ANOMALY DETECTION IN ENERGY STORAGE SYSTEMS

1 . A system comprising:

one or more energy storage units; and

a computing device comprising at least one processor and a memory storing instructions that, when executed by the at least one processor, cause the computing device to:

receive usage data of a battery located within the one or more energy storage units during a time period;

input the usage data to a machine learning model;

generate, based on processing of the usage data by the machine learning model, a predicted temperature of the battery at the end of the time period;

receive, from a temperature sensor of the battery, a measured temperature of the battery at the end of the time period;

determine a difference between the predicted temperature and the measured temperature;

based on the determined difference, send an indication of a state of the battery; and

based on the state of the battery, configure usage of the battery.

2 . The system of claim 1 , wherein each energy storage unit of the one or more energy storage units comprises an enclosure including a plurality of batteries.

3 . The system of claim 1 , wherein the usage data of the battery comprises one or more of:

a starting state of charge of the battery for the time period;

an ending state of charge of the battery for the time period;

a sum of current squared of the battery for the time period;

a voltage pattern of the battery associated with the time period;

a dispatch pattern associated with the battery for the time period;

a measured temperature of the battery at the start of the time period;

an ambient temperature or humidity associated with the battery during the time period;

one or more utilization parameters of a cooling system for the battery during the time period; or

a relative location of the battery within an enclosure of an energy storage unit including the battery.

4 . The system of claim 1 , wherein the machine learning model comprises one of a random forest model or a neural network.

5 . The system of claim 1 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

receive particular usage data of the battery during periods of time when the battery is manually deemed to exhibit normal behavior;

receive measured temperature data of the battery at the end of each of the periods of time; and

train the machine learning model using the particular usage data and the measured temperature data.

6 . The system of claim 1 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

for each particular battery of a plurality of batteries of the one or more energy storage units:

generate, using the machine learning model, a predicted temperature of the particular battery at the end of the time period;

receive a measured temperature of the particular battery at the end of the time period; and

determine a temperature difference between the predicted temperature of the particular battery and the measured temperature of the particular battery.

7 . The system of claim 6 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

cause display of a user interface indicating the plurality of batteries, relative locations of the plurality of batteries, and the temperature difference for each of the plurality of batteries.

8 . The system of claim 7 , wherein the user interface indicates the relative locations of the plurality of batteries in a three-dimensional perspective, and indicates the temperature difference for each of the plurality of batteries using a color scale.

9 . The system of claim 1 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

determine, for the battery, a difference between a predicted temperature and a measured temperature for each of a plurality of periods of time; and

determine, based on the difference for each of the plurality of periods of time, whether an anomaly of the battery is detected.

10 . The system of claim 1 , wherein the computing device is associated with a cloud architecture.

11 . The system of claim 1 , wherein the computing device is local to the one or more energy storage units.

12 . The system of claim 1 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

based on the difference satisfying a threshold, send an indication that an anomaly of the battery is detected; and

based on detecting the anomaly of the battery, adjust the usage of the battery.

13 . The system of claim 12 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

input the usage data to a second machine learning model;

generate, based on processing of the usage data by the second machine learning model, a predicted voltage of the battery at the end of the time period;

receive, from a voltage sensor of the battery, a measured voltage of the battery at the end of the time period;

determine a voltage difference between the predicted voltage and the measured voltage; and

send the indication that the anomaly of the battery is detected when the voltage difference satisfies a threshold.

14 . The system of claim 12 , wherein the instructions, when executed by the at least one processor, cause the computing device to:

adjust the usage of the battery by one or more of: suspending the usage of the battery, reducing the usage of the battery, or modifying a usage pattern of the battery.

15 . A method comprising:

receiving, by a computing device, usage data of a battery located within one or more energy storage units during a time period;

inputting the usage data to a machine learning model;

generating, based on processing of the usage data by the machine learning model, a predicted temperature of the battery at the end of the time period;

receiving, from a temperature sensor of the battery, a measured temperature of the battery at the end of the time period;

determining a difference between the predicted temperature and the measured temperature;

based on the determined difference, sending an indication of a state of the battery; and

based on the state of the battery, configuring usage of the battery.

16 . The method of claim 15 , further comprising:

for each particular battery of a plurality of batteries of the one or more energy storage units:

generating, using the machine learning model, a predicted temperature of the particular battery at the end of the time period;

receiving a measured temperature of the particular battery at the end of the time period; and

determining a temperature difference between the predicted temperature of the particular battery and the measured temperature of the particular battery.

17 . The method of claim 16 , further comprising:

causing display of a user interface indicating the plurality of batteries, relative locations of the plurality of batteries, and the temperature difference for each of the plurality of batteries.

18 . A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause the at least one processor to:

receive usage data of a battery located within one or more energy storage units during a time period;

input the usage data to a machine learning model;

generate, based on processing of the usage data by the machine learning model, a predicted temperature of the battery at the end of the time period;

receive, from a temperature sensor of the battery, a measured temperature of the battery at the end of the time period;

determine a difference between the predicted temperature and the measured temperature;

based on the determined difference, send an indication of a state of the battery; and

based on the state of the battery, configure usage of the battery.

19 . The non-transitory computer-readable medium of claim 18 , wherein the instructions, when executed by the at least one processor, cause the at least one processor to:

for each particular battery of a plurality of batteries of the one or more energy storage units:

generate, using the machine learning model, a predicted temperature of the particular battery at the end of the time period;

receive a measured temperature of the particular battery at the end of the time period; and

determine a temperature difference between the predicted temperature of the particular battery and the measured temperature of the particular battery.

20 . The non-transitory computer-readable medium of claim 19 , wherein the instructions, when executed by the at least one processor, cause the at least one processor to:

cause display of a user interface indicating the plurality of batteries, relative locations of the plurality of batteries, and the temperature difference for each of the plurality of batteries.