articulation Controller algorithm, until its estimated accuracy meets preset requirement.
6. the online method for dynamic estimation of the accumulator capacity according to claim 5 based on impact load response characteristic, Be characterized in that: the Cerebellar Model Articulation Controller algorithm interbed selects 8 layers, and chooses 1,3,9 layers in voltage decomposition energy, electricity Stream, temperature decompose the third layer energy in energy, input of totally five characteristic signals as neural network, and neural network output is Accumulator capacity.
CN201910106341.XA 2019-02-02 2019-02-02 The online method for dynamic estimation of accumulator capacity based on impact load response characteristic CN109870654A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910106341.XA CN109870654A (en) 2019-02-02 2019-02-02 The online method for dynamic estimation of accumulator capacity based on impact load response characteristic

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910106341.XA CN109870654A (en) 2019-02-02 2019-02-02 The online method for dynamic estimation of accumulator capacity based on impact load response characteristic

Publications (1)

Publication Number Publication Date
CN109870654A true CN109870654A (en) 2019-06-11

Family

ID=66918605

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910106341.XA CN109870654A (en) 2019-02-02 2019-02-02 The online method for dynamic estimation of accumulator capacity based on impact load response characteristic

Country Status (1)

Country Link
CN (1) CN109870654A (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1809758A (en) * 2003-06-23 2006-07-26 矢崎总业株式会社 Method and apparatus for judging deterioration of battery
CN102023283A (en) * 2009-09-16 2011-04-20 徐先 On-line measuring method of accumulator capacity
CN102967831A (en) * 2012-09-17 2013-03-13 常州大学 On-line detection system and detection method of lead-acid storage battery performance
CN103344921A (en) * 2013-07-08 2013-10-09 华南师范大学 Lithium-ion power battery health state evaluating system and method
CN104049213A (en) * 2013-03-12 2014-09-17 三星Sdi株式会社 Apparatus for predicting state of health of battery pack by using discrete wavelet transform
CN105021995A (en) * 2015-07-28 2015-11-04 王树华 Portable intelligent storage battery parameter test method and tester
CN105807230A (en) * 2016-03-11 2016-07-27 郑贵林 Storage battery residual capacity and health state rapid detection method and device
CN106599777A (en) * 2016-11-02 2017-04-26 华南理工大学 Cable partial discharge signal identification method based on energy percentage
CN108845264A (en) * 2018-06-01 2018-11-20 西安交通大学 A kind of cell health state estimation method based on small echo
CN108896926A (en) * 2018-07-18 2018-11-27 湖南宏迅亿安新能源科技有限公司 A kind of appraisal procedure, assessment system and the associated component of lithium battery health status

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1809758A (en) * 2003-06-23 2006-07-26 矢崎总业株式会社 Method and apparatus for judging deterioration of battery
CN102023283A (en) * 2009-09-16 2011-04-20 徐先 On-line measuring method of accumulator capacity
CN102967831A (en) * 2012-09-17 2013-03-13 常州大学 On-line detection system and detection method of lead-acid storage battery performance
CN104049213A (en) * 2013-03-12 2014-09-17 三星Sdi株式会社 Apparatus for predicting state of health of battery pack by using discrete wavelet transform
CN103344921A (en) * 2013-07-08 2013-10-09 华南师范大学 Lithium-ion power battery health state evaluating system and method
CN105021995A (en) * 2015-07-28 2015-11-04 王树华 Portable intelligent storage battery parameter test method and tester
CN105807230A (en) * 2016-03-11 2016-07-27 郑贵林 Storage battery residual capacity and health state rapid detection method and device
CN106599777A (en) * 2016-11-02 2017-04-26 华南理工大学 Cable partial discharge signal identification method based on energy percentage
CN108845264A (en) * 2018-06-01 2018-11-20 西安交通大学 A kind of cell health state estimation method based on small echo
CN108896926A (en) * 2018-07-18 2018-11-27 湖南宏迅亿安新能源科技有限公司 A kind of appraisal procedure, assessment system and the associated component of lithium battery health status

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
李学斌: ""基于容量分解的电力直流电源系统蓄电池容量计算方法"", 《电网技术》 *

Similar Documents

Publication Publication Date Title
CN104753461B (en) Method for diagnosing and classifying faults of photovoltaic power generation arrays on basis of particle swarm optimization support vector machines
Catalão et al. Short-term wind power forecasting in Portugal by neural networks and wavelet transform
CN102289590B (en) Method for estimating operating state of SF6 high-voltage circuit breaker and intelligent system
Ducange et al. An intelligent system for detecting faults in photovoltaic fields
CN101067644B (en) Storage battery performance analytical expert diagnosing method
Mellit et al. Modeling and simulation of a stand-alone photovoltaic system using an adaptive artificial neural network: Proposition for a new sizing procedure
CN102520366B (en) Electric car cell safety and health assessment system and method thereof
CN105787584B (en) Wind turbine group fault early warning method based on cloud platform
Laghari et al. Application of computational intelligence techniques for load shedding in power systems: A review
Zhou et al. Hybrid genetic algorithm method for efficient and robust evaluation of remaining useful life of supercapacitors
Eddahech et al. Behavior and state-of-health monitoring of Li-ion batteries using impedance spectroscopy and recurrent neural networks
CN105303468A (en) Comprehensive evaluation method of smart power grid construction based on principal component cluster analysis
CN103886374B (en) A kind of cable connector conductor temperature Forecasting Methodology based on RBF neural
CN103336230B (en) Device and method for predicting cable insulation failure rate in power distribution network
CN101067645B (en) Method for analysing valve control type lead-acid accumulator battery performance
CN104319768B (en) A kind of micro-capacitance sensor is powered and method for supervising
CN103279807B (en) A kind of static risk assessment method for power grid in severe weather
CN104750915B (en) A kind of Buck circuits multi-parameter on-line identification method
CN106291351B (en) High-voltage circuitbreaker fault detection method based on convolutional neural networks algorithm
DE102011054144A1 (en) Control method for the detection of excessive current
CN102288857B (en) Fault arc identification and detection method and detection protection device
CN102412606B (en) Charging controller and control method thereof
CN104091008B (en) Based on the user power utilization equipment safety appraisal procedure improving interval based AHP
CN104485681B (en) A kind of monitoring method of wind energy turbine set energy-storage system
CN105334065B (en) A kind of electric powered motor multifunction electronic tests system

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