Electric vehicle battery monitoring system
Systems and methods for monitoring and controlling a battery are disclosed. Systems can include a battery having an output voltage and an output current when delivering power, a load driven by power delivered from the battery, battery output voltage and current sensing circuits, and processing circuitry coupled to the battery output voltage and current sensing circuits. The processing circuitry may implement a recurrent neural network for battery state estimation.
1. A battery powered apparatus comprising:
a battery comprising one or more electrochemical cells, the battery having an output voltage and an output current when delivering power;
a load driven by power delivered from the battery;
a battery output current sensing circuit; and
a battery management system comprising processing circuitry coupled to the battery output current sensing circuit, the processing circuitry configured to model a predicted output voltage of the battery with a battery model neural network comprising:
a first recurrent neural network comprising a first input layer, a first hidden layer, and a first output layer, wherein a node of the first output layer generates a predicted temperature of the battery, and
a second neural network comprising a second input layer, a second hidden layer, and a second output layer, wherein the second input layer includes a node that receives the predicted temperature from the node of the first output layer of the first recurrent neural network, and wherein the second output layer includes a node that generates the predicted output voltage of the battery,
wherein a particular node of the first hidden layer receives a first input from a node of the first input layer and a second input from a previous time state of the first hidden layer, and wherein a particular node of the second hidden layer receives a first input from a node of the second input layer and a second input from a previous time state of the second hidden layer.
2. The battery powered apparatus of claim 1 , wherein the second input layer of the second recurrent neural network has two additional input nodes corresponding to battery output current and integrated battery output current, respectively.
3. The battery powered apparatus of claim 1 , wherein the first recurrent neural network includes first gating for generating the previous time state of the first hidden layer.
4. The battery powered apparatus of claim 1 , further comprising:
a sensor configured for sensing an output voltage of the battery; and
a temperature sensor configured for sensing a temperature of the battery.
5. The battery powered apparatus of claim 4 , wherein the battery management system is configured to:
retrain the first neural network based on a difference between the predicted temperature and the sensed temperature; and
retrain the second neural network based on a difference between the predicted temperature and the sensed output temperature.
6. The electric vehicle of claim 4 , wherein the processing circuitry is configured to:
retrain the first neural network based on a difference between the predicted temperature and the sensed temperature; and
retrain the second neural network based on a difference between the predicted temperature and the sensed output temperature.
7. A method of operating a battery powered apparatus, the method comprising:
driving a load of the apparatus with a battery;
sensing the output current of the battery at defined intervals while driving the load;
predicting an output voltage that will be exhibited by the battery under selected output current conditions using at least first and second recurrent neural networks each comprising an input layer, a hidden layer, and an output layer, wherein a particular node of the hidden layer receives a first input from a node of the input layer node and a second input from a previous time state of the hidden layer;
using the first recurrent neural network to generate a predicted temperature of the battery;
providing at least the predicted temperature as an input to the second neural network, wherein the second neural network generates the predicted output voltage; and
driving the load of the apparatus with the selected current sourced from the battery.
8. The method of claim 7 , comprising sensing an output voltage of the battery and sensing a temperature of the battery.
9. The method of claim 8 , comprising, while driving the load of the apparatus with the battery:
retraining the first neural network based on a difference between the predicted temperature and the sensed temperature; and
retraining the second neural network based on a difference between the predicted temperature and the sensed output temperature.
10. An electric vehicle with a battery monitoring system, the vehicle comprising:
a battery;
a powertrain comprising at least one electric motor driven by the battery;
a battery output current sensing circuit; and
processing circuitry coupled to a battery output voltage sensing circuit and the battery output current sensing circuit, the processing circuitry configured to model behavior of the battery with at least first and second recurrent neural networks each comprising an input layer, a hidden layer, and an output layer, wherein a particular node of the hidden layer receives a first input from a node of the input layer and a second input from a previous time state of the hidden layer, wherein the first recurrent neural network models a predicted temperature of the battery, and wherein the processing circuitry provides the predicted temperature as input into the second recurrent neural network.
11. The electric vehicle of claim 10 , wherein the second recurrent neural network has a single output layer node corresponding to a predicted output voltage of the battery.
12. The electric vehicle of claim 10 , wherein the second recurrent neural network has a first input layer node corresponding to the predicted temperature and two additional input layer nodes corresponding to battery output current and integrated battery output current, respectively.
13. The electric vehicle of claim 12 , comprising powertrain control circuitry that is configured to receive an output from the second recurrent neural network and control the powertrain based at least in part on the output.
14. The electric vehicle of claim 10 , wherein the first recurrent neural network has a single output layer node corresponding to the predicted temperature.
15. The electric vehicle of claim 14 , wherein the second neural network has an input layer node corresponding to the single output layer node of the first recurrent neural network.
16. The electric vehicle of claim 15 , wherein the second recurrent neural network has a single output layer node that generates a predicted output voltage of the battery.
17. The electric vehicle of claim 10 , wherein the processing circuitry is further configured to calculate a state of health of the battery using outputs of at least one of the first and second recurrent neural networks.
18. The electric vehicle of claim 10 , wherein the processing circuitry is further configured to estimate an instantaneous power limit of the battery pack in real time based at least in part on outputs of at least one of the first and second recurrent neural networks.
19. The electric vehicle of claim 18 , further comprising updating a pedal map of the electric vehicle based on the estimated instantaneous power limit.
20. The electric vehicle of claim 10 , further comprising:
a sensor configured for sensing an output voltage of the battery; and
a temperature sensor configured for sensing a temperature of the battery.