WO2022173384A1 - Early fault detection method for a battery - Google Patents
Early fault detection method for a battery Download PDFInfo
- Publication number
- WO2022173384A1 WO2022173384A1 PCT/TR2021/050125 TR2021050125W WO2022173384A1 WO 2022173384 A1 WO2022173384 A1 WO 2022173384A1 TR 2021050125 W TR2021050125 W TR 2021050125W WO 2022173384 A1 WO2022173384 A1 WO 2022173384A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- battery
- cell
- artificial intelligence
- fault
- Prior art date
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 31
- 238000009826 distribution Methods 0.000 claims abstract description 57
- 238000013473 artificial intelligence Methods 0.000 claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 8
- 238000013528 artificial neural network Methods 0.000 claims description 14
- 230000006399 behavior Effects 0.000 claims description 9
- 238000005259 measurement Methods 0.000 claims description 8
- 230000001131 transforming effect Effects 0.000 claims description 2
- 210000004027 cell Anatomy 0.000 description 95
- 238000004422 calculation algorithm Methods 0.000 description 10
- 230000015572 biosynthetic process Effects 0.000 description 4
- 238000007635 classification algorithm Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 229910001317 nickel manganese cobalt oxide (NMC) Inorganic materials 0.000 description 2
- 230000001172 regenerating effect Effects 0.000 description 2
- 238000010998 test method Methods 0.000 description 2
- 230000002547 anomalous effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L3/00—Electric devices on electrically-propelled vehicles for safety purposes; Monitoring operating variables, e.g. speed, deceleration or energy consumption
- B60L3/0023—Detecting, eliminating, remedying or compensating for drive train abnormalities, e.g. failures within the drive train
- B60L3/0046—Detecting, eliminating, remedying or compensating for drive train abnormalities, e.g. failures within the drive train relating to electric energy storage systems, e.g. batteries or capacitors
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L58/00—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles
- B60L58/10—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries
- B60L58/12—Methods or circuit arrangements for monitoring or controlling batteries or fuel cells, specially adapted for electric vehicles for monitoring or controlling batteries responding to state of charge [SoC]
-
- 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]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- 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]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
- B60L2260/00—Operating Modes
- B60L2260/40—Control modes
- B60L2260/50—Control modes by future state prediction
-
- 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
- G01R31/3835—Arrangements for monitoring battery or accumulator variables, e.g. SoC involving only voltage measurements
Definitions
- the object of the invention is to early detection of the cell fault of the cells in the batteries.
- the invention is a fault detection method for a battery comprising creating a battery pack virtual model providing a voltage threshold value indicating voltage distribution values of at least one battery cell at the battery and usability of predefined a battery cell by means of a virtual data generator in a control unit; collecting a first data part and at least a second data part from the created battery pack virtual model by means of a data collector; generating a training dataset by merging the first and second data parts by a controller; training an artificial intelligence module to detect and distinguish the cells that behave differently from distribution values by a controller detecting the cell voltage distribution values according to the changing parameters by utilizing the training data and process as the input data.
- a preferred application comprising the step of obtaining the first and second data parts with a data collector by simultaneous and repeatedly measuring current value distribution of the battery cell at the virtual model of the battery pack and matching the data and corresponding various parameters at each point of measurement. Simultaneous and repeated measurement facilitates for the artificial intelligence module to learn battery cell voltage distribution values under various parameters.
- a preferred application comprises the step of obtaining a training data by merging the first and the second data parts and battery cell voltage distribution value at the point of measurement corresponding to the parameters selected from the group comprising the maximum usable battery charge, discharge interval and temperature change.
- the training data is used to train the artificial intelligence module.
- the artificial intelligence is trained with the training data until it reaches the desired performance level. Thus, safety is ensured for application of artificial intelligence to the batteries on the vehicle.
- a preferred application comprises the step of transforming the information classified by the artificial intelligence module into a result data.
- the information obtained by the fault detection method can be converted into a report and suitable for storage.
- a preferred application of the invention the step of saving the result data in a memory unit by means of the artificial intelligence module.
- the loss of result data is prevented and the output data generated by other test results can be accessed at any time.
- Figure 1 is a schematic representation of the control unit and the virtual model of the battery pack created by the control unit of the subject matter fault detection method.
- Figure 2 is a schematic illustration of the artificial intelligence module and signal processor and analytics module of the inventive fault detection method.
- Figure 3 is a schematic representation of the process steps of the inventive fault detection method.
- FIG. 1 shows schematically the control unit of the inventive fault detection method and the battery model created with the control unit.
- a control unit (10) operates according to the working principle of the fault detection method used for batteries.
- a virtual data generator (12) in the control unit (12) visually transforms the battery pack virtual model (20) indicating the cell voltage distribution values and voltage threshold value of the batteries corresponding to a certain temperature change, a certain maximum battery charge and discharge interval parameter indicating the usable status of the battery cells (1) into a virtual space.
- the virtual model of the battery pack (20) is created by the virtual data generator (12).
- the virtual model of the battery pack (20) is arranged to indicate the cell voltage distribution values and voltage threshold value of the batteries corresponding to a certain temperature change, a certain maximum battery charge and discharge interval parameter.
- the virtual modal of the battery pack (20) arranged such that indicate the states of the battery cells (1 ) with different predetermined voltage values under various parameters.
- multiple data parts of the cell voltage distribution value and voltage threshold value of the battery are collected with a data collector (14) arranged in the control unit (10).
- the data parts obtained by the data collector (14) are formed by the simultaneous and repeated measurement of the battery cell voltage distribution in the battery pack virtual model (20) and matching the data parts at each corresponding measurement point.
- the obtained data parts are combined into a training data by means of a controller (16) located in the control unit (10).
- the control unit (10) is having a memory unit (18) that stores the results obtained by the test method.
- FIG. 2 shows schematically the artificial intelligence module and signal processor and analytical module of the test method of the subject matter invention.
- a signal processor and analytical module (40) that is trained to determine the battery cell voltage distribution and cell voltage threshold value according to the changing parameters of the cells in the batteries.
- the signal processor and analytics module (40) calculates the voltage distribution value of the battery cell (1) and creates input data (40), and by using the input data, the artificial intelligence module (30) is a multi-layered neural network (32) detects and distinguishes cells that behave differently from these distribution values.
- the neural network (32) comprises multiple neurons configured to take the cell voltage distribution values as input data and apply an algorithm that detects and distinguishes cells that behave differently from these distribution values.
- the algorithm taking the cell voltage distribution values as input data and using the input data, detects and distinguishes cells that behave differently from these distribution values is trained with the supervised learning method.
- the training data consists of inputs paired with correct corresponding outputs.
- the algorithm models patterns in the data associated with the desired outputs.
- a supervised learning algorithm will take new invisible inputs and determine what result the new inputs will achieve, based on previous training data.
- the object of the supervised learning model is to predict the correct outcome for newly presented input data.
- the supervised learning method can be preferred as a classification or regression model.
- the classification model provides data points with a category assigned to a classification algorithm during training.
- the task of a classification algorithm is then to take an input value and assign it a class or category to which it is appropriate based on the training data provided.
- an input layer (322), at least one middle layer (324) and an output layer (326) in the neural network (32) are used to input the cell voltage distribution values of the battery cell (1 ) according to the changing parameters and trained through supervised training to detect and distinguish cells that behave differently from these distribution values.
- the cell voltage distribution value is calculated by the signal processor and analytics module (30) and created as input data (50) and transmitted to the artificial intelligence module (30).
- the neurons of the input layer (322) in the neural network (32) located in the artificial intelligence module (30) distinguish the cells that give signals that there is a problem by displaying different behavior among these distribution values.
- Cell fault status is classified by the output layer (326) as low severity, medium severity, and high severity (70).
- the obtained information is converted into a result data (80) by the artificial intelligence module (30) and recorded in the memory unit (18).
- the virtual data generator (12) in the control unit (10) creates the virtual model (20) of the battery pack.
- the voltage distribution values of the battery according to the changing parameters and the voltage threshold values according to the changing parameters are collected by the data collector (14) as data parts.
- a collected first piece of data, a second piece of data, and a third piece of data are formed according to the varying maximum battery charge and discharge interval and the measurement point corresponding to the temperature change.
- the data parts are combined into training data by the controller (16) in the control unit (10).
- the control unit (10) trains the artificial intelligence module (30). With the neuroprocessor and analytical module (40), cell voltage distribution values are generated as input data (50).
- the artificial intelligence module (30) is trained with data parts from the battery pack virtual model (20) until it learns completely to detect and distinguish cells that give early signals about problem formation by taking the battery cell voltage distribution values as input according to the changing parameters of the batteries and displaying different behavior from these distribution values.
- Training the artificial intelligence module (30) using training data could be, for example, a supervised learning model using a recurrent neural network (RNN) model or the like.
- RNN recurrent neural network
- the voltage distribution value of the battery cell (1 ) is transmitted by the signal processor and analytical module (40) as input data by the artificial intelligence module (30), and the value of the cell voltage distribution is used by the artificial intelligence module (30) as input data, and different behavior is observed among these distribution values and detecting cells that give early signals of problem formation, it is determined whether the battery cell (1) has a cell fault condition (60). Depending on the severity of the cell fault, the artificial intelligence module (30) classifies the cell fault (70). The classified cell fault is converted into a result data (80) and stored in the memory unit. In an exemplary application, LiNiMnCo02 (NMC), were chosen as the batteries utilized for an electric vehicle.
- NMC LiNiMnCo02
- the detected voltage distribution values are processed with the artificial intelligence algorithm neural network (32) to obtain information whether there is an early signal of cell fault formation or not. If the cell has a fault, it is classified as low severity, average severity, or high severity cell fault, depending on the severity of the fault. In this way, whether the battery cell (1 ) used for the vehicle has a cell fault or not, the faults of the battery cell (1) are actively detected in the background while the vehicle is running, and it can be determined that whether it is not suitable for the use. If a battery with a cell fault can be made suitable for an active ride or a new battery can be used is determined by considering the severity of the cell fault.
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Transportation (AREA)
- Mechanical Engineering (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Secondary Cells (AREA)
Abstract
The invention relates to a fault detection method for a battery comprising creating (20) a battery pack virtual model providing a voltage threshold value indicating voltage distribution values of at least one battery cell (1) at the battery and usability of predefined a battery cell (1) by means of a virtual data generator (12) in a control unit (10); collecting a first data part and at least a second data part from the created battery pack virtual model by means of a data collector (14); generating a training dataset by merging the first and second data parts by a controller (16); training an artificial intelligence module (30) to detect and distinguish the cells that behave differently from distribution values by a controller (16) detecting the cell voltage distribution values according to the changing parameters by utilizing the training data and process as the input data; training an artificial intelligence module (30) such that determining (60) whether the battery cell (1) has a cell fault condition utilizing the input data detecting and distinguishing the battery cells (1) that behave differently from the distribution values.
Description
EARLY FAULT DETECTION METHOD FOR A BATTERY
TECHNICAL FIELD
The present invention relates to a fault detection system of a battery, in particular a fault detection system powered by artificial intelligence.
BACKGROUND OF THE ART
Electric vehicles are vehicles in which only an electric motor is utilized as a driving part, and hybrid vehicles are vehicles both have an internal combustion engine and an electric motor in drivetrain. A battery pack with a plurality of battery cells is provided on the underside of an electric vehicle and a hybrid vehicle.
The lifetime of the battery used in an electric vehicle or a hybrid electric vehicle is reduced depending on the number of charges, depth of discharge, charge and discharge current, charge environment (temperature change) or similar environmental conditions. This situation also depends on the driving behaviour of the user and charging temperatures, fast charging frequency, amount of charge with regenerative brake, timing of charge with regenerative brake and similar conditions may cause a reduction in lifetime. Anomalies may occur in the battery utilized in an electric vehicle or hybrid electric vehicle due to production faults, deterioration over time or similar reasons.
US2020278398A1 discloses an anomaly detection system for a secondary battery which detects the remaining capacity of the secondary battery on an electric vehicle, cautions against the secondary battery with anomalous characteristics, stops using the secondary battery, changes the secondary battery, or changes charging conditions of the secondary battery is provided. The anomaly detection system is provided; the system compares a value obtained by estimating internal resistance or SOC of a secondary battery based on the measured value of a current or a voltage of the secondary battery with the use of a nonlinear Kalman filter and a value input to an anomaly detection system (network) of Al to predict a change in the internal resistance; the system regards a case where the difference is large as an anomaly; and the system detects an anomaly.
BRIEF DESCRIPTION OF THE INVENTION
The object of the invention is to early detection of the cell fault of the cells in the batteries.
In order to achieve the mentioned objectives, the invention is a fault detection method for a battery comprising creating a battery pack virtual model providing a voltage threshold value indicating voltage distribution values of at least one battery cell at the battery and usability of predefined a battery cell by means of a virtual data generator in a control unit; collecting a first data part and at least a second data part from the created battery pack virtual model by means of a data collector; generating a training dataset by merging the first and second data parts by a controller; training an artificial intelligence module to detect and distinguish the cells that behave differently from distribution values by a controller detecting the cell voltage distribution values according to the changing parameters by utilizing the training data and process as the input data. The fault detection method comprising the steps of determining the cell voltage distributions of the battery cell and generating an input data by a neuroprocessor and analytics module connected with the artificial intelligence module in a signal transmitting manner; determining whether the battery cell has a cell fault condition, by means of artificial intelligence module utilizing the input data detecting and distinguishing the battery cells that behave differently from the distribution values. Thus, it can be determined whether the batteries that can be used for the vehicle have a cell fault early before the use. In addition, the fault detection method works actively in the background while the vehicle is running, and the cell fault situation that may occur in the battery cell during use can be detected in advance. In this way, problems that may occur during the use of batteries in the vehicle are prevented and malfunctions that may occur in the vehicle are prevented.
In a preferred application, the artificial intelligence module comprises the process step of classification of the cell fault status as a low severity level, average severity level and high severity level. Sorting by severity allows to determine which cell is in the most risky state. In an alternative embodiment, the classification can be based on present, not present flags. In another alternative application, the degree of fault can be calculated as a % value.
A preferred application comprising the step of early detection of cell fault by detecting and distinguishing the cells with different behavior from these distribution values by means of a neural network provided in the artificial intelligence module taking the cell voltage distribution values as the input data and assigning the severity level of the cell fault depending on the cell voltage distribution. The mathematical calculations made by the neural network provide reliable results. Thus, cell fault detection for battery cells can be accurately done.
A preferred application comprises the step of training the neural network by supervised learning by utilizing training data comprising an input layer, at least one middle layer and an output layer assigning current value of the battery cell according to the changing parameters detecting early fault occurrence by determining and distinguishing cells with different behavior among the cell value distribution. It is ensured that the input layer, middle layer and output layer in the neural network learn all the dynamics of the battery cell.
A preferred application comprising the step of obtaining the first and second data parts with a data collector by simultaneous and repeatedly measuring current value distribution of the battery cell at the virtual model of the battery pack and matching the data and corresponding various parameters at each point of measurement. Simultaneous and repeated measurement facilitates for the artificial intelligence module to learn battery cell voltage distribution values under various parameters.
A preferred application comprises the step of obtaining a training data by merging the first and the second data parts and battery cell voltage distribution value at the point of measurement corresponding to the parameters selected from the group comprising the maximum usable battery charge, discharge interval and temperature change. The training data is used to train the artificial intelligence module. In order for the artificial intelligence to use the voltage distribution values of the battery cell as input data and to detect and distinguish cells that behave differently from these distribution values by using the input data, the artificial intelligence is trained with the training data until it reaches the desired performance level. Thus, safety is ensured for application of artificial intelligence to the batteries on the vehicle.
A preferred application comprises the step of transforming the information classified by the artificial intelligence module into a result data. Thus, the information obtained by the fault detection method can be converted into a report and suitable for storage.
A preferred application of the invention the step of saving the result data in a memory unit by means of the artificial intelligence module. Thus, the loss of result data is prevented and the output data generated by other test results can be accessed at any time.
BRIEF DESCRIPTION OF THE FIGURES
Figure 1 is a schematic representation of the control unit and the virtual model of the battery pack created by the control unit of the subject matter fault detection method.
Figure 2 is a schematic illustration of the artificial intelligence module and signal processor and analytics module of the inventive fault detection method.
Figure 3 is a schematic representation of the process steps of the inventive fault detection method.
DETAILED DESCRIPTION OF THE INVENTION
In this detailed explanation, the invention is explained without any limitation and only with exemplary references to better explain the subject matter invention.
Figure 1 shows schematically the control unit of the inventive fault detection method and the battery model created with the control unit. A control unit (10) operates according to the working principle of the fault detection method used for batteries. A virtual data generator (12) in the control unit (12) visually transforms the battery pack virtual model (20) indicating the cell voltage distribution values and voltage threshold value of the batteries corresponding to a certain temperature change, a certain maximum battery charge and discharge interval parameter indicating the usable status of the battery cells (1) into a virtual space. The virtual model of the battery pack (20) is created by the virtual data generator (12). The virtual model of the battery pack (20) is arranged to indicate the cell voltage distribution values and voltage threshold value of the batteries corresponding to a certain temperature change, a certain maximum battery charge and discharge interval parameter. The virtual modal of the battery pack (20) arranged such that indicate the states of the battery cells (1 ) with different predetermined voltage values under various parameters. According to the changing parameters, multiple data parts of the cell voltage distribution value and voltage threshold value of the battery are collected with a data collector (14) arranged in the control unit (10). The data parts obtained by the data collector (14) are formed by the simultaneous and repeated measurement of the battery cell voltage distribution in the battery pack virtual model (20) and matching the data parts at each corresponding measurement point. The obtained data parts are combined into a training data by means of a controller (16) located in the control unit (10). The control unit (10) is having a memory unit (18) that stores the results obtained by the test method.
Figure 2 shows schematically the artificial intelligence module and signal processor and analytical module of the test method of the subject matter invention. Along with the training data, there is a signal processor and analytical module (40) that is trained to determine the
battery cell voltage distribution and cell voltage threshold value according to the changing parameters of the cells in the batteries. The signal processor and analytics module (40) calculates the voltage distribution value of the battery cell (1) and creates input data (40), and by using the input data, the artificial intelligence module (30) is a multi-layered neural network (32) detects and distinguishes cells that behave differently from these distribution values. The neural network (32) comprises multiple neurons configured to take the cell voltage distribution values as input data and apply an algorithm that detects and distinguishes cells that behave differently from these distribution values. The algorithm taking the cell voltage distribution values as input data and using the input data, detects and distinguishes cells that behave differently from these distribution values is trained with the supervised learning method. When the supervised learning algorithm is trained, the training data consists of inputs paired with correct corresponding outputs. During training, the algorithm models patterns in the data associated with the desired outputs. After training, a supervised learning algorithm will take new invisible inputs and determine what result the new inputs will achieve, based on previous training data. The object of the supervised learning model is to predict the correct outcome for newly presented input data. The supervised learning method can be preferred as a classification or regression model. The classification model provides data points with a category assigned to a classification algorithm during training. The task of a classification algorithm is then to take an input value and assign it a class or category to which it is appropriate based on the training data provided. With the classification training model, the information whether the battery cell (1 ) has a battery fault is determined by two classes as having a cell fault that can be selected or not having a cell fault. Training data on battery cells (1 ) with and without cell fault are given to the algorithm. The model will find the features associated with any of the data and create the matching function: Y = f(x). Subsequently, when a battery cell (1 ) is provided which is unknown whether it has a faulty cell, the model can determine whether the battery cell (1 ) has a cell fault. With the classification training model, an input layer (322), at least one middle layer (324) and an output layer (326) in the neural network (32) are used to input the cell voltage distribution values of the battery cell (1 ) according to the changing parameters and trained through supervised training to detect and distinguish cells that behave differently from these distribution values.
In Figure 3, the processing steps executed by the artificial intelligence module are shown schematically. The cell voltage distribution value is calculated by the signal processor and analytics module (30) and created as input data (50) and transmitted to the artificial intelligence module (30). With the input data, the neurons of the input layer (322) in the neural network (32) located in the artificial intelligence module (30) distinguish the cells that give signals that there is a problem by displaying different behavior among these distribution values. It is determined
(60) whether the battery cell (1 ) has a cell fault condition or not, with the input values passed through non-linear mathematical operations by the middle layers (324) of the neural network (32) included in the artificial intelligence module (30). Cell fault status is classified by the output layer (326) as low severity, medium severity, and high severity (70). The obtained information is converted into a result data (80) by the artificial intelligence module (30) and recorded in the memory unit (18).
In the working principle of the fault detection method according to the subject matter invention, the virtual data generator (12) in the control unit (10) creates the virtual model (20) of the battery pack. In the created battery pack virtual model (20), the voltage distribution values of the battery according to the changing parameters and the voltage threshold values according to the changing parameters are collected by the data collector (14) as data parts. A collected first piece of data, a second piece of data, and a third piece of data are formed according to the varying maximum battery charge and discharge interval and the measurement point corresponding to the temperature change. The data parts are combined into training data by the controller (16) in the control unit (10). Using the training data, the control unit (10) trains the artificial intelligence module (30). With the neuroprocessor and analytical module (40), cell voltage distribution values are generated as input data (50). The artificial intelligence module (30) is trained with data parts from the battery pack virtual model (20) until it learns completely to detect and distinguish cells that give early signals about problem formation by taking the battery cell voltage distribution values as input according to the changing parameters of the batteries and displaying different behavior from these distribution values. Training the artificial intelligence module (30) using training data could be, for example, a supervised learning model using a recurrent neural network (RNN) model or the like. After the artificial intelligence module (30) learns the algorithms to take the voltage distribution values of the battery cells as input data according to the changing parameters, it is trained to learn to detect and distinguish the cells that give early signals about the formation of the problem by showing different behavior from these distribution values for the cells (1 ) of the batteries. The voltage distribution value of the battery cell (1 ) is transmitted by the signal processor and analytical module (40) as input data by the artificial intelligence module (30), and the value of the cell voltage distribution is used by the artificial intelligence module (30) as input data, and different behavior is observed among these distribution values and detecting cells that give early signals of problem formation, it is determined whether the battery cell (1) has a cell fault condition (60). Depending on the severity of the cell fault, the artificial intelligence module (30) classifies the cell fault (70). The classified cell fault is converted into a result data (80) and stored in the memory unit.
In an exemplary application, LiNiMnCo02 (NMC), were chosen as the batteries utilized for an electric vehicle. A battery early fault detection algorithm is run by the artificial intelligence module (30) to detect whether the battery cells (1 ) have a cell fault. In an example application, the voltage distribution values of the battery cells (1) examined with the battery early fault diagnosis algorithm are determined between -30 °C and 45 °C while the electric vehicle is moving and the maximum discharge power of 200 kW is not exceeded. While the maximum discharge power of 200 kW at 25 °C temperature was not exceeded, when the voltage distribution values were examined, it was determined that there was no cell fault condition. While the maximum discharge power of 200 kW at 35 °C temperature was not exceeded, when the voltage distribution values were examined, it was determined that it had a cell fault condition. The detected voltage distribution values are processed with the artificial intelligence algorithm neural network (32) to obtain information whether there is an early signal of cell fault formation or not. If the cell has a fault, it is classified as low severity, average severity, or high severity cell fault, depending on the severity of the fault. In this way, whether the battery cell (1 ) used for the vehicle has a cell fault or not, the faults of the battery cell (1) are actively detected in the background while the vehicle is running, and it can be determined that whether it is not suitable for the use. If a battery with a cell fault can be made suitable for an active ride or a new battery can be used is determined by considering the severity of the cell fault.
REFERENCE NUMBERS
1 Battery Cell
10 Control Unit
12 Virtual Data Generator
14 Data Collector
16 Controller
18 Memory Unit
20 Battery Pack Virtual Model
30 Artificial Intelligence Module
32 Neural Network
322 Input Layer
324 Middle Layer
326 Output Layer
40 Signal Processor and Analytics Module 50 Creating Input Data 60 Determining Cell Fault Condition 70 Classification of Cell fault 80 Converting into Result Data
Claims
1- A fault detection method for a battery comprising creating (20) a battery pack virtual model providing a voltage threshold value indicating voltage distribution values of at least one battery cell (1 ) at the battery and usability of predefined a battery cell (1) by means of a virtual data generator (12) in a control unit (10); collecting a first data part and at least a second data part from the created battery pack virtual model by means of a data collector (14); generating a training dataset by merging the first and second data parts by a controller (16); training an artificial intelligence module (30) to detect and distinguish the cells that behave differently from distribution values by a controller (16) detecting the cell voltage distribution values according to the changing parameters by utilizing the training data and process as the input data characterized by comprising the steps of determining the cell voltage distributions of the battery cell (1) and generating an input data (50) by a neuroprocessor and analytics module (40) connected with the artificial intelligence module (30) in a signal transmitting manner; determining (60) whether the battery cell (1 ) has a cell fault condition, by means of artificial intelligence module (30) utilizing the input data detecting and distinguishing the battery cells (1 ) that behave differently from the distribution values.
2- A fault detection method in accordance with Claim 1 , wherein the artificial intelligence module (30) comprises the process step of classification of the cell fault status as a low severity level, average severity level and high severity level (70).
3- A fault detection method according to any of the previous claims, comprising the step of early detection of cell fault by detecting and distinguishing the cells with different behavior from these distribution values by means of a neural network (32) provided in the artificial intelligence module (30) taking the cell voltage distribution values as the input data and assigning the severity level of the cell fault depending on the cell voltage distribution.
4- A fault detection method according to claim 3, comprising the step of training the neural network (32) by supervised learning by utilizing training data comprising an input layer (322), at least one middle layer (324) and an output layer (326) assigning current value of the battery cell according to the changing parameters detecting early fault occurrence by determining and distinguishing cells with different behavior among the cell value distribution.
5- A fault detection method according to any of the previous claims, comprising the step obtaining the first and second data parts with a data collector by simultaneous and repeatedly measuring current value distribution of the battery cell (1 ) at the virtual model (20) of the battery
pack and matching the data and corresponding various parameters at each point of measurement.
6- A fault detection method according to claim 5, comprising the step of obtaining a training data by merging the first and the second data parts and battery cell (1) voltage distribution value at the point of measurement corresponding to the parameters selected from the group comprising the maximum usable battery charge, discharge interval and temperature change.
7- A fault detection method according to any of the previous claims, comprising the step of transforming the information classified by the artificial intelligence module (30) into a result data (80).
8- A fault detection method according to Claim 7, comprising the step of saving the result data in a memory unit (18) by means of the artificial intelligence module (30).
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP21925959.5A EP4291908A1 (en) | 2021-02-11 | 2021-02-11 | Early fault detection method for a battery |
PCT/TR2021/050125 WO2022173384A1 (en) | 2021-02-11 | 2021-02-11 | Early fault detection method for a battery |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/TR2021/050125 WO2022173384A1 (en) | 2021-02-11 | 2021-02-11 | Early fault detection method for a battery |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022173384A1 true WO2022173384A1 (en) | 2022-08-18 |
Family
ID=82838560
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/TR2021/050125 WO2022173384A1 (en) | 2021-02-11 | 2021-02-11 | Early fault detection method for a battery |
Country Status (2)
Country | Link |
---|---|
EP (1) | EP4291908A1 (en) |
WO (1) | WO2022173384A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116577673A (en) * | 2023-07-12 | 2023-08-11 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
EP4336194A1 (en) * | 2022-09-09 | 2024-03-13 | Volytica Diagnostics GmbH | Computer program and method for analysing inhomogeneities, and anomaly detection and prediction of electric energy stores |
CN118219910A (en) * | 2024-05-22 | 2024-06-21 | 浙江祥晋汽车零部件股份有限公司 | Power exchange station and power exchange method of power exchange station |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090112395A1 (en) * | 2007-10-31 | 2009-04-30 | Toyota Motor Engineering & Manufacturing North America, Inc.. | System for detecting a battery malfunction and performing battery mitigation for an hev |
CN104714175A (en) * | 2013-12-12 | 2015-06-17 | 北京有色金属研究总院 | Battery system fault diagnosis method and system |
CN111007401A (en) * | 2019-12-16 | 2020-04-14 | 国网江苏省电力有限公司电力科学研究院 | Electric vehicle battery fault diagnosis method and device based on artificial intelligence |
-
2021
- 2021-02-11 EP EP21925959.5A patent/EP4291908A1/en active Pending
- 2021-02-11 WO PCT/TR2021/050125 patent/WO2022173384A1/en active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090112395A1 (en) * | 2007-10-31 | 2009-04-30 | Toyota Motor Engineering & Manufacturing North America, Inc.. | System for detecting a battery malfunction and performing battery mitigation for an hev |
CN104714175A (en) * | 2013-12-12 | 2015-06-17 | 北京有色金属研究总院 | Battery system fault diagnosis method and system |
CN111007401A (en) * | 2019-12-16 | 2020-04-14 | 国网江苏省电力有限公司电力科学研究院 | Electric vehicle battery fault diagnosis method and device based on artificial intelligence |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP4336194A1 (en) * | 2022-09-09 | 2024-03-13 | Volytica Diagnostics GmbH | Computer program and method for analysing inhomogeneities, and anomaly detection and prediction of electric energy stores |
WO2024052188A1 (en) * | 2022-09-09 | 2024-03-14 | volytica diagnostics GmbH | Computer program and method for analysing inhomogeneities and anomaly detection and prediction of electrical energy stores |
BE1030866B1 (en) * | 2022-09-09 | 2024-04-09 | volytica diagnostics GmbH | Computer program and method for the analysis of inhomogeneities as well as anomaly detection and prediction of electrical energy storage systems |
CN116577673A (en) * | 2023-07-12 | 2023-08-11 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
CN116577673B (en) * | 2023-07-12 | 2023-09-12 | 深圳先进储能材料国家工程研究中心有限公司 | Distributed neural network-based energy storage power station fault diagnosis method and system |
CN118219910A (en) * | 2024-05-22 | 2024-06-21 | 浙江祥晋汽车零部件股份有限公司 | Power exchange station and power exchange method of power exchange station |
Also Published As
Publication number | Publication date |
---|---|
EP4291908A1 (en) | 2023-12-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ismail et al. | Battery state of charge estimation using an Artificial Neural Network | |
Li et al. | Big data driven lithium-ion battery modeling method based on SDAE-ELM algorithm and data pre-processing technology | |
WO2022173384A1 (en) | Early fault detection method for a battery | |
Zhang et al. | Synchronous estimation of state of health and remaining useful lifetime for lithium-ion battery using the incremental capacity and artificial neural networks | |
Rezvanizaniani et al. | Review and recent advances in battery health monitoring and prognostics technologies for electric vehicle (EV) safety and mobility | |
CN109061537B (en) | Electric vehicle lithium ion battery sensor fault diagnosis method based on observer | |
CN109116242B (en) | Data processing method and device for power battery | |
KR101992051B1 (en) | Method and system for predicting state of charge of battery | |
KR20180055192A (en) | Method and apparatus for estimating state of battery | |
Shah et al. | State of charge, remaining useful life and knee point estimation based on artificial intelligence and Machine learning in lithium-ion EV batteries: A comprehensive review | |
Zhang et al. | Intelligent state of charge estimation of battery pack based on particle swarm optimization algorithm improved radical basis function neural network | |
CN110568360B (en) | Lithium battery aging diagnosis method based on fuzzy logic algorithm | |
Xie et al. | Quantified assessment of internal short-circuit state for 18 650 batteries using an extreme learning machine-based pseudo-distributed model | |
CN116505629B (en) | Control management method, device, equipment and storage medium for solid-state battery | |
He et al. | Voltage abnormality-based fault diagnosis for batteries in electric buses with a self-adapting update model | |
Yao et al. | Fault identification of lithium-ion battery pack for electric vehicle based on ga optimized ELM neural network | |
Chen et al. | State of health estimation for lithium-ion battery based on long short term memory networks | |
Wang et al. | Charging diagnosis of power battery based on adaptive STCKF and BLS for electric vehicles | |
CN116819328A (en) | Electric automobile power battery fault diagnosis method, system, equipment and medium | |
Chen et al. | Meta rule-based energy management strategy for battery/supercapacitor hybrid electric vehicles | |
Yang et al. | Multi-fault diagnosis for battery pack based on adaptive correlation sequence and sparse classification model | |
Machlev | EV battery fault diagnostics and prognostics using deep learning: Review, challenges & opportunities | |
Zhou et al. | Data-driven battery state-of-health estimation and prediction using IC based features and coupled model | |
Saxena et al. | PHM of Li‐ion Batteries | |
Shete et al. | Fault diagnosis of electric vehicle’s battery by deploying neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21925959 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2021925959 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2021925959 Country of ref document: EP Effective date: 20230911 |