Thermal runaway prediction by detecting abnormal battery voltage and SOC degradation

Abstract

A vehicle, system, and method for monitoring a vehicle battery pack for the occurrence of thermal runaway. The system includes a plurality of voltage sensors and a processor. The plurality of voltage sensors obtain a plurality of voltage measurements at each of a plurality of battery cells in the battery pack. The processor is configured to determine an average value based on the plurality of voltage measurements, compare a voltage measurement obtained from a selected battery cell to the average value, and generate a notification signal when a difference between the voltage measurement from the selected battery cell and the average value is greater than or equal to a prediction threshold.

Description

Technical Field

[0001] The present invention relates to a system and method for monitoring the occurrence of thermal runaway in a battery pack, and more particularly to a system and method for predicting the onset of thermal runaway to allow preventative measures to be taken. Background Art

[0002] Electric vehicles operate using battery packs employing multiple battery cells. If a battery cell experiences an internal short circuit, it will self-discharge. The current flowing in the short circuit causes the battery cell's temperature to increase, which in turn reduces the internal short-circuit resistance, allowing the short-circuit current to increase. This feedback loop can lead to an uncontrolled heating condition known as thermal runaway, or TRA. If left unchecked, thermal runaway can cause battery failure. Therefore, it is desirable to provide predictive capabilities that can predict impending thermal runaway so that preventative measures can be taken before damage occurs to the battery pack and, consequently, the electric vehicle. Summary of the Invention

[0003] In one exemplary embodiment, a method for monitoring the occurrence of thermal runaway in a battery pack is disclosed. Multiple voltage measurements are obtained at each of a plurality of battery cells in the battery pack. An average value is determined based on the multiple voltage measurements. The voltage measurement obtained from a selected battery cell is compared to the average value. When the difference between the voltage measurement from the selected battery cell and the average value is greater than or equal to a prediction threshold, a notification signal is generated.

[0004] In addition to one or more features described herein, the prediction threshold is a multiple of the standard deviation of the plurality of voltage measurements. The method further includes determining a voltage parameter from the voltage measurements, wherein the voltage parameter is one of a voltage of a selected battery cell, a voltage residual of the selected battery cell, a state of charge of the selected battery cell, a rate of change of the voltage of the selected battery cell over time, and a rate of change of the state of charge of the selected battery cell over time. The method further includes obtaining a plurality of voltage measurements and a voltage measurement from the selected battery cell at the same time step. The method further includes comparing the rate of change of the voltage of the selected battery cell over a time window with a prediction threshold, wherein the prediction threshold is based on an average voltage of the battery pack over the time window and a capacity of the battery pack. The method further includes determining a short-circuit resistance within the selected battery cell based on a change in the deviation of the state of charge of the selected battery cell from the average value over time, wherein the change in the deviation is determined using cell balancing correction. The method further includes correcting the voltage of the selected battery cell for natural discharge due to cell balancing.

[0005] In another exemplary embodiment, a system for monitoring the occurrence of thermal runaway in a vehicle battery pack is disclosed. The system includes a plurality of voltage sensors and a processor. The plurality of voltage sensors obtain a plurality of voltage measurements at each of a plurality of battery cells in the battery pack. The processor is configured to determine an average value based on the plurality of voltage measurements, compare a voltage measurement obtained from a selected battery cell with the average value, and generate a notification signal when a difference between the voltage measurement from the selected battery cell and the average value is greater than or equal to a prediction threshold.

[0006] In addition to one or more features described herein, the prediction threshold is a multiple of a standard deviation of the plurality of voltage measurements. The processor is further configured to determine a voltage parameter based on the voltage measurements, wherein the voltage parameter is one of a voltage of a selected battery cell, a voltage residual of the selected battery cell, a state of charge of the selected battery cell, a rate of change of the voltage of the selected battery cell over time, and a rate of change of the state of charge of the selected battery cell over time. According to the system, the plurality of voltage sensors are further configured to obtain the plurality of voltage measurements and the voltage measurements from the selected battery cell at the same time step. The processor is further configured to compare the rate of change of the voltage of the selected battery cell over a time window to a prediction threshold, wherein the prediction threshold is based on an average voltage of the battery pack over the time window and a capacity of the battery pack. The processor is further configured to determine the resistance of the selected battery cell based on a change in the deviation of the state of charge of the selected battery cell from the average over time, wherein the change in the deviation is determined using cell balancing correction. The processor is further configured to correct the voltage of the selected battery cell for natural discharge due to cell balancing.

[0007] In yet another exemplary embodiment, a vehicle is disclosed. The vehicle includes a battery pack, a plurality of voltage sensors, and a processor. The battery pack includes a plurality of battery cells. The plurality of voltage sensors obtain a plurality of voltage measurements for each of the plurality of battery cells. The processor is configured to determine an average value based on the plurality of voltage measurements, compare a voltage measurement obtained from a selected battery cell with the average value, and generate a notification signal when a difference between the voltage measurement from the selected battery cell and the average value is greater than or equal to a prediction threshold.

[0008] In addition to one or more features described herein, the prediction threshold is a multiple of the standard deviation of the multiple voltage measurements. The processor is further configured to determine a voltage parameter based on the voltage measurements, wherein the voltage parameter is one of the voltage of the selected battery cell, the voltage residual of the selected battery cell, the state of charge of the selected battery cell, the rate of change of the voltage of the selected battery cell over time, and the rate of change of the state of charge of the selected battery cell over time. The multiple voltage sensors are further configured to obtain the multiple voltage measurements and the voltage measurements from the selected battery cell at the same time step. The processor is further configured to compare the rate of change of the voltage of the selected battery cell over a time window with a prediction threshold, and the prediction threshold is based on the average voltage of the battery pack over the time window and the capacity of the battery pack. The processor is further configured to determine the resistance of the battery cell based on the change in the deviation of the state of charge of the selected battery cell from the average value over time, wherein the change in the deviation is determined using cell balancing correction.

[0009] The above features and advantages and other features and advantages of the present disclosure will become apparent from the following detailed description when taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Additional features, advantages, and details appear, by way of example only, in the following detailed description, which refers to the accompanying drawings, in which:

[0011] Figure 1 An electric vehicle in an illustrative embodiment is shown;

[0012] Figure 2 A schematic diagram showing a battery unit of an electric vehicle;

[0013] Figure 3 A circuit diagram representing a battery cell in a short circuit condition is shown;

[0014] Figure 4 an illustrative graph showing the relationship between battery cell temperature and time for a battery cell having an internal short circuit;

[0015] Figure 5 An illustrative graph showing the evolution of battery cell voltage over time is shown;

[0016] Figure 6 shows a graph illustrating a method for detecting an internal short circuit and then predicting thermal runaway by measuring battery cell voltage;

[0017] Figure 7 shows a graph illustrating a method for detecting an internal short circuit and associated thermal runaway condition from a residual of a battery cell voltage;

[0018] Figure 8 Shown for prediction Figure 6 and 7 A flow chart of a method for a thermal runaway condition is shown;

[0019] Figure 9 shows an illustrative charging graph for a battery pack;

[0020] Figure 10 shows a graph illustrating circumstances under which an alarm or notification signal may be generated;

[0021] Figure 11 is a graph depicting the open circuit voltage of a battery cell as a function of the state of charge of the battery cell;

[0022] Figure 12 is a graph depicting the state of charge of a battery cell as a function of the open circuit voltage of the battery cell;

[0023] Figure 13 showing a graph illustrating a second method for measuring voltage drop in a battery cell;

[0024] Figure 14 shows a graph illustrating a third method for determining the presence of an internal short circuit resistance;

[0025] Figure 15 A flow chart showing a method for predicting thermal runaway based on changes in battery cell charging stages; and

[0026] Figure 16 A flow chart illustrating a method for predicting a thermal runaway condition and providing an alert is shown. DETAILED DESCRIPTION

[0027] The following description is merely exemplary in nature and is in no way intended to limit the present disclosure, its application, or uses. It should be understood that throughout the drawings, corresponding reference numerals indicate like or corresponding parts and features.

[0028] According to an exemplary embodiment, Figure 1 An electric vehicle 100 is shown. The electric vehicle 100 includes a battery pack 102, an electrical load 108 that operates using power supplied by the battery pack, and a control system 110 that monitors the battery pack. The battery pack 102 includes a plurality of battery cells 104a...104n. A plurality of voltage sensors 106a...106n obtain voltage measurements from the plurality of battery cells 104a...104n, respectively, and transmit the voltage measurements to the control system 110.

[0029] The electrical loads 108 may include motors and / or other electrical components of the electric vehicle 100, such as dashboard lights, exterior lights, entertainment systems, etc. The control system 110 includes a processor 112 and a memory storage device 114 storing various programs or instructions 116. The processor 112 may access the programs or instructions from the memory storage device 114 and execute the programs or instructions to perform various operations disclosed herein for predicting the onset of thermal runaway (TRA) and generating appropriate alerts or taking appropriate actions.

[0030] The control system 110 is in communication with the alarm notification unit 118 and can notify the alarm notification unit when a thermal runaway event is predicted based on the calculations disclosed herein. In one embodiment, the alarm notification unit 118 can send a message to a user, such as The remote server 120 may send a notification signal to the remote server 120. The remote server 120 may respond to the notification signal by alerting local emergency response units, such as firefighters. In other embodiments, the alarm notification unit 118 may provide a visual signal to the vehicle driver or sound an alarm. In various embodiments, rather than generating an alarm, the data may be sent to the remote server 120 for processing and review at a later time.

[0031] Figure 2 A schematic diagram 200 of a battery cell (e.g., battery cell 104a) is shown. Battery cell 104a includes a cathode 202, an anode 204, and an insulating medium 206 separating the cathode from the anode. Insulating medium 206 prevents current from flowing between cathode 202 and anode 204. Schematic diagram 200 also shows a short circuit 208 across insulating medium 206, which may occur due to degradation of the insulating medium or due to excessive use of battery cell 104a.

[0032] Figure 3 A circuit diagram 300 is shown representing a battery cell 104a in a short circuit condition. The circuit diagram 300 includes a voltage source 302, an internal resistance R 304 of the battery cell, and an internal short circuit resistance R 306 due to the internal short circuit. SC 306. The battery terminal voltage Vt of the circuit diagram 300 is given by equation (1):

[0033]

[0034] Where V OC is the open circuit voltage of the battery. V OC Depends on the state of charge (SOC) of the battery at a given time step k. The change in state of charge at a given time is given by equation (2):

[0035]

[0036] Where SOC(k+1) is the state of charge at the (k+1)th time step, and SOC(k) is the state of charge at the kth time step. Cap is the cell capacity of the battery cell, and dt is the sampling time interval between the kth time step and the k+1th time step. Equation (2) can be rearranged to account for the internal short-circuit resistance, as shown in Equation (3).

[0037]

[0038] Figure 4 A graph 400 shows the relationship between battery cell temperature and time in an illustrative embodiment of a battery cell having an internal short circuit. Time is shown in seconds along the abscissa and temperature is shown in degrees Celsius along the ordinate. The internal short circuit resistance is R SC =0.39Ω. The heat transfer equation for the battery cell is given in equation (4).

[0039]

[0040] Among them, M b is the battery cell mass, C pb is the thermal coefficient, Tb is the battery cell temperature, Tc is the coolant temperature, Ti is the temperature of the adjacent battery cell, and hA is the heat transfer coefficient or thermal conductivity. During use, the temperature of the battery increases over time. Point 402 represents the temperature at which thermal runaway occurs.

[0041] Figure 5 A graph 500 is shown, illustrating the evolution of battery cell voltage over time in an illustrative embodiment. Time is shown in seconds along the abscissa, and voltage is shown along the ordinate. A first curve 502 illustrates the voltage of a normal battery cell. A second curve 504 illustrates the voltage of a battery cell with an internal short circuit. Over time, the voltage of the battery cell with an internal short circuit (i.e., second curve 504) deviates from the voltage of the normal battery cell (i.e., first curve 502). Graph 500 illustrates a fault time 506, at which the internal short circuit occurred, and a TRA time 508, at which TRA occurs due to the internal short circuit at fault time 506. A time window 510 marks the time interval during which voltage can be measured to predict the onset of TRA.

[0042] Figure 6 A graph 600 is shown that illustrates a method for detecting an internal short circuit and then predicting thermal runaway by measuring battery cell voltages. Time is shown in seconds along the abscissa and voltage is shown along the ordinate. Graph 600 includes a set of voltages 602 for healthy battery cells from a battery pack. Graph 600 also illustrates the thermal runaway behavior of various internal short circuit resistances R SC The voltage curve 604 corresponds to the voltage of the battery cell RSC =0.3Ω. Voltage curve 606 corresponds to R SC =0.43Ω. Voltage curve 608 corresponds to R SC =0.88Ω. Voltage curve 610 corresponds to R SC =2Ω. Voltage curve 612 corresponds to R SC =7Ω. Voltage curve 614 corresponds to Rsc=15Ω. Point 618 represents the time on voltage curve 608 when thermal runaway occurs.

[0043] Time window 620 is used to determine the presence of an internal short circuit. Time window 620 can be a moving time window and includes a back time edge 622 and a current time edge 624 that are held for a selected duration relative to each other. In one embodiment, the calculation is performed using measurements obtained at the current time edge 624.

[0044] At the current time edge 624, a voltage measurement is obtained for each of the "n" cells in the battery pack. An average or mean voltage is determined from the voltages. A plurality of prediction thresholds are established based on the average voltage and the standard deviation about the average voltage. In various embodiments, the prediction thresholds are separated from the average by an integer multiple of the standard deviation. However, this is not a necessary limitation of the present invention. For illustrative purposes, a first prediction threshold 630 is located at 3σ from the average, a second prediction threshold 632 is located at 4σ from the average, and a third prediction threshold 634 is located at 6σ from the average. Multiple prediction thresholds are used to test for conditions that lead to thermal runaway. Each battery cell voltage is tested against the prediction threshold. If one of the battery cell voltages exceeds the average voltage by an amount greater than the plurality of prediction thresholds, an alarm is generated.

[0045] As an example, the voltage curve 612 is between the second prediction threshold 632 and the third prediction threshold 634 at the current time edge, as shown at point 636. In this case, no alarm is generated. In another example, as shown at point 638, the voltage curve 608 is outside the third prediction threshold 634 at the current time edge, so an alarm is generated.

[0046] In one embodiment, the average voltage is determined using only voltage measurements taken at the current time edge 624. In another embodiment, the average voltage and standard deviation σ are determined using voltage measurements taken during the time window 620 (i.e., between the subsequent time edge 622 and the current time edge 624) to provide a more accurate average voltage.

[0047] Figure 7 A graph 700 is shown illustrating a method for detecting internal short circuits and associated thermal runaway conditions from battery cell voltage residuals. Time is shown in seconds along the abscissa and residual values ​​are shown along the ordinate axis.

[0048] The residual of the jth battery cell at the i-th time step is given by the difference between the battery cell voltage and the average voltage of a group of battery cells in the battery module or battery pack at the i-th time step, as shown in equation (5):

[0049] r j (i) = V j (i)-V mean (i) Equation (5)

[0050] Where j = 1 ... n, and n is the number of battery cells in the battery module or battery pack. Graph 700 includes a residual group 702 of the residuals of the normal battery cells of the battery pack. An average value or average residual is derived from the residuals of n battery cells. Graph 700 also shows the residuals of the normal battery cells of the battery pack with various internal short-circuit resistances R SC The residual curve 708 corresponds to the residual of the battery cell. SC = 0.3 ohms. The residual curve 710 corresponds to R SC = 0.43 ohms. The residual curve 712 corresponds to R SC = 0.88 ohms. The residual curve 714 corresponds to R SC = 2 ohms. The residual curve 716 corresponds to R SC =7 ohms. The residual curve 718 corresponds to R SC =15 ohms.

[0051] Time window 720 shows a time interval over which measurements can be obtained to perform calculations for predicting the start of TRA. Time window 720 includes a back time edge 722 and a current time edge 724. The calculation using the residual Figure 6 Same as the description of battery cell voltage in .

[0052] Figure 7 The prediction thresholds shown in are integer multiples of the standard deviation of the mean residual. Multiple prediction thresholds are used to test for conditions that lead to thermal runaway. The prediction thresholds are established based on the mean residual and the standard deviation about the mean residual. In various embodiments, the prediction thresholds are separated from the mean residual by integer multiples of the standard deviation. However, this is not a necessary limitation of the present invention. For illustrative purposes, the first residual prediction threshold 704 is at 4σ and the second residual prediction threshold 706 is at 6σ. The residual of each battery cell is tested against the prediction threshold. If one of the residuals exceeds the mean voltage by more than the prediction threshold, an alarm is generated.

[0053] Figure 8 shows the predictions for Figure 6 and 7Flowchart 800 of a method for a thermal runaway condition is shown. The method begins at block 802. At block 804, the method tests whether the high voltage contacts are disconnected or whether constant current charging of the battery is occurring. If these conditions do not exist, the method loops back to block 802. If either condition exists, the method proceeds to block 806. In block 806, a mean and standard deviation are calculated based on voltage measurements obtained from the battery pack. The mean and standard deviation are based on measurements obtained at the current time edge of the time window. In block 808, a prediction threshold is set based on the standard deviation. For illustrative purposes, three prediction thresholds are established. Each cell voltage or residual is tested against the prediction threshold established in block 808 to determine whether a TRA condition exists.

[0054] Blocks 810, 812, and 814 provide a test cycle for the battery voltage or residual. In block 810, the battery voltage or residual is compared to a first prediction threshold. If the voltage or residual is less than or equal to the first prediction threshold, the method loops back to block 810. If the battery voltage or residual is greater than the first prediction threshold, the method continues to block 812. In block 812, the battery voltage or residual is compared to a second prediction threshold. If the voltage or residual is less than or equal to the second prediction threshold, the method loops back to block 812. If the battery voltage or residual is greater than the second prediction threshold, the method continues to block 814. In block 814, the battery voltage or residual is compared to a third prediction threshold. If the voltage or residual is less than or equal to the third prediction threshold, the method loops back to block 814. If the battery voltage or residual is greater than the third prediction threshold, the method continues to block 816. When testing the battery voltage or residual in blocks 810, 812, and 814, the cycle time dt is measured to determine how long it takes for the battery voltage to pass the prediction threshold.

[0055] At block 816, a test is performed to determine the validity of the test cycle. If the cycle time dt is less than or equal to a selected time threshold, the method proceeds to block 818, where it is determined that the predictive test is invalid. In various embodiments, the selected time threshold may be a predetermined fraction of a second. If, at block 816, the cycle time dt is greater than the selected time threshold, the method proceeds to block 820. At block 820, an alarm is generated, and if the battery is currently charging, the charging process is stopped.

[0056] Figure 9An illustrative charging graph 900 for a battery pack is shown. Time is shown in seconds along the abscissa and voltage is shown along the ordinate. Group curve 902 shows the charging voltage of a plurality of normal battery cells of the battery pack. Curve 904 shows the charging voltage of a battery cell having an internal short-circuit resistance. Point 906 indicates the time at which thermal runaway occurs. Point 908 indicates the time at which an alarm can be generated using the methods disclosed herein. Obviously, an alarm can be issued before thermal runaway occurs, so that there is sufficient time to take preventative measures. In the illustrative charging graph 900, a warning is provided approximately 1500 seconds (approximately 25 minutes) before thermal runaway occurs.

[0057] Figure 10 Graph 1000 is shown, illustrating circumstances under which an alarm or notification signal may be generated. Time is shown in seconds along the abscissa, and voltage is shown along the ordinate. A battery pack curve 1002 shows the residual charge voltage of a plurality of normal battery cells of the battery pack over time. Curve 1004 shows a battery cell approaching thermal runaway. Curve 1004 crosses the outermost prediction threshold 1006 at approximately 2400 seconds, generating an alarm.

[0058] Figure 11 is a graph 1100 depicting the open circuit voltage V of a battery cell as a function of the state of charge SOC of the battery cell. OC The curve of the graph 1100 represents the function f used to convert the state of charge to the voltage measurement value. soc2voc . Figure 12 is a graph 1200 depicting the open circuit voltage V OC The curve of the graph 1200 represents the function f for converting the voltage measurement value to the state of charge SOC of the battery cell. voc2soc . Figure 11 and 12 The curve is determined by the battery chemistry.

[0059] Figure 13 A graph 1300 is shown illustrating a second method for measuring voltage drop in a battery cell. The second method uses voltage measurements taken at two different time steps within a time interval and determines TRA by comparing the voltage change within the time interval to one or more prediction thresholds. The prediction thresholds are based in part on an average of the voltage measurements over a period of time.

[0060] Graph 1300 shows the voltage measurements of the jth battery cell of a battery pack. The voltage measurements are obtained at periodic intervals, as shown by V(1), V(2), V(3), ..., V(i), where V(i) is the i-th measurement. The fundamental time interval separating two temporally adjacent voltage measurements is the time interval ΔT. Thus, the voltage V(i) obtained at the i-th time step is separated from the first voltage V(1) obtained at the first time step by iΔT.

[0061] Once the measurement is obtained, if cell balancing is in progress within the battery, the voltage is corrected for natural discharge. For the voltage measurement V(i), use Figure 12 The function shown in Figure 5 is used to determine the corresponding state of charge SOC(i). Then, the effect of natural battery discharge is eliminated from SOC(i) by the steps shown in Equation (6):

[0062] SOC(i)→SOC(i)+Ah(i) / Cap Equation (6)

[0063] Where Ah(i) is the equilibrium ampere-hours (Ah) of natural discharge and Cap is the battery capacity. Then use Figure 11 The function shown determines the corrected voltage from the corrected SOC. The change in battery voltage within the time window i*ΔT is calculated as shown in equation (7).

[0064] dV(i)=V(i+1)-V(1) Equation (7)

[0065] For example, dV(1) = V(2) - V(1) and dV(3) = V(3) - V(1)

[0066] In one embodiment, the voltages from multiple cells in a module or pack may be used to determine an average value dV m (i). If the average value dV m (i) is substantially greater than or larger than a selected threshold, then that value may be subtracted from each voltage measurement before determining the rate of change between time intervals.

[0067] To monitor the internal short circuit of the battery cell, the voltage drop rate dV of the jth battery cell at the i-th time step is tested against the prediction threshold Thr shown in equation (8). j (i)

[0068]

[0069] Where the predicted threshold is on the right-hand side of equation (8). Cap is the cell capacity of the battery cell, and x is a specified cell short-circuit resistance, such as approximately 500Ω. In one embodiment, the average voltage is given by equation (9):

[0070]

[0071] In another embodiment, the threshold Thr is the difference between the average value of the voltage drops across n battery cells in the battery module or battery pack and a multiple of the standard deviation σ, as shown in equation (10):

[0072]

[0073] A similar calculation can be made using the rate of decrease in state of charge (SOC). Figure 12 The function represented in determines the SOC from the voltage measurement, as shown in the function of equation (11).

[0074] SOC(i)=f voc2soc (V(i)) Equation (11)

[0075] Therefore, the change in SOC is given by equation (12):

[0076] dSOC j (i) = f voc2soc (V j (i))-f voc2soc (V j (1)) Equation (12)

[0077] Similar to the voltage drop method, if cell balancing of the battery pack is being performed, the natural discharge caused by cell balancing is removed from the determined state of charge shown in equation (11). Before calculating the change in SOC in equation (12), the change in the SOC of the battery cell over the time frame is compared with the predicted threshold value, as shown in equation (13):

[0078]

[0079] If dSOC j (i) is greater than a threshold, an alarm is generated. In another embodiment, the threshold is based on the difference between the average ΔSOC (dSOC) across all cells of the battery pack and a multiple of the standard deviation of ΔSOC, as shown in equation (14).

[0080]

[0081] Figure 14 A graph 1400 illustrating a third method for determining the presence of an internal short circuit resistance is shown. Voltage measurements are obtained for each of the 'n' battery cells of the battery pack at a first time (eg, approximately 400 seconds) and a second time (eg, approximately 1600 seconds). Figure 12 The function expressed in determines the state of charge SOC of the jth battery cell at the i-th time step from the voltagej (i).

[0082] Figure 15 A flowchart 1500 is shown of a method for predicting TRA based on changes in the charge stage of a battery cell. At block 1502, voltage measurements are obtained for 'n' battery cells of a battery pack and a first sampling time (i=1). At block 1504, i=1 (SOC) is determined for each of the 'n' cells. j At block 1506, the average state of charge value of the 'n' cells is determined, and the deviation of the SOC of the 'n' cells is determined, as given in equation (15):

[0083]

[0084] Where dSOC j (1) is the deviation of the state of charge of the jth cell at time i=1, is the average state of charge of 'n' cells at time i=1, and SOC j (1) is the SOC of the jth cell at time i=1.

[0085] At block 1508, voltage measurements are obtained for the 'n' cells at a second sampling time (i=2). At block 1510, the state of charge (SOC) at i=2 is determined for each of the 'n' cells. j (2)). At block 1512, the average state of charge value of the 'n' cells is determined, and the deviation of the SOC of each of the 'n' cells is determined, as given in equation (16):

[0086]

[0087] Where dSOC j (2) is the deviation of the state of charge of the jth cell at time i=2, is the average state of charge of 'n' cells at time i=2, and SOC j (2) is the SOC of the jth cell at time i=2.

[0088] In block 1514, the change in state of charge between time i=1 and time i=2 is determined. The effect of cell balancing on a single cell can be compensated based on using a balance percentage based on ampere hours over the time interval Δt. In one embodiment, the balance percentage is given by equation (17):

[0089]

[0090] where Ahr(j) is the equilibrium ampere-hour of the jth unit, is the average ampere-hour of the 'n' cells in the battery pack, and Cap is the cell capacity. The change in dSOCs of the jth cell is given by Equation (18):

[0091] ddSOC(j)=dSOC j (2)-dSOC j (1)+BalPct Equation (18)

[0092] In block 1516, the short-circuit resistance R is then calculated based on the result of equation (18), as shown in equation (19):

[0093]

[0094] In block 1518, each short circuit resistance Rj is compared to a resistance threshold, and if Rj is less than the resistance threshold, an alarm is generated, indicating a short circuit.

[0095] Figure 16 A flowchart 1600 of a second method for predicting a thermal runaway condition and providing an alert is shown. The method is applied to each cell and begins at box 1602. At box 1604, the method tests whether the high voltage contacts are disconnected or whether constant current charging of the battery is occurring. If these are not present, the method loops back to box 1602. If either condition exists, the method proceeds to box 1606. In box 1606, an average and standard deviation are calculated based on measurements taken from the battery pack. The measurement is the voltage difference of the battery cells calculated by equation (7). In one embodiment, the average can be the average of the voltage differences of all cells in the module or battery pack, and the standard deviation is the standard deviation of these voltage differences. In box 1608, a prediction threshold is set based on calibration or based on the standard deviation. The threshold can be, for example, the prediction threshold of equations (8) and (10), or a derivative thereof. In box 1610, the voltage deviation of 'n' cells is measured at different time intervals (i·ΔT). In box 1612, the deviation of the cell at time (i) is measured relative to the threshold. If the deviation is less than the threshold, the method proceeds to block 1614 where the time interval is increased by one time step. From block 1614, the method loops back to block 1612. If the deviation is greater than the threshold, the method proceeds to block 1616 where an alarm is generated. Figure 16 A similar prediction flow chart in can be applied to the deviation of the state of charge, where the voltage deviation-based method dV(i) can be replaced with dSOC(i) and tested against the thresholds in equations (13) and (14).

[0096] Although the above disclosure has been described with reference to exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope thereof. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the present disclosure without departing from the essential scope of the present disclosure. Therefore, it is intended that the present disclosure is not limited to the particular embodiments disclosed, but is intended to include all embodiments falling within its scope.

Claims

1. A method for monitoring the occurrence of thermal runaway in a battery pack, comprising: obtaining a plurality of voltage measurements at each of a plurality of battery cells of a battery pack; determining an average value based on the plurality of voltage measurements; comparing voltage measurements obtained from the selected battery cell with the average value; and generating a notification signal when a difference between a voltage measurement from a selected battery cell and an average value is greater than or equal to a prediction threshold; A short circuit resistance within the selected battery cell is determined based on a change in a state of charge of the selected battery cell from an average value over time, wherein the change in the deviation is determined using a cell balancing correction.

2. The method according to claim 1 also includes determining a voltage parameter from the voltage measurement value, wherein the voltage parameter is one of the following: (i) the voltage of the selected battery cell; (ii) the voltage residual of the selected battery cell; (iii) the charge state of the selected battery cell; (iv) the rate of change of the voltage of the selected battery cell over time; and (v) the rate of change of the charge state of the selected battery cell over time. 3 . The method of claim 1 , further comprising obtaining a plurality of voltage measurements and voltage measurements from selected battery cells at the same time step. 4 . The method of claim 1 , further comprising comparing a voltage change rate of the selected battery cell over a time window with a prediction threshold, wherein the prediction threshold is based on an average voltage of the battery pack over the time window and a capacity of the battery pack.

5. A system for monitoring the occurrence of thermal runaway in a vehicle battery pack, comprising: a plurality of voltage sensors for obtaining a plurality of voltage measurements at each of a plurality of battery cells of the battery pack; and The processor is configured to: determining an average value based on the plurality of voltage measurements; comparing voltage measurements obtained from the selected battery cell with the average value; and generating a notification signal when a difference between a voltage measurement from a selected battery cell and an average value is greater than or equal to a prediction threshold; The processor is further configured to determine a resistance of the selected battery cell based on a change in a deviation of a state of charge of the selected battery cell from an average value over time, wherein the change in the deviation is determined using a cell balancing correction.

6. The system according to claim 5, wherein: The processor is also configured to determine a voltage parameter based on the voltage measurement value, wherein the voltage parameter is one of the following: (i) the voltage of the selected battery cell; (ii) the voltage residual of the selected battery cell; (iii) the charge state of the selected battery cell; (iv) the rate of change of the voltage of the selected battery cell over time; and (v) the rate of change of the charge state of the selected battery cell over time.

7. The system according to claim 5, wherein: The plurality of voltage sensors are further configured to obtain the plurality of voltage measurements and the voltage measurement from the selected battery cell at a same time step.

8. The system of claim 5, wherein the processor is further configured to compare a voltage change rate of the selected battery cell over a time window with the prediction threshold, and the prediction threshold is based on an average voltage of the battery pack over the time window and a capacity of the battery pack.

Images