CN112347692A - Method and device for realizing predictive maintenance of battery of uninterruptible power supply and electronic device - Google Patents

Method and device for realizing predictive maintenance of battery of uninterruptible power supply and electronic device Download PDF

Info

Publication number
CN112347692A
CN112347692A CN202010997018.9A CN202010997018A CN112347692A CN 112347692 A CN112347692 A CN 112347692A CN 202010997018 A CN202010997018 A CN 202010997018A CN 112347692 A CN112347692 A CN 112347692A
Authority
CN
China
Prior art keywords
battery
data
power supply
voltage
time
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
CN202010997018.9A
Other languages
Chinese (zh)
Other versions
CN112347692B (en
Inventor
林青雷
曾念寅
贺良
张柯歌
雷聪
廖志伟
张华山
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.)
Shenzhen Qianhai Yespowering Iot Technology Co ltd
Original Assignee
Shenzhen Qianhai Yespowering Iot Technology Co ltd
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 Shenzhen Qianhai Yespowering Iot Technology Co ltd filed Critical Shenzhen Qianhai Yespowering Iot Technology Co ltd
Priority to CN202010997018.9A priority Critical patent/CN112347692B/en
Publication of CN112347692A publication Critical patent/CN112347692A/en
Application granted granted Critical
Publication of CN112347692B publication Critical patent/CN112347692B/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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21—Design, administration or maintenance of databases
    • G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2474—Sequence data queries, e.g. querying versioned data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/245—Classification techniques relating to the decision surface
    • G06F18/2451—Classification techniques relating to the decision surface linear, e.g. hyperplane
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J9/00—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting
    • H02J9/04—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting in which the distribution system is disconnected from the normal source and connected to a standby source
    • H02J9/06—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting in which the distribution system is disconnected from the normal source and connected to a standby source with automatic change-over, e.g. UPS systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Probability & Statistics with Applications (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Algebra (AREA)
  • Quality & Reliability (AREA)
  • Fuzzy Systems (AREA)
  • Operations Research (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Power Engineering (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)
  • Secondary Cells (AREA)

Abstract

The embodiment of the application provides a method and a device for realizing predictive maintenance of a battery of an uninterruptible power supply and an electronic device, and relates to the technical field of electronics, wherein the method comprises the following steps: establishing a digital twin model corresponding to the uninterruptible power supply, acquiring historical data and current state data of battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes, giving a label value to the power supply performance of the data of the battery, performing feature construction on the basic attributes, balancing the data quantity of the batteries with different power supply performance, inputting the balanced data of the battery into a prediction model of the battery fault of a nonlinear support vector machine for training, and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply. The method, the device and the electronic device can improve the failure prediction efficiency and reduce the cost.

Description

Method and device for realizing predictive maintenance of battery of uninterruptible power supply and electronic device
Technical Field
The present disclosure relates to the field of electronic technologies, and in particular, to a method and an apparatus for implementing battery predictive maintenance of an uninterruptible power supply, and an electronic apparatus.
Background
Uninterruptible Power Supplies (UPS) can play a vital role in the fields of machine rooms, data centers, and the like in the industries of Power, medical treatment, communication, and the like. When the commercial power is normally input, the power is supplied to the load after the voltage is stabilized by the UPS; the UPS may provide a stable backup power supply when mains power is interrupted. If the UPS fails unexpectedly, immeasurable loss can be caused to related industries. Maintenance of the UPS becomes an important technical issue.
In the prior art, UPS operation and maintenance technologies can be mainly classified into incident maintenance, i.e., maintenance when a fault occurs, and preventive maintenance, i.e., periodic maintenance when no fault occurs, but both technologies have the defects of large hysteresis, low efficiency, high cost, and the like.
Disclosure of Invention
The embodiment of the application provides a method for realizing battery predictive maintenance of an uninterruptible power supply, which can solve the problems of large hysteresis, low efficiency and high cost in the preventive maintenance of a UPS battery.
An embodiment of the present application provides a method for implementing battery predictive maintenance of an uninterruptible power supply, including:
establishing a digital twin model corresponding to the uninterrupted power supply;
continuously collecting real-time data of the uninterruptible power supply and transmitting the real-time data to the digital twin model;
acquiring historical data and current state data of the battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, and acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes;
according to the characteristics of the power supply performance of the battery, a label value is given to the power supply performance of the data of the battery;
expanding the data dimension of the battery by performing feature construction on the basic attributes by using a feature engineering method, wherein the constructed features comprise: basic features, unit features, temporal features, and combination features;
balancing the data volume of the batteries of different power supply performance;
inputting the balanced data of the battery into a prediction model of the battery fault of a nonlinear support vector machine for training;
and performing performance evaluation on the trained prediction model, and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
Another aspect of the present disclosure provides an apparatus for implementing battery predictive maintenance of an uninterruptible power supply, including:
the modeling module is used for establishing a digital twin model corresponding to the uninterrupted power supply;
the acquisition module is used for continuously acquiring real-time data of the uninterruptible power supply and transmitting the real-time data to the digital twin model;
the processing module is used for acquiring historical data and current state data of the battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, and acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes;
the tag module is used for endowing a tag value for the power supply performance of the data of the battery according to the characteristics of the power supply performance of the battery;
a feature module for expanding the data dimension of the battery by performing feature construction on the basic attributes by using a feature engineering method, the constructed features including: basic features, unit features, temporal features, and combination features;
the balancing module is used for balancing the data volume of the batteries with different power supply performances;
the training module is used for inputting the balanced data of the battery into a prediction model of the battery fault of the nonlinear support vector machine for training;
and the evaluation module is used for evaluating the performance of the trained prediction model and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
Another aspect of the embodiments of the present application provides an electronic device, including: the device comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor executes the computer program to realize the method for realizing the predictive maintenance of the battery of the uninterruptible power supply.
In the embodiment, by establishing the digital twin model corresponding to the UPS, inputting the battery data of the UPS into the prediction model of the battery fault of the non-linear support vector machine for training in the digital twin model, performing performance verification on the trained prediction model, and using the verified prediction model as the deployment prediction model of the battery fault of the UPS, the dilemma that historical data and fault information of the UPS are few and difficult to acquire and analyze is effectively solved, and the feature engineering technology is applied to expand the data dimension in view of the situations that the dimension of the UPS battery is few and the training accuracy of the model is affected.
Drawings
Fig. 1 is a flowchart illustrating a method for implementing predictive battery maintenance for an ups according to an embodiment of the present disclosure.
Fig. 2 is a schematic structural diagram of an apparatus for implementing battery predictive maintenance of an uninterruptible power supply according to an embodiment of the present disclosure.
Fig. 3 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present disclosure.
Fig. 4 is a schematic diagram of data interaction between an ups physical entity and an electronic device according to an embodiment of the present disclosure.
Detailed Description
In order to make the objects, features and advantages of the present invention more apparent and understandable, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Referring to fig. 1, fig. 1 is a schematic flow chart of a method for implementing battery predictive maintenance of an Uninterruptible Power Supply (UPS) according to an embodiment of the present disclosure, in which a battery system of the UPS is formed by a plurality of battery pack units, each battery pack unit is formed by a plurality of batteries, and the batteries may be specifically storage batteries or other types of batteries that can provide power for the UPS, and the method includes:
s101, establishing a digital twin model corresponding to the uninterrupted power supply;
various data in the UPS are collected through the multi-source sensor, and a digital twin model corresponding to the UPS entity is constructed. Namely, a digital twin model of the UPS is established on the basis of the physical space UPS, and predictive operation and maintenance analysis is carried out on the UPS in a digital space in a data driving mode.
S102, continuously collecting real-time data of the uninterruptible power supply, and transmitting the real-time data to the digital twin model;
and continuously transmitting the collected real-time data of the UPS entity to a digital twin model of the UPS to complete data synchronization.
S103, acquiring historical data and current state data of battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, and acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes of the battery;
basic properties include voltage, current, resistance, temperature and depth of discharge.
The basic rules of the battery power supply performance refer to the characteristics and rules of basic attributes of different power supply capacities of the battery, including the characteristics and rules of the basic attributes of a fault battery, the characteristics and rules of the basic attributes of a normal battery, the characteristics and rules of the basic attributes of the remaining service life of the battery, and the like.
The historical information and the current state information are obtained by taking time as an independent variable, and specifically comprise collected time nodes, factory numbers of the battery, and current (I), voltage (U), resistance (R, namely internal resistance), temperature (Temp), depth of discharge (D), fault information and the like of the battery. The collected time nodes are usually collected by taking the collected time nodes as one node every 10 minutes, and parameters such as resistance, discharge depth and the like which are not easy to change along with time are collected by taking the collected time nodes as one node every 10 days.
Removing invalid data in the collected historical data and the current state data, wherein the invalid data comprises wrong abnormal data and incomplete data with missing items, and determining statistical values of current, voltage, resistance, temperature and depth of discharge in the remaining valid data, wherein the statistical values comprise: the average value, the variance, the extreme value and the like, drawing a characteristic curve of the statistic, and obtaining basic rules of the fault battery and the normal battery from the characteristic curve, namely the basic rule of the statistic of the fault battery, the basic rule of the statistic of the normal battery, the basic rule of the statistic of the service life of the battery and the like.
S104, according to the characteristics of the power supply performance of the battery, giving a label value to the power supply performance of the data of the battery;
the power supply performance includes: whether the battery has a fault or not and the remaining service life of the battery, wherein the power supply performance of the battery with the fault is abnormal, and the power supply performance of the battery without the fault is normal.
The step of giving a label value to the power supply performance of the data of the battery refers to giving a label value to the data of the battery to be input into the training model for training, and marking whether the battery has a fault and the time for replacing the next new battery.
Specifically, the manner of determining a normal battery and its remaining useful life is as follows: if the internal resistance value of the battery is increased to a first resistance threshold value, preferably 5 milli-ohms (m Ω), or if the dc voltage of the battery is less than a second resistance threshold value, preferably 8 volts (V), the battery needs to be replaced in time, and the calculation formula of the time point for replacing the battery is as follows:
tchange1＝min{tthR,tthV}-tinit；
wherein, tchange1Indicates the time point of the normal battery replacement (10 minutes as one time point), tthRThe time when the internal resistance value of the normal battery reaches the first resistance threshold value of 5m omega, tthVThe time required for the direct current voltage to drop to the first voltage threshold value of 8V; t is tinitThe time origin calculated in this embodiment is an initial value of the battery use time, starting from a period of time after the battery starts to be used. t is tinitIs 4320, i.e., 30 (days) × 24 (hours) × 6 (6 per hour, 10 minutes), the calculated time origin is the second month of initial battery use.
The manner of determining the abnormal battery in which the unexpected malfunction occurs and the time when it needs to be replaced is as follows: in the early stage of the internal fault of the battery, the float charging voltage of the battery tends to be reduced and fluctuates sharply, and when the float charging voltage of the fault battery is reduced to the float charging voltage threshold value, the fault battery reaches the replacement time point.
First, a float voltage drop index VD of the current time point t is calculatedt
Figure BDA0002692880480000041
N and T belong to the range of time nodes to be calculated; n is a counting variable of a time node, starting from 1 to N; i is a counting variable starting from T-2T and going to T-T. Preferably, N is the number of time nodes of one month and T is the number of time nodes of one week. Specifically, N is 30 × 24 × 6 in units of days, hours, and collection count per hour, respectively, and T is 7 × 24 × 6 in units of days (days per week), hours, and collection count per hour, respectively; calculating the average value of the float voltage in the time range from (T-2T) to (T-T)
Figure BDA0002692880480000042
Figure BDA0002692880480000051
Calculating the difference between the voltage at the moment (t-n) and the average value, and taking the smaller value of the difference and 0 after comparison; w is anIs the weight value corresponding to n moments and the sum of all the weight values
Figure BDA0002692880480000052
Is 1.
Wherein, in the above formula
Figure BDA0002692880480000053
Represents the sum of the Ui values every two adjacent weeks.
t-n denotes n time nodes, U, ahead of the current time tt-nIndicating the voltage value at the time of calculation (t-n).
The calculation formula of the time point of the battery replacement at which the internal failure occurs is as follows:
tchange2＝min{t|VDt≥2σmean}
wherein, tchange2Indicating the point in time of faulty battery replacement, σmeanThe min operation is used for calculating the voltage drop index VD as the average value of the standard deviation of the float charge voltage of all normal batteriestNot less than float voltage threshold 2 sigmameanMinimum time required, float voltage threshold 2 sigmamean。
tchange1And tchange2For determining the time to assign the tag value to the battery when the battery reaches tchange1And tchange2At any "replacement time point" in the battery pack, the data for these batteries is marked as "-1", i.e., as a faulty battery.
S105, expanding the data dimension of the battery by constructing the characteristics of the basic attributes of the battery by using a characteristic engineering method;
in order to increase the dimension of battery data, the basic attributes of the battery are subjected to feature expansion through a feature engineering method, and the expanded features comprise basic features, unit features, time features and combination features.
The basic characteristics comprise 5 basic characteristics including voltage, current, resistance, temperature and discharge depth of a battery, and all the basic characteristic values can be directly obtained from a digital twin model of the UPS;
cell characteristics include voltage statistics of battery cells: mean value of voltage MeanVCoefficient of variation CoffVAnd Pearson ModeVAnd, resistance statistics: mean value of resistance MeanRCoefficient of variation CoffRAnd Pearson ModeRAnd 6 unit features.
Multiple storage batteries are connected in series to form individual battery pack units, the characteristic values of the battery pack units are calculated by a statistical method, and specifically, a voltage average Mean is calculated according to the acquired voltage of the batteriesVCoefficient of variation CoffVAnd Pearson ModeVThe formula of (1) is as follows:
Figure BDA0002692880480000054
Figure BDA0002692880480000061
ModeV＝MeanV-3(MeanV-MedianV)
where N denotes the number of blocks of the secondary battery in the battery unit, MedianVThe median of the voltages of the respective cells in the battery pack is shown.
It should be noted that: calculating the Mean value Mean of the resistance according to the obtained resistance of the batteryRCoefficient of variation CoffRAnd Pearson ModeRThe method of (3) is the same as the above-mentioned method of calculating the voltage average value, the coefficient of variation and the pearson coefficient, and is not described herein again.
The time characteristic comprising the voltage differential Dif of the batteryVDifferential resistance DifRVoltage trend quantity trendVAnd a resistance tendency amount trendR4 time features in total;
according to the acquired voltage and resistance of the battery, a voltage differential value, a resistance differential value and an early warning trend amount are calculated, and the relation of the battery parameter of the UPS changing along with time and the voltage trend amount trend can be well reflectedVThe calculation formula of (a) is as follows:
Figure BDA0002692880480000062
n, M are the number of time nodes, N and M are adjacent positive integers, and N is less than M.
When V isiOr VjLess than the voltage trend amount trendVThe Bool value is 1 and greater than or equal to trend at the lower boundary value of (1)VAt the lower boundary value of (1), Bool is 0, preferably trendVThe lower boundary value of (1) is 9V (volts). When trendVWhen the value of (d) is greater than 1, it indicates that the tendency to reach the set warning value is increasing.
Similarly, the resistance trend amount trendRThe calculation formula of (a) is as follows:
Figure BDA0002692880480000063
when the resistance R isiOr RjGreater than the resistance trend amount trendRWhen the upper boundary value of (1), Bool is 1 and is less than or equal to trendRAnd when the boundary value is higher, the Bool value is 0. Preferably, trendRHas an upper boundary value of 6m omega.
The combined characteristics include impedance to voltage ratio, i.e., the ratio of resistance to voltage, calculated as follows:
Figure BDA0002692880480000064
s106, balancing the data volume of the batteries with different power supply performances;
the battery is divided into a normal battery and a fault battery according to the power supply performance, the quantity of the fault battery is small, the quantity of fault data collected from a digital twin model of the UPS is far smaller than that of the normal battery, and in order to achieve a better model training effect, an improved downsampling (under-sampling) method and a K-fold cross validation (K-fold cross validation) method are adopted to solve the problem of imbalance between the two data quantities.
Specifically, the improved downsampling method can reduce the loss of data or information in the downsampling process, specifically can be balanceCascad, and the improved downsampling method is characterized in that a training set of a normal battery is generated through one-time downsampling operation by utilizing the principle of incremental training, and the training set is put into a classifier model; if the classification result is correct, taking out the sample, and continuing to perform downsampling operation on the rest training data to obtain a trained second classifier, … …, and sequentially obtaining an Nth classifier; the results from all classifiers are combined, where N may take 10 as the final training result.
The method of K-fold cross validation is adopted to avoid the problem of overfitting, the battery data is divided into K parts, one of the K parts is selected as test data, the rest K-1 parts are selected as training data, and K in the K parts can be 3, namely, the method of 3-fold cross validation is adopted.
S107, inputting the balanced battery data into a prediction model of the battery fault of the nonlinear support vector machine for training;
in this embodiment, the multidimensional feature constructed on the battery data belongs to table data, and in consideration of the accuracy, efficiency, and convenience of model training, a non-linear Support Vector Machine (SVM) classification model is adopted as an optimal model for the battery predictive operation and maintenance problem of the UPS, and the optimal model is used as a prediction model of the battery fault.
For the nonlinear classification problem in the input space, the algorithm is as follows:
inputting: each characteristic parameter D { (x) of the training data set of the battery data1,y1),(x2,y2),Λ(xN,yN) In which xiFor sample eigenvalues, each xiIncluding the expanded features in step S105: basic features, unit features, temporal features, and combination features; yi ∈ { -1, +1} is a label value, i ═ 1,2,. N, N denotes the number of samples participating in training. When y isiWhen equal to-1, a faulty battery is indicated; when y isiWhen +1 is equal, a normal battery is indicated.
And (3) outputting: separating the hyperplane and the classification decision function. Selecting a Gaussian kernel function K (x, z) and a penalty parameter C larger than 0, constructing and solving a convex quadratic programming problem to obtain an optimal solution
Figure BDA0002692880480000071
Wherein N represents the number of samples participating in training;
wherein, the Gaussian kernel function is as follows:
Figure BDA0002692880480000072
wherein z is the center of the Gaussian kernel function, and sigma is the width parameter of the Gaussian kernel function;
selection of alpha*A component of
Figure BDA0002692880480000073
Calculating a bias parameter b*The calculation formula is as follows:
Figure BDA0002692880480000074
calculating a classification decision function, wherein the selected gaussian kernel function corresponds to a gaussian radial basis function, and the classification decision function can be expressed as the following formula:
Figure BDA0002692880480000081
the method comprises the steps of training a nonlinear SVM classification model, specifically, using a Stochastic Gradient Descent (SGD) as an optimization algorithm, and operating the SGD by solving a minimum value as a cost function because an optimization target of the SVM is a convex function. The iteration rule of the nonlinear SVM can be expressed as:
Figure BDA0002692880480000082
where γ denotes a learning rate, and ω and b are a weight and a bias coefficient, respectively.
And S108, performing performance evaluation on the trained prediction model, and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
Before being input into a prediction model of a battery fault of the SVM for training, a data set is divided into a training set and a verification set in a ratio of 4: 1. Stopping iterative training when the optimal solution is obtained through iterative training, and evaluating the training effect of the model in the data of the training set and the data of the verification set respectively. When the predetermined better effect is obtained in the data of the training set and the verification set, the prediction model can be confirmed to be more suitable for solving the battery classification problem (namely, the battery is classified into a normal battery and a fault battery). And selecting the model with the best effect as a verified prediction model, and further using the model as a deployment prediction model of the battery fault of the uninterruptible power supply as a prediction model deployed on an application field to perform prediction of the battery fault.
The classification result is output as a normal battery and a fault battery, and is confirmed and judged by technical experts, if the power supply performance type is judged correctly (true positive), an early warning signal is sent out, and a maintenance request is provided; and if the power supply performance type is judged to be wrong (false positive), adding a new data sample, changing the model parameters, adjusting the model structure, and retraining to achieve the optimal training effect.
The accuracy accuracuracy, precision precison-recall curve and F1 score F are used for verifying the performance of model training, specifically:
Figure BDA0002692880480000083
Figure BDA0002692880480000084
Figure BDA0002692880480000085
Figure BDA0002692880480000086
wherein, TP represents that the prediction result is a normal battery, and the actual result is also a normal battery; TN indicates that the prediction result is a fault battery, and the actual result is also a fault battery; FP shows that the prediction result is a normal battery and the actual result is a fault battery; FN indicates that the predicted result is a faulty battery and the actual result is a normal battery. In order to prevent the error of the single index result which is inconsistent with the actual result in the detection, the precision (precision) and the recall (recall) are combined for use. Taking recall as an abscissa (taking 0.1 as a step length, dividing into 11 points of 0.0-1.0), and taking a precision value obtained by calculation as an ordinate to obtain a precision-recall curve (P-R curve for short). The Area enclosed by the graph and the horizontal axis (recall) is called AP (Area precision), and the mean value of all categories is called mAP (mean Area precision). If the value of mAP is larger, the training effect and the classification capability are better.
And deploying the trained model algorithm. The UPS digital twin body is used as a data middle station to form a closed loop structure of data acquisition, characteristic extraction, data processing, classification judgment and result feedback, the closed loop structure is used as a controller of the whole UPS predictability operation and maintenance system, a UPS physical entity is used as a given input, a UPS digital twin body is used as an output and feedback, and intellectualization is realized by combining a machine learning method, so that the complete UPS closed loop predictability operation and maintenance system is constructed.
In the embodiment, by establishing a digital twin model corresponding to the UPS, inputting the UPS battery data into a prediction model of the battery fault of the non-linear support vector machine for training in the digital twin model, performing performance verification on the prediction model obtained by training, and using the prediction model passing the verification as a deployment prediction model of the battery fault of the UPS, the dilemma that the UPS battery historical data and fault information are less and difficult to collect and analyze is effectively solved, and the characteristic engineering technology is applied to expand the data dimension aiming at the conditions that the UPS battery data dimension is less and the model training accuracy is influenced.
Referring to fig. 2, fig. 2 is a schematic structural diagram of an apparatus for implementing battery predictive maintenance of an uninterruptible power supply according to an embodiment of the present disclosure. The implementation device comprises:
the modeling module 201 is used for establishing a digital twin model corresponding to the uninterrupted power supply;
the acquisition module 202 is configured to continuously acquire real-time data of the uninterruptible power supply and transmit the real-time data to the digital twin model;
the processing module 203 is configured to obtain historical data and current state data of the battery operation from the digital twin model, obtain data of basic attributes of the battery in a time sequence from the historical data and the current state data, and obtain a basic rule of the battery power supply performance according to the data of the basic attributes;
the tag module 204 is configured to assign a tag value to the power supply performance of the data of the battery according to the characteristic of the power supply performance of the battery;
a feature module 205, configured to expand the data dimension of the battery by performing feature construction on the basic attributes by using a feature engineering method, where the constructed features include: basic features, unit features, temporal features, and combination features;
a balancing module 206 for balancing data amounts of the batteries with different power supply performances;
the training module 207 is used for inputting the balanced data of the battery into a prediction model of the battery fault of the nonlinear support vector machine for training;
and the evaluation module 208 is configured to perform performance evaluation on the trained prediction model, and use the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
Further, the historical data and the current state data comprise time nodes, the identity of the battery, current, voltage, internal resistance, temperature and depth of discharge;
the processing module 203 is further configured to remove invalid data in the historical data and the current state data; determining statistics of the obtained current, voltage, resistance, temperature and depth of discharge; and drawing a characteristic curve of the statistic, and obtaining a basic rule of basic attributes of the fault battery and the normal battery from the characteristic curve.
The tag module 204 is further configured to assign tag values to a normal battery with a normal power supply performance and a faulty battery with an abnormal power supply performance in the batteries according to the characteristics of the power supply performance of the batteries;
calculating and assigning label values to the replacement time points of the normal battery and the fault battery;
wherein the resistance value of the normal battery smoothly increases with time, and when the resistance value of the normal battery reaches a preset first resistance threshold value, or when the direct-current voltage of the normal battery is smaller than a preset first voltage threshold value, the normal battery reaches the replacement time point, and the calculation mode of the replacement time point of the normal battery is as follows:
tchange1＝min{tthR,tthV}-tinit
wherein, tchange1Represents the time point of the normal battery replacement, tthRIs the time point, t, when the resistance value of the normal battery reaches the first resistance threshold valuethVIs the time point t when the DC voltage of the normal battery is reduced to the first voltage threshold valueinitThe initial value of the battery service time is the initial value;
when the float voltage of the fault battery is reduced to a float voltage threshold value, the fault battery reaches the replacement time point, and the replacement time point of the fault battery is calculated as follows:
tchange2＝min{t|VDt≥2σmean}
wherein, tchange2Indicating the point in time of said faulty battery replacement, σmeanThe min operation is used for calculating the float voltage drop index VD as the average value of the standard deviation of the float voltage of all the normal batteriestNot less than float voltage threshold 2 sigmameanThe minimum time required;
calculating the float voltage drop index VD of the current time point ttThe calculation formula of (a) is as follows:
Figure BDA0002692880480000111
wherein N and T are time node ranges to be calculated; n is a counting variable of a time node, and N is more than or equal to 1 and less than or equal to N; i is a counting variable of a node of the time node, and n is more than or equal to T-2T and less than or equal to T-T; w is anWeights corresponding for n time nodesThe weight value of the weight is set to be,
Figure BDA0002692880480000112
is the sum of all weighted values;
Figure BDA0002692880480000113
represents the sum of the Ui values every two adjacent weeks; (t-n) represents n time points before the current time point t; u shapet-nIndicating the voltage value at the time of calculation (t-n);
Figure BDA0002692880480000114
the difference between the voltage at the time (t-n) and the average value is calculated, and the smaller value of the difference and 0 is taken.
A characteristic module 205, further configured to obtain a current, a voltage, a resistance, a temperature, and a depth of discharge of the battery from the digital twin model as basic characteristics;
calculating, as unit characteristics, voltage statistics and resistance statistics of a battery pack unit composed of the batteries, the voltage statistics including a voltage average value, a coefficient of variation, and a pearson mode, and the resistance statistics including a resistance average value, a coefficient of variation, and a pearson mode, based on the voltages and resistances of the batteries;
calculating voltage differential, resistance differential, voltage trend quantity and resistance trend quantity of the battery as time characteristics according to the voltage and the resistance of the battery;
and calculating the impedance-voltage ratio of the battery as a combined characteristic according to the voltage and the resistance of the battery.
A features module 205 to:
Figure BDA0002692880480000115
Figure BDA0002692880480000116
n, M is the number of time nodes, N and M are adjacent positive integers, and N is more than M;bool is Bool value; when V isiOr VjLess than the voltage trend amount trendVWhen V is greater than the lower boundary value, Bool is 1iOr VjGreater than or equal to trendVWhen the lower boundary value of (1), Bool is 0; when the resistance R isiOr RjGreater than the resistance trend amount trendRWhen the upper boundary value of (1) is not more than 1 and not more than trendRWhen the upper boundary value of (1) is greater than (0).
The balancing module 206 is further configured to balance the data amount of the normal cell and the data amount of the failed cell by combining the modified downsampling method with the K-fold cross validation method.
Further, the apparatus further comprises:
the battery fault prediction module comprises a dividing module, a judging module and a judging module, wherein the dividing module divides the battery data into training set data and verification set data according to a preset proportion, and the training set data is used for being input into a prediction model of the battery fault of the nonlinear support vector machine for training;
then, the training module 207 is further configured to input an input parameter of the prediction model, which includes a feature parameter formed by the constructed feature and the label value;
selecting a Gaussian kernel function K and a penalty parameter C, and obtaining an optimal solution alpha by C > 0*；
Wherein, the formula of the Gaussian kernel function K is as follows:
Figure BDA0002692880480000121
optimal solution alpha*Expressed as:
Figure BDA0002692880480000122
selection of alpha*A component of
Figure BDA0002692880480000123
Computing a bias term b*The calculation formula of (a) is as follows:
Figure BDA0002692880480000124
solving the optimal solution alpha*The classification decision function of (a) is:
Figure BDA0002692880480000125
and training the prediction model according to the input parameters and the classification decision function and a random gradient descent algorithm.
And the evaluation module 208 is further configured to verify the performance of model training in the verification set data through the accuracy, the accuracy-recall curve and the F1 score obtained through training, and determine the prediction model with the performance reaching a preset value as the verified prediction model.
In the embodiment, by establishing a digital twin model corresponding to the UPS, inputting the UPS battery data into a prediction model of the battery fault of the non-linear support vector machine for training in the digital twin model, performing performance verification on the prediction model obtained by training, and using the prediction model passing the verification as a deployment prediction model of the battery fault of the UPS, the dilemma that the UPS battery historical data and fault information are less and difficult to collect and analyze is effectively solved, and the characteristic engineering technology is applied to expand the data dimension aiming at the conditions that the UPS battery data dimension is less and the model training accuracy is influenced.
Referring to fig. 3, fig. 3 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present disclosure. The electronic apparatus described in this embodiment includes:
the memory 301, the processor 302, and a computer program stored in the memory 301 and executable on the processor 302, when the processor 302 executes the computer program, the method for implementing battery predictive maintenance of the uninterruptible power supply described in the embodiment of fig. 1 is implemented.
Further, the electronic device further includes:
at least one input device 303 and at least one output device 304.
The memory 301, the processor 302, the input device 303, and the output device 304 are connected via a bus 305.
The input device 303 may be a camera, a touch panel, a physical button, or the like. The output device 304 may specifically be a display screen.
The Memory 301 may be a Random Access Memory (RAM) Memory or a non-volatile Memory (non-volatile Memory), such as a magnetic disk Memory. The memory 301 is used to store a set of executable program code, and the processor 302 is coupled to the memory 301.
Further, an embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium may be an electronic device configured in the foregoing embodiments, and the computer-readable storage medium may be a storage unit configured in the main control chip and the data acquisition chip in the foregoing embodiments. The computer readable storage medium has stored thereon a computer program, which when executed by a processor implements the method for implementing battery predictive maintenance for an uninterruptible power supply as described in the embodiment of fig. 1.
Referring to fig. 4, fig. 4 is a schematic diagram of data interaction between the UPS physical entity 100 and the electronic device 200, wherein the UPS physical entity includes a battery, and the battery may be a storage battery. The electronic device 200 is provided with a device for implementing battery predictive maintenance of the uninterruptible power supply, and a digital twin model corresponding to the UPS physical entity 100 is established in the electronic device 200, wherein the digital twin model comprises a battery digital twin model corresponding to the battery. The electronic device 200 may be a computer or other intelligent terminal having a processor. The implementation method for battery predictive maintenance of the uninterruptible power supply in the embodiment shown in fig. 1 can be executed in the electronic device 200, and specific reference can be made to the description of the foregoing embodiments.
It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
The above description is provided for the implementation method, apparatus and electronic apparatus for predictive battery maintenance of an uninterruptible power supply, and for those skilled in the art, there may be variations in the specific implementation and application scope according to the ideas of the embodiments of the present application.

Claims (10)

1. A method for implementing predictive maintenance of a battery of an uninterruptible power supply, comprising:
establishing a digital twin model corresponding to the uninterrupted power supply;
continuously collecting real-time data of the uninterruptible power supply and transmitting the real-time data to the digital twin model;
acquiring historical data and current state data of the battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, and acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes;
according to the characteristics of the power supply performance of the battery, a label value is given to the power supply performance of the data of the battery;
expanding the data dimension of the battery by performing feature construction on the basic attributes by using a feature engineering method, wherein the constructed features comprise: basic features, unit features, temporal features, and combination features;
balancing the data volume of the batteries of different power supply performance;
inputting the balanced data of the battery into a prediction model of the battery fault of a nonlinear support vector machine for training;
and performing performance evaluation on the trained prediction model, and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
2. The method of claim 1, wherein the historical data and current state data include time nodes, an identity of the battery, current, voltage, internal resistance, temperature, and depth of discharge;
then, obtaining the basic attribute data of the battery with time as a sequence from the historical data and the current state data, and obtaining the basic rule of the battery power supply performance according to the data of the basic attribute of the battery, wherein the basic rule comprises the following steps:
removing invalid data in the historical data and the current state data;
determining statistics of the obtained current, voltage, resistance, temperature and depth of discharge;
and drawing a characteristic curve of the statistic, and obtaining a basic rule of basic attributes of the fault battery and the normal battery from the characteristic curve.
3. The method of claim 1 or 2, wherein said assigning a tag value to the powering performance of the data of the battery according to the characteristic of the powering performance of the battery comprises:
according to the characteristics of the power supply performance of the battery, label values are respectively given to a normal battery with normal power supply performance and a fault battery with abnormal power supply performance in the battery;
calculating and assigning label values to the replacement time points of the normal battery and the fault battery;
wherein the resistance value of the normal battery smoothly increases with time, and when the resistance value of the normal battery reaches a preset first resistance threshold value, or when the direct-current voltage of the normal battery is smaller than a preset first voltage threshold value, the normal battery reaches the replacement time point, and the calculation mode of the replacement time point of the normal battery is as follows:
tchange1＝min{tthR,tthV}-tinit
wherein, tchange1Represents the time point of the normal battery replacement, tthRIs the time point, t, when the resistance value of the normal battery reaches the first resistance threshold valuethVIs the time point t when the DC voltage of the normal battery is reduced to the first voltage threshold valueinitThe initial value of the battery service time is the initial value;
when the float voltage of the fault battery is reduced to a float voltage threshold value, the fault battery reaches the replacement time point, and the replacement time point of the fault battery is calculated as follows:
tchange2＝min{t|VDt≥2σmean}
wherein, tchange2Indicating the point in time of said faulty battery replacement, σmeanThe min operation is used for calculating the float voltage drop index VD as the average value of the standard deviation of the float voltage of all the normal batteriestNot less than float voltage threshold 2 sigmameanThe minimum time required;
calculating the float voltage drop index VD of the current time point ttThe calculation formula of (a) is as follows:
Figure FDA0002692880470000024
wherein N and T are time node ranges to be calculated; n is a counting variable of a time node, and N is more than or equal to 1 and less than or equal to N; i is a counting variable of a node of the time node, and n is more than or equal to T-2T and less than or equal to T-T; w is anIs the weight value corresponding to the n time nodes,
Figure FDA0002692880470000021
is the sum of all weighted values;
Figure FDA0002692880470000022
represents the sum of the Ui values every two adjacent weeks;(t-n) represents n time points before the current time point t; u shapet-nIndicating the voltage value at the time of calculation (t-n);
Figure FDA0002692880470000023
the difference between the voltage at the time (t-n) and the average value is calculated, and the smaller value of the difference and 0 is taken.
4. The method of claim 3, wherein the feature engineering method is used to feature the base attributes, and wherein expanding the dimensions of the input data comprises:
obtaining the current, the voltage, the resistance, the temperature and the discharge depth of the battery from the digital twin model as basic characteristics;
calculating, as unit characteristics, voltage statistics and resistance statistics of a battery pack unit composed of the batteries, the voltage statistics including a voltage average value, a coefficient of variation, and a pearson mode, and the resistance statistics including a resistance average value, a coefficient of variation, and a pearson mode, based on the voltages and resistances of the batteries;
calculating voltage differential, resistance differential, voltage trend quantity and resistance trend quantity of the battery as time characteristics according to the voltage and the resistance of the battery;
and calculating the impedance-voltage ratio of the battery as a combined characteristic according to the voltage and the resistance of the battery.
5. The method of claim 4, wherein calculating the voltage trend amount and the resistance trend amount of the battery as the time characteristic from the voltage and the resistance of the battery comprises:
calculating the voltage trend quantity and the resistance trend quantity according to the following calculation formula:
Figure FDA0002692880470000031
Figure FDA0002692880470000032
n, M is the number of time nodes, N and M are adjacent positive integers, and N is more than M; bool is Bool value; when V isiOr VjLess than the voltage trend amount trendVWhen V is greater than the lower boundary value, Bool is 1iOr VjGreater than or equal to trendVWhen the lower boundary value of (1), Bool is 0; when the resistance R isiOr RjGreater than the resistance trend amount trendRWhen the upper boundary value of (1) is not more than 1 and not more than trendRWhen the upper boundary value of (1) is greater than (0).
6. The method of claim 5, wherein the balancing the data volume of the batteries of different power delivery performance comprises:
and balancing the data quantity of the normal battery and the fault battery by combining an improved downsampling method with a K-fold cross validation method.
7. The method of claim 6, wherein the inputting the balanced data of the battery into a predictive model of battery failure of a non-linear support vector machine for training comprises:
dividing the battery data into training set data and verification set data according to a preset proportion, wherein the training set data is used for being input into a prediction model of the battery fault of the nonlinear support vector machine for training;
inputting the balanced data of the battery into a prediction model of the battery fault of the nonlinear support vector machine for training comprises:
inputting the input parameters of the prediction model into a feature parameter formed by the constructed feature and the label value;
selecting a Gaussian kernel function K and a penalty parameter C, and obtaining an optimal solution alpha by C > 0*；
Wherein, the formula of the Gaussian kernel function K is as follows:
Figure FDA0002692880470000033
optimal solution alpha*Expressed as:
Figure FDA0002692880470000034
wherein N represents the number of samples participating in training;
selection of alpha*A component of
Figure FDA0002692880470000041
Computing a bias term b*The calculation formula of (a) is as follows:
Figure FDA0002692880470000042
solving the optimal solution alpha*The classification decision function of (a) is:
Figure FDA0002692880470000043
and training the prediction model according to the input parameters and the classification decision function and a random gradient descent algorithm.
8. The method of claim 7, wherein the performing a performance evaluation on the trained predictive model and using the validated predictive model as a deployment predictive model for the battery failure of the uninterruptible power supply comprises:
and verifying the performance of model training in the verification set data through the accuracy, the accuracy-recall curve and the F1 score obtained through training, and confirming the prediction model with the performance reaching a preset value as the verified prediction model.
9. An apparatus for implementing predictive battery maintenance for an uninterruptible power supply, comprising:
the modeling module is used for establishing a digital twin model corresponding to the uninterrupted power supply;
the acquisition module is used for continuously acquiring real-time data of the uninterruptible power supply and transmitting the real-time data to the digital twin model;
the processing module is used for acquiring historical data and current state data of the battery operation from the digital twin model, acquiring data of basic attributes of the battery with time as a sequence from the historical data and the current state data, and acquiring a basic rule of the power supply performance of the battery according to the data of the basic attributes;
the tag module is used for endowing a tag value for the power supply performance of the data of the battery according to the characteristics of the power supply performance of the battery;
a feature module for expanding the data dimension of the battery by performing feature construction on the basic attributes by using a feature engineering method, the constructed features including: basic features, unit features, temporal features, and combination features;
the balancing module is used for balancing the data volume of the batteries with different power supply performances;
the training module is used for inputting the balanced data of the battery into a prediction model of the battery fault of the nonlinear support vector machine for training;
and the evaluation module is used for evaluating the performance of the trained prediction model and taking the verified prediction model as a deployment prediction model of the battery fault of the uninterruptible power supply.
10. An electronic device, comprising: memory, processor and computer program stored on the memory and executable on the processor, characterized in that the processor implements the method for implementing predictive maintenance of a battery of an uninterruptible power supply according to any of claims 1 to 8 when executing the computer program.
CN202010997018.9A 2020-09-21 2020-09-21 Method and device for realizing battery predictive maintenance of uninterruptible power supply and electronic device Active CN112347692B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010997018.9A CN112347692B (en) 2020-09-21 2020-09-21 Method and device for realizing battery predictive maintenance of uninterruptible power supply and electronic device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010997018.9A CN112347692B (en) 2020-09-21 2020-09-21 Method and device for realizing battery predictive maintenance of uninterruptible power supply and electronic device

Publications (2)

Publication Number Publication Date
CN112347692A true CN112347692A (en) 2021-02-09
CN112347692B CN112347692B (en) 2024-02-06

Family

ID=74357349

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010997018.9A Active CN112347692B (en) 2020-09-21 2020-09-21 Method and device for realizing battery predictive maintenance of uninterruptible power supply and electronic device

Country Status (1)

Country Link
CN (1) CN112347692B (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112993338A (en) * 2021-03-04 2021-06-18 上海电气集团股份有限公司 Monitoring method, device and equipment of energy storage system
CN113191502A (en) * 2021-04-21 2021-07-30 烽火通信科技股份有限公司 Artificial intelligence model on-line training method and system
CN113254311A (en) * 2021-05-31 2021-08-13 广州衡昊数据科技有限公司 Monitoring system and monitoring method of distributed power supply system
CN113406508A (en) * 2021-06-23 2021-09-17 苏州东吴智控科技有限公司 Battery detection and maintenance method and device based on digital twinning
CN113655382A (en) * 2021-06-23 2021-11-16 华北水利水电大学 Method for predicting UPS battery fault based on hydrogen generation rate ratio
CN115648944A (en) * 2022-10-25 2023-01-31 中国汽车技术研究中心有限公司 Early warning method, device, equipment and storage medium for power battery
CN116051079A (en) * 2023-03-28 2023-05-02 深圳市中科恒辉科技有限公司 Uninterrupted power source predictive maintenance system
CN116260235A (en) * 2023-03-11 2023-06-13 傲视恒安科技（北京）有限公司 Power supply switching method and device, electronic equipment and readable storage medium
CN116306287A (en) * 2022-06-30 2023-06-23 华北电力大学 Fan prediction data determining method, system, application and readable storage medium
CN117592977A (en) * 2024-01-19 2024-02-23 陕西万禾数字科技有限公司 Intelligent digital twin guarantee system for equipment full life cycle
CN117706379A (en) * 2024-02-06 2024-03-15 北京航空航天大学 Method and device for constructing dynamic safety boundary of battery and readable storage medium

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160195587A1 (en) * 2015-01-07 2016-07-07 Samsung Electronics Co., Ltd. Apparatus and method estimating state of battery pack including plural battery cells
EP3121614A1 (en) * 2015-07-21 2017-01-25 Samsung Electronics Co., Ltd. Method and apparatus for estimating state of health of battery
US20180136285A1 (en) * 2016-11-16 2018-05-17 Samsung Electronics Co., Ltd. Method and apparatus for estimating state of battery
CN108139446A (en) * 2015-10-15 2018-06-08 江森自控科技公司 For predicting the battery test system of cell testing results
CN108535656A (en) * 2018-03-22 2018-09-14 中北大学 Lithium ion battery remaining life prediction technique and system based on PCA-NARX neural networks
CN108805217A (en) * 2018-06-20 2018-11-13 山东大学 A kind of health state of lithium ion battery method of estimation and system based on support vector machines
CN109633477A (en) * 2019-01-04 2019-04-16 深圳市计量质量检测研究院 The method for real-time monitoring of battery state-of-health based on EKF-GPR and daily fragment data
CN109901075A (en) * 2017-12-11 2019-06-18 福特全球技术公司 Method for predicting battery life
CN109978252A (en) * 2019-03-22 2019-07-05 山东科技大学 A kind of high-power integrated fuel cell system running state prediction and evaluation method
CN110456635A (en) * 2019-07-02 2019-11-15 北京航空航天大学 The control method of power system of electric automobile based on the twin technology of number
CN110492186A (en) * 2019-07-02 2019-11-22 北京航空航天大学 A kind of management method of the power battery module based on cloud control technology
CN110568374A (en) * 2019-09-02 2019-12-13 东北电力大学 Lithium ion battery residual service life prediction method based on consideration of internal and external characteristics
CN111007417A (en) * 2019-12-06 2020-04-14 重庆大学 Battery pack SOH and RUL prediction method and system based on inconsistency evaluation
CN111027165A (en) * 2019-07-19 2020-04-17 北京航空航天大学 Power battery pack management system and method based on digital twinning
CN111157899A (en) * 2020-01-20 2020-05-15 南京邮电大学 Method for estimating SOC of battery based on model fusion idea
CN111398837A (en) * 2020-04-01 2020-07-10 重庆大学 Vehicle battery health state estimation method based on data driving
CN111651858A (en) * 2020-04-27 2020-09-11 平安城市建设科技(深圳)有限公司 Digital twin model generation method, device and computer readable storage medium
CN112904220A (en) * 2020-12-30 2021-06-04 厦门大学 UPS (uninterrupted Power supply) health prediction method and system based on digital twinning and machine learning, electronic equipment and storable medium
CN113109715A (en) * 2021-03-29 2021-07-13 东华大学 Battery health condition prediction method based on feature selection and support vector regression

Patent Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160195587A1 (en) * 2015-01-07 2016-07-07 Samsung Electronics Co., Ltd. Apparatus and method estimating state of battery pack including plural battery cells
EP3121614A1 (en) * 2015-07-21 2017-01-25 Samsung Electronics Co., Ltd. Method and apparatus for estimating state of health of battery
CN108139446A (en) * 2015-10-15 2018-06-08 江森自控科技公司 For predicting the battery test system of cell testing results
US20180136285A1 (en) * 2016-11-16 2018-05-17 Samsung Electronics Co., Ltd. Method and apparatus for estimating state of battery
CN109901075A (en) * 2017-12-11 2019-06-18 福特全球技术公司 Method for predicting battery life
CN108535656A (en) * 2018-03-22 2018-09-14 中北大学 Lithium ion battery remaining life prediction technique and system based on PCA-NARX neural networks
CN108805217A (en) * 2018-06-20 2018-11-13 山东大学 A kind of health state of lithium ion battery method of estimation and system based on support vector machines
CN109633477A (en) * 2019-01-04 2019-04-16 深圳市计量质量检测研究院 The method for real-time monitoring of battery state-of-health based on EKF-GPR and daily fragment data
CN109978252A (en) * 2019-03-22 2019-07-05 山东科技大学 A kind of high-power integrated fuel cell system running state prediction and evaluation method
CN110456635A (en) * 2019-07-02 2019-11-15 北京航空航天大学 The control method of power system of electric automobile based on the twin technology of number
CN110492186A (en) * 2019-07-02 2019-11-22 北京航空航天大学 A kind of management method of the power battery module based on cloud control technology
CN111027165A (en) * 2019-07-19 2020-04-17 北京航空航天大学 Power battery pack management system and method based on digital twinning
CN110568374A (en) * 2019-09-02 2019-12-13 东北电力大学 Lithium ion battery residual service life prediction method based on consideration of internal and external characteristics
CN111007417A (en) * 2019-12-06 2020-04-14 重庆大学 Battery pack SOH and RUL prediction method and system based on inconsistency evaluation
CN111157899A (en) * 2020-01-20 2020-05-15 南京邮电大学 Method for estimating SOC of battery based on model fusion idea
CN111398837A (en) * 2020-04-01 2020-07-10 重庆大学 Vehicle battery health state estimation method based on data driving
CN111651858A (en) * 2020-04-27 2020-09-11 平安城市建设科技(深圳)有限公司 Digital twin model generation method, device and computer readable storage medium
CN112904220A (en) * 2020-12-30 2021-06-04 厦门大学 UPS (uninterrupted Power supply) health prediction method and system based on digital twinning and machine learning, electronic equipment and storable medium
CN113109715A (en) * 2021-03-29 2021-07-13 东华大学 Battery health condition prediction method based on feature selection and support vector regression

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
刘大同 等: "锂离子电池健康评估和寿命预测综述", 仪器仪表学报, vol. 36, no. 1, pages 1 - 16 *
孙猛猛: "基于数据驱动方法的锂离子电池健康状态估计", CNKI优秀硕士学位论文全文库工程科技Ⅱ辑, no. 1, pages 1 - 93 *
应何杰: "锂离子动力电池安全性评估方法研究", CNKI优秀硕士学位论文全文库工程科技Ⅱ辑, no. 7, pages 1 - 84 *
王宁 等: "基于VPSO-SVM的磷酸铁锂电池寿命预测", 重庆理工大学学报(自然科学), vol. 33, no. 11, pages 173 - 177 *
胡晓亚 等: "锂离子电池健康状态估计方法研究综述", 电源学报, vol. 20, no. 1, pages 126 - 133 *
郑方丹: "基于数据驱动的多时间尺度锂离子电池状态评估技术研究", CNKI博士学位论文全文库工程科技Ⅱ辑, no. 12, pages 1 - 164 *
高栋: "基于新健康指标的锂离子电池剩余寿命预测", CNKI优秀硕士学位论文全文库工程科技Ⅱ辑, no. 7, pages 1 - 91 *

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112993338A (en) * 2021-03-04 2021-06-18 上海电气集团股份有限公司 Monitoring method, device and equipment of energy storage system
CN113191502A (en) * 2021-04-21 2021-07-30 烽火通信科技股份有限公司 Artificial intelligence model on-line training method and system
CN113254311B (en) * 2021-05-31 2022-12-13 广州衡昊数据科技有限公司 Monitoring system and monitoring method of distributed power supply system
CN113254311A (en) * 2021-05-31 2021-08-13 广州衡昊数据科技有限公司 Monitoring system and monitoring method of distributed power supply system
CN113655382B (en) * 2021-06-23 2023-11-14 华北水利水电大学 Method for predicting UPS battery fault based on hydrogen generation rate ratio
CN113655382A (en) * 2021-06-23 2021-11-16 华北水利水电大学 Method for predicting UPS battery fault based on hydrogen generation rate ratio
CN113406508A (en) * 2021-06-23 2021-09-17 苏州东吴智控科技有限公司 Battery detection and maintenance method and device based on digital twinning
CN116306287B (en) * 2022-06-30 2024-02-09 华北电力大学 Fan prediction data determining method, system, application and readable storage medium
CN116306287A (en) * 2022-06-30 2023-06-23 华北电力大学 Fan prediction data determining method, system, application and readable storage medium
CN115648944A (en) * 2022-10-25 2023-01-31 中国汽车技术研究中心有限公司 Early warning method, device, equipment and storage medium for power battery
CN115648944B (en) * 2022-10-25 2024-09-27 中国汽车技术研究中心有限公司 Early warning method, device, equipment and storage medium of power battery
CN116260235B (en) * 2023-03-11 2023-08-22 傲视恒安科技（北京）有限公司 Power supply switching method and device, electronic equipment and readable storage medium
CN116260235A (en) * 2023-03-11 2023-06-13 傲视恒安科技（北京）有限公司 Power supply switching method and device, electronic equipment and readable storage medium
CN116051079B (en) * 2023-03-28 2023-06-27 深圳市中科恒辉科技有限公司 Uninterrupted power source predictive maintenance system
CN116051079A (en) * 2023-03-28 2023-05-02 深圳市中科恒辉科技有限公司 Uninterrupted power source predictive maintenance system
CN117592977A (en) * 2024-01-19 2024-02-23 陕西万禾数字科技有限公司 Intelligent digital twin guarantee system for equipment full life cycle
CN117592977B (en) * 2024-01-19 2024-04-09 陕西万禾数字科技有限公司 Intelligent digital twin guarantee system for equipment full life cycle
CN117706379A (en) * 2024-02-06 2024-03-15 北京航空航天大学 Method and device for constructing dynamic safety boundary of battery and readable storage medium
CN117706379B (en) * 2024-02-06 2024-04-12 北京航空航天大学 Method and device for constructing dynamic safety boundary of battery and readable storage medium

Also Published As

Publication number Publication date
CN112347692B (en) 2024-02-06

Similar Documents

Publication Publication Date Title
CN112347692B (en) Method and device for realizing battery predictive maintenance of uninterruptible power supply and electronic device
Cadini et al. State-of-life prognosis and diagnosis of lithium-ion batteries by data-driven particle filters
Tao et al. A novel support vector regression method for online reliability prediction under multi-state varying operating conditions
JP7031649B2 (en) Evaluation device, computer program and evaluation method
CN116113961A (en) Battery life prediction using machine learning model
CN116449218B (en) Lithium battery health state estimation method
Goodwin et al. A pattern recognition approach for peak prediction of electrical consumption
US20230273262A1 (en) Battery capacity estimation apparatus and method
JP7546702B2 (en) Battery abnormality diagnosis device and method
KR20240015240A (en) Method and System for Estimating Online Charging and Health Status of Lithium Batteries Based on Neural Network Model Banks
CN112800231A (en) Power data verification method and device, computer equipment and storage medium
CN117543791B (en) Power supply detection method, device, equipment and storage medium for power supply
EP4300109A1 (en) Device and method for predicting low voltage failure of secondary battery, and battery control system comprising same device
WO2023052910A1 (en) System and method for estimating state of health and remaining useful life of a battery
CN116794510A (en) Fault prediction method, device, computer equipment and storage medium
Asefi et al. Power system anomaly detection and classification utilizing WLS-EKF state estimation and machine learning
Gou et al. Remaining useful life prediction for lithium-ion battery using ensemble learning method
EP4235198A1 (en) Apparatus and method for managing battery
Reza et al. Recent advancement of remaining useful life prediction of lithium-ion battery in electric vehicle applications: A review of modelling mechanisms, network configurations, factors, and outstanding issues
Mehta et al. A Comprehensive study of Machine Learning Techniques used for estimating State of Charge for Li-ion Battery
US20230184838A1 (en) Device and method for diagnosing battery
Li et al. Holistic comparison of different kernel functions for support vector regression based on state-of-health prediction of lithium-ion battery
Sun et al. Data-driven state-of-charge estimation of a lithium-ion battery pack in electric vehicles based on real-world driving data
Mostafa Battery total capacity estimation based on the sunflower algorithm
CN117150368B (en) Printed circuit board fault diagnosis method, system, equipment and storage medium

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
CB02 Change of applicant information

Address after: 518000 Room 301, building 1, fengyesheng factory building, Chuangye 2nd Road, Dalang community, Xin'an street, Bao'an District, Shenzhen City, Guangdong Province

Applicant after: Shenzhen Youdian IOT Technology Co.,Ltd.

Address before: Room 201, Building A, No. 1 Qianwan Road, Qianhai Shenzhen-Hong Kong Cooperation Zone, Nanshan District, Shenzhen City, Guangdong Province

Applicant before: SHENZHEN QIANHAI YESPOWERING IOT TECHNOLOGY Co.,Ltd.

CB02 Change of applicant information
GR01 Patent grant
GR01 Patent grant