CN116643181A - Storage battery state monitoring system - Google Patents

Storage battery state monitoring system Download PDF

Info

Publication number
CN116643181A
CN116643181A CN202211026238.2A CN202211026238A CN116643181A CN 116643181 A CN116643181 A CN 116643181A CN 202211026238 A CN202211026238 A CN 202211026238A CN 116643181 A CN116643181 A CN 116643181A
Authority
CN
China
Prior art keywords
sequence
voltage
prediction
storage battery
sample
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202211026238.2A
Other languages
Chinese (zh)
Other versions
CN116643181B (en
Inventor
罗家仁
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.)
Zhejiang Changxing Zhenge Technology Co ltd
Original Assignee
Zhejiang Changxing Zhenge 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 Zhejiang Changxing Zhenge Technology Co ltd filed Critical Zhejiang Changxing Zhenge Technology Co ltd
Priority to CN202211026238.2A priority Critical patent/CN116643181B/en
Publication of CN116643181A publication Critical patent/CN116643181A/en
Application granted granted Critical
Publication of CN116643181B publication Critical patent/CN116643181B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
    • G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/389—Measuring internal impedance, internal conductance or related variables
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/396—Acquisition or processing of data for testing or for monitoring individual cells or groups of cells within a battery
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/10—Energy storage using batteries

Abstract

The invention relates to the technical field of measuring electric variables, in particular to a storage battery state monitoring system, which comprises a data acquisition module and a data processing module, wherein the data acquisition module is used for acquiring data and sending the acquired data to the data processing module, and the data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps: outputting a plurality of predicted sequences related to whether the storage battery to be detected has faults or not according to the received data and the Gaussian process regression model; and determining a predicted voltage sequence and a voltage difference sequence, further carrying out combined operation on the output and the determined sequence, inputting the output and the determined sequence into a storage battery state monitoring network, and outputting the storage battery state to realize detection of the storage battery state to be detected. The invention can realize the test and the monitoring of the state of the storage battery by utilizing the trained Gaussian process regression model and the storage battery state monitoring network, and effectively improves the accuracy of the state monitoring of the storage battery.

Description

Storage battery state monitoring system
Technical Field
The invention relates to the technical field of measuring electric variables, in particular to a storage battery state monitoring system.
Background
Monitoring of the battery is critical to the safety of the battery. Currently, in monitoring the state of a battery, the following methods are generally adopted: the voltage in the process of charging the storage battery is input into a battery state network, and the state of the storage battery is output through the battery state network. The battery state network is often a network that is trained using voltages during charging of a plurality of storage batteries that are known to be faulty.
However, when the above manner is adopted, there are often the following technical problems:
the cause of the change in the battery voltage is often not just a battery failure, but there are often some non-failure factors (e.g., the accuracy of a sensor that measures the battery voltage is not high), so if the training battery state network only considers the voltage during the charging of the battery, and does not consider the non-failure factors that may cause the voltage change, the state of the battery detected through the trained battery state network is often caused to be inaccurate, resulting in a low accuracy of the state monitoring of the battery.
Disclosure of Invention
The summary of the invention is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. The summary of the invention is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
Some embodiments of the present invention provide a battery condition monitoring system that solves one or more of the above-mentioned problems of the background art.
Some embodiments of the present invention provide a system for monitoring a state of a storage battery, the system including a data acquisition module and a data processing module, the data acquisition module being configured to acquire data and send the acquired data to the data processing module, the data including: the data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps:
inputting the voltage sequence and the current sequence into a trained Gaussian process regression model, and outputting a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault predicted value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period through the Gaussian process regression model;
determining a predicted voltage sequence of the storage battery to be detected in the predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence;
Determining a voltage difference sequence of the storage battery to be detected in the prediction time period according to the detection voltage sequence and the prediction voltage sequence;
performing joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence;
and inputting the combined sequence into a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected.
Further, before the step of inputting the voltage sequence and the current sequence into the trained gaussian process regression model, the step implemented by the data processing module further includes:
acquiring a standard voltage sequence and a standard current sequence in a standard time period;
when the duration corresponding to the first preset time period is not equal to the duration corresponding to the standard time period, respectively carrying out data alignment on the standard voltage sequence and the voltage sequence, and obtaining an aligned voltage sequence corresponding to the voltage sequence and an aligned current sequence corresponding to the current sequence;
And determining the alignment voltage sequence and the alignment current sequence as the voltage sequence and the current sequence, respectively.
Further, the determining the voltage difference sequence of the to-be-detected storage battery in the prediction time period according to the detected voltage sequence and the prediction voltage sequence includes:
when the duration corresponding to the second preset time period is not equal to the duration corresponding to the predicted time period, acquiring a reference voltage sequence in the predicted time period;
data alignment is carried out on the detection voltage sequence and the reference voltage sequence, and a trimming voltage sequence corresponding to the detection voltage sequence is obtained;
and determining a difference value of each trimming voltage in the trimming voltage sequence and a predicted voltage in the predicted voltage sequence corresponding to the trimming voltage as a voltage difference value to obtain the voltage difference value sequence.
Further, the training process of the gaussian process regression model includes:
constructing a Gaussian process regression model;
obtaining a sample set of a sample storage battery, wherein samples in the sample set comprise: the sample battery measures voltage sequence and measuring current sequence in the third default time quantum, and the label that the sample in above-mentioned sample set corresponds includes: normal voltage sequence, measurement linear error sequence and voltage fault value sequence of the sample storage battery in a fourth preset time period;
And training the Gaussian process regression model by using the sample set and the labels corresponding to the samples in the sample set to obtain the trained Gaussian process regression model.
Further, the training process of the storage battery state monitoring network includes:
constructing a storage battery state monitoring network;
inputting the sample set into a trained Gaussian process regression model, and outputting a prediction information set corresponding to the sample set through the trained Gaussian process regression model, wherein the prediction information in the prediction information set comprises: a predicted normal voltage sequence, a predicted measured linear error sequence, a predicted voltage fault value sequence, a predicted voltage uncertainty sequence, a predicted linear error uncertainty sequence and a predicted fault uncertainty sequence within the predicted time period;
for each piece of prediction information in the prediction information set, determining a voltage prediction value sequence corresponding to the prediction information according to a prediction normal voltage sequence, a prediction measurement linear error sequence and a prediction voltage fault value sequence which are included in the prediction information;
for each piece of prediction information in the prediction information set, determining a sample voltage difference value sequence corresponding to the prediction information according to a voltage prediction value sequence corresponding to the prediction information and a prediction normal voltage sequence included in the prediction information;
For each piece of prediction information in the prediction information set, performing joint operation on a sample voltage difference value sequence corresponding to the piece of prediction information, a prediction normal voltage sequence included in the piece of prediction information, a prediction measurement linear error sequence, a prediction voltage fault value sequence, a prediction voltage uncertainty sequence, a prediction linear error uncertainty sequence and a prediction fault uncertainty sequence to obtain a prediction joint sequence corresponding to the piece of prediction information;
and training the storage battery state monitoring network by utilizing the prediction joint sequences corresponding to the prediction information in the prediction information set to obtain the trained storage battery state monitoring network.
Further, the sample set includes: the system comprises a first sample set, a second sample set and a third sample set, wherein the first sample set is a plurality of samples of a sample storage battery collected through a sensor when the sample storage battery is charged under various constant currents and constant pressures, the second sample set is a plurality of samples of the sample storage battery which is collected through a target sensor and is charged by a target charger when the sample storage battery is charged, and the third sample set is a plurality of samples collected when the sample storage battery is simulated to be charged and fault.
Further, the normal voltage in the normal voltage sequence included in the tag corresponding to the first sample in the first sample set is a voltage of the sample storage battery under the influence of the target factor, the measured linear error in the measured linear error sequence included in the tag corresponding to each first sample in the first sample set is an absolute value of a difference between the measured voltage in the measured voltage sequence included in the first sample and the normal voltage in the normal voltage sequence included in the first sample, and the voltage fault value in the voltage fault value sequence included in the tag corresponding to the first sample in the first sample set is zero.
Further, the normal voltage in the normal voltage sequence included in the label corresponding to the second sample in the second sample set is the voltage of the sample storage battery under the influence of the target factor, and the measured linear error in the measured linear error sequence and the voltage fault value in the voltage fault value sequence included in the label corresponding to the second sample in the second sample set are both zero.
Further, the normal voltage in the normal voltage sequence included in the label corresponding to the third sample in the third sample set is the voltage of the sample storage battery under the influence of the target factor, the measurement linear error in the measurement linear error sequence included in the label corresponding to the third sample in the third sample set is zero, and the voltage fault value in the voltage fault value sequence included in the label corresponding to the third sample in the third sample set is a preset fault value.
Further, the formula corresponding to the predicted voltage sequence of the storage battery to be detected in the predicted time period is determined as follows:
u t ＝U t +τ t +μ t
wherein u is t Is the t-th predicted voltage in the predicted voltage sequence, U t Is the t-th normal predicted voltage in the normal predicted voltage sequence, τ t Is the t-th measurement linear prediction error, mu, in the measurement linear prediction error sequence t Is the t-th voltage failure prediction value in the above-described voltage failure prediction value sequence.
The above embodiments of the present invention have the following advantages:
according to the storage battery state monitoring system provided by the embodiments of the invention, the trained Gaussian process regression model and the storage battery state monitoring network are utilized, so that the state of the storage battery can be tested and monitored, the problem of low accuracy of state monitoring of the storage battery is solved, and the accuracy of state monitoring of the storage battery is improved. The battery condition monitoring system may include a data acquisition module and a data processing module. The data acquisition module is used for acquiring data and sending the acquired data to the data processing module. The data includes: the voltage sequence and the current sequence of the storage battery to be detected in the first preset time period and the detection voltage sequence in the second preset time period. The state of the storage battery is often related to whether the voltage of the storage battery is abnormal or not, so that the voltage of the storage battery to be detected is obtained, and the state of the storage battery to be detected can be detected conveniently. The data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps: firstly, the voltage sequence and the current sequence are input into a trained Gaussian process regression model, and a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault prediction value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period are output through the Gaussian process regression model. And then, determining a predicted voltage sequence of the storage battery to be detected in the predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence. The larger the voltage change value of the storage battery to be detected due to faults is, the larger the voltage prediction uncertainty in the voltage prediction uncertainty sequence, the linear error prediction uncertainty in the linear error prediction uncertainty sequence and the fault prediction uncertainty in the fault prediction uncertainty sequence are output by the trained Gaussian process regression model. Therefore, the state of the storage battery to be detected can be conveniently detected later, and the accuracy of state monitoring of the storage battery is improved. Secondly, dividing the predicted voltage sequence of the storage battery to be detected in the predicted prediction time period into a normal predicted voltage sequence, a measured linear prediction error sequence and a voltage fault prediction value sequence of the storage battery to be detected in the prediction time period. 3 factors affecting the reading of the voltage of the storage battery to be detected, which is acquired by the sensor, are considered, so that the accuracy of state monitoring of the storage battery is improved. The factors corresponding to the normal predicted voltage sequence can represent the change of the voltage of the storage battery to be detected caused by the voltage and the current in the charging environment. The factor corresponding to the measurement of the linear prediction error sequence may characterize a change in the battery voltage to be detected due to a relatively low accuracy of the voltage sensor measuring the battery voltage to be detected or a failure of the voltage sensor. The factor corresponding to the voltage failure prediction value sequence may represent a change in the voltage of the battery to be detected due to the battery failure to be detected. And secondly, determining a voltage difference value sequence of the storage battery to be detected in the prediction time period according to the detection voltage sequence and the prediction voltage sequence. In practical situations, when the storage battery to be detected fails, the voltage difference in the voltage difference sequence tends to be large. Therefore, considering the voltage difference value in the voltage difference value sequence, the accuracy of fault judgment of the storage battery to be detected can be improved. And then, carrying out joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence. And finally, inputting the combined sequence into a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected. Since the joint sequence is a joint of the voltage difference sequence, the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence, the joint sequence contains more data related to whether the storage battery to be detected has a fault or not. Therefore, the combined sequence is input into the storage battery state monitoring network, so that the storage battery state monitoring network can detect more data related to whether the storage battery to be detected has faults or not, and the accuracy of the storage battery state monitoring network for monitoring the state of the storage battery to be detected is improved. The invention can realize the test and the monitoring of the state of the storage battery by utilizing the trained Gaussian process regression model and the storage battery state monitoring network, solves the problem of low accuracy of the state monitoring of the storage battery, and improves the accuracy of the state monitoring of the storage battery.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of some embodiments of a battery condition monitoring system according to the present disclosure;
FIG. 2 is a flow chart of some embodiments of a method for implementing a data processing module according to the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The embodiment provides a battery state monitoring system, which comprises a data acquisition module and a data processing module, wherein the data acquisition module is used for acquiring data and sending the acquired data to the data processing module, and the data comprises: the data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps:
inputting the voltage sequence and the current sequence into a trained Gaussian process regression model, and outputting a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault predicted value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period through the Gaussian process regression model;
determining a predicted voltage sequence of the storage battery to be detected in the predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence;
Determining a voltage difference sequence of the storage battery to be detected in the prediction time period according to the detection voltage sequence and the prediction voltage sequence;
performing joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence;
and inputting the combined sequence into a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected.
The following will be a detailed development of each of the above modules:
referring to fig. 1, a schematic diagram of some embodiments of a battery condition monitoring system according to the present invention is shown. The battery state monitoring system comprises a data acquisition module 101 and a data processing module 102. The data acquisition module 101 may be configured to acquire data and send the acquired data to the data processing module 102. The data may include: the voltage sequence and the current sequence of the storage battery to be detected in the first preset time period and the detection voltage sequence in the second preset time period. The data processing module 102 may be configured to receive data sent by the data acquisition module.
The storage battery to be detected may be a storage battery for detecting whether the storage battery to be detected has a voltage abnormality in a charging process. The first preset time period and the second preset time period may be preset time periods of the to-be-detected storage battery in a charging process, respectively. And adding the duration corresponding to the first preset time period and the duration corresponding to the second preset time period, wherein the obtained sum can be equal to the total duration of the charging process of the storage battery to be detected. The starting time of the first preset time period may be a time when the to-be-detected storage battery starts to be charged. The end time of the second preset time period may be the time when the battery to be detected ends charging. The ending time of the first preset time period may be the same as the starting time of the second preset time period.
The voltage in the voltage sequence may be the voltage of the battery to be detected, which is acquired by a voltage sensor, in the first preset period. The current in the current sequence may be the current of the battery to be detected, which is collected by the current sensor, in the first preset time period. The time interval between the instants at which adjacent voltages in the sequence of voltages are acquired may be the same as the time interval between the instants at which adjacent currents in the sequence of currents are acquired. The above-mentioned voltage sensor may be a sensor for detecting the voltage of the battery during the charging of the battery. The current sensor may be a sensor for detecting a current of the battery during charging of the battery. For example, the voltage sensor and the current sensor may be electrochemical detection sensors. The detected voltage in the detected voltage sequence may be a voltage of the battery to be detected acquired by the voltage sensor within the second preset period.
With reference to FIG. 2, a flowchart of some embodiments of a method for implementing a data processing module in accordance with the present invention is shown. The data processing module realizes the following steps:
step 201, the voltage sequence and the current sequence are input into a trained gaussian process regression model, and a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault prediction value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period are output through the gaussian process regression model.
In some embodiments, the voltage sequence and the current sequence may be input into a trained gaussian process regression model, and the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence, and the fault prediction uncertainty sequence of the battery to be detected in a prediction time period are output through the gaussian process regression model.
The prediction period may be a period of time preset after the first preset period of time. The normal predicted voltage in the normal predicted voltage sequence may be a predicted voltage of the battery to be detected that is not affected by the target factor. The target factor may be a factor that causes a voltage change when the battery is charged. For example, the target factors described above may include, but are not limited to: the accuracy of the voltage sensor for measuring the voltage of the battery is relatively low, the voltage sensor malfunctions, loss of active materials and lithium inside the battery due to charge and discharge, breakage of battery electrodes, short circuits, abnormal heat generation, and separator breakdown. When the accuracy of the voltage sensor is smaller than the preset accuracy, the accuracy of the voltage sensor can be considered to be relatively low. The measurement linear prediction error in the measurement linear prediction error sequence may be a predicted voltage variation amount due to a relatively low accuracy of a voltage sensor measuring the battery voltage to be detected or a failure of the voltage sensor. The voltage failure prediction value in the voltage failure prediction value sequence may be a predicted voltage variation amount due to the battery failure to be detected. The above-mentioned battery faults to be detected may include, but are not limited to: the battery internal active material and lithium are lost due to charge and discharge, battery electrode breakage, abnormal heat generation, short circuit, and separator breakdown. The voltage prediction uncertainty in the sequence of voltage prediction uncertainties may be an uncertainty of a normal predicted voltage prediction. The linear error prediction uncertainty in the linear error prediction uncertainty sequence may be an uncertainty of measuring a linear prediction error prediction. The fault prediction uncertainty in the sequence of fault prediction uncertainties may be an uncertainty in the prediction of the voltage fault prediction value. The time intervals between the times corresponding to adjacent elements in the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault predicted value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence, and the fault prediction uncertainty sequence may be the same. The voltage prediction uncertainty in the voltage prediction uncertainty sequence, the linear error prediction uncertainty in the linear error prediction uncertainty sequence, and the fault prediction uncertainty in the fault prediction uncertainty sequence may be characterized by standard deviations of predictions of normal prediction voltage, measured linear prediction error, and voltage fault prediction values, respectively, obtained by a gaussian process regression model.
With the progress of technology, the speed of charging the storage battery is increasing, for example, in a very fast state of charge, the storage battery is usually charged to 80% of its capacity within 10 minutes. The high power requirements of fast charging significantly increase the risk of battery failure. Therefore, the scheme mainly detects the state of the storage battery in the charging process.
Optionally, before this step, the step implemented by the data processing module may further include the following steps:
first, a standard voltage sequence and a standard current sequence in a standard time period are obtained.
The standard time period may be a time period during which the target storage battery is charged when the target storage battery is used for a period of time less than a preset period of time. The start time of the above-described standard period may be a time at which the target storage battery starts to be charged. And adding the time length corresponding to the standard time period and the time length corresponding to the prediction time period, wherein the obtained sum can be equal to the total time length of the target storage battery charging process. The target storage battery can be a storage battery with the specification and model of which the specification and model are the same as those of the storage battery to be detected, and the service time of the storage battery without faults is less than the preset time length. The preset duration may be a preset total duration of usage in which the battery to be detected starts to degrade (also referred to as aging). The standard voltages in the standard voltage sequence may be voltages of the target storage battery without being affected by the target factor. The standard current in the standard current sequence may be a current of the target storage battery without being affected by the target factor.
And a second step of respectively carrying out data alignment on the standard voltage sequence and the voltage sequence when the duration corresponding to the first preset time period is not equal to the duration corresponding to the standard time period, so as to obtain an aligned voltage sequence corresponding to the voltage sequence and an aligned current sequence corresponding to the current sequence.
After data alignment, the alignment voltage sequence, the alignment current sequence, the standard voltage sequence and the standard current sequence have the same length.
The battery tends to be more susceptible to degradation as the battery is used longer. The duration of the charging process of a degraded battery tends to be longer than the duration of the charging process before degradation of the battery. Different storage batteries tend to have different degradation degrees, so that the situation of battery degradation can be effectively avoided through data alignment, and the prediction accuracy of Gaussian process regression is improved.
For example, the standard time period may be a time period of the first 2/3 of the total time period of the target battery charging process. The first preset time period may be a time period of the first 2/3 of the total time period of the battery charging process to be detected. When the to-be-detected storage battery is degraded, the duration corresponding to the first preset time period is often longer than the duration corresponding to the standard time period. The standard voltage sequence and the voltage sequence may be matched by DTW (Dynamic Time Warping, dynamic time warping algorithm). Each standard voltage in the standard voltage sequence can be matched to obtain a plurality of voltages in the voltage sequence, and the standard voltage can be updated to be the average value of the plurality of voltages to obtain an aligned voltage sequence. Likewise, the standard current sequence and the current sequence may be matched by DTW. Each standard current in the standard current sequence can be matched to obtain a plurality of currents in the current sequence, and the standard current can be updated to be the average value of the plurality of currents to obtain an aligned current sequence.
For example, the standard voltage sequence may be a sequence of acquisition of a target battery voltage value every second in 3 minutes. The voltage sequence may be a sequence obtained by collecting a voltage value of the storage battery to be detected every second in 4 minutes, due to the change of the charging period of the storage battery caused by the degradation of the battery. By DTW matching, an aligned voltage sequence within 3 minutes can be obtained. By the method, the input of the Gaussian process regression model can be ensured to be an alignment voltage sequence and an alignment current sequence with fixed lengths.
And thirdly, determining the alignment voltage sequence and the alignment current sequence as the voltage sequence and the current sequence respectively.
Optionally, the training process of the gaussian process regression model may include the following steps:
first, a Gaussian process regression model is constructed.
The construction of the gaussian process regression model can be realized in an existing manner, and is not described in detail herein.
And secondly, acquiring a sample set of the sample storage battery.
Wherein, the samples in the sample set may include: a measured voltage sequence and a measured current sequence of the sample battery over a third preset time period. The labels corresponding to the samples in the sample set may include: normal voltage sequence, measurement linear error sequence and voltage fault value sequence of the sample storage battery in a fourth preset time period. The specification and model of the sample storage battery can be the same as the specification and model of the storage battery to be detected.
The third preset time period and the fourth preset time period may be preset time periods of the sample storage battery during charging, respectively. The starting time of the third preset time period may be a time when the sample battery starts to be charged. The end time of the fourth preset time period may be a time at which the sample battery ends charging. The sum of the duration corresponding to the fourth preset time period and the duration corresponding to the third preset time period may be the same as the total duration of the sample battery charging process. The ending time of the third preset time period may be the same as the starting time of the fourth preset time period. The measured voltage in the measured voltage sequence and the measured current in the measured current sequence may be the voltage and the current, respectively, measured by the sensor during charging of the sample battery. The sensors may include, but are not limited to: a voltage sensor and a current sensor. The time interval between the moments corresponding to adjacent elements in the measured voltage sequence and the measured current sequence may be the same. The normal voltage in the normal voltage sequence may be a voltage of the sample battery without being affected by the target factor. The measurement linearity errors in the measurement linearity error sequence may be the amount of voltage change due to a relatively low accuracy of the voltage sensor measuring the sample battery voltage or a voltage sensor failure. The voltage failure value in the sequence of voltage failure values may be an amount of voltage change due to a sample battery failure.
For example, the standard time period may be a time period of the first 2/3 of the total time period of the target battery charging process. The first preset time period may be a time period of the first 2/3 of the total time period of the battery charging process to be detected. The predicted period of time may be a period of time that is the last 1/3 of the total period of time of the target battery charging process. The second preset time period may be a time period of the latter 1/3 of the total time period of the battery charging process to be detected. The third preset time period may be a time period of the first 2/3 of the total time period of the sample battery charging process. The duration corresponding to the fourth preset time period may be a time period of the last 1/3 of the total time period of the sample battery charging process.
The sample set may include: a first sample set, a second sample set, and a third sample set. The first sample set may be a plurality of samples of the sample storage battery collected by the sensor when the sample storage battery is charged under a plurality of constant current and constant voltage. A first sample may correspond to a constant current and constant voltage. Constant current and constant voltage during charging often affect the normal voltage of the sample battery. The second sample set is a plurality of samples of the sample storage battery which is collected by the target sensor and is charged by the target charger during charging. The target sensor may be a relatively high precision sensor that can measure voltage and current. When the accuracy of the target sensor is not less than the preset accuracy, the accuracy of the target sensor can be considered to be relatively high. The target charger may be a charger adapted to the sample battery. The third sample set may be a plurality of samples taken when the simulated sample battery charge fails.
The normal voltage in the normal voltage sequence included in the tag corresponding to the first sample in the first sample set may be a voltage of the sample storage battery without being affected by the target factor. The measured linear error in the measured linear error sequence included in the tag corresponding to each first sample in the first sample set may be an absolute value of a difference between the measured voltage in the measured voltage sequence included in the first sample and the normal voltage in the normal voltage sequence included in the first sample. For example, the second measured linear error in the sequence of measured linear errors may be the absolute value of the difference between the second measured voltage in the sequence of measured voltages and the second normal voltage in the sequence of normal voltages. The voltage fault value in the voltage fault value sequence included in the label corresponding to the first sample in the first sample set may be zero.
The normal voltage in the normal voltage sequence included in the tag corresponding to the second sample in the second sample set may be a voltage of the sample storage battery under the influence of the target factor. The tag corresponding to the second sample in the second sample set may include a normal voltage in a normal voltage sequence equal to the voltage acquired by the target sensor. The labels corresponding to the second samples in the second sample set may include a measured linear error in the measured linear error sequence and a voltage fault value in the voltage fault value sequence, which may both be zero.
The normal voltage in the normal voltage sequence included in the label corresponding to the third sample in the third sample set may be a voltage of the sample storage battery under the influence of the target factor. The measurement linear error in the measurement linear error sequence included in the label corresponding to the third sample in the third sample set may be zero. The voltage fault value in the voltage fault value sequence included in the label corresponding to the third sample in the third sample set may be a simulated fault voltage. The voltage fault value in the voltage fault value sequence included in the label corresponding to the third sample in the third sample set may be a preset fault value.
Thirdly, training the Gaussian process regression model by using the sample set and the labels corresponding to the samples in the sample set to obtain the trained Gaussian process regression model.
Wherein, gaussian white noise can be added in the kernel of the Gaussian process regression model to improve the robustness of the Gaussian process regression model. The kernel function of the gaussian process regression model may be a square-index kernel function.
For example, first, the sample set described above may be input to a gaussian process regression model by which a normal voltage sequence, a measured linear error sequence, and a voltage failure value sequence within a fourth preset period of time are predicted. And respectively comparing the predicted normal voltage sequence, the measured linear error sequence and the voltage fault value sequence in a fourth preset time period with the normal voltage sequence, the measured linear error sequence and the voltage fault value sequence of the sample storage battery in the fourth preset time period, which are included in the label corresponding to the sample, until the predicted normal voltage sequence, the measured linear error sequence and the voltage fault value sequence of the sample storage battery meeting the preset precision in the fourth preset time period are obtained, so that training of a Gaussian process regression model is realized. The preset accuracy may be an accuracy set in advance.
For another example, when the duration corresponding to the third preset time period is longer than the duration corresponding to the standard time period, the labels corresponding to the samples in the sample set and the samples in the sample set may be aligned according to the standard voltage sequence by DTW matching, and further, the gaussian process regression model is trained by using the aligned labels corresponding to the sample set and the samples in the sample set, so as to obtain a trained gaussian process regression model. For each standard voltage in the standard voltage sequence, a plurality of measured voltages in the measured voltage sequence in the third preset time period can be obtained by matching through DTW matching, and the sample set after data alignment can be obtained through the following modes: and updating the standard voltage to be the average value of the plurality of measurement voltages, and obtaining a measurement voltage sequence after data alignment. And updating the standard voltage to be the average value of the measurement currents corresponding to the plurality of measurement voltages, and obtaining a measurement current sequence after data alignment. When the duration corresponding to the fourth preset time period is longer than the duration corresponding to the predicted time period, a measurement voltage sequence of the sample storage battery in the fourth preset time period can be obtained first. Then, according to the reference voltage sequence, the reference voltage sequence and the measurement voltage sequence in the fourth preset time period can be aligned in data through DTW matching. For each reference voltage in the reference voltage sequence, by DTW matching, a plurality of measurement voltages in the measurement voltage sequence in the fourth preset time period may be obtained by matching, and the labels corresponding to the samples in the sample set after the data alignment may be obtained by: and updating the reference voltage to be the average value of the normal voltages of the sample storage battery corresponding to the plurality of measured voltages in a fourth preset time period, so that a normal voltage sequence of the sample storage battery after data alignment in the fourth preset time period can be obtained. And updating the reference voltage to be the average value of the measurement linear errors of the sample storage battery corresponding to the plurality of measurement voltages in a fourth preset time period, so that a measurement linear error sequence of the sample storage battery after data alignment in the fourth preset time period can be obtained. And updating the reference voltage to be the average value of the voltage fault values of the sample storage battery corresponding to the plurality of measured voltages in a fourth preset time period, so that a voltage fault value sequence of the sample storage battery after data alignment in the fourth preset time period can be obtained. The training method for training the gaussian process regression model by using the sample set after data alignment and the labels corresponding to the samples in the sample set may refer to the training method for training the gaussian process regression model when the labels corresponding to the samples in the sample set and the sample set are not aligned.
Step 202, determining a predicted voltage sequence of the storage battery to be detected in a predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence.
In some embodiments, the predicted voltage sequence of the battery to be detected in the predicted time period may be determined according to the normal predicted voltage sequence, the measured linear prediction error sequence, and the voltage failure prediction value sequence.
As an example, the formula for determining that the predicted voltage sequence of the battery to be detected in the predicted period corresponds to the above formula may be:
u t ＝U t +τ t +μ t
wherein u is t Is the t-th predicted voltage in the predicted voltage sequence. U (U) t Is the t-th normal predicted voltage in the normal predicted voltage sequence. τ t Is the t-th measured linear prediction error in the sequence of measured linear prediction errors. Mu (mu) t Is the t-th voltage failure prediction value in the above-described voltage failure prediction value sequence.
And 203, determining a voltage difference sequence of the storage battery to be detected in a prediction time period according to the detection voltage sequence and the prediction voltage sequence.
In some embodiments, the voltage difference sequence of the battery to be detected in the prediction period may be determined according to the detected voltage sequence and the predicted voltage sequence.
As an example, this step may include the steps of:
the first step, when the duration corresponding to the second preset time period is not equal to the duration corresponding to the predicted time period, acquiring a reference voltage sequence in the predicted time period.
The reference voltages in the reference voltage sequence may be voltages of the target storage battery without being affected by target factors.
And secondly, carrying out data alignment on the detection voltage sequence and the reference voltage sequence to obtain an aligned voltage sequence corresponding to the detection voltage sequence.
For example, the predicted period of time may be a period of time that is the last 1/3 of the total period of time of the target battery charging process. The second preset time period may be a time period of the latter 1/3 of the total time period of the battery charging process to be detected. When the to-be-detected storage battery is degraded, the duration corresponding to the second preset time period is often longer than the duration corresponding to the predicted time period. The detection voltage sequence and the reference voltage sequence may be matched by DTW. Each reference voltage in the reference voltage sequence may be matched to obtain a plurality of detection voltages in the detection voltage sequence, and the reference voltage may be updated to be an average value of the plurality of detection voltages to obtain a complete voltage sequence.
And thirdly, determining a difference value of each regulated voltage in the regulated voltage sequence and a predicted voltage in the predicted voltage sequence corresponding to the regulated voltage as a voltage difference value, and obtaining the voltage difference value sequence.
As yet another example, when the duration corresponding to the second preset time period is equal to the duration corresponding to the predicted time period, a difference between each detected voltage in the detected voltage sequence and a predicted voltage in the predicted voltage sequence corresponding to the detected voltage may be determined as a voltage difference, and the voltage difference sequence may be obtained.
And 204, performing joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence.
In some embodiments, the voltage difference sequence, the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence, and the fault prediction uncertainty sequence may be combined to obtain a combined sequence.
As an example, the voltage difference sequence, the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence, and the fault prediction uncertainty sequence may be jointly operated by a concatate function in a Numpy packet in Python language to obtain a joint sequence.
Step 205, inputting the combined sequence to a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected.
In some embodiments, the combined sequence may be input to a battery state monitoring network after training, and the battery state corresponding to the to-be-detected battery is output through the battery state monitoring network, so as to implement detection of the to-be-detected battery state.
The storage battery state corresponding to the storage battery to be detected can be that the storage battery fails or the storage battery is normal. The battery state monitoring network may be used to detect whether the battery to be detected is faulty.
Optionally, the training process of the battery state monitoring network may include the following steps:
first, a storage battery state monitoring network is constructed.
The structure of the storage battery state monitoring network can be a mixed structure consisting of LSTM (Long-Short Term Memory, long-short-term memory model cyclic neural network) and FCN (Full Connected Network, fully connected network).
The construction of the battery state monitoring network can be realized in an existing manner, and is not described in detail herein.
And secondly, inputting the sample set into a trained Gaussian process regression model, and outputting a prediction information set corresponding to the sample set through the trained Gaussian process regression model.
Wherein, the prediction information in the prediction information set may include: a predicted normal voltage sequence, a predicted measured linear error sequence, a predicted voltage fault value sequence, a predicted voltage uncertainty sequence, a predicted linear error uncertainty sequence, and a predicted fault uncertainty sequence within the predicted time period.
The predicted normal voltage in the predicted normal voltage sequence may be a predicted voltage of the sample battery without being affected by the target factor. The predicted measured linearity error in the predicted measured linearity error sequence may be a predicted amount of voltage change due to a relatively low accuracy of a voltage sensor measuring the sample battery voltage or a voltage sensor failure. The predicted voltage failure value in the predicted voltage failure value sequence may be a predicted voltage variation due to a sample battery failure. Sample battery faults may include, but are not limited to: the battery internal active material and lithium are lost due to charge and discharge, battery electrode breakage, short circuit, and separator breakdown. The predicted voltage uncertainty in the sequence of predicted voltage uncertainties may be an uncertainty in predicting a normal voltage prediction. The predicted linear error uncertainty in the sequence of predicted linear error uncertainties may be an uncertainty of a predicted measured linear error prediction. The predicted fault uncertainty in the sequence of predicted fault uncertainties may be an uncertainty in the prediction of the predicted voltage fault value. The time intervals between the moments corresponding to adjacent elements in the predicted normal voltage sequence, the predicted measured linear error sequence, the predicted voltage fault value sequence, the predicted voltage uncertainty sequence, the predicted linear error uncertainty sequence, and the predicted fault uncertainty sequence may be the same.
By taking into account the uncertainty of the predicted normal voltage, the predicted measured linear error, and the predicted voltage fault value, a highly accurate battery state monitoring network can be trained with relatively few samples.
And thirdly, for each piece of prediction information in the prediction information set, determining a voltage prediction value sequence corresponding to the prediction information according to a prediction normal voltage sequence, a prediction measurement linear error sequence and a prediction voltage fault value sequence which are included in the prediction information.
For example, for each piece of the prediction information in the set of prediction information, a sum of each piece of the predicted normal voltage in the predicted normal voltage sequence included in the piece of prediction information, and a predicted voltage failure value of the predicted measurement linear error corresponding to the predicted normal voltage and corresponding to the predicted normal voltage may be determined as a voltage prediction value, and a voltage prediction value sequence corresponding to the piece of prediction information may be obtained.
Fourth, for each piece of prediction information in the set of prediction information, determining a sample voltage difference sequence corresponding to the piece of prediction information according to a voltage prediction value sequence corresponding to the piece of prediction information and a predicted normal voltage sequence included in the piece of prediction information.
For example, for each piece of the prediction information in the set of prediction information, a difference between each voltage prediction value in the sequence of voltage prediction values corresponding to the piece of prediction information and a predicted normal voltage corresponding to the voltage prediction value may be determined as a sample voltage difference value, and a sample voltage difference value sequence corresponding to the piece of prediction information may be obtained.
And fifthly, for each piece of prediction information in the prediction information set, performing joint operation on a sample voltage difference value sequence corresponding to the piece of prediction information, a predicted normal voltage sequence, a predicted measurement linear error sequence, a predicted voltage fault value sequence, a predicted voltage uncertainty sequence, a predicted linear error uncertainty sequence and a predicted fault uncertainty sequence which are included in the piece of prediction information, so as to obtain a predicted joint sequence corresponding to the piece of prediction information.
For example, for each piece of the prediction information in the prediction information set, a prediction joint sequence corresponding to the prediction information can be obtained by performing joint operation on a sample voltage difference value sequence corresponding to the prediction information, a prediction normal voltage sequence included in the prediction information, a prediction measurement linear error sequence, a prediction voltage fault value sequence, a prediction voltage uncertainty sequence, a prediction linear error uncertainty sequence and a prediction fault uncertainty sequence through a concatate function in a Numpy packet in a Python language.
And sixthly, training the storage battery state monitoring network by utilizing the prediction joint sequences corresponding to the prediction information in the prediction information set to obtain the trained storage battery state monitoring network.
Because the state of the storage battery corresponding to the prediction joint sequence is known, the storage battery state monitoring network can be trained by utilizing the prediction joint sequence corresponding to each prediction information in the prediction information set, and the trained storage battery state monitoring network is obtained.
The prediction joint sequence is a combination of a sample voltage difference value sequence, a prediction normal voltage sequence, a prediction measurement linear error sequence, a prediction voltage fault value sequence, a prediction voltage uncertainty sequence, a prediction linear error uncertainty sequence and a prediction fault uncertainty sequence, so that the prediction joint sequence contains more data related to whether the sample storage battery has a fault or not. Therefore, the prediction joint sequence is input into the storage battery state monitoring network, the storage battery state monitoring network can detect more data related to whether the sample storage battery has faults or not, and the accuracy of the storage battery state monitoring network for monitoring the state of the storage battery to be detected is improved.
According to the storage battery state monitoring system provided by the embodiments of the invention, the trained Gaussian process regression model and the storage battery state monitoring network are utilized, so that the state of the storage battery can be tested and monitored, the problem of low accuracy of state monitoring of the storage battery is solved, and the accuracy of state monitoring of the storage battery is improved. The battery condition monitoring system may include a data acquisition module and a data processing module. The data acquisition module is used for acquiring data and sending the acquired data to the data processing module. The data includes: the voltage sequence and the current sequence of the storage battery to be detected in the first preset time period and the detection voltage sequence in the second preset time period. The state of the storage battery is often related to whether the voltage of the storage battery is abnormal or not, so that the voltage of the storage battery to be detected is obtained, and the state of the storage battery to be detected can be detected conveniently. The data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps: firstly, the voltage sequence and the current sequence are input into a trained Gaussian process regression model, and a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault prediction value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period are output through the Gaussian process regression model. And then, determining a predicted voltage sequence of the storage battery to be detected in the predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence. The larger the voltage change value of the storage battery to be detected due to faults is, the larger the voltage prediction uncertainty in the voltage prediction uncertainty sequence, the linear error prediction uncertainty in the linear error prediction uncertainty sequence and the fault prediction uncertainty in the fault prediction uncertainty sequence are output by the trained Gaussian process regression model. Therefore, the state of the storage battery to be detected can be conveniently detected later, and the accuracy of state monitoring of the storage battery is improved. Secondly, dividing the predicted voltage sequence of the storage battery to be detected in the predicted prediction time period into a normal predicted voltage sequence, a measured linear prediction error sequence and a voltage fault prediction value sequence of the storage battery to be detected in the prediction time period. 3 factors affecting the reading of the voltage of the storage battery to be detected, which is acquired by the sensor, are considered, so that the accuracy of state monitoring of the storage battery is improved. The factors corresponding to the normal predicted voltage sequence can represent the change of the voltage of the storage battery to be detected caused by the voltage and the current in the charging environment. The factor corresponding to the measurement of the linear prediction error sequence may characterize a change in the battery voltage to be detected due to a relatively low accuracy of the voltage sensor measuring the battery voltage to be detected or a failure of the voltage sensor. The factor corresponding to the voltage failure prediction value sequence may represent a change in the voltage of the battery to be detected due to the battery failure to be detected. And secondly, determining a voltage difference value sequence of the storage battery to be detected in the prediction time period according to the detection voltage sequence and the prediction voltage sequence. In practical situations, when the storage battery to be detected fails, the voltage difference in the voltage difference sequence tends to be large. Therefore, considering the voltage difference value in the voltage difference value sequence, the accuracy of fault judgment of the storage battery to be detected can be improved. And then, carrying out joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence. And finally, inputting the combined sequence into a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected. Since the joint sequence is a joint of the voltage difference sequence, the normal predicted voltage sequence, the measured linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence, the joint sequence contains more data related to whether the storage battery to be detected has a fault or not. Therefore, the combined sequence is input into the storage battery state monitoring network, so that the storage battery state monitoring network can detect more data related to whether the storage battery to be detected has faults or not, and the accuracy of the storage battery state monitoring network for monitoring the state of the storage battery to be detected is improved. The invention can realize the test and the monitoring of the state of the storage battery by utilizing the trained Gaussian process regression model and the storage battery state monitoring network, solves the problem of low accuracy of the state monitoring of the storage battery, and improves the accuracy of the state monitoring of the storage battery.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.

Claims (10)

1. The storage battery state monitoring system is characterized by comprising a data acquisition module and a data processing module, wherein the data acquisition module is used for acquiring data and sending the acquired data to the data processing module, and the data comprises: the data processing module is used for receiving the data sent by the data acquisition module and realizing the following steps:
inputting the voltage sequence and the current sequence into a trained Gaussian process regression model, and outputting a normal predicted voltage sequence, a measured linear prediction error sequence, a voltage fault predicted value sequence, a voltage prediction uncertainty sequence, a linear error prediction uncertainty sequence and a fault prediction uncertainty sequence of the storage battery to be detected in a prediction time period through the Gaussian process regression model;
Determining a predicted voltage sequence of the storage battery to be detected in the predicted time period according to the normal predicted voltage sequence, the measured linear prediction error sequence and the voltage fault predicted value sequence;
determining a voltage difference value sequence of the storage battery to be detected in the prediction time period according to the detection voltage sequence and the prediction voltage sequence;
performing joint operation on the voltage difference value sequence, the normal prediction voltage sequence, the measurement linear prediction error sequence, the voltage fault prediction value sequence, the voltage prediction uncertainty sequence, the linear error prediction uncertainty sequence and the fault prediction uncertainty sequence to obtain a joint sequence;
and inputting the combined sequence into a storage battery state monitoring network after training, and outputting the storage battery state corresponding to the storage battery to be detected through the storage battery state monitoring network so as to realize detection of the storage battery state to be detected.
2. The system of claim 1, wherein the step of the data processing module implementing prior to the inputting the voltage sequence and the current sequence into the trained gaussian process regression model further comprises:
Acquiring a standard voltage sequence and a standard current sequence in a standard time period;
when the duration corresponding to the first preset time period is not equal to the duration corresponding to the standard time period, respectively carrying out data alignment on the standard voltage sequence and the voltage sequence, and obtaining an aligned voltage sequence corresponding to the voltage sequence and an aligned current sequence corresponding to the current sequence;
and determining the alignment voltage sequence and the alignment current sequence as the voltage sequence and the current sequence respectively.
3. The system of claim 1, wherein the determining a sequence of voltage differences for the battery to be detected over the predicted time period from the sequence of detected voltages and the sequence of predicted voltages comprises:
when the duration corresponding to the second preset time period is not equal to the duration corresponding to the prediction time period, acquiring a reference voltage sequence in the prediction time period;
data alignment is carried out on the detection voltage sequence and the reference voltage sequence, and a trimming voltage sequence corresponding to the detection voltage sequence is obtained;
and determining the difference value between each trimming voltage in the trimming voltage sequence and the predicted voltage in the predicted voltage sequence corresponding to the trimming voltage as a voltage difference value, and obtaining the voltage difference value sequence.
4. The system of claim 2, wherein the training process of the gaussian process regression model comprises:
constructing a Gaussian process regression model;
obtaining a sample set of sample batteries, wherein samples in the sample set comprise: the sample battery measures voltage sequence and measuring current sequence in the third preset time period, the label that the sample in sample set corresponds includes: normal voltage sequence, measurement linear error sequence and voltage fault value sequence of the sample storage battery in a fourth preset time period;
and training the Gaussian process regression model by using the sample set and the labels corresponding to the samples in the sample set to obtain a trained Gaussian process regression model.
5. The system of claim 4, wherein the battery condition monitoring network training process comprises:
constructing a storage battery state monitoring network;
inputting the sample set into a trained Gaussian process regression model, and outputting a prediction information set corresponding to the sample set through the trained Gaussian process regression model, wherein the prediction information in the prediction information set comprises: a predicted normal voltage sequence, a predicted measured linear error sequence, a predicted voltage fault value sequence, a predicted voltage uncertainty sequence, a predicted linear error uncertainty sequence and a predicted fault uncertainty sequence within the predicted time period;
For each piece of prediction information in the prediction information set, determining a voltage prediction value sequence corresponding to the prediction information according to a prediction normal voltage sequence, a prediction measurement linear error sequence and a prediction voltage fault value sequence which are included in the prediction information;
for each piece of prediction information in the prediction information set, determining a sample voltage difference value sequence corresponding to the prediction information according to a voltage prediction value sequence corresponding to the prediction information and a prediction normal voltage sequence included in the prediction information;
for each piece of prediction information in the prediction information set, performing joint operation on a sample voltage difference value sequence corresponding to the prediction information, a prediction normal voltage sequence included in the prediction information, a prediction measurement linear error sequence, a prediction voltage fault value sequence, a prediction voltage uncertainty sequence, a prediction linear error uncertainty sequence and a prediction fault uncertainty sequence to obtain a prediction joint sequence corresponding to the prediction information;
and training the storage battery state monitoring network by utilizing the prediction joint sequences corresponding to the prediction information in the prediction information set to obtain the trained storage battery state monitoring network.
6. The system of claim 4, wherein the sample set comprises: the system comprises a first sample set, a second sample set and a third sample set, wherein the first sample set is a plurality of samples of a sample storage battery collected through a sensor when the sample storage battery is charged under various constant currents and constant pressures, the second sample set is a plurality of samples of the sample storage battery which is collected through a target sensor and is charged by a target charger when the sample storage battery is charged, and the third sample set is a plurality of samples collected when the sample storage battery is simulated to be charged and fault.
7. The system of claim 6, wherein the normal voltage in the normal voltage sequence included in the tag corresponding to the first sample in the first sample set is a voltage of the sample battery without being affected by the target factor, and the measured linear error in the measured linear error sequence included in the tag corresponding to each first sample in the first sample set is an absolute value of a difference between the measured voltage in the measured voltage sequence included in the first sample and the normal voltage in the normal voltage sequence included in the first sample, and the voltage fault value in the voltage fault value sequence included in the tag corresponding to the first sample in the first sample set is zero.
8. The system of claim 6, wherein the normal voltage in the normal voltage sequence included in the label corresponding to the second sample in the second sample set is a voltage of the sample battery without being affected by the target factor, and the measured linear error in the measured linear error sequence and the voltage fault value in the voltage fault value sequence included in the label corresponding to the second sample in the second sample set are both zero.
9. The system of claim 6, wherein the normal voltage in the normal voltage sequence included in the label corresponding to the third sample in the third sample set is a voltage of the sample battery under the influence of no target factor, the measured linear error in the measured linear error sequence included in the label corresponding to the third sample in the third sample set is zero, and the voltage fault value in the voltage fault value sequence included in the label corresponding to the third sample in the third sample set is a preset fault value.
10. The system of claim 1, wherein the formula corresponding to the predicted voltage sequence of the battery to be detected over the predicted time period is determined as:
u t ＝U t +τ t +μ t
Wherein u is t Is the t-th predicted voltage in the predicted voltage sequence, U t Is the t-th normal predicted voltage, τ, in the sequence of normal predicted voltages t Is the t-th linear prediction error, mu, in the linear prediction error sequence t Is the t-th voltage failure prediction value in the sequence of voltage failure prediction values.
CN202211026238.2A 2022-08-25 2022-08-25 Storage battery state monitoring system Active CN116643181B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211026238.2A CN116643181B (en) 2022-08-25 2022-08-25 Storage battery state monitoring system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211026238.2A CN116643181B (en) 2022-08-25 2022-08-25 Storage battery state monitoring system

Publications (2)

Publication Number Publication Date
CN116643181A true CN116643181A (en) 2023-08-25
CN116643181B CN116643181B (en) 2024-04-16

Family

ID=87621762

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211026238.2A Active CN116643181B (en) 2022-08-25 2022-08-25 Storage battery state monitoring system

Country Status (1)

Country Link
CN (1) CN116643181B (en)

Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104090080A (en) * 2014-07-16 2014-10-08 胡小青 Monitoring method for abnormal state of oil-immersed transformer
CN104200113A (en) * 2014-09-10 2014-12-10 山东农业大学 Internet of Things data uncertainty measurement, prediction and outlier-removing method based on Gaussian process
CN106124884A (en) * 2014-07-16 2016-11-16 曹新民 The warning system of Monitoring Power Transformer abnormality
CN107153162A (en) * 2017-06-06 2017-09-12 山东大学 A kind of power battery pack multiple faults on-line detecting system and method
CN108008176A (en) * 2017-11-22 2018-05-08 福州大学 A kind of photovoltaic array real-time state monitoring and fault location system
CN108544925A (en) * 2018-04-02 2018-09-18 北京理工大学 Battery management system
CN109975026A (en) * 2019-03-29 2019-07-05 潍柴动力股份有限公司 Engine failure monitoring method and relevant device
CN109990923A (en) * 2016-06-23 2019-07-09 福州丹诺西诚电子科技有限公司 Improve the fault diagnosis method and system of the temperature sensor of utilization of resources rate
CN110789396A (en) * 2019-11-04 2020-02-14 广州小鹏汽车科技有限公司 Battery pack charge state data detection method, device, medium, terminal and automobile
CN111695618A (en) * 2020-06-01 2020-09-22 清华大学深圳国际研究生院 Electric vehicle motor fault detection method based on OBD data
CN112349981A (en) * 2020-11-04 2021-02-09 中天鸿锂清源股份有限公司 Battery management system
CN113657360A (en) * 2021-10-21 2021-11-16 长沙德壹科技有限公司 Lithium battery health state estimation method, device, equipment and readable storage medium
CN114325395A (en) * 2020-09-30 2022-04-12 北京昇科能源科技有限责任公司 Method and device for determining battery state
CN114330150A (en) * 2022-03-15 2022-04-12 四川大学 Lithium battery health state estimation method with electric heating sensors mutually corrected
CN114611597A (en) * 2022-03-07 2022-06-10 山东大学 Lithium ion battery health state estimation method and system based on RGAN
CN114935721A (en) * 2022-05-30 2022-08-23 深圳先进技术研究院 Lithium ion battery charge state estimation method based on fiber grating sensor

Patent Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104090080A (en) * 2014-07-16 2014-10-08 胡小青 Monitoring method for abnormal state of oil-immersed transformer
CN106124884A (en) * 2014-07-16 2016-11-16 曹新民 The warning system of Monitoring Power Transformer abnormality
CN104200113A (en) * 2014-09-10 2014-12-10 山东农业大学 Internet of Things data uncertainty measurement, prediction and outlier-removing method based on Gaussian process
CN109990923A (en) * 2016-06-23 2019-07-09 福州丹诺西诚电子科技有限公司 Improve the fault diagnosis method and system of the temperature sensor of utilization of resources rate
CN107153162A (en) * 2017-06-06 2017-09-12 山东大学 A kind of power battery pack multiple faults on-line detecting system and method
CN108008176A (en) * 2017-11-22 2018-05-08 福州大学 A kind of photovoltaic array real-time state monitoring and fault location system
CN108544925A (en) * 2018-04-02 2018-09-18 北京理工大学 Battery management system
CN109975026A (en) * 2019-03-29 2019-07-05 潍柴动力股份有限公司 Engine failure monitoring method and relevant device
CN110789396A (en) * 2019-11-04 2020-02-14 广州小鹏汽车科技有限公司 Battery pack charge state data detection method, device, medium, terminal and automobile
CN111695618A (en) * 2020-06-01 2020-09-22 清华大学深圳国际研究生院 Electric vehicle motor fault detection method based on OBD data
CN114325395A (en) * 2020-09-30 2022-04-12 北京昇科能源科技有限责任公司 Method and device for determining battery state
CN112349981A (en) * 2020-11-04 2021-02-09 中天鸿锂清源股份有限公司 Battery management system
CN113657360A (en) * 2021-10-21 2021-11-16 长沙德壹科技有限公司 Lithium battery health state estimation method, device, equipment and readable storage medium
CN114611597A (en) * 2022-03-07 2022-06-10 山东大学 Lithium ion battery health state estimation method and system based on RGAN
CN114330150A (en) * 2022-03-15 2022-04-12 四川大学 Lithium battery health state estimation method with electric heating sensors mutually corrected
CN114935721A (en) * 2022-05-30 2022-08-23 深圳先进技术研究院 Lithium ion battery charge state estimation method based on fiber grating sensor

Also Published As

Publication number Publication date
CN116643181B (en) 2024-04-16

Similar Documents

Publication Publication Date Title
CN108254696B (en) Battery health state evaluation method and system
US6332113B1 (en) Electronic battery tester
US5945829A (en) Midpoint battery monitoring
JP5349810B2 (en) Storage device abnormality detection device, method, and program
US20170115355A1 (en) Maximum capacity estimator for battery state of health and state of charge determinations
US11686699B2 (en) System and method for anomaly detection and total capacity estimation of a battery
US10024924B2 (en) Remaining battery life prediction device and battery pack
US9450427B2 (en) State of charge determination for an electrical accumulator
EP2397863A1 (en) System for monitoring the status of a battery
CN108333548A (en) Insulation resistance measuring apparatus and fault self-diagnosis method
US11846680B2 (en) Battery resistance diagnosis device and method
CN110888065B (en) Battery pack state of charge correction method and device
CN102540092A (en) Method and device for monitoring the maximum available capacity of battery
US11688887B2 (en) Slave BMS inspection system and method
KR20200017367A (en) Apparatus for battery diagnosis
CN107430172B (en) Battery monitoring apparatus and method using discharge pulse measurements
CN116643181B (en) Storage battery state monitoring system
CN116338487B (en) Battery cell detection method and device, electronic equipment and storage medium
CN111413564B (en) Supercapacitor failure early warning method, system and equipment
CN114514433A (en) Apparatus and method for diagnosing battery
US11796605B2 (en) Battery cell diagnostic device and method
Peng et al. Robust Physics Guided Data-Driven Fleet Battery Pack Fault Detection Under Dynamic Operating Conditions
CN113296011B (en) Circuit conducted signal active detection analysis system
Reichl et al. Impulse diagnostical test of batterries
US20220291290A1 (en) Measurement appratus of power storage device and measurement method

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