CN117074965B - Lithium ion battery remaining life prediction method and system - Google Patents

Lithium ion battery remaining life prediction method and system Download PDF

Info

Publication number
CN117074965B
CN117074965B CN202311337131.4A CN202311337131A CN117074965B CN 117074965 B CN117074965 B CN 117074965B CN 202311337131 A CN202311337131 A CN 202311337131A CN 117074965 B CN117074965 B CN 117074965B
Authority
CN
China
Prior art keywords
data
battery
life
lithium battery
abnormal
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.)
Active
Application number
CN202311337131.4A
Other languages
Chinese (zh)
Other versions
CN117074965A (en
Inventor
陈明香
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 Shentong World Technology Co ltd
Original Assignee
Shenzhen Shentong World 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 Shentong World Technology Co ltd filed Critical Shenzhen Shentong World Technology Co ltd
Priority to CN202311337131.4A priority Critical patent/CN117074965B/en
Publication of CN117074965A publication Critical patent/CN117074965A/en
Application granted granted Critical
Publication of CN117074965B publication Critical patent/CN117074965B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/367Software therefor, e.g. for battery testing using modelling or look-up tables
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/382Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/392Determining battery ageing or deterioration, e.g. state of health
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02EREDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/10Energy storage using batteries

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Secondary Cells (AREA)
  • Tests Of Electric Status Of Batteries (AREA)

Abstract

The invention relates to the technical field of lithium ion batteries, in particular to a method and a system for predicting the residual life of a lithium ion battery. The method comprises the following steps: acquiring lithium battery electric data and lithium battery working condition data, and constructing a lithium battery characteristic matrix; detecting abnormal time window of the lithium battery feature matrix to obtain time window data; performing dimension reduction processing on the time window data to obtain low-dimension window data, and constructing a battery life prediction model; performing sequence prediction on the low-dimensional window data so as to obtain a residual life sequence; carrying out morphological analysis according to the residual life sequence to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics to obtain a life dynamic prediction model; and predicting the residual life of the lithium battery by using the life dynamic prediction model to obtain the predicted residual life of the lithium ion battery. The invention predicts the residual life of the lithium battery based on data mining.

Description

Lithium ion battery remaining life prediction method and system
Technical Field
The invention relates to the technical field of lithium ion batteries, in particular to a method and a system for predicting the residual life of a lithium ion battery.
Background
Lithium ion batteries are widely used in the fields of modern electronic devices, electric automobiles and the like, but their performance gradually declines with time and use, which results in limitation of battery life. The lifetime of a battery refers to the time that the battery can maintain its performance under specific conditions of use. Because of the relatively high cost of battery replacement and maintenance, accurate prediction of the remaining life of a battery is critical to improving battery sustainability and reducing maintenance costs.
Disclosure of Invention
Accordingly, there is a need for a method for predicting remaining life of a lithium ion battery, so as to solve at least one of the above-mentioned problems.
In order to achieve the above purpose, a method for predicting the remaining life of a lithium ion battery includes the following steps:
step S1: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor, and constructing a lithium battery feature matrix according to the lithium battery electrical data and the lithium battery working condition data;
step S2: detecting abnormal time window of the lithium battery feature matrix, so as to obtain time window data;
step S3: performing embedded dimension reduction processing on the time window data to obtain low-dimension window data, and constructing a battery life prediction model according to the low-dimension window data;
Step S4: performing sequence prediction on the low-dimensional window data by using a battery life prediction model so as to obtain a residual life sequence;
step S5: carrying out morphological analysis according to the residual life sequence to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics to obtain a life dynamic prediction model;
step S6: and predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using a life dynamic prediction model, so as to obtain the predicted residual life of the lithium ion battery.
According to the invention, the electrical data and the working condition data of the battery are obtained through the preset sensor, and the data comprise information such as voltage, current, temperature and the like. The data can be integrated by constructing the lithium battery feature matrix, so that multiple parameters are comprehensively considered, and a basis is provided for subsequent analysis. This helps to capture the behavior and performance of the battery. Time window anomaly detection helps identify anomalies in the data, for example, a battery may exhibit unusual behavior in certain situations, such as an abnormal discharge or charging process. By identifying these anomalies, it is possible to more accurately capture signs of deterioration of battery life, discover problems in advance, and take action. The embedded dimension reduction process helps reduce the dimensionality of the data while retaining key information to avoid overfitting. The battery life prediction model may build a model of battery life degradation from the reduced-dimension data, which helps accurately predict future performance. The low-dimensional window data is sequence predicted using a battery life prediction model, which helps predict the life remaining of the battery. Through the sequence prediction, the performance of the battery in a future time period can be estimated, and the possible problems of the battery are warned in advance, so that the maintenance plan is more targeted. Morphological analysis helps capture features in the battery life sequence, such as the trend of decay rate, periodic behavior, etc. According to the morphological characteristics, the dynamic parameter optimization can be carried out on the battery life prediction model so as to adapt to battery performance changes in different stages, and the adaptability and the accuracy of the prediction model are improved. The remaining life prediction of the battery electrical data and the operating condition data using the life dynamic prediction model allows the system to predict the remaining life of the battery based on the current battery state and environmental conditions. This helps to optimize battery usage and maintenance strategies, reducing unnecessary replacement and repair costs.
Optionally, step S1 specifically includes:
step S11: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor;
step S12: carrying out correlation analysis on lithium battery electric data and lithium battery working condition data so as to obtain a correlation matrix;
step S13: detecting abnormal battery operation of the lithium battery electrical data and lithium battery working condition data, so as to obtain abnormal battery operation data;
step S14: extracting time domain features of lithium battery electric data and lithium battery working condition data, thereby obtaining lithium battery time domain features;
step S15: extracting frequency domain features of lithium battery electrical data and lithium battery working condition data, thereby obtaining lithium battery frequency domain features;
step S16: and performing incidence matrix mapping on the time domain features of the lithium battery, the frequency domain features of the lithium battery and the abnormal working data through the correlation matrix, so as to obtain a lithium battery feature matrix.
The invention obtains the battery electric data and the working condition data through the sensor as a key starting point. These data include information about current, voltage, temperature, etc., which represent the actual operating conditions of the battery. This helps to capture the battery's behavior and provides the basis data for subsequent analysis. Correlation analysis helps to understand the relationship between different parameters. The correlation matrix may show the degree of correlation between the various parameters. This helps determine which parameters have the greatest impact on battery performance and life, thereby guiding the subsequent feature extraction and modeling process. The battery malfunction detection helps to identify abnormal behavior during battery operation. These anomalies may be caused by internal battery problems, external environmental factors, or system failures. By identifying and recording these anomalies, it is possible to help predict degradation of battery life and take corrective action to repair or replace in advance. Time domain feature extraction facilitates capturing time-dependent features from battery electrical data. This may include statistical properties of the current and voltage, such as mean, variance, waveform shape, etc. Time domain feature extraction helps to identify battery performance variations and trends at different points in time. The frequency domain feature extraction helps to translate battery electrical data into information on the frequency domain. This may include spectral analysis for detecting frequency components that may be present in the battery. Frequency domain feature extraction may help identify oscillations or periodic behavior in the battery that may be related to life degradation. And performing associated matrix mapping on the time domain features, the frequency domain features and the abnormal working data through the associated matrix, thereby being beneficial to integrating various feature information and constructing a comprehensive lithium battery feature matrix. This feature matrix integrates data from different sources to more fully describe the state and performance of the battery. This provides a richer input to the subsequent life prediction model.
Optionally, step S12 specifically includes:
feature screening is carried out on the lithium battery electrical data and the lithium battery working condition data, so that electrical feature data and working condition feature data are obtained;
performing covariance matrix calculation on the electrical characteristic data so as to obtain an electrical covariance matrix;
performing covariance matrix calculation on the working condition characteristic data so as to obtain a working condition covariance matrix;
and combining matrix elements according to the electric covariance matrix and the working condition covariance matrix, so as to obtain a correlation matrix.
The electrical data and operating condition data in the present invention typically include a large number of features, some of which may be less relevant or redundant. Feature screening can help reduce the dimensionality of the data, improving computational efficiency and model performance. The screening of the most relevant features is helpful for constructing a more accurate model, reducing the influence of noise and improving the prediction accuracy. The characteristics screened are generally easier to interpret, which is important for understanding the influencing factors of the performance of lithium batteries. The electrical covariance matrix helps analyze correlations between electrical features. This may reveal which electrical features are in a linear relationship with each other, helping to understand the physical process inside the battery. The covariance matrix may be used to detect multiple collinearity between features, i.e., some features are highly correlated. This helps to avoid introducing redundant information during the modeling process. The operating mode covariance matrix provides relationship information between operating mode features. This is useful for understanding the behavior of lithium batteries under different operating conditions. By combining the electrical covariance matrix and the operating mode covariance matrix, a comprehensive correlation matrix can be obtained, wherein the correlation matrix comprises the relationship between electrical characteristics and operating mode characteristics. This helps to comprehensively consider the impact of the electrical performance and operating conditions of a lithium battery on its life and performance.
Optionally, step S13 specifically includes:
step S131: carrying out time sequence combination on the lithium battery electrical data and the lithium battery working condition data so as to obtain data to be detected abnormally;
step S132: performing abnormal work score calculation on the data to be detected abnormal through an abnormal work score calculation formula, so as to obtain abnormal work score data;
step S133: carrying out statistical analysis according to the abnormal work score data so as to obtain an abnormal work threshold;
step S134: and carrying out classified calculation on the data to be detected abnormally through an abnormal working threshold value, so as to obtain the working abnormal data.
The electrical data and the operating mode data in the invention are usually from different sources, and can be integrated into one data set through time sequence combination. This helps to establish a correlation between the operating state and performance of the lithium battery. The abnormal operation score calculation can identify an abnormal situation in the battery operation. This may include battery temperature anomalies, voltage fluctuations, capacity drops, etc. By calculating the abnormal work score, the degree of abnormality can be quantified, facilitating further analysis and processing. These work scores may help detect anomalies early in the process, thereby taking precautions, extending the life of the lithium battery, and improving safety. Statistical analysis may help determine a threshold for abnormal job scores. These thresholds may be determined based on historical data and performance requirements of the model. The setting of the threshold determines when the battery is considered abnormal, triggering further processing. The statistical analysis can help to eliminate accidental anomalies, improve the accuracy of anomaly detection and reduce false alarms. The data to be detected can be classified into normal and abnormal operation data according to the abnormality threshold. This helps to identify and record anomalies in the battery in a timely manner. The anomaly data can be used to further analyze the root cause of the problem, helping to make fault analysis and maintenance decisions. This can save maintenance costs and improve the reliability of the battery system.
Alternatively, the abnormal work score calculation formula in step S132 is specifically:
in the method, in the process of the invention,for abnormal work score, < >>For lithium battery voltage, ">For observing time, < >>For the current value of the lithium battery, +.>For the capacity of lithium batteries, +.>For the battery temperature value, ">Charge rate for lithium battery->For lithium battery discharge rate,/->Is a comprehensive parameter of working condition>For the charge rate weight, +.>Is discharge rate weight +.>The average operating time of the lithium battery.
The invention constructs an abnormal work score calculation formula for calculating the abnormal work score of the data to be detected abnormally. The formula fully considers influencing abnormal work scoresLithium battery voltage>Observation time->Current value of lithium cell->Capacity of lithium battery->Battery temperature value->Battery charge rate->Lithium battery discharge rate->Condition comprehensive parameter->Charge rate weight->Discharge rate weight->Average operating time of lithium battery->A functional relationship is formed:
wherein the method comprises the steps ofThis part represents the square of the speed of the voltage change over time. The change speed of the voltage can reflect the dynamic change inside the battery, and the battery usually generates voltage fluctuation due to different factors (such as current change, temperature change and the like) in operation. The square of this section indicates that higher voltage change rates have a greater impact on the abnormal operation score. / >This part contains the current value of the battery +.>And Capacity value->Is a relationship of (3). It first calculates +.>And->Then square it and add 1, and finally take the logarithm. The purpose of this section is to take into account the correlation between current and battery capacity, as high currents can cause the battery to discharge rapidly, affecting the performance and life of the battery. />This part represents the square root of the temperature value of the battery. The temperature of the battery has an important influence on its performance and safety. In general, the higher the temperature, the more vulnerable the battery is, and therefore the square root of the temperature is used to represent the temperature-dependent effects in the abnormal operation score. />This part is the charge rate +.>And charge rate weight->Is a product of (a) and (b). It takes into account the effect of the battery charge rate on the operating conditions. If the charge rate is high, the battery may be overcharged or heated, thereby causing an abnormal situation. />This part is the discharge rate +.>And discharge rate weight->Is a product of (a) and (b). It takes into account the effect of the battery discharge rate on the operating conditions. Higher discharge rates may lead to overdischarge or rapid wear of the battery and may also cause abnormal situations. />This part is the condition combination parameter +. >Is multiplied by the weight of the average operating time of the lithium battery +.>. The working condition comprehensive parameters comprise a plurality of factors, and a plurality of working condition factors such as current, temperature, voltage and the like are comprehensively considered. Squaring emphasizes the importance of the operating mode complex parameters to the abnormal operating score, whereas +.>A trade-off consideration for operating time is expressed. The formula takes into account a number of factors related to battery performance, including current, voltage, temperature, charge and discharge rates, and the like. The abnormal working score obtained through calculation can be used for monitoring the working state of the battery in real time. If the score exceeds a certain threshold, it may indicate that the battery is experiencing an abnormal condition, such as overdischarge, overcharge, temperature abnormality, etc. This helps to find potential problems early, take precautions, and reduce the risk of battery damage.
Optionally, step S2 specifically includes:
step S21: defining a time window and a sliding step length according to the data to be detected abnormally, so as to obtain the time window and the sliding step length;
step S22: matrix segmentation is carried out on the lithium battery characteristic matrix through a time window and a sliding step length, so that a time window data set is obtained;
step S23: constructing a periodic anomaly detection model according to the time window data set and the data to be detected anomalies;
Step S24: performing periodic anomaly detection on the lithium battery feature matrix through a periodic anomaly detection model, so as to obtain charging periodic anomaly data;
step S25: and carrying out abnormal marking on the lithium battery characteristic matrix by using the charging period abnormal data so as to obtain time window data.
According to the invention, the data can be divided into small time windows by detecting the abnormal time window of the lithium battery characteristic matrix, so that the working mode and the performance of the battery can be better understood. This helps to detect changes in battery behavior over different time periods, including fluctuations in battery charge, discharge, temperature, etc. Identifying abnormal behavior of the battery within different time windows, such as voltage fluctuations, current anomalies, etc., helps to more finely monitor the performance and health of the battery. Defining the time window and the sliding step size is helpful to flexibly adjust the division mode of the data so as to adapt to different analysis requirements. The definition of the time window may affect the sensitivity and efficiency of subsequent anomaly detection. Depending on the particular application scenario, different time windows and sliding steps may be selected, for example, shorter time windows and sliding steps may be required for real-time monitoring, while longer time windows may be required for historical data analysis. And cutting the lithium battery characteristic matrix through a time window and a sliding step length to obtain a series of time window data sets. This facilitates a segmented analysis of the data to better understand the battery's operating mode and performance changes over different time periods. These time window datasets can be used for subsequent anomaly detection and model construction to help discover the abnormal behavior of the battery over different time periods. According to the time window data set and the data to be detected abnormally, a periodic abnormal detection model is constructed, so that the periodic abnormal behavior of the battery can be recognized. This can help distinguish between normal duty cycles and potential problems. The periodic anomaly detection model can be used for predicting the abnormal condition of the battery in a future period and finding the abnormal behavior in the working period of the battery in real-time monitoring. And the periodic anomaly detection can be carried out on the lithium battery characteristic matrix through the periodic anomaly detection model. This helps identify abnormal behavior of the battery during a particular duty cycle, such as abnormal current or voltage fluctuations of the battery during each charge cycle. Periodic anomaly detection can help discover periodic problems early, thereby taking steps to reduce the risk of battery damage or failure. By using the charge cycle abnormal data to perform abnormal marking on the lithium battery characteristic matrix, abnormal conditions and time window data can be associated, so that the time and mode of abnormal behaviors can be better known. The anomaly flags help to visualize and aggregate anomaly data for subsequent decisions and actions.
Optionally, step S24 specifically includes:
extracting the time dimension of the time window data set, thereby obtaining the time dimension;
extracting sensor space data from the data to be detected abnormally so as to obtain the sensor space data;
carrying out space sequence arrangement according to the sensor space data so as to obtain a space dimension;
extracting lithium battery characteristics from the time window data set, thereby obtaining characteristic dimensions;
constructing a 3D tensor according to the space dimension, the time dimension and the characteristic dimension;
and constructing a periodic anomaly detection model through the time window dataset and the 3D tensor.
The time dimension is extracted to organize the data according to the time sequence, so that the analysis of the change trend of the data along with time is facilitated. This may reveal a time dependence of battery performance and anomalies. The time dimension extraction helps to build time series data, allowing analysis of the historical evolution of battery behavior. The extraction of sensor spatial data can be focused on the sensor measurements, an important component in understanding battery status. By extracting the sensor data, the physical characteristics of the battery system can be captured. These data are used to analyze sensor changes, such as voltage, temperature, current, etc., to detect potential anomalies or problems. By arranging the sensor spatial data, a data structure with spatial dimensions can be constructed, which helps to better understand the interrelationship between the different sensors in the battery system. This helps to capture spatially anomalous patterns in the battery system, such as interactions between sensors, or anomalies in specific locations. Extracting lithium battery characteristics can capture key characteristics of battery performance, such as capacity, internal resistance, charge/discharge efficiency, and the like. This helps focus on key parameters in anomaly detection. The extracted features can be used to build an anomaly detection model to identify abnormal behavior in terms of battery performance. Combining the temporal dimension, the spatial dimension, and the feature dimension into a 3D tensor structure helps to comprehensively consider multiple data dimensions. This allows simultaneous analysis of information of different dimensions. The construction of 3D tensors allows anomaly detection using multidimensional data, for example taking into account anomaly behaviour in time, space and features simultaneously. By using the time window dataset and the 3D tensor, constructing a periodic anomaly detection model can help detect periodic anomalies in the battery system. This helps to find abnormal behavior of the battery under certain time and space conditions. The model can be used for monitoring and predicting the periodic problems in the battery system in real time so as to take maintenance and preventive measures early and improve the reliability and performance of the battery.
Optionally, step S4 specifically includes:
carrying out charge cycle anomaly flag extraction on the low-dimensional window data so as to obtain anomaly flag data points;
calculating abnormal mark data points and lithium battery working condition data through an abnormal charge influence calculation formula, so as to obtain abnormal charge film loudness;
the abnormal film filling loudness calculation formula specifically comprises:
in the method, in the process of the invention,to abnormally fill the film loudness->For charging time, +.>For lithium battery voltage, ">For observing time, < >>For marking data points for anomaliesLithium battery capacity->For the average operating time of lithium batteries, < >>For internal resistance of battery->Characteristic index for outlier marker data points, +.>For the feature total number of outlier marker data points, +.>The rate of change over time of the characteristic value for the outlier marker data point;
the invention constructs an abnormal charging film loudness calculation formula for calculating abnormal mark data points and lithium battery working condition data. The formula fully considers influencing the abnormal charge influence degreeCharging time of->Lithium battery voltage>Observation time->Lithium battery capacity of abnormal marker data points>Average operating time of lithium battery->Internal resistance of battery->Characteristic index of outlier marker data points +. >Total number of features of outlier marker data points +.>Characteristic value change rate of abnormal marker data point with time +.>A functional relationship is formed:
wherein the method comprises the steps ofIs the battery voltage +.>Relative to time->And represents the slope of the voltage. The voltage may change over time as the battery is charged or discharged. This part represents the rate of change of the voltage. />This is the average operating time of the lithium battery +.>Battery capacity with outlier marker data points +.>Natural logarithm of the ratio. It represents the capacity usage of the battery over an average operating time. If the battery capacity->Far below average working time +.>Indicating that the battery may be at a higher levelRun out in a short time, thereby affecting battery life. />This part represents the charging time +.>And (2) internal resistance of battery>The square root of the ratio. Internal resistance of battery->Has important influence on the performance and charging process of the battery. If the charging time is->Relative to the internal resistance of the battery>Larger, may cause heat accumulation inside the battery, thereby affecting the life of the battery. />This means +.>Summing is performed. />The total number of features, which are outlier marker data points, represents a comprehensive consideration of all features. / >This is a feature +.>Representing the relative importance of different features to the loudness of an abnormally charged film. Different features may affect the life of the battery in different ways, thus taking into account their contribution by giving different weights. />This is a feature +.>Over time->Is a rate of change of (c). It represents the features->The change in time was observed. Different features may change in different ways during charging, this part taking into account the effects of these changes. Finally, the results of the above-mentioned parts are integrated (from 0 to +.>) Obtaining the abnormal filling film loudness +.>. The influence can comprehensively consider the influence of voltage change, battery capacity use, the relationship between charging time and internal resistance and the change of a plurality of characteristics on the service life of the battery. By this calculation formula, the influence of abnormal charge can be quantified, which is helpful for further analysis and management of the battery system to improve the reliability and lifetime of the battery.
Predicting the future battery life of the low-dimensional window data through a battery life prediction model, so as to obtain a future life sequence;
and performing sequence correction on the future life sequence based on the abnormal film filling loudness, so as to obtain a residual life sequence.
By identifying and marking the charge cycle abnormal data points, the invention can determine abnormal conditions, such as too fast charge or too slow charge, occurring in the battery charging process. The outlier marker data points can be used in the outlier charge movie loudness calculation in subsequent steps to help understand the impact of outlier charge on battery life. By calculating the abnormal charge film loudness, the potential extent of the influence of abnormal charge on battery life can be quantified. This helps identify which abnormal charge conditions most severely affect the life of the battery. Predicting the low-dimensional window data using a battery life prediction model may estimate the life of the future battery. This may help to plan maintenance and better manage battery usage. By correcting the future life sequence based on the abnormal charge film loudness, the remaining life of the battery can be estimated more accurately. This helps to avoid unnecessary maintenance or replacement while improving the reliability of the battery. The corrected remaining life sequence may be used to make a more accurate maintenance schedule to ensure that the battery does not fail until the end of its life.
Optionally, step S5 specifically includes:
smoothing the residual life sequence to obtain a smooth life sequence;
Trend analysis is carried out according to the smooth life sequence, so that life trend is obtained;
performing Fourier transform on the smooth life sequence, thereby obtaining a life spectrum;
performing Fourier transformation according to the charging period abnormal data, so as to obtain a charging period abnormal frequency spectrum;
periodically analyzing according to the life spectrum to obtain life periodicity, and parameter marking the life periodicity according to the abnormal charging period spectrum to obtain marked life periodicity;
carrying out peak extraction on the smooth life sequence so as to obtain a life peak point;
combining the life peak point, the marked life periodicity and the life trend data, thereby obtaining morphological characteristics;
and carrying out dynamic parameter optimization on the battery life prediction model according to morphological characteristics, thereby obtaining the life dynamic prediction model.
The smoothing processing in the invention can remove noise and fluctuation in the data, so that the service life data has better interpretability and the randomness is reduced. The smoothed data can be used to more accurately estimate battery life and reduce errors due to noise. Trend analysis can help identify the trend of battery life over time, whether it is gradually decaying or continuously stabilizing. Trend analysis provides important insight into the state of health of the battery, helping to determine if the battery is at the end of life. The fourier transform may transform the lifetime data from the time domain to the frequency domain, revealing potentially periodic or frequency components. Life spectrum analysis can be used to detect periodic patterns in battery life, which is very useful for finding periodic factors in life. Fourier transforms are used to analyze the charge cycle anomaly data to help find charge cycle problems that may lead to life anomalies. Abnormal spectral analysis may be used to identify abnormal behavior during a charging cycle, which may affect battery life. Periodic analysis may help determine periodic patterns over battery life and flag these patterns through the charge cycle anomaly spectrum for further analysis. This helps identify the periodicity factor that leads to battery life anomalies and may guide maintenance strategies. Extracting life peak points helps identify key events or incidents in battery life. The peak point of life may be used to mark a significant change in battery performance, which may be related to the critical point of life. Combining peak life points, signature life periodicity, and life trends into morphological features can provide comprehensive information about battery life status. These features can be used to optimize a life prediction model to more accurately estimate the life of the battery. By optimizing dynamic parameters of the battery life prediction model based on morphological characteristics, a more accurate life prediction model can be established, and the model can monitor the life of the battery in real time and can be adjusted according to actual conditions. This model can be used to predict the remaining life of the battery, helping to take appropriate maintenance and replacement strategies, improving the reliability and performance of the battery.
Optionally, the present specification further provides a lithium ion battery remaining life prediction system, which includes:
the characteristic matrix construction module is used for acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor and constructing a lithium battery characteristic matrix according to the lithium battery electrical data and the lithium battery working condition data;
the time window detection module is used for detecting the abnormality of the time window of the lithium battery characteristic matrix so as to obtain time window data;
the model construction module is used for carrying out embedded dimension reduction processing on the time window data so as to obtain low-dimensional window data, and constructing a battery life prediction model according to the low-dimensional window data;
the sequence prediction module is used for performing sequence prediction on the low-dimensional window data by using the battery life prediction model so as to obtain a residual life sequence;
the morphological analysis module is used for carrying out morphological analysis according to the residual life sequence so as to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics so as to obtain a life dynamic prediction model;
and the residual life prediction module is used for predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using the life dynamic prediction model so as to obtain the predicted residual life of the lithium ion battery.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of a non-limiting implementation, made with reference to the accompanying drawings in which:
FIG. 1 is a flow chart of the steps of the method for predicting the remaining life of a lithium ion battery according to the present invention;
FIG. 2 is a detailed step flow chart of step S1 of the present invention;
FIG. 3 is a detailed step flow chart of step S2 of the present invention;
the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The following is a clear and complete description of the technical method of the present patent in conjunction with the accompanying drawings, and it is evident that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
Furthermore, the drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. The functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor methods and/or microcontroller methods.
It will be understood that, although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
In order to achieve the above objective, referring to fig. 1 to 3, the present invention provides a method for predicting remaining life of a lithium ion battery, the method comprising the following steps:
step S1: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor, and constructing a lithium battery feature matrix according to the lithium battery electrical data and the lithium battery working condition data;
in this embodiment, a preset sensor is used to collect electrical data and working condition data of the lithium battery. The electrical data may include voltage, current, temperature, etc., while the operating condition data may include state of charge, state of discharge, temperature, humidity, etc. These data will be integrated and processed to construct a feature matrix for the lithium battery. For example, voltage and current data per second may be collected and characteristics such as average current, maximum voltage, rate of change of temperature, etc. calculated. The features are organized in a feature matrix with time stamps, where each row represents data within a time window.
Step S2: detecting abnormal time window of the lithium battery feature matrix, so as to obtain time window data;
in this embodiment, a time window method is used to perform anomaly detection on the feature matrix. First, a fixed size time window is defined, for example, every 10 minutes. Then, abnormality detection is performed within each time window. Abnormal data points can be detected by counting the statistical indicators, such as mean, variance, etc., of the characteristic data within each time window. For example, if the current in a certain time window exceeds twice the standard deviation of the mean, it can be marked as an outlier
Step S3: performing embedded dimension reduction processing on the time window data to obtain low-dimension window data, and constructing a battery life prediction model according to the low-dimension window data;
in this embodiment, embedded dimension reduction processing is performed on the time window data, so as to reduce the dimension of the data and preserve key information. Dimensionality reduction techniques such as Principal Component Analysis (PCA) may be used. And then, constructing a battery life prediction model by using the dimension-reduced data. This may be a machine learning model, such as a Support Vector Machine (SVM), neural network, or regression model. The training data of the model is the reduced-dimension time window data, and the label is the known battery life data.
Step S4: performing sequence prediction on the low-dimensional window data by using a battery life prediction model so as to obtain a residual life sequence;
in this embodiment, a trained battery life prediction model is used to perform sequence prediction on the time window data after the dimension reduction, so as to obtain a remaining life sequence. This can be obtained by predicting the lifetime in the future time window step by step and then accumulating it.
Step S5: carrying out morphological analysis according to the residual life sequence to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics to obtain a life dynamic prediction model;
in this example, morphology analysis was performed based on the remaining lifetime sequence to identify lifetime degradation trends or mutations. For example, if the remaining life drops sharply in a short time, it may indicate that a problem occurs in the battery. Based on morphological characteristics, dynamic parameter optimization can be performed on the battery life prediction model. This means that the parameters of the model can be adjusted to accommodate changes in battery life.
Step S6: and predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using a life dynamic prediction model, so as to obtain the predicted residual life of the lithium ion battery.
In the embodiment, the life prediction model optimized by dynamic parameters is utilized to predict the residual life of the electrical data and the working condition data of the lithium battery. This will give an estimated remaining useful life, helping the user to better manage the battery, replace it in time or maintain it.
According to the invention, the electrical data and the working condition data of the battery are obtained through the preset sensor, and the data comprise information such as voltage, current, temperature and the like. The data can be integrated by constructing the lithium battery feature matrix, so that multiple parameters are comprehensively considered, and a basis is provided for subsequent analysis. This helps to capture the behavior and performance of the battery. Time window anomaly detection helps identify anomalies in the data, for example, a battery may exhibit unusual behavior in certain situations, such as an abnormal discharge or charging process. By identifying these anomalies, it is possible to more accurately capture signs of deterioration of battery life, discover problems in advance, and take action. The embedded dimension reduction process helps reduce the dimensionality of the data while retaining key information to avoid overfitting. The battery life prediction model may build a model of battery life degradation from the reduced-dimension data, which helps accurately predict future performance. The low-dimensional window data is sequence predicted using a battery life prediction model, which helps predict the life remaining of the battery. Through the sequence prediction, the performance of the battery in a future time period can be estimated, and the possible problems of the battery are warned in advance, so that the maintenance plan is more targeted. Morphological analysis helps capture features in the battery life sequence, such as the trend of decay rate, periodic behavior, etc. According to the morphological characteristics, the dynamic parameter optimization can be carried out on the battery life prediction model so as to adapt to battery performance changes in different stages, and the adaptability and the accuracy of the prediction model are improved. The remaining life prediction of the battery electrical data and the operating condition data using the life dynamic prediction model allows the system to predict the remaining life of the battery based on the current battery state and environmental conditions. This helps to optimize battery usage and maintenance strategies, reducing unnecessary replacement and repair costs.
Optionally, step S1 specifically includes:
step S11: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor;
in the embodiment, the electrical data and the working condition data of the lithium battery are collected in real time through the pre-deployed sensor. The electric data comprise parameters such as voltage, current, temperature and the like, and the working condition data comprise information such as charging state, discharging state, using environment of the battery and the like. These data will be collected continuously and used for subsequent analysis and processing.
Step S12: carrying out correlation analysis on lithium battery electric data and lithium battery working condition data so as to obtain a correlation matrix;
in this embodiment, the correlation between the electrical data and the working condition data can be identified by performing correlation analysis on the electrical data and the working condition data. For example, a correlation coefficient between voltage and temperature may be calculated to determine if there is a correlation. This analysis can help to understand which factors have a significant impact on battery performance. After obtaining the correlation matrix, it is possible to know which data items have a strong correlation between them, which will play an important role in subsequent feature extraction and anomaly detection.
Step S13: detecting abnormal battery operation of the lithium battery electrical data and lithium battery working condition data, so as to obtain abnormal battery operation data;
In this embodiment, the battery operation abnormality detection is performed using the electrical data and the operating condition data. This may be achieved by various methods, such as threshold detection, statistical methods, or machine learning models. For example, if the battery voltage continues to drop over a period of time while the temperature increases, the system may detect this abnormal behavior and record as abnormal operation data.
Step S14: extracting time domain features of lithium battery electric data and lithium battery working condition data, thereby obtaining lithium battery time domain features;
in this embodiment, for the electrical data, time domain feature extraction may be performed to obtain information about the behavior of the battery. This includes statistical features such as mean, variance, peak, etc., and temporal waveform features such as periodicity, waveform shape, etc. For example, the mean and standard deviation of the current signal may be calculated to understand the stability and variation of the battery current.
Step S15: extracting frequency domain features of lithium battery electrical data and lithium battery working condition data, thereby obtaining lithium battery frequency domain features;
in this embodiment, frequency domain feature extraction may also be performed on the electrical data to obtain information in the frequency domain. This may be achieved by fourier transform or wavelet transform, etc. The frequency domain features include frequency components, power spectral density, etc. For example, the frequency content of the voltage signal may be calculated to detect the presence or absence of high frequency noise or oscillations.
Step S16: and performing incidence matrix mapping on the time domain features of the lithium battery, the frequency domain features of the lithium battery and the abnormal working data through the correlation matrix, so as to obtain a lithium battery feature matrix.
In this embodiment, through the correlation matrix, the correlation matrix mapping may be performed on the time domain features, the frequency domain features and the abnormal working data, so as to construct a feature matrix of the lithium battery. This feature matrix will include various feature items, each row representing feature data over a period of time.
The invention obtains the battery electric data and the working condition data through the sensor as a key starting point. These data include information about current, voltage, temperature, etc., which represent the actual operating conditions of the battery. This helps to capture the battery's behavior and provides the basis data for subsequent analysis. Correlation analysis helps to understand the relationship between different parameters. The correlation matrix may show the degree of correlation between the various parameters. This helps determine which parameters have the greatest impact on battery performance and life, thereby guiding the subsequent feature extraction and modeling process. The battery malfunction detection helps to identify abnormal behavior during battery operation. These anomalies may be caused by internal battery problems, external environmental factors, or system failures. By identifying and recording these anomalies, it is possible to help predict degradation of battery life and take corrective action to repair or replace in advance. Time domain feature extraction facilitates capturing time-dependent features from battery electrical data. This may include statistical properties of the current and voltage, such as mean, variance, waveform shape, etc. Time domain feature extraction helps to identify battery performance variations and trends at different points in time. The frequency domain feature extraction helps to translate battery electrical data into information on the frequency domain. This may include spectral analysis for detecting frequency components that may be present in the battery. Frequency domain feature extraction may help identify oscillations or periodic behavior in the battery that may be related to life degradation. And performing associated matrix mapping on the time domain features, the frequency domain features and the abnormal working data through the associated matrix, thereby being beneficial to integrating various feature information and constructing a comprehensive lithium battery feature matrix. This feature matrix integrates data from different sources to more fully describe the state and performance of the battery. This provides a richer input to the subsequent life prediction model.
Optionally, step S12 specifically includes:
feature screening is carried out on the lithium battery electrical data and the lithium battery working condition data, so that electrical feature data and working condition feature data are obtained;
in this embodiment, feature screening is performed on electrical data and working condition data of the lithium battery. The purpose of feature screening is to select the most informative features to reduce the dimensionality and complexity of the data. Screening may be performed using various methods, such as variance threshold, correlation analysis, feature importance assessment, dimension reduction techniques (e.g., principal component analysis), and the like.
Performing covariance matrix calculation on the electrical characteristic data so as to obtain an electrical covariance matrix;
the electrical characteristic data is used in this embodiment to calculate an electrical covariance matrix. The covariance matrix is used to represent the covariance relationship between the individual electrical features. Each element (i, j) represents the covariance between feature i and feature j. The diagonal elements of the covariance matrix are the variances of the individual features. For example, if there are voltage, current and temperature as electrical characteristics, the electrical covariance matrix will show a linear relationship between them.
Performing covariance matrix calculation on the working condition characteristic data so as to obtain a working condition covariance matrix;
Similar to the electrical data in this embodiment, the operating condition covariance matrix is calculated using the operating condition characteristic data. This matrix represents the covariance relationship between the operating mode features, helping to understand the relevance between the various operating mode parameters. For example, the operating condition covariance matrix may display an association between the state of charge, the state of discharge, and the ambient temperature.
And combining matrix elements according to the electric covariance matrix and the working condition covariance matrix, so as to obtain a correlation matrix.
In this embodiment, elements of the electrical covariance matrix and the operating mode covariance matrix are combined, so as to obtain a correlation matrix. This matrix will contain information about the correlation between the electrical characteristics and the operating characteristics.
The electrical data and operating condition data in the present invention typically include a large number of features, some of which may be less relevant or redundant. Feature screening can help reduce the dimensionality of the data, improving computational efficiency and model performance. The screening of the most relevant features is helpful for constructing a more accurate model, reducing the influence of noise and improving the prediction accuracy. The characteristics screened are generally easier to interpret, which is important for understanding the influencing factors of the performance of lithium batteries. The electrical covariance matrix helps analyze correlations between electrical features. This may reveal which electrical features are in a linear relationship with each other, helping to understand the physical process inside the battery. The covariance matrix may be used to detect multiple collinearity between features, i.e., some features are highly correlated. This helps to avoid introducing redundant information during the modeling process. The operating mode covariance matrix provides relationship information between operating mode features. This is useful for understanding the behavior of lithium batteries under different operating conditions. By combining the electrical covariance matrix and the operating mode covariance matrix, a comprehensive correlation matrix can be obtained, wherein the correlation matrix comprises the relationship between electrical characteristics and operating mode characteristics. This helps to comprehensively consider the impact of the electrical performance and operating conditions of a lithium battery on its life and performance.
Optionally, step S13 specifically includes:
step S131: carrying out time sequence combination on the lithium battery electrical data and the lithium battery working condition data so as to obtain data to be detected abnormally;
in this embodiment, the electrical data and the working condition data of the lithium battery are combined according to a time sequence. The data of the electrical parameters (such as voltage, current, etc.) and the working conditions (such as temperature, charge-discharge state, etc.) are integrated according to the time stamp. Such time series integration may help to establish a time series that better understand the electrical characteristics of a lithium battery under different operating conditions. For example, using the time stamp as a key identifier, the electrical data and the operating condition data are combined in time sequence to form a complete time series data set containing the electrical characteristics and the operating conditions.
Step S132: performing abnormal work score calculation on the data to be detected abnormal through an abnormal work score calculation formula, so as to obtain abnormal work score data;
in this embodiment, an abnormal work score calculation formula is used to process the combined data to be detected abnormally, and an abnormal work score of each data point is calculated.
Step S133: carrying out statistical analysis according to the abnormal work score data so as to obtain an abnormal work threshold;
In this embodiment, statistical analysis is performed on the calculated abnormal work score data, and the distribution characteristics of the scores are explored. This may include calculation of statistics such as mean, standard deviation, quantiles, etc., and plotting charts such as histograms or probability density maps to aid in determining abnormal operating thresholds. For example, the mean and standard deviation of the abnormal operation scores are calculated, and these statistics are then used to determine an abnormal operation threshold, such as defining data points with scores exceeding the mean plus several standard deviations as abnormal.
Step S134: and carrying out classified calculation on the data to be detected abnormally through an abnormal working threshold value, so as to obtain the working abnormal data.
In this embodiment, the determined abnormal operation threshold is used to perform classification calculation on the data to be detected abnormally. The abnormal working score of the data point is compared with an abnormal working threshold, and if the score exceeds the threshold, the data point is marked as working abnormal data to represent that the working state of the lithium battery is abnormal at the moment. For example, the calculated abnormal work score is compared to a previously determined threshold, and data points that score above the threshold are marked as abnormal work. These abnormal data points can be further analyzed to understand the abnormal behavior of the lithium battery under certain conditions.
The electrical data and the operating mode data in the invention are usually from different sources, and can be integrated into one data set through time sequence combination. This helps to establish a correlation between the operating state and performance of the lithium battery. The abnormal operation score calculation can identify an abnormal situation in the battery operation. This may include battery temperature anomalies, voltage fluctuations, capacity drops, etc. By calculating the abnormal work score, the degree of abnormality can be quantified, facilitating further analysis and processing. These work scores may help detect anomalies early in the process, thereby taking precautions, extending the life of the lithium battery, and improving safety. Statistical analysis may help determine a threshold for abnormal job scores. These thresholds may be determined based on historical data and performance requirements of the model. The setting of the threshold determines when the battery is considered abnormal, triggering further processing. The statistical analysis can help to eliminate accidental anomalies, improve the accuracy of anomaly detection and reduce false alarms. The data to be detected can be classified into normal and abnormal operation data according to the abnormality threshold. This helps to identify and record anomalies in the battery in a timely manner. The anomaly data can be used to further analyze the root cause of the problem, helping to make fault analysis and maintenance decisions. This can save maintenance costs and improve the reliability of the battery system.
Alternatively, the abnormal work score calculation formula in step S132 is specifically:
in the method, in the process of the invention,for abnormal work score, < >>For lithium battery voltage, ">For observing time, < >>For the current value of the lithium battery, +.>For the capacity of lithium batteries, +.>For the battery temperature value, ">Charge rate for lithium battery->For lithium battery discharge rate,/->Is a comprehensive parameter of working condition>For the charge rate weight, +.>Is discharge rate weight +.>The average operating time of the lithium battery.
The invention constructs an abnormal work score calculation formula for calculating the abnormal work score of the data to be detected abnormally. The formula fully considers influencing abnormal work scoresLithium battery voltage>Observation time->Current value of lithium cell->Capacity of lithium battery->Battery temperature value->Battery charge rate->Lithium battery discharge rate->Condition comprehensive parameter->Charge rate weight->Discharge rate weight->Average operating time of lithium battery->A functional relationship is formed:
wherein the method comprises the steps ofThis part represents the square of the speed of the voltage change over time. The change speed of the voltage can reflect the dynamic change inside the battery, and the battery usually generates voltage fluctuation due to different factors (such as current change, temperature change and the like) in operation. The square of this section indicates that higher voltage change rates have a greater impact on the abnormal operation score. / >This part contains the current value of the battery +.>And Capacity value->Is a relationship of (3). It first calculates +.>And->Then square it and add 1, and finally take the logarithm. The purpose of this section is to take into account the correlation between current and battery capacity, as high currents can cause the battery to discharge rapidly, affecting the performance and life of the battery. />This part represents the square root of the temperature value of the battery. The temperature of the battery has an important influence on its performance and safety. In general, the higher the temperature, the more vulnerable the battery is, and therefore the square root of the temperature is used to represent the temperature-dependent effects in the abnormal operation score. />This part is the charge rate +.>And charge rateWeight->Is a product of (a) and (b). It takes into account the effect of the battery charge rate on the operating conditions. If the charge rate is high, the battery may be overcharged or heated, thereby causing an abnormal situation. />This part is the discharge rate +.>And discharge rate weight->Is a product of (a) and (b). It takes into account the effect of the battery discharge rate on the operating conditions. Higher discharge rates may lead to overdischarge or rapid wear of the battery and may also cause abnormal situations. />This part is the condition combination parameter +. >Is multiplied by the weight of the average operating time of the lithium battery +.>. The working condition comprehensive parameters comprise a plurality of factors, and a plurality of working condition factors such as current, temperature, voltage and the like are comprehensively considered. Squaring emphasizes the importance of the operating mode complex parameters to the abnormal operating score, whereas +.>A trade-off consideration for operating time is expressed. The formula takes into account a number of factors related to battery performance, including current, voltage, temperature, charge and discharge rates, and the like. The abnormal working score obtained through calculation can be used for monitoring the working state of the battery in real time. If the score exceeds a certain threshold, it may indicate that the battery is experiencing an abnormal condition, such as overdischarge, overcharge, temperature abnormality, etc. This helps to find potential problems early, takingPrecautions reduce the risk of battery damage. />
Optionally, step S2 specifically includes:
step S21: defining a time window and a sliding step length according to the data to be detected abnormally, so as to obtain the time window and the sliding step length;
in this embodiment, a time window and a sliding step are defined for subsequent anomaly detection. Assuming that the performance of the lithium battery is monitored, it is necessary to divide the data of one day into a plurality of time windows for abnormality detection. The time windows may be defined as one per hour, meaning that each time window includes one hour of data. The sliding step size may be defined every 30 minutes, so that every 30 minutes a new time window is created to ensure that an abnormal situation can be detected in time.
Step S22: matrix segmentation is carried out on the lithium battery characteristic matrix through a time window and a sliding step length, so that a time window data set is obtained;
in this embodiment, after the definition of the time window and the sliding step length is performed, the lithium battery feature matrix may be divided into matrix segments according to these parameters, so as to obtain a time window data set. For example, for data of one day, 24 time window data sets, each containing one hour of data, would be obtained. These data sets will become inputs for subsequent steps.
Step S23: constructing a periodic anomaly detection model according to the time window data set and the data to be detected anomalies;
in this embodiment, a periodic anomaly detection model is constructed using a time window dataset and data to be anomaly detected. Assuming a seasonal decomposition method is used, each time window dataset is decomposed seasonally to extract seasonal components. These components will be used to build a model, such as a seasonal time series model or a machine learning model, to detect periodic anomalies.
Step S24: performing periodic anomaly detection on the lithium battery feature matrix through a periodic anomaly detection model, so as to obtain charging periodic anomaly data;
In the embodiment, the constructed periodic anomaly detection model is utilized to detect periodic anomalies of the whole lithium battery feature matrix. It involves applying a model to each time window dataset and identifying anomalies in the charging cycle. For example, if an abnormal seasonal component is detected within a certain time window, it is marked as periodic abnormal data.
Step S25: and carrying out abnormal marking on the lithium battery characteristic matrix by using the charging period abnormal data so as to obtain time window data.
In this embodiment, the charging cycle anomaly data is used to mark the time window data in the lithium battery feature matrix to obtain an anomaly flag for the time window data. This will facilitate subsequent analysis and processing. For example, if an anomaly is detected within a certain time window, the corresponding time window is marked as anomalous so that the operator can identify the problem and take appropriate action to resolve.
According to the invention, the data can be divided into small time windows by detecting the abnormal time window of the lithium battery characteristic matrix, so that the working mode and the performance of the battery can be better understood. This helps to detect changes in battery behavior over different time periods, including fluctuations in battery charge, discharge, temperature, etc. Identifying abnormal behavior of the battery within different time windows, such as voltage fluctuations, current anomalies, etc., helps to more finely monitor the performance and health of the battery. Defining the time window and the sliding step size is helpful to flexibly adjust the division mode of the data so as to adapt to different analysis requirements. The definition of the time window may affect the sensitivity and efficiency of subsequent anomaly detection. Depending on the particular application scenario, different time windows and sliding steps may be selected, for example, shorter time windows and sliding steps may be required for real-time monitoring, while longer time windows may be required for historical data analysis. And cutting the lithium battery characteristic matrix through a time window and a sliding step length to obtain a series of time window data sets. This facilitates a segmented analysis of the data to better understand the battery's operating mode and performance changes over different time periods. These time window datasets can be used for subsequent anomaly detection and model construction to help discover the abnormal behavior of the battery over different time periods. According to the time window data set and the data to be detected abnormally, a periodic abnormal detection model is constructed, so that the periodic abnormal behavior of the battery can be recognized. This can help distinguish between normal duty cycles and potential problems. The periodic anomaly detection model can be used for predicting the abnormal condition of the battery in a future period and finding the abnormal behavior in the working period of the battery in real-time monitoring. And the periodic anomaly detection can be carried out on the lithium battery characteristic matrix through the periodic anomaly detection model. This helps identify abnormal behavior of the battery during a particular duty cycle, such as abnormal current or voltage fluctuations of the battery during each charge cycle. Periodic anomaly detection can help discover periodic problems early, thereby taking steps to reduce the risk of battery damage or failure. By using the charge cycle abnormal data to perform abnormal marking on the lithium battery characteristic matrix, abnormal conditions and time window data can be associated, so that the time and mode of abnormal behaviors can be better known. The anomaly flags help to visualize and aggregate anomaly data for subsequent decisions and actions.
Optionally, step S24 specifically includes:
extracting the time dimension of the time window data set, thereby obtaining the time dimension;
the time dimension is extracted from the time window dataset in this embodiment. Assuming one day of data, each time window represents one hour of observation. The time dimension may be defined as the number of hours from midnight, so the time dimension of the first time window is 0, the second is 1, and so on, up to 23. This will help to take into account the effect of time in subsequent analysis.
Extracting sensor space data from the data to be detected abnormally so as to obtain the sensor space data;
in this embodiment, sensor space data is extracted from data to be detected for abnormality. Such data may include readings of various sensors, such as temperature, current, voltage, etc. The data for each sensor is taken as a dimension to obtain sensor space data. For example, if there are 3 sensors, one 3-dimensional data point will be obtained, with each dimension corresponding to one sensor reading.
Carrying out space sequence arrangement according to the sensor space data so as to obtain a space dimension;
in this embodiment, the sensor spatial data is arranged to obtain a spatial dimension. It is assumed that there are a plurality of time windows of sensor data, which can be arranged in a time series, wherein each time point contains the data of all sensors. This will form a spatial dimension in which each point in time represents sensor data for a window in time.
Extracting lithium battery characteristics from the time window data set, thereby obtaining characteristic dimensions;
in this embodiment, the feature dimension of the lithium battery is obtained by extracting the feature of the lithium battery from the time window dataset. This may include extracting various features from the battery performance data, such as charge rate, rate of change of temperature, current ripple, etc. Each extracted feature will become part of the feature dimension.
Constructing a 3D tensor according to the space dimension, the time dimension and the characteristic dimension;
in this embodiment, the time dimension, the space dimension and the feature dimension are integrated to construct a 3D tensor. The dimension of this tensor will be the number of time windows x the spatial dimension x the feature dimension. For example, if there are 30 time windows, 5 sensors, 10 features, a 3D tensor of 30x5x10 would be obtained, with each element representing sensor data and features within one time window.
And constructing a periodic anomaly detection model through the time window dataset and the 3D tensor.
The periodic anomaly detection model is built using a time window dataset and a built 3D tensor in this embodiment. This model may be a machine learning based model, such as a Recurrent Neural Network (RNN) or Convolutional Neural Network (CNN), or a statistical model, such as a seasonal decomposition approach. This model may be used to detect periodic anomalies, such as abnormal changes in battery performance over a particular time window.
The time dimension is extracted to organize the data according to the time sequence, so that the analysis of the change trend of the data along with time is facilitated. This may reveal a time dependence of battery performance and anomalies. The time dimension extraction helps to build time series data, allowing analysis of the historical evolution of battery behavior. The extraction of sensor spatial data can be focused on the sensor measurements, an important component in understanding battery status. By extracting the sensor data, the physical characteristics of the battery system can be captured. These data are used to analyze sensor changes, such as voltage, temperature, current, etc., to detect potential anomalies or problems. By arranging the sensor spatial data, a data structure with spatial dimensions can be constructed, which helps to better understand the interrelationship between the different sensors in the battery system. This helps to capture spatially anomalous patterns in the battery system, such as interactions between sensors, or anomalies in specific locations. Extracting lithium battery characteristics can capture key characteristics of battery performance, such as capacity, internal resistance, charge/discharge efficiency, and the like. This helps focus on key parameters in anomaly detection. The extracted features can be used to build an anomaly detection model to identify abnormal behavior in terms of battery performance. Combining the temporal dimension, the spatial dimension, and the feature dimension into a 3D tensor structure helps to comprehensively consider multiple data dimensions. This allows simultaneous analysis of information of different dimensions. The construction of 3D tensors allows anomaly detection using multidimensional data, for example taking into account anomaly behaviour in time, space and features simultaneously. By using the time window dataset and the 3D tensor, constructing a periodic anomaly detection model can help detect periodic anomalies in the battery system. This helps to find abnormal behavior of the battery under certain time and space conditions. The model can be used for monitoring and predicting the periodic problems in the battery system in real time so as to take maintenance and preventive measures early and improve the reliability and performance of the battery.
Optionally, step S4 specifically includes:
carrying out charge cycle anomaly flag extraction on the low-dimensional window data so as to obtain anomaly flag data points;
in this embodiment, the charge cycle abnormality flag is extracted from the low-dimensional window data. This can be achieved by analyzing the abnormal behavior during the charging of the battery. For example, a sudden change in the voltage, current, or temperature of the battery during charging or an out of normal range condition may be detected. These abnormal conditions will be marked as abnormal charge events and used for subsequent analysis.
Calculating abnormal mark data points and lithium battery working condition data through an abnormal charge influence calculation formula, so as to obtain abnormal charge film loudness;
the abnormal full-film loudness of each abnormal marker data point is calculated using an abnormal full-film loudness calculation formula in this embodiment.
The abnormal film filling loudness calculation formula specifically comprises:
in the method, in the process of the invention,to abnormally fill the film loudness->For charging time, +.>For lithium battery voltage, ">For observing time, < >>Lithium battery capacity for outlier marker data points, +.>For the average operating time of lithium batteries, < >>For internal resistance of battery->Characteristic index for outlier marker data points, +. >For the feature total number of outlier marker data points, +.>The rate of change over time of the characteristic value for the outlier marker data point;
the invention constructs an abnormal charging film loudness calculation formula for calculating abnormal mark data points and lithium battery working condition data. The formula fully considers influencing the abnormal charge influence degreeCharging time of->Lithium battery voltage>Observation time->Lithium battery capacity of abnormal marker data points>Average operating time of lithium battery->Internal resistance of battery->Characteristic index of outlier marker data points +.>Total number of features of outlier marker data points +.>Characteristic value change rate of abnormal marker data point with time +.>A functional relationship is formed: />
Wherein the method comprises the steps ofIs the battery voltage +.>Relative to time->And represents the slope of the voltage. The voltage may change over time as the battery is charged or discharged. This part represents the rate of change of the voltage. />This is the average operating time of the lithium battery +.>Battery capacity with outlier marker data points +.>Natural logarithm of the ratio. It represents the capacity usage of the battery over an average operating time. If the battery capacity->Far below average working time +.>It is stated that the battery may run out in a short time, thereby affecting battery life. / >This part represents the charging time +.>And (2) internal resistance of battery>The square root of the ratio. Internal resistance of battery->Has important influence on the performance and charging process of the battery. If the charging time is->Relative to the internal resistance of the battery>Larger, may cause heat accumulation inside the battery, thereby affecting the life of the battery. />This means +.>Summing is performed. />The total number of features, which are outlier marker data points, represents a comprehensive consideration of all features. />This is a feature +.>Representing the relative importance of different features to the loudness of an abnormally charged film. Different features may affect the life of the battery in different ways, thus taking into account their contribution by giving different weights. />This is a feature +.>Over time->Is a rate of change of (c). It is a kind ofRepresentation feature->The change in time was observed. Different features may change in different ways during charging, this part taking into account the effects of these changes. Finally, the results of the above-mentioned parts are integrated (from 0 to +.>) Obtaining the abnormal filling film loudness +.>. The influence can comprehensively consider the influence of voltage change, battery capacity use, the relationship between charging time and internal resistance and the change of a plurality of characteristics on the service life of the battery. By this calculation formula, the influence of abnormal charge can be quantified, which is helpful for further analysis and management of the battery system to improve the reliability and lifetime of the battery.
Predicting the future battery life of the low-dimensional window data through a battery life prediction model, so as to obtain a future life sequence;
in this embodiment, a battery life prediction model is used to predict future battery life for low-dimensional window data. The model will take into account the battery's usage, charging history, and other factors to estimate the battery's life.
And performing sequence correction on the future life sequence based on the abnormal film filling loudness, so as to obtain a residual life sequence.
In this embodiment the future life sequence is corrected based on the abnormal film filling loudness. If the abnormal charge film at a certain time point has high loudness, the battery is greatly influenced, so that the future life prediction at the time point can be adjusted, the life prediction value is reduced, and the reduction of the life of the battery is reflected. In this way, a corrected remaining life sequence is obtained that more accurately reflects the remaining life of the battery.
By identifying and marking the charge cycle abnormal data points, the invention can determine abnormal conditions, such as too fast charge or too slow charge, occurring in the battery charging process. The outlier marker data points can be used in the outlier charge movie loudness calculation in subsequent steps to help understand the impact of outlier charge on battery life. By calculating the abnormal charge film loudness, the potential extent of the influence of abnormal charge on battery life can be quantified. This helps identify which abnormal charge conditions most severely affect the life of the battery. Predicting the low-dimensional window data using a battery life prediction model may estimate the life of the future battery. This may help to plan maintenance and better manage battery usage. By correcting the future life sequence based on the abnormal charge film loudness, the remaining life of the battery can be estimated more accurately. This helps to avoid unnecessary maintenance or replacement while improving the reliability of the battery. The corrected remaining life sequence may be used to make a more accurate maintenance schedule to ensure that the battery does not fail until the end of its life.
Optionally, step S5 specifically includes:
smoothing the residual life sequence to obtain a smooth life sequence;
the remaining life sequence is smoothed in this embodiment to reduce noise and fluctuations. This may be accomplished using various filtering techniques, such as moving average filtering, exponential smoothing, or Kalman filtering. For example, moving average filtering may be used to calculate an average over a window to smooth the lifetime sequence, thereby obtaining a smoothed lifetime sequence.
Trend analysis is carried out according to the smooth life sequence, so that life trend is obtained;
in this embodiment, a trend analysis is performed on the smooth lifetime sequence to capture the trend of lifetime over time. This can be done using linear regression, polynomial fitting, or time series analysis methods. Through this step, parameters of life trend, such as slope and intercept, can be obtained to reflect whether life is increasing or decreasing.
Performing Fourier transform on the smooth life sequence, thereby obtaining a life spectrum;
the smoothed lifetime sequence is fourier transformed in this embodiment to convert the time domain signal to a frequency domain signal. This may be accomplished using Fast Fourier Transform (FFT) techniques or the like. Fourier transforms will help analyze the periodic components in the lifetime sequence and obtain the lifetime spectrum, which contains amplitude and phase information for the different frequency components.
Performing Fourier transformation according to the charging period abnormal data, so as to obtain a charging period abnormal frequency spectrum;
in this embodiment, fourier transformation is performed on the charging cycle anomaly data to obtain a charging cycle anomaly spectrum. This will help to analyze the frequency domain characteristics of the charge cycle anomalies.
Periodically analyzing according to the life spectrum to obtain life periodicity, and parameter marking the life periodicity according to the abnormal charging period spectrum to obtain marked life periodicity;
in this embodiment, the life spectrum is periodically analyzed to determine the life periodicity. This can be achieved by identifying the dominant frequency components and amplitudes in the lifetime spectrum. Meanwhile, the life cycle can be marked according to the characteristic of the abnormal frequency spectrum of the charging cycle. For example, if a frequency component associated with a charge cycle anomaly is found in the spectrum, it may be taken as one of the parameters that marks the life cycle.
Carrying out peak extraction on the smooth life sequence so as to obtain a life peak point;
in this embodiment, peak extraction is performed on the smoothed lifetime sequence. This means that significant peak points in the life sequence are identified, which points may represent critical events or incidents of life. The peaks may be extracted using signal processing techniques such as peak detection algorithms.
Combining the life peak point, the marked life periodicity and the life trend data, thereby obtaining morphological characteristics;
in this embodiment, the information such as the lifetime peak point, the lifetime periodicity of the mark, and the lifetime trend is combined to create the morphological feature. These features will contain key information of the life sequence such as fluctuations, periodicity, trends and important events of life.
And carrying out dynamic parameter optimization on the battery life prediction model according to morphological characteristics, thereby obtaining the life dynamic prediction model.
In this embodiment, according to morphological characteristics, dynamic parameter optimization can be performed on the battery life prediction model. This may be accomplished by machine learning techniques or optimization algorithms to ensure that the model is able to more accurately predict battery life. According to the change of morphological characteristics, the model parameters can be adaptively adjusted to better adapt to the change condition of the service life of the battery.
The smoothing processing in the invention can remove noise and fluctuation in the data, so that the service life data has better interpretability and the randomness is reduced. The smoothed data can be used to more accurately estimate battery life and reduce errors due to noise. Trend analysis can help identify the trend of battery life over time, whether it is gradually decaying or continuously stabilizing. Trend analysis provides important insight into the state of health of the battery, helping to determine if the battery is at the end of life. The fourier transform may transform the lifetime data from the time domain to the frequency domain, revealing potentially periodic or frequency components. Life spectrum analysis can be used to detect periodic patterns in battery life, which is very useful for finding periodic factors in life. Fourier transforms are used to analyze the charge cycle anomaly data to help find charge cycle problems that may lead to life anomalies. Abnormal spectral analysis may be used to identify abnormal behavior during a charging cycle, which may affect battery life. Periodic analysis may help determine periodic patterns over battery life and flag these patterns through the charge cycle anomaly spectrum for further analysis. This helps identify the periodicity factor that leads to battery life anomalies and may guide maintenance strategies. Extracting life peak points helps identify key events or incidents in battery life. The peak point of life may be used to mark a significant change in battery performance, which may be related to the critical point of life. Combining peak life points, signature life periodicity, and life trends into morphological features can provide comprehensive information about battery life status. These features can be used to optimize a life prediction model to more accurately estimate the life of the battery. By optimizing dynamic parameters of the battery life prediction model based on morphological characteristics, a more accurate life prediction model can be established, and the model can monitor the life of the battery in real time and can be adjusted according to actual conditions. This model can be used to predict the remaining life of the battery, helping to take appropriate maintenance and replacement strategies, improving the reliability and performance of the battery.
Optionally, the present specification further provides a lithium ion battery remaining life prediction system, which includes:
the characteristic matrix construction module is used for acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor and constructing a lithium battery characteristic matrix according to the lithium battery electrical data and the lithium battery working condition data;
the time window detection module is used for detecting the abnormality of the time window of the lithium battery characteristic matrix so as to obtain time window data;
the model construction module is used for carrying out embedded dimension reduction processing on the time window data so as to obtain low-dimensional window data, and constructing a battery life prediction model according to the low-dimensional window data;
the sequence prediction module is used for performing sequence prediction on the low-dimensional window data by using the battery life prediction model so as to obtain a residual life sequence;
the morphological analysis module is used for carrying out morphological analysis according to the residual life sequence so as to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics so as to obtain a life dynamic prediction model;
and the residual life prediction module is used for predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using the life dynamic prediction model so as to obtain the predicted residual life of the lithium ion battery.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
The foregoing is only a specific embodiment of the invention to enable those skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (9)

1. The method for predicting the residual life of the lithium ion battery is characterized by comprising the following steps of:
step S1: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor, and constructing a lithium battery feature matrix according to the lithium battery electrical data and the lithium battery working condition data;
Step S2: detecting abnormal time window of the lithium battery feature matrix, so as to obtain time window data;
step S3: performing embedded dimension reduction processing on the time window data to obtain low-dimension window data, and constructing a battery life prediction model according to the low-dimension window data;
step S4: performing sequence prediction on the low-dimensional window data by using a battery life prediction model so as to obtain a residual life sequence;
step S5, including:
smoothing the residual life sequence to obtain a smooth life sequence;
trend analysis is carried out according to the smooth life sequence, so that life trend is obtained;
performing Fourier transform on the smooth life sequence, thereby obtaining a life spectrum;
performing Fourier transformation according to the charging period abnormal data, so as to obtain a charging period abnormal frequency spectrum;
periodically analyzing according to the life spectrum to obtain life periodicity, and parameter marking the life periodicity according to the abnormal charging period spectrum to obtain marked life periodicity;
carrying out peak extraction on the smooth life sequence so as to obtain a life peak point;
combining the life peak point, the marked life periodicity and the life trend data, thereby obtaining morphological characteristics;
Carrying out dynamic parameter optimization on the battery life prediction model according to morphological characteristics so as to obtain a life dynamic prediction model;
step S6: and predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using a life dynamic prediction model, so as to obtain the predicted residual life of the lithium ion battery.
2. The method according to claim 1, wherein step S1 is specifically:
step S11: acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor;
step S12: carrying out correlation analysis on lithium battery electric data and lithium battery working condition data so as to obtain a correlation matrix;
step S13: detecting abnormal battery operation of the lithium battery electrical data and lithium battery working condition data, so as to obtain abnormal battery operation data;
step S14: extracting time domain features of lithium battery electric data and lithium battery working condition data, thereby obtaining lithium battery time domain features;
step S15: extracting frequency domain features of lithium battery electrical data and lithium battery working condition data, thereby obtaining lithium battery frequency domain features;
step S16: and performing incidence matrix mapping on the time domain features of the lithium battery, the frequency domain features of the lithium battery and the abnormal working data through the correlation matrix, so as to obtain a lithium battery feature matrix.
3. The method according to claim 2, wherein step S12 is specifically:
feature screening is carried out on the lithium battery electrical data and the lithium battery working condition data, so that electrical feature data and working condition feature data are obtained;
performing covariance matrix calculation on the electrical characteristic data so as to obtain an electrical covariance matrix;
performing covariance matrix calculation on the working condition characteristic data so as to obtain a working condition covariance matrix;
and combining matrix elements according to the electric covariance matrix and the working condition covariance matrix, so as to obtain a correlation matrix.
4. A method according to claim 3, wherein step S13 is specifically:
step S131: carrying out time sequence combination on the lithium battery electrical data and the lithium battery working condition data so as to obtain data to be detected abnormally;
step S132: performing abnormal work score calculation on the data to be detected abnormal through an abnormal work score calculation formula, so as to obtain abnormal work score data;
step S133: carrying out statistical analysis according to the abnormal work score data so as to obtain an abnormal work threshold;
step S134: and carrying out classified calculation on the data to be detected abnormally through an abnormal working threshold value, so as to obtain the working abnormal data.
5. The method according to claim 4, wherein the abnormal work score calculation formula in step S132 is specifically:
in the method, in the process of the invention,for abnormal work score, < >>For lithium battery voltage, ">For observing time, < >>For the current value of the lithium battery, +.>For the capacity of lithium batteries, +.>For the battery temperature value, ">Charge rate for lithium battery->For lithium battery discharge rate,/->Is a comprehensive parameter of working condition>For the charge rate weight, +.>Is discharge rate weight +.>The average operating time of the lithium battery.
6. The method according to claim 5, wherein step S2 is specifically:
step S21: defining a time window and a sliding step length according to the data to be detected abnormally, so as to obtain the time window and the sliding step length;
step S22: matrix segmentation is carried out on the lithium battery characteristic matrix through a time window and a sliding step length, so that a time window data set is obtained;
step S23: constructing a periodic anomaly detection model according to the time window data set and the data to be detected anomalies;
step S24: performing periodic anomaly detection on the lithium battery feature matrix through a periodic anomaly detection model, so as to obtain charging periodic anomaly data;
step S25: and carrying out abnormal marking on the lithium battery characteristic matrix by using the charging period abnormal data so as to obtain time window data.
7. The method according to claim 6, wherein step S24 is specifically:
extracting the time dimension of the time window data set, thereby obtaining the time dimension;
extracting sensor space data from the data to be detected abnormally so as to obtain the sensor space data;
carrying out space sequence arrangement according to the sensor space data so as to obtain a space dimension;
extracting lithium battery characteristics from the time window data set, thereby obtaining characteristic dimensions;
constructing a 3D tensor according to the space dimension, the time dimension and the characteristic dimension;
and constructing a periodic anomaly detection model through the time window dataset and the 3D tensor.
8. The method according to claim 7, wherein step S4 is specifically:
carrying out charge cycle anomaly flag extraction on the low-dimensional window data so as to obtain anomaly flag data points;
calculating abnormal mark data points and lithium battery working condition data through an abnormal charge influence calculation formula, so as to obtain abnormal charge film loudness;
the abnormal film filling loudness calculation formula specifically comprises:
in the method, in the process of the invention,to abnormally fill the film loudness->For charging time, +.>For lithium battery voltage, ">For observing time, < > >Lithium battery capacity for outlier marker data points, +.>For the average operating time of lithium batteries, < >>For internal resistance of battery->Characteristic index for outlier marker data points, +.>For the feature total number of outlier marker data points, +.>As the rate of change of the characteristic value of the outlier marker data point over time,characteristic weight factors for abnormal marker data points;
predicting the future battery life of the low-dimensional window data through a battery life prediction model, so as to obtain a future life sequence;
and performing sequence correction on the future life sequence based on the abnormal film filling loudness, so as to obtain a residual life sequence.
9. A lithium ion battery remaining life prediction system for performing the lithium ion battery remaining life prediction method of claim 1, the lithium ion battery remaining life prediction system comprising:
the characteristic matrix construction module is used for acquiring lithium battery electrical data and lithium battery working condition data through a preset sensor and constructing a lithium battery characteristic matrix according to the lithium battery electrical data and the lithium battery working condition data;
the time window detection module is used for detecting the abnormality of the time window of the lithium battery characteristic matrix so as to obtain time window data;
The model construction module is used for carrying out embedded dimension reduction processing on the time window data so as to obtain low-dimensional window data, and constructing a battery life prediction model according to the low-dimensional window data;
the sequence prediction module is used for performing sequence prediction on the low-dimensional window data by using the battery life prediction model so as to obtain a residual life sequence;
the morphological analysis module is used for carrying out morphological analysis according to the residual life sequence so as to obtain morphological characteristics, and carrying out dynamic parameter optimization on the battery life prediction model according to the morphological characteristics so as to obtain a life dynamic prediction model;
and the residual life prediction module is used for predicting the residual life of the lithium battery electrical data and the lithium battery working condition data by using the life dynamic prediction model so as to obtain the predicted residual life of the lithium ion battery.
CN202311337131.4A 2023-10-17 2023-10-17 Lithium ion battery remaining life prediction method and system Active CN117074965B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311337131.4A CN117074965B (en) 2023-10-17 2023-10-17 Lithium ion battery remaining life prediction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311337131.4A CN117074965B (en) 2023-10-17 2023-10-17 Lithium ion battery remaining life prediction method and system

Publications (2)

Publication Number Publication Date
CN117074965A CN117074965A (en) 2023-11-17
CN117074965B true CN117074965B (en) 2023-12-12

Family

ID=88704663

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311337131.4A Active CN117074965B (en) 2023-10-17 2023-10-17 Lithium ion battery remaining life prediction method and system

Country Status (1)

Country Link
CN (1) CN117074965B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117272844B (en) * 2023-11-22 2024-04-09 武汉宏联电线电缆有限公司 Method and system for predicting service life of distribution board
CN117710759B (en) * 2024-02-06 2024-04-26 青岛哈尔滨工程大学创新发展中心 System and method for predicting faults and service lives of manned submersible vehicles

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102074757A (en) * 2010-12-24 2011-05-25 惠州市亿能电子有限公司 Method for estimating charge states of lithium ion battery
CN112765772A (en) * 2020-12-25 2021-05-07 武汉理工大学 Power battery residual life prediction method based on data driving
CN112986831A (en) * 2021-04-30 2021-06-18 上海海事大学 Lithium ion battery life prediction method based on correlation coefficient particle filtering
CN113640674A (en) * 2021-06-30 2021-11-12 昆明理工大学 Lithium ion battery available capacity estimation method based on optimized Gaussian process regression
CN114814631A (en) * 2022-04-25 2022-07-29 浙江大学 Cloud computing and feature selection based lithium battery online life prediction method
CN116609668A (en) * 2023-04-26 2023-08-18 淮阴工学院 Lithium ion battery health state and residual life prediction method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10401433B2 (en) * 2015-01-21 2019-09-03 Samsung Electronics Co., Ltd. Method and apparatus for estimating battery life

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102074757A (en) * 2010-12-24 2011-05-25 惠州市亿能电子有限公司 Method for estimating charge states of lithium ion battery
CN112765772A (en) * 2020-12-25 2021-05-07 武汉理工大学 Power battery residual life prediction method based on data driving
CN112986831A (en) * 2021-04-30 2021-06-18 上海海事大学 Lithium ion battery life prediction method based on correlation coefficient particle filtering
CN113640674A (en) * 2021-06-30 2021-11-12 昆明理工大学 Lithium ion battery available capacity estimation method based on optimized Gaussian process regression
CN114814631A (en) * 2022-04-25 2022-07-29 浙江大学 Cloud computing and feature selection based lithium battery online life prediction method
CN116609668A (en) * 2023-04-26 2023-08-18 淮阴工学院 Lithium ion battery health state and residual life prediction method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
纯电动公交车动力电池寿命测试行驶工况的研究;时玮等;汽车工程;第35卷(第02期);第138-151页 *

Also Published As

Publication number Publication date
CN117074965A (en) 2023-11-17

Similar Documents

Publication Publication Date Title
CN117074965B (en) Lithium ion battery remaining life prediction method and system
CN112731159B (en) Method for pre-judging and positioning battery faults of battery compartment of energy storage power station
EP2869153A1 (en) Health management system, fault diagnosis system, health management method, and fault diagnosis method
KR101948604B1 (en) Method and device for equipment health monitoring based on sensor clustering
CN114167838B (en) Multi-scale health assessment and fault prediction method for servo system
US20150066431A1 (en) Use of partial component failure data for integrated failure mode separation and failure prediction
CN116683588B (en) Lithium ion battery charge and discharge control method and system
CN116976707B (en) User electricity consumption data anomaly analysis method and system based on electricity consumption data acquisition
CN117289168B (en) Product quality monitoring system based on sensor data and intelligent algorithm
CN116660759B (en) Battery life prediction method and device based on BMS battery management system
CN116643178A (en) SOC estimation method and related device of battery management system
CN116415126A (en) Method, device and computing equipment for anomaly detection of doctor blades of paper machine
CN116714469A (en) Charging pile health monitoring method, device, terminal and storage medium
CN117630672A (en) Method and device for predicting a device battery of a diagnostic device using a converter model
WO2024090453A1 (en) Information processing method, information processing system, and program
US20240210479A1 (en) Battery performance monitoring and optimization using unsupervised clustering
CN117452236B (en) Method and system for detecting service life of battery of new energy automobile
CN117783745B (en) Data online monitoring method and system for battery replacement cabinet
WO2024057996A1 (en) Electricity storage element degradation state calculating device, degradation state calculating method, degradation state calculating program, degradation state estimating device, degradation state estimating method, abnormality detecting device, and abnormality detecting method
CN117092541B (en) Analysis method for calculating battery health by direct-current charging big data
JP2024063628A (en) Information processing method, information processing system, and program
CN117930024A (en) Intelligent battery health state monitoring method based on informer model
JP2024063629A (en) Information processing method, information processing system, and program
CN117471346A (en) Method and system for determining remaining life and health status of retired battery module
JP2024041524A (en) Abnormality detection device, abnormality detection method, and program

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant